14.11.2012 Views

Faecal occult blood testing for population health screening May 2004

Faecal occult blood testing for population health screening May 2004

Faecal occult blood testing for population health screening May 2004

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong><br />

<strong>testing</strong> <strong>for</strong><br />

<strong>population</strong> <strong>health</strong><br />

<strong>screening</strong><br />

<strong>May</strong> <strong>2004</strong><br />

MSAC reference 18<br />

Assessment report


© Commonwealth of Australia <strong>2004</strong><br />

ISBN 0 642 82545 9<br />

ISSN (Print) 1443-7120<br />

ISSN (Online) 1443-7139<br />

First printed October <strong>2004</strong><br />

This work is copyright. Apart from any use as permitted under the Copyright Act 1968 no part<br />

may be reproduced by any process without written permission from AusInfo. Requests and<br />

inquiries concerning reproduction and rights should be directed to the Manager, Legislative<br />

Services, AusInfo, GPO Box 1920, Canberra, ACT, 2601.<br />

Electronic copies of the report can be obtained from the Medical Service Advisory Committee’s Internet site<br />

at:<br />

http://www.msac.gov.au/<br />

Hard copies of the report can be obtained from:<br />

The Secretary<br />

Medical Services Advisory Committee<br />

Department of Health and Ageing<br />

Mail Drop 107<br />

GPO Box 9848<br />

Canberra ACT 2601<br />

Enquiries about the content of the report should be directed to the above address.<br />

The Medical Services Advisory Committee is an independent committee which has been established to<br />

provide advice to the Commonwealth Minister <strong>for</strong> Health and Ageing on the strength of evidence available<br />

on new and existing medical technologies and procedures in terms of their safety, effectiveness and costeffectiveness.<br />

This advice will help to in<strong>for</strong>m Government decisions about which medical services should<br />

attract funding under Medicare.<br />

MSAC recommendations do not necessarily reflect the views of all individuals who participated in the<br />

MSAC evaluation.<br />

This report was prepared by the Medical Services Advisory Committee with the assistance of Mr Lachlan<br />

Standfield, Dr Suzanne Dyer, Mr Dominic Tilden, Mr Paul Mernagh and Dr Patrick Fitzgerald from<br />

M-TAG Pty Ltd.<br />

Publication approval number: 3533


Contents<br />

Executive summary................................................................................................. ix<br />

Introduction .............................................................................................................1<br />

Background ............................................................................................................ 2<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>............................................................................................. 2<br />

Description of the technology and the procedure............................................ 2<br />

Intended purpose .................................................................................................. 3<br />

Clinical need/burden of disease .................................................................................... 4<br />

Incidence of colorectal cancer............................................................................. 5<br />

Mortality due to colorectal cancer....................................................................... 5<br />

Existing procedures......................................................................................................... 5<br />

Reference standard .......................................................................................................... 6<br />

Comparator....................................................................................................................... 6<br />

Marketing status of the technology............................................................................... 6<br />

Current reimbursement arrangement ........................................................................... 7<br />

Approach to assessment .......................................................................................... 8<br />

Review of literature ......................................................................................................... 8<br />

Search strategy ....................................................................................................... 8<br />

Selection criteria................................................................................................... 10<br />

Quality assessment .............................................................................................. 12<br />

Additional searches ............................................................................................. 12<br />

Expert advice.................................................................................................................. 13<br />

Results of assessment .............................................................................................14<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>........................................................................................... 14<br />

Available evidence............................................................................................... 14<br />

Is it safe?.......................................................................................................................... 16<br />

Adverse events associated with faecal <strong>occult</strong> <strong>blood</strong> tests.............................. 16<br />

Adverse events associated with colonoscopy, sigmoidoscopy and<br />

double contrast barium enema.......................................................................... 16<br />

Psychiatric morbidity associated with colorectal cancer <strong>screening</strong> .............. 17<br />

Safety data reported in the FOBT studies ....................................................... 17<br />

Is it effective? ................................................................................................................. 18<br />

Diagnostic per<strong>for</strong>mance..................................................................................... 18<br />

Participation ......................................................................................................... 22<br />

Change in clinical management and clinical outcomes.................................. 22<br />

Results................................................................................................................... 22<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> iii


What are the economic considerations?..................................................................... 33<br />

Assessment of value <strong>for</strong> money of <strong>population</strong> <strong>health</strong> <strong>screening</strong> <strong>for</strong><br />

colorectal cancer .................................................................................................. 33<br />

Results of the economic model......................................................................... 48<br />

Sensitivity analyses............................................................................................... 54<br />

Discussion ............................................................................................................ 58<br />

Conclusions ...........................................................................................................61<br />

Safety ............................................................................................................................... 61<br />

Effectiveness .................................................................................................................. 61<br />

Cost-effectiveness.......................................................................................................... 62<br />

Summary of outcomes ........................................................................................... 65<br />

Appendix A MSAC terms of reference and membership..................................... 67<br />

Appendix B Advisory panel .................................................................................. 69<br />

Appendix C Studies included in the review...........................................................71<br />

Appendix D Literature searches ........................................................................... 74<br />

Appendix E Excluded references ......................................................................... 77<br />

Appendix F Positive predictive values ................................................................. 96<br />

Appendix G Diagnostic odds ratios.....................................................................102<br />

Appendix H Quality scoring ................................................................................105<br />

Appendix I Participation in FOBT <strong>screening</strong> rates...........................................106<br />

Appendix J Relative cost-effectiveness of Hem<strong>occult</strong> versus Fecatwin<br />

Sensitive/ Feca EIA........................................................................108<br />

Appendix K Sensitivity analysis of Hem<strong>occult</strong> versus HemeSelect ................... 112<br />

Abbreviations ........................................................................................................ 114<br />

References ......................................................................................................... 116<br />

iv <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Tables<br />

Table 1 Hierarchy of evidence ............................................................................................ 12<br />

Table 2 Head-to-head studies of FOBT in <strong>screening</strong> <strong>population</strong>s................................ 14<br />

Table 3 Characteristics of the head-to-head FOBT studies per<strong>for</strong>med in a<br />

<strong>screening</strong> <strong>population</strong> .............................................................................................. 20<br />

Table 4 Meta-analyses of sensitivity <strong>for</strong> carcinoma from FOBT head-to-head<br />

studies....................................................................................................................... 27<br />

Table 5 Meta-analyses of specificity <strong>for</strong> carcinoma from FOBT head-to-head<br />

studies....................................................................................................................... 28<br />

Table 6 Meta-analyses of relative TPR <strong>for</strong> carcinoma from main FOBT headto-head<br />

studies......................................................................................................... 30<br />

Table 7 Meta-analyses of relative FPR <strong>for</strong> carcinoma from main FOBT headto-head<br />

studies......................................................................................................... 31<br />

Table 8 Disease prevalence in the eligible <strong>screening</strong> <strong>population</strong> at program<br />

entry. ......................................................................................................................... 38<br />

Table 9 Distribution of CRC by Dukes’ stage classification in the eligible<br />

<strong>screening</strong> <strong>population</strong> at program entry................................................................ 38<br />

Table 10 Prevalence used <strong>for</strong> individuals entering the economic model........................ 39<br />

Table 11 Incidence of CRC ................................................................................................... 39<br />

Table 12 Incidence of progressive adenomas, age-adjusted ............................................. 40<br />

Table 13 Incidence of all adenomas, age-adjusted ............................................................. 40<br />

Table 14 Mean duration of undiagnosed cancerous and precancerous <strong>health</strong><br />

states prior to progression..................................................................................... 40<br />

Table 15 Distributional spread of known CRC, by stage.................................................. 41<br />

Table 16 Probability of individual with CRC presenting as symptomatic, by<br />

stage .......................................................................................................................... 41<br />

Table 17 Five-year survival rates <strong>for</strong> CRC, by stage .......................................................... 42<br />

Table 18 Quarterly probability of death from diagnosed CRC........................................ 42<br />

Table 19 Probability of positive result using Hem<strong>occult</strong> and HemeSelect..................... 43<br />

Table 20 Probability of positive result using Hem<strong>occult</strong> Sensa and HemeSelect.......... 44<br />

Table 21 Probability of positive result using Hem<strong>occult</strong> and Fecatwin<br />

Sensitive/Feca EIA ................................................................................................ 44<br />

Table 22 Detection rates associated with colonoscopy..................................................... 45<br />

Table 23 Cost of FOBT per invited individual................................................................... 45<br />

Table 24 Cost of FOBT per participant .............................................................................. 46<br />

Table 25 Cost of colonoscopy without polyp removal ..................................................... 46<br />

Table 26 Cost of colonoscopy with polyp removal ........................................................... 46<br />

Table 27 Lifetime CRC treatment costs, from time of diagnosis .................................... 47<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> v


Table 28 Compliance with FOBT <strong>screening</strong> rates used in the economic model .......... 47<br />

Table 29 Life-expectancy from the beginning of the economic model .......................... 49<br />

Table 30 Relative cost-effectiveness of Hem<strong>occult</strong> Sensa over HemeSelect ................. 51<br />

Table 31 Life-expectancy from the beginning of the economic model .......................... 53<br />

Table 32 Relative cost-effectiveness of Hem<strong>occult</strong> over HemeSelect ............................ 54<br />

Table 33 The effect of increasing the frequency of <strong>screening</strong> to annual <strong>testing</strong><br />

on relative cost-effectiveness: Hem<strong>occult</strong> Sensa versus HemeSelect ............. 55<br />

Table 34 The effect of lowering the eligible age to 50 years on relative costeffectiveness:<br />

Hem<strong>occult</strong> Sensa versus HemeSelect.......................................... 56<br />

Table 35 The effect of increasing the eligible age to 80 years on relative costeffectiveness:<br />

Hem<strong>occult</strong> Sensa versus HemeSelect.......................................... 56<br />

Table 36 The effect of increasing the frequency of <strong>screening</strong> to annual <strong>testing</strong><br />

on relative cost-effectiveness: Hem<strong>occult</strong> versus HemeSelect ........................ 57<br />

Table 37 The effect of lowering the eligible age to 50 years on relative costeffectiveness:<br />

Hem<strong>occult</strong> versus HemeSelect..................................................... 58<br />

Table 38 The effect of increasing the eligible age to 80 years on relative costeffectiveness:<br />

Hem<strong>occult</strong> versus HemeSelect..................................................... 58<br />

Table 39 Relevant studies identified..................................................................................... 71<br />

Table 40 FOBT Medline search strategy (1966 to February week 3 <strong>2004</strong>) .................... 74<br />

Table 41 FOBT EMBASE search strategy (1980 to <strong>2004</strong> week 09) ............................... 75<br />

Table 42 Meta-analyses of PPV <strong>for</strong> carcinoma from FOBT head-to-head<br />

studies....................................................................................................................... 98<br />

Table 43 Meta-analyses of PPV <strong>for</strong> neoplasms (carcinoma or adenoma) from<br />

FOBT head-to-head studies................................................................................ 100<br />

Table 44 Meta-analyses of diagnostic odds ratios <strong>for</strong> carcinoma from FOBT<br />

head-to-head studies............................................................................................. 103<br />

Table 45 Quality scoring scale <strong>for</strong> FOBT comparative <strong>screening</strong> studies.................... 105<br />

Table 46 Round 1 participation with FOBT <strong>screening</strong> ................................................... 106<br />

Table 47 Participation data from Jorgensen et al (2002)................................................. 106<br />

Table 48 Later-round participation with FOBT <strong>screening</strong> <strong>for</strong> those who<br />

participated in round 1......................................................................................... 107<br />

Table 49 Proportion of new participants after the first <strong>screening</strong> round ..................... 107<br />

Table 50 Life-expectancy from the beginning of the economic model in the<br />

Hem<strong>occult</strong> versus Fecatwin Sensitive/Feca EIA analysis.............................. 109<br />

Table 51 Incremental cost-effectiveness of Hem<strong>occult</strong> versus Fecatwin<br />

Sensitive/Feca EIA .............................................................................................. 111<br />

Table 52 The effect of selecting alternative sensitivity and specificity values <strong>for</strong><br />

Hem<strong>occult</strong>: Hem<strong>occult</strong> versus HemeSelect...................................................... 112<br />

vi <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Figures<br />

Figure 1 Reasons <strong>for</strong> exclusion of published studies of FOBT identified by the<br />

literature search ....................................................................................................... 11<br />

Figure 2 Simplified natural history of CRC used in the economic model ..................... 36<br />

Figure 3 Screening pathway used in the economic model................................................ 37<br />

Figure 4 Component costs of FOBT <strong>screening</strong> – Hem<strong>occult</strong> Sensa versus<br />

HemeSelect .............................................................................................................. 49<br />

Figure 5 Pattern of neoplasm detection in the Hem<strong>occult</strong> Sensa versus<br />

HemeSelect <strong>screening</strong> analysis.............................................................................. 50<br />

Figure 6 Method of neoplasm detection in the Hem<strong>occult</strong> Sensa versus<br />

HemeSelect analysiss.............................................................................................. 51<br />

Figure 7 Component costs of FOBT <strong>screening</strong> – Hem<strong>occult</strong> versus<br />

HemeSelect .............................................................................................................. 52<br />

Figure 8 Pattern of detection in the Hem<strong>occult</strong> versus HemeSelect <strong>screening</strong><br />

analysis...................................................................................................................... 53<br />

Figure 9 Method of neoplasm detection in the Hem<strong>occult</strong> versus HemeSelect<br />

analysis...................................................................................................................... 54<br />

Figure 10 Summary receiver-operating characteristic curve <strong>for</strong> Hem<strong>occult</strong><br />

versus HemeSelect sensitivity analyses .............................................................. 104<br />

Figure 11 Component costs in the Hem<strong>occult</strong> versus Fecatwin Sensitive/Feca<br />

EIA analysis........................................................................................................... 109<br />

Figure 12 The pattern of detection in the Hem<strong>occult</strong> versus Fecatwin<br />

Sensitive/Feca EIA .............................................................................................. 110<br />

Figure 13 Method of CRC/cancerous adenoma detection in the Hem<strong>occult</strong><br />

versus Fecatwin Sensitive/Feca EIA analysis................................................... 110<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> vii


Executive summary<br />

The procedure<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> tests (FOBTs) are used to detect <strong>blood</strong> in faeces that is not obvious<br />

by general inspection (ie, <strong>occult</strong> <strong>blood</strong>). This technology has been utilised <strong>for</strong> both<br />

diagnostic and <strong>screening</strong> purposes; however, the purpose of this review is to investigate<br />

the relative per<strong>for</strong>mance of different FOBTs <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>. The review<br />

was undertaken to provide support to the Australian Government Bowel Cancer<br />

Screening Pilot Program (the pilot).<br />

There are two types of FOBTs in common use: the long-established guaiac tests and the<br />

newer immunochemical tests. ‘Two-tiered’ combinations of these tests have also been<br />

developed.<br />

Guaiac FOBTs<br />

The guaiac FOBTs act by detecting the intact haem molecule from haemoglobin. These<br />

tests are simple to per<strong>for</strong>m colorimetric tests, often conducted in the home. Two small<br />

samples from stools obtained on three consecutive days are applied to a piece of paper<br />

impregnated with guaiac gum. Upon application of a developing solution, the presence of<br />

trace amounts of haem results in a blue colour change due to the pseudo-peroxidase<br />

actions of haem. Guaiac FOBTs are able to detect bleeding originating anywhere<br />

between the mouth and anus. However, the sensitivity of guaiac FOBTs <strong>for</strong> detecting<br />

upper gastrointestinal (GI) bleeding is less than <strong>for</strong> the lower GI tract. The accuracy of<br />

guaiac FOBTs can be affected by several factors including medications, diet and<br />

excessive amounts of reducing agents in faecal samples (eg, vitamin C).<br />

The guaiac FOBTs include brand names such as: Hem<strong>occult</strong>, Hem<strong>occult</strong> II, Hem<strong>occult</strong><br />

Sensa, Hemo FEC, Coloscreen, Fecatest (discontinued) and Shionogi B. The two<br />

versions of the Hem<strong>occult</strong> test (Hem<strong>occult</strong> and Hem<strong>occult</strong> II) differ only with regard to<br />

their configuration, there<strong>for</strong>e this document will refer to both versions simply as<br />

Hem<strong>occult</strong>, without differentiation.<br />

Immunochemical FOBTs<br />

The immunochemical FOBTs involve the use of an anti-human monoclonal antibody,<br />

targeted at intact human <strong>blood</strong>-borne proteins (usually haemoglobin). These tests<br />

there<strong>for</strong>e have the theoretical advantage of not being affected by haem, peroxidases or<br />

anti-oxidases in the diet. In addition, immunochemical tests will not detect proximal<br />

bleeds, since in this case haemoglobin will be digested be<strong>for</strong>e being passed in the faeces.<br />

However, these tests are generally more complex and expensive than the guaiac tests and<br />

often require laboratory processing.<br />

The immunochemical FOBTs include brand names such as: HemeSelect (discontinued)<br />

and Immudia-HemSp (international equivalents), FlexSure OBT (discontinued),<br />

Quicktest, DIMA FOB-10, Bayer Detect, Magstream HemSp, !nSure, !nForm and<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> ix


BM-Test Colon Albumin. HemeSelect and Immudia-HemSp are reverse passive<br />

haemagglutination (RPHA) tests.<br />

Two-tier tests<br />

The ‘two-tier’ approach has been developed in response to the high positivity rates of<br />

some guaiac FOBTs. By following positive results from a highly sensitive guaiac test with<br />

an immunochemical test, there should be an increase in specificity and a reduction in the<br />

otherwise high rates of unnecessary diagnostic follow-up. Studies using Hem<strong>occult</strong> Sensa<br />

and HemeSelect in such a two-tier approach have been conducted (Allison et al 1996;<br />

Rae and Cleator 1994).<br />

Fecatwin Sensitive/Feca EIA was an example of the combination of a guaiac and<br />

immunochemical test per<strong>for</strong>med in two-tiers. Samples were initially screened with the<br />

Fecatwin Sensitive guaiac test and then positive samples were tested with the Feca EIA<br />

immunochemical test. This test is no longer commercially available.<br />

Medical Services Advisory Committee – role and approach<br />

The Medical Services Advisory Committee (MSAC) is a key element of a measure taken<br />

by the Commonwealth Government to strengthen the role of evidence in <strong>health</strong><br />

financing decisions in Australia. MSAC advises the Commonwealth Minister <strong>for</strong> Health<br />

and Ageing on the evidence relating to the safety, effectiveness and cost-effectiveness of<br />

new and existing medical technologies and procedures, and under what circumstances<br />

public funding should be supported.<br />

A rigorous assessment of the available evidence is thus the basis of decision-making<br />

when funding is sought under Medicare. Medical Technology Assessment Group Pty Ltd<br />

(M-TAG) was contracted to undertake a systematic review and economic evaluation of<br />

FOBTs <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>. An advisory panel with appropriate expertise<br />

then evaluated this evidence and provided advice to MSAC.<br />

MSAC’s assessment of faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> <strong>population</strong><br />

<strong>health</strong> <strong>screening</strong><br />

Clinical need<br />

Colorectal cancer (CRC) is the most common type of cancer in Australia apart from nonmelanocytic<br />

skin cancer, with 12,405 cases being diagnosed in 2000 (Australian Institute<br />

of Health and Welfare (AIHW) and Australasian Association of Cancer Registries<br />

(AACR) 2003). Of these, 6863 patients were men, equating to an age-standardised rate of<br />

80.2/100,000, and the remaining 5542 were women, equating to an age-standardised rate<br />

of 53.8/100,000. Hence, the incidence of CRC in 2000 was second only to prostate<br />

cancer in Australian men and to breast cancer in Australian women (again, excluding<br />

non-melanocytic skin cancer).<br />

Early stage disease usually responds well to surgery. However, the prognosis <strong>for</strong><br />

metastatic disease is poor. Approximately 4718 people died from CRC in Australia in<br />

x <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Safety<br />

2000, making it second only to lung cancer as a cause of cancer deaths nationally in 2000<br />

(AIHW and AACR 2003). Approximately five per cent of Australians develop CRC<br />

during their lifetime; half of these people die from the disease within five years of<br />

diagnosis.<br />

FOBTs are used as a <strong>screening</strong> test <strong>for</strong> the identification of CRC. This is based on the<br />

use of faecal <strong>occult</strong> <strong>blood</strong> as a marker <strong>for</strong> CRC. However, an inherent problem with the<br />

technology is that many other non-neoplastic disease processes can cause GI bleeding.<br />

There<strong>for</strong>e, the sensitivity and specificity of a test <strong>for</strong> the detection of <strong>blood</strong> will not<br />

necessarily translate into equivalent accuracy <strong>for</strong> the detection of CRC in a <strong>population</strong><br />

<strong>health</strong> <strong>screening</strong> setting. Positive FOBT results due to other GI diseases will give a<br />

reduction in the expected specificity of the technique as a CRC <strong>screening</strong> measure.<br />

Immunochemical tests are expected to be less susceptible than guaiac tests to this effect,<br />

since only lower GI bleeding is detected.<br />

FOBTs are non-invasive and there<strong>for</strong>e unlikely to cause adverse events. However, the<br />

use of FOBTs in a <strong>screening</strong> setting is likely to increase the number of colonoscopies,<br />

sigmoidoscopies and barium enemas per<strong>for</strong>med in the screened <strong>population</strong>. There<strong>for</strong>e,<br />

the increased numbers of adverse events associated with these procedures are important.<br />

A 1997 review of six prospective studies estimated that around 1 in 1000 patients suffer<br />

per<strong>for</strong>ation, 3 in 1000 suffer major haemorrhage and between 1 and 3 in 10,000 die as a<br />

result of colonoscopy (Winawer et al 1997). A 1989 review of nine studies of diagnostic<br />

colonoscopy gives a slightly higher figure of around 1.7 in 1000 patients suffering<br />

per<strong>for</strong>ation and similar figures <strong>for</strong> haemorrhage and death (Habr-Gama and Waye 1989).<br />

The risk of per<strong>for</strong>ation and haemorrhage increases with the per<strong>for</strong>mance of<br />

polypectomy. There are also other occasional serious complications associated with<br />

bowel preparation prior to colonoscopy or the use of sedation in conjunction with this<br />

procedure.<br />

Per<strong>for</strong>ation rates associated with sigmoidoscopy appear low and occur in less than 2<br />

cases per 10,000 examinations (National Health and Medical Research Council<br />

(NHMRC) 1999). Serious complications associated with double contrast barium enema<br />

(DCBE) also appear rare and have been estimated at 3 per 10,000 tests, with a death rate<br />

of 3 in 100,000 tests (Winawer et al 1997).<br />

The Minnesota (US) randomised controlled trial (RCT) of FOBT <strong>screening</strong> <strong>for</strong> CRC<br />

reported a per<strong>for</strong>ation rate of 0.03% (4/12,246) and a rate of serious bleeding of 0.09%<br />

(11/12,246) in association with colonoscopy (Mandel et al 1993). The Nottingham (UK)<br />

FOBT trial reported a rate of 0.5% (7/1474) <strong>for</strong> colonoscopy complications and no<br />

complications of DCBE, from a total of 1778 subjects receiving follow-up (Robinson et<br />

al 1999). There were no colonoscopy related deaths reported in either trial.<br />

Concern has also been expressed about the psychological impact of CRC <strong>screening</strong>, and<br />

this is an important issue. However, detailed exploration of this would require a<br />

dedicated systematic review and a cost-effectiveness comparison incorporating qualityof-life<br />

measures, which is beyond the scope of this review.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> xi


The head-to-head studies examining the relative per<strong>for</strong>mance of different FOBTs<br />

identified in this assessment did not report any safety data. There<strong>for</strong>e, the per<strong>for</strong>mance<br />

of alternative FOBTs <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong> cannot be assessed on the basis of<br />

safety data.<br />

Effectiveness<br />

Three head-to-head studies estimated the sensitivity and specificity of various FOBTs<br />

using the interval cancer rate (ie, the number of cancers that are detected during intervals<br />

between <strong>screening</strong>) as a proxy <strong>for</strong> the false negative rate. This allowed calculation of the<br />

relative sensitivities and specificities <strong>for</strong> each of the FOBTs included in these studies.<br />

Slightly more studies reported data that enabled calculation of the relative true positive<br />

rate (TPR) and the relative false positive rate (FPR) (ie, the ratio of one test’s TPR or<br />

FPR to the other) of the FOBTs compared. There were no studies of a suitable quality<br />

available to enable assessment of the relative accuracy of the different FOBTs <strong>for</strong> the<br />

detection of adenomas.<br />

On the basis of two of these studies, HemeSelect was found to be significantly more<br />

sensitive than Hem<strong>occult</strong> in detecting carcinoma (77.1% vs 30.0%, respectively; risk<br />

difference (RD) 0.46; 95% confidence interval (CI): 0.231, 0.631). In contrast, the<br />

Hem<strong>occult</strong> test was significantly more specific than HemeSelect (97.8% vs 93.5%,<br />

respectively; RD –0.043; 95% CI: –0.070, –0.026). This finding was maintained in a<br />

sensitivity analysis incorporating a study conducted in a <strong>population</strong> with a high<br />

proportion of subjects that were at increased risk. Similarly, the TPR was significantly in<br />

favour of RPHA (HemeSelect and Immudia-HemSp pooled) rather than Hem<strong>occult</strong><br />

(relative TPR 0.59; 95% CI: 0.39, 0.88), whereas the FPR of Hem<strong>occult</strong> was lower than<br />

that <strong>for</strong> RPHA (relative FPR 0.53; 95% CI: 0.33, 0.87). There<strong>for</strong>e, HemeSelect was more<br />

effective than Hem<strong>occult</strong> at detecting patients with CRC but was not as effective at<br />

excluding participants without the disease. The practical impact of this trade-off cannot<br />

be determined by the sensitivity and specificity values alone and has been investigated via<br />

the use of an economic model in the cost-effectiveness section of this assessment report.<br />

In a single study, the sensitivity of Hem<strong>occult</strong> and Fecatwin Sensitive/Feca EIA were<br />

not significantly different (50.0% vs 83.3%, respectively; RD 0.33; 95% CI: –0.166,<br />

0.832). However, Hem<strong>occult</strong> was significantly more specific than Fecatwin<br />

Sensitive/Feca EIA (97.1% vs 92.2%, respectively; RD –0.049; 95% CI: –0.066, –0.032).<br />

This means that at this time there is no statistically significant evidence to suggest that<br />

Hem<strong>occult</strong> or Fecatwin Sensitive/Feca EIA are better at identifying patients with CRC.<br />

However, there is statistically significant evidence to suggest that Hem<strong>occult</strong> is superior<br />

at excluding participants without the disease. Similarly, the TPR was not significantly<br />

different when Hem<strong>occult</strong> was compared to Fecatwin Sensitive/Feca EIA (relative TPR<br />

0.60; 95%CI: 0.14, 2.51). However, the FPR of Hem<strong>occult</strong> was significantly lower than<br />

Fecatwin Sensitive/Feca EIA (relative FPR 0.35; 95% CI: 0.24, 0.51). There<strong>for</strong>e, based<br />

on these limited data, Hem<strong>occult</strong> was statistically more accurate than Fecatwin<br />

Sensitive/Feca EIA, when used <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>. It should be noted,<br />

however, that the subjects in this study were not required to adhere to the dietary<br />

restrictions generally recommended with the use of a guaiac FOBT.<br />

The sensitivity of Hem<strong>occult</strong> Sensa and HemeSelect were not significantly different<br />

(79.4% vs 68.8%, respectively; RD –0.107; 95% CI: –0.317, 0.103), based on data from a<br />

single study. In contrast, HemeSelect proved to be significantly more specific than<br />

xii <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Hem<strong>occult</strong> Sensa (94.4% vs 86.7%, respectively; RD 0.077; 95% CI: 0.068, 0.086). These<br />

results were maintained in a sensitivity analysis incorporating a study conducted in a<br />

<strong>population</strong> with a high proportion of subjects that were at increased risk. Similarly, the<br />

TPR was not significantly different when Hem<strong>occult</strong> Sensa was compared with<br />

HemeSelect (relative TPR 1.07; 95%CI: 0.70, 1.62). However, the FPR of HemeSelect<br />

was lower than that of Hem<strong>occult</strong> Sensa (relative FPR 2.23; 95% CI: 1.14, 4.37). This<br />

means that at this time there is no statistically significant evidence to suggest that<br />

Hem<strong>occult</strong> Sensa or HemeSelect are better at identifying patients with CRC. However,<br />

there is statistically significant evidence to suggest that HemeSelect is superior at<br />

excluding participants without the disease. There<strong>for</strong>e, based on these limited data,<br />

HemeSelect was statistically more accurate than Hem<strong>occult</strong> Sensa, when used <strong>for</strong><br />

<strong>population</strong> <strong>health</strong> <strong>screening</strong>.<br />

Newer immunochemical tests are currently commercially available. Other FOBTs with<br />

similar technical characteristics (ie, in vitro diagnostic accuracy <strong>for</strong> the detection of<br />

haemoglobin) may provide similar outcomes to those of HemeSelect. However, there is<br />

currently a lack of evidence <strong>for</strong> this and the per<strong>for</strong>mance of other immunochemical tests<br />

in the context of <strong>population</strong> <strong>health</strong> <strong>screening</strong> <strong>for</strong> CRC remains to be determined.<br />

It is important to note that the measures of the sensitivity and specificity of the FOBTs<br />

<strong>for</strong> the detection of CRC varied considerably between studies conducted in different<br />

<strong>population</strong>s. In addition, estimates of these measures differed significantly between<br />

different tests of the same class. There<strong>for</strong>e, relative findings <strong>for</strong> any individual pairs of<br />

guaiac and immunochemical tests cannot be generalised across comparisons and findings<br />

<strong>for</strong> any individual test do not represent all tests of that class.<br />

A modelled economic evaluation was used to determine the effectiveness and costeffectiveness<br />

of the different tests in a <strong>population</strong> <strong>screening</strong> setting. This model was built<br />

upon the sensitivity and specificity data summarised above and was able to determine the<br />

broader <strong>population</strong>-wide impact of the sensitivity and specificity of these tests.<br />

Cost-effectiveness<br />

An economic model was designed to assess the relative cost-effectiveness of various<br />

FOBTs used to detect CRC in a <strong>population</strong> <strong>health</strong> <strong>screening</strong> setting. Consequently, headto-head<br />

comparisons between FOBTs are made. It is not intended that decision-makers<br />

compare tests not compared head-to-head within the economic model. The base-case<br />

scenario assumes biennial <strong>screening</strong> in individuals aged 55–74 years.<br />

The economic model indicates that FOBTs with greater sensitivity <strong>for</strong> colorectal<br />

neoplasia detection will offer better overall survival outcomes. A key driver of the<br />

difference in the total cost associated with the FOBTs is the specificity. Tests with lower<br />

specificities are associated with higher diagnostic follow-up costs due to the increased<br />

levels of resource wastage. The model attempts to present the trade-off between these<br />

values in an economic context by providing estimates of the incremental cost per lifeyear<br />

gained <strong>for</strong> each pair of tests compared.<br />

The incremental cost per life-year gained of HemeSelect was $3172 in a comparison<br />

against Hem<strong>occult</strong>, and $21,533 <strong>for</strong> Hem<strong>occult</strong> Sensa in a comparison against<br />

HemeSelect. These findings should, however, be constrained to the context of the headto-head<br />

data upon which they are based. That is, the magnitude of the difference in the<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> xiii


sensitivity between tests contains a level of uncertainty. This is particularly due to the low<br />

prevalence of CRC in the <strong>population</strong>s tested in the FOBT studies and the scarcity and<br />

poor quality of data providing estimates of the sensitivity <strong>for</strong> adenoma detection.<br />

Consequently, use of these data in the economic model may render the results of the<br />

economic model equally uncertain.<br />

However, the absolute estimates of the difference in specificity between tests are more<br />

reliable than the estimates of the sensitivity. Similarly, the results concerning costs of the<br />

FOBTs, inclusive of diagnostic follow-up and treatment, are more reliable than the<br />

modelled life-expectancy, due to the quality of the estimates that they are based upon.<br />

There<strong>for</strong>e, the tests associated with the higher specificity were shown to be less costly<br />

overall, with the impact of sensitivity upon effectiveness of uncertain magnitude. This<br />

implies that the most promising tests in an economic sense may be those associated with<br />

a high specificity.<br />

The results of these analyses <strong>for</strong> the immunochemical tests used may reflect those <strong>for</strong><br />

FOBTs with similar technical characteristics (ie, in vitro diagnostic accuracy <strong>for</strong> the<br />

detection of haemoglobin). However, there is currently a lack of suitable comparative<br />

evidence <strong>for</strong> the per<strong>for</strong>mance of other immunochemical FOBTs in the context of<br />

<strong>population</strong> <strong>health</strong> <strong>screening</strong> <strong>for</strong> CRC.<br />

Other important findings of the economic model include the following.<br />

• There was no apparent class effect in determining the relative cost-effectiveness of<br />

different FOBTs.<br />

• Increasing participation will lead to increased cancer detection and an increase in<br />

the effectiveness of any <strong>screening</strong> program. Participation will not, however, have a<br />

major impact upon the relative cost-effectiveness between different FOBTs as a<br />

change in participation will shift both costs and effectiveness in the same direction,<br />

thus having a minimal effect on the incremental cost per life-year gained.<br />

• Altering key variables endogenous to the <strong>screening</strong> program generally yielded little<br />

difference in the relative cost-effectiveness between FOBTs. However, these<br />

sensitivity analyses did demonstrate consistent trends which provide insight into<br />

methods that may maximise the cost-effectiveness of <strong>screening</strong> in comparison with<br />

no-<strong>screening</strong>.<br />

• Increasing the <strong>screening</strong> frequency from biennial to annual increases both costs<br />

and effectiveness. Within the context of the main analyses, it appears that annual<br />

FOBT <strong>screening</strong> is not cost-effective compared with biennial <strong>screening</strong>.<br />

• Changing the minimum eligible <strong>screening</strong> age from 55 to 50 years appears to offer<br />

benefits in terms of cost-effectiveness.<br />

• Increasing the maximum <strong>screening</strong> age from 75 to 80 years does not appear to<br />

offer the same degree of benefit.<br />

It is important to note that the reliability of the above findings is dependent upon the<br />

reliability of the clinical data available to make head-to-head comparisons between the<br />

xiv <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


FOBTs. An improvement in the quantity, quality and external validity of head-to-head<br />

study data is required to validate the findings of the economic model.<br />

Summary of outcomes<br />

The MSAC considers that faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> is useful <strong>for</strong> <strong>population</strong> <strong>health</strong><br />

<strong>screening</strong> to reduce CRC mortality.<br />

The available evidence indicated that there was no apparent class effect of the guaiac<br />

versus immunochemical FOBTs with regard to their effectiveness or cost-effectiveness.<br />

Different brands of FOBTs possess different sensitivities and specificities <strong>for</strong> the<br />

detection of CRC within an average risk <strong>screening</strong> <strong>population</strong> setting. The specificity of<br />

the FOBTs was a major determinant of the total associated costs of FOBT <strong>screening</strong>,<br />

inclusive of diagnostic follow-up and treatment.<br />

An economic model indicated that biennial <strong>screening</strong> was more cost-effective than<br />

annual <strong>screening</strong>, within the context of the main analysis. Lowering the minimum eligible<br />

<strong>screening</strong> age from 55 to 50 years offered benefits in terms of cost-effectiveness.<br />

Increasing the maximum <strong>screening</strong> age from 75 to 80 years did not offer the same degree<br />

of benefit.<br />

The immunochemical tests included in the assessment are no longer available and they<br />

have been replaced by newer assays. There was no available evidence suitable <strong>for</strong> the<br />

assessment of the comparative per<strong>for</strong>mance of currently available immunochemical tests<br />

within an average risk <strong>population</strong> <strong>health</strong> <strong>screening</strong> setting. However, the results of the<br />

analyses of the immunochemical tests used may reflect those <strong>for</strong> FOBTs with similar<br />

technical characteristics (ie, in vitro diagnostic accuracy <strong>for</strong> the detection of haemoglobin).<br />

There<strong>for</strong>e, it is suggested that currently available and new immunochemical tests with<br />

promising technical characteristics be evaluated against established FOBTs within the<br />

context of an ongoing <strong>screening</strong> program.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> xv


Introduction<br />

The Medical Services Advisory Committee (MSAC) has undertaken a review to provide<br />

advice on the use of different commercially available faecal <strong>occult</strong> <strong>blood</strong> tests (FOBTs)<br />

<strong>for</strong> <strong>population</strong> <strong>screening</strong> in Australia.<br />

MSAC evaluates new and existing <strong>health</strong> technologies and procedures <strong>for</strong> which funding is<br />

sought under the Medicare Benefits Scheme in terms of their safety, effectiveness and costeffectiveness,<br />

while taking into account other issues such as access and equity. MSAC<br />

adopts an evidence-based approach to its assessments, based on reviews of the scientific<br />

literature and other in<strong>for</strong>mation sources, including clinical expertise.<br />

MSAC’s terms of reference and membership are at Appendix A. MSAC is a<br />

multidisciplinary expert body, comprising members drawn from such disciplines as<br />

diagnostic imaging, pathology, surgery, internal medicine and general practice, clinical<br />

epidemiology, <strong>health</strong> economics, consumer <strong>health</strong> and <strong>health</strong> administration.<br />

This review was undertaken to provide support to the Australian Government Bowel<br />

Cancer Screening Pilot Program (the pilot). The primary aim of the pilot is to provide<br />

in<strong>for</strong>mation about the feasibility, acceptability and cost-effectiveness of FOBT <strong>screening</strong><br />

amongst the Australian <strong>population</strong> in both rural and urban areas. The results of the pilot<br />

will in<strong>for</strong>m decisions about whether, and how, to introduce a national bowel cancer<br />

<strong>screening</strong> program.<br />

This report summarises the assessment of the relative per<strong>for</strong>mance and costeffectiveness<br />

of commercially available FOBTs <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>. This<br />

report also provides suggestions as to the most promising FOBTs available at this time<br />

and the most cost-effective setting <strong>for</strong> FOBT <strong>screening</strong>.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 1


Background<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

Description of the technology and the procedure<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> tests (FOBTs) are used to detect <strong>blood</strong> in faeces that is not obvious<br />

by general inspection (ie, <strong>occult</strong> <strong>blood</strong>). This technology has been utilised <strong>for</strong> both<br />

diagnostic and <strong>screening</strong> purposes; however, the purpose of this review is to investigate<br />

the relative per<strong>for</strong>mance of different FOBTs <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>.<br />

In general, FOBTs involve sample preparation prior to colorimetric visualisation. This<br />

may be conducted in the home, clinician’s office or a pathology laboratory. The most<br />

accepted regimen <strong>for</strong> faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> involves <strong>testing</strong> two faecal samples from<br />

each of three stools over three consecutive days. There are several alternative sampling<br />

methods associated with faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>. These include utilising either a<br />

wooden spatula or a brush, methods which involve a different degree of stool handling.<br />

The samples are then usually transferred onto a sample card which can be delivered by<br />

post to the appropriate location <strong>for</strong> development and interpretation. A single positive<br />

sample in the series is counted as a positive result. A high roughage diet is sometimes<br />

recommended prior to conducting the tests, in order to uncover lesions that bleed<br />

intermittently. FOBTs should not be conducted whilst the subject is experiencing<br />

menstrual or peri-anal bleeding.<br />

There are two types of FOBTs in common use: the long established guaiac tests and the<br />

newer immunochemical tests.<br />

Guaiac FOBTs<br />

The guaiac FOBTs act by detecting the intact haem molecule from haemoglobin. These<br />

tests are simple-to-per<strong>for</strong>m colorimetric tests, often conducted in the home. Two small<br />

samples from stools obtained on three consecutive days are applied to a piece of paper<br />

impregnated with the guaiac gum. Upon application of a developing solution, the<br />

presence of trace amounts of haem results in a blue colour change due to the pseudoperoxidase<br />

actions of haem. Guaiac FOBTs are able to detect bleeding originating<br />

anywhere between the mouth and anus. However, the sensitivity of guaiac FOBTs <strong>for</strong><br />

detecting upper gastrointestinal (GI) bleeding is less than <strong>for</strong> the lower GI tract.<br />

The accuracy of guaiac FOBTs can be affected by medications (eg, non steroidal antiinflammatory<br />

drugs (NSAIDs), anticoagulants, iron supplements), contamination of the<br />

sample with toilet water, time delay between sample collection and developing, and<br />

dietary factors (eg, haem in meat or peroxidases or catalases in many fresh fruits or<br />

vegetables or rare red meat). Since the tests are dependent upon the pseudo-peroxidase<br />

activity of haem, false negative tests will arise from excessive amounts of reducing agents<br />

in faecal samples (eg, vitamin C). Hydration of faecal samples is also known to increase<br />

test sensitivity with a loss of specificity. Delaying the development of the test <strong>for</strong> a few<br />

days after the collection of the sample minimises the effects of dietary peroxidases on<br />

test results; however, this may also diminish the chance of detecting a weak positive<br />

result.<br />

2 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


The guaiac FOBTs include brand names such as: Hem<strong>occult</strong>, Hem<strong>occult</strong> Sensa, Hemo<br />

FEC, Coloscreen, Fecatest (discontinued) and Shionogi B.<br />

Immunochemical FOBTs<br />

The immunochemical FOBTs involve the use of an anti-human monoclonal antibody,<br />

targeted at intact human <strong>blood</strong>-borne proteins (usually haemoglobin). These tests<br />

there<strong>for</strong>e have the theoretical advantage of not being affected by haem, peroxidases or<br />

anti-oxidases in the diet. In addition, immunochemical tests will not detect proximal<br />

bleeds, since in this case haemoglobin is digested be<strong>for</strong>e being passed in the faeces.<br />

However, these tests are generally more complex and expensive than the guaiac tests and<br />

often require laboratory processing.<br />

The immunochemical FOBTs include brand names such as: HemeSelect (discontinued)<br />

and Immudia-HemSp (international equivalents) (Young 1998), FlexSure OBT<br />

(discontinued), Magstream HemSp, Bayer Detect, Quicktest, DIMA FOB-10, !nSure,<br />

!nForm and BM-Test Colon Albumin. HemeSelect and Immudia-HemSp are reverse<br />

passive haemagglutination (RPHA) tests. These RPHA tests use chicken-erythrocytes<br />

coated with antibodies to human haemoglobin. The coated chicken-erythrocytes are<br />

agglutinated in the presence of haemoglobin in human faecal samples. HemeSelect,<br />

Immudia-HemSp, !nSure and FlexSure OBT utilise an anti-human monoclonal antibody<br />

to haemoglobin. BM-Test Colon Albumin utilises an antibody specific <strong>for</strong> human<br />

albumin.<br />

Two-tier tests<br />

There have been suggestions of using a “two-tier” approach to faecal <strong>occult</strong> <strong>blood</strong><br />

<strong>testing</strong>, where a guaiac and then an immunochemical test are used in sequence <strong>for</strong><br />

colorectal cancer (CRC) <strong>screening</strong>. This approach has been developed in response to the<br />

high positivity rates of some guaiac FOBTs. Following positive results from a highly<br />

sensitive guaiac test with an immunochemical test should increase specificity and reduce<br />

the associated high rates of unnecessary diagnostic follow-up. Studies using Hem<strong>occult</strong><br />

Sensa and HemeSelect in such a two-tier approach have been conducted (Allison et al<br />

1996; Rae and Cleator 1994).<br />

Fecatwin Sensitive/Feca EIA was a combination of a guaiac and immunochemical test<br />

per<strong>for</strong>med in two-tiers. Samples were initially screened with the Fecatwin Sensitive guaiac<br />

test and then positive samples were tested with the Feca EIA immunochemical test. This<br />

test is no longer commercially available.<br />

Intended purpose<br />

For diagnostic purposes, the use of both a guaiac and an immunochemical test together<br />

has been suggested, to differentiate between upper and lower GI tract bleeding.<br />

However, the purpose of this review is to investigate the relative effectiveness of<br />

different commercially available FOBTs when used <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>. The<br />

diagnostic use of FOBTs would need to be examined in a separate review.<br />

FOBTs are used as a <strong>screening</strong> test <strong>for</strong> the identification of CRC. This is based on the<br />

use of faecal <strong>occult</strong> <strong>blood</strong> as a marker <strong>for</strong> CRC. However, an inherent problem with the<br />

technology is that many other non-neoplastic disease processes can cause GI bleeding.<br />

There<strong>for</strong>e, the sensitivity and specificity of a test <strong>for</strong> the detection of <strong>blood</strong> will not<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 3


necessarily translate into equivalent accuracy <strong>for</strong> the detection of CRC in a <strong>population</strong><br />

<strong>health</strong> <strong>screening</strong> setting. Positive FOBT results due to other GI diseases will give a<br />

reduction in the expected specificity of the technique as a CRC <strong>screening</strong> measure.<br />

Immunochemical tests are expected to be less susceptible than guaiac tests to this effect,<br />

since only lower GI bleeding will be detected. There<strong>for</strong>e, the key issue to be addressed in<br />

this review is the relative per<strong>for</strong>mance of guaiac versus immunochemical FOBTs when<br />

used <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>.<br />

GI bleeding, particularly from the upper GI tract, may be caused or exacerbated by the<br />

intake of medications such as NSAIDs and anticoagulants. Pathophysiological causes of<br />

GI bleeding are diverse in nature and may include: ulcers and erosions; haemorrhoids;<br />

fissures (perianal disease); colitis; diverticulitis; vascular ectasia/angiodysplasia; radiation<br />

mucosal injury; previous gastric surgery; inflammation; polyps; and cancer of the colon.<br />

