Submission for Lancet Protocol Review Title A study ... - NHS Lothian

Submission for Lancet Protocol Review
Title
A study to develop and validate a Clinical Decision Rule using history,
examination, electrocardiographic and biochemical markers, to predict one
month outcome for patients presenting with syncope to the Emergency
Department (the ROSE study)
Short Title
Risk stratification Of Syncope in the Emergency department: the ROSE study
Principal Investigators
Dr Matthew Reed, Chief Scientist Office Clinical Academic Research Training
Fellow and Locum Consultant in Emergency Medicine, Department of
Emergency Medicine, Edinburgh Royal Infirmary, Edinburgh, UK.
Dr Alasdair Gray, Consultant and Honorary Reader in Emergency Medicine,
Department of Emergency Medicine, Edinburgh Royal Infirmary, Edinburgh,
UK.
Professor David Newby, Professor of Cardiology, Centre for Cardiovascular
Sciences, Edinburgh Royal Infirmary, Edinburgh, UK.
Dr Andrew Coull, Consultant Physician and Honorary Senior Lecturer,
Department of Medicine for the Elderly, Edinburgh Royal Infirmary, Edinburgh,
UK.
Professor Robin Prescott, Professor of Statistics, Medical Statistical Unit,
University of Edinburgh.
Collaborator
Mr Robert Lee, Statistician, Medical Statistical Unit, University of Edinburgh.
Main study centre
Department of Emergency Medicine, Edinburgh Royal Infirmary, 51 Little
France Crescent, Edinburgh, EH16 4SA, UK.
Summary
We aim to conduct the largest international derivation and validation study of
undifferentiated syncope based in the Emergency Department looking at onemonth
outcome, and the first in the UK. 550 consecutive patients aged 16
years or over will be recruited into the derivation study and 550 into the
validation study over an 18-month period. All patients will undergo a
standardised assessment using 32 historical variables, 14 examination
Version 14: 14/06/2007 1

variables and 24 ECG based variables. All patients will also have a full blood
count, urea, creatinine, liver function, glucose, electrolytes, high sensitivity
CRP and near-patient BNP tests. Admitted patients and a random selection of
100 discharged derivation group patients will undergo a formal laboratory
based Troponin I at least 12 hours post syncope. Endpoint measures are
serious outcome at 1 month. A Clinical Decision Rule using history,
examination, ECG, and biochemical markers will be developed and validated
to predict 1 month outcome for patients presenting with syncope to the
Emergency Department.
Background and previous research
Syncope is a transient loss of consciousness with an inability to maintain
postural tone followed by a spontaneous recovery [1]. It accounts for 3% of
ED visits and 1-6% of hospital medical admissions, affecting 6 per 1000
people per year [2,3]. Clinical assessment of syncope is difficult due to the
heterogeneous nature of underlying causes, ranging from benign
neurocardiogenic syncope, to potentially fatal arrhythmias.
In 1983, Kapoor et al [4] published the first prospective syncope study. 12month
mortality was 14%. Mortality was greatest in patients in whom a
cardiovascular cause was identified (30%). Subsequent studies have shown
that underlying heart disease in patients with syncope is associated with a
poor prognosis [5]. Recent emphasis has focused on risk stratification of
patients with syncope. Although guidelines have been issued [6-10],
evidence in respect to ED management is sparse.
There are 5 risk stratification studies [11-16]. Most involved small numbers of
patients and used different characteristics and outcome measures in their risk
stratification tools. Only two were prospective and had mixed results [11,13].
Only one study, a US based study looked at short-term adverse outcome
[15,16], which is relevant to emergency medicine practice. There were two
fundamental problems with this study. Firstly, two of the selected predictor
variables included in the rule would usually necessitate immediate hospital
admission (systolic blood pressure less than 90mmHg and a haematocrit of
less than 30%). This greatly reduces the benefit of the rule in ED admission
decision-making. Secondly, the rule was originally derived for 7 day serious
outcome, however later validated for 1 month outcome. No studies have been
examined in a UK population.
With growing pressures on acute medical beds and an increasingly elderly
population, there is a need for a large study of this common presenting
symptom to identify high-risk populations requiring further investigation, and
low-risk patients who may be discharged safely. Accurate identification of
patients would enable specific targeting of resources and prevent excessive
investigation of patients with benign causes.
No risk stratification studies have investigated the role of biochemical markers
in risk stratification. We hypothesize that Brain Natriuretic Peptide (BNP) and
high sensitivity C-reactive protein (HS-CRP) acting as objective markers of
underlying cardiac disease, may be excellent ED markers of 1 month
outcome. BNP [17,18] is an excellent marker of prognosis in patients with
heart failure or cardiac disease, and several prospective epidemiological
studies from the United States and Europe have demonstrated that HS-CRP
Version 14: 14/06/2007 2

is a good predictor of future coronary events [19-21]. It is also well
established that prognosis in syncope is related to the presence of underlying
heart disease [5], and all existing syncope CDRs include either a history of
congestive heart failure [11,14,15] or underlying cardiac disease [12,13].
Troponin I (TnI) is commonly measured 12 hours post syncope to rule out
acute myocardial infarction (AMI) and has only been assessed as a
prognostic marker in syncope in one previous study [22]. This study took
Troponin levels as early as four hours after the syncopal episode, when
sensitivity for AMI is poor (0.64) [23]. We hypothesize that a routine TnI is
unnecessary and aim to assess the value of a 12-hour troponin as both an
AMI rule-out marker, and as a syncope prognostic marker.
Results of pilot study
A pilot study was performed between November 2005 and January 2006. 99
consecutive patients with undifferentiated syncope were enrolled (1.6 per
day). 32 patients were high-risk, 51 medium and 16 low according to our
existing departmental syncope guidelines (based on the European Society of
Cardiology [9,10], American College of Physicians [6,7] and American
College of Emergency Physicians guidelines [8]). 72 patients had BNP
measured, 25 (35%) were 100 pg/ml or greater, and 3 were over 1000 pg/ml.
44 patients were admitted to hospital and 55 were discharged from the ED. 30
of those admitted had troponin I measured. Only one of these was raised
(14.40 ng/ml) and this was thought to be due to an AMI. 66 patients had CRP
measured, 16 were raised (>5 mg/l). There were 11 serious outcomes. 5
patients had died by 3 months, and 6 others had an alternative serious
outcome. 8 of the 11 patients had a serious outcome by 1 week. The
percentage risk of serious outcome at 7 days, 1 month and 3 months was
therefore 8.1%, 8.1% and 11.1% respectively. 9 of 11 serious outcome
patients had BNP measured, 6 were >100, 3 were >1000 and all patients with
a BNP >1000 were in the serious outcome group (2 died, 1 AMI) [24]. The
OESIL score, the SFSR and our existing departmental guidelines all showed
some ability to risk stratify syncope patients [25].
Primary aim
To develop and validate a Clinical Decision Rule (CDR) using history,
examination, electrocardiographic (ECG) and biochemical markers, to predict
one month outcome for patients presenting with syncope to the ED.
Research questions
(1) Can a syncope CDR using specific components from the history and
examination, ECG characteristics and biochemical markers (HS-CRP and
BNP) predict one-month outcome in ED patients