The most common sources of <strong>occult</strong> bleeding detected by FOBTs are colonic adenomas<br />

≥ 1 cm (12−14%), peptic ulcer disease (7−10%), oesophagitis (6−9%), colon carcinoma<br />

(5−6%) and angiodysplasia (3−13%) (Zuckerman et al 2000). Fifteen to 20 per cent of<br />

peptic ulcer patients experience bleeding, particularly in association with the use of<br />

NSAIDs and Helicobacter pylori infection (Kurata et al 1982; Hawkey 2000). In a series of<br />

reflux oesophagitis patients, 8.2 per cent experienced bleeding, primarily in those with<br />

severe disease (Costa et al 2001). In elderly patients, a major source of lower GI bleeding<br />

is angiodysplasia; however, rare types of congenital or hereditary vascular ecstasia also<br />

occurs in younger patients (Sharma and Gorbien 1995). Haemorrhoids are not readily<br />

detected by FOBTs and it was suggested that this was due to bleeding being intermittent<br />

or not adherent to the particular region of sample used <strong>for</strong> <strong>testing</strong> (Nakama et al 1997).<br />

Studies have estimated that 50–60 per cent of CRCs are likely to bleed. Overall, 11 per<br />

cent of adenomatous polyps have a propensity to bleed with larger adenomas (≥ 1.5 cm)<br />

the most likely to bleed (Foutch et al 1998; Simon 1985). There<strong>for</strong>e, the use of FOBTs as<br />

a <strong>screening</strong> measure to detect early colorectal neoplasia has been widely investigated.<br />

A meta-analysis of mortality results from randomised controlled trials (RCTs) showed<br />

that people allocated to FOBT <strong>screening</strong> (Hem<strong>occult</strong>, a guaiac test) had a relative risk<br />

reduction in CRC mortality of 16 per cent (relative risk 0.84; 95% confidence interval<br />

(CI): 0.77, 0.89) (Towler et al 1998). A comprehensive discussion of many of the issues<br />

surrounding the use of FOBT as a CRC <strong>population</strong> <strong>health</strong> <strong>screening</strong> test has recently<br />

been published by Allison (2003).<br />

Clinical need/burden of disease<br />

Colorectal polyps may be hyperplastic or adenomatous. Hyperplastic polyps do not<br />

develop into carcinomas. However, a proportion of adenomatous polyps (adenomas) will<br />

become malignant. Approximately 10 per cent of adenomas greater than 1 cm in size will<br />

become cancerous within 10 years, and a lesser proportion of smaller adenomas (Stryker<br />

et al 1987). The majority of CRC develops from adenomas. There<strong>for</strong>e, detection and<br />

resection of neoplasms at this precancerous stage would effectively reduce the incidence<br />

of CRC (Winawer et al 1993).<br />

4 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Classification of CRC may be by one of several staging systems. The first well described<br />

and widely used staging system was Dukes’ staging of the extent of tumour spread.<br />

Dukes’ stages A–D correspond to a localised tumour, a localised tumour which has<br />

spread beyond the muscularis propria, a localised tumour associated with lymph node<br />

involvement and a tumour associated with metastases, respectively. The pathological<br />

staging system (pTNM) designates four stages I–IV, which differs from Dukes’ staging<br />

only in regard to the fact that classification as stage IV is not dependent upon the<br />

presence of tumour in a line of resection.<br />

Incidence of colorectal cancer<br />

CRC is the most common type of cancer in Australia, aside from non-melanocytic skin<br />

cancer, with 12,405 cases being diagnosed in 2000 (Australian Institute of Health and<br />

Welfare (AIHW) and Australasian Association of Cancer Registries (AACR) 2003). Of<br />

these, 6863 patients were men, equating to an age-standardised rate of 80.2/100,000, and<br />

the remaining 5542 were women, equating to an age-standardised rate of 53.8/100,000.<br />

Hence, the incidence of CRC in 2000 was second only to prostate cancer in Australian<br />

men and to breast cancer in Australian women (again, excluding non-melanocytic skin<br />

cancer).<br />

Mortality due to colorectal cancer<br />

Early stage disease usually responds well to surgery. However, the prognosis <strong>for</strong><br />

metastatic disease is poor. Approximately 4718 people died from CRC in Australia in<br />

2000, making it second only to lung cancer as a cause of cancer deaths nationally in 2000<br />

(AIHW and AACR 2003). Approximately 5 per cent of Australians develop CRC during<br />

their lifetime and half of these people die from the disease within five years of diagnosis.<br />

Existing procedures<br />

Many procedures are available to detect malignancies in the colon and rectum. These<br />

include: colonoscopy, sigmoidoscopy, double contrast barium enema (DCBE) and more<br />

recently, virtual colonoscopy (thin-section helical computerised tomography). The ‘goldstandard’<br />

diagnostic procedure <strong>for</strong> the investigation of CRC is colonoscopy, which<br />

enables examination of the full length of the colon by endoscopy under sedation.<br />

Sigmoidoscopy is used <strong>for</strong> examination of the rectum and sigmoid (descending) colon,<br />

but not the ascending or transverse colon. Flexible sigmoidoscopy allows examination of<br />

a greater extent of the bowel than rigid sigmoidoscopy and is superior in terms of patient<br />

com<strong>for</strong>t. DCBE does not visualise the rectum and rectosigmoid region well and is more<br />

likely to miss a Dukes’ A cancer than colonoscopy (National Health and Medical<br />

Research Council (NHMRC) 1999). Virtual colonoscopy uses computed tomography<br />

scanning to obtain two-dimensional images, which are reconstructed into threedimensional<br />

images with the use of advanced software. The resulting images simulate the<br />

endoluminal view of the colorectum seen at colonoscopy without the need <strong>for</strong> an<br />

invasive procedure.<br />

In general, these diagnostic modalities are more accurate than the FOBTs. However,<br />

these procedures are less practical options <strong>for</strong> use in a <strong>population</strong>-<strong>screening</strong> program due<br />

to their high level of invasiveness, their acceptability to the general public, and their cost.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 5


Reference standard<br />

It is mandatory that any positive FOBT be investigated by an appropriate diagnostic<br />

procedure. The ‘gold-standard’ diagnostic procedure and the procedure of choice <strong>for</strong> this<br />

investigation is colonoscopy. If the colonoscopy is ‘incomplete’, then a DCBE is<br />

generally used to ensure complete visualisation of the colon (NHMRC 1999). A DCBE<br />

plus flexible sigmoidoscopy may replace the colonoscopy if there are difficulties with<br />

local availability or expertise, or if the patient prefers (NHMRC 1999).<br />

Comparator<br />

Since this report focuses on the relative effectiveness of different commercially available<br />

FOBTs and not the absolute effectiveness of these tests, the comparators identified <strong>for</strong><br />

this evaluation are the other FOBTs that have been or are currently commercially<br />

available.<br />

Marketing status of the technology<br />

The guaiac FOBTs include brand names such as: Hem<strong>occult</strong> (Beckman Coulter Inc,<br />

USA), Hem<strong>occult</strong> Sensa (Beckman Coulter Inc, USA), Hemo FEC (Roche Diagnostics,<br />

Switzerland), Coloscreen (Helena Laboratories, USA), Fecatest (Nordic Products,<br />

Scandinavia) and Shionogi B (Shionogi Pharmaceutical Co, Japan). SmithKline and<br />

French Laboratories first produced the Hem<strong>occult</strong> brand in 1970, and the currently<br />

marketed version of this product is Hem<strong>occult</strong> II, produced by Beckman Coulter Inc.<br />

The two versions of this test differ only with regard to their configuration (Mandel et al<br />

1989), there<strong>for</strong>e this document will refer to both versions simply as Hem<strong>occult</strong>, without<br />

differentiation. Hem<strong>occult</strong> Sensa was produced in 1982 by the same company as an<br />

enhanced Hem<strong>occult</strong> product and is currently marketed as Hem<strong>occult</strong> II Sensa.<br />

Beckman Coulter Inc. has also recently marketed the Hem<strong>occult</strong> II Sensa elite FOBT.<br />

Fecatest does not appear to be currently commercially available.<br />

The immunochemical FOBTs include brand names such as: HemeSelect, Immudia-<br />

HemSp, Bayer Detect and Magstream HemSp (Fujirebio Inc, Japan); FlexSure OBT<br />

(Beckman Coulter Inc., USA); Quicktest (Laboratory Diagnostics, Australia); DIMA<br />

FOB-10 (Asquith Laboratories, Australia); !nSure (Enterix Inc., USA); !nForm (Enterix<br />

P/L, Australia) and BM-Test Colon Albumin (Boehringer Mannheim, Germany).<br />

Immudia-Hem Sp was originally developed and marketed in Japan. HemeSelect<br />

(discontinued) is the international equivalent of Immudia-Hem Sp (Young 1998).<br />

FlexSure OBT was previously available commercially; however, sales have been<br />

discontinued. BM-Test Colon Albumin is now produced by Roche Diagnostics.<br />

The two-tiered FOBT, Fecatwin Sensitive/Feca EIA (Nordic Products) does not appear<br />

to be currently commercially available.<br />

6 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Current reimbursement arrangement<br />

FOBTs are currently reimbursed on the Medicare Benefits Scheme (MBS).<br />

The current wording of the listing of FOBTs on the MBS <strong>for</strong> item number 66764 is:<br />

“Examination <strong>for</strong> faecal <strong>occult</strong> <strong>blood</strong> (including tests <strong>for</strong> haemoglobin and its<br />

derivatives in the faeces) by: (a) an immunological method; and (b) a chemical<br />

method (except reagent strip or dip stick); with a maximum of 3 examinations<br />

on specimens collected on separate days in a 28-day period - 1 examination by<br />

both methods.”<br />

Item numbers 66767 and 66770 cover two and three examinations (respectively)<br />

by both methods described in item 66764, per<strong>for</strong>med on separately collected and<br />

identified specimens.<br />

The current wording <strong>for</strong> item number 73809 is:<br />

“Chemical tests <strong>for</strong> <strong>occult</strong> <strong>blood</strong> in faeces by reagent stick, strip, tablet or similar<br />

method.”<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 7


Approach to assessment<br />

It has been argued that determining person-centred outcomes (eg, mortality) is the<br />

ultimate purpose of per<strong>for</strong>ming a test, and that the strongest evidence should come from<br />

randomised controlled trials (RCTs). There<strong>for</strong>e, the important effects of the range of<br />

available faecal <strong>occult</strong> <strong>blood</strong> tests (FOBTs) should ideally be measured in a series of<br />

person-centred outcomes in RCTs. Level I evidence (ie, a systematic review of all<br />

relevant RCTs) shows that people allocated to FOBT <strong>screening</strong> (Hem<strong>occult</strong>, a guaiac<br />

test) have a relative risk reduction in colorectal cancer (CRC) mortality of 16 per cent<br />

(relative risk 0.84; 95% confidence interval (CI): 0.77, 0.89) (Towler et al 1998). However,<br />

it is unlikely that every new FOBT will be included in an RCT, or that patient-centred<br />

outcomes will be collected <strong>for</strong> each test.<br />

There<strong>for</strong>e, this assessment report focuses on the review of head-to-head studies of<br />

FOBTs in ‘average risk’ general <strong>population</strong> subjects. These studies usually report<br />

outcomes such as the number of patients detected with CRC or adenoma(s) rather than<br />

CRC mortality. These studies are likely to provide the most robust estimate of the<br />

relative per<strong>for</strong>mance of commercially available FOBTs <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong><br />

(Cochrane Methods Group on Systematic Review of Screening and Diagnostic Tests<br />

1996).<br />

Review of literature<br />

Search strategy<br />

The medical literature was searched to identify relevant studies and reviews <strong>for</strong> the<br />

period to February <strong>2004</strong>. Searches were conducted via the following primary databases:<br />

• Medline 1966 to current<br />

• Embase 1980 to current<br />

• Econlit 1969 to current.<br />

The search terms used included the following:<br />

• <strong>occult</strong> <strong>blood</strong>; faecal <strong>blood</strong>; faecal haemoglobin; faecal globin; faecal haem<br />

• fobt; heme<strong>occult</strong>; haem<strong>occult</strong>; fecatwin; colocare; okokit; hemofec; flexsure;<br />

hemeselect; feca-eia; iatrohemcheck; imdiahem; hemochaser; monohaem; hemodia;<br />

hemoglobin; annual bowel check; haptoglobin; guaiac; immunochemical test; elisa;<br />

in<strong>for</strong>m<br />

• mass <strong>screening</strong><br />

• colorectal neoplasms.<br />

Complete details of the literature searches per<strong>for</strong>med using the Medline and Embase<br />

databases are presented in Appendix D.<br />

8 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Searches of the following secondary databases/sites were also per<strong>for</strong>med:<br />

• British Columbia Office of Health Technology Assessment<br />

• Canadian Coordinating Office <strong>for</strong> Health Technology Assessment (CCOHTA)<br />

• Centre <strong>for</strong> Health Program Evaluation (Monash University, Australia)<br />

• Cochrane Library database<br />

• Danish Centre <strong>for</strong> Evaluation and Health Technology Assessment (DACEHTA)<br />

• Health Economics Research Group (Brunel University, UK)<br />

• Health In<strong>for</strong>mation Research Unit (HIRU) internal database (McMaster University,<br />

Canada)<br />

• International Standard Randomised Controlled Trial Number (ISRTCN) register<br />

(BioMed Central)<br />

• National Cancer Control Initiative (NCCI) publication list<br />

• National Health and Medical Research Council (NHMRC) Australia publication list<br />

• National Health Service Centre <strong>for</strong> Reviews and Dissemination databases (NHS<br />

CRD), including the Database of Abstracts of Reviews and Effects (DARE), the<br />

National Health Service Economic Evaluation Database (NHSEED) and the Health<br />

Technology Assessment (HTA) database (University of York, UK)<br />

• National In<strong>for</strong>mation Center on Health Services Research and Health Care<br />

Technology, the Health Services Technology Assessment Texts (HSTAT) database<br />

(US)<br />

• Swedish Council on Technology Assessment in Health Care (SBU)<br />

• US Preventative Services Task Force (USPSTF).<br />

Other in<strong>for</strong>mation sources used to gain relevant data included:<br />

• market assessments and economic evaluation reports<br />

• Therapeutic Goods Administration (TGA)<br />

• content experts.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 9


Selection criteria<br />

The ideal study design <strong>for</strong> a study of the accuracy of diagnostic tests is that in which each<br />

test being compared is per<strong>for</strong>med in all individuals. In addition, <strong>for</strong> the generalisability<br />

and external validity of studies being assessed to apply to the Australian <strong>population</strong>,<br />

studies must be per<strong>for</strong>med in a <strong>screening</strong> <strong>population</strong>, rather than a hospital-based highrisk<br />

<strong>population</strong>. For this reason the inclusion and exclusion criteria were as follows.<br />

Inclusion criteria<br />

• All human head-to-head studies comparing at least two FOBTs that have been, or<br />

are currently, commercially available.<br />

• Population <strong>health</strong> <strong>screening</strong> studies.<br />

• Use of an appropriate reference standard (eg, colonoscopy and/or double contrast<br />

barium enema (DCBE) where colonoscopy is incomplete or contra-indicated).<br />

Exclusion criteria<br />

• Trials with fewer than 100 patients.<br />

• Non-systematic reviews and opinion pieces.<br />

• Non-comparative studies.<br />

The flow chart in Figure 1 summarises the exclusion of studies from the safety and<br />

effectiveness review of FOBTs. A total of 1214 references were identified by the search,<br />

of which 30 met the criteria to be considered as evidence in the effectiveness review. A<br />

complete list of the citations identified in the literature search, identified as FOBT studies<br />

and later excluded at higher levels is included in Appendix E, together with reasons <strong>for</strong><br />

exclusion from the review.<br />

Publications that duplicated all or some of the patient data were included in the first<br />

instance. They were then reviewed, and excluded if necessary. Publications that failed to<br />

report outcomes adequately (eg, sensitivity, specificity, true positive rates, false positive<br />

rates, positive predictive value) were also excluded after review.<br />

10 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Identified by the search<br />

(n = 1214)<br />

Original studies<br />

(n = 726)<br />

Original in vivo human studies<br />

(n = 714)<br />

Original in vivo human studies of<br />

FOBT<br />

(n = 229)<br />

Original in vivo human studies of<br />

FOBT in <strong>population</strong> <strong>health</strong><br />

<strong>screening</strong><br />

(n = 177)<br />

Original in vivo human studies of<br />

FOBT in <strong>population</strong> <strong>health</strong><br />

<strong>screening</strong> with appropriate<br />

reported outcomes<br />

(n = 123)<br />

Original in vivo human studies of<br />

FOBT in <strong>population</strong> <strong>health</strong><br />

<strong>screening</strong> in ≥100 individuals<br />

(n = 123)<br />

Original in vivo human<br />

head-to-head studies of FOBT<br />

in <strong>population</strong> <strong>health</strong> <strong>screening</strong><br />

in ≥ 100 individuals<br />

(n = 30)<br />

Figure 1 Reasons <strong>for</strong> exclusion of published studies of FOBT identified by the literature search<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

Excluded if a nonsystematic<br />

review, editorial,<br />

letter, news article, note,<br />

survey, opinion piece or<br />

economic analysis<br />

(n = 488)<br />

Excluded if non-human or<br />

in vitro study<br />

(n =12)<br />

Excluded if the study was<br />

not of FOBT<br />

(n = 485)<br />

Excluded if the study was<br />

not a <strong>population</strong> <strong>health</strong><br />

<strong>screening</strong> study<br />

(n = 52)<br />

Excluded if inappropriate<br />

FOBT usage or outcomes<br />

(n = 54)<br />

Excluded if < 100 screened<br />

individuals<br />

(n = 0)<br />

Excluded if not a<br />

head-to-head study of<br />

two different FOBTs<br />

(n = 93)<br />

Available<br />

evidence initially<br />

included in the<br />

the review of<br />

comparative<br />

FOBT<br />

effectiveness<br />

(n = 30)<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 11


Quality assessment<br />

The evidence presented in the selected studies was assessed and classified using the<br />

hierarchy of evidence <strong>for</strong> <strong>screening</strong> tests defined by the New Zealand Health Technology<br />

Assessment group (NZHTA) (Broadstock 2000). The designations of the hierarchy of<br />

evidence are shown in Table 1.<br />

Table 1 Hierarchy of evidence<br />

Level of<br />

evidence<br />

1 All tests done on each person (within-subjects design)<br />

Study design<br />

2a Different tests done on randomly allocated individuals (between subjects design, randomised controlled trial)<br />

2b Different tests done on randomly allocated groups (between-subjects design)<br />

3a Different tests done on different individuals, not randomly allocated, and recruited concurrently (between<br />

subjects design)<br />

3b Different tests done on different individuals, not randomly allocated, with historical cohort (between subjects<br />

design)<br />

Source: Broadstock (2000).<br />

A more detailed assessment of study quality was undertaken using a modification of the<br />

diagnostic-specific checklist published by the Cochrane Screening and Diagnostic Tests<br />

Methods group (Cochrane Methods Group on Systematic review of Screening and<br />

Diagnostic Tests 1996). Where a study contained one or more components, the<br />

assessment of quality was only of the component used in this assessment. Two<br />

evaluators independently scored each of the included studies, with discrepancies resolved<br />

by discussion and consensus. This enabled a quality score to be assigned to each study.<br />

Details of the checklist and scoring are provided in Appendix H.<br />

Other important in<strong>for</strong>mation that was assessed included:<br />

• appropriateness of <strong>screening</strong> <strong>population</strong> to proposed <strong>population</strong><br />

• appropriateness of the test to proposed indication (eg, timing of FOBT)<br />

• appropriateness of <strong>screening</strong> method/intervention used<br />

• appropriateness of outcome measurement (eg, sensitivity specificity, etc.)<br />

• strength and magnitude of <strong>screening</strong> effect.<br />

Additional searches<br />

Additional searches were conducted to examine the following:<br />

• psychological impact of faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

• economic evaluations of FOBTs<br />

• in<strong>for</strong>mation required <strong>for</strong> the modelled economic evaluation.<br />

12 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Expert advice<br />

An advisory panel with expertise in epidemiology, gastroenterology, pathology and<br />

gastrointestinal surgery was established to evaluate the evidence and provide advice to<br />

the Medical Services Advisory Committee (MSAC) from a clinical perspective. In<br />

selecting members <strong>for</strong> advisory panels, the MSAC’s practice is to approach the<br />

appropriate medical colleges, specialist societies and associations and consumer bodies<br />

<strong>for</strong> nominees. Membership of the advisory panel is provided at Appendix B.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 13


Results of assessment<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

Available evidence<br />

Head-to-head studies<br />

Table 2 shows the 30 head-to-head studies of faecal <strong>occult</strong> <strong>blood</strong> tests (FOBTs)<br />

per<strong>for</strong>med in a <strong>screening</strong> setting that were identified in the literature search. Five of these<br />

studies described duplicate data and were excluded from further analysis (Armitage and<br />

Hardcastle 1987; Castiglione et al 1996; Hardcastle et al 1986; Robinson et al 1995;<br />

Walter et al 1991). One study could not be obtained (Aisawa et al 1988). One study did<br />

not utilise an appropriate reference standard (Vaananen and Tenhunen 1988). Four other<br />

studies were excluded due to inadequate data separation (Cole et al 2003; Lee and<br />

Costello 1982; Li et al 2000; Winawer et al 1980a). One study was excluded because a<br />

large portion of the included patient <strong>population</strong> was symptomatic (Rae and Cleator<br />

1994). One study was excluded as the immunochemical test studied was never<br />

commercially produced (Frommer et al 1988). Three other studies were excluded from<br />

the assessment as data enabling calculation of a suitable outcome measure were not<br />

reported (Kettner et al 1990; St John and Young 2003; Verne et al 1993). Fourteen headto-head<br />

studies remained <strong>for</strong> evaluation of the effectiveness of faecal <strong>occult</strong> <strong>blood</strong><br />

<strong>testing</strong>.<br />

Table 2 Head-to-head studies of FOBT in <strong>screening</strong> <strong>population</strong>s<br />

Trial Trial design Reviewed <strong>for</strong><br />

effectiveness<br />

assessment<br />

Reviewed <strong>for</strong><br />

safety<br />

assessment<br />

Aisawa et al 1988 Unavailable x x<br />

Allison et al 1996b Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

� �<br />

Armitage et al 1985 Prospective, parallel-arm, patients randomly assigned to two<br />

sequential FOBTs or control group<br />

Armitage and<br />

Hardcastle 1987<br />

Barrison and<br />

Parkins 1985a<br />

Castiglione et al<br />

1996<br />

Castiglione et al<br />

1997<br />

Prospective, parallel-arm, patients randomly assigned to two<br />

sequential FOBTs or control group (cohort of large RCT)<br />

Prospective, single group, with each participant receiving<br />

multiple FOBTs (GP group)<br />

Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Cole et al 2003 Prospective, parallel arm, patients randomly assigned to one<br />

of three FOBTs<br />

Frommer et al 1988 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

� �<br />

x �<br />

� �<br />

x �<br />

� �<br />

x �<br />

x �<br />

Fujita et al 1992 Parallel arm, with historical cohort � �<br />

Hardcastle et al<br />

1986<br />

Prospective, parallel-arm, patients randomly assigned to two<br />

sequential FOBTs or control group (group 2, cohort of large<br />

RCT)<br />

x �<br />

14 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Trial Trial design Reviewed <strong>for</strong><br />

effectiveness<br />

assessment<br />

Iwase 1992 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Kettner et al 1990 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Lee and Costello<br />

1982<br />

Prospective, single group, with each participant receiving<br />

multiple FOBTs (groups 1 & 2)<br />

Levin et al 1997 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Li et al 2000 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Petrelli et al 1994 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Rae and Cleator<br />

1994<br />

Robinson et al<br />

1995<br />

Robinson et al<br />

1996<br />

Prospective, single group, with each participant receiving<br />

multiple FOBTs (arm 2)<br />

Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Prospective, single group, with each participant receiving<br />

multiple FOBTs (study 1)<br />

Rozen et al 1995 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Rozen et al 1997 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Rozen et al 2000 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

St John et al 1993 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

St John and Young<br />

2003<br />

Vaananen and<br />

Tenhunen 1988<br />

Prospective, randomised, with cross-over between dietary<br />

restriction and no dietary restriction, with each participant<br />

receiving multiple FOBTs<br />

Prospective, single group, with each participant receiving<br />

multiple FOBTs (<strong>health</strong>y subject group)<br />

Verne et al 1993 Prospective, parallel-arm, patients randomly assigned to one<br />

of three FOBTs with or without dietary restriction<br />

Walter et al 1991 Prospective, single group, with each participant receiving<br />

multiple FOBTs<br />

Winawer et al<br />

1980b<br />

Prospective, pseudo randomised to control or intervention<br />

groups, with different FOBTs received over different times<br />

periods<br />

Zappa et al 2001 Different tests per<strong>for</strong>med on non-randomly allocated<br />

individuals, with the likelihood of receiving different tests<br />

changing over time as immunochemical tests are introduced<br />

Abbreviations: GP, general practice; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; RCT, randomised controlled trial.<br />

Reviewed <strong>for</strong><br />

safety<br />

assessment<br />

� �<br />

x �<br />

x �<br />

� �<br />

x �<br />

� �<br />

x �<br />

x �<br />

� �<br />

� �<br />

� �<br />

� �<br />

� �<br />

x �<br />

x �<br />

x �<br />

x �<br />

x �<br />

� �<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 15


Is it safe?<br />

Adverse events associated with faecal <strong>occult</strong> <strong>blood</strong> tests<br />

FOBTs are non-invasive and there<strong>for</strong>e unlikely to cause adverse events. However, the<br />

use of FOBTs in a <strong>screening</strong> setting is likely to increase the number of colonoscopies,<br />

sigmoidoscopies and barium enemas per<strong>for</strong>med in the screened <strong>population</strong>. There<strong>for</strong>e,<br />

the increased numbers of adverse events associated with these procedures are important.<br />

The rates of adverse events associated with colonoscopy, sigmoidoscopy and double<br />

contrast barium enema (DCBE) are discussed below.<br />

Adverse events associated with colonoscopy, sigmoidoscopy and double contrast<br />

barium enema<br />

Colonoscopy is per<strong>for</strong>med as a day-case procedure and generally requires sedation. In a<br />

study of colonoscopy data over a five-year period, diagnostic colonoscopy was associated<br />

with a complication rate of 0.14 per cent, compared with a rate of 1.8 per cent <strong>for</strong><br />

therapeutic colonoscopy (Reiertsen et al 1987). In a 1997 review of six prospective<br />

studies it was estimated that around 1 in 1000 patients suffer per<strong>for</strong>ation, 3 in 1000 suffer<br />

major haemorrhage and between 1 and 3 in 10,000 die as a result of colonoscopy<br />

(Winawer et al 1997). A 1989 review of nine studies of diagnostic colonoscopy gives a<br />

slightly higher figure of around 1.7 in 1000 patients suffering per<strong>for</strong>ation and similar<br />

figures <strong>for</strong> haemorrhage and death (Habr-Gama and Waye 1989). The risk of per<strong>for</strong>ation<br />

and haemorrhage increased with the per<strong>for</strong>mance of polypectomy. There are also other<br />

occasional serious complications associated with bowel preparation prior to colonoscopy<br />

or the use of sedation in conjunction with this procedure (National Health and Medical<br />

Research Council (NHMRC) 1999).<br />

Per<strong>for</strong>ation rates associated with sigmoidoscopy are low and occur in fewer than 2 cases<br />

per 10,000 examinations (NHMRC 1999). Sedation is not generally used during<br />

sigmoidoscopy.<br />

DCBE is per<strong>for</strong>med as an outpatient procedure. Sedation is not used. Serious<br />

complications are rare and include bowel per<strong>for</strong>ation and cardiac complications. The<br />

complication rate of DCBE has been estimated at 3 per 10,000 tests, with a death rate of<br />

3 in 100,000 tests (Winawer et al 1997).<br />

The Minnesota (US) randomised controlled trial (RCT) of FOBT <strong>screening</strong> <strong>for</strong> colorectal<br />

cancer (CRC) reported safety outcomes relevant to <strong>population</strong> <strong>health</strong> <strong>screening</strong>. This<br />

study utilised the Hem<strong>occult</strong> test in combination with appropriate dietary restrictions.<br />

A per<strong>for</strong>ation rate of 0.03% (4/12,246) and a rate of serious bleeding of 0.09%<br />

(11/12,246) were reported in association with colonoscopy (Mandel et al 1993). All<br />

four instances of per<strong>for</strong>ation, and three of the eleven cases of serious bleeding, required<br />

surgery.<br />

The Nottingham (UK) FOBT trial reported complication rates in a smaller sample<br />

of colonoscopies. This study also used the Hem<strong>occult</strong> test in combination with<br />

dietary restrictions. In this study, a rate of 0.5% complications of colonoscopy and<br />

no complications of DCBE from a total of 1778 subjects receiving follow-up was<br />

16 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


eported (Robinson et al 1999). Of the seven patients experiencing complications<br />

associated with colonoscopy (7/1474), six required surgery. There were no colonoscopyrelated<br />

deaths reported in either study.<br />

Psychiatric morbidity associated with colorectal cancer <strong>screening</strong><br />

Concern has been expressed about the psychological impact of CRC <strong>screening</strong>,<br />

particularly <strong>for</strong> the group of participants that return false positives results. A recent<br />

publication by Parker et al (2002) investigated the psychiatric morbidity associated with<br />

CRC <strong>screening</strong> in an RCT. The RCT aimed to determine: (a) whether the offer of faecal<br />

<strong>occult</strong> <strong>blood</strong> <strong>testing</strong> caused increased, sustained anxiety; (b) whether the existence of<br />

psychiatric morbidity was a factor in the decision to accept or refuse <strong>screening</strong>; and (c)<br />

the anxiety levels of participants returning positive FOBTs at <strong>screening</strong>, clinical<br />

investigation and follow up.<br />

The study found that participation in a faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> program <strong>for</strong> CRC<br />

did not cause sustained anxiety or psychiatric morbidity. The study also showed that a<br />

positive <strong>screening</strong> result caused short-term anxiety in most subjects, but this returned to<br />

a lower level one month after clinical investigation (anxiety was measured on the<br />

Spielberger anxiety inventory). Similarly, the study showed no sustained anxiety in<br />

participants returning false positive results. This is important, as it has been suggested<br />

that the high false positive rate in faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> will undermine public<br />

confidence in the <strong>screening</strong> programs. In fact, when questioned one month after<br />

investigation, 85 per cent of participants stated that they would accept <strong>screening</strong> if it was<br />

offered again in 2 years.<br />

Another trial that investigated the experiences of participants receiving false positive<br />

results from CRC <strong>screening</strong> found that 26 per cent were quite distressed or very<br />

distressed from the initial false positive result (Mant et al 1990). The remaining 74 per<br />

cent of patients were slightly distressed or not at all distressed regarding the false positive<br />

FOBT result. A total of 98.1 per cent of the participants returning false positive results in<br />

this study believed that it was worthwhile having the test.<br />

This is an important issue; however, exploration of this in detail would required a<br />

dedicated systematic review and exploration via a cost-effectiveness comparison<br />

incorporating quality-of-life measures, which is beyond the scope of this review.<br />

Safety data reported in the FOBT studies<br />

The head-to-head studies examining the relative per<strong>for</strong>mance of different FOBTs<br />

identified in this assessment did not report any safety data.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 17


Is it effective?<br />

Diagnostic per<strong>for</strong>mance<br />

This assessment report focuses on the relative effectiveness of the different commercially<br />

available FOBTs in the <strong>screening</strong> setting. Table 3 presents the characteristics of the 14<br />

head-to-head studies of FOBT identified in the literature search and subsequently<br />

determined as suitable <strong>for</strong> inclusion. Each study was rated according to a hierarchy of<br />

evidence <strong>for</strong> <strong>screening</strong> tests as described by the New Zealand Health Technology<br />

Assessment (NZHTA) group (Broadstock 2000). Each trial was evaluated <strong>for</strong> quality<br />

using a quality-scoring checklist based on the recommendations of the Cochrane<br />

collaboration (see Appendix H).<br />

Five studies compared Hem<strong>occult</strong> to HemeSelect (Allison et al 1996; Castiglione et al<br />

1997; Petrelli et al 1994; Robinson et al 1996; Rozen et al 1997). Four studies compared<br />

Hem<strong>occult</strong> Sensa to HemeSelect (Allison et al 1996; Petrelli et al 1994; Rozen et al 1997;<br />

St John et al 1993). One study compared Hem<strong>occult</strong> to Fecatwin (guaiac) (Barrison and<br />

Parkins 1985) and another to Fecatwin Sensitive/Feca EIA (Armitage et al 1985). Single<br />

studies were also identified that compared: Hem<strong>occult</strong> with Flexsure OBT (Rozen et al<br />

1997); Hem<strong>occult</strong> Sensa with Flexsure OBT (Rozen et al 2000); Hem<strong>occult</strong> with<br />

Immudia-HemSp (Iwase 1992); Shionogi B with Immudia-HemSp (Fujita et al 1992);<br />

Hem<strong>occult</strong> Sensa with BM-Test Colon Albumin (Rozen et al 1995); Hem<strong>occult</strong> with<br />

Hem<strong>occult</strong> Sensa (Levin et al 1997) and; Hem<strong>occult</strong> with RPHA tests (Zappa et al 2001).<br />

Twelve studies represented level 1 evidence according to the evidence hierarchy designed<br />

<strong>for</strong> the assessment of studies of <strong>screening</strong> tests (Broadstock 2000). Two studies were<br />

rated as level 3b evidence and were excluded from further review due to poor study<br />

quality (Fujita et al 1992; Zappa et al 2001). Since the key issue to be addressed in this<br />

review is the relative per<strong>for</strong>mance of guaiac versus immunochemical FOBTs when used<br />

<strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>, studies that made comparisons between different types<br />

of guaiac tests were also excluded from further review (Barrison and Parkins 1985; Levin<br />

et al 1997). In addition, a single study that examined the use of rehydrated Hem<strong>occult</strong><br />

was excluded from the analysis (Castiglione et al 1997). Nine studies remained <strong>for</strong> review.<br />

Studies conducted in <strong>screening</strong> <strong>population</strong>s that included a high proportion of patients at<br />

increased risk were included in sensitivity analyses of estimates of test sensitivity,<br />

specificity and positive predictive value (Rozen et al 1995; Rozen et al 1997; Rozen et al<br />

2000; St John et al 1993).<br />

In accordance with the inclusion criteria, all studies were head-to-head studies conducted<br />

in ‘average risk’ general <strong>population</strong> subjects which utilised an appropriate reference<br />

standard. The included studies all used a within-subjects design. Outside of these<br />

parameters, many of the included studies had quality limitations. All of the studies had<br />

received industry support in some <strong>for</strong>m. All of the studies included in the main analyses<br />

were prospectively designed. However, it is unclear whether or not the studies by Rozen<br />

et al, which are included in the sensitivity analyses, were prospectively designed (Rozen et<br />

al 1995; Rozen et al 1997; Rozen et al 2000).<br />

An inherent problem with <strong>screening</strong> studies, including all of the studies in this<br />

assessment, is the issue of verification bias. In studies of <strong>screening</strong> tests it is generally<br />

18 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


unacceptable and unethical to subject asymptomatic people with a negative test to what<br />

may be a battery of invasive, time-consuming, and costly clinical investigations.<br />

There<strong>for</strong>e, the use and measurement of the reference standard was not independent of<br />

the test results. In only one of the studies (Allison et al 1996) were results of the different<br />

FOBTs determined in a blinded fashion. In addition, the selection of subjects in the<br />

included trials was not reported to be random in any of the studies. In two of the studies<br />

in the sensitivity analyses the selection of subjects was consecutive (Rozen et al 1997;<br />

Rozen et al 2000). In all of the studies in the main analyses, external validity was good;<br />

however, the external validity was reduced in the sensitivity analyses where studies with a<br />

high proportion of patients at increased risk were included.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 19


Table 3 Characteristics of the head-to-head FOBT studies per<strong>for</strong>med in a <strong>screening</strong> <strong>population</strong><br />

Long-term followup<br />

Randomisation /<br />

multiple tests in a<br />

single patient<br />

Reference standard<br />

FOBTs compared<br />

Population (ITS)<br />

Study<br />

QS/12<br />

LoE<br />

Yes, 2 years review<br />

of insurance records<br />

<strong>for</strong> neoplasms (96%<br />

of subjects)<br />

Multiple tests<br />

Positive FOBT: colonoscopy;<br />

Hem<strong>occult</strong> Sensa positive: FSa Yes, 1 year review of<br />

hospital pathology<br />

records <strong>for</strong><br />

neoplasms<br />

(Hardcastle et al<br />

1986)<br />

Yes, 2 years<br />

follow-up <strong>for</strong><br />

colonic symptom<br />

presentation to GP<br />

Multiple tests<br />

• Hem<strong>occult</strong><br />

• Hem<strong>occult</strong> Sensa<br />

• HemeSelect<br />

• Hem<strong>occult</strong> Sensa +<br />

Hemeselect<br />

• Hem<strong>occult</strong> (no dietary<br />

restriction)<br />

• Fecatwin Sensitive/<br />

Feca EIAb Members of non-profit US <strong>health</strong><br />

insurer electing <strong>health</strong> appraisal, ><br />

50 years; N = 10,702<br />

Allison et al<br />

1996<br />

7<br />

1<br />

Multiple tests<br />

Positive: history, exam, RS and 60 cm<br />

fibreoptic sigmoidoscopy, then if<br />

carcinoma: DCBE; adenoma:<br />

colonoscopy; no neoplasms: DCBE<br />

If negative follow-up: repeat FOBT<br />

If positive, gastroscopy<br />

Positive: history, exam, RS, sample<br />

obtained on exam tested with Haemostix<br />

If positive or pathology suspected: full<br />

investigation<br />

If negative: repeat FOBT after 3 months<br />

Individuals from GP registers (UK),<br />

45–75 years; N = 3225; excluded<br />

known large bowel disease patients<br />

Armitage et al<br />

1985<br />

6<br />

• Hem<strong>occult</strong> (no dietary<br />

restriction)<br />

• Fecatwin (guaiac)<br />

GP patient attendees (UK),<br />

> 40 years; N = 640 (group 1)<br />

Barrison and<br />

Parkins 1985<br />

Excl<br />

Yes, 2 year repeat<br />

FOBT <strong>screening</strong> in<br />

25% of subjects<br />

20 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

Multiple tests<br />

No<br />

Multiple tests<br />

Positive HO and/or positive/borderline<br />

HemeSelect: pancolonoscopy or left<br />

Colonoscopy plus DCBE when<br />

pancolonoscopy not possible<br />

All: FS<br />

Positive: DCBE or colonoscopy<br />

• Hem<strong>occult</strong><br />

(rehydrated)<br />

• HemeSelect<br />

General <strong>population</strong> (Italy),<br />

40–70 years; N = 24,282<br />

Castiglione<br />

et al 1997<br />

Excl<br />

• Hem<strong>occult</strong><br />

• Immudia–HemSp<br />

Individuals presenting <strong>for</strong> <strong>health</strong><br />

care examination (Japan), most 50–<br />

59 years (mean 52 years); N = 6437<br />

Iwase 1992<br />

5<br />

No<br />

Multiple tests<br />

Positive: full colonoscopy or FS + DCBE<br />

No<br />

Multiple tests<br />

Positive: mixed, including exam, repeat<br />

FOBT, DCBE, sigmoidoscopy,<br />

colonoscopy or an upper gastrointestinal<br />

barium series. 70% had DCBE or<br />

colonoscopy<br />

• Hem<strong>occult</strong><br />

(unhydrated)<br />

• Hem<strong>occult</strong><br />

(rehydrated)<br />

• Hem<strong>occult</strong> Sensa<br />

• Hem<strong>occult</strong><br />

• Hem<strong>occult</strong> Sensa<br />

• HemeSelect<br />

General <strong>population</strong> (US, distribution<br />

through pharmacies and community<br />

groups following media campaign),<br />

≥ 50 years; N = 85,931<br />

Levin et al 1997<br />

Excl<br />

General <strong>population</strong> (US, distribution<br />

through selected pharmacies<br />

following media campaign), > 30<br />

years; N = 39,000<br />

Petrelli et al<br />

1994<br />

5


Long-term follow-up<br />

Randomisation/<br />

multiple tests in<br />

a single patient<br />

Reference standard<br />

FOBTs<br />

compared<br />

Population (ITS)<br />

Study<br />

QS/<br />

12<br />

LoE<br />

Yes, median 35 months by GP<br />

in<strong>for</strong>mation and review of hospital<br />

pathology records <strong>for</strong> neoplasms<br />

Multiple tests<br />

Positive: exam,<br />

sigmoidoscopy and<br />

colonoscopy/FS & DCBE<br />

• Hem<strong>occult</strong><br />

• HemeSelect<br />

No<br />

Multiple tests<br />

No<br />

Multiple tests<br />

All had endoscopy, FS<br />

(41%) or total colonoscopy<br />

(59%, including FOBT<br />

positive)<br />

All had endoscopy, 59% at<br />

time of study<br />

• Hem<strong>occult</strong> Sensa<br />

• BM-Test Colon<br />

Albumin<br />

Individuals from GP registers (UK),<br />

50–75 years; N = 4018. Known colorectal<br />

neoplasia, advanced visceral malignancy or<br />

unfit <strong>for</strong> further investigation excluded<br />

Consecutive attendees at CRC <strong>screening</strong><br />

service (95% asymptomatic) (Israel); N = nr<br />

(527 screened)<br />

Robinson et<br />

al 1996<br />

6<br />

1<br />

Rozen et al<br />

1995<br />

4<br />

Yes, > 2 years clinical follow-up in<br />

4 subjects with positive FOBT who<br />

refused colonoscopy<br />

Multiple tests<br />

All had full colonoscopy or<br />

FS, 48% at time of study<br />

• Hem<strong>occult</strong><br />

• Hem<strong>occult</strong> Sensa<br />

• FlexSure OBT<br />

• HemeSelect<br />

• Hem<strong>occult</strong> Sensa<br />

• FlexSure OBT<br />

Consecutive attendees at CRC <strong>screening</strong><br />

service (97% asymptomatic) (Israel); N = nr<br />

(403 screened)<br />

Rozen et al<br />

1997<br />

5<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 21<br />

No<br />

Multiple tests<br />

Positive: colonoscopy<br />

• Hem<strong>occult</strong> Sensa<br />

• HemeSelect<br />

Consecutive attendees at CRC <strong>screening</strong><br />

service (97% asymptomatic) (Israel); N = nr<br />

(1410 screened)<br />

Participants in <strong>screening</strong> program (individuals<br />

with FH CRC (80%) and community<br />

volunteers) (Australia), 45–79 years; N = nr<br />

(1355 screened)<br />

Rozen et al<br />

2000<br />

5<br />

St John et al<br />

1993<br />

5<br />

None available<br />

None available<br />

No<br />

None available<br />

Excl<br />

2a<br />

2b<br />

3a<br />

3b<br />

Parallel arms with<br />

historical cohort<br />

Positive, symptomatic or<br />

FH CRC: exam, FS and<br />

DCBE<br />

• Shionogi B<br />

• Immudia-HemSp<br />

Mass <strong>screening</strong> (Japan), > 40 years;<br />

N = 18,497<br />

Fujita et al<br />

1992<br />

Yes, cancer registry data over<br />

5 years and negative FOBTs<br />

rescreened in 2.5 years<br />

Parallel arms, nonrandomised<br />

Total colonoscopy or<br />

colonoscopy & DCBE<br />

• Hem<strong>occult</strong><br />

• RPHA<br />

(HemeSelect;<br />

Immudia–HemSp<br />

General <strong>population</strong> (Italy), 50–70 years;<br />

N = nr (41,744 screened)<br />

Zappa et al<br />

2001<br />

Excl<br />

Abbreviations: CRC, colorectal cancer; DCBE, double contrast barium enema; Excl, excluded from further review; FH, family history; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; FS, flexible sigmoidoscopy; GP, general practice; HO, Hem<strong>occult</strong>;<br />

LoE, level of evidence; nr, not reported; QS, quality score ; RPHA, reverse-passive haemagglutination; RS, rigid sigmoidoscopy.<br />

aDue to the unacceptably high number of patients whose only positive test was Hem<strong>occult</strong> Sensa and whose colonoscopic examinations revealed no colorectal carcinoma, patients were advised to first undergo flexible sigmoidoscopy<br />

and repeated Hem<strong>occult</strong> <strong>testing</strong> at 6 and 12 months. Colonoscopy was offered to anyone who was found to have a colorectal neoplasm on sigmoidoscopy; bFecatwin Sensitive = Fecatwin/Feca EIA.


Participation<br />

An important component of any <strong>population</strong> based <strong>screening</strong> program is the participation<br />

rate of those targeted <strong>for</strong> <strong>screening</strong>. There are many reasons <strong>for</strong> non-participation in<br />

FOBT <strong>screening</strong> programs including: intercurrent illness; the lack of appreciation of the<br />

concept of asymptomatic illness; the fear of further tests and surgery; the unpleasantness<br />

of the stool collection procedure and the fear of cancer itself (Hynam et al 1995).<br />

Non-participation in the targeted <strong>population</strong> negatively impacts on the cost and<br />

effectiveness of the <strong>screening</strong> program. Since participation may vary between FOBTs<br />

(eg, due to the need to adhere to dietary restrictions), it is important that estimates of the<br />

relative participation rates associated with each FOBT are incorporated into any<br />

assessment of cost-effectiveness. Estimates of the likely participation rates of guaiac<br />

versus immunochemical FOBTs in an Australian <strong>population</strong> <strong>health</strong> <strong>screening</strong> program<br />

were obtained from a study conducted in an Australian average risk <strong>population</strong> (Cole et<br />

al 2003) and data from the Australian Government Bowel Cancer Screening Pilot<br />

Program (the pilot). The effects of participation rates on the relative cost-effectiveness of<br />

different FOBTs are discussed in the economic section of this report.<br />

Change in clinical management and clinical outcomes<br />

Results<br />

The identification of early stage disease or potential precursors of disease by FOBTs and<br />

subsequent curative procedures such as preventative polypectomy and surgery can<br />

significantly reduce CRC morbidity and mortality. FOBT <strong>screening</strong> <strong>for</strong> CRC in the<br />

general <strong>population</strong> has been shown to reduce CRC mortality in a number of large<br />

randomised and non-randomised controlled trials. A systematic review and meta-analysis<br />

of RCTs per<strong>for</strong>med by the Cochrane Collaboration has shown that <strong>screening</strong> <strong>for</strong> CRC in<br />

asymptomatic patients using a FOBT (Hem<strong>occult</strong>) can reduce the relative risk of CRC<br />

mortality by around 16 per cent (relative risk 0.84; 95% CI: 0.77, 0.89) (Towler et al<br />

1998).<br />

Diagnostic per<strong>for</strong>mance<br />

The accepted methodology <strong>for</strong> investigating the accuracy of new diagnostic tests is to<br />

compare the diagnosis made with the new test with the true disease status. However, it is<br />

often not feasible to determine the disease status of a patient unequivocally. There<strong>for</strong>e, in<br />

many disease states a proxy measure, such as another diagnostic test or clinical<br />

judgement, must be used. The best available measure of disease is called the reference<br />

standard. Both the test result and the comparator result must be independently compared<br />

with the reference standard to assess sensitivity, specificity and accuracy. However, in<br />

studies of <strong>screening</strong> tests it is often unacceptable and unethical to subject asymptomatic<br />

people with a negative test to what may be a battery of invasive, time-consuming, and<br />

costly clinical investigations.<br />

22 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Three different methods are used to circumvent this problem and investigate the relative<br />

per<strong>for</strong>mance of the FOBTs identified in this assessment. An explanation of these<br />

methods and their limitations are listed below.<br />

1. The relative sensitivity and specificity of each of the FOBTs were estimated using the<br />

interval cancer rate (ie, the number of cancers that are detected during intervals<br />

between <strong>screening</strong>) as a proxy <strong>for</strong> the false negative rate (FNR).<br />

The sensitivity of a test is a measure of how good the test is at picking up people<br />

who have the condition:<br />

True positive/(True positive + False negative).<br />

The specificity of a test is a measure of how good the test is at correctly excluding<br />

people without the condition:<br />

True negative/(True negative + False positive).<br />

The diagnostic odds ratio is a measure of test accuracy combining sensitivity and<br />

specificity measures:<br />

(True positive × True negative)/(False negative × False positive)].<br />

These data are reported in Appendix G.<br />

2. A useful measure of the relative per<strong>for</strong>mance of <strong>screening</strong> tests, where sensitivity and<br />

specificity data are not available, is the relative true positive rate (TPR) and the<br />

relative false positive rate (FPR). These values provide a measure of the relative<br />

per<strong>for</strong>mance of the tests, when both tests are conducted in each person and negative<br />

<strong>screening</strong> test results are not followed-up with an appropriate reference standard<br />

(Chock et al 1997).<br />

The actual measure of TPR and FPR will depend on the disease prevalence within<br />

the study <strong>population</strong>. However, the relative TPR and the relative FPR provide a<br />

useful and unbiased measure of comparative FOBT per<strong>for</strong>mance.<br />

The TPR of a test is:<br />

The number of true positive results (confirmed by the reference<br />

standard)/the number of participants with disease confirmed by the<br />

reference standard.<br />

The FPR of a test is:<br />

The number of false positive results (excluded by the reference standard)/the<br />

number of participants without disease, confirmed by the reference standard.<br />

These values are unknown in <strong>screening</strong> studies where negative test results are not<br />

followed up with the reference standard.<br />

The relative TPR is the ratio of the TPR of one test to the TPR of the other. The<br />

relative FPR is the ratio of the FPR of one test to the FPR of the other. These ratios<br />

can be determined without knowing the total number of patients with the disease.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 23


3. The positive predictive value (PPV) <strong>for</strong> each FOBT was also determined; these data<br />

are reported in Appendix F, <strong>for</strong> completeness.<br />

The PPV is the probability that if a person tests positive that they actually have the<br />

condition:<br />

True positive/(True positive + False positive).<br />

Meta-analyses<br />

This project required separate meta-analyses of various outcomes measuring the<br />

efficacy of the test being examined. The analyses were also carried out <strong>for</strong> various<br />

<strong>population</strong>s.<br />

Typically, each set of trial results consisted of two or more observed proportions, one <strong>for</strong><br />

each test used. For each trial, the tests were usually repeated on the same set of patients.<br />

These circumstances, in general, preclude the use of classical meta-analyses methods <strong>for</strong><br />

summary data, as these methods were developed <strong>for</strong> two-arm trials with independent sets<br />

of subjects in each arm.<br />

In addition, there was usually a small number of trials involved in each meta-analysis.<br />

Classical random-effects meta-analysis (Der Simonian and Laird 1986) implicitly assumes<br />

that a between-studies variance estimate used in the calculation of the weights <strong>for</strong> the<br />

method is known with precision, and this is not likely to be true <strong>for</strong> a small number of<br />

trials.<br />

A further source of variability may be that diagnostic test results are sensitive to the<br />

threshold set <strong>for</strong> the test, which may vary between studies (see Lijmer et al 2002 <strong>for</strong><br />

details).<br />

The circumstances outlined above suggest that a more sophisticated random-effects<br />

analysis is required, and a Bayesian approach was deemed to provide a more appropriate,<br />

conservative approach (Higgins and Whitehead 1996; Smith et al 1995). The analyses<br />

were carried out using Markov Chain Monte Carlo (MCMC) simulation as implemented<br />

in the statistical package WinBUGS (version 1.3, Spiegelhalter et al 2000).<br />

The method is based on fitting logistic regression models incorporating random effects.<br />

A random intercept model of the <strong>for</strong>m was adopted:<br />

logit(p ij) = α + b i + β j × x ij,<br />

where,<br />

• p ij = prob(Y ij = 1) is the probability that the event occurs<br />

• α is an intercept, interpreted as the log-odds <strong>for</strong> the event occurring in a<br />

reference category, in this case the guaiac test group<br />

• b i is a random effect <strong>for</strong> the i th study, which is common across all categories of<br />

interest<br />

24 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


• β j is a fixed effect which is the deviation, on the log-odds scale, of the event rate<br />

in the j th category from that of the reference category, and<br />

• x ij is an indicator <strong>for</strong> the j th category in study i.<br />

The random study effects were assumed to be normally distributed with 0 mean and<br />

2 2<br />

variance σb , and it was also assumed that 1/σb , also known as the (between-studies)<br />

“precision” parameter, has a gamma distribution (Smith et al 1995). The parameters were<br />

chosen so that the prior distribution (gamma(0.001,0.001)) was “unin<strong>for</strong>mative”, which<br />

2<br />

leads to a conservative posterior distribution <strong>for</strong> 1/σb .<br />

Unin<strong>for</strong>mative normal prior distributions (N(0,10 6 )) were set <strong>for</strong> the fixed parameters (α<br />

and β).<br />

The results given below include the median estimate of the fixed effects, together with<br />

95% credibility intervals, which are the range of estimated values that lie between 2.5%<br />

and 97.5% percentiles of the values which make up the posterior distribution <strong>for</strong> each<br />

parameter of interest.<br />

The difference between effects of any pair of diagnostic measures <strong>for</strong> different tests can<br />

be obtained by taking the difference between the estimated β’s corresponding to the<br />

various diagnostic test groups.<br />

For the relative true positive and false positive rates, frequentist meta-analysis methods<br />

were used, with inverse-variance weighting (Whitehead and Whitehead 1991). The<br />

standard heterogeneity test based on the (Cochran’s)“Q” statistic was used to choose<br />

between results from fixed- and random-effects variants.<br />

Sensitivity and specificity<br />

Three studies estimated the sensitivity, specificity and accuracy of various FOBTs using<br />

the interval cancer rate as a proxy <strong>for</strong> the FNR (Allison et al 1996; Armitage et al 1985;<br />

Robinson et al 1996). The analyses of the sensitivity and specificity data from these<br />

studies constitute the main analyses reported in Table 4 and Table 5, respectively. The<br />

sensitivity is the proportion of all patients with CRC who test positive. That is, how good<br />

the test is at picking up participants who have the condition. The specificity is the<br />

proportion of all patients without CRC who test negative. That is, how good the test is at<br />

correctly excluding people without the condition.<br />

Another three studies estimated the sensitivity and specificity of various FOBTs in<br />

<strong>screening</strong> <strong>population</strong>s with a high proportion of participants who were at higher risk of<br />

disease, by subjecting all participants to the endoscopic examination (Rozen et al 1995;<br />

Rozen et al 1997; Rozen et al 2000). However, in these studies the endoscopic<br />

examination had been per<strong>for</strong>med up to several years previously in a proportion of the<br />

patients <strong>testing</strong> negative with FOBT. In addition, some of the subjects with a negative<br />

FOBT had received flexible sigmoidoscopy rather than colonoscopy. There<strong>for</strong>e, these<br />

studies were still subject to verification bias. These studies have been included only in a<br />

sensitivity analyses. These studies also provided the only estimates of sensitivity and<br />

specificity of various FOBTs <strong>for</strong> adenoma detection. There<strong>for</strong>e, there were no data of a<br />

suitable quality available to assess relative accuracy of the tests <strong>for</strong> the detection of<br />

adenomas.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 25


HemeSelect was found to be significantly more sensitive than Hem<strong>occult</strong> (77.1% vs<br />

30.0%, respectively). The odds ratio (OR) and risk difference (RD) <strong>for</strong> this comparison<br />

was 7.88 (95% confidence interval (CI): 3.09, 21.47) and 0.46 (95% CI: 0.231, 0.631),<br />

respectively. These differences were maintained in the sensitivity analysis. No statistically<br />

significant difference was identified in any of the other comparisons per<strong>for</strong>med. The<br />

level of heterogeneity was relatively low in all of the analyses per<strong>for</strong>med.<br />

The Hem<strong>occult</strong> test was significantly more specific than HemeSelect (97.8% vs 93.5%,<br />

respectively). The OR and RD <strong>for</strong> this comparison was 0.33 (95% CI: 0.28, 0.39) and<br />

–0.043 (95% CI: –0.070, –0.026), respectively. This difference was maintained in the<br />

sensitivity analysis.<br />

The specificity of Hem<strong>occult</strong> was significantly greater than that <strong>for</strong> Fecatwin<br />

Sensitive/Feca EIA (97.1% vs 92.2%, respectively). The OR and RD <strong>for</strong> this comparison<br />

was 0.35 (95% CI: 0.24, 0.51) and –0.049 (95% CI: –0.066, –0.032), respectively.<br />

The HemeSelect test proved to be significantly more specific than Hem<strong>occult</strong> Sensa<br />

(94.4% vs 86.7%, respectively). The OR and RD <strong>for</strong> this comparison was 2.58 (95% CI:<br />

2.29, 2.91) and 0.077 (95% CI: 0.068, 0.086), respectively. Again, this difference was<br />

maintained in the sensitivity analysis.<br />

In a study with a large portion of high-risk patients, the immunochemical test Flexsure<br />

OBT did not have a significantly different specificity to that of Hem<strong>occult</strong> (96.7% vs<br />

94.7%, respectively). In another study in a similar patient <strong>population</strong>, the specificity of<br />

Flexsure OBT was significantly greater than that of Hem<strong>occult</strong> Sensa (98.5% vs 94.9%,<br />

respectively). The OR and RD <strong>for</strong> this comparison was 3.51 (95% CI: 2.14, 5.74) and<br />

0.036 (95% CI: 0.023, 0.040), respectively.<br />

Similarly, in a study with a large portion of high-risk patients, the immunochemical test<br />

BM-Test Colon Albumin also had a significantly greater specificity than Hem<strong>occult</strong><br />

Sensa (89.5% vs 83.7%, respectively). The OR and RD <strong>for</strong> this comparison was 1.65<br />

(95% CI: 1.15, 2.38) and 0.057 (95% CI: 0.016, 0.099), respectively.<br />

26 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Table 4 Meta-analyses of sensitivity <strong>for</strong> carcinoma from FOBT head-to-head studies<br />

Heterogeneity<br />

(p)<br />

RD<br />

pooled results<br />

(95% CI)<br />

OR<br />

pooled results<br />

(95% CI)<br />

Immunochemical<br />

pooled<br />

sensitivity (%)<br />

Guaiac pooled<br />

sensitivity (%)<br />

Sensitivity n/N<br />

(true +ve/total<br />

carcinoma)<br />

(%)<br />

FOBT<br />

Studies<br />

FOBT comparison<br />

Main guaiac to immunochemical comparisons<br />

13/35 (37.1)<br />

22/32 (68.8)<br />

0.011<br />

(0.0001–5.98)<br />

0.46<br />

(0.231 – 0.631)<br />

7.88<br />

(3.09 – 21.47)<br />

HemeSelect<br />

77.1<br />

HO<br />

30.0<br />

–<br />

0.33<br />

(–0.166 – 0.832)<br />

5.0<br />

(0.344 – 72.78)<br />

Fecatwin<br />

83.3<br />

–<br />

–0.107<br />

(–0.317 – 0.103)<br />

0.57<br />

(0.186 – 1.745)<br />

HemeSelect<br />

68.8<br />

HO<br />

50.0<br />

HO Sensa<br />

79.4<br />

1/11 (9.1) a<br />

11/11 (100)<br />

3/6 (50.0)<br />

5/6 (83.3)<br />

27/34 (79.4)<br />

22/32 (68.8)<br />

HO<br />

HemeSelect<br />

HO<br />

HemeSelect<br />

HO<br />

FecaTwin<br />

HO Sensa<br />

HemeSelect<br />

HO vs HemeSelect<br />

Allison 1996<br />

Robinson 1996<br />

HO vs Fecatwin<br />

Armitage 1985 b<br />

HO Sensa vs HemeSelect<br />

Allison 1996<br />

0.009<br />

(0.00009–2.00)<br />

0.44<br />

(0.255 – 0.580)<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 27<br />

0.012<br />

(0.00009–5.99)<br />

–0.355<br />

(–1.428 – 0.701)<br />

–<br />

0.20<br />

(–0.354 – 0.754)<br />

–<br />

0.0<br />

(–0.518 – 0.518)<br />

–<br />

0.20<br />

(–0.354 – 0.754)<br />

13/35 (37.1)<br />

22/32 (68.8)<br />

1/11 (9.1) HO<br />

HemeSelect<br />

7.07<br />

33.6<br />

78.1<br />

(3.00 – 17.67)<br />

HO Sensa<br />

HemeSelect<br />

1.23<br />

77.5<br />

70.7<br />

(0.110 – 2.408)<br />

HO<br />

Flexsure<br />

2.67<br />

60.0<br />

80.0<br />

(0.158 – 45.143)<br />

HO Sensa<br />

Flexsure<br />

1.0<br />

42.9<br />

42.9<br />

(0.120 – 8.307)<br />

HO Sensa<br />

Colon Albumin<br />

2.67<br />

60.0<br />

80.0<br />

(0.158 – 45.143)<br />

a<br />

Sensitivity analyses<br />

HO vs HemeSelect<br />

HO<br />

Allison 1996<br />

HemeSelect<br />

HO<br />

Robinson 1996<br />

HemeSelect<br />

11/11 (100)<br />

HO<br />

3/5 (60.0)<br />

Rozen 1997<br />

HemeSelect<br />

4/5 (80.0)<br />

HO Sensa vs HemeSelect<br />

HO Sensa<br />

27/34 (79.4)<br />

Allison 1996<br />

HemeSelect<br />

22/32 (68.8)<br />

HO Sensa<br />

3/5 (60.0)<br />

Rozen 1997<br />

HemeSelect<br />

4/5 (80.0)<br />

HO vs Flexsure OBT<br />

HO<br />

3/5 (60.0)<br />

Rozen 1997<br />

Flexure<br />

4/5 (80.0)<br />

HO Sensa vs Flexsure OBT<br />

HO Sensa<br />

3/7 (42.9)<br />

Rozen 2000<br />

Flexsure<br />

3/7 (42.9)<br />

HO Sensa vs BM-Test Colon<br />

HO Sensa<br />

3/5 (60.0)<br />

Albumin<br />

Rozen 1995<br />

BM-Test Colon Albumin 4/5 (80.0)<br />

Abbreviations: CI, confidence interval; HO, Hem<strong>occult</strong>; HO Sensa, Hem<strong>occult</strong> Sensa; Fecatwin, Fecatwin Sensitive/Feca EIA; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; OR, odds ratio; RD, risk difference.<br />

aAssumes that the patient developing cancer during follow-up <strong>for</strong> adenoma initially tested negative with Hem<strong>occult</strong>; bData from Hardcastle et al 1986.


Table 5 Meta-analyses of specificity <strong>for</strong> carcinoma from FOBT head-to-head studies<br />

Heterogeneity<br />

(p)<br />

RD<br />

pooled results<br />

(95% CI)<br />

OR<br />

pooled results<br />

(95% CI)<br />

Immunochemical<br />

pooled<br />

specificity (%)<br />

Guaiac pooled<br />

specificity (%)<br />

Specificity n/N<br />

(True –ve/Total<br />

no Carcinoma)<br />

(%)<br />

FOBT<br />

Studies<br />

FOBT comparison<br />

Main guaiac to immunochemical comparisons<br />

0.033<br />

(0.0004–2.517)<br />

–0.043<br />

(–0.070 – –0.026)<br />

0.33<br />

(0.28 – 0.39)<br />

HemeSelect<br />

93.5<br />

HO<br />

97.8<br />

–<br />

–0.049<br />

(–0.066 – –0.032)<br />

0.35<br />

(0.24 – 0.51)<br />

Fecatwin<br />

92.2<br />

–<br />

0.077<br />

(0.068 – 0.086)<br />

2.58<br />

(2.29 – 2.91)<br />

HemeSelect<br />

94.4<br />

HO<br />

97.1<br />

HO Sensa<br />

86.7<br />

7845/8030 (97.7)<br />

7043/7461 (94.4)<br />

1462/1478 (98.9)<br />

1344/1478 (90.9)<br />

1261/1298 (97.2)<br />

1197/1298 (92.2)<br />

6824/7870 (86.7)<br />

7043/7461 (94.4)<br />

HO<br />

HemeSelect<br />

HO<br />

HemeSelect<br />

HO<br />

FecaTwin<br />

HO Sensa<br />

HemeSelect<br />

HO vs HemeSelect<br />

Allison 1996<br />

Robinson 1996<br />

HO vs Fecatwin<br />

Armitage 1985 a<br />

HO Sensa vs HemeSelect<br />

Allison 1996<br />

0.033<br />

(0.0004–2.517)<br />

–0.038<br />

(–0.045 – –0.031)<br />

28 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

0.342<br />

(0.029–12.48)<br />

0.054<br />

(0.024 – 0.121)<br />

–<br />

0.020<br />

(–0.008 – 0.048)<br />

–<br />

0.036<br />

(0.023 – 0.049)<br />

–<br />

0.057<br />

(0.016 – 0.099)<br />

Sensitivity analyses<br />

HO vs HemeSelect<br />

HO<br />

7845/8030 (97.7)<br />

Allison 1996<br />

HemeSelect<br />

7043/7461 (94.4)<br />

HO<br />

1462/1478 (98.9)<br />

HO<br />

HemeSelect<br />

0.36<br />

Robinson 1996<br />

HemeSelect<br />

1344/1478 (90.9)<br />

97.8<br />

94.0<br />

(0.31 – 0.42)<br />

HO<br />

377/398 (94.7)<br />

Rozen 1997<br />

HemeSelect<br />

390/398 (98.0)<br />

HO Sensa vs HemeSelect<br />

HO Sensa<br />

6824/7870 (86.7)<br />

Allison 1996<br />

HemeSelect<br />

7043/7461 (94.4) HO Sensa<br />

HemeSelect<br />

2.62<br />

HO Sensa<br />

366/398 (92.0)<br />

90.8<br />

96.3<br />

(2.33 – 2.95)<br />

Rozen 1997<br />

HemeSelect<br />

390/398 (98.0)<br />

HO vs Flexsure OBT<br />

HO<br />

377/398 (94.7)<br />

HO<br />

Flexsure<br />

1.64<br />

Rozen 1997<br />

Flexure<br />

385/398 (96.7)<br />

94.7<br />

96.7<br />

(0.81 – 3.34)<br />

HO Sensa vs Flexsure OBT<br />

HO Sensa<br />

1332/1403 (94.9) HO Sensa<br />

Flexsure<br />

3.51<br />

Rozen 2000<br />

Flexsure<br />

1382/1403 (98.5<br />

94.9<br />

98.5<br />

(2.14 – 5.74)<br />

HO Sensa vs BM-Test Colon<br />

HO Sensa<br />

437/522 (83.7)<br />

HO Sensa<br />

Colon Albumin<br />

1.65<br />

Albumin<br />

Rozen 1995<br />

BM-Test Colon Albumin 467/522 (89.5)<br />

83.7<br />

89.5<br />

(1.15 – 2.38)<br />

Abbreviations: CI, confidence interval; HO, Hem<strong>occult</strong>; HO Sensa, Hem<strong>occult</strong> Sensa; Fecatwin, Fecatwin Sensitive/Feca EIA; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; OR, odds ratio; RD, risk difference.<br />

aData from Hardcastle et al 1986.


Relative true positive rate and relative false positive rate<br />

Screening tests are usually compared in a sequential “within-subject” design, with only<br />

positive test results being verified with the reference standard. The actual measure of<br />

TPR and FPR will depend on the disease prevalence within the study <strong>population</strong>.<br />

However, the relative TPR and the FPR (ie, the ratio of one test’s TPR or FPR to the<br />

other) provide a useful and unbiased measure of comparative FOBT per<strong>for</strong>mance.<br />

Table 6 and Table 7 present meta-analyses of relative TPR and relative FPR data from<br />

the head-to-head studies of FOBTs.<br />

Table 6 shows that the TPR of Hem<strong>occult</strong> to reverse-passive haemagglutination<br />

(RPHA) tests (HemeSelect and Immudia-HemSp pooled) <strong>for</strong> detecting carcinoma was<br />

significantly higher <strong>for</strong> HemeSelect (relative TPR 0.59; 95% CI: 0.39, 0.88). The TPR was<br />

not significantly different when Hem<strong>occult</strong> was compared to Fecatwin Sensitive/Feca<br />

EIA (relative TPR 0.60; 95% CI: 0.14, 2.51) or when Hem<strong>occult</strong> Sensa was compared<br />

with HemeSelect (relative TPR 1.07; 95% CI: 0.70, 1.62).<br />

Table 7 shows that the FPR of Hem<strong>occult</strong> was superior to RPHA (relative FPR 0.53;<br />

95% CI: 0.33, 0.87). Similarly, the FPR of Hem<strong>occult</strong> was superior to Fecatwin<br />

Sensitive/Feca EIA (relative FPR 0.35; 95% CI: 0.24, 0.51). The FPR of HemeSelect was<br />

superior to Hem<strong>occult</strong> Sensa (relative FPR 2.23; 95% CI:1.14, 4.37). There was<br />

significant heterogeneity in the meta-analyses <strong>for</strong> Hem<strong>occult</strong> versus RPHA and<br />

Hem<strong>occult</strong> Sensa versus HemeSelect; however, this may well be due to the small number<br />

of included studies.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 29


Table 6 Meta-analyses of relative TPR <strong>for</strong> carcinoma from main FOBT head-to-head studies<br />

Heterogeneity<br />

p<br />

Pooled relative TPR<br />

guaiac:immunochemical<br />

Guaiac relative TPR<br />

True positives<br />

n<br />

FOBT<br />

Studies<br />

FOBT comparison<br />

p (95% CI)<br />

13<br />

HO<br />

Allison 1996 a<br />

0.59<br />

HO vs RPHA (pooled<br />

HemeSelect and Immudia)<br />

22<br />

HemeSelect<br />

8<br />

HO<br />

Iwase 1992<br />

0.50<br />

0.35<br />

0.59 (0.39–0.88)<br />

FEM<br />

16<br />

Immudia d<br />

16<br />

HO<br />

Petrelli 1994 b<br />

0.76<br />

21<br />

HemeSelect<br />

1<br />

HO<br />

Robinson 1995 c<br />

0.11<br />

9<br />

HemeSelect<br />

3<br />

HO<br />

Armitage 1985<br />

HO vs Fecatwin<br />

30 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

-<br />

0.60 (0.14–2.51)<br />

0.60<br />

5<br />

Fecatwin<br />

27<br />

HO Sensa<br />

Allison 1996 a<br />

HO Sensa vs HemeSelect<br />

1.23<br />

0.48<br />

1.07 (0.70–1.62)<br />

FEM<br />

22<br />

HemeSelect<br />

19<br />

HO Sensa<br />

Petrelli 1994 b<br />

0.90<br />

21<br />

HemeSelect<br />

Abbreviations: CI, confidence interval; HO, Hem<strong>occult</strong>; HO Sensa, Hem<strong>occult</strong> Sensa; Fecatwin, Fecatwin Sensitive/Feca EIA; FEM, fixed effects method; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; RPHA, reverse-passive<br />

haemagglutination; TPR, true positive rate.<br />

aThe total number of tests per<strong>for</strong>med differs slightly between test type, as both test were not per<strong>for</strong>med in 100% of patients. To simplify the analyses, it was assumed that the reason <strong>for</strong> any test not being done was independent<br />

of test type and outcome, so that the missing data could be treated as ignorable. bTotal positives calculated from number of patients screened and percentage positivity rate as reported; c Data from Robinson et al (1996) are not<br />

used due to slight discrepancies between the publications and a degree of uncertainty surrounding the data. dOver 3 days.


Table 7 Meta-analyses of relative FPR <strong>for</strong> carcinoma from main FOBT head-to-head studies<br />

Heterogeneity<br />

p<br />

Pooled relative FPR<br />

guaiac:immunochemical<br />

Guaiac relative FPR<br />

False positives<br />

n<br />

FOBT<br />

Studies<br />

FOBT comparison<br />

p (95% CI)<br />

185<br />

HO<br />

Allison 1996 a<br />

0.44<br />

HO vs RPHA (pooled<br />

HemeSelect and Immudia)<br />

418<br />

HemeSelect<br />

402<br />

HO<br />

Iwase 1992<br />

1.07<br />

< 0.0001<br />

0.53 (0.33–0.87)<br />

REM<br />

375<br />

Immudia d<br />

440<br />

HO<br />

Petrelli 1994 b<br />

1.06<br />

417<br />

HemeSelect<br />

16<br />

HO<br />

Robinson 1995 c<br />

0.12<br />

136<br />

HemeSelect<br />

37<br />

HO<br />

Armitage 1985<br />

HO vs Fecatwin<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 31<br />

-<br />

0.35 (0.24–0.51)<br />

0.35<br />

106<br />

Fecatwin<br />

1046<br />

HO Sensa<br />

Allison 1996 a<br />

HO Sensa vs HemeSelect<br />

2.50<br />

0.006<br />

2.23 (1.14–4.37)<br />

REM<br />

418<br />

HemeSelect<br />

830<br />

HO Sensa<br />

Petrelli 1994 b<br />

1.99<br />

417<br />

HemeSelect<br />

Abbreviations: CI, confidence interval; HO, Hem<strong>occult</strong>; HO Sensa, Hem<strong>occult</strong> Sensa; Fecatwin, Fecatwin Sensitive/Feca EIA; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; FPR, false positive rate; REM, random effects method; RPHA,<br />

reverse-passive haemagglutination.<br />

aThe total number of tests per<strong>for</strong>med differs slightly between test type, as both test were not per<strong>for</strong>med in 100% of patients. To simplify the analyses, it was assumed that the reason <strong>for</strong> any test not being done was independent<br />

of test type and outcome, so that the missing data could be treated as ignorable. bTotal positives calculated from number of patients screened and percentage positivity rate as reported; c Data from Robinson 96 is not used due to<br />

slight discrepancies between the publications and a degree of uncertainty surrounding the data. dOver 3 days.


Summary of diagnostic per<strong>for</strong>mance of FOBTs<br />

Three head-to-head studies estimated the sensitivity and specificity of various FOBTs <strong>for</strong><br />

CRC detection using the interval cancer rate as a proxy <strong>for</strong> the FNR. This allowed the<br />

calculation of the relative sensitivities and specificities <strong>for</strong> each of the FOBTs included in<br />

these studies. The relative TPR and the relative FPR (ie, the ratio of one test’s TPR or<br />

FPR to the other) provide a useful and unbiased measure of comparative FOBT<br />

per<strong>for</strong>mance (Chock et al 1997).<br />

On the basis of two of these studies, HemeSelect was found to be significantly more<br />

sensitive than Hem<strong>occult</strong> (77.1% vs 30.0%, respectively; RD 0.46; 95% CI: 0.231, 0.631).<br />

In contrast, the Hem<strong>occult</strong> test was significantly more specific than HemeSelect (97.8%<br />

vs 93.5%, respectively; RD –0.043; 95% CI: –0.070, –0.026). There<strong>for</strong>e, HemeSelect is<br />

more effective at detecting patients with CRC but is not as effective at excluding<br />

participants without the disease. The relative TPR <strong>for</strong> detecting carcinoma was<br />

significantly in favour of RPHA rather than Hem<strong>occult</strong> (relative TPR 0.59; 95% CI: 0.39,<br />

0.88), whereas the FPR of Hem<strong>occult</strong> was lower than <strong>for</strong> RPHA (relative FPR 0.53; 95%<br />

CI: 0.33, 0.87). There<strong>for</strong>e, these measures are not adequate to determine which of these<br />

tests is more accurate. The practical impact of this trade-off has been investigated via the<br />

use of an economic model in the cost-effectiveness section of this assessment report.<br />

The sensitivities of Hem<strong>occult</strong> and Fecatwin Sensitive/Feca EIA were not significantly<br />

different (50% vs 83.3%, respectively; RD 0.33; 95% CI: –0.166, 0.832). However,<br />

Hem<strong>occult</strong> was significantly more specific than Fecatwin Sensitive/Feca EIA (97.1% vs<br />

92.2%, respectively; RD –0.049; 95% CI: –0.066, –0.032). This means that at this time<br />

there is no statistically significant evidence to suggest that Hem<strong>occult</strong> or Fecatwin<br />

Sensitive/Feca EIA are better at identifying patients with CRC. However, there is<br />

statistically significant evidence to suggest that Hem<strong>occult</strong> is superior at excluding<br />

participants without the disease. The TPR <strong>for</strong> detecting carcinoma was not significantly<br />

different when Hem<strong>occult</strong> was compared to Fecatwin Sensitive/Feca EIA (relative TPR<br />

0.60; 95% CI: 0.14, 2.51). However, the FPR of Hem<strong>occult</strong> was lower than Fecatwin<br />

Sensitive/Feca EIA (relative FPR 0.35; 95% CI: 0.24, 0.51). This would imply that, based<br />

on the available data, Hem<strong>occult</strong> is statistically the more accurate test. It should be noted<br />

that the subjects in this study were not required to adhere to the dietary restrictions<br />

generally recommended with the use of a guaiac FOBT.<br />

The sensitivity of Hem<strong>occult</strong> Sensa and HemeSelect were not significantly different<br />

(79.4% vs 68.8%, respectively; RD –0.107; 95% CI: –0.317, 0.103). In contrast,<br />

HemeSelect proved to be significantly more specific than Hem<strong>occult</strong> Sensa (94.4% vs<br />

86.7%, respectively; RD 0.077; 95% CI: 0.068, 0.086). This means that at this time there<br />

is no statistically significant evidence to suggest that Hem<strong>occult</strong> Sensa or HemeSelect are<br />

better at identifying patients with CRC. However, there is statistically significant evidence<br />

to suggest that HemeSelect is superior at excluding participants without the disease. The<br />

TPR was not significantly different when Hem<strong>occult</strong> Sensa was compared with<br />

HemeSelect (relative TPR 1.07; 95% CI: 0.70, 1.62). However, the FPR of HemeSelect<br />

was lower than that of Hem<strong>occult</strong> Sensa (relative FPR 2.23; 95% CI: 1.14, 4.37). This<br />

would imply that, based on these data, HemeSelect is statistically the more accurate test.<br />

Only three head-to-head studies estimated the sensitivity and specificity of various<br />

FOBTs <strong>for</strong> adenoma detection. These studies were all conducted in <strong>screening</strong><br />

<strong>population</strong>s with a high proportion of participants who were at higher risk of disease, by<br />

subjecting all participants to endoscopic examination. In addition, these studies were<br />

32


subject to verification bias due to the large time lapse between per<strong>for</strong>ming the<br />

endoscopic examination and the FOBT in many patients with a negative FOBT result.<br />

There<strong>for</strong>e, there were no data of a suitable quality to assess relative accuracy of the tests<br />

<strong>for</strong> the detection of adenomas.<br />

The relative cost-effectiveness of pairs of FOBTs based on the sensitivity and specificity<br />

data as derived from these head-to-head studies of FOBTs is explored below. There<strong>for</strong>e,<br />

the available data have determined the economic comparisons which can be per<strong>for</strong>med.<br />

The relative cost-effectiveness of Hem<strong>occult</strong> versus HemeSelect; Hem<strong>occult</strong> Sensa<br />

versus HemeSelect; and Hem<strong>occult</strong> versus Fecatwin Sensitive/Feca EIA are examined<br />

utilising data from the main analyses of sensitivity and specificity. Due to concerns of<br />

external validity, where the only data available <strong>for</strong> a head-to-head comparison are from a<br />

<strong>population</strong> at increased risk (Hem<strong>occult</strong> versus FlexSure OBT, Hem<strong>occult</strong> Sensa versus<br />

FlexSure OBT, and Hem<strong>occult</strong> Sensa versus BM-Test Colon Albumin comparisons), this<br />

is not extended to an economic analysis.<br />

What are the economic considerations?<br />

The cost-effectiveness of <strong>population</strong>-based FOBT <strong>screening</strong> <strong>for</strong> CRC has been<br />

established repeatedly in the <strong>health</strong> economics literature (Wagner et al 1996; Khandker et<br />

al 2000; Salkeld et al 1996; Bolin et al 1999; O’Leary et al <strong>2004</strong>). Given this, the economic<br />

model developed <strong>for</strong> this assessment seeks to determine the most promising FOBT<br />

within a <strong>screening</strong> setting. To this end, the relative per<strong>for</strong>mance and cost-effectiveness of<br />

a variety of FOBTs in a number of settings is determined. In particular, key<br />

characteristics of the <strong>screening</strong> program – such as the age of the <strong>screening</strong> <strong>population</strong><br />

and the interval between <strong>screening</strong> rounds – are altered to determine which option yields<br />

the best value <strong>for</strong> money <strong>for</strong> the Australian <strong>health</strong>care system.<br />

Assessment of value <strong>for</strong> money of <strong>population</strong> <strong>health</strong> <strong>screening</strong> <strong>for</strong><br />

colorectal cancer<br />

Why an economic model is required<br />

Economic models of <strong>health</strong> care interventions have a range of advantages and limitations<br />

compared with observational studies and other prospectively designed data collection<br />

experiments. The costs of an FOBT <strong>screening</strong> program <strong>for</strong> the detection of CRC are<br />

expected to be partially offset by longer-term benefits in terms of both costs and <strong>health</strong><br />

outcomes. Consequently, an economic model is required to quantify these offsets<br />

accurately. In addition, in order to compare alternative FOBTs, an economic model is<br />

required to measure the cost-effectiveness of various <strong>screening</strong> options under different<br />

conditions.<br />

A decision-analytic model with Markov processes was developed to estimate the<br />

downstream costs associated with an FOBT <strong>screening</strong> program. A Monte Carlo<br />

simulation is used as it allows individuals to be tracked through time. Consequently, the<br />

Markov model allows individuals to cycle through the <strong>screening</strong> program until the<br />

detection of colorectal cancer, ineligibility <strong>for</strong> <strong>screening</strong> or death from any cause. It<br />

should be noted, however, that the reliability of any model is dependent upon the<br />

uncertainty surrounding the clinical data and the key model assumptions.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 33


Key assumptions used in the economic model<br />

• The economic model is designed to assess the relative cost-effectiveness of various<br />

FOBTs used to detect CRC in a <strong>screening</strong> <strong>population</strong>. Consequently, head-to-head<br />

comparisons between FOBTs are made. It is not intended that decision-makers<br />

compare tests not appearing directly alongside one another.<br />

• The natural progression of CRC is defined such that patients with undiagnosed<br />

cancer pass sequentially through each clinical stage after an appropriate time interval.<br />

Patients do not ‘skip’ stages in the sequential sequence.<br />

• Individuals with diagnosed CRC are at greatest risk of mortality <strong>for</strong> the first 5 years<br />

following diagnosis. These individuals are assumed to revert to a ‘normal’ lifeexpectancy<br />

if they survive their first 5 years with diagnosed cancer.<br />

• The likelihood of CRC detection with FOBT, or procedures used later in the<br />

diagnostic work-up, is assumed to be independent of the stage of cancer.<br />

• It is assumed that the presence or absence of dietary restrictions is the key<br />

determinant of differences in participation rates <strong>for</strong> the two classes of FOBTs<br />

(guaiac and immunochemical).<br />

• Participation in a <strong>screening</strong> program is assumed to remain constant despite<br />

arguments that the increased community awareness associated with a general<br />

<strong>population</strong> <strong>health</strong> <strong>screening</strong> program may lead to improved participation.<br />

• Participation in a <strong>screening</strong> program is assumed to be dependent on past behaviour.<br />

That is, future patterns of participation are affected by behaviour in previous rounds.<br />

• Individuals with increased risk of CRC, through family history <strong>for</strong> example, are not<br />

treated as part of the <strong>screening</strong> <strong>population</strong>. Similarly, symptomatic patients are<br />

treated as external to the general <strong>screening</strong> <strong>population</strong>. These individuals are treated<br />

as external due to the different diagnostic pathway they are subject to.<br />

• All individuals in the economic model are treated as having a maximum lifeexpectancy<br />

of 100 years.<br />

• For simplicity, indirect/societal costs are not included in the economic model, as<br />

they are likely to be negligible in the context of the total cost of the <strong>screening</strong><br />

program. This assumption is consistent with other models in the literature.<br />

• The results of the economic model are presented in terms of the incremental cost<br />

per life-year gained with one <strong>screening</strong> program over another.<br />

• A discount rate of 5% per annum was applied to all costs and <strong>health</strong> outcomes.<br />

34


The literature on FOBT <strong>screening</strong> <strong>for</strong> colorectal cancer<br />

FOBT has been repeatedly shown to be a cost-effective <strong>screening</strong> option <strong>for</strong> CRC in the<br />

general <strong>population</strong> (Wagner et al 1996; Khandker et al 2000; Salkeld et al 1996; Bolin et<br />

al 1999; O’Leary et al <strong>2004</strong>). Although studies in the literature utilise a wide range of<br />

assumptions and <strong>screening</strong> <strong>population</strong>s, the majority of studies favour FOBT <strong>screening</strong><br />

over no <strong>screening</strong>.<br />

There are a limited number of studies that have been conducted within the Australian<br />

setting. Again, these tend to show FOBT to be a cost-effective method in <strong>screening</strong> <strong>for</strong><br />

CRC. In these Australian studies, Salkeld et al (1996) estimate the cost-effectiveness of<br />

annual FOBT <strong>screening</strong> as $24,660 per life-year gained, Bolin et al (1999) estimate the<br />

cost-effectiveness of annual FOBT <strong>screening</strong> as $36,132 per life-year gained and O’Leary<br />

et al (<strong>2004</strong>) give an estimate of $46,900. Note that while the methodologies and<br />

assumptions used in these studies vary greatly, the results provide good evidence of<br />

FOBT being a cost-effective method of CRC <strong>screening</strong>.<br />

Studies to date have focused on the incremental cost-effectiveness of FOBT <strong>screening</strong><br />

versus no <strong>screening</strong> and a comparison between the available FOBTs has not yet been<br />

conducted.<br />

Patient <strong>population</strong> used in the economic model<br />

Individuals from the general <strong>population</strong> are invited to be screened <strong>for</strong> CRC under the<br />

proposed program. They are selected based on age alone and other risk factors do not<br />

play a role. Individuals identified as being at heightened risk (possibly due to a family<br />

history of CRC), are not included in the <strong>screening</strong> program and will instead undergo a<br />

separate diagnostic work-up. The age of the eligible <strong>population</strong> is treated as exogenous in<br />

the economic model. Annual and biennial <strong>screening</strong> intervals are tested and the most<br />

cost-effective alternative is recommended based on the results.<br />

Structure of the economic model<br />

A Markov model with Monte Carlo simulations is used to follow a sample of 20,000<br />

patients through the FOBT <strong>screening</strong> program from the first invitation <strong>for</strong> <strong>screening</strong><br />

until either death or ineligibility due to age. The timing of the first invitation to <strong>screening</strong><br />

is dependent on the characteristics of the <strong>screening</strong> program and is to be determined by<br />

the results of the economic model, as is the interval between <strong>screening</strong> rounds.<br />

The model consists of two modelled components:<br />

• the natural history of CRC (see Figure 2), which determines how patients progress<br />

through various <strong>health</strong> states; and<br />

• the <strong>screening</strong> pathway (see Figure 3), which aims to intercept the natural<br />

progression of CRC in such a way that treatment can be initiated and life-expectancy<br />

increased.<br />

Individuals in the model are advanced through the Markov process in 3-monthly cycles<br />

until death.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 35


Figure 2 Simplified natural history of CRC used in the economic model<br />

Abbreviations: CRC, colorectal cancer.<br />

As illustrated in Figure 2, individuals eligible <strong>for</strong> <strong>screening</strong> are classified as either ‘well’,<br />

having an adenoma(s), or having CRC. The model categorises cancer according to<br />

Dukes’ stages, defining the <strong>health</strong> state as either undiagnosed or diagnosed. Undiagnosed<br />

CRC progresses through the various stages according to defined distributions, whereas<br />

diagnosis of CRC is treated as a terminal endpoint in the model. Upon diagnosis (which<br />

can occur via the <strong>screening</strong> pathway or after an individual presents with symptoms<br />

outside of the <strong>screening</strong> program), the life-expectancy of the individual is calculated<br />

according to the five-year survival rate of the relevant cancer stage and, beyond that, the<br />

‘normal’ life-expectancy of an individual of that age. Individuals who are either ‘well’ or<br />

have premalignant adenoma(s) are only able to exit the model via non-CRC death.<br />

The <strong>screening</strong> pathway used in the economic model is illustrated in Figure 3. This is a<br />

simplified representation outlining the key stages of the <strong>screening</strong> process. It illustrates<br />

how important issues such as participation and diagnostic follow-up play a role in the<br />

final patient outcomes. As Figure 3 shows, the eligible <strong>population</strong> is invited to be<br />

screened, at which stage individuals choose either to participate or not to participate. If<br />

an individual chooses to participate, he/she can receive either a positive or negative<br />

FOBT result. This result is dependent upon the diagnostic accuracy of the particular test<br />

used. If an individual tests negative, he/she is invited <strong>for</strong> re-<strong>screening</strong> after an<br />

appropriate interval (to be determined by the economic model). If a positive result is<br />

given, the individual receives a diagnostic work-up and will either receive treatment if<br />

CRC is confirmed or be re-invited into the <strong>screening</strong> program after five years.<br />

36<br />

Well<br />

Adenoma<br />

Other death<br />

Undiagnosed<br />

Dukes’ A<br />

Undiagnosed<br />

Dukes’ B<br />

Undiagnosed<br />

Dukes’ C<br />

Undiagnosed<br />

Dukes’ D<br />

Diagnosed<br />

Dukes’ A<br />

Diagnosed<br />

Dukes’ B<br />

Diagnosed<br />

Dukes’ C<br />

Diagnosed<br />

Dukes’ D<br />

Other death<br />

CRC death


Figure 3 Screening pathway used in the economic model<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 37


Variables used in the economic model<br />

Clinical variables<br />

As individuals from the eligible patient <strong>population</strong> enter the economic model, they are<br />

assigned to one of seven <strong>health</strong> states:<br />

• patient is free of disease (well)<br />

• patient has non-progressive adenoma(s) present<br />

• patient has progressive adenoma(s) present (though undetected)<br />

• patient has Dukes’ A stage CRC (which has not been diagnosed)<br />

• patient has Dukes’ B stage CRC (which has not been diagnosed)<br />

• patient has Dukes’ C stage CRC (which has not been diagnosed)<br />

• patient has Dukes’ D stage CRC (which has not been diagnosed).<br />

The likelihood of individuals entering the economic model in each of these states has<br />

been calculated from CRC and adenoma prevalence data reported in an Australian<br />

flexible sigmoidoscopy <strong>screening</strong> trial (Collett et al 2000). These data have been<br />

distributed by Dukes’ stage according to data from a colonoscopy <strong>screening</strong> study<br />

(Lieberman et al 2000). These calculations are presented in Table 8–Table 10.<br />

Table 8 Disease prevalence in the eligible <strong>screening</strong> <strong>population</strong> at program entry.<br />

Row Health state Prevalence Reference <strong>for</strong> data source<br />

A Adenoma(s) present 0.1351 Collett et al (2000)<br />

B CRC present 0.0054 Collett et al (2000)<br />

Abbreviations: CRC, colorectal cancer.<br />

Table 9 Distribution of CRC by Dukes’ stage classification in the eligible <strong>screening</strong> <strong>population</strong> at<br />

program entry<br />

Row Stage of CRC Prevalence Reference <strong>for</strong> data source<br />

A Dukes’ stage A 0.5000 Lieberman et al (2000)<br />

B Dukes’ stage B 0.2330 Lieberman et al (2000)<br />

C Dukes’ stage C 0.2000 Lieberman et al (2000)<br />

D Dukes’ stage D 0.0670 Lieberman et al (2000)<br />

Abbreviations: CRC, colorectal cancer.<br />

38 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Table 10 Prevalence used <strong>for</strong> individuals entering the economic model<br />

Row Stage of CRC Prevalence Reference<br />

A Adenoma(s) present 0.1351 A = Table 8 Row A<br />

B Dukes’ stage A 0.0027 B = Table 8 Row B × Table 9 Row A<br />

C Dukes’ stage B 0.0013 C = Table 8 Row B × Table 9 Row B<br />

D Dukes’ stage C 0.0011 D = Table 8 Row B × Table 9 Row C<br />

E Dukes’ stage D 0.0004 E = Table 8 Row B × Table 9 Row D<br />

F Patient is ‘well’ 0.8595 F = 1 – (A + B + C + D + E)<br />

Abbreviations: CRC, colorectal cancer.<br />

Following entry into the economic model, individuals are assigned a probability of<br />

progressing to other <strong>health</strong> states. Prior to detection of either adenomas or CRC,<br />

progression must occur sequentially from ‘well’ to adenoma(s) present to each of the<br />

cancer stages in order of severity. Patients cannot skip stages of CRC prior to diagnosis<br />

and treatment.<br />

The probability of progression is calculated on a quarterly basis.<br />

The probability of an individual developing an adenoma is derived by tracking the<br />

incidence of CRC back to the likely time of adenoma development and adjusting <strong>for</strong> the<br />

proportion of adenomas that undergo malignant trans<strong>for</strong>mation (estimated by Stryker et<br />

al 1987, to be 24%). A duration of 20 years between the onset of a progressive adenoma<br />

and clinical diagnosis was assumed (Loeve et al 2000). Age-adjusted incidence figures<br />

were used to trace CRC incidence back to the time of onset of the adenoma. These rates<br />

are presented in Table 11–Table 13.<br />

Table 11 Incidence of CRC<br />

Row Age (years) Incidence per 100,000<br />

individualsa Incidence<br />

A 50–54 56.9 0.0006<br />

B 55–59 107.2 0.0011<br />

C 60–64 166.5 0.0017<br />

D 65–69 239.7 0.0024<br />

E 70–74 316.4 0.0032<br />

F 75–79 378.3 0.0038<br />

G 80–84 410.0 0.0041<br />

H 85+ 454.1 0.0045<br />

Abbreviations: CRC, colorectal cancer.<br />

aGender-adjusted rates.<br />

Source: Cancer In Australia 1999, Table 20 (AIHW).<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 39


Table 12 Incidence of progressive adenomas, age-adjusted<br />

Row Age (years) Incidence Reference<br />

A 50–54 0.0032 A = Table 11 Row E<br />

B 55–59 0.0038 B = Table 11 Row F<br />

C 60–64 0.0041 C = Table 11 Row G<br />

D 65–69 0.0045 C = Table 11 Row H<br />

E 70–74 0.0045 D = Table 11 Row H<br />

F 75–79 0.0045 E = Table 11 Row H<br />

G 80–84 0.0045 F = Table 11 Row H<br />

H 85+ 0.0045 G = Table 11 Row H<br />

Abbreviations: CRC, colorectal cancer.<br />

Table 13 Incidence of all adenomas, age-adjusted<br />

Row Age (years) Incidence Reference<br />

A 50–54 0.0132 A = Table 12 Row A/24% a<br />

B 55–59 0.0158 B = Table 12 Row B/24% a<br />

C 60–64 0.0171 C = Table 12 Row C/24% a<br />

D 65–69 0.0189 C = Table 12 Row D/24% a<br />

E 70–74 0.0189 D = Table 12 Row E/24% a<br />

F 75–79 0.0189 E = Table 12 Row F/24% a<br />

G 80–84 0.0189 F = Table 12 Row G/24% a<br />

H 85+ 0.0189 G = Table 12 Row H/24% a<br />

Abbreviations: CRC, colorectal cancer.<br />

a24% assumed from Stryker et al (1987).<br />

Once patients have progressed from undiagnosed adenoma(s) to Dukes’ stage A CRC,<br />

the probability of moving through the stages of cancer is governed by an exponential<br />

distribution using a unique mean duration <strong>for</strong> each of the cancer stages. These<br />

distributions are taken from the literature whereby the estimated average duration of<br />

undiagnosed CRC is given. The durations used in establishing these distributions are<br />

presented in Table 14.<br />

Table 14 Mean duration of undiagnosed cancerous and precancerous <strong>health</strong> states prior to<br />

progression<br />

Health state Mean duration (years) Reference<br />

Adenoma 16.4a Loeve et al (1999)<br />

Undiagnosed Dukes’ stage A 2.0 Loeve et al (1999)<br />

Undiagnosed Dukes’ stage B 1.0 Loeve et al (1999)<br />

Undiagnosed Dukes’ stage C 1.5 Loeve et al (1999)<br />

Undiagnosed Dukes’ stage D 0.8 Loeve et al (1999)<br />

aDuration between onset of a progressive adenoma and clinical detection (20 years) minus average duration of undiagnosed CRC (assumed<br />

from Loeve et al (1999) to be 3.6 years.<br />

Alternatively, individuals may circumvent <strong>screening</strong> altogether if they present with<br />

symptoms. The probability of this occurring is based on current incidence data, since in<br />

the absence of a general <strong>screening</strong> program individuals are generally only diagnosed once<br />

they become symptomatic. It is assumed that individuals will not present with symptoms<br />

at the adenoma stage, only once they have developed CRC.<br />

40 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Table 15 and Table 16 outline the calculations <strong>for</strong> the likelihood of individuals<br />

presenting by stage of CRC. The proportion of individuals with CRC whose cancer is<br />

classified as either Dukes’ stage C or D is taken from NSW stage of spread data as<br />

regional cancer and metastatic cancer, as reported in Bell et al (1996). The proportion of<br />

those whose cancer is classified as either Dukes’ A or B is calculated using the<br />

in<strong>for</strong>mation in Bell et al (1996) <strong>for</strong> localised CRC and distributed into these Dukes’<br />

stages according to the distribution from the no-<strong>screening</strong> arm of the Nottingham FOBT<br />

trial (Mapp et al 1999).<br />

Table 15 Distributional spread of known CRC, by stage<br />

Row Dukes’ stage Proportion of patients with Dukes stage<br />

cancer at diagnosis<br />

Reference<br />

A Dukes’ stage A 0.0910 Bell et al (1996) and Mapp et al (1999)<br />

B Dukes’ stage B 0.2680 Bell et al (1996) and Mapp et al (1999)<br />

C Dukes’ stage C 0.4880 Bell et al (1996)<br />

D Dukes’ stage D 0.1540 Bell et al (1996)<br />

Abbreviations: CRC, colorectal cancer.<br />

Table 16 Probability of individual with CRC presenting as symptomatic, by stage<br />

Row Dukes’<br />

stage<br />

Probability of<br />

presenting as<br />

symptomatic<br />

A Dukes’ A 0.0910 Table 15 Row A<br />

Reference<br />

B Dukes’ B 0.2948 B = Table 15 Row B/(1 – Table 15 Row A)<br />

C Dukes’ C 0.7613 C = Table 15 Row C/(1 – Table 15 Row A – Table 15 Row B)<br />

D Dukes’ D 1.0000 D = Table 15 Row D/(1 – Table 15 Row A – Table 15 Row B – Table 15 Row C)<br />

Abbreviations: CRC, colorectal cancer.<br />

Note that detection of CRC with FOBT <strong>screening</strong> reduces the number of individuals<br />

presenting with symptoms. For example, in a <strong>population</strong> not subject to <strong>screening</strong>, the<br />

probabilities appearing in Table 16 apply to the entire <strong>population</strong>. Alternatively,<br />

detection via FOBT participation lowers the number of individuals these probabilities<br />

apply to. Consequently, the total number of individuals presenting with symptoms will be<br />

lower in the <strong>screening</strong> arms of the economic model.<br />

Once an individual has been diagnosed with CRC, either through FOBT <strong>screening</strong> or<br />

otherwise, he/she is treated as remaining in that <strong>health</strong> state until death. At this point,<br />

the life-expectancy of the individual is calculated and CRC treatment costs accrue. In<br />

calculating the life-expectancy of these individuals, five-year survival rates from a<br />

Victorian Cancer registry study of CRC surgical patients have been used (McLeish et al<br />

2002), with exponential interpolation adopted over the five-year period. Whilst survival<br />

rates may currently be higher than those reported in this study, different survival rates<br />

will not affect the outcome of the relative per<strong>for</strong>mance of the different FOBTs. These<br />

five-year survival rates are presented in Table 17, whereas the probabilities of death per<br />

quarter are presented in Table 18.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 41


Table 17 Five-year survival rates <strong>for</strong> CRC, by stage<br />

Row Cancer stage Five-year survival rate Reference<br />

A Dukes’ stage A 0.8913 McLeish et al (2002)<br />

B Dukes’ stage B 0.7948 McLeish et al (2002)<br />

C Dukes’ stage C 0.3478 McLeish et al (2002)<br />

D Dukes’ stage D 0.0000 McLeish et al (2002)<br />

Abbreviations: CRC, colorectal cancer.<br />

Table 18 Quarterly probability of death from diagnosed CRC<br />

Row Cancer stage Probability of death used in the<br />

economic model<br />

A Dukes’ stage A 0.0057 Calculated a<br />

B Dukes’ stage B 0.0114 Calculated a<br />

C Dukes’ stage C 0.0514 Calculated a<br />

D Dukes’ stage D 0.2057 Calculated a<br />

Abbreviations: CRC, colorectal cancer.<br />

aProbability = 1 – (1 – 5-year survival rate) [1/(5 years/4)] = 1 – (5-year mortality rate) [1/(5 years/4)]<br />

Reference<br />

If an individual with diagnosed CRC survives the initial five-year period, he/she is<br />

assumed to continue with a ‘normal’ life-expectancy. This assumption is based on expert<br />

opinion and supported by a cursory comparison of 5- and 10-year survival data in the<br />

Victorian Cancer registry study of CRC surgical patients (McLeish et al 2002).<br />

‘Normal’ life-expectancy takes the <strong>for</strong>m of non-CRC death. The probabilities used are<br />

calculated by removing the mortality from CRC (Australian Institute of Health and<br />

Welfare (AIHW) 2002) from Australian life tables. Exponential interpolation is used to<br />

convert this to a quarterly probability of death.<br />

Diagnostic variables<br />

Due to the aim of the economic model – to find the most promising test and <strong>screening</strong><br />

setting <strong>for</strong> a <strong>population</strong> <strong>health</strong> based <strong>screening</strong> program <strong>for</strong> CRC – the cost-effectiveness<br />

of a number of FOBTs is investigated using the model. The systematic review and metaanalyses<br />

in the effectiveness section of this report has been used as the source <strong>for</strong> the<br />

diagnostic accuracy of the various FOBTs. These data have been derived from head-tohead<br />

comparisons of guaiac versus immunochemical FOBTs. Specifically, the following<br />

comparisons are made:<br />

• Hem<strong>occult</strong> (guaiac) versus HemeSelect (immunochemical)<br />

• Hem<strong>occult</strong> Sensa (guaiac) versus HemeSelect (immunochemical)<br />

• Hem<strong>occult</strong> (guaiac) versus Fecatwin sensitive/Feca EIA (two-tiered combination of<br />

guaiac and immunochemical).<br />

Due to the lack of an estimate of accuracy <strong>for</strong> the detection of adenomas, and the lack of<br />

dietary restrictions imposed during the study, the last comparison of Hem<strong>occult</strong> versus<br />

Fecatwin sensitive/Feca EIA appears in a sensitivity analysis contained in Appendix J,<br />

rather than in the main body of this report.<br />

42 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Relevant sensitivity and specificity rates have been determined from the effectiveness<br />

section of this report. Since estimates of sensitivity and specificity will be dependent<br />

upon the <strong>population</strong> prevalence and spectrum of disease in the <strong>population</strong> in each study,<br />

different values <strong>for</strong> the same FOBT appear in different head-to-head comparisons. As a<br />

result, it is inappropriate to draw conclusions outside of individual head-to-head<br />

comparisons. Conclusions can only be drawn regarding the relative cost-effectiveness of<br />

one test over another within the head-to-head comparisons. Sensitivity and specificity<br />

data <strong>for</strong> adenoma have been derived from a study conducted in a <strong>screening</strong> <strong>population</strong><br />

with a high proportion of participants who were at higher risk of disease, by subjecting<br />

all participants to the reference standard (Rozen et al 1997).<br />

It is important to note that the external validity of the data available <strong>for</strong> estimating the<br />

sensitivity and specificity of adenoma detection is of concern. The only studies providing<br />

estimates of test accuracy <strong>for</strong> the detection of adenoma were those conducted in a<br />

<strong>population</strong> with a high proportion of subjects at increased risk. In addition, in these<br />

studies the endoscopic examination had been per<strong>for</strong>med up to several years previously in<br />

a proportion of the patients <strong>testing</strong> negative with FOBT. There<strong>for</strong>e these studies were<br />

subject to significant verification bias.<br />

The probability of a positive FOBT result in each of the comparisons is presented in<br />

Table 19– Table 21. Note that it is assumed that the probability of detection of CRC is<br />

the same regardless of the stage to which the cancer has progressed. This assumption is<br />

necessary due to the limitations of the available data.<br />

The data presented in Table 19 <strong>for</strong> Hem<strong>occult</strong> are varied in a sensitivity analysis based<br />

on a summary receiver-operating characteristic curve derived from data presented in the<br />

efficacy section of this report (see Appendix K).<br />

Table 19 Probability of positive result using Hem<strong>occult</strong> and HemeSelect<br />

True <strong>health</strong> state<br />

Probability of<br />

detection using<br />

Hem<strong>occult</strong> HemeSelect<br />

Reference<br />

Well a 0.022 0.065 1 – specificity (0.978 <strong>for</strong> Hem<strong>occult</strong> and 0.935 <strong>for</strong> HemeSelect) b<br />

Adenoma present 0.190 0.190 Rozen (1997)<br />

Dukes’ stage A CRC 0.300 0.771 Table 4 Sensitivity<br />

Dukes’ stage B CRC 0.300 0.771 Table 4 Sensitivity<br />

Dukes’ stage C CRC 0.300 0.771 Table 4 Sensitivity<br />

Dukes’ stage D CRC 0.300 0.771 Table 4 Sensitivity<br />

Abbreviations: CRC, colorectal cancer.<br />

aFalse positive rate <strong>for</strong> CRC/adenoma.<br />

bSpecificity taken from systematic assessment of effectiveness in this document (Table 5).<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 43


Table 20 Probability of positive result using Hem<strong>occult</strong> Sensa and HemeSelect<br />

True <strong>health</strong> state<br />

Probability of<br />

detection using<br />

Hem<strong>occult</strong> HemeSelect<br />

Sensa<br />

Reference<br />

Wella 0.133 0.056 1 – specificity (0.867 <strong>for</strong> Hem<strong>occult</strong> Sensa and 0.944 <strong>for</strong><br />

HemeSelect) b<br />

Adenoma present 0.238 0.190 Rozen (1997)<br />

Dukes’ stage A CRC 0.794 0.688 Table 4 Sensitivity<br />

Dukes’ stage B CRC 0.794 0.688 Table 4 Sensitivity<br />

Dukes’ stage C CRC 0.794 0.688 Table 4 Sensitivity<br />

Dukes’ stage D CRC 0.794 0.688 Table 4 Sensitivity<br />

Abbreviations: CRC, colorectal cancer.<br />

aFalse positive rate <strong>for</strong> CRC/adenoma.<br />

bSpecificity taken from systematic assessment of effectiveness in this document (Table 5).<br />

Table 21 Probability of positive result using Hem<strong>occult</strong> and Fecatwin Sensitive/Feca EIA<br />

True <strong>health</strong> state<br />

Probability of<br />

detection using<br />

Hem<strong>occult</strong> Fecatwin<br />

Reference<br />

Well a 0.029 0.078 1 – specificity (0.971 <strong>for</strong> Hem<strong>occult</strong> and 0.922 <strong>for</strong> Fecatwin) b<br />

Adenoma present 0.000 0.000 Assumption c<br />

Dukes’ stage A CRC 0.500 0.833 Table 4 Sensitivity<br />

Dukes’ stage B CRC 0.500 0.833 Table 4 Sensitivity<br />

Dukes’ stage C CRC 0.500 0.833 Table 4 Sensitivity<br />

Dukes’ stage D CRC 0.500 0.833 Table 4 Sensitivity<br />

Abbreviations: CRC, colorectal cancer; Fecatwin, Fecatwin Sensitive/Feca EIA.<br />

aFalse positive rate <strong>for</strong> CRC/adenoma.<br />

bSpecificity taken from systematic assessment of effectiveness in this document (Table 5).<br />

cAssumption necessary due to absence of data. By assuming that no adenomas are detected with either test, the relative cost-effectiveness is<br />

undistorted.<br />

Note that it has been necessary to assume a zero detection rate <strong>for</strong> adenomas in the<br />

Hem<strong>occult</strong> versus Fecatwin Sensitive/Feca EIA analysis. This is due to an absence of<br />

data. Consequently, the results of this head-to-head analysis appear in a sensitivity<br />

analysis in Appendix J, as there are limitations placed on the effectiveness captured.<br />

Diagnostic follow-up is assumed to take the <strong>for</strong>m of complete colonoscopy. Within the<br />

pilot, complete colonoscopy has accounted <strong>for</strong> almost 80% of all diagnostic follow-up to<br />

date. Consequently, the diagnostic accuracy of complete colonoscopy is used in the<br />

economic model as the default <strong>for</strong> diagnostic follow-up. For the purposes of the<br />

economic model, the differences between the accuracy of complete colonoscopy and<br />

other methods is unlikely to have a significant effect.<br />

Complete colonoscopy is assumed to have a 95% sensitivity (assumption based on<br />

NHMRC (1999)). The sensitivity of colonoscopy in the detection of adenomas is 85%<br />

(Loeve et al 1999). Complete colonoscopy is assumed to have a 100% specificity. This<br />

assumption eliminates the need to model outcomes resulting from false positive results.<br />

These probabilities are presented in Table 22.<br />

44 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Table 22 Detection rates associated with colonoscopy<br />

Parameter Value Reference<br />

Sensitivity associated with detecting CRC 0.95 Assumption based on NHMRC (1999)<br />

Specificity associated with detecting CRC 1 Assumption<br />

Sensitivity associated with detecting adenoma 0.85 Loeve et al (1999)<br />

Specificity associated with detecting adenoma 1 Assumption<br />

Abbreviations: CRC, colorectal cancer; NHMRC, National Health and Medical Research Council.<br />

Colonoscopy is associated with a 0.0041 probability of an adverse event (Winawer et al<br />

1997). This rate incorporates the risk of per<strong>for</strong>ation, severe haemorrhage and death.<br />

Cost variables<br />

The cost of establishing a <strong>screening</strong> program has been omitted on the basis that it is a<br />

negligible cost on a per invited individual basis. The cost will be distributed across all<br />

individuals invited to <strong>screening</strong> over the entire life of the <strong>screening</strong> program. This will<br />

result in a negligible individual cost which will have virtually no effect on the relative<br />

cost-effectiveness of the different FOBTs.<br />

The primary cost associated with a <strong>population</strong> based <strong>screening</strong> program is the cost of the<br />

FOBT itself. This cost can, however, be broken down into its components – that is the<br />

cost that applies to all invited members of the eligible patient <strong>population</strong> and the cost<br />

that applies only to those who participate in <strong>screening</strong>. This separation is necessary as the<br />

<strong>screening</strong> program is designed such that a FOBT kit is mailed to each individual invited<br />

to participate in the <strong>screening</strong> program. If an individual chooses to participate, they<br />

return the completed kit at the expense of the program <strong>for</strong> pathology <strong>testing</strong>. The latter<br />

cost, there<strong>for</strong>e, only arises in the case of participation. The relevant costs are illustrated in<br />

Table 23 and Table 24.<br />

Table 23 Cost of FOBT per invited individual<br />

Test Cost<br />

Hem<strong>occult</strong> $2.30<br />

Hem<strong>occult</strong> Sensa $3.77<br />

HemeSelect $6.30 a<br />

Fecatwin Sensitive/Feca EIA $5.35b Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

aHemeSelect is currently unavailable. This estimate is based on the cost of another immunochemical FOBT that uses a similar methodology.<br />

bCost calculation based on prices <strong>for</strong> other guaiac and immunochemical tests. All participants receive the guaiac component and 25% receive<br />

the immunochemical component and no dietary restriction. Calculation is based on the expected cost of these two tests. Transport cost was<br />

not included in cost provided and has been assumed to be $5.50.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 45


Table 24 Cost of FOBT per participant<br />

Test Cost<br />

Hem<strong>occult</strong> $23.93<br />

Hem<strong>occult</strong> Sensa $16.50<br />

HemeSelect $24.20 a<br />

Fecatwin Sensitive/Feca EIA $21.18 a<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

aTransport cost was not included in cost provided and has been assumed to be $5.50.<br />

Further diagnostic work-up, assumed to take the <strong>for</strong>m of complete colonoscopy, is<br />

priced depending on whether abnormalities are detected or not. In the case of<br />

abnormalities present at colonoscopy, participants will receive a polypectomy, thereby<br />

increasing the cost of colonoscopy. Cost increases are due to the polypectomy itself and<br />

the pathology <strong>testing</strong> associated with it.<br />

All colonoscopies, however, attract a cost <strong>for</strong> the service, <strong>for</strong> anaesthetic and <strong>for</strong> other<br />

associated costs (including day theatre charges, pharmaceuticals, etc.). The costs of<br />

colonoscopy with and without polyp removal are presented in Table 25 and Table 26.<br />

Associated costs have been sourced from Weller et al (1995) and converted to 2002<br />

prices using the AIHW total <strong>health</strong> price index (AIHW 2003).<br />

Table 25 Cost of colonoscopy without polyp removal<br />

Procedure Cost Reference<br />

Colonoscopy $277.80 MBSa Item no. 32090<br />

Anaesthetic $99.00 MBSa Item no 20810 and 23023<br />

Associated costs $606.35 Weller et al (1995) and AIHW total <strong>health</strong> price index (2003)<br />

Total cost $983.15<br />

aMedicare Benefits Schedule, November 2003.<br />

Table 26 Cost of colonoscopy with polyp removal<br />

Procedure Cost Reference<br />

Colonoscopy + polypectomy $389.90 MBS a Item no. 32093<br />

Pathology $109.30 MBS a Item no 72823 and 73903<br />

Anaesthetic $99.00 MBS a Item no 20810 and 23023<br />

Associated costs $606.35 Weller et al (1995) and AIHW total <strong>health</strong> price index (2003)<br />

Total cost $1204.55<br />

aMedicare Benefits Schedule, November 2003.<br />

The pathology cost quoted in Table 26 assumes that more than one polyp is found and<br />

removed. This assumption is biased against the <strong>screening</strong> program.<br />

Colonoscopy is additionally associated with a risk of an adverse event. As per<strong>for</strong>ation is<br />

the most common of these, <strong>for</strong> costing purposes it is assumed that all adverse events<br />

take this <strong>for</strong>m. This cost has been calculated by updating the cost found in Salkeld et al<br />

(1996) to 2002 prices. The 2002 cost of treating a per<strong>for</strong>ation is assumed to be<br />

$15,010.46.<br />

46 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Following diagnosis of CRC, patients accrue treatment costs. Lifetime costs have been<br />

used in the economic model, capturing all costs of treatment following diagnosis and<br />

disease progression. These costs are illustrated in Table 27.<br />

Table 27 Lifetime CRC treatment costs, from time of diagnosis<br />

Cancer stage Cost (2002 prices) Reference<br />

Dukes’ stage A CRC $15,808.18 O’Leary et al (<strong>2004</strong>) and AIHW total <strong>health</strong> price index (2003)<br />

Dukes’ stage B CRC $30,757.73 O’Leary et al (<strong>2004</strong>) and AIHW total <strong>health</strong> price index (2003)<br />

Dukes’ stage C CRC $23,757.67 O’Leary et al (<strong>2004</strong>) and AIHW total <strong>health</strong> price index (2003)<br />

Dukes’ stage D CRC $5775.07 O’Leary et al (<strong>2004</strong>) and AIHW total <strong>health</strong> price index (2003)<br />

Abbreviations: CRC, colorectal cancer.<br />

Characteristics of the <strong>screening</strong> program<br />

Participation is an integral part of any <strong>screening</strong> program and is a driver of overall costeffectiveness<br />

in comparison with no <strong>screening</strong>, by affecting the way costs are distributed<br />

amongst those invited to be screened. In the economic model, the difference in<br />

participation rates between tests comes into play at the point of the initial invitation <strong>for</strong><br />

<strong>screening</strong> and in later rounds when an individual’s participation is treated as a function of<br />

previous participation.<br />

Participation is assumed to vary between tests, based on the type of test being used.<br />

Guaiac tests, <strong>for</strong> instance, have a lower rate of participation due to the dietary restrictions<br />

that are imposed. Participation rates have been calculated from a number of sources,<br />

including preliminary estimates from the pilot. These calculations appear in detail in<br />

Appendix I. The participation rates used in the economic model are presented in<br />

Table 28. Note that there is scope <strong>for</strong> individuals who have previously not participated<br />

in <strong>screening</strong> to begin participating after the first round. This feature is based on expert<br />

advice indicating that the number of individuals picked up in later rounds is roughly<br />

equivalent to the number lost.<br />

Table 28 Compliance with FOBT <strong>screening</strong> rates used in the economic model<br />

Patient <strong>population</strong> Probability of<br />

compliance with<br />

guaiac FOBT<br />

Eligible <strong>population</strong> invited <strong>for</strong> <strong>screening</strong><br />

in first round<br />

Those who have complied with initial<br />

invitation <strong>for</strong> <strong>screening</strong> (later rounds)<br />

Those who have not complied with<br />

previous invitations <strong>for</strong> <strong>screening</strong> (later<br />

rounds)<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

Probability of<br />

compliance with<br />

immunochemical<br />

FOBT<br />

Reference<br />

0.3483 0.4540 Calculated – see Appendix I<br />

0.8618 0.8842 Calculated – see Appendix I<br />

0.0739 0.0963 Calculated – see Appendix I<br />

Additionally, compliance is an issue <strong>for</strong> individuals recommended <strong>for</strong> diagnostic followup.<br />

The model assumes 92.4% of individuals comply with follow-up, once<br />

recommended. This figure is taken from available in<strong>for</strong>mation from the pilot. It captures<br />

compliance with colonoscopy (complete or incomplete but adequate) or DCBE.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 47


In addition to participation and compliance issues, the relative cost-effectiveness of the<br />

various FOBTs is likely to be linked to both the age of the <strong>population</strong> eligible <strong>for</strong><br />

<strong>screening</strong> and the interval between <strong>screening</strong> rounds <strong>for</strong> individuals.<br />

In the base-case scenario of biennial <strong>screening</strong>, it is assumed that the eligible <strong>population</strong><br />

consists of all individuals aged 55–74 years who are not considered to be at increased risk<br />

of CRC. Other ages considered in the sensitivity analyses include:<br />

• individuals aged between 50 and 74 years<br />

• individuals aged between 55 and 79 years.<br />

Although this is not an exhaustive list of options, it does give an impression of the<br />

sensitivity of the relative cost-effectiveness of the FOBTs investigated to variations in the<br />

age of eligibility <strong>for</strong> <strong>screening</strong>.<br />

The <strong>screening</strong> interval is assumed to be 2 years. In a sensitivity analysis, annual <strong>screening</strong><br />

is also tested as an alternative. These two options are the most common in the literature.<br />

Results of the economic model<br />

The purpose of the economic model is to compare the various FOBTs within the<br />

context of head-to-head studies. As previously outlined, comparisons cannot be made<br />

outside these studies. The discussion of the results of the model will reflect this.<br />

The base-case scenario assumes biennial <strong>screening</strong> in individuals aged 55–74 years. The<br />

effect of varying these assumptions (frequency of <strong>screening</strong> and age of <strong>screening</strong><br />

<strong>population</strong>), as well as other assumptions such as participation rates, is also examined and<br />

discussed briefly in relation to relative cost-effectiveness.<br />

It is worth noting that the model accurately replicates true disease incidence and<br />

prevalence. This is highlighted in a number of the figures below (Figure 5 on page 50<br />

and Figure 8 on page 53, <strong>for</strong> example), which illustrate the number of individuals in<br />

each stage of disease from a sampled pool of 20,000.<br />

In addition, it should be noted that there are substantial differences between the<br />

economic model analyses results and published FOBT RCT data. In particular, the<br />

model follows a cohort of individuals entering the model at the lowest eligibility age<br />

throughout the life of the program, whereas trial data will present findings <strong>for</strong> rounds of<br />

<strong>screening</strong> of individuals distributed across a spectrum of ages in the <strong>population</strong> entering<br />

a <strong>screening</strong> program.<br />

Hem<strong>occult</strong> Sensa versus HemeSelect<br />

Costs<br />

The Hem<strong>occult</strong> Sensa test is associated with higher costs overall, an average of $1532 per<br />

individual invited to <strong>screening</strong> versus $1428 <strong>for</strong> the HemeSelect test. This difference is<br />

largely driven by more frequent use of colonoscopy in the Hem<strong>occult</strong> Sensa arm of the<br />

economic model, leading to a higher average cost of colonoscopy per eligible individual.<br />

48 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


In the case of Hem<strong>occult</strong> Sensa, colonoscopy accounted <strong>for</strong> 22% of all costs, whereas they<br />

accounted <strong>for</strong> 13% of HemeSelect total costs. A graphic breakdown of costs appears below<br />

in Figure 4.<br />

Cost per individual invited to <strong>screening</strong><br />

$1,600<br />

$1,400<br />

$1,200<br />

$1,000<br />

$800<br />

$600<br />

$400<br />

$200<br />

$0<br />

Hem<strong>occult</strong> Sensa HemeSelect<br />

Type of FOBT<br />

Figure 4 Component costs of FOBT <strong>screening</strong> – Hem<strong>occult</strong> Sensa versus HemeSelect<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

Cancer treatment<br />

Colonoscopy<br />

FOBT<br />

For both FOBTs, the cost of treating cancer is lower than under the current no <strong>screening</strong><br />

scenario. The model estimates that the average cost of cancer treatment on a per eligible<br />

participant basis is $1187. This cost is reduced to $1084 and $1101 in the Hem<strong>occult</strong><br />

Sensa and HemeSelect arms of the model, respectively. The lower cost is the result of<br />

earlier detection of cancer, thereby reducing lifetime treatment costs, and detection of<br />

adenomas, which can be treated be<strong>for</strong>e the chance of malignant trans<strong>for</strong>mation.<br />

Outcomes<br />

As expected from the relatively similar participation rates, invitees in the Hem<strong>occult</strong><br />

Sensa and HemeSelect arms of the economic model received, on average, a similar<br />

number of <strong>screening</strong> tests, that is 5.6 and 6.4, respectively. The difference is driven by the<br />

differing participation rates. There<strong>for</strong>e, a greater difference in participation rates would<br />

create a greater difference between tests. Although both tests yield higher lifeexpectancies<br />

<strong>for</strong> individuals, the difference between Hem<strong>occult</strong> Sensa and HemeSelect is<br />

marginal (Table 29).<br />

Table 29 Life-expectancy from the beginning of the economic model<br />

Patient group Life-expectancy from the<br />

beginning of the model (years)<br />

Patients screened with Hem<strong>occult</strong> Sensa 15.497<br />

Incremental life-expectancy from<br />

the beginning of the model (years)<br />

Patients screened with HemeSelect 15.492 0.005<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 49


The positive difference in life-expectancy between <strong>screening</strong> and no <strong>screening</strong> is largely<br />

driven by earlier detection. In the case of Hem<strong>occult</strong> Sensa versus HemeSelect, the most<br />

important driver of this difference is the difference in the number of adenomas detected.<br />

This is, in turn, driven by the different sensitivity rates <strong>for</strong> adenoma detection associated<br />

with each FOBT. Improved detection of Dukes’ A stage CRC also plays a role in<br />

improving the life-expectancy of individuals in the <strong>screening</strong> arms of the economic model.<br />

The pattern of detection is highlighted in Figure 5.<br />

Number detected from 20,000 invited individuals<br />

600<br />

500<br />

400<br />

300<br />

200<br />

100<br />

0<br />

Total<br />

cancers<br />

detected by<br />

<strong>screening</strong><br />

Adenoma Dukes' A Dukes' B Dukes' C Dukes' D<br />

Hem<strong>occult</strong> Sensa<br />

HemeSelect<br />

Figure 5 Pattern of neoplasm detection in the Hem<strong>occult</strong> Sensa versus HemeSelect <strong>screening</strong> analysis<br />

Note that adenoma detection illustrated includes detection of all adenomas, whether premalignant or otherwise.<br />

Earlier detection of cancer improves the chances of survival, due to better chances of<br />

successful treatment. For example, 89% of patients detected with Dukes’ A survive five<br />

years of disease be<strong>for</strong>e moving on to have a ‘normal’ life-expectancy. This contrasts with<br />

a zero chance of survival <strong>for</strong> patients detected with Dukes’ D stage CRC (Table 17).<br />

FOBT <strong>screening</strong> consistently detects a greater number of cancers in patients at early<br />

stages. This is to be expected, given that CRC will be detected be<strong>for</strong>e patients are<br />

symptomatic. Importantly, <strong>screening</strong> detects substantially more cancers at Dukes’ stage<br />

A, thereby increasing the life-expectancy of the <strong>screening</strong> <strong>population</strong>.<br />

As shown in Figure 5, <strong>screening</strong> with either FOBT also detects a significant number of<br />

adenomas, premalignant or otherwise. Through the detection of adenomas, and their<br />

consequent removal, the incidence of CRC can be markedly reduced. In turn, this<br />

increases the life-expectancy of those undergoing <strong>screening</strong> and reduces potential costs.<br />

Detection of adenomas through <strong>screening</strong> with either FOBT will prevent the<br />

development of CRC in approximately 2.2–2.5% of the invited <strong>population</strong>. It is<br />

important to note, however, that this figure is correlated with the participation rate of the<br />

invited <strong>population</strong>. By assumption, patients not undergoing <strong>screening</strong> do not have<br />

adenomas detected, as they remain asymptomatic. There<strong>for</strong>e, a greater difference in<br />

participation rates would create a greater difference between tests.<br />

50 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Of the two tests, Hem<strong>occult</strong> Sensa detects more adenomas and early-stage cancers via<br />

<strong>screening</strong> than HemeSelect, due to better test sensitivity <strong>for</strong> the detection of CRC.<br />

The distribution of stage of neoplasm detection <strong>for</strong> this analysis is presented in Figure 6.<br />

Number detected from 20,000 invited individuals<br />

3000<br />

2500<br />

2000<br />

1500<br />

1000<br />

500<br />

0<br />

Total adenomas<br />

detected by<br />

<strong>screening</strong><br />

Total cancers<br />

detected by<br />

<strong>screening</strong><br />

Total cancers<br />

presenting<br />

symptomatically<br />

Total cancers<br />

undetected at death<br />

Figure 6 Method of neoplasm detection in the Hem<strong>occult</strong> Sensa versus HemeSelect analysis<br />

Hem<strong>occult</strong> Sensa<br />

HemeSelect<br />

No <strong>screening</strong><br />

Cost-effectiveness<br />

The available evidence indicates that each of the tests offer acceptable value-<strong>for</strong>-money<br />

against the current practice of no <strong>screening</strong>. Incremental costs are low, with a net benefit<br />

in terms of life-expectancy. Additionally, adverse events arising from the <strong>screening</strong><br />

program are minimal.<br />

More importantly, based on the current data, Hem<strong>occult</strong> Sensa appears to offer benefits<br />

over HemeSelect at what would normally be considered a reasonable price premium.<br />

Specifically, Hem<strong>occult</strong> Sensa is associated with an extra cost of $21,534 per additional<br />

life-year gained (Table 30). Hem<strong>occult</strong> Sensa thus appears to offer acceptable value-<strong>for</strong>money<br />

over HemeSelect in a general <strong>population</strong> biennial FOBT <strong>screening</strong> program <strong>for</strong><br />

CRC in patients aged between 55 and 74 years. However, the difference in lifeexpectancy<br />

between the tests is marginal.<br />

Table 30 Relative cost-effectiveness of Hem<strong>occult</strong> Sensa over HemeSelect<br />

Hem<strong>occult</strong> Sensa HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1531.88 $1427.84 $104.04<br />

Life-expectancy from the beginning of the model 15.497 15.492 0.005<br />

Incremental costs per life-year gained $21,533.72<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 51


Hem<strong>occult</strong> versus HemeSelect<br />

Costs<br />

In a head-to-head comparison of Hem<strong>occult</strong> and HemeSelect, HemeSelect is the more<br />

costly alternative. HemeSelect is associated with an average cost of $1449 per individual<br />

invited to <strong>screening</strong>. This compares to an average of $1334 in the Hem<strong>occult</strong> arm of the<br />

economic model. The average cost of colonoscopy per eligible individual (driven by the<br />

frequency of colonoscopy) is a key driver in this differential, with colonoscopy costs<br />

representing 14% of costs in the HemeSelect arm and 7% in the Hem<strong>occult</strong> arm.<br />

Cancer treatment costs, however, are lower in the HemeSelect arm, due to the greater<br />

sensitivity <strong>for</strong> detection. Earlier detection of CRC and adenomas means that lifetime<br />

treatment costs are reduced. These costs are lower in both of the <strong>screening</strong> arms than<br />

they are in the non-<strong>screening</strong> arm, <strong>for</strong> the same reason.<br />

A breakdown of the estimated costs associated with the use of Hem<strong>occult</strong> versus<br />

HemeSelect in a <strong>population</strong> <strong>health</strong> <strong>screening</strong> program appears in Figure 7.<br />

Cost per individual invited to <strong>screening</strong><br />

$1,600<br />

$1,400<br />

$1,200<br />

$1,000<br />

$800<br />

$600<br />

$400<br />

$200<br />

$0<br />

Hem<strong>occult</strong> HemeSelect<br />

Type of FOBT<br />

Figure 7 Component costs of FOBT <strong>screening</strong> – Hem<strong>occult</strong> versus HemeSelect<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

Cancer treatment<br />

Colonoscopy<br />

FOBT<br />

Outcomes<br />

The difference in the average number of FOBTs administered per invited individual is<br />

minimal, being 6.3 in the HemeSelect arm and 6.1 in the Hem<strong>occult</strong> arm. As in the<br />

previous comparison (Hem<strong>occult</strong> Sensa versus HemeSelect), this small difference is<br />

driven by the different participation rates.<br />

The difference in the life-expectancy of individuals invited to <strong>screening</strong> between these<br />

two FOBTs is marginal, despite notable differences in the sensitivity and specificity of<br />

the tests <strong>for</strong> CRC detection (Table 31). This is largely the result of the relatively low<br />

prevalence combined with repeated <strong>screening</strong>, equivalent sensitivity <strong>for</strong> adenoma<br />

detection and relatively similar participation rates.<br />

52 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Table 31 Life-expectancy from the beginning of the economic model<br />

Patient group Life-expectancy from the beginning<br />

of the model (years)<br />

Patients screened with Hem<strong>occult</strong> 15.465<br />

Incremental life-expectancy from the<br />

beginning of the model (years)<br />

Patients screened with HemeSelect 15.501 –0.036<br />

Screening with either Hem<strong>occult</strong> or HemeSelect consistently detects more abnormalities<br />

than no <strong>screening</strong>. Importantly, there are improved rates of detection of both adenomas<br />

and early-stage CRC. Detection of adenomas has the potential to reduce the incidence of<br />

CRC, while early detection of cancer both improves life-expectancy and reduces lifetime<br />

treatment cancer costs. The distribution of stage of neoplasm detection <strong>for</strong> this analysis<br />

is presented in Figure 8.<br />

Number detected from 20,000 invited individuals<br />

500<br />

450<br />

400<br />

350<br />

300<br />

250<br />

200<br />

150<br />

100<br />

50<br />

0<br />

Total cancers<br />

detected by<br />

<strong>screening</strong><br />

Adenoma Dukes' A Dukes' B Dukes' C Dukes' D<br />

Figure 8 Pattern of detection in the Hem<strong>occult</strong> versus HemeSelect <strong>screening</strong> analysis<br />

Note that adenoma detection illustrated includes detection of all adenomas, whether premalignant or otherwise.<br />

Hem<strong>occult</strong><br />

HemeSelect<br />

Of the two tests, HemeSelect appears more efficient in detecting CRC, though the<br />

difference in life-expectancy is marginal. Note that with equal sensitivity rates <strong>for</strong><br />

adenoma detection between the FOBTs, the life-expectancy difference is driven by the<br />

difference in CRC sensitivity rates alone. This substantial difference causes a larger<br />

incremental difference in life-expectancy than occurs in the Hem<strong>occult</strong> Sensa versus<br />

HemeSelect analysis.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 53


Number detected from 20,000 invited individuals<br />

3000<br />

2500<br />

2000<br />

1500<br />

1000<br />

500<br />

0<br />

Total adenomas<br />

detected by<br />

<strong>screening</strong><br />

Total cancers<br />

detected by<br />

<strong>screening</strong><br />

Total cancers<br />

presenting<br />

symptomatically<br />

Total cancers<br />

undetected at death<br />

Figure 9 Method of neoplasm detection in the Hem<strong>occult</strong> versus HemeSelect analysis<br />

Hem<strong>occult</strong><br />

HemeSelect<br />

No <strong>screening</strong><br />

Figure 9 gives a breakdown of the method of detection in the Hem<strong>occult</strong> versus<br />

HemeSelect analysis, illustrating that fewer cancers are left either undetected or are<br />

detected symptomatically in the <strong>screening</strong> arms compared with the no <strong>screening</strong> arm.<br />

Cost-effectiveness<br />

Superficially, HemeSelect appears to offer a <strong>health</strong> benefit at what would normally be<br />

considered an acceptable price premium. Specifically, the relative cost-effectiveness of<br />

HemeSelect over Hem<strong>occult</strong> is $3172 per life-year gained (Table 32).<br />

Table 32 Relative cost-effectiveness of Hem<strong>occult</strong> over HemeSelect<br />

Hem<strong>occult</strong> HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1333.65 $1448.89 –$115.24<br />

Life-expectancy from the beginning of the model (years) 15.465 15.501 –0.036<br />

Incremental costs per life-year gained $3172.10<br />

Sensitivity analyses<br />

Several sensitivity analyses have been conducted around key variables in order to<br />

illustrate how sensitive the relative cost-effectiveness measures are to change. Variables<br />

have been selected based on either their uncertainty or the scope of decision-makers to<br />

alter them through the design of the <strong>screening</strong> program. A justification <strong>for</strong> each scenario<br />

is presented below.<br />

54 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


• The frequency of <strong>screening</strong> has been increased from the base-case assumption of<br />

biennial <strong>testing</strong> to annual <strong>testing</strong> in order to improve understanding of how the<br />

frequency may affect costs and outcomes.<br />

• To better understand how the age of the eligible <strong>population</strong> may affect the costeffectiveness<br />

of various FOBTs, the age of the screened <strong>population</strong> has been<br />

adjusted by five years in either direction. Through the correlation between risk and<br />

age, it is expected that this will affect cost-effectiveness.<br />

• An analysis has been conducted varying the sensitivity and specificity of CRC<br />

detection using the Hem<strong>occult</strong> test in the Hem<strong>occult</strong> versus HemeSelect analysis.<br />

This has been conducted in response to the uncertainty surrounding the estimates<br />

used. The results of this analysis appear in Appendix K.<br />

Hem<strong>occult</strong> Sensa versus HemeSelect<br />

The effect of increasing the frequency of <strong>screening</strong> to annual <strong>testing</strong><br />

Increasing the frequency of <strong>screening</strong>, as would be expected, increases costs.<br />

Additionally, there are obvious improvements in life-expectancy. In the comparison of<br />

Hem<strong>occult</strong> Sensa versus HemeSelect, increasing the frequency of <strong>screening</strong> to annual<br />

<strong>testing</strong> causes HemeSelect to dominate over Hem<strong>occult</strong> Sensa. That is, HemeSelect<br />

offers marginally better outcomes at a lower cost than Hem<strong>occult</strong> Sensa (Table 33).<br />

Increasing the frequency to annual <strong>screening</strong> is not a consistently cost-effective variation<br />

of the <strong>screening</strong> program by comparison with biennial <strong>screening</strong>. That is, life-expectancy<br />

does not increase to a greater degree than associated costs.<br />

Table 33 The effect of increasing the frequency of <strong>screening</strong> to annual <strong>testing</strong> on relative costeffectiveness:<br />

Hem<strong>occult</strong> Sensa versus HemeSelect<br />

Base-case analysis – biennial <strong>screening</strong><br />

Hem<strong>occult</strong> Sensa HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1531.88 $1427.84 $104.04<br />

Life-expectancy from the beginning of the model (years) 15.497 15.492 0.005<br />

Incremental costs per life-year gained $21,533.72<br />

Annual <strong>screening</strong><br />

Total cost associated with <strong>screening</strong> option $1719.30 $1558.90 $160.41<br />

Life-expectancy from the beginning of the model (years) 15.506 15.529 –0.023<br />

Incremental costs per life-year gained –$6908.59<br />

The effect of lowering the minimum eligible age to 50 years<br />

Lowering the minimum eligible age effectively increases the number of <strong>screening</strong> rounds<br />

an individual is eligible <strong>for</strong>. Consequently costs increase due to more intensive<br />

monitoring.<br />

Furthermore, increased exposure to <strong>screening</strong> improves the effectiveness of the program<br />

by improving the life-expectancy of the eligible <strong>population</strong>. Importantly <strong>for</strong> the relative<br />

cost-effectiveness, the incremental life-expectancy difference between the two tests<br />

increased. Consequently, the conclusion that Hem<strong>occult</strong> Sensa offers a marginal benefit<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 55


at a reasonable cost is accentuated (Table 34). Lowering the eligible <strong>screening</strong> age from<br />

55 to 50 years improves the cost-effectiveness of Hem<strong>occult</strong> Sensa over HemeSelect.<br />

Additionally, lowering the minimum eligible <strong>screening</strong> age to 50 years is more costeffective,<br />

compared with the base-case age of 55 years, <strong>for</strong> both FOBTs. That is, lifeexpectancy<br />

by comparison with no <strong>screening</strong> increases to a greater degree than the<br />

associated increases in costs when the minimum eligible age is reduced to 50 years.<br />

Table 34 The effect of lowering the eligible age to 50 years on relative cost-effectiveness: Hem<strong>occult</strong><br />

Sensa versus HemeSelect<br />

Base-case analysis – minimum eligible age = 55 years<br />

Hem<strong>occult</strong> Sensa HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1531.88 $1427.84 $104.04<br />

Life-expectancy from the beginning of the model (years) 15.497 15.492 0.005<br />

Incremental costs per life-year gained $21,533.72<br />

Minimum eligible age = 50 years<br />

Total cost associated with <strong>screening</strong> option $1677.62 $1521.14 $156.48<br />

Life-expectancy from the beginning of the model (years) 15.537 15.514 0.022<br />

Incremental costs per life-year gained $6973.76<br />

The effect of increasing the maximum eligible age to 80 years<br />

Altering the maximum age of the eligible patient <strong>population</strong> has very little effect on the<br />

Hem<strong>occult</strong> Sensa versus HemeSelect analysis (Table 35). Costs and outcomes <strong>for</strong><br />

Hem<strong>occult</strong> Sensa increase marginally. This result is expected, as adenomas detected at<br />

later ages have reduced chances of undergoing malignant trans<strong>for</strong>mation. Detection of<br />

adenomas drives the result in the base-case analysis to a certain degree.<br />

Table 35 The effect of increasing the eligible age to 80 years on relative cost-effectiveness: Hem<strong>occult</strong><br />

Sensa versus HemeSelect<br />

Base-case analysis – maximum eligible age = 75 years<br />

Hem<strong>occult</strong> Sensa HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1531.88 $1427.84 $104.04<br />

Life-expectancy from the beginning of the model (years) 15.497 15.492 0.005<br />

Incremental costs per life-year gained $21,533.72<br />

Maximum eligible age = 80 years<br />

Total cost associated with <strong>screening</strong> option $1557.13 $1449.54 $107.60<br />

Life-expectancy from the beginning of the model (years) 15.499 15.485 0.014<br />

Incremental costs per life-year gained $7740.19<br />

The conclusion of this analysis remains unchanged – though the relative costeffectiveness<br />

of Hem<strong>occult</strong> Sensa is improved due to the growth in the incremental<br />

benefit outstripping the growth in the incremental costs.<br />

56 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Additionally, <strong>for</strong> both FOBTs, increasing the maximum eligible <strong>screening</strong> age to 80 years<br />

is less cost-effective compared with the base-case age of 75 years. That is, changes in lifeexpectancy<br />

by comparison with no <strong>screening</strong> increase to a lesser degree than the<br />

associated increases in cost when the maximum age is increased to 80 years.<br />

Hem<strong>occult</strong> versus HemeSelect<br />

The effect of increasing the frequency of <strong>screening</strong> to annual <strong>testing</strong><br />

Increasing the frequency of <strong>screening</strong> had little effect overall in the comparison of<br />

Hem<strong>occult</strong> versus HemeSelect. Costs <strong>for</strong> both tests increased, along with life-expectancy,<br />

as one would expect from increased <strong>screening</strong>. The conclusion drawn, however, remains<br />

the same. Under the changed assumptions, HemeSelect still offers a greater lifeexpectancy<br />

benefit at a reasonable cost premium (Table 36), though the cost per lifeyear<br />

saved increases.<br />

Increasing the frequency of <strong>screening</strong> to annual <strong>testing</strong> is not a consistently cost-effective<br />

variation of the <strong>screening</strong> program by comparison with biennial <strong>screening</strong>. That is, lifeexpectancy<br />

does not increase to a greater degree than associated costs.<br />

Table 36 The effect of increasing the frequency of <strong>screening</strong> to annual <strong>testing</strong> on relative costeffectiveness:<br />

Hem<strong>occult</strong> versus HemeSelect<br />

Base-case analysis – biennial <strong>screening</strong><br />

Hem<strong>occult</strong> HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1333.65 $1448.89 –$115.24<br />

Life-expectancy from the beginning of the model<br />

(years)<br />

15.465 15.501 –0.036<br />

Incremental costs per life-year gained $3172.10<br />

Annual <strong>screening</strong><br />

Total cost associated with <strong>screening</strong> option $1391.24 $1586.53 –$195.30<br />

Life-expectancy from the beginning of the model (years) 15.494 15.535 –0.042<br />

Incremental costs per life-year gained $4677.36<br />

The effect of lowering the minimum eligible age to 50 years<br />

Lowering the minimum eligible age of the <strong>screening</strong> <strong>population</strong> effectively increases the<br />

intensity of <strong>screening</strong>. As expected, this is associated with an increase in the cost of<br />

<strong>screening</strong> on a per invitee basis. Additionally, the difference between the life-expectancy<br />

of individuals in the <strong>screening</strong> arms and the no-<strong>screening</strong> arm increased – though the<br />

life-expectancy associated with Hem<strong>occult</strong> surpassed that of HemeSelect.<br />

Consequently, the overall conclusion is reversed. By lowering the minimum eligible<br />

<strong>screening</strong> age, Hem<strong>occult</strong> offers a superior life-expectancy at a lower overall cost than<br />

HemeSelect (Table 37).<br />

Additionally, lowering the minimum eligible <strong>screening</strong> age to 50 years is more costeffective,<br />

compared with the base-case age of 55 years, <strong>for</strong> both FOBTs. That is, lifeexpectancy<br />

by comparison with no <strong>screening</strong> increases to a greater degree than the<br />

associated increases in costs when the minimum age is lowered to 50 years.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 57


Discussion<br />

Table 37 The effect of lowering the eligible age to 50 years on relative cost-effectiveness: Hem<strong>occult</strong><br />

versus HemeSelect<br />

Base-case analysis – minimum eligible age = 55 years<br />

Hem<strong>occult</strong> HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1333.65 $1448.89 –$115.24<br />

Life-expectancy from the beginning of model (years) 15.465 15.501 –0.036<br />

Incremental costs per life-year gained $3172.10<br />

Minimum eligible age = 50 years<br />

Total cost associated with <strong>screening</strong> option $1353.50 $1545.76 –$192.26<br />

Life-expectancy from the beginning of the model (years) 15.527 15.521 0.006<br />

Incremental costs per life-year gained –$32,405.24<br />

The effect of increasing the maximum eligible age to 80 years<br />

Increasing the maximum eligible <strong>screening</strong> age also has the effect of increasing costs due<br />

to more intensive <strong>screening</strong>, and the life-expectancy of the eligible <strong>population</strong> is similar.<br />

Again, however, the overall conclusion remains unchanged – HemeSelect offers greater<br />

life-expectancy at a reasonable level of relative cost-effectiveness (Table 38).<br />

Additionally, <strong>for</strong> both FOBTs, increasing the maximum eligible <strong>screening</strong> age to 80 years<br />

is less cost-effective compared with the base-case age of 75 years. That is, changes in lifeexpectancy<br />

by comparison with no <strong>screening</strong> increase to a lesser degree than the<br />

associated increases in cost when the maximum age is increased to 80 years.<br />

Table 38 The effect of increasing the eligible age to 80 years on relative cost-effectiveness: Hem<strong>occult</strong><br />

versus HemeSelect<br />

Base-case analysis – maximum eligible age = 75 years<br />

Hem<strong>occult</strong> HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1333.65 $1448.89 –$115.24<br />

Life-expectancy from the beginning of the model (years) 15.465 15.501 –0.036<br />

Incremental costs per life-year gained $3172.10<br />

Maximum eligible age = 80 years<br />

Total cost associated with <strong>screening</strong> option $1342.41 $1470.68 –$128.27<br />

Life-expectancy from the beginning of the model (years) 15.468 15.498 –0.030<br />

Incremental costs per life-year gained $4227.07<br />

The results of the analyses presented above make intuitive sense. On the basis of the<br />

available head-to-head comparison data, FOBT <strong>screening</strong> <strong>for</strong> CRC appears to be<br />

associated with a marked shift in the stage of diagnosis of CRC (Figure 5 and Figure 8).<br />

Most importantly, early detection of adenomas prior to malignant trans<strong>for</strong>mation allows<br />

CRC development to be circumvented through earlier treatment initiation, thereby<br />

avoiding increased risk of mortality and expensive cancer treatment costs. Based on these<br />

data, the model confirms that early diagnosis of CRC and adenomas results in greater<br />

life-expectancy in the eligible <strong>screening</strong> <strong>population</strong>. Although the costs associated with<br />

58 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


<strong>screening</strong> are higher, the difference appears to be at the lower margin of what is generally<br />

considered acceptable.<br />

The model aims to identify the most promising FOBTs <strong>for</strong> general <strong>population</strong> CRC<br />

<strong>screening</strong>. As discussed above, head-to-head comparisons have been made between<br />

various FOBTs and it is inappropriate to draw comparisons across the head-to-head<br />

scenarios presented.<br />

The economic model nevertheless reveals a general trend. There appears to be, based on<br />

the available data, small differences in the life-expectancy gains provided between the<br />

pairs of FOBTs tested. These differences are driven by the varying sensitivity rates<br />

utilised in each of the analyses. The level of certainty associated with this result would,<br />

however, be improved if the economic model were based on a greater body of clinical<br />

evidence.<br />

The base-case scenario of the model assumes biennial <strong>screening</strong> in individuals aged 55–<br />

74 years. The analyses show that Hem<strong>occult</strong> Sensa is cost-effective relative to<br />

HemeSelect, with a cost of $21,533 per life-year gained. In a separate analysis,<br />

HemeSelect was shown to be cost-effective relative to Hem<strong>occult</strong>, at a cost of $3172 per<br />

life-year gained. Again, it is inappropriate to draw conclusions across these comparisons.<br />

In light of the small differences in absolute life-expectancy between the different FOBTs<br />

and the uncertainty surrounding these estimates, considerations of the overall costs of<br />

the different FOBTs may become important. The total cost of colonoscopy is a key<br />

determinant in this cost difference (Figure 4 and Figure 7), primarily reflecting the<br />

specificity of the various tests. Thus, the overall costs associated with HemeSelect ($1420<br />

per individual invited to <strong>screening</strong>) were lower than those of Hem<strong>occult</strong> Sensa ($1502<br />

per individual invited to <strong>screening</strong>). Similarly, the overall cost of Hem<strong>occult</strong> ($1316 per<br />

individual invited to <strong>screening</strong>) was lower than that of HemeSelect ($1440 per individual<br />

invited to <strong>screening</strong>). There<strong>for</strong>e, the average cost of colonoscopy associated with each<br />

FOBT is a key driver of overall costs by determining the intensity of resource wastage<br />

through false positive results.<br />

The issues surrounding the importance of the cost differences must be emphasised, as<br />

there is a degree of uncertainty associated with the costs themselves. Specifically, the cost<br />

of an adverse event following colonoscopy is based on the best available estimate,<br />

though it may be incorrect in the current environment. This is unlikely to have an effect<br />

on the overall conclusion, however, given the low proportion of individuals subject to<br />

this cost.<br />

Additionally, there is uncertainty surrounding the estimates of the costs of some of the<br />

FOBTs themselves. The costs used in the model are based on the best currently available<br />

in<strong>for</strong>mation. These prices may change once a <strong>screening</strong> program is in place. This is<br />

unlikely to have a major effect on the relative cost-effectiveness.<br />

A final issue surrounding costs concerns the cost estimate <strong>for</strong> colonoscopy. The<br />

components making up the ‘associated costs’ of colonoscopy are unclear, making reestimation<br />

in the current setting difficult. As colonoscopy costs are a significant driver of<br />

the cost difference in all analyses, this is of notable importance.<br />

Significantly, there are factors other than relative cost-effectiveness which are of<br />

importance to decision-makers when considering the establishment of a <strong>population</strong><br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 59


<strong>health</strong> <strong>screening</strong> program. This may include, <strong>for</strong> example, the importance of public<br />

confidence in a <strong>screening</strong> program. The World Health Organization recommendations<br />

<strong>for</strong> choice of FOBT <strong>for</strong> CRC <strong>screening</strong> indicate that resource availability to fund the<br />

additional costs associated with follow-up colonoscopy is a key factor in the decisionmaking<br />

process (Young et al 2002). The reliability of subject compliance with the diet<br />

and drug restrictions required with the use of guaiac tests may also influence choices<br />

made (Young et al 2002).<br />

Improving participation will logically lead to the detection of more neoplasms in practice,<br />

there<strong>for</strong>e increasing the effectiveness of any <strong>screening</strong> program. Participation will not,<br />

however, have a major impact upon the relative cost-effectiveness between different<br />

FOBTs as a change in participation will shift both costs and effectiveness in the same<br />

direction, thus having a minimal impact on the incremental cost-effectiveness ratio.<br />

Altering key variables of uncertainty generally yielded little difference in the relative costeffectiveness<br />

between FOBTs or in the conclusions drawn. These changes, however, do<br />

provide insight into general ways in which the <strong>screening</strong> program may be designed to<br />

maximise its cost-effectiveness compared with the no-<strong>screening</strong> option. Importantly,<br />

these conclusions are drawn from the head-to-head comparison data, making them<br />

impossible to quantify with certainty without utilising data comparing a particular<br />

FOBT’s efficacy with no <strong>screening</strong>. Despite this, they do provide good indications as to<br />

the direction of change brought about by alterations in key variables/designs of the<br />

<strong>screening</strong> program.<br />

Increasing the <strong>screening</strong> frequency from biennial to annual increases both costs and<br />

effectiveness, both of which are the result of more intensive monitoring of the <strong>screening</strong><br />

<strong>population</strong>. This is an intuitive result. In terms of cost-effectiveness, previous studies<br />

tend to find annual <strong>screening</strong> less cost-effective than biennial <strong>screening</strong> (Bolin et al 1999;<br />

O’Leary et al 2002; Gyrd-Hansen 1998).<br />

Lowering the minimum eligible <strong>screening</strong> age from 55 to 50 years appears to offer<br />

benefits. Although costs do increase, earlier detection of adenomas and, to a lesser<br />

extent, early-stage CRC, has considerable benefits in terms of cost-effectiveness. The<br />

available data indicate that lowering the <strong>screening</strong> age to 50 may be a viable method of<br />

increasing the cost-effectiveness of FOBT <strong>screening</strong> versus no <strong>screening</strong>.<br />

Increasing the maximum <strong>screening</strong> age to 80 years does not appear to offer the same<br />

degree of benefit. This makes intuitive sense. Expanding the <strong>screening</strong> <strong>population</strong> in<br />

such a way is unlikely to offer benefits in terms of early adenoma detection and<br />

consequent reduction of CRC-induced mortality. It is not expected that such a design<br />

would prove cost-effective.<br />

All of the issues above are subject to the disclaimer that the available clinical outcomes<br />

data used in the economic model were substantially limited. Consequently, the relative<br />

cost-effectiveness ratios may be subject to change should more data become available.<br />

60 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Conclusions<br />

Safety<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> tests (FOBTs) are non-invasive and there<strong>for</strong>e unlikely to cause<br />

adverse events. Safety issues that have been raised previously in association with the use<br />

of FOBTs in a <strong>screening</strong> setting relate to a) the potential to increase the number of<br />

adverse events associated with follow-up diagnostic procedures and b) the psychological<br />

impact of colorectal cancer (CRC) <strong>screening</strong>.<br />

International randomised controlled trials (RCTs) have reported low complication rates<br />

associated with diagnostic follow-up from the use of the Hem<strong>occult</strong> FOBT <strong>for</strong><br />

<strong>population</strong> <strong>health</strong> <strong>screening</strong>.<br />

The head-to-head studies identified in this assessment examining the relative<br />

per<strong>for</strong>mance of different FOBTs did not report any safety data. There<strong>for</strong>e, the<br />

per<strong>for</strong>mance of alternative FOBTs <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong> cannot be assessed<br />

on the basis of safety data.<br />

Effectiveness<br />

There are two types of FOBTs in common use: the long established guaiac tests and the<br />

newer immunochemical tests. A key issue addressed in this review was the relative<br />

per<strong>for</strong>mance of guaiac versus immunochemical FOBTs when used <strong>for</strong> <strong>population</strong> <strong>health</strong><br />

<strong>screening</strong>. Very few head-to-head studies providing estimates of the sensitivity and<br />

specificity of different commercially produced FOBTs in the detection of CRC were<br />

identified. Slightly more studies were available which enabled a comparison of the<br />

relative true positive rates (TPRs) and false positive rates (FPRs) of these tests in<br />

<strong>population</strong> <strong>health</strong> <strong>screening</strong>. There were no studies of a suitable quality available to<br />

enable assessment of the relative accuracy of the different FOBTs <strong>for</strong> the detection of<br />

adenomas.<br />

A comparison of Hem<strong>occult</strong>, a guaiac test, and HemeSelect, a reversed passive<br />

haemagglutination (RPHA) test (which is a type of immunochemical test) was conducted.<br />

On the basis of two trials utilising the interval cancer rate as an indicator of false negative<br />

results, HemeSelect was found to be significantly more sensitive than Hem<strong>occult</strong>;<br />

however, Hem<strong>occult</strong> was significantly more specific than HemeSelect. This result was<br />

maintained in a sensitivity analysis incorporating a study conducted in a <strong>population</strong> with a<br />

high proportion of participants that were at increased risk of disease. Similarly, the<br />

relative TPR <strong>for</strong> detecting CRC was significantly in favour of pooled RPHA tests<br />

(HemeSelect and Immudia-HemSp) rather than Hem<strong>occult</strong>, whereas the FPR of<br />

Hem<strong>occult</strong> was lower than <strong>for</strong> RPHA. The practical impact of this trade-off cannot be<br />

determined by these measures alone and have been investigated via the use of an<br />

economic model.<br />

A comparison of the guaiac Hem<strong>occult</strong> test and Fecatwin Sensitive/Feca EIA, a ‘twotier’<br />

test which incorporated both guaiac and immunochemical tests conducted in a tiered<br />

fashion, was also per<strong>for</strong>med. In a single study, the sensitivities of Hem<strong>occult</strong> and<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 61


Fecatwin Sensitive/Feca EIA were not significantly different; however, Hem<strong>occult</strong> was<br />

significantly more specific than Fecatwin Sensitive/Feca EIA. Similarly, the TPR <strong>for</strong><br />

detecting CRC was not significantly different when Hem<strong>occult</strong> was compared to<br />

Fecatwin Sensitive/Feca EIA; however, the FPR of Hem<strong>occult</strong> was lower than Fecatwin<br />

Sensitive/Feca EIA. There<strong>for</strong>e, based on these limited data, Hem<strong>occult</strong> is statistically the<br />

more accurate test of these two, when used <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>.<br />

A comparison of an alternative guaiac test, Hem<strong>occult</strong> Sensa, versus the<br />

immunochemical test HemeSelect was also per<strong>for</strong>med. The sensitivity of Hem<strong>occult</strong><br />

Sensa and HemeSelect were not significantly different; however, HemeSelect proved to<br />

be significantly more specific than Hem<strong>occult</strong> Sensa, based on data from a single study.<br />

This result was maintained in a sensitivity analysis incorporating a study conducted in a<br />

<strong>population</strong> with a high proportion of participants who were at increased risk of disease.<br />

The TPR was not significantly different when Hem<strong>occult</strong> Sensa was compared with<br />

HemeSelect, whereas the FPR of HemeSelect was significantly lower than that of<br />

Hem<strong>occult</strong> Sensa. There<strong>for</strong>e, based on these limited data, HemeSelect is statistically the<br />

more accurate test of the two, when used <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong>.<br />

Newer immunochemical tests are currently commercially available. Other<br />

immunochemical FOBTs with similar technical characteristics (ie, in vitro diagnostic<br />

accuracy <strong>for</strong> the detection of haemoglobin) may provide similar outcomes to those of<br />

HemeSelect. However, there is currently a lack of evidence <strong>for</strong> this and the per<strong>for</strong>mance<br />

of other immunochemical tests in the context of <strong>population</strong> <strong>health</strong> <strong>screening</strong> <strong>for</strong> CRC<br />

remains to be determined.<br />

It is important to note that the measures of the sensitivity and specificity of the FOBTs<br />

<strong>for</strong> the detection of CRC varied considerably between studies conducted in different<br />

<strong>population</strong>s. In addition, estimates of these measures differed significantly between<br />

different tests of the same class. There<strong>for</strong>e, relative findings <strong>for</strong> any individual pairs of<br />

guaiac and immunochemical tests cannot be generalised across comparisons and findings<br />

<strong>for</strong> any individual test do not represent all tests of that type.<br />

A modelled economic evaluation was used to determine the effectiveness and costeffectiveness<br />

of the different tests in a <strong>population</strong> <strong>screening</strong> setting. This model was built<br />

upon the sensitivity and specificity data summarised above and was able to determine the<br />

broader <strong>population</strong>-wide impact of the sensitivity and specificity of these tests.<br />

Cost-effectiveness<br />

The economic model indicates that FOBTs with greater sensitivity <strong>for</strong> colorectal<br />

neoplasia detection will offer better overall survival outcomes. A key driver of the<br />

difference in the total cost associated with the FOBTs is the specificity. Tests with lower<br />

specificities are associated with higher diagnostic follow-up costs due to the increased<br />

levels of resource wastage.<br />

The model attempts to present the trade-off between these values in an economic<br />

context by providing estimates of the incremental cost per life-year gained <strong>for</strong> each pair<br />

of tests compared. The base-case scenario assumes biennial <strong>screening</strong> in individuals aged<br />

55–74 years.<br />

62 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


The incremental cost per life-year gained of HemeSelect was $3172 in a comparison<br />

against Hem<strong>occult</strong>. For Hem<strong>occult</strong> Sensa in a comparison against HemeSelect, the<br />

incremental cost per life-year gained was $21,533. These findings should, however, be<br />

constrained to the context of the head-to-head data upon which they are based. That is,<br />

the magnitude of the difference in the sensitivity between tests contains a level of<br />

uncertainty. This is particularly due to the low prevalence of CRC in the <strong>population</strong>s<br />

tested in the FOBT studies and the scarcity and poor quality of data providing estimates<br />

of the sensitivity <strong>for</strong> adenoma detection. Consequently, use of these data in the economic<br />

model may render the results of the economic model equally uncertain.<br />

However, the absolute estimates of the difference in specificity between tests are more<br />

reliable than the estimates of the sensitivity. Similarly, the results concerning costs of the<br />

FOBTs, inclusive of diagnostic follow-up and treatment, are more reliable than the<br />

modelled life-expectancy, due to the quality of the estimates that they are based upon.<br />

There<strong>for</strong>e, the tests associated with the higher specificity were shown to be less costly<br />

overall, with the impact of sensitivity upon effectiveness of uncertain magnitude. This<br />

implies that the most promising tests in an economic sense may be those associated with<br />

a high specificity.<br />

The results of these analyses <strong>for</strong> the immunochemical tests used may reflect those <strong>for</strong><br />

FOBTs with similar technical characteristics (ie, diagnostic accuracy <strong>for</strong> the detection of<br />

haemoglobin). However, there is currently a lack of suitable comparative evidence <strong>for</strong> the<br />

per<strong>for</strong>mance of other immunochemical FOBTs in the context of <strong>population</strong> <strong>health</strong><br />

<strong>screening</strong> <strong>for</strong> CRC.<br />

Other important findings of the economic model include the following.<br />

• There was no apparent class effect in determining the relative cost-effectiveness of<br />

different FOBTs.<br />

• Increasing participation will lead to increased cancer detection and an increase in the<br />

effectiveness of any <strong>screening</strong> program. Participation will not, however, have a major<br />

impact upon the relative cost-effectiveness between different FOBTs as a change in<br />

participation will shift both costs and effectiveness in the same direction, thus having<br />

a minimal effect on the incremental cost per life-year gained.<br />

• Altering key variables endogenous to the <strong>screening</strong> program generally yielded little<br />

difference in the relative cost-effectiveness between FOBTs. However, these<br />

sensitivity analyses did demonstrate consistent trends which provide insight into<br />

methods which may maximise the cost-effectiveness of <strong>screening</strong> by comparison<br />

with no-<strong>screening</strong>.<br />

• Increasing the <strong>screening</strong> frequency from biennial to annual increases both costs and<br />

effectiveness. Within the context of the main analyses, it appears that annual FOBT<br />

<strong>screening</strong> is not cost-effective compared to biennial <strong>screening</strong>.<br />

• Changing the minimum eligible <strong>screening</strong> age from 55 to 50 years appears to offer<br />

benefits in terms of cost-effectiveness.<br />

• Increasing the maximum <strong>screening</strong> age from 75 to 80 years does not appear to offer<br />

the same degree of benefit.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 63


It is important to note that the reliability of the above findings is dependent upon the<br />

reliability of the clinical data available to make head-to-head comparisons between the<br />

FOBTs. An improvement in the quantity, quality and external validity of head-to-head<br />

study data is required to validate the findings of the economic model.<br />

64 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Summary of outcomes<br />

The Medical Services Advisory Committee (MSAC) considers that faecal <strong>occult</strong> <strong>blood</strong><br />

<strong>testing</strong> is useful <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong> to reduce colorectal cancer (CRC)<br />

mortality.<br />

The available evidence indicated that there was no apparent class effect of the guaiac<br />

versus immunochemical faecal <strong>occult</strong> <strong>blood</strong> tests (FOBTs) with regard to their<br />

effectiveness or cost-effectiveness.<br />

Different brands of FOBTs possess different sensitivities and specificities <strong>for</strong> the<br />

detection of CRC within an average-risk <strong>screening</strong> <strong>population</strong> setting. The specificity of<br />

the FOBTs was a major determinant of the total associated costs of FOBT <strong>screening</strong>,<br />

inclusive of diagnostic follow-up and treatment.<br />

An economic model indicated that biennial <strong>screening</strong> was more cost-effective than<br />

annual <strong>screening</strong>, within the context of the main analysis. Lowering the minimum eligible<br />

<strong>screening</strong> age from 55 to 50 years offered benefits in terms of cost-effectiveness.<br />

Increasing the maximum <strong>screening</strong> age from 75 to 80 years did not offer the same degree<br />

of benefit.<br />

The immunochemical tests included in the assessment are no longer available as they<br />

have been replaced by newer assays. There was no available evidence suitable <strong>for</strong> the<br />

assessment of the comparative per<strong>for</strong>mance of currently available immunochemical tests<br />

within an average-risk <strong>population</strong> <strong>health</strong> <strong>screening</strong> setting. However, the results of the<br />

analyses of the immunochemical tests used may reflect those <strong>for</strong> FOBTs with similar<br />

technical characteristics (ie, in vitro diagnostic accuracy <strong>for</strong> the detection of haemoglobin).<br />

There<strong>for</strong>e, it is suggested that currently available and new immunochemical tests with<br />

promising technical characteristics be evaluated against established FOBTs within the<br />

context of an ongoing <strong>screening</strong> program.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 65


Appendix A MSAC terms of reference and<br />

membership<br />

MSAC's terms of reference are to:<br />

• advise the Minister <strong>for</strong> Health and Ageing on the strength of evidence pertaining to<br />

new and emerging medical technologies and procedures in relation to their safety,<br />

effectiveness and cost-effectiveness and under what circumstances public funding<br />

should be supported;<br />

• advise the Minister <strong>for</strong> Health and Ageing on which new medical technologies and<br />

procedures should be funded on an interim basis to allow data to be assembled to<br />

determine their safety, effectiveness and cost-effectiveness;<br />

• advise the Minister <strong>for</strong> Health and Ageing on references related either to new and/or<br />

existing medical technologies and procedures; and<br />

• undertake <strong>health</strong> technology assessment work referred by the Australian Health<br />

Ministers’ Advisory Council (AHMAC) and report its findings to AHMAC.<br />

The membership of MSAC comprises a mix of clinical expertise covering pathology,<br />

nuclear medicine, surgery, specialist medicine and general practice, plus clinical<br />

epidemiology and clinical trials, <strong>health</strong> economics, consumers, and <strong>health</strong> administration<br />

and planning:<br />

Member Expertise or affiliation<br />

Dr Stephen Blamey (Chair) General surgery<br />

Associate Professor John Atherton Cardiology<br />

Professor Bruce Barraclough General surgery<br />

Professor Syd Bell Pathology<br />

Dr Michael Cleary Emergency medicine<br />

Dr Paul Craft Clinical epidemiology and oncology<br />

Dr Gerry FitzGerald AHMAC representative<br />

Dr Kwun Fong Thoracic medicine<br />

Professor Jane Hall Health economics<br />

Dr Terri Jackson Health economics<br />

Professor Brendon Kearney Health administration and planning<br />

Associate Professor Richard King Internal medicine<br />

Dr Ray Kirk Health research<br />

Dr Michael Kitchener Nuclear medicine<br />

Dr Ewa Piejko General practice<br />

Ms Sheila Rimmer Consumer <strong>health</strong> issues<br />

Dr Jeffrey Robinson Obstetrics and gynaecology<br />

Professor John Simes Clinical epidemiology and clinical trials<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 67


Professor Bryant Stokes Neurological surgery<br />

Professor Ken Thomson Radiology<br />

Dr Douglas Travis Urology<br />

68 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Appendix B Advisory panel<br />

Advisory panel <strong>for</strong> MSAC reference 18<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> <strong>population</strong> <strong>health</strong> <strong>screening</strong><br />

Professor Brendon Kearney AM (Chair)<br />

MB BS FRACP FRACMA<br />

Executive Director Clinical Systems<br />

SA Department of Human Services<br />

Adelaide, SA<br />

Professor Syd Bell<br />

MD BS FRCPA FAFPHM(RACP)<br />

SEALS Prince of Wales Hospital<br />

Randwick, NSW<br />

Dr David Deam<br />

MB BS MAACB FRCPA<br />

Gribbles Pathology<br />

Clayton, VIC<br />

Professor Les Irwig<br />

MBBCh PhD FFPHM<br />

School of Public Health<br />

University of Sydney<br />

Sydney, NSW<br />

Ms Alex Lloyd<br />

Department of Health and Ageing<br />

Canberra<br />

ACT<br />

Associate Professor Michael Solomon<br />

MB BCH BAO(Hons) LRCSI LRCPI MSc<br />

FRACS<br />

Royal Prince Alfred Hospital<br />

Newton, NSW<br />

Elizabeth Symons<br />

RN GradDip AE&T PGradDipEval<br />

Independent Consumer Representative<br />

Highton, VIC<br />

Member of MSAC<br />

Member of MSAC<br />

Nominated by the Royal<br />

College of Pathologists<br />

of Australasia<br />

Co-opted member<br />

Project manager<br />

Nominated by the<br />

Colorectal Surgical<br />

Society of Australasia<br />

Nominated by the<br />

Consumers’ Health<br />

Forum<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 69


Professor Graeme Young<br />

MBBS MD FRACP<br />

Department of Gastroenterology and Hepatology<br />

Flinders Medical Centre and Repatriation Hospital<br />

Bed<strong>for</strong>d Park, SA<br />

Nominated by the<br />

Gastroenterological<br />

Society of Australia<br />

70 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Appendix C Studies included in the review<br />

Table 39 Relevant studies identified<br />

Specificity<br />

carcinoma<br />

n/N (%)<br />

Sensitivity<br />

carcinoma<br />

n/N (%)<br />

PPV<br />

neoplasm<br />

n/N (%)<br />

PPV<br />

carcinoma<br />

n/N (%)<br />

FOBTs<br />

Included in final<br />

assessment?<br />

(QS)<br />

LoE<br />

Long-term<br />

follow-up<br />

Randomisation/<br />

multiple tests in a<br />

single patient<br />

Reference standard<br />

Population<br />

(ITS)<br />

Study<br />

7845/8030<br />

(97.7)<br />

13/35<br />

(37.1)<br />

46/198<br />

(23.2)<br />

13/198<br />

(6.6)<br />

Hem<strong>occult</strong> II<br />

Y<br />

(7)<br />

1<br />

6824/7870<br />

(86.7)<br />

27/34<br />

(79.4)<br />

99/1073<br />

(9.2)<br />

27/1073<br />

(2.5)<br />

Hem<strong>occult</strong> II<br />

Sensa<br />

Yes, 2 years<br />

review of<br />

insurance<br />

records <strong>for</strong><br />

neoplasms<br />

(96% of<br />

subjects)<br />

Multiple tests<br />

Positive FOBT:<br />

colonoscopy;<br />

Hem<strong>occult</strong> II Sensa<br />

positive: FSa 7043/7461<br />

(94.4)<br />

22/32<br />

(68.8)<br />

90/440<br />

(20.5)<br />

22/440<br />

(5.0)<br />

HemeSelect<br />

1261/1298<br />

(97.2)<br />

3/6<br />

(50.0)<br />

28/40<br />

(70.0)<br />

3/40<br />

(7.5)<br />

Hem<strong>occult</strong> (no<br />

dietary<br />

restriction)<br />

Y<br />

(6)<br />

1<br />

1197/1298<br />

(92.2)<br />

5/6<br />

(83.3)<br />

27/106<br />

(25.5)<br />

5/106<br />

(4.7)<br />

Feca EIA b<br />

Yes, 1 year<br />

review of<br />

hospital<br />

pathology<br />

records <strong>for</strong><br />

neoplasms<br />

(Hardcastle et<br />

al 1986)<br />

Multiple tests<br />

Positive: history, exam, RS<br />

and 60 cm fibreoptic<br />

sigmoid.<br />

Then if carcinoma: DCBE;<br />

adenoma: colonoscopy; no<br />

neoplasm: DCBE<br />

If negative follow-up:<br />

repeat FOBT, if positive,<br />

gastroscopy<br />

Hem<strong>occult</strong> (no<br />

dietary<br />

restriction)<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 71<br />

–<br />

–<br />

–<br />

–<br />

N<br />

Guaiac to guaiac<br />

comparison<br />

1<br />

Yes, 2 years<br />

follow-up <strong>for</strong><br />

colonic<br />

symptom<br />

presentation to<br />

GP<br />

Multiple tests<br />

Fecatwin<br />

(guaiac)<br />

Positive: history, exam,<br />

RS, sample obtained on<br />

exam tested with<br />

Haemostix. Then if positive<br />

or pathology suspected:<br />

full investigation<br />

If negative: repeat FOBT<br />

after 3 months<br />

Members of<br />

non-profit US<br />

private <strong>health</strong><br />

insurer<br />

electing <strong>health</strong><br />

appraisal,<br />

≥ 50 years;<br />

N = 10,702<br />

Individuals<br />

from GP<br />

registers (UK),<br />

45–75 years;<br />

N = 3225;<br />

excluded<br />

known large<br />

bowel disease<br />

patients<br />

GP patient<br />

attendees<br />

(UK),<br />

> 40 years;<br />

N = 640<br />

(group 1)<br />

Allison et al<br />

1996a<br />

Armitage et<br />

al 1985<br />

Barrison and<br />

Parkins<br />

1985b<br />

Rehydrated<br />

Hem<strong>occult</strong> I<br />

–<br />

–<br />

–<br />

–<br />

N<br />

Rehydrated<br />

1<br />

Yes, 2 year<br />

repeat FOBT<br />

<strong>screening</strong> in<br />

25% of<br />

subjects<br />

Multiple tests<br />

HemeSelect<br />

No<br />

Shionogi B<br />

Parallel arm with<br />

historical cohort<br />

Immudia-<br />

HemSp<br />

N<br />

Low quality of<br />

evidence<br />

3b<br />

Positive HO and/or<br />

positive/borderline<br />

HSelect: pancolonoscopy<br />

or left colonoscopy plus<br />

DCBE when<br />

pancolonoscopy not<br />

possible<br />

Positive, symptomatic or<br />

FH CRC: exam, FS and<br />

DCBE<br />

General<br />

<strong>population</strong><br />

(Italy)<br />

40–70 years;<br />

N = 24,282<br />

Castiglione<br />

et al 1997<br />

Mass<br />

<strong>screening</strong><br />

(Japan)<br />

> 40 years;<br />

N = 18,497<br />

Fujita et al<br />

1992


Specificity<br />

carcinoma<br />

Sensitivity<br />

carcinoma<br />

PPV<br />

neoplasm<br />

PPV<br />

carcinoma<br />

FOBTs<br />

Included in final<br />

assessment?<br />

(QS/12)<br />

LoE<br />

Long-term<br />

follow-up<br />

Randomisation/<br />

multiple tests in<br />

a single patient<br />

Reference standard<br />

Population (ITS)<br />

Study<br />

16/410<br />

(3.9)<br />

8/410<br />

(2.0)<br />

No<br />

Multiple tests<br />

Hem<strong>occult</strong> II<br />

All: FS<br />

Positive: DCBE or colonoscopy<br />

Individuals presenting <strong>for</strong><br />

<strong>health</strong> care examination<br />

(Japan); most 50–59 years<br />

(mean 52 years); N = 6437<br />

Iwase 1992<br />

–<br />

–<br />

Y<br />

(5)<br />

1<br />

44/391<br />

(11.3)<br />

16/391<br />

(4.1)<br />

Immudia-<br />

HemSp<br />

–<br />

–<br />

–<br />

–<br />

Hem<strong>occult</strong> II<br />

(unhydrated)<br />

Hydrated<br />

Hem<strong>occult</strong> II<br />

Hem<strong>occult</strong><br />

Sensa<br />

No<br />

Multiple tests<br />

Positive: full colonoscopy or FS<br />

+ DCBE<br />

N<br />

Guaiac to guaiac<br />

comparison<br />

1<br />

60/456<br />

(13.2) c<br />

16/456<br />

(3.5) c<br />

No<br />

Multiple tests<br />

Hem<strong>occult</strong> II<br />

72 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

–<br />

–<br />

104/849<br />

(12.2) c<br />

19/849<br />

(2.2) c<br />

Hem<strong>occult</strong><br />

Sensa<br />

Y<br />

(5)<br />

1<br />

Positive: mixed, including<br />

exam, repeat FOBT, DCBE,<br />

sigmoid, colonoscopy or an<br />

upper gastrointestinal barium<br />

series. 70% had DCBE or<br />

colonoscopy<br />

General <strong>population</strong> (US,<br />

distribution through<br />

pharmacies and<br />

community groups<br />

following media campaign)<br />

≥ 50 years; N = 85,931<br />

General <strong>population</strong> (US,<br />

distribution through<br />

selected pharmacies<br />

following media campaign)<br />

> 30 years; N = 39,000<br />

Levin et al<br />

1997<br />

Petrelli et al<br />

1994<br />

66/438<br />

(15.1) c<br />

21/438<br />

(4.8) c<br />

HemeSelect<br />

1462/1478<br />

(98.9) b<br />

1/11<br />

(9.1) a,b<br />

9/17<br />

(52.9)<br />

1/17<br />

(5.9)<br />

Hem<strong>occult</strong><br />

Y<br />

(6)<br />

1<br />

1344/1478<br />

(90.9) b<br />

11/11<br />

(100) b<br />

57/145<br />

(39.3)<br />

9/145<br />

(6.2)<br />

HemeSelect<br />

Yes, median<br />

35 months by<br />

GP in<strong>for</strong>mation<br />

and review of<br />

hospital<br />

pathology<br />

records <strong>for</strong><br />

neoplasm<br />

Multiple tests<br />

Positive: exam, sigmoid and<br />

colonoscopy/FS & DCBE<br />

Individuals from GP<br />

registers (UK)<br />

50–75 years; N = 4018;<br />

known colorectal<br />

neoplasm, advanced<br />

visceral malignancy or<br />

unfit <strong>for</strong> further<br />

investigation excluded<br />

Robinson et<br />

al 1995<br />

No<br />

Multiple tests<br />

437/522<br />

(83.7)<br />

3/5<br />

(60.0)<br />

21/88<br />

(23.9<br />

3/88<br />

(3.4)<br />

Hem<strong>occult</strong><br />

Sensa<br />

All had endoscopy, FS (41%) or<br />

total colonoscopy (59%,<br />

including FOBT positive)<br />

Consecutive attendees at<br />

CRC <strong>screening</strong> service<br />

(95% asymptomatic);<br />

N = nr (527 screened)<br />

Rozen et al<br />

1995<br />

Y<br />

(Sensitivity<br />

analysis)<br />

(4)<br />

1<br />

467/522<br />

(89.5)<br />

4/5 (80.0)<br />

17/59<br />

(28.8)<br />

4/59<br />

(6.8)<br />

BM-Test Colon<br />

Albumin


Specificity<br />

carcinoma<br />

Sensitivity<br />

carcinoma<br />

PPV<br />

neoplasm<br />

PPV<br />

carcinoma<br />

FOBTs<br />

Included in final<br />

assessment?<br />

(QS)<br />

LoE<br />

Long-term<br />

follow-up<br />

Randomisation/<br />

multiple tests in<br />

a single patient<br />

Reference standard<br />

Population (ITS)<br />

Study<br />

377/398<br />

(94.7)<br />

3/5<br />

(60.0)<br />

7/24<br />

(29.2)<br />

3/24<br />

(12.5)<br />

No<br />

Multiple tests<br />

Hem<strong>occult</strong> II<br />

All had endoscopy, 59% at time<br />

of study<br />

366/398<br />

(92.0)<br />

3/5<br />

(60.0)<br />

8/35<br />

(22.9)<br />

385/398<br />

(96.7)<br />

4/5<br />

(80.0)<br />

6/17<br />

(35.3)<br />

390/398<br />

(98.0)<br />

4/5<br />

(80.0)<br />

8/12<br />

(66.7)<br />

3/35<br />

(8.6)<br />

4/17<br />

23.5<br />

4/12<br />

(33.3)<br />

Hem<strong>occult</strong><br />

Sensa<br />

Consecutive attendees at<br />

CRC <strong>screening</strong> service<br />

(97% asymptomatic);<br />

N = nr (403 screened)<br />

Rozen et al<br />

1997<br />

FlexSure OBT<br />

Y<br />

(Sensitivity<br />

analysis)<br />

(5)<br />

1<br />

HemeSelect<br />

1332/1403<br />

(95.2)<br />

3/7<br />

(42.9)<br />

19/74<br />

(25.7)<br />

3/74<br />

(4.1)<br />

Hem<strong>occult</strong><br />

Sensa<br />

1382/1403<br />

(98.6)<br />

3/7<br />

(42.9)<br />

10/24<br />

(41.7)<br />

3/24<br />

(12.5)<br />

FlexSure OBT<br />

Y<br />

(Sensitivity<br />

analysis)<br />

(5)<br />

1<br />

Yes, > 2 years<br />

clinical followup<br />

in 4<br />

subjects with<br />

positive FOBT<br />

who refused<br />

colonoscopy<br />

Multiple tests<br />

All had full colonoscopy or FS,<br />

48% at time of study<br />

Consecutive attendees at<br />

CRC <strong>screening</strong> service<br />

(97% asymptomatic);<br />

N = nr (1410 screened)<br />

Rozen et al<br />

2000<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 73<br />

16/68<br />

(23.5)<br />

1/68<br />

(1.5)<br />

Hem<strong>occult</strong><br />

Sensa<br />

No<br />

Multiple tests<br />

Positive: colonoscopy<br />

–<br />

–<br />

11/41<br />

(26.8)<br />

1/41<br />

(2.4)<br />

HemeSelect<br />

Y<br />

(Sensitivity<br />

analysis)<br />

(5)<br />

1<br />

Participants in <strong>screening</strong><br />

program (screenees:<br />

individuals with FH CRC<br />

(80%) and community<br />

volunteers); 45–79 years;<br />

N = nr (1355 screened)<br />

St John et al<br />

1993<br />

(Screenees)<br />

Hem<strong>occult</strong> II<br />

–<br />

–<br />

–<br />

–<br />

N<br />

Low quality of<br />

evidence<br />

3b<br />

Yes, cancer<br />

registry data<br />

over 5 years &<br />

negative<br />

FOBTs<br />

rescreened in<br />

2.5 years<br />

Parallel arms,<br />

non-randomised<br />

Total colonoscopy or<br />

colonoscopy & DCBE<br />

General <strong>population</strong> (Italy)<br />

50–70 years; N = nr<br />

(41,744 screened)<br />

Zappa et al<br />

2001<br />

RPHA<br />

(HemeSelect;<br />

Immudia--<br />

HemSp)<br />

Abbreviations: CRC, colorectal cancer; DCBE, double contrast barium enema; Fecatwin, Fecatwin Sensitive/Feca EIA; FH, family history; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; FS, flexible sigmoidoscopy; GP, general practice; HSelect,<br />

HemeSelect; HO, Hem<strong>occult</strong>; ITS, intention to screen; LoE, level of evidence; nr, not reported; PPV, positive predictive value; QS, quality score; RPHA, reverse-passive haemagglutination; RS, rigid sigmoidoscopy.<br />

aAssumes that the patient developing cancer during follow-up <strong>for</strong> adenoma initially tested negative with Hem<strong>occult</strong>. bTaken from Robinson et al (1996). cTotal positives calculated from number of patients screened and percentage<br />

positivity rate as reported, authors were contacted <strong>for</strong> data confirmation but no response was received.


Appendix D Literature searches<br />

Medline search strategy<br />

The search strategy used to identify relevant studies of faecal <strong>occult</strong> <strong>blood</strong> tests (FOBTs)<br />

in Medline is presented in Table 40.<br />

Table 40 FOBT Medline search strategy (1966 to February week 3 <strong>2004</strong>)<br />

Search history Results<br />

1 <strong>occult</strong> <strong>blood</strong>/ 2654<br />

2 ((fecal or faecal) adj2 <strong>blood</strong>).ti,ab. 1679<br />

3 (fobt$1 or hem<strong>occult</strong> or haem<strong>occult</strong>).ti,ab. 748<br />

4 ((fecal or faecal) adj2 (haemoglobin or hemoglobin)).ti,ab. 58<br />

5 ((fecal or faecal) adj2 (globin or haem or heme)).ti,ab. 19<br />

6 or/1–5 3472<br />

7 exp colorectal neoplasms/ 76,016<br />

8 (colorect$ adj5 screen$).ti,ab. 2188<br />

9 or/6–7 77,631<br />

10 9 and exp mass <strong>screening</strong>/ 2793<br />

11 6 and exp "sensitivity and specificity"/ 471<br />

12 6 and pc.fs. 938<br />

13 6 and di.fs. 1745<br />

14 6 and ep.fs. 553<br />

15 or/8,10–14 4881<br />

16 haem<strong>occult</strong>.ti,ab. 131<br />

17 fecatwin.ti,ab. 13<br />

18 colocare.ti,ab. 3<br />

19 okokit.ti,ab. 3<br />

20 hemofec.ti,ab. 10<br />

21 flexsure.ti,ab. 22<br />

22 hemeselect.ti,ab. 28<br />

23 (feca-eia or (feca adj eia) or fecaeia).ti,ab. 10<br />

24 ((iatro adj hemcheck) or iatro-hemcheck or iatrohemcheck).ti,ab. 2<br />

25 (imdia-hem or (imdia adj hem) or imdiahem).ti,ab. 0<br />

26 hemochaser.ti,ab. 2<br />

27 monohaem.ti,ab. 27<br />

28 (hemodia or ochemodia or oc-hemodia).ti,ab. 11<br />

29 (hemoglobin-haptoglobin or (hemoglobin adj haptoglobin) or hemoglobinhaptoglobin).ti,ab. 131<br />

30 haptoglobin.ti,ab. 3661<br />

31 annual bowel check.ti,ab. 0<br />

32 guaiac.ti,ab. 294<br />

33 (immunochemical and test$).ti,ab. 1699<br />

34 elisa.ti,ab. 50,550<br />

35 in<strong>for</strong>m.ti,ab. 5142<br />

36 or/30,32–35 61,105<br />

37 36 and 6 291<br />

74 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Search history Results<br />

38 or/15–29,31,37 5103<br />

39 38 and exp clinical trials/ 178<br />

40 38 and clinical trial.pt. 322<br />

41 38 and (clinical adj trial$1).ti,ab. 59<br />

42 or/39–41 491<br />

43 (case adj report).ti,ab. 96,592<br />

44 letter.pt. 495,739<br />

45 historical article.pt. 206,244<br />

46 review of reported cases.pt. 48,682<br />

47 review multicase.pt. 7963<br />

48 or/43–47 838,778<br />

49 42 not 48 476<br />

EMBASE search strategy<br />

The search strategy used to identify relevant studies of FOBTs in EMBASE is presented<br />

in Table 41.<br />

Table 41 FOBT EMBASE search strategy (1980 to <strong>2004</strong> week 09)<br />

Search history Results<br />

1 <strong>occult</strong> <strong>blood</strong>/ or <strong>occult</strong> <strong>blood</strong> test/ 2411<br />

2 ((fecal or faecal) adj2 <strong>blood</strong>).ti,ab. 1435<br />

3 (fobt$1 or hem<strong>occult</strong> or haem<strong>occult</strong>).ti,ab. 617<br />

4 ((fecal or faecal) adj2 (haemoglobin or hemoglobin)).ti,ab. 51<br />

5 ((fecal or faecal) adj2 (globin or haem or heme)).ti,ab. 16<br />

6 or/1–5 2986<br />

7 colorectal cancer/ 18,806<br />

8 colorectal carcinoma/ 6335<br />

9 colorectal tumor/ 1163<br />

10 (colorect$ adj5 screen$).ti,ab. 1928<br />

11 or/6–9 27,581<br />

12 11 and exp <strong>screening</strong>/ 3158<br />

13 6 and exp "sensitivity and specificity"/ 41<br />

14 6 and pc.fs. 287<br />

15 6 and di.fs. 1725<br />

16 6 and ep.fs. 473<br />

17 exp feces analysis/ 8053<br />

18 17 and exp diagnostic test/ 1404<br />

19 or/10,12–16,18 5155<br />

20 haem<strong>occult</strong>.ti,ab,tn,mf. 105<br />

21 fecatwin.ti,ab,tn,mf. 16<br />

22 colocare.ti,ab,tn,mf. 4<br />

23 okokit.ti,ab,tn,mf. 2<br />

24 hemofec.ti,ab,tn,mf. 3<br />

25 flexsure.ti,ab,tn,mf. 24<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 75


Search history Results<br />

26 hemeselect.ti,ab,tn,mf. 28<br />

27 (feca-eia or (feca adj eia) or fecaeia).ti,ab,tn,mf. 8<br />

28 ((iatro adj hemcheck) or iatro-hemcheck or iatrohemcheck).ti,ab,tn,mf. 2<br />

29 (imdia-hem or (imdia adj hem) or imdiahem).ti,ab,tn,mf. 0<br />

30 hemochaser.ti,ab,tn,mf. 2<br />

31 monohaem.ti,ab,tn,mf. 20<br />

32 (hemodia or ochemodia or oc-hemodia).ti,ab,tn,mf. 13<br />

33 (hemoglobin-haptoglobin or (hemoglobin adj haptoglobin) or hemoglobinhaptoglobin).ti,ab,tn,mf. 79<br />

34 haptoglobin.ti,ab,tn,mf. 2224<br />

35 annual bowel check.ti,ab,tn,mf. 0<br />

36 guaiac.ti,ab,tn,mf. 251<br />

37 (immunochemical and test$).ti,ab,tn,mf. 1230<br />

38 elisa.ti,ab,tn,mf. 41,694<br />

39 in<strong>for</strong>m.ti,ab,tn,mf. 4202<br />

40 or/34,36–39 49,392<br />

41 40 and 6 217<br />

42 40 and 18 122<br />

43 or/19–33,35,41–42 5318<br />

44 43 and exp clinical trial/ 539<br />

45 43 and (clinical adj trial$1).ti,ab. 85<br />

46 43 and ct.fs. 185<br />

47 or/44–46 594<br />

48 case study/ 3312<br />

49 (case adj report).ti,ab. 81,218<br />

50 abstract-report/ 71,109<br />

51 letter/ 268,678<br />

52 or/48–51 423,107<br />

53 47 not 52 578<br />

76 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Appendix E Excluded references<br />

488 citations were excluded as they were a non-systematic review, editorial, letter; news article,<br />

note, survey, opinion piece, or economic analysis.<br />

12 citations were excluded as they were a non-human or in vitro study.<br />

485 citations were excluded as they were not of FOBT.<br />

Reasons <strong>for</strong> exclusion of all other citations are given below:<br />

1. Ahlquist DA, Wieand HS, Moertel CG, McGill DB, Loprinzi CL, O'Connell MJ, Mailliard<br />

JA, Gerstner JB, Pandya K, Ellefson RD (1993). Accuracy of fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong><br />

<strong>for</strong> colorectal neoplasia. A prospective study using Hem<strong>occult</strong> and HemoQuant tests.<br />

JAMA 269: 1262-1267.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

2. Aisawa T, Saito H, Kawaguchi H, Uno Y, Munakata A, Yoshida Y (1988). Mass <strong>screening</strong><br />

<strong>for</strong> colon cancer using an immunologic fecal <strong>occult</strong> <strong>blood</strong> test by reversed passive<br />

hemagglutination reaction (RPHA)-comparison of single RPHA with 2-day <strong>testing</strong>,<br />

Proceedings of the 8th Asia-Pacific Congress of Gastroenterology FP23-FP25.<br />

Reason <strong>for</strong> exclusion: unavailable<br />

3. Archambault F, Durand G, Faivre J, Voilquin JP, Archambault P, Riou F, Ageorges P,<br />

Archambault ML, Delville JM, Fesneau M, et al (1989). Acceptability of the Hem<strong>occult</strong> test<br />

in general medical practice. Results of a pilot study [French]. Bulletin du Cancer 76: 1071-1075.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

4. Armitage NC, Hardcastle JD (1987). Screening <strong>for</strong> colorectal cancer–the Nottingham<br />

experience. Annals of the Academy of Medicine, Singapore 16: 432-436.<br />

Reason <strong>for</strong> exclusion: study duplication<br />

5. Armitage N, Hardcastle J, Leicester R (1989). Screening <strong>for</strong> colorectal cancer. Practitioner 233:<br />

830-833.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

6. Asao T, Kuwano H, Ide M, Hirayama I, Nakamura J-I, Fujita K-I, Horiuti R (1992).<br />

Spasmolytic effect of peppermint oil in barium during double-contrast barium enema<br />

compared with Buscopan. Clinical Radiology 58: 01.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

7. Baier M, Calonge N, Cutter G, McClatchey M, Schoentgen S, Hines S, Marcus A, Ahnen D<br />

(2000). Validity of self-reported colorectal cancer <strong>screening</strong> behavior. Cancer Epidemiology,<br />

Biomarkers & Prevention 9: 229-232.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

8. Barrison IG, Littlewood ER, Primavesi J, Sharples A, Gilmore IT, Parkins RA (1981).<br />

Screening <strong>for</strong> <strong>occult</strong> gastrointestinal bleeding in hospital patients. Journal of the Royal Society of<br />

Medicine 74: 41-43.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

9. Bauer JJ, Finger MJ, Heidenberg HB, Preston DM, Moses FM, Watson RA, Irby PB (1997).<br />

Incidence of stool guaiac conversion following extracorporeal shock wave lithotripsy. Urology<br />

50: 192-194.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 77


10. Bech K, Kronborg O, Fenger C (1991). Adenomas and hyperplastic polyps in <strong>screening</strong><br />

studies. World Journal of Surgery 15: 7-13.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

11. Bech K, Kronborg O (1992). Requirement of hospital beds in connection with <strong>screening</strong> <strong>for</strong><br />

colorectal cancer. The first 5-years of a randomized <strong>population</strong> survey [Danish]. Ugeskrift <strong>for</strong><br />

Laeger 154: 696-699.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

12. Bejes C, Marvel MK (1992). Attempting the improbable: offering colorectal cancer<br />

<strong>screening</strong> to all appropriate patients. Family Practice Research Journal 12: 83-90.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

13. Bennett DH, Robinson MR, Preece P, Moshakis V, Vellacott KD, Besbeas S, Kewenter J,<br />

Kronborg O, Moss S, Chamberlain J, Hardcastle JD (1995). Colorectal cancer <strong>screening</strong>: the<br />

effect of combining flexible sigmoidoscopy with a faecal <strong>occult</strong> <strong>blood</strong> test [abstract]. Gut 36:<br />

F23.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

14. Berry DP, Clarke P, Hardcastle JD, Vellacott KD (1997). Randomized trial of the addition<br />

of flexible sigmoidoscopy to faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> colorectal neoplasia <strong>population</strong><br />

<strong>screening</strong>. British Journal of Surgery 84: 1274-1276.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

15. Blackshear JL, Baker VS, Holland A, Litin SC, Ahlquist DA, Hart RG, Ellefson R, Koehler J<br />

(1996). Fecal hemoglobin excretion in elderly patients with atrial fibrillation: Combined<br />

aspirin and low-dose warfarin vs conventional warfarin therapy. Archives of Internal Medicine<br />

156: 658-660.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

16. Brevinge H, Lindholm E, Buntzen S, Kewenter J (1997). Screening <strong>for</strong> colorectal neoplasia<br />

with faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> compared with flexible sigmoidoscopy directly in a 55-56<br />

years' old <strong>population</strong>. International Journal of Colorectal Disease 12: 291-295.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

17. Cargill VA, Conti M, Neuhauser D, McClish D (1991). Improving the effectiveness of<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer by involving nurse clinicians. Medical Care 29: 1-5.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

18. Castiglione G, Zappa M, Grazzini G, Mazzotta A, Biagini M, Salvadori P, Ciatto S (1996).<br />

Immunochemical vs guaiac faecal <strong>occult</strong> <strong>blood</strong> tests in a <strong>population</strong>-based <strong>screening</strong><br />

program <strong>for</strong> colorectal cancer. British Journal of Cancer 74: 141-144.<br />

Reason <strong>for</strong> exclusion: study duplication<br />

19. Chen TH, Yen MF, Lai MS, Koong SL, Wang CY, Wong JM, Prevost TC, Duffy SW (1999).<br />

Evaluation of a selective <strong>screening</strong> <strong>for</strong> colorectal carcinoma: the Taiwan Multicenter Cancer<br />

Screening (TAMCAS) project. Cancer 86: 1116-1128.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

20. Cole SR, Young GP (2001). Effect of dietary restriction on participation in faecal <strong>occult</strong><br />

<strong>blood</strong> test <strong>screening</strong> <strong>for</strong> colorectal cancer. Medical Journal of Australia 175: 195-198.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

78 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


21. Cole SR, Young GP, Esterman A, Cadd B, Morcom J (2003). A randomised trial of the<br />

impact of new faecal haemoglobin test technologies on <strong>population</strong> participation in <strong>screening</strong><br />

<strong>for</strong> colorectal cancer. Journal of Medical Screening 10: 117-122.<br />

Reason <strong>for</strong> exclusion: inadequate data separation<br />

22. Courtier R, Casamitjana M, Macia F, Panades A, Castells X, Gil M-J, Hidalgo JM, Sanchez-<br />

Ortega JM (2002). Participation in a colorectal cancer <strong>screening</strong> program: Influence of the<br />

method of contacting the target <strong>population</strong>. European Journal of Cancer Prevention 11: 209-213.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

23. Deyhle P, Nuesch HJ, Kobler E (1976). The haem<strong>occult</strong> test in <strong>screening</strong> <strong>for</strong> carcinoma of<br />

the colon [German]. Schweizerische Medizinische Wochenschrift 106: 297.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

24. Duane WC, Heneghan MA, McCarthy CF, Fine KD (1996). Occult gastrointestinal bleeding<br />

in celiac sprue [4]. New England Journal of Medicine 335: 753.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

25. Duggan AE, Elliott C, Logan RF (1999). Testing <strong>for</strong> Helicobacter pylori infection: validation<br />

and diagnostic yield of a near patient test in primary care. British Medical Journal 319: 1236-<br />

1239.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

26. Durst J, Neumann G, Schmidt K (1976). Occult <strong>blood</strong> in stool. A field trial in cancer<br />

<strong>screening</strong> [German]. Deutsche Medizinische Wochenschrift 101: 440-443.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

27. Ederer F, Church TR, Mandel JS (1997). Fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> in the Minnesota<br />

study: role of chance detection of lesions [see comments]. Journal of the National Cancer<br />

Institute 89: 1423-1428.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

28. Elwood TW, Erickson A, Lieberman S (1978). Comparative educational approaches to<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer. American Journal of Public Health 68: 135-138.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

29. Faivre J, Tazi MA, El Mrini T, Lejeune C, Benhamiche AM, Dassonville F (1999). <strong>Faecal</strong><br />

<strong>occult</strong> <strong>blood</strong> <strong>screening</strong> and reduction of colorectal cancer mortality: A case-control study.<br />

British Journal of Cancer 79: 680-683.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

30. Fattah AS, Nakama H, Kamijo N, Fujimori K, Zhang B (1998). Colorectal adenomatous<br />

polyps detected by immunochemical <strong>occult</strong> <strong>blood</strong> <strong>screening</strong>. Hepato-Gastroenterology 45: 712-<br />

716.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

31. Fernandez JL, Gallegos M, Brochero A, Arevalo C, Piccioni H, Gutierrez G (1999).<br />

Screening <strong>for</strong> colorectal cancer with an immunological fecal <strong>occult</strong> <strong>blood</strong> test [Spanish]. Acta<br />

Gastroenterologica Latinoamericana 29: 73-78.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

32. Fludger S, Turner AM, Harvey RF, Haslam N (2002). Controlled prospective study of faecal<br />

<strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer in Bury, black pudding capital of the world.<br />

British Medical Journal 325: 1444-1445.<br />

Reason <strong>for</strong> exclusion: wrong usage<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 79


33. Foliente RL, Wise GR, Collen MJ, Abdulian JD, Chen YK (1995). Colocare self-test versus<br />

Hem<strong>occult</strong> II Sensa <strong>for</strong> fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>. American Journal of Gastroenterology 90:<br />

2160-2163.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

34. Fric P, Zavoral M, Dvorakova H, Zoubek V, Roth Z (1994). An adapted program of<br />

colorectal cancer <strong>screening</strong>: 7 years experience and cost-benefit analysis. Hepato-<br />

Gastroenterology 41: 413-416.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

35. Friedman LC, Everett TE, Peterson L, Ogbonnaya KI, Mendizabal V (2001). Compliance<br />

with fecal <strong>occult</strong> <strong>blood</strong> test <strong>screening</strong> among low-income medical outpatients: a randomized<br />

controlled trial using a videotaped intervention. Journal of Cancer Education 16: 85-88.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

36. Frommer DJ, Kapparis A, Brown MK (1988). Improved <strong>screening</strong> <strong>for</strong> colorectal cancer by<br />

immunological detection of <strong>occult</strong> <strong>blood</strong>. British Medical Journal Clinical Research Edition 296:<br />

1092-1094.<br />

Reason <strong>for</strong> exclusion: not a commercially available FOBT<br />

37. Fujita M, Nakano Y, Ota J, Kumanishi Y, Taguchi T (1987). Mass <strong>screening</strong> <strong>for</strong> colorectal<br />

cancer by <strong>testing</strong> <strong>for</strong> <strong>occult</strong> <strong>blood</strong> under restricted diet and a questionnaire in Osaka. Cancer<br />

Detection & Prevention 353-359.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

38. Garcia-Diaz E, Castro-Fernandez M, Romero-Gomez M, Vargas-Romero J (2003). The<br />

effectiveness of (IgG-ELISA) serology as an alternative diagnostic method <strong>for</strong> detecting<br />

Helicobacter pylori infection in patients with gastro-intestinal bleeding due to gastroduodenal<br />

ulcer. Revista Espanola de Enfermedades Digestivas 94: 01.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

39. Gerbes L, Jungst D, Kobberling J (1994). Does yearly fecal exam <strong>for</strong> <strong>occult</strong> <strong>blood</strong> lower<br />

colorectal cancer mortality? [German]. Zeitschrift fur Gastroenterologie 32: 603-606.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

40. Gilbert JA, Ahlquist DA, Mahoney DW, Zinsmeister AR, Rubin J, Ellefson RD (1996).<br />

Fecal marker variability in colorectal cancer: Calprotectin versus hemoglobin. Scandinavian<br />

Journal of Gastroenterology 31: 1001-1005.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

41. Gnauck R (1977). Screening <strong>for</strong> colorectal cancer with the haem<strong>occult</strong> test [German]. Leber,<br />

Magen, Darm 7: 32-35.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

42. Greenberg PD, Bertario L, Gnauck R, Kronborg O, Hardcastle JD, Epstein MS, Sadowski<br />

D, Sudduth R, Zuckerman GR, Rockey DC (2000). A prospective multicenter evaluation of<br />

new fecal <strong>occult</strong> <strong>blood</strong> tests in patients undergoing colonoscopy. American Journal of<br />

Gastroenterology 95: 1331-1338.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

43. Griesenberg D, Nurnberg R, Bahlo M, Klapdor R (1999). CEA, TPS, CA 19-9 and CA 72-4<br />

and the fecal <strong>occult</strong> <strong>blood</strong> test in the preoperative diagnosis and follow-up after resective<br />

surgery of colorectal cancer. Anticancer Research 19: 2443-2450.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

80 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


44. Gyrd-Hansen D, Sogaard J, Kronborg O (1997). Analysis of <strong>screening</strong> data: colorectal<br />

cancer. International Journal of Epidemiology 26: 1172-1181.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

45. Gyrd-Hansen D, Sogaard J, Kronborg O (1998). Colorectal cancer <strong>screening</strong>:efficiency and<br />

effectiveness. Health Economics 7: 9-20.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

46. Hahn M, Fennerty MB, Corless CL, Magaret N, Lieberman DA, Faigel DO (2000).<br />

Noninvasive tests as a substitute <strong>for</strong> histology in the diagnosis of Helicobacter pylori<br />

infection. Gastrointestinal Endoscopy 52: 20-26.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

47. Hardcastle J (1991). Randomized control trial of faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal<br />

cancer: results <strong>for</strong> the first 144,103 patients. European Journal of Cancer Prevention 1(Suppl 2):<br />

21.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

48. Hardcastle JD, Farrands PA, Balfour TW, Chamberlain J, Amar SS, Sheldon MG (1983).<br />

Controlled trial of faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> in the detection of colorectal cancer. Lancet 2:<br />

1-4.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

49. Hardcastle JD, Armitage NC, Chamberlain J, Balfour TW, Amar SS (1985). A control trial<br />

of faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer: 2-year results. British Journal of Surgery<br />

72(Suppl): S69-S71.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

50. Hardcastle JD, Armitage NC, Chamberlain J, Amar SS, James PD, Balfour TW (1986). Fecal<br />

<strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer in the general <strong>population</strong>. Results of a<br />

controlled trial. Cancer 58: 397-403.<br />

Reason <strong>for</strong> exclusion: study duplication<br />

51. Hardcastle JD (1987). Population <strong>screening</strong> <strong>for</strong> colorectal cancer. Proceedings of the Div Surg R<br />

British Hospitals 3: 29-31.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

52. Hardcastle JD, Thomas WM, Chamberlain J, Pye G, Sheffield J, James PD, Balfour TW,<br />

Amar SS, Armitage NC, Moss SM (1989). Randomised, controlled trial of faecal <strong>occult</strong><br />

<strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer. Results <strong>for</strong> first 107,349 subjects. Lancet 1: 1160-1164.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

53. Hardcastle JD, Chamberlain JO, Robinson MH, Moss SM, Amar SS, Balfour TW, James<br />

PD, Mangham CM (1996). Randomised controlled trial of faecal-<strong>occult</strong>-<strong>blood</strong> <strong>screening</strong> <strong>for</strong><br />

colorectal cancer. Lancet 348: 1472-1477.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

54. Harris MA, Byles JE, Cockburn J, D'Este C (2000). A general practice-based recruitment<br />

strategy <strong>for</strong> colorectal cancer <strong>screening</strong>. Australian and New Zealand Journal of Public Health<br />

24(4): 441-443.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

55. Hart AR, Gay SP, Donnelly A, Griffin L, Inglis A, <strong>May</strong>berry MK, Wicks ACB, <strong>May</strong>berry JF<br />

(1994). Screening <strong>for</strong> colorectal cancer in Market Harborough, UK: A community-based<br />

program. European Journal of Gastroenterology & Hepatology 6: 519-522.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 81


56. Hart AR, Barone TL, Gay SP, Inglis A, Griffin L, Tallon CA, <strong>May</strong>berry JF (1997). The<br />

effect on compliance of a <strong>health</strong> education leaflet in colorectal cancer <strong>screening</strong> in general<br />

practice in central England. Journal of Epidemiology & Community Health 51: 187-191.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

57. Hastings JB (1974). Mass <strong>screening</strong> <strong>for</strong> colorectal cancer. American Journal of Surgery 127:<br />

228-233.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

58. Hisamichi S, Fukao A, Fujii Y, Tsuji I, Komatsu S, Inawashiro H, Tsubono Y (1991). Mass<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer in Japan. Cancer Detection and Prevention 15: 351-356.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

59. Holmes-Rovner M, Williams GA, Hoppough S, Quillan L, Butler R, Given CW (2002).<br />

Colorectal cancer <strong>screening</strong> barriers in persons with low income. Cancer Practice 10: 240-247.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

60. Hovendal CP, Kronborg O, Hem J, Grinsted P, Fenger C (1990). Rectoscopy and<br />

Hem<strong>occult</strong> II in irritable colon. A prospective study [Danish]. Ugeskrift <strong>for</strong> Laeger 152: 2732-<br />

2734.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

61. Howarth GF, Robinson MH, Jenkins D, Hardcastle JD, Logan RF (2002). High prevalence<br />

of undetected ulcerative colitis: data from the Nottingham fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> trial.<br />

American Journal of Gastroenterology 97: 690-694.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

62. Huicho L, Campos M, Rivera J, Guerrant RL (1996). Fecal <strong>screening</strong> tests in the approach<br />

to acute infectious diarrhea: a scientific overview. Pediatric Infectious Disease Journal 15(6): 486-<br />

494.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

63. Jani AL, Hamilos D (2003). Bloody diarrhea, fever, and pancytopenia in a patient with active<br />

ulcerative colitis. Annals of Allergy, Asthma, & Immunology 90: 383-388.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

64. Jensen J, Kewenter J, Asztely M, Lycke G, Wojciechowski J (1990). Double contrast barium<br />

enema and flexible rectosigmoidoscopy: a reliable diagnostic combination <strong>for</strong> detection of<br />

colorectal neoplasm. British Journal of Surgery 77: 270-272.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

65. Jorgensen OD, Kronborg O, Fenger C (2002). A randomised study of <strong>screening</strong> <strong>for</strong><br />

colorectal cancer using faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>: results after 13 years and seven biennial<br />

<strong>screening</strong> rounds. Gut 50: 29-32.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

66. Joseph A (1988). Compliance with fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>: the role of restrictive diets.<br />

American Journal of Public Health 78: 839-841.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

67. Kalra L, Hamlyn AN (1988). Comparative evaluation of investigations <strong>for</strong> colorectal<br />

carcinoma in symptomatic patients. Postgraduate Medical Journal 64: 666-668.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

82 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


68. Kato S, Ozawa K, Okuda M, Fujisawa T, Kagimoto S, Konno M, Maisawa S, Iinuma K<br />

(2003). Accuracy of the stool antigen test <strong>for</strong> the diagnosis of childhood Helicobacter pylori<br />

infection: A multicenter Japanese study. American Journal of Gastroenterology 98: 300.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

69. Kemppainen M, Hakkinen I, Raiha I, Pomoell R, Sourander L (1994). Finding colorectal<br />

tumours with an immunological faecal <strong>occult</strong> <strong>blood</strong> test in symptomatic primary <strong>health</strong> care<br />

patients. Age & Ageing 23: 365-370.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

70. Kettner JD, Whatrup C, Verne JE, Young K, Williams CB, Northover JM (1990). Is there a<br />

preference <strong>for</strong> different ways of per<strong>for</strong>ming faecal <strong>occult</strong> <strong>blood</strong> tests? [erratum appears in Int<br />

J Colorectal Dis 1990 3:176]. International Journal of Colorectal Disease 5: 82-86.<br />

Reason <strong>for</strong> exclusion: suitable outcome data not reported<br />

71. Kewenter J, Bjork S, Haglind E, Smith L, Svanvik J, Ahren C (1988). Screening and<br />

re<strong>screening</strong> <strong>for</strong> colorectal cancer. A controlled trial of fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> in 27,700<br />

subjects. Cancer 62: 645-651.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

72. Kewenter J, Engaras B, Haglind E, Jensen J (1990). Value of re<strong>testing</strong> subjects with a<br />

positive Hem<strong>occult</strong> in <strong>screening</strong> <strong>for</strong> colorectal cancer. British Journal of Surgery 77: 1349-1351.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

73. Kewenter J, Brevinge H, Engaras B, Haglind E, Ahren C (1994). Follow-up after <strong>screening</strong><br />

<strong>for</strong> colorectal neoplasms with fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> in a controlled trial. Diseases of the<br />

Colon & Rectum 37: 115-119.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

74. Kewenter J, Brevinge H, Engaras B, Haglind E, Ahren C (1994). Results of <strong>screening</strong>,<br />

re<strong>screening</strong>, and follow-up in a prospective randomized study <strong>for</strong> detection of colorectal<br />

cancer by fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>. Results <strong>for</strong> 68,308 subjects. Scandinavian Journal of<br />

Gastroenterology 29: 468-473.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

75. Kewenter J, Brevinge H (1996). Endoscopic and surgical complications of work-up in<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer. Diseases of the Colon & Rectum 39: 676-680.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

76. Klaaborg K, Madsen MS, Sondergaard O, Kronborg O (1986). Participation in mass<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer with fecal <strong>occult</strong> <strong>blood</strong> test. Scandinavian Journal of<br />

Gastroenterology 21: 1180-1184.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

77. Kristinsson J, Nygaard K, Aadland E, Barstad S, Sauar J, Hofstad B, Stray N, Stallemo A,<br />

Haug B, Ugstad M, Ton H, Fuglerud P (2001). Screening of first degree relatives of patients<br />

operated <strong>for</strong> colorectal cancer: evaluation of fecal calprotectin vs. hem<strong>occult</strong> II. Digestion 64:<br />

104-110.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

78. Kronborg O, Fenger C, Sondergaard O, Pedersen KM, Olsen J (1987). Initial mass<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer with fecal <strong>occult</strong> <strong>blood</strong> test. A prospective randomized study<br />

at Funen in Denmark. Scandinavian Journal of Gastroenterology 22: 677-686.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 83


79. Kronborg O, Fenger C, Olsen J, Bech K, Sondergaard O (1989). Repeated <strong>screening</strong> <strong>for</strong><br />

colorectal cancer with fecal <strong>occult</strong> <strong>blood</strong> test. A prospective randomized study at Funen,<br />

Denmark. Scandinavian Journal of Gastroenterology 24: 599-606.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

80. Kronborg O, Fenger C, Worm J, Pedersen SA, Hem J, Bertelsen K, Olsen J (1992). Causes<br />

of death during the first 5 years of a randomized trial of mass <strong>screening</strong> <strong>for</strong> colorectal cancer<br />

with fecal <strong>occult</strong> <strong>blood</strong> test. Scandinavian Journal of Gastroenterology 27: 47-52.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

81. Kronborg O (1993). [Screening interval <strong>for</strong> intestinal cancer with Hem<strong>occult</strong> II every other<br />

year]. [Danish]. Ugeskrift <strong>for</strong> Laeger 155: 1457-1459.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

82. Kronborg O, Fenger C, Olsen J, Jorgensen OD, Sondergaard O (1996). Randomised study<br />

of <strong>screening</strong> <strong>for</strong> colorectal cancer with faecal-<strong>occult</strong>-<strong>blood</strong> test. Lancet 348: 1467-1471.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

83. Kronborg O, Fenger C, Olsen J, Jorgensen OD, Sondergaard O (1997). Randomized<br />

<strong>population</strong> study of <strong>screening</strong> <strong>for</strong> intestinal cancer with Hem<strong>occult</strong>-II [Danish]. Ugeskrift <strong>for</strong><br />

Laeger 159: 4977-4981.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

84. Kroser JA, Faigel DO, Furth EE, Metz DC (1998). Comparison of rapid office-based<br />

serology with <strong>for</strong>mal laboratory-based ELISA <strong>testing</strong> <strong>for</strong> diagnosis of Helicobacter pylori<br />

gastritis. Digestive Diseases & Sciences 43: 108.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

85. Lang CA, Ransohoff DF (1994). Fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer: Is<br />

mortality reduced by chance selection <strong>for</strong> <strong>screening</strong> colonoscopy? Journal of the American<br />

Medical Association 27: 1011-1013.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

86. Lee CY (1990). A randomized controlled trial to motivate worksite fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

[Korean]. Kanho Hakhoe Chi [Journal of Nurses Academic Society] 20: 300-306.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

87. Lee CY (1991). A randomized controlled trial to motivate worksite fecal <strong>occult</strong> <strong>blood</strong><br />

<strong>testing</strong>. Yonsei Medical Journal 32: 131-138.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

88. Lee FI, Costello FT (1982). Assessment of Fecatest and Hem<strong>occult</strong> <strong>for</strong> faecal <strong>occult</strong> <strong>blood</strong><br />

<strong>testing</strong>. British Medical Journal Clinical Research Edition 285: 93.<br />

Reason <strong>for</strong> exclusion: inadequate data separation<br />

89. Leese PT, White KD, Frampton M, Baker JS, Cocchetto DM (1990). Absence of increased<br />

fecal <strong>blood</strong> loss in adult volunteers after oral administration of conventional tablets and<br />

osmotic tablets of albuterol. Current Therapeutic Research Clinical Exp. 48: 440-450.<br />

Reason <strong>for</strong> exclusion: wrong usage<br />

90. Lewis RJ, Lerman SE, Schnatter AR, Hughes JI, Vernon SW (1994). Colorectal polyp<br />

incidence among polypropylene manufacturing workers. Journal of Occupational Medicine 36:<br />

174-181.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

84 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


91. Li S, Nie Z, Li N, Li J, Zhang P, Yang Z, Mu S, Du Y, Hu J, Yuan S, Qu H, Zhang T, Wang<br />

S, Dong E, Qi D (2000). Colorectal cancer <strong>screening</strong> <strong>for</strong> the natural <strong>population</strong> of Beijing<br />

with sequential fecal <strong>occult</strong> <strong>blood</strong> test: A multicenter study. Chinese Medical Journal 116: 0-202.<br />

Reason <strong>for</strong> exclusion: inadequate data separation<br />

92. Lieberman DA, Weiss DG (2001). One-time <strong>screening</strong> <strong>for</strong> colorectal cancer with combined<br />

fecal <strong>occult</strong>-<strong>blood</strong> <strong>testing</strong> and examination of the distal colon. New England Journal of Medicine<br />

345: 555-560.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

93. Lieberman DA, Weiss DG, Hoey J, Wooltorton E (2001). Colorectal cancer <strong>screening</strong>: You<br />

can't be positive about a negative result. Canadian Medical Association Journal 165: 1248.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

94. Lindholm E, Berglund B, Haglind E, Kewenter J (1995). Factors associated with<br />

participation in <strong>screening</strong> <strong>for</strong> colorectal cancer with faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>. Scandinavian<br />

Journal of Gastroenterology 30: 171-176.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

95. Little J, Logan RF, Hawtin PG, Hardcastle JD, Turner ID (1993). Colorectal adenomas and<br />

diet: a case-control study of subjects participating in the Nottingham faecal <strong>occult</strong> <strong>blood</strong><br />

<strong>screening</strong> program. British Journal of Cancer 67: 177-184.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

96. Litzelman DK, Dittus RS, Miller ME, Tierney WM (1993). Requiring physicians to respond<br />

to computerized reminders improves their compliance with preventive care protocols.<br />

Journal of General Internal Medicine 8: 311-317.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

97. Liu H-H, Huang TW, Chen H-L, Wang T-H, Lin J-T (2003). Clinicopathologic significance<br />

of immunohistochemical fecal <strong>occult</strong> <strong>blood</strong> test in subjects receiving bidirectional<br />

endoscopy. Hepato-Gastroenterology 50: 1390-1392.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

98. Logan RF, Little J, Hawtin PG, Hardcastle JD (1993). Effect of aspirin and non-steroidal<br />

anti-inflammatory drugs on colorectal adenomas: case-control study of subjects participating<br />

in the Nottingham faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> program. British Medical Journal 307: 285-<br />

289.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

99. Lopez PJ, Nieto EM, Rodriguez AC, Molinero MJ, Olmo DG, Albero J (2000). Economic<br />

evaluation of colorectal cancer <strong>screening</strong> with fecal <strong>occult</strong> <strong>blood</strong> detection. Revista Espanola<br />

de Enfermedades Digestivas 92(5): 342-348.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

100. Lurie JD, Welch HG (1999). Diagnostic <strong>testing</strong> following fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> in the<br />

elderly. Journal of the National Cancer Institute 91: 1641-1646.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

101. Luthgens K, Maier A, Kampert I, Sieg A, Schmidt-Gayk H (1998). Hemoglobinhaptoglobin-complex:<br />

A highly sensitive assay <strong>for</strong> the detection of fecal <strong>occult</strong> <strong>blood</strong>. Clinical<br />

Laboratory 44: 543-551.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 85


102. Lynch NM, McHutchison JG, Young GP, Deacon M, St J, Barraclough D (1989).<br />

Gastrointestinal <strong>blood</strong> loss from a new buffered aspirin (Ostoprin): measurement by<br />

radiochromium and Hemoquant techniques. Australian & New Zealand Journal of Medicine 19:<br />

89-96.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

103. Malaty HM, Logan ND, Graham DY, Ramchatesingh JE, Reddy SG (2000). Helicobacter<br />

pylori infection in asymptomatic children: comparison of diagnostic tests. Helicobacter 5: 155-<br />

159.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

104. Mandel JS, Bond JH, Bradley M, Snover DC, Church TR, Williams S, Watt G, Schuman<br />

LM, Ederer F, Gilbertsen V (1989). Sensitivity, specificity, and positive predictivity of the<br />

Hem<strong>occult</strong> test in <strong>screening</strong> <strong>for</strong> colorectal cancers. The University of Minnesota's Colon<br />

Cancer Control Study. Gastroenterology 97: 597-600.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

105. Mandel JS, Bond JH, Church TR, Snover DC, Bradley GM, Schuman LM, Ederer F (1993).<br />

Reducing mortality from colorectal cancer by <strong>screening</strong> <strong>for</strong> fecal <strong>occult</strong> <strong>blood</strong>. Minnesota<br />

Colon Cancer Control Study.[comment][erratum appears in N Engl J Med 1993 329:672].<br />

New England Journal of Medicine 328: 1365-1371.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

106. Mandel JS, Church TR, Ederer F, Bond JH (1999). Colorectal cancer mortality: effectiveness<br />

of biennial <strong>screening</strong> <strong>for</strong> fecal <strong>occult</strong> <strong>blood</strong>. Journal of the National Cancer Institute 91: 434-437.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

107. Mandel JS, Church TR, Bond JH, Ederer F, Geisser MS, Mongin SJ, Snover DC, Schuman<br />

LM (2000). The effect of fecal <strong>occult</strong>-<strong>blood</strong> <strong>screening</strong> on the incidence of colorectal cancer.<br />

New England Journal of Medicine 343: 1603-1607.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

108. Mandel JS, Church TR, Bond JH (2001). <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> reduced the<br />

incidence of colorectal cancer. Evidence Based Medicine 6: 89.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

109. Mant D, Fitzpatrick R, Hogg A, Fuller A, Farmer A, Verne J, Northover J (1990).<br />

Experiences of patients with false positive results from colorectal cancer <strong>screening</strong>. British<br />

Journal of General Practice 40: 423-425.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

110. Mant D, Fuller A, Northover J, Astrop P, Chivers A, Crockett A, Clements S, Lawrence M<br />

(1992). Patient compliance with colorectal cancer <strong>screening</strong> in general practice. British Journal<br />

of General Practice 42: 18-20.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

111. Manus B, Bragelmann R, Armbrecht U, Stolte M, Stockbrugger RW (1996). Screening <strong>for</strong><br />

gastrointestinal neoplasia: efficacy and cost of two different approaches in a clinical<br />

rehabilitation centre. European Journal of Cancer Prevention 5(1): 49-55.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

112. Mapp TJ, Hardcastle JD, Moss SM, Robinson MH (1999). Survival of patients with<br />

colorectal cancer diagnosed in a randomized controlled trial of faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong>.<br />

British Journal of Surgery 86: 1286-1291.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

86 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


113. Moller JB, Kronborg O, Fenger C (1992). Interval cancers in <strong>screening</strong> with fecal <strong>occult</strong><br />

<strong>blood</strong> test <strong>for</strong> colorectal cancer. Scandinavian Journal of Gastroenterology 27: 779-782.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

114. Moran A, Husband D, Jones AF, Asquith P (1995). Diagnostic value of a guaiac <strong>occult</strong><br />

<strong>blood</strong> test and faecal alpha 1-antitrypsin. Gut 36: 87-89.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

115. Moss SM, Hardcastle JD, Coleman DA, Robinson MH, Rodrigues VC (1999). Interval<br />

cancers in a randomized controlled trial of <strong>screening</strong> <strong>for</strong> colorectal cancer using a faecal<br />

<strong>occult</strong> <strong>blood</strong> test. International Journal of Epidemiology 28: 386-390.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

116. Myers RE, Trock BJ, Lerman C, Wolf T, Ross E, Engstrom PF (1990). Adherence to<br />

colorectal cancer <strong>screening</strong> in an HMO <strong>population</strong>. Preventive Medicine 19: 502-514.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

117. Myers RE, Ross EA, Wolf TA, Balshem A, Jepson C, Millner L (1991). Behavioral<br />

interventions to increase adherence in colorectal cancer <strong>screening</strong>. Medical Care 29: 1039-<br />

1050.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

118. Myers RE, Balshem AM, Wolf TA, Ross EA, Millner L (1993). Adherence to continuous<br />

<strong>screening</strong> <strong>for</strong> colorectal neoplasia. Medical Care 31: 508-519.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

119. Myers RE, Ross E, Jepson C, Wolf T, Balshem A, Millner L, Leventhal H (1994). Modeling<br />

adherence to colorectal cancer <strong>screening</strong>.[erratum appears in Prev Med 1994 23:545]. Preventive<br />

Medicine 23: 142-151.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

120. Myers RE, Turner B, Weinberg D, Hauck WW, Hyslop T, Brigham T, Rothermel T, Grana<br />

J, Schlackman N (2001). Complete diagnostic evaluation in colorectal cancer <strong>screening</strong>:<br />

research design and baseline findings. Preventive Medicine 33: 249-260.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

121. Nakama H, Abdul F, Zhang B, Kamijo N, Fujimori K, Miyata K (1997). Detection rate of<br />

immunochemical fecal <strong>occult</strong> <strong>blood</strong> test <strong>for</strong> colorectal adenomatous polyps with severe<br />

dysplasia. Journal of Gastroenterology 32: 492-495.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

122. Nakama H, Kamijo N, Miyata K, Abdul Fattah ASM, Zhang B, Uehara Y (1998). Sensitivity<br />

and specificity of several immunochemical tests <strong>for</strong> colerectal cancer. Hepato-Gastroenterology<br />

45: 1579-1582.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

123. Nakama H, Kayano T, Katsuura T, Kamigaito T, Shimada S, Nishikawa N, Yoshii S,<br />

Kamijo N (1999). Comparison of predictive value <strong>for</strong> colorectal cancer in subjects with and<br />

without rectal bleeding. Hepato-Gastroenterology 46: 1730-1732.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

124. Nakama H, Yamamoto M, Kamijo N, Li T, Wei N, Fattah ASMA, Zhang B (1999).<br />

Colonoscopic evaluation of immunochemical fecal <strong>occult</strong> <strong>blood</strong> test <strong>for</strong> detection of<br />

colorectal neoplasia. Hepato-Gastroenterology 46: 228-231.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 87


125. Nakama H, Fattah A, Zhang B, Uehara Y, Wang C (2000). A comparative study of<br />

immunochemical fecal tests <strong>for</strong> detection of colorectal adenomatous polyps. Hepato-<br />

Gastroenterology 47: 386-389.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

126. Nakama H, Zhang B, Fattah ASMA (2000). A cost-effective analysis of the optimum<br />

number of stool specimens collected <strong>for</strong> immunochemical <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong><br />

colorectal cancer. European Journal of Cancer 36(5): 647-650.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

127. Nakama H, Fattah AS, Zhang B, Kamijo N (2000). Digital rectal examination sampling of<br />

stool is less predictive of significant colorectal pathology than stool passed spontaneously.<br />

European Journal of Gastroenterology and Hepatology 12: 1235-1238.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

128. Nakama H, Zhang B (2001). Diagnostic reliability of immunochemical faecal <strong>occult</strong> <strong>blood</strong><br />

test <strong>for</strong> small colorectal flat adenomas [1]. Digestive & Liver Disease 33: 301.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

129. Nakama H, Zhang B, Zhang X (2001). Evaluation of the optimum cut-off point in<br />

immunochemical <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> in <strong>screening</strong> <strong>for</strong> colorectal cancer. European Journal of<br />

Cancer 37(3): 398-401.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

130. Nichols S, Koch E, Lallemand RC, Heald RJ, Izzard L, Machin D, Mullee MA (1986).<br />

Randomised trial of compliance with <strong>screening</strong> <strong>for</strong> colorectal cancer. British Medical Journal<br />

Clinical Research Edition 293: 107-110.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

131. Niv Y, Lev-El M, Fraser G, Abuksis G, Tamir A (2002). Protective effect of faecal <strong>occult</strong><br />

<strong>blood</strong> test <strong>screening</strong> <strong>for</strong> colorectal cancer: worse prognosis <strong>for</strong> <strong>screening</strong> refusers. Gut 50:<br />

33-37.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

132. Nivatvongs S, Gilbertsen VA, Goldberg SM, Williams SE (1982). Distribution of largebowel<br />

cancers detected by <strong>occult</strong> <strong>blood</strong> test in asymptomatic patients. Diseases of the Colon &<br />

Rectum 25: 420-421.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

133. Odes HS, Rozen P, Ron E, Bass D, Bat L, Keren S, Fireman Z, Shemesh E, Krugliak P,<br />

Fraser G, et al (1992). Screening <strong>for</strong> colorectal neoplasia: a multicenter study in Israel. Israel<br />

Journal of Medical Sciences 28: 21-28.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

134. Ohlsson B, Breland U, Ekberg H, Graffner H, Tranberg KG (1995). Follow-up after<br />

curative surgery <strong>for</strong> colorectal carcinoma. Randomized comparison with no follow-up.<br />

Diseases of the Colon & Rectum 38: 619-626.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

135. Olsen J, Kronborg O (1993). Coffee, tobacco and alcohol as risk factors <strong>for</strong> cancer and<br />

adenoma of the large intestine. International Journal of Epidemiology 22: 398-402.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

88 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


136. Ore L, Hagoel L, Lavi I, Rennert G (2001). Screening with faecal <strong>occult</strong> <strong>blood</strong> test (FOBT)<br />

<strong>for</strong> colorectal cancer: assessment of two methods that attempt to improve compliance.<br />

European Journal of Cancer Prevention 10: 251-256.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

137. Park SI, Saxe JC, Weesner RE (1993). Does use of the Coloscreen Self-Test improve patient<br />

compliance with fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong>? American Journal of Gastroenterology 88: 1391-<br />

1394.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

138. Parker MA, Robinson MH, Scholefield JH, Hardcastle JD (2002). Psychiatric morbidity and<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer. Journal of Medical Screening 9: 7-10.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

139. Plaskon PP, Fadden MJ (1995). Cancer <strong>screening</strong> utilization: is there a role <strong>for</strong> social work in<br />

cancer prevention? Social Work in Health Care 21: 59-70.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

140. Porschen R, Haack G (1997). Does fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> result in decreased mortality<br />

from large intestine cancer? [German]. Zeitschrift fur Gastroenterologie 35: 595-596.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

141. Powe BD (1995). Fatalism among elderly African Americans. Effects on colorectal cancer<br />

<strong>screening</strong>. Cancer Nursing 18: 385-392.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

142. Powe BD (2002). Promoting fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> in rural African American women.<br />

Cancer Practice 10: 139-146.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

143. Pye G, Christie M, Chamberlain JO, Moss SM, Hardcastle JD (1988). A comparison of<br />

methods <strong>for</strong> increasing compliance within a general practitioner based <strong>screening</strong> project <strong>for</strong><br />

colorectal cancer and the effect on practitioner workload. Journal of Epidemiology & Community<br />

Health 42: 66-71.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

144. Pye G, Marks CG, Martin S, Marks V, Jackson J, Hardcastle JD (1989). An evaluation of<br />

Fecatwin/Feca EIA; a faecal <strong>occult</strong> <strong>blood</strong> test <strong>for</strong> detecting colonic neoplasia. European<br />

Journal of Surgical Oncology 15: 446-448.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

145. Rae AJ, Cleator IGM (1994). The two-tier fecal <strong>occult</strong> <strong>blood</strong> test: Cost-effective <strong>screening</strong>.<br />

Canadian Journal of Gastroenterology 8: 362-368.<br />

Reason <strong>for</strong> exclusion: not an appropriate <strong>screening</strong> <strong>population</strong><br />

146. Rasmussen M, Kronborg O, Fenger C, Jorgensen OD (1999). Possible advantages and<br />

drawbacks of adding flexible sigmoidoscopy to hem<strong>occult</strong>-II in <strong>screening</strong> <strong>for</strong> colorectal<br />

cancer. A randomized study. Scandinavian Journal of Gastroenterology 34: 73-78.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

147. Rasmussen M, Kronborg O (2002). Upper gastrointestinal cancer in a <strong>population</strong>-based<br />

<strong>screening</strong> program with fecal <strong>occult</strong> <strong>blood</strong> test <strong>for</strong> colorectal cancer [summary <strong>for</strong> patients in<br />

J Fam Pract 2002 51:601]. Scandinavian Journal of Gastroenterology 37: 95-98.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 89


148. Rasmussen M, Fenger C, Kronborg O (2003). Diagnostic yield in a biennial Hem<strong>occult</strong>-II<br />

<strong>screening</strong> program compared to a once-only <strong>screening</strong> with flexible sigmoidoscopy and<br />

Hem<strong>occult</strong>-II. Scandinavian Journal of Gastroenterology 38: 114-118.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

149. Richardson CR (2001). Do dietary restrictions reduce fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> adherence?<br />

Journal of Family Practice 50: 1081.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

150. Robinson MH, Thomas WM, Pye G, Hardcastle JD, Mangham CM (1993). Is dietary<br />

restriction always necessary in Hem<strong>occult</strong> <strong>screening</strong> <strong>for</strong> colorectal neoplasia? European<br />

Journal of Surgical Oncology 19: 539-542.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

151. Robinson MH, Thomas WM, Hardcastle JD, Chamberlain J, Mangham CM (1993). Change<br />

towards earlier stage at presentation of colorectal cancer. British Journal of Surgery 80: 1610-<br />

1612.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

152. Robinson MH, Pye G, Thomas WM, Hardcastle JD, Mangham CM (1994). Hem<strong>occult</strong><br />

<strong>screening</strong> <strong>for</strong> colorectal cancer: the effect of dietary restriction on compliance. European<br />

Journal of Surgical Oncology 20: 545-548.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

153. Robinson MH, Moss SM, Hardcastle JD, Whynes DK, Chamberlain JO, Mangham CM<br />

(1995). Effect of re<strong>testing</strong> with dietary restriction in Hem<strong>occult</strong> <strong>screening</strong> <strong>for</strong> colorectal<br />

cancer. Journal of Medical Screening 2: 41-44.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

154. Robinson MH, Kronborg O, Williams CB, Bostock K, Rooney PS, Hunt LM, Hardcastle JD<br />

(1995). <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> and colonoscopy in the surveillance of subjects at high<br />

risk of colorectal neoplasia. British Journal of Surgery 82: 318-320.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

155. Robinson MH, Hardcastle JD, Moss SM, Amar SS, Chamberlain JO, Armitage NC,<br />

Scholefield JH, Mangham CM (1999). The risks of <strong>screening</strong>: data from the Nottingham<br />

randomised controlled trial of faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer. Gut 45:<br />

588-592.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

156. Robinson MHE, Marks CG, Farrands PA, Bostock K, Hardcastle JD (1996). Screening <strong>for</strong><br />

colorectal cancer with an immunological faecal <strong>occult</strong> <strong>blood</strong> test: 2-year follow-up. British<br />

Journal of Surgery 83: 500-501.<br />

Reason <strong>for</strong> exclusion: study duplication<br />

157. Rockey DC, Auslander A, Greenberg PD (1999). Detection of upper gastrointestinal <strong>blood</strong><br />

with fecal <strong>occult</strong> <strong>blood</strong> tests. American Journal of Gastroenterology 94: 344-350.<br />

Reason <strong>for</strong> exclusion: wrong usage<br />

158. Rodney WM, Ruggiero C (1985). The coloscreen self-test <strong>for</strong> detection of fecal <strong>occult</strong> <strong>blood</strong>.<br />

Journal of Family Practice 21: 200-204.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

90 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


159. Roseth AG, Aadland E, Jahnsen J, Raknerud N (1997). Assessment of disease activity in<br />

ulcerative colitis by faecal calprotectin, a novel granulocyte marker protein. Digestion 58: 176-<br />

180.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

160. Rozen P, Knaani J, Samuel Z (1999). Eliminating the need <strong>for</strong> dietary restrictions when<br />

using a sensitive guaiac fecal <strong>occult</strong> <strong>blood</strong> test. Digestive Diseases and Sciences 44: 756-760.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

161. Saito H, Tsuchida S, Nakaji S, Kakizaki R, Aisawa T, Munakata A, Yoshida Y (1985). An<br />

immunologic test <strong>for</strong> fecal <strong>occult</strong> <strong>blood</strong> by counter immunoelectrophoresis. Higher<br />

sensitivity and higher positive reactions in colorectal cancer than single radial<br />

immunodiffusion and hem<strong>occult</strong> test. Cancer 56: 1549-1552.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

162. Saito H, Soma Y, Nakajima M, Koeda J, Kawaguchi H, Kakizaki R, Chiba R, Aisawa T,<br />

Munakata A (2000). A case-control study evaluating <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal<br />

cancer with hem<strong>occult</strong> test and an immunochemical hemagglutination test. Oncology Reports 7:<br />

815-819.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

163. Saitoh O, Matsumoto H, Sugimori K, Sugi K, Nakagawa K, Miyoshi H, Hirata I, Matsuse R,<br />

Uchida K, Ohshiba S (1995). Intestinal protein loss and bleeding assessed by fecal<br />

hemoglobin, transferrin, albumin, and alpha-1-antitrypsin levels in patients with colorectal<br />

diseases. Digestion 56: 67-75.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

164. Saitoh O, Kojima K, Kayazawa M, Sugi K, Tanaka S, Nakagawa K, Teranishi T, Matsuse R,<br />

Uchida K, Morikawa H, Hirata I, Katsu K (2000). Comparison of tests <strong>for</strong> fecal lactoferrin<br />

and fecal <strong>occult</strong> <strong>blood</strong> <strong>for</strong> colorectal diseases: a prospective pilot study. Internal Medicine 39:<br />

778-782.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

165. Salkeld G, Young G, Irwig L, Haas M, Glasziou P (1996). Cost-effectiveness analysis of<br />

<strong>screening</strong> by faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> colorectal cancer in Australia. Australian & New<br />

Zealand Journal of Public Health 20: 138-143.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

166. Santucci L, Fiorucci S, Pelli AM, Sicilia A, Morelli A, Patoia L (1991). - Macromolecular<br />

encapsulation as a protective factor against gastric disorders caused by non-steroidal antiinflammatory<br />

drugs. Journal of Drug Development 4(Suppl): 27-37.<br />

Reason <strong>for</strong> exclusion: wrong usage<br />

167. Schoemaker D, Toouli J, Black R, Watts J, Sarre R, and Rich C (1996). Colorectal cancer<br />

follow-up: are colonoscopy, CT liver and chest x-ray worthwhile? Abstract. International<br />

Journal of Colorectal Disease 11: 129.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

168. Scholefield JH, Robinson MHE, Mangham CM, Hardcastle JD (1998). Screening <strong>for</strong><br />

colorectal cancer reduces emergency admissions. European Journal of Surgical Oncology 24: 47-<br />

50.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 91


169. Scholefield JH, Moss S, Sufi F, Mangham CM, Hardcastle JD (2002). Effect of faecal <strong>occult</strong><br />

<strong>blood</strong> <strong>screening</strong> on mortality from colorectal cancer: results from a randomised controlled<br />

trial. Gut 50: 840-844.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

170. Siba S (1980). Results of Hem<strong>occult</strong> <strong>screening</strong> in Hungary (multicenter studies) [Hungarian].<br />

Orvosi Hetilap 121: 1701-1703.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

171. Sieg A, Scheida M, John MR, Hertel A, Schroter M, Luthgens K, Gayk H (1998). Validity of<br />

new immunological human fecal hemoglobin and albumin tests in detecting colorectal<br />

neoplasms - An endoscopy-controlled study. Zeitschrift fur Gastroenterologie 36: 485-490.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

172. Sieg A, Thoms C, Luthgens K, John MR, Schmidt-Gayk H (1999). Detection of colorectal<br />

neoplasms by the highly sensitive hemoglobin-haptoglobin complex in feces. International<br />

Journal of Colorectal Disease 14: 267-271.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

173. Skaife P, Seow-Choen F, Eu KW, Tang CL (2003). A novel indicator <strong>for</strong> surveillance<br />

colonoscopy following colorectal cancer resection. Colorectal Disease 5: 48.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

174. St John DJ, Young GP, McHutchison JG, Deacon MC, Alexeyeff MA (1992). Comparison<br />

of the specificity and sensitivity of Hem<strong>occult</strong> and HemoQuant in <strong>screening</strong> <strong>for</strong> colorectal<br />

neoplasia. Annals of Internal Medicine 117: 376-382.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

175. St John DJB, Young GP (2003). Is There a Need to Restrict Diet in Occult Blood Screening<br />

<strong>for</strong> Colorectal Cancer? [abstract]. Gastroenterology 102: A402.<br />

Reason <strong>for</strong> exclusion: suitable outcome data not reported<br />

176. Staib L, Link KH, Beger HG (2000). Follow-up in colorectal cancer: cost-effectiveness<br />

analysis of established and novel concepts. Langenbecks Archives of Surgery 385: 412-420.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

177. Struewing, JP, Pape DM, Snow DA (1991). Improving colorectal cancer <strong>screening</strong> in a<br />

medical residents' primary care clinic. American Journal of Preventive Medicine 7: 75-81.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

178. Tape TG, Campbell JR (1993). Computerized medical records and preventive <strong>health</strong> care:<br />

success depends on many factors. American Journal of Medicine 94: 619-625.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

179. Thomas W, White CM, Mah J, Geisser MS, Church TR, Mandel JS (1995). Longitudinal<br />

compliance with annual <strong>screening</strong> <strong>for</strong> fecal <strong>occult</strong> <strong>blood</strong>. Minnesota Colon Cancer Control<br />

Study. American Journal of Epidemiology 142: 176-182.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

180. Thomas WM, Pye G, Hardcastle JD, Chamberlain J, Charnley RM (1989). Role of dietary<br />

restriction in Hem<strong>occult</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer. British Journal of Surgery 76: 976-978.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

92 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


181. Thomas WM, Pye G, Hardcastle JD, Mangham CM (1990). <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>screening</strong><br />

<strong>for</strong> colorectal neoplasia: a randomized trial of three days or six days of tests. British Journal of<br />

Surgery 77: 277-279.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

182. Thomas WM, Pye G, Hardcastle JD, Walker AR (1992). Screening <strong>for</strong> colorectal carcinoma:<br />

an analysis of the sensitivity of haem<strong>occult</strong>. British Journal of Surgery 79: 833-835.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

183. Thompson NJ, Boyko EJ, Dominitz JA, Belcher DW, Chesebro BB, Stephens LM, Chapko<br />

MK (2000). A randomized controlled trial of a clinic-based support staff intervention to<br />

increase the rate of fecal <strong>occult</strong> <strong>blood</strong> test ordering. Preventive Medicine 30: 244-251.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

184. Thompson RS, Michnich ME, Gray J, Friedlander L, Gilson B (1986). Maximizing<br />

compliance with hem<strong>occult</strong> <strong>screening</strong> <strong>for</strong> colon cancer in clinical practice. Medical Care 24:<br />

904-914.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

185. Tilley BC, Vernon SW, Myers R, Glanz K, Lu M, Sanders K, Smereka C (1995). Planning<br />

the next step. A <strong>screening</strong> promotion and nutrition intervention trial in the work site. Annals<br />

of the New York Academy of Sciences 768: 296-299.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

186. Tilley BC, Vernon SW, Glanz K, Myers R, Sanders K, Lu M, Hirst K, Kristal AR, Smereka<br />

C, Sowers MF (1997). Worksite cancer <strong>screening</strong> and nutrition intervention <strong>for</strong> high-risk<br />

auto workers: design and baseline findings of the Next Step Trial. Preventive Medicine 26: 227-<br />

235.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

187. Tilley BC, Vernon SW, Myers R, Glanz K, Lu M, Hirst K, Kristal AR (1999). The Next Step<br />

Trial: impact of a worksite colorectal cancer <strong>screening</strong> promotion program. Preventive Medicine<br />

28: 276-283.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

188. Vaananen P, Tenhunen R (1988). Rapid immunochemical detection of fecal <strong>occult</strong> <strong>blood</strong> by<br />

use of a latex-agglutination test. Clinical Chemistry 34: 1763-1766.<br />

Reason <strong>for</strong> exclusion: no appropriate reference standard<br />

189. Verne J, Kettner J, Mant D, Farmer A, Mortenson N, Northover J (1993). Self-administered<br />

faecal <strong>occult</strong> <strong>blood</strong> tests do not increase compliance with <strong>screening</strong> <strong>for</strong> colorectal cancer:<br />

results of a randomized controlled trial. European Journal of Cancer Prevention 2: 301-305.<br />

Reason <strong>for</strong> exclusion: suitable outcome data not reported<br />

190. Verne JE, Aubrey R, Love SB, Talbot IC, Northover JM (1998). Population based<br />

randomized study of uptake and yield of <strong>screening</strong> by flexible sigmoidoscopy compared with<br />

<strong>screening</strong> by faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>. British Medical Journal 317: 182-185.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

191. Vinker S, Nakar S, Rosenberg E, Kitai E (2002). The role of family physicians in increasing<br />

annual fecal <strong>occult</strong> <strong>blood</strong> test <strong>screening</strong> coverage: a prospective intervention study. Israel<br />

Medical Association Journal 4: 424-425.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 93


192. Walker AR, Whynes DK, Hardcastle JD (1991). Rehydration of guaiac-based faecal <strong>occult</strong><br />

<strong>blood</strong> tests in mass <strong>screening</strong> <strong>for</strong> colorectal cancer. An economic perspective. Scandinavian<br />

Journal of Gastroenterology 26: 215-218.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

193. Walter SD, Frommer DJ, Cook RJ (1991). The estimation of sensitivity and specificity in<br />

colorectal cancer <strong>screening</strong> methods. Cancer Detection and Prevention 15: 465-469.<br />

Reason <strong>for</strong> exclusion: study duplication<br />

194. Wardle J, Taylor T, Sutton S, Atkin W (1999). Does publicity about cancer <strong>screening</strong> raise<br />

fear of cancer? Randomised trial of the psychological effect of in<strong>for</strong>mation about cancer<br />

<strong>screening</strong>. British Medical Journal 319: 1037-1038.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

195. Wechselberger F, Sfetsos G (1979). Field study on a <strong>health</strong>y, fit group of workers aged 40-60<br />

yr <strong>for</strong> early detection of colorectal carcinoma with the 'haem<strong>occult</strong>' test [German].<br />

Arbeitsmedizin Sozialmedizin Praventivmedizin 14: 154-156.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

196. Weinrich SP, Weinrich MC, Stromborg MF, Boyd MD, Weiss HL (1993). Using elderly<br />

educators to increase colorectal cancer <strong>screening</strong>. Gerontologist 33: 491-496.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

197. Weinrich SP, Weinrich MC, Boyd MD, Atwood J, Cervenka B (1994). Teaching older adults<br />

by adapting <strong>for</strong> aging changes. Cancer Nursing 17: 494-500.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

198. Weller D, Thomas D, Hiller J, Woodward A, Edwards J (1994). Screening <strong>for</strong> colorectal<br />

cancer using an immunochemical test <strong>for</strong> faecal <strong>occult</strong> <strong>blood</strong>: Results of the first 2 years of a<br />

South Australian program. Australian & New Zealand Journal of Surgery 64: 464-469.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

199. Weller D, Moss J, Hiller J, Thomas J, Edwards J (1995). Screening <strong>for</strong> colorectal cancer:<br />

what are the costs? International Journal of Technology Assessment in Health Care 11: 26-39.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

200. Whynes DK, Walker AR, Hardcastle JD (1992). Effect of subject age on costs of <strong>screening</strong><br />

<strong>for</strong> colorectal cancer. Journal of Epidemiology & Community Health 46: 577-581.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

201. Whynes DK, Neilson AR, Robinson MH, Hardcastle JD (1994). Colorectal cancer <strong>screening</strong><br />

and quality of life. Quality of Life Research 3: 191-198.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

202. Whynes DK, Neilson AR, Walker AR, Hardcastle JD (1998). <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>screening</strong><br />

<strong>for</strong> colorectal cancer: is it cost-effective? Health Economics 7: 21-29.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

203. Whynes DK, Frew EJ, Manghan CM, Scholefield JH, Hardcastle JD (2003). Colorectal<br />

cancer, <strong>screening</strong> and survival: The influence of socio-economic deprivation. Public Health<br />

117: 389-395.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

94 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


204. Winawer SJ, Andrews M, Flehinger B, Sherlock P, Schottenfeld D, Miller DG (1980).<br />

Progress report on controlled trial of fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> the detection of<br />

colorectal neoplasia. Cancer 45: 2959-2964.<br />

Reason <strong>for</strong> exclusion: inadequate data separation<br />

205. Winawer SJ, Flehinger BJ, Schottenfeld D, Miller DG (1993). Screening <strong>for</strong> colorectal cancer<br />

with fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> and sigmoidoscopy. Journal of the National Cancer Institute 85:<br />

1311-1318.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

206. Winickoff RN, Coltin KL, Morgan MM, Buxbaum RC, Barnett GO (1984). Improving<br />

physician per<strong>for</strong>mance through peer comparison feedback. Medical Care 22: 527-534.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

207. Wolf AM, Schorling JB (2000). Does in<strong>for</strong>med consent alter elderly patients' preferences <strong>for</strong><br />

colorectal cancer <strong>screening</strong>? Results of a randomized trial. Journal of General Internal Medicine<br />

15: 24-30.<br />

Reason <strong>for</strong> exclusion: wrong outcome<br />

208. Xing PX, Young G, McKenzie IFC (2001). Development of a fecal <strong>occult</strong> <strong>blood</strong> test using a<br />

monoclonal antibody to haptoglobin. Redox Report 6: 363-365.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

209. Yamamoto M, Nakama H (2000). Cost-effectiveness analysis of immunochemical <strong>occult</strong><br />

<strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer among three fecal sampling methods. Hepato-<br />

Gastroenterology 47: 396-399.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

210. Yokoyama S, Shatney CH, Mochizuki H, Hase K, Johnson DL, Cummings S, Trollope ML,<br />

Tamakuma S, Galandiuk S (1997). The potential role of fecal carbonic anhydrase II in<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer. American Surgeon 63: 243-246.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

211. Yoshinaga M, Motomura S, Takeda H, Yanagisawa Z, Ikeda K (1995). Evaluation of the<br />

sensitivity of an immunochemical fecal <strong>occult</strong> <strong>blood</strong> test <strong>for</strong> colorectal neoplasia. American<br />

Journal of Gastroenterology 90: 1076-1079.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

212. Zacks M (1997). Fecal-<strong>occult</strong>-<strong>blood</strong> test <strong>screening</strong> <strong>for</strong> colorectal cancer. Journal of Family<br />

Practice 44: 247-248.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

213. Zappa M, Castiglione G, Grazzini G, Falini P, Giorgi D, Paci E, Ciatto S (1997). Effect of<br />

faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> on colorectal mortality: results of a <strong>population</strong>-based case-control<br />

study in the district of Florence, Italy. International Journal of Cancer 73: 208-210.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

214. Zheng S, Chen K, Liu X, Ma X, Yu H, Chen K, Yao K, Zhou L, Wang L, Qiu P, Deng Y,<br />

Zhang S (2003). Cluster randomization trial of sequence mass <strong>screening</strong> <strong>for</strong> colorectal<br />

cancer. Diseases of the Colon & Rectum 46: 51-58.<br />

Reason <strong>for</strong> exclusion: not a head-to-head FOBT study<br />

215. Ziegler EE, Fomon SJ, Nelson SE, Rebouche CJ, Edwards BB, Rogers RR, Lehman LJ<br />

(1990). Cow milk feeding in infancy: further observations on <strong>blood</strong> loss from the<br />

gastrointestinal tract. Journal of Pediatrics 116: 11-18.<br />

Reason <strong>for</strong> exclusion: wrong patient group<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 95


Appendix F Positive predictive values<br />

Positive predictive value<br />

Table 42 and Table 43 present the results of meta-analyses comparing the positive<br />

predictive values (PPV) of individual guaiac and immunochemical faecal <strong>occult</strong> <strong>blood</strong><br />

tests (FOBTs). These meta-analyses are based on the results of the nine head-to-head<br />

studies of FOBTs identified in the literature search. PPV is the probability that if a<br />

person tests positive with a FOBT then that person actually has a carcinoma and/or an<br />

adenoma (ie, avoid false-positives). PPV does not incorporate the probability that a<br />

person who tests negative may actually have the condition (ie, false-negatives). This is<br />

difficult to avoid in <strong>screening</strong> studies where it is generally not acceptable to subject all<br />

participants to an invasive ‘reference standard’ such as colonoscopy.<br />

PPV is only a useful comparative measure of diagnostic per<strong>for</strong>mance between tests when<br />

the tests are applied to <strong>population</strong>s with the same prevalence of disease. This means that,<br />

in general, PPV is only useful <strong>for</strong> comparing test per<strong>for</strong>mance within a study. There<strong>for</strong>e,<br />

the most valid trial design <strong>for</strong> determining the comparative PPV of FOBTs is where both<br />

tests are applied to the same person in a ‘within-subjects’ analysis. It is only valid to<br />

compare PPV values in ‘between-subjects’ analysis (eg, a parallel arm study) when the<br />

<strong>population</strong>s undergoing each test carry the same spectrum of disease. This is most likely<br />

to occur in randomised controlled trials (RCTs). All of the studies included in this<br />

analysis employed a ‘within-subjects’ design.<br />

Table 42 and Table 43 include a meta-analysis where studies on HemeSelect and<br />

Immudia-HemSp are pooled, since these FOBTs are equivalent international versions of<br />

the same test (St John et al 1993; Zappa et al 2001).<br />

Table 42 presents the results of the meta-analyses <strong>for</strong> carcinoma alone. The table is split<br />

into two sections, one reporting the results of the main meta-analyses and the other<br />

reporting the results of a series of sensitivity meta-analyses. The main meta-analyses are<br />

strictly limited to the assessment of only those studies in which the participants came<br />

entirely from a <strong>screening</strong> <strong>population</strong> at average risk. The sensitivity meta-analyses<br />

includes four additional studies in which some of the participants were symptomatic or at<br />

increased risk (Rozen et al 1995; Rozen et al 1997; Rozen et al 2000; St John et al 1993).<br />

The immunochemical test HemeSelect had a greater PPV <strong>for</strong> carcinoma than Hem<strong>occult</strong><br />

Sensa (4.8% versus 2.4%, respectively). The PPV odds ratio (OR) <strong>for</strong> carcinoma <strong>for</strong><br />

HemeSelect versus Hem<strong>occult</strong> Sensa was significantly different and favoured the<br />

HemeSelect test (OR 2.10; 95% confidence interval (CI): 1.37, 3.26). The PPV risk<br />

difference (RD) was also significantly different in favour of the HemeSelect test (RD<br />

0.025; 95% CI: 0.009, 0.048). The difference in PPV between HemeSelect and<br />

Hem<strong>occult</strong> Sensa was maintained in the sensitivity analyses (4.9% versus 2.3%,<br />

respectively). The PPV OR <strong>for</strong> carcinoma <strong>for</strong> HemeSelect versus Hem<strong>occult</strong> Sensa in<br />

the sensitivity analysis was statistically different and favoured the HemeSelect test (OR<br />

2.19; 95% CI: 1.45, 3.31). The PPV RD was also statistically significant in favour of the<br />

HemeSelect test in the sensitivity analysis (RD 0.026; 95% CI: 0.012, 0.042). The PPVs<br />

<strong>for</strong> all of the other immunochemical to guaiac test comparisons did not differ<br />

significantly. In all of the main meta-analyses the level of heterogeneity between studies<br />

was low. In the sensitivity meta-analyses the level of heterogeneity between studies<br />

increases, reflecting the differences in study participants.<br />

96 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Table 43 presents the results of the meta-analyses <strong>for</strong> carcinoma or adenoma<br />

(neoplasms). The table is split into two sections, one reporting the results of the main<br />

meta-analyses and the other reporting the results of a series of sensitivity meta-analyses.<br />

The main meta-analyses are strictly limited to the assessment of only those studies in<br />

which the participants came entirely from a <strong>screening</strong> <strong>population</strong> at average risk. The<br />

sensitivity meta-analyses include four additional studies in which some of the participants<br />

were symptomatic or at increased risk (Rozen et al 1995; Rozen et al 1997; Rozen et al<br />

2000; St John et al 1993).<br />

The PPV <strong>for</strong> the detection of neoplasms (adenoma or carcinoma) <strong>for</strong> the reverse-passive<br />

haemagglutination (RPHA) tests (HemeSelect and Immudia-HemSp) was greater than<br />

that <strong>for</strong> Hem<strong>occult</strong> (15.5% vs 11.8%, respectively). The PPV OR <strong>for</strong> neoplasms<br />

(adenoma or carcinoma) <strong>for</strong> RPHA versus Hem<strong>occult</strong> was statistically significant and<br />

favoured RPHA (OR 1.38; 95% CI: 1.09, 1.75). Similarly, the PPV RD <strong>for</strong> neoplasms <strong>for</strong><br />

RPHA versus Hem<strong>occult</strong> was also statistically significant and favoured the Immudia-<br />

HemSp test (RD 0.038; 95% CI: 0.01, 0.065). However, the data showed a significant<br />

degree of heterogeneity.<br />

In contrast, the PPV <strong>for</strong> Fecatwin Sensitive/Feca EIA was lower than that <strong>for</strong><br />

Hem<strong>occult</strong> (25.5% vs 70.0%, respectively). The OR <strong>for</strong> neoplasm PPV <strong>for</strong> Hem<strong>occult</strong><br />

was statistically significant and favoured Hem<strong>occult</strong> (OR 0.15; 95% CI: 0.07, 0.328).<br />

Similarly, the RD <strong>for</strong> neoplasm PPV <strong>for</strong> Hem<strong>occult</strong> versus Fecatwin Sensitive/Feca EIA<br />

was different and favoured Hem<strong>occult</strong> (RD –0.445; 95% CI: –0.610, –0.281).<br />

The PPV <strong>for</strong> neoplasms was greater <strong>for</strong> HemeSelect than <strong>for</strong> Hem<strong>occult</strong> Sensa (17.7%<br />

vs 10.5%, respectively). The PPV OR <strong>for</strong> HemeSelect versus Hem<strong>occult</strong> Sensa was<br />

statistically significant and favoured HemeSelect (OR 1.83; 95% CI: 1.45, 2.28). The PPV<br />

RD <strong>for</strong> HemeSelect versus Hem<strong>occult</strong> Sensa was statistically significant and also<br />

favoured HemeSelect (RD 0.071; 95% CI: 0.037, 0.103).<br />

In the main meta-analyses the level of heterogeneity between the Hem<strong>occult</strong> and<br />

HemeSelect studies was higher than seen in the assessment of PPV <strong>for</strong> carcinoma. This<br />

is likely to be a reflection of the variation in definitions used to classify neoplasms in each<br />

of the included studies. The relatively high levels of heterogeneity between studies were<br />

also observed in the sensitivity meta-analyses.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 97


Table 42 Meta-analyses of PPV <strong>for</strong> carcinoma from FOBT head-to-head studies<br />

Heterogeneity<br />

(95% CI)<br />

Pooled results (95% CI)<br />

Pooled PPV %<br />

RD<br />

OR<br />

Immunochemical<br />

Guaiac<br />

n/N<br />

(True +ve/<br />

Total +ve)<br />

FOBT<br />

Studies<br />

FOBT<br />

comparison<br />

0.011<br />

(0.0001, 0.51)<br />

0.012<br />

(–0.005, 0.028)<br />

1.35<br />

(0.89; 2.04)<br />

RPHA<br />

5.4<br />

HO<br />

3.6<br />

–<br />

–0.028<br />

(–0.119, 0.063)<br />

0.61<br />

(0.14; 2.68)<br />

Fecatwin<br />

4.7<br />

HO<br />

7.5<br />

0.004<br />

(0.00009, 1.682)<br />

0.025<br />

(0.009, 0.048)<br />

2.10<br />

(1.37; 3.26)<br />

HemeSelect<br />

4.8<br />

HO Sensa<br />

2.4<br />

13/198<br />

22/440<br />

16/456<br />

21/438<br />

1/17<br />

9/145<br />

8/410<br />

16/391<br />

3/40<br />

5/106<br />

27/1073<br />

22/440<br />

19/849<br />

21/438<br />

HO<br />

HemeSelect<br />

HO<br />

HemeSelect<br />

Allison 1996<br />

Petrelli 1994c Main guaiac to immunochemical comparisons<br />

HO vs RPHA<br />

(pooled<br />

HemeSelect and<br />

Immudia-<br />

HemSp)<br />

HO<br />

HemeSelect<br />

Robinson<br />

1995a HO<br />

Immudia-HemSpb Iwase 1992<br />

HO<br />

Fecatwin<br />

Armitage<br />

1985<br />

HO vs Fecatwin<br />

HO Sensa<br />

HemeSelect<br />

Allison 1996<br />

HO Sensa vs<br />

HemeSelect<br />

HO Sensa<br />

HemeSelect<br />

13/198<br />

22/440<br />

HO<br />

HemeSelect<br />

16/456<br />

21/438<br />

HO<br />

HemeSelect<br />

0.204<br />

(0.0003, 3.91)<br />

0.013<br />

(–0.005, 0.036)<br />

1.34<br />

(0.90; 2.02)<br />

RPHA<br />

5.4<br />

HO<br />

4.1<br />

1/17<br />

9/145<br />

8/410<br />

16/391<br />

HO<br />

HemeSelect<br />

HO<br />

Immudia-HemSpb 3/24<br />

4/12<br />

HO<br />

HemeSelect<br />

98 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

Petrelli 1994 f<br />

Allison 1996<br />

Petrelli 1994 c<br />

Sensitivity analyses<br />

HO vs RPHA<br />

(pooled<br />

HemeSelect and<br />

Immudia-<br />

HemSp)<br />

Robinson<br />

1995a Iwase 1992<br />

Rozen 1997


Heterogeneity<br />

(95% CI)<br />

Pooled results (95% CI)<br />

Pooled PPV %<br />

RD<br />

OR<br />

Immunochemical<br />

Guaiac<br />

n/N<br />

(True +ve /<br />

Total +ve)<br />

FOBT<br />

Studies<br />

FOBT<br />

comparison<br />

27/1073<br />

22/440<br />

19/849<br />

21/438<br />

4.70<br />

(0.052, 2302)<br />

0.026<br />

(0.012, 0.042)<br />

2.19<br />

(1.45, 3.31)<br />

HemeSelect<br />

4.9<br />

HO Sensa<br />

2.3<br />

3/35<br />

4/12<br />

1/68<br />

1/41<br />

–<br />

0.110<br />

(–0.230, 0.352)<br />

2.15<br />

(0.41, 11.2)<br />

Flexsure OBT<br />

23.5<br />

HO<br />

12.5<br />

3/24<br />

4/17<br />

HO Sensa<br />

HemeSelect<br />

HO Sensa<br />

HemeSelect<br />

HO Sensa<br />

HemeSelect<br />

HO Sensa<br />

HemeSelect<br />

HO<br />

Flexsure OBT<br />

HO Sensa<br />

Flexsure OBT<br />

HO Sensa<br />

BM-Test Colon<br />

Sensitivity analyses (continued)<br />

HO Sensa vs Allison 1996<br />

HemeSelect<br />

Petrelli 1994 c<br />

Rozen 1997<br />

St John 1993<br />

Rozen 1997<br />

HO vs Flexsure<br />

OBT<br />

HO Sensa vs Rozen 2000<br />

3/74<br />

HO Sensa<br />

Flexsure OBT<br />

3.38<br />

0.08<br />

Flexsure OBT<br />

–<br />

3/24<br />

4.1<br />

12.5<br />

(0.63, 18.0) (–0.055, 0.224)<br />

HO Sensa vs Rozen 1995<br />

3/88<br />

BM-Test Colon<br />

HO Sensa BM-Test Colon Albumin<br />

2.06<br />

0.034<br />

4/59<br />

–<br />

Albumin<br />

3.4<br />

6.8<br />

(0.44, 9.56) (–0.041, 0.108)<br />

Albumin<br />

Abbreviations: CI, confidence interval; Fecatwin, Fecatwin Sensitive/Feca EIA; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; HO, Hem<strong>occult</strong>; HO Sensa, Hem<strong>occult</strong> Sensa; OR, odds ratio; PPV, positive predictive value; RD, risk difference;<br />

RPHA, reverse-passive haemagglutination.<br />

aData from Robinson et al (1996) are not used due to slight discrepancies between the publications and a degree of uncertainty surrounding the data. bOver 3 days. cTotal positives calculated from number of patients screened<br />

and percentage positivity rate as reported, authors were contacted <strong>for</strong> data confirmation but no response was received.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 99


Table 43 Meta-analyses of PPV <strong>for</strong> neoplasms (carcinoma or adenoma) from FOBT head-to-head<br />

studies<br />

Heterogeneity<br />

(95% CI)<br />

Pooled results (95% CI)<br />

Pooled PPV %<br />

RD<br />

OR<br />

Immunochemical<br />

Guaiac<br />

n/N<br />

(True +ve/<br />

Total +ve)<br />

FOBT<br />

Studies<br />

FOBT<br />

comparison<br />

46/198<br />

90/440<br />

HO<br />

HemeSelect<br />

60/456<br />

66/438<br />

HO<br />

HemeSelect<br />

Allison 1996a Petrelli 1994b Main guaiac to immunochemical comparisons<br />

HO vs RPHA<br />

(pooled<br />

HemeSelect and<br />

Immudia-<br />

HemSp)<br />

3.86<br />

(0.32, 2247)<br />

0.038<br />

(0.01, 0.065)<br />

1.38<br />

(1.09, 1.75)<br />

RPHA<br />

15.5<br />

HO<br />

11.8<br />

9/17<br />

57/145<br />

HO<br />

HemeSelect<br />

Robinson<br />

1995c 16/410<br />

44/391<br />

HO<br />

Immudia-HemSpe Iwase 1992 d<br />

–<br />

–0.445<br />

(–0.610, –0.281)<br />

0.15<br />

(0.07, 0.328)<br />

Fecatwin<br />

25.5<br />

HO<br />

70.0<br />

28/40<br />

27/106<br />

HO<br />

Fecatwin<br />

Armitage<br />

1985<br />

HO vs Fecatwin<br />

99/1073<br />

90/440<br />

HO Sensa<br />

HemeSelect<br />

Allison 1996 a<br />

HO Sensa vs<br />

HemeSelect<br />

0.002<br />

(0.00007, 1.113)<br />

0.071<br />

(0.037, 0.103)<br />

1.83<br />

(1.45, 2.28)<br />

HemeSelect<br />

17.7<br />

HO Sensa<br />

10.5<br />

104/849<br />

66/438<br />

HO Sensa<br />

HemeSelect<br />

46/198<br />

90/440<br />

HO<br />

HemeSelect<br />

60/456<br />

66/438<br />

HO<br />

HemeSelect<br />

2.53<br />

(0.20, 1205)<br />

0.061<br />

(0.026, 0.095)<br />

1.45<br />

(1.18, 1.88)<br />

RPHA<br />

22.2<br />

HO<br />

16.1<br />

9/17<br />

57/145<br />

HO<br />

HemeSelect<br />

16/410<br />

44/391<br />

HO<br />

Immudia-HemSpe 7/24<br />

8/12<br />

HO<br />

HemeSelect<br />

100 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong><br />

Petrelli 1994 b<br />

Allison 1996 a<br />

Petrelli 1994 b<br />

Sensitivity analyses<br />

HO vs RPHA<br />

(pooled<br />

HemeSelect and<br />

Immudia-<br />

HemSp)<br />

Robinson<br />

1995c Iwase 1992 d<br />

Rozen 1997


Heterogeneity<br />

(95% CI)<br />

Pooled results (95% CI)<br />

Pooled PPV %<br />

RD<br />

OR<br />

Immunochemical<br />

Guaiac<br />

n/N<br />

(True +ve/<br />

Total +ve)<br />

FOBT<br />

Studies<br />

FOBT<br />

comparison<br />

99/1073<br />

90/440<br />

104/849<br />

66/438<br />

1.72<br />

(0.12, 849)<br />

0.121<br />

(0.075, 0.170)<br />

1.83<br />

(1.48, 2.26)<br />

HemeSelect<br />

34.8<br />

HO Sensa<br />

22.6<br />

8/35<br />

8/12<br />

16/68<br />

11/41<br />

–<br />

0.061<br />

(0.230, 0.352)<br />

1.32<br />

(0.35, 5.00)<br />

Flexsure OBT<br />

35.3<br />

HO<br />

29.2<br />

7/24<br />

6/17<br />

HO Sensa<br />

HemeSelect<br />

HO Sensa<br />

HemeSelect<br />

HO Sensa<br />

HemeSelect<br />

HO Sensa<br />

HemeSelect<br />

HO<br />

Flexsure OBT<br />

HO Sensa<br />

Flexsure OBT<br />

HO Sensa<br />

BM-Test Colon<br />

Allison 1996a Sensitivity analyses (continued)<br />

HO Sensa vs<br />

HemeSelect<br />

Petrelli 1994 b<br />

Rozen 1997<br />

St John 1993<br />

Rozen 1997<br />

HO vs Flexsure<br />

OBT<br />

HO Sensa vs Rozen 2000<br />

19/74<br />

HO Sensa<br />

Flexsure OBT<br />

2.07<br />

0.160<br />

Flexsure OBT<br />

–<br />

10/24<br />

25.7<br />

41.7<br />

(0.79, 5.43) (0.061, 0.381)<br />

HO Sensa vs Rozen 1995<br />

21/88<br />

BM-Test Colon<br />

HO Sensa BM-Test Colon Albumin<br />

1.29<br />

0.049<br />

17/59<br />

–<br />

Albumin<br />

23.9<br />

28.8<br />

(0.61, 2.73) (–0.096, 0.195)<br />

Albumin<br />

Abbreviations: CI, confidence interval; Fecatwin, Fecatwin Sensitive/Feca EIA; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test; HO, Hem<strong>occult</strong>; HO Sensa, Hem<strong>occult</strong> Sensa; OR, odds ratio; PPV, positive predictive value; RD, risk difference;<br />

RPHA, reverse-passive haemagglutination.<br />

aAdenoma data are only <strong>for</strong> adenomas > 1cm. bTotal positives calculated from number of patients screened and percentage positivity rate as reported, authors were contacted <strong>for</strong> data confirmation however no response was<br />

received. cData from Robinson et al (1996) are not used due to slight discrepancies between the publications and a degree of uncertainty surrounding the data. dOver 3 days.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 101


Appendix G Diagnostic odds ratios<br />

Accuracy<br />

The diagnostic odds ratios (DORs) were determined <strong>for</strong> those tests <strong>for</strong> which the<br />

sensitivity and specificity were available, by utilising the interval cancer rate (Table 44).<br />

In addition, studies estimating the sensitivity and specificity in <strong>screening</strong> <strong>population</strong>s with<br />

a high proportion of participants who were at higher risk of disease, by subjecting all<br />

participants to the reference standard, were included in a sensitivity analysis (Rozen et al<br />

1995; Rozen et al 1997; Rozen et al 2000; St John et al 1993). However, in these studies<br />

the endoscopic examination had been per<strong>for</strong>med up to several years previously in a<br />

proportion of the patients <strong>testing</strong> negative with FOBT. There<strong>for</strong>e, these studies were still<br />

subject to verification bias. These studies have been included in a sensitivity analyses. In<br />

the sensitivity analyses the use of different reference standards (ie, colonoscopy plus<br />

interval cancer rates, versus colonoscopy in all subjects) is likely to lead to a threshold<br />

effect across studies.<br />

The heterogeneity in the DORs indicated that the pooled DOR estimate is likely to be<br />

imprecise. Nevertheless, the DOR was significantly higher <strong>for</strong> HemeSelect than<br />

Hem<strong>occult</strong> (DOR 0.384; 95% confidence interval (CI): 0.140, 0.993) and this difference<br />

was maintained in the sensitivity analysis.<br />

The DORs were also higher <strong>for</strong> the immunochemical tests in single head-to-head studies<br />

in the main analyses when Hem<strong>occult</strong> was compared with Fecatwin Sensitive/Feca EIA<br />

and when Hem<strong>occult</strong> Sensa was compared with HemeSelect. The difference in the<br />

DORs between Hem<strong>occult</strong> Sensa and HemeSelect was not significantly different in the<br />

sensitivity analysis.<br />

The practical impact of the relative DORs is not apparent from these data in isolation.<br />

The clinical significance of these trade-offs between sensitivity and specificity has been<br />

investigated via the use of an economic model in the cost-effectiveness section of this<br />

assessment report.<br />

A summary receiver-operating characteristic curve (SROC) was obtained <strong>for</strong> the<br />

Hem<strong>occult</strong> test from the Hem<strong>occult</strong> to HemeSelect sensitivity analyses by back<br />

trans<strong>for</strong>mation of a linear regression of the difference between logit (true positive rate,<br />

TPR) and logit (false negative rate, FNR) on the sum of the two logits (Figure 10). The<br />

equation <strong>for</strong> the curve was:<br />

Sensitivity =<br />

1+<br />

e<br />

1<br />

a<br />

1+<br />

b<br />

FPR<br />

1 −b<br />

1−b<br />

× ( )<br />

1−<br />

FPR<br />

where, FPR = 1 – specificity, a = 4.087515, and b = 0.262321.<br />

This SROC curve was used to determine alternative hypothetical pairs of sensitivity and<br />

specificity data <strong>for</strong> use in a sensitivity analyses in the economic model (Appendix K).<br />

102 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Table 44 Meta-analyses of diagnostic odds ratios <strong>for</strong> carcinoma from FOBT head-to-head studies<br />

Relative DORg (95% CI)<br />

Pooled DOR<br />

immunochemicalf Pooled DOR<br />

guaiacf DOR<br />

Specificity<br />

Sensitivity<br />

FOBT<br />

Studies<br />

FOBT comparison<br />

Main guaiac to immunochemical comparisons<br />

HO vs HemeSelect Allison 1996<br />

0.384<br />

(0.140, 0.993)<br />

HemeSelect:<br />

45.7<br />

HO:<br />

17.7<br />

25.1<br />

37.1<br />

0.9<br />

97.7<br />

94.4<br />

37.1<br />

68.8<br />

HO<br />

HemeSelect<br />

–<br />

Fecatwin:<br />

59.3<br />

HO:<br />

34.1<br />

–<br />

HemeSelect:<br />

37.1<br />

HO Sensa:<br />

25.2<br />

210.6e 34.1<br />

59.3<br />

25.2<br />

37.1<br />

98.9<br />

90.9<br />

9.1a 100<br />

HO<br />

HemeSelect<br />

Robinson 1996<br />

97.2<br />

92.2<br />

50.0<br />

83.3<br />

86.7<br />

94.4<br />

79.4<br />

68.8<br />

HO<br />

Fecatwin<br />

HO Sensa<br />

HemeSelect<br />

Armitage 1985 b<br />

HO vs Fecatwin<br />

Allison 1996<br />

HO Sensa vs<br />

HemeSelect<br />

25.1<br />

37.1<br />

0.9<br />

97.7<br />

94.4<br />

0.394<br />

(0.149, 0.956)<br />

HemeSelect:<br />

55.3<br />

HO:<br />

21.8<br />

98.9<br />

90.9<br />

37.1<br />

68.8<br />

9.1a 100<br />

HO<br />

HemeSelect<br />

Allison 1996<br />

Sensitivity analyses<br />

HO vs HemeSelect<br />

HO<br />

HemeSelect<br />

Robinson 1996<br />

HemeSelect:<br />

0.930<br />

43.55 (0.327, 2.793)<br />

–<br />

–<br />

–<br />

d<br />

HO Sensa:<br />

23.55<br />

Flexsure OBT:<br />

118.5<br />

Flexsure OBT:<br />

49.4<br />

BM-Test Colon Alb:<br />

34.0<br />

d<br />

Rozen 1997 HO<br />

60.0<br />

94.7<br />

HemeSelect<br />

80.0<br />

98.0<br />

195<br />

HO Sensa vs<br />

Allison 1996 HO Sensa<br />

79.4<br />

86.7<br />

25.2<br />

HemeSelect<br />

HemeSelect<br />

68.8<br />

94.4<br />

37.1<br />

Rozen 1997 HO Sensa<br />

60.0<br />

92.0<br />

17.2<br />

HemeSelect<br />

80.0<br />

98.0<br />

195<br />

Rozen 2000 HO<br />

60.0<br />

94.7<br />

26.9<br />

HO:<br />

Flexsure OBT<br />

80.0<br />

96.7<br />

118.5<br />

26.9<br />

Rozen 1995 HO Sensa<br />

42.9<br />

94.9<br />

14.1<br />

HO Sensa:<br />

Flexsure OBT<br />

42.9<br />

98.5<br />

49.4<br />

14.1<br />

Rozen 1995 HO Sensa<br />

60.0<br />

83.7<br />

7.7<br />

HO Sensa:<br />

BM-Test Colon Albumin<br />

80.0<br />

89.5<br />

34.0<br />

7.7<br />

c<br />

HO vs Flexsure OBT<br />

HO Sensa vs Flexsure<br />

OBT<br />

HO Sensa vs BM-Test<br />

Colon Albumin<br />

Abbreviations: CI, confidence interval; DOR, diagnostic odds ratio; HO, Hem<strong>occult</strong>; HO Sensa, Hem<strong>occult</strong> Sensa; Fecatwin, Fecatwin Sensitive/Feca EIA; FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

aAssumes that the patient developing cancer during follow-up <strong>for</strong> adenoma initially tested negative with Hem<strong>occult</strong>. bData from Hardcastle 86. c Results affected by BUGS convergence problem noted during the analysis.<br />

dEstimated by inverse-variance pooling method. eEstimate only, as false negative rate was 0. fMedian parameter estimate. gGuaiac test relative to immunochemical test.<br />

210.6e 26.9<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 103


Figure 10 Summary receiver-operating characteristic curve <strong>for</strong> Hem<strong>occult</strong> versus HemeSelect<br />

sensitivity analyses<br />

Abbreviations: H’<strong>occult</strong>, Hem<strong>occult</strong>; H’select, HemeSelect; SROC, summary receiver-operating characteristic.<br />

104 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Appendix H Quality scoring<br />

Table 45 Quality scoring scale <strong>for</strong> FOBT comparative <strong>screening</strong> studies<br />

Evaluation criteria<br />

Quality<br />

score (/12)<br />

Criteria <strong>for</strong> study design quality<br />

A. Was the study prospectively designed?<br />

No (unclear) 0<br />

Yes 1<br />

B. Was there a within-subjects study design?<br />

No, different tests done on different individuals, not randomly allocated (case-control) 0<br />

No, different tests done on randomly allocated individuals (parallel randomised or<br />

quasi-randomised)<br />

1<br />

Yes, tests per<strong>for</strong>med on each individual (single group with sequential tests) 2<br />

C. Was the test being evaluated compared with a valid reference standard?<br />

No (not reported) 0<br />

Yes 1<br />

D. What was the industry’s relationship to the study?<br />

Full or partial support 0<br />

No support 1<br />

Criteria <strong>for</strong> minimising verification bias<br />

E. Was the decision to per<strong>for</strong>m the reference standard independent of the test results?<br />

No (not reported, positives only or positives and sample of negatives) 0<br />

Yes (all tests) 1<br />

Criteria <strong>for</strong> minimising selection bias<br />

F. Was the selection of sample of participants:<br />

Not random (or not reported) 0<br />

Consecutive 1<br />

Random 2<br />

Criteria <strong>for</strong> blinding<br />

G. Were the test and the reference standard measured independently (blind) of each other?<br />

No (or not reported) 0<br />

Yes, the test was measured independently of the reference standard and the reference<br />

standard independently of the test (or methodology not subject to observer bias)<br />

H. Were the test and the comparator measured independently (blind) of each other?<br />

No (or not reported) 0<br />

Yes, the test was measured independently of the reference standard and the reference<br />

1<br />

standard independently of the test (or methodology not subject to observer bias)<br />

Criteria relevant to the external validity of the results<br />

I. Was the participant <strong>population</strong> consistent with proposed Australian <strong>screening</strong> setting (ie,<br />

<strong>population</strong>-based, average-risk, older age group)?<br />

No (not reported) 0<br />

Partially 1<br />

Yes 2<br />

Abbreviation: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 105<br />

1


Appendix I Participation in FOBT<br />

<strong>screening</strong> rates<br />

The participation rates used in the economic model differentiate between <strong>screening</strong><br />

rounds and between type of faecal <strong>occult</strong> <strong>blood</strong> test (FOBT) used. This is necessary in<br />

order to capture the negative effect of non-participation on the cost-effectiveness of the<br />

FOBTs investigated. The calculations behind the rates used in the economic model<br />

appear below. The values used in the economic model appear in italics.<br />

As shown in Table 46, the participation rate <strong>for</strong> guaiac FOBTs is calculated by<br />

multiplying the ratio of guaiac-to-immunochemical participation rates taken from Cole et<br />

al (2003) by the immunochemical participation rate in the Australian Bowel Cancer<br />

Screening Pilot Program (the pilot).<br />

Table 46 Round 1 participation with FOBT <strong>screening</strong><br />

Row Parameter Guaiac FOBT Immunochemical<br />

FOBT<br />

A Participation rates from<br />

the literature<br />

B Participation rates used<br />

in the economic model<br />

C Non-participation rates<br />

implicit in the economic<br />

model<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

0.2340 0.3050 Cole et al (2003)<br />

Reference<br />

0.3483 0.4540 Immunochemical = data from the pilot<br />

Guaiac = (Row A guaiac rate/Row A<br />

immunochemical rate) × Row B<br />

immunochemical rate<br />

0.6517 0.5460 C = 1 – Row B<br />

An individual’s participation after the first invitation to <strong>screening</strong> is assumed to be a<br />

function of the individual’s previous decision regarding participation. The calculations<br />

supporting the values used in the economic model are presented below. Table 47 gives<br />

details of the calculation of proportional round 2 losses from <strong>screening</strong> from the Danish<br />

trial (Jorgensen et al 2002). Table 48 provides details of the calculation of estimated<br />

losses from <strong>screening</strong> in the economic model.<br />

Table 47 Participation data from Jorgensen et al (2002)<br />

Row Parameter Value Reference<br />

A Round 1 participation 0.6700 Jorgensen et al (2002)<br />

B Round 1 non-participation 0.3300 B = 1 – Row A<br />

C Participation in round 2 <strong>for</strong> those individuals who participated in<br />

round 1<br />

0.9300 Jorgensen et al (2002)<br />

D Proportion of individuals not participating in round 2 0.0700 D = 1 – Row C<br />

E Proportion of invitees previously participating but lost in round 2 0.0469 E = Row A × Row D<br />

106 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Table 48 Later-round participation with FOBT <strong>screening</strong> <strong>for</strong> those who participated in round 1<br />

Row Parameter Guaiac<br />

FOBT<br />

A Ratio of estimated round 1<br />

loss to round 1 loss in<br />

Jorgensen et al (2002)<br />

B Estimated proportion of<br />

invitees lost from <strong>screening</strong><br />

in round 2<br />

C Proportion of individuals<br />

participating in round 2.<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

Note that figures may not multiply exactly due to rounding.<br />

Immunochemical<br />

FOBT<br />

Reference<br />

1.97 1.65 A = (Table 46 Row C / Table 47 Row B) × 100<br />

0.1382 0.1155 B = Row A × Table 47 Row D<br />

0.862 0.884 C = 1 – Row B<br />

In addition to these figures, the economic model also allows <strong>for</strong> a proportion of first<br />

round non-participants to participate in later <strong>screening</strong> rounds.<br />

Expert opinion indicates that the number of new participants in the second round of<br />

<strong>screening</strong> is roughly equivalent to the number of individuals lost from <strong>screening</strong> after<br />

participating in round 1. The calculations used to estimate the proportion of new<br />

participants used in the economic model are presented in Table 49.<br />

Table 49 Proportion of new participants after the first <strong>screening</strong> round<br />

Row Parameter Guaiac<br />

FOBT<br />

A Proportion of invitees<br />

previously participating but<br />

lost in round 2<br />

B Proportion of new<br />

participants after the first<br />

<strong>screening</strong> round<br />

Abbreviations: FOBT, faecal <strong>occult</strong> <strong>blood</strong> test.<br />

Immunochemical<br />

FOBT<br />

Reference<br />

0.0481 0.0526 D = Table 46 Row B × Table 48 Row B<br />

0.0739 0.0963 B = A/Table 46 Row C<br />

Rates of participation calculated <strong>for</strong> round 2 are used <strong>for</strong> all further rounds of <strong>screening</strong>.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 107


Appendix J Relative cost-effectiveness<br />

of Hem<strong>occult</strong> versus<br />

Fecatwin Sensitive/<br />

Feca EIA<br />

The variables used in this analysis of the economic model were outlined previously in the<br />

main document. The reason <strong>for</strong> not including the results in that section is that the data<br />

are of a lower quality, particularly due to the absence of reliable estimates on the<br />

sensitivity of the tests in detection of adenomas. This is an important feature, as the<br />

ability to detect precancerous adenomas prior to malignant trans<strong>for</strong>mation is an integral<br />

feature of faecal <strong>occult</strong> <strong>blood</strong> test (FOBT) <strong>screening</strong> <strong>for</strong> colorectal cancer (CRC). As<br />

previously discussed, early detection of adenomas may lead to increased life-expectancy<br />

and reduced lifetime treatment costs by avoiding the incidence of CRC. In addition, it<br />

should be noted that the subjects in this study were not required to adhere to the dietary<br />

restrictions generally recommended with the use of a guaiac FOBT. Consequently, it<br />

would be inappropriate to place the results of this head-to-head comparison alongside<br />

those with better quality data.<br />

The results of the analysis are discussed below.<br />

Costs<br />

Fecatwin Sensitive/Feca EIA is associated with higher costs overall. The expected cost<br />

of the test itself is more than the alternatives. Participation is also higher, leading to a<br />

marginally higher average cost per individual invited to <strong>screening</strong>. Furthermore,<br />

diagnostic-follow up costs are substantially more in the Fecatwin Sensitive/Feca EIA<br />

arm of the economic model, a result of higher sensitivity and a lower specificity <strong>for</strong><br />

detection of CRC. This is the key driver of the incremental price difference between the<br />

two tests, with diagnostic follow-up costs accounting <strong>for</strong> 14% of all costs in the Fecatwin<br />

Sensitive/Feca EIA arm of the model, compared with 5% in the Hem<strong>occult</strong> arm.<br />

Costs are broken down into their components in Figure 11.<br />

108 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Cost<br />

$1,800.00<br />

$1,600.00<br />

$1,400.00<br />

$1,200.00<br />

$1,000.00<br />

$800.00<br />

$600.00<br />

$400.00<br />

$200.00<br />

$0.00<br />

Hem<strong>occult</strong> Fecatwin<br />

Type of FOBT<br />

Figure 11 Component costs in the Hem<strong>occult</strong> versus Fecatwin Sensitive/Feca EIA analysis<br />

Outcomes<br />

As a result of the higher participation rates, Fecatwin Sensitive/Feca EIA is associated<br />

with only a marginally higher average number of completed FOBTs (6.0 compared with<br />

5.8). It is unlikely, there<strong>for</strong>e, that this will be a key driver of the relative costeffectiveness.<br />

Alongside higher overall costs, Fecatwin Sensitive/Feca EIA is also associated with a<br />

numerically greater life-expectancy from the outset of the model. As with the analyses<br />

presented in the main document, however, the difference is small and statistically<br />

insignificant. This is illustrated in Table 50.<br />

Table 50 Life-expectancy from the beginning of the economic model in the Hem<strong>occult</strong> versus Fecatwin<br />

Sensitive/Feca EIA analysis<br />

Patient group Life-years gained<br />

(per patient)<br />

Patients screened with Hem<strong>occult</strong> 14.399<br />

Incremental life-years<br />

gained<br />

Patients screened with Fecatwin 14.442 –0.043<br />

Cancer treatment<br />

Colonoscopy<br />

FOBT<br />

The positive difference in life-expectancy between <strong>screening</strong> and no <strong>screening</strong> is driven<br />

by the better detection of early-stage CRC. This difference is most pronounced in terms<br />

of Dukes’ A CRC, though it remains in the case of Dukes’ B.<br />

Figure 12 illustrates the difference in detection between the two tests, based on this<br />

analysis.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 109


Number detected by <strong>screening</strong> or symptoms<br />

350<br />

300<br />

250<br />

200<br />

150<br />

100<br />

50<br />

0<br />

Note that despite the assumed inability to detect adenomas, there is still a greater<br />

number of abnormalities detected overall when compared with the no <strong>screening</strong> option.<br />

Figure 13 illustrates the method of detection of all neoplasms.<br />

Total cancers detected by<br />

<strong>screening</strong><br />

Figure 12 The pattern of detection in the Hem<strong>occult</strong> versus Fecatwin Sensitive/Feca EIA<br />

2500<br />

2000<br />

1500<br />

1000<br />

500<br />

0<br />

Total neoplasms detected by<br />

<strong>screening</strong><br />

Dukes' A Dukes' B Dukes' C Dukes' D<br />

Total cancers detected by<br />

<strong>screening</strong><br />

Total cancers presenting<br />

symptomatically<br />

Total cancers undetected at<br />

death<br />

Figure 13 Method of CRC/cancerous adenoma detection in the Hem<strong>occult</strong> versus Fecatwin<br />

Sensitive/Feca EIA analysis<br />

Hem<strong>occult</strong><br />

Fecatwin<br />

Hem<strong>occult</strong><br />

Fecatwin<br />

No <strong>screening</strong><br />

110 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Cost-effectiveness<br />

While Fecatwin Sensitive/Feca EIA offers a numerically superior benefit in terms of lifeexpectancy<br />

at a reasonable cost, the difference is marginal (Table 51).<br />

Table 51 Incremental cost-effectiveness of Hem<strong>occult</strong> versus Fecatwin Sensitive/Feca EIA<br />

Hem<strong>occult</strong> Fecatwin Incremental<br />

Total cost associated with <strong>screening</strong> option $1478.28 $1674.08 –$195.80<br />

Life-years gained 14.399 14.442 –0.043<br />

Incremental costs per life-year gained $4553.49<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 111


Appendix K Sensitivity analysis of<br />

Hem<strong>occult</strong> versus<br />

HemeSelect<br />

The following analysis is based on variations to the sensitivity and specificity of the<br />

Hem<strong>occult</strong> faecal <strong>occult</strong> <strong>blood</strong> test (FOBT) in detection of colorectal cancer (CRC). It is<br />

motivated by a desire to explore the possible inaccuracies of the estimates used in the<br />

main analysis.<br />

The values used in this analysis have been calculated from the summary receiveroperating<br />

characteristic (SROC) curve appearing in Appendix G, using the following<br />

<strong>for</strong>mulas:<br />

False positive rate (FPR) = 1 – specificity (Equation 1)<br />

Sensitivity =<br />

1+<br />

e<br />

where a = 4.087515 and b = 0.262321.<br />

1<br />

a<br />

1+<br />

b<br />

FPR<br />

1 −b<br />

1−b<br />

× ( )<br />

1−<br />

FPR<br />

(Equation 2)<br />

Using Equations 1 and 2 and selecting a Hem<strong>occult</strong> specificity of 0.9550, a sensitivity of<br />

0.5778 is calculated. These values are then used in the economic model comparing<br />

Hem<strong>occult</strong> with HemeSelect. All other values used in the main analysis remain the same.<br />

The results of this analysis appear alongside the base-case analysis in Table 52.<br />

Table 52 The effect of selecting alternative sensitivity and specificity values <strong>for</strong> Hem<strong>occult</strong>: Hem<strong>occult</strong><br />

versus HemeSelect<br />

Hem<strong>occult</strong><br />

Base-case analysis – Hem<strong>occult</strong> sensitivity 30.0%, specificity 97.8%<br />

HemeSelect Incremental<br />

Total cost associated with <strong>screening</strong> option $1333.65 $1448.89 –$115.24<br />

Life-years gained 15.465 15.501 –0.036<br />

Incremental costs per life-year gained $3172.10<br />

Different accuracy estimate – Hem<strong>occult</strong> sensitivity 57.8%, specificity 95.5%<br />

Total cost associated with <strong>screening</strong> option $1377.35 $1448.89 –$71.54<br />

Life-years gained 15.475 15.501 –0.026<br />

Incremental costs per life-year gained $2752.99<br />

With regards to outcomes, the variation means that Hem<strong>occult</strong> detects slightly more<br />

CRC in the <strong>screening</strong> <strong>population</strong>. The pattern of detection, however, remains similar to<br />

that of the main analysis.<br />

112 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Overall, altering the sensitivity and specificity has little effect on the relative costeffectiveness<br />

result. Individuals subject to Hem<strong>occult</strong> have a slightly higher lifeexpectancy<br />

due to better accuracy in detecting the presence of CRC, although this is<br />

achieved at a marginally higher cost. Nonetheless, HemeSelect remains cost-effective<br />

compared with Hem<strong>occult</strong> and, importantly, the incremental cost-effectiveness ratio<br />

improves due to faster growth in costs than outcomes in the Hem<strong>occult</strong> arm of the<br />

model.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 113


Abbreviations<br />

AACR Australasian Association of Cancer Registries<br />

AHMAC Australian Health Ministers’ Advisory Council<br />

AIHW Australian Institute of Health and Welfare<br />

CCOHTA Canadian Coordinating Office <strong>for</strong> Health Technology Assessment<br />

CI Confidence interval<br />

CRC Colorectal cancer<br />

DACEHTA Danish Centre <strong>for</strong> Evaluation and Health Technology Assessment<br />

DARE Database of Abstracts of Reviews and Effects<br />

DCBE Double contrast barium enema<br />

DOR Diagnostic odds ratio<br />

FNR False negative rate<br />

FOBT <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> test<br />

FPR False positive rate<br />

GI Gastrointestinal<br />

HIRU Health In<strong>for</strong>mation Research Unit<br />

HSTAT Health Services Technology Assessment Texts<br />

HTA Health Technology Assessment<br />

ISRTCN International Standard Randomised Controlled Trial Number<br />

MBS Medicare Benefits Scheme<br />

MCMC Markov chain Monte Carlo<br />

MSAC Medical Services Advisory Committee<br />

NCCI National Cancer Control Initiative<br />

NHMRC National Health and Medical Research Council<br />

NHS CRD National Health Service Centre <strong>for</strong> Reviews and Dissemination<br />

NHSEED National Health Service Economic Evaluation Database<br />

114 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


NSAID Non-steroidal anti-inflammatory drug<br />

NZHTA New Zealand Health Technology Assessment<br />

OR Odds ratio<br />

Pilot The Australian Government Bowel Cancer Screening Pilot Program<br />

PPV Positive predictive value<br />

RCT Randomised controlled trial<br />

RD Risk difference<br />

RPHA Reverse-passive haemagglutination<br />

SBU Swedish Council on Technology Assessment in Health Care<br />

SROC Summary receiver-operating characteristic<br />

TGA Therapeutic Goods Administration<br />

TPR True positive rate<br />

USPSTF United States Preventative Services Task Force<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 115


References<br />

Aisawa T, Saito H, Kawaguchi H, Uno Y, Munakata A, Yoshida Y (1988), Mass<br />

<strong>screening</strong> <strong>for</strong> colon cancer using an immunologic fecal <strong>occult</strong> <strong>blood</strong> test by reversed<br />

passive hemagglutination reaction (RPHA)-comparison of single RPHA with 2-day<br />

<strong>testing</strong>, Proceedings of the 8th Asia-Pacific Congress of Gastroenterology FP23–FP25.<br />

Allison JE, Tekawa IS, Ransom LJ, Adrain AL (1996), A comparison of fecal <strong>occult</strong><strong>blood</strong><br />

tests <strong>for</strong> colorectal-cancer <strong>screening</strong>, New England Journal of Medicine 334: 155–159.<br />

Allison JE (2003), Screening <strong>for</strong> colorectal cancer 2003: Is there still a role <strong>for</strong> the<br />

FOBT? Techniques in Gastrointestinal Endoscopy 5: 127–133.<br />

Armitage N, Hardcastle JD, Amar SS, Balfour TW, Haynes J, James PD (1985), A<br />

comparison of an immunological faecal <strong>occult</strong> <strong>blood</strong> test Fecatwin sensitive/FECA EIA<br />

with Hem<strong>occult</strong> in <strong>population</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer, British Journal of Cancer 51:<br />

799–804.<br />

Armitage NC, Hardcastle JD (1987), Screening <strong>for</strong> colorectal cancer – the Nottingham<br />

experience, Annals of the Academy of Medicine Singapore 16: 432–436.<br />

Australian Institute of Health and Welfare (AIHW) 2003, Health expenditure Australia<br />

2001–02, Health and Welfare Expenditure Series no. 17, AIHW Cat. no. HWE 24,<br />

Canberra: AIHW.<br />

Australian Institute of Health and Welfare (AIHW), Australasian Association of Cancer<br />

Registries (AACR) (2003). Cancer in Australia 2000. AIHW cat. no. CAN 18. Canberra:<br />

AIHW (Cancer Series no. 23).<br />

Australian Institute of Health and Welfare (AIHW), Australasian Association of Cancer<br />

Registries (AACR) (2002), Cancer in Australia 1999, AIHW cat. no. CAN 15, Canberra:<br />

AIHW (Cancer Series no. 20)<br />

Barrison IG, Parkins RA (1985), The clinical value of Hem<strong>occult</strong> and Fecatwin in the<br />

detection of colorectal neoplasia in hospital and general practice patients, Postgraduate<br />

Medical Journal 61: 701–704.<br />

Bell J, Coates M, Day P, Armstrong B (1996), Colorectal cancer in NSW in 1972 to 1993,<br />

Cancer Council, NSW Health Department.<br />

Bolin TD, Korman MG, Stanton R, Talley N, Newstead GL, Donnelly N, Hall W, Ho<br />

MT, Lapsley H (1999), Positive cost-effectiveness of early diagnosis of colorectal cancer,<br />

Colorectal Disease 1: 113–122<br />

Broadstock M (2000), Effectiveness and cost-effectiveness of automated and semiautomated<br />

cervical <strong>screening</strong> devices: a systematic review, New Zealand Health Technology<br />

Assessment (NZHTA) Report 3(1).<br />

Castiglione G, Zappa M, Grazzini G, Mazzotta A, Biagini M, Salvadori P, Ciatto S<br />

(1996), Immunochemical vs guaiac faecal <strong>occult</strong> <strong>blood</strong> tests in a <strong>population</strong>-based<br />

<strong>screening</strong> program <strong>for</strong> colorectal cancer, British Journal of Cancer 74: 141–144.<br />

116 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Castiglione G, Zappa M, Grazzini G, Sani C, Mazzotta A, Mantellini P (1997), Cost<br />

analysis in a <strong>population</strong> based <strong>screening</strong> program <strong>for</strong> colorectal cancer: comparison of<br />

immunochemical and guaiac faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>, Journal of Medical Screening 4: 142–<br />

146.<br />

Commonwealth Department of Health and Ageing (CDHA) (2002), Medicare Benefits<br />

Schedule. Available from: http://www.<strong>health</strong>.gov.au/pubs/mbs/mbs/css/index.htm.<br />

Chock C, Irwig L, Berry G, Glasziou P (1997), Comparing dichotomous <strong>screening</strong> tests<br />

when individuals negative on both tests are not verified, Journal of Clinical Epidemiology 50:<br />

1211–1217.<br />

Cochrane Methods Group on Systematic Review of Screening and Diagnostic Tests:<br />

Recommended Methods, updated 6 June 1996. Available from:<br />

http://www.cochrane.org/cochrane/sadtdoc1.htm [Accessed 28 October 2002].<br />

Cole SR, Young GP, Esterman A, Cadd B, Morcom J (2003), A randomised trial of the<br />

impact of new faecal haemoglobin test technologies on <strong>population</strong> participation in<br />

<strong>screening</strong> <strong>for</strong> colorectal cancer, Journal of Medical Screening 10: 117–122.<br />

Collett JA, Olynyk JK, Platell CF (2000), Flexible sigmoidoscopy <strong>screening</strong> <strong>for</strong> colorectal<br />

cancer in average-risk people: update of a community-based project, Medical Journal of<br />

Australia 173: 463–466.<br />

Costa ND, Cadiot G, Merle C, Jolly D, Bouche O, Thiefin G, Zeitoun P (2001), Bleeding<br />

reflux esophagitis: a prospective 1-year study in a university hospital, American Journal of<br />

Gastroenterology 96: 47–51.<br />

DerSimonian R, Laird N (1986), Meta-analysis in clinical trials, Controlled Clinical Trials 7:<br />

177–188.<br />

Foutch PG, Manne RK, Sanowski RA, Gaines JA (1988), Risk factors <strong>for</strong> <strong>blood</strong> loss<br />

from adenomatous polyps of the large bowel: a colonoscopic evaluation with<br />

histopathological correlation, Journal of Clinical Gastroenterology 10: 50–56.<br />

Frommer DJ, Kapparis A, Brown MK (1988), Improved <strong>screening</strong> <strong>for</strong> colorectal cancer<br />

by immunological detection of <strong>occult</strong> <strong>blood</strong>, British Medical Journal, Clinical Research Edition<br />

296: 1092–1094.<br />

Fujita T, Kumanishi Y, Okuyama Y, Sugiyama R, Mannen I, Nanbu K, Ota J, Taguchi T<br />

(1992), Feasibility of 3-day fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> using immunochemical<br />

hemagglutination method in mass <strong>screening</strong> <strong>for</strong> colorectal cancer. In: Young GY, Saito<br />

H (eds), Fecal Occult Blood Tests: Current Issues and New Tests, Proceedings of a satellite meeting of<br />

the World Congresses of Gastroenterology, Sydney, Australia, August 1990, SmithKline<br />

Diagnostics Inc, San Jose, 82–89.<br />

Gyrd-Hensen D (1998), <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> tests: a cost-effectiveness analysis,<br />

International Journal of Technology Assessment in Health Care 14: 290–301.<br />

Habr-Gama A, Waye JD (1989), Complications and hazards of gastrointestinal<br />

endoscopy, World Journal of Surgery 13: 193–201.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 117


Hardcastle JD, Armitage NC, Chamberlain J, Amar SS, James PD, Balfour TW (1986),<br />

Fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer in the general <strong>population</strong>. Results of a<br />

controlled trial, Cancer 58: 397–403.<br />

Hawkey CJ (2000), Risk of ulcer bleeding in patients infected with Helicobacter pylori taking<br />

non-steroidal anti-inflammatory drugs, Gut 46: 310–311.<br />

Higgins JPT, Whitehead A (1996), Borrowing strength from external trials in a metaanalysis,<br />

Statistics in Medicine 15: 2733–2749.<br />

Hynam KA, Hart AR, Gay SP, Inglis A, Wicks AC, <strong>May</strong>berry JF (1995), Screening <strong>for</strong><br />

colorectal cancer: reasons <strong>for</strong> refusal of faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> in a general practice in<br />

England, Journal of Epidemiology and Community Health 49: 84–86.<br />

Iwase T (1992), Evaluation of an immunochemical fecal <strong>occult</strong> <strong>blood</strong> test by reversed<br />

passive hemagglutination compared with Hem<strong>occult</strong> II <strong>screening</strong> <strong>for</strong> colorectal cancer,<br />

In: Young GY, Saito H (eds). Fecal Occult Blood Tests: Current Issues and New Tests. Proceedings<br />

of a satellite meeting of the World Congresses of Gastroenterology, Sydney, Australia, August 1990.<br />

SmithKline Diagnostics Inc, San Jose, 90–94.<br />

Jorgensen OD, Kronborg O, Fenger C (2002), A randomised study of <strong>screening</strong> <strong>for</strong><br />

colorectal cancer using faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>: results after 13 years and seven<br />

biennial <strong>screening</strong> rounds, Gut 50: 29–32.<br />

Kettner JD, Whatrup C, Verne JE, Young K, Williams CB, Northover JM (1990), Is<br />

there a preference <strong>for</strong> different ways of per<strong>for</strong>ming faecal <strong>occult</strong> <strong>blood</strong> tests? [erratum<br />

appears in International Journal of Colorectal Disease 5:176], International Journal of Colorectal<br />

Disease 5: 82–86.<br />

Khandker RK, Dulski JD, Kilpatrick JB, Ellis RP, Mitchell JB, Baine WB (2000), A<br />

decision model and cost-effectiveness analysis of colorectal cancer <strong>screening</strong> and<br />

surveillance guidelines <strong>for</strong> average-risk adults, International Journal of Technology Assessment in<br />

Health Care 16: 799–810.<br />

Kurata JH, Honda GD, Frankl H (1982), Hospitalization and mortality rates <strong>for</strong> peptic<br />

ulcers: a comparison of a large <strong>health</strong> maintenance organization and United States data,<br />

Gastroenterology 83: 1008–1016.<br />

Lee FI, Costello FT (1982), Assessment of Fecatest and Hem<strong>occult</strong> <strong>for</strong> faecal <strong>occult</strong><br />

<strong>blood</strong> <strong>testing</strong>, British Medical Journal, Clinical Research Edition 285: 938.<br />

Levin B, Hess K, Johnson C (1997), Screening <strong>for</strong> colorectal cancer. A comparison of 3<br />

fecal <strong>occult</strong> <strong>blood</strong> tests, Archives of Internal Medicine 157: 970–976.<br />

Li S, Nie Z, Li N, Li J, Zhang P, Yang Z, Mu S, Du Y, Hu J, Yuan S, Qu H, Zhang T,<br />

Wang S, Dong E, Qi D (2000), Colorectal cancer <strong>screening</strong> <strong>for</strong> the natural <strong>population</strong> of<br />

Beijing with sequential fecal <strong>occult</strong> <strong>blood</strong> test: A multicenter study, Chinese Medical Journal<br />

116: 200–202.<br />

Lieberman DA, Weiss DG, Bond JH, Ahnen DJ, Garewal H, Chejfec G (2000), Use of<br />

colonoscopy to screen asymptomatic adults <strong>for</strong> colorectal cancer. Veterans Affairs<br />

Cooperative Study Group 380, New England Journal of Medicine 343: 162–168.<br />

118 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


Lijmer JG, Bossuyt PMM, Heisterkamp SH (2002), Exploring sources of heterogeneity in<br />

systematic reviews of diagnostic tests, Statistics in Medicine 21: 1525–1537.<br />

Loeve F, Boer R, van Oortmarssen GJ, van Ballegooijen M, Habbema JD (1999), The<br />

MISCAN-COLON simulation model <strong>for</strong> the evaluation of colorectal cancer <strong>screening</strong>,<br />

Computers and Biomedical Research 32: 13–33<br />

Loeve F, Brown ML, Boer R, van Ballegooijen M, van Oortmarssen GJ, Habbema JD<br />

(2000), Endoscopic colorectal cancer <strong>screening</strong>: a cost-saving analysis, Journal of the<br />

National Cancer Institute 92: 557–563.<br />

Mandel JS, Bond JH, Bradley M, Snover DC, Church TR, Williams S, Watt G, Schuman<br />

LM, Ederer F, Gilbertsen V (1989), Sensitivity, specificity, and positive predictivity of the<br />

Hem<strong>occult</strong> test in <strong>screening</strong> <strong>for</strong> colorectal cancers. The University of Minnesota's Colon<br />

Cancer Control Study, Gastroenterology 97: 597–600.<br />

Mandel JS, Bond JH, Church TR, Snover DC, Bradley GM, Schuman LM, Ederer F<br />

(1993), Reducing mortality from colorectal cancer by <strong>screening</strong> <strong>for</strong> fecal <strong>occult</strong> <strong>blood</strong>.<br />

Minnesota Colon Cancer Control Study [comment] [erratum appears in New England<br />

Journal of Medicine 329:672], New England Journal of Medicine 328: 1365–1371.<br />

Mant D, Fitzpatrick R, Hogg A, Fuller A, Farmer A, Verne J, Northover J (1990),<br />

Experiences of patients with false positive results from colorectal cancer <strong>screening</strong>,<br />

British Journal of General Practice 40: 423–425.<br />

Mapp TJ, Hardcastle JD, Moss SM, Robinson MH (1999), Survival of patients with<br />

colorectal cancer diagnosed in a randomized controlled trial of faecal <strong>occult</strong> <strong>blood</strong><br />

<strong>screening</strong>, British Journal of Surgery 86: 1286–1291.<br />

McLeish JA, Thursfield VJ, Giles GG (2002), Survival from colorectal cancer in Victoria:<br />

10-year follow up of the 1987 management survey, Australian & New Zealand Journal of<br />

Surgery 72: 352–356.<br />

Nakama H, Kamijo N, Fujimori K, Horiuchi A, Abdul FS, Zhang B (1997),<br />

Immunochemical fecal <strong>occult</strong> <strong>blood</strong> test is not suitable <strong>for</strong> diagnosis of hemorrhoids,<br />

American Journal of Medicine 102: 551–554.<br />

NHMRC (1999), Guidelines <strong>for</strong> the prevention, detection and management of Colorectal<br />

Cancer (CRC), National Health and Medical Research Council, Commonwealth of<br />

Australia. Available from www.ausinfo.gov.au/general/gen_hottobuy.htm.<br />

O'Leary BA, Olynyk JK, Neville AM, Platell CF (<strong>2004</strong>), Cost-effectiveness of colorectal<br />

cancer <strong>screening</strong>: comparison of community-based flexible sigmoidoscopy with fecal<br />

<strong>occult</strong> <strong>blood</strong> <strong>testing</strong> and colonoscopy, Journal of Gastroenterology & Hepatology 19: 38–47.<br />

Parker MA, Robinson MH, Scholefield JH, Hardcastle JD (2002), Psychiatric morbidity<br />

and <strong>screening</strong> <strong>for</strong> colorectal cancer, Journal of Medical Screening 9: 7–10.<br />

Petrelli N, Michalek AM, Freedman A, Baroni M, Mink I, Rodriguez-Bigas M (1994),<br />

Immunochemical versus guaiac <strong>occult</strong> <strong>blood</strong> stool tests: results of a community-based<br />

<strong>screening</strong> program, Surgical Oncology 3: 27–36.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 119


Rae AJ, Cleator IGM (1994), The two-tier fecal <strong>occult</strong> <strong>blood</strong> test: Cost-effective<br />

<strong>screening</strong>, Canadian Journal of Gastroenterology 8: 362–368.<br />

Reiertsen O, Skjoto J, Jacobsen CD, Rosseland AR (1987), Complications of fibreoptic<br />

endoscopy – five years experience in a central hospital, Endoscopy 19: 1–6.<br />

Robinson MH, Marks CG, Farrands PA, Whynes DK, Bostock K, Hardcastle JD (1995),<br />

Is an immunological faecal <strong>occult</strong> <strong>blood</strong> test better than Hem<strong>occult</strong>? A cost-benefit<br />

study, European Journal of Surgical Oncology 21: 261–264.<br />

Robinson MHE, Marks CG, Farrands PA, Bostock K, Hardcastle JD (1996), Screening<br />

<strong>for</strong> colorectal cancer with an immunological faecal <strong>occult</strong> <strong>blood</strong> test: 2-year follow-up,<br />

British Journal of Surgery 83: 500–501.<br />

Robinson MH, Hardcastle JD, Moss SM, Amar SS, Chamberlain JO, Armitage NC,<br />

Scholefield JH, Mangham CM (1999), The risks of <strong>screening</strong>: data from the Nottingham<br />

randomised controlled trial of faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal cancer, Gut 45:<br />

588–592.<br />

Rozen P, Knaani J, Papo N (1995), Evaluation and comparison of an immunochemical<br />

and a guaiac faecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> test <strong>for</strong> colorectal neoplasia, European Journal of<br />

Cancer Prevention 4: 475–481.<br />

Rozen P, Knaani J, Samuel Z (1997), Per<strong>for</strong>mance characteristics and comparison of two<br />

immunochemical and two guaiac fecal <strong>occult</strong> <strong>blood</strong> <strong>screening</strong> tests <strong>for</strong> colorectal<br />

neoplasia, Digestive Diseases & Sciences 42: 2064–2071.<br />

Rozen P, Knaani J, Samuel Z (2000), Comparative <strong>screening</strong> with a sensitive guaiac and<br />

specific immunochemical <strong>occult</strong> <strong>blood</strong> test in an endoscopic study, Cancer 89: 46–52.<br />

Salkeld G, Young G, Irwig L, Haas M, Glaziou P (1996), Cost-effectiveness analysis of<br />

<strong>screening</strong> by faecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> colorectal cancer in Australia, Australian and<br />

New Zealand Journal of Public Health 20: 138–143.<br />

Sharma R, Gorbien MJ (1995), Angiodysplasia and lower gastrointestinal tract bleeding in<br />

elderly patients [Review] [75 refs], Archives of Internal Medicine 155: 807–812.<br />

Simon JB (1985), Occult <strong>blood</strong> <strong>screening</strong> <strong>for</strong> colorectal carcinoma: a critical review.<br />

Gastroenterology 88: 820–837.<br />

Simon JB (1987), The pros and cons of fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> colorectal<br />

neoplasms [Review] [156 refs], Cancer & Metastasis Reviews 6: 397–411.<br />

Smith TC, Spiegelhalter DJ, Thomas A (1995), Bayesian approaches to random-effects<br />

meta-analysis, Statistics in Medicine 14: 2685–2699.<br />

Spiegelhalter DJ, Thomas A, Best N (2000), WinBUGS Version 1.3 User Manual.<br />

Cambridge: University of Cambridge, Institute of Public Health.<br />

St John DJ, Young GP, Alexeyeff MA, Deacon MC, Cuthbertson AM, Macrae FA,<br />

Penfold JC (1993), Evaluation of new <strong>occult</strong> <strong>blood</strong> tests <strong>for</strong> detection of colorectal<br />

neoplasia, Gastroenterology 104: 1661–1668.<br />

120 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>


St John DJB, Young GP (2003), Is there a need to restrict diet in <strong>occult</strong> <strong>blood</strong> <strong>screening</strong><br />

<strong>for</strong> colorectal cancer? Gastroenterology 102: A402.<br />

Stryker SJ, Wolff BG, Culp CE, Libbe SD, Ilstrup DM, MacCarty RL (1987), Natural<br />

history of untreated colonic polyps, Gastroenterology 93: 1009–1013.<br />

Towler B, Irwig L, Glasziou P, Kewenter J, Weller D, Silagy C (1998), A systematic<br />

review of the effects of <strong>screening</strong> <strong>for</strong> colorectal cancer using the faecal <strong>occult</strong> <strong>blood</strong> test,<br />

hem<strong>occult</strong> [comment], British Medical Journal 317: 559–565.<br />

Vaananen P, Tenhunen R (1988), Rapid immunochemical detection of fecal <strong>occult</strong> <strong>blood</strong><br />

by use of a latex-agglutination test, Clinical Chemistry 34: 1763–1766.<br />

Verne J, Kettner J, Mant D, Farmer A, Mortenson N, Northover J (1993), Selfadministered<br />

faecal <strong>occult</strong> <strong>blood</strong> tests do not increase compliance with <strong>screening</strong> <strong>for</strong><br />

colorectal cancer: results of a randomized controlled trial, European Journal of Cancer<br />

Prevention 2: 301–305.<br />

Wagner JL, Tunis S, Brown M, Ching A, Almeida R (1996), Cost-effectiveness of<br />

colorectal cancer <strong>screening</strong> in average-risk adults, Prevention and early detection of colorectal<br />

cancer. Young GP, Rozen P, Levin B, eds, London: WB Saunders, pp 321–356.<br />

Walter SD, Frommer DJ, Cook RJ (1991), The estimation of sensitivity and specificity in<br />

colorectal cancer <strong>screening</strong> methods, Cancer Detection and Prevention 15: 465–469.<br />

Weller D, Moss J, Hiller J, Thomas D, Edwards J (1995), Screening <strong>for</strong> colorectal cancer:<br />

what are the costs? International Journal of Technology Assessment in Health Care, 11: 26–39.<br />

Winawer SJ, Andrews M, Flehinger B, Sherlock P, Schottenfeld D, Miller DG (1980a),<br />

Progress report on controlled trial of fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> the detection of<br />

colorectal neoplasia, Cancer 45: 2959–2964.<br />

Winawer SJ, Andrews M, Flehinger B, Sherlock P, Schottenfeld D, Miller DG (1980b),<br />

Progress report on controlled trial of fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> <strong>for</strong> the detection of<br />

colorectal neoplasia, Cancer 45: 2959–2964.<br />

Winawer SJ, Zauber AG, O'Brien MJ, <strong>May</strong> N, Gottlieb L, Sternberg SS, Waye JD, Bond<br />

J, Schapiro M, Stewart ET, Panish J, Ackroyd F, Kurtz RC, Shike M, Lightdale CJ,<br />

Gerdes H, Hornsby-Lewis L, Edelman M, Fleisher M, et al (1993), Randomized<br />

comparison of surveillance intervals after colonoscopic removal of newly diagnosed<br />

adenomatous polyps, New England Journal of Medicine 328: 901–906.<br />

Winawer SJ, Fletcher RH, Miller L, Godlee F, Stolar MH, Mulrow CD, Woolf SH, Glick<br />

SN, Ganiats TG, Bond JH, Rosen L, Zapka JG, Olsen SJ, Giardiello FM, Sisk JE, Van<br />

Antwerp R, Brown-Davis C, Marciniak DA, <strong>May</strong>er RJ (1997), Colorectal cancer<br />

<strong>screening</strong>: clinical guidelines and rationale, Gastroenterology 112: 594–642.<br />

Whitehead A, Whitehead J (1991), A general parametric approach to meta-analysis of<br />

randomized clinical trials, Statistics in Medicine 10: 1665–1677.<br />

Young GP (1998), Screening <strong>for</strong> colorectal cancer: alternative faecal <strong>occult</strong> <strong>blood</strong> tests,<br />

European Journal of Gastroenterology and Hepatology 10: 205–212.<br />

<strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong> 121


Young GP, St John DJB, Winawer SJ, Rozen PR (2002), Choice of fecal <strong>occult</strong> <strong>blood</strong><br />

tests <strong>for</strong> colorectal cancer <strong>screening</strong>: recommendations based on per<strong>for</strong>mance<br />

characteristics in <strong>population</strong> studies – a WHO (World Health Organization) and OMED<br />

(World Organization <strong>for</strong> Digestive Endoscopy) report, American Journal of Gastroenterology<br />

97: 2549–2507.<br />

Zappa M, Castiglione G, Paci E, Grazzini G, Rubeca T, Turco P, Crocetti E, Ciatto S<br />

(2001), Measuring interval cancers in <strong>population</strong>-based <strong>screening</strong> using different assays of<br />

fecal <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>: the district of Florence experience, International Journal of Cancer<br />

92: 151–154.<br />

Zuckerman GR, Prakash C, Askin MP, Lewis BS (2000), AGA technical review on the<br />

evaluation and management of <strong>occult</strong> and obscure gastrointestinal bleeding,<br />

Gastroenterology 118: 201–221<br />

122 <strong>Faecal</strong> <strong>occult</strong> <strong>blood</strong> <strong>testing</strong>

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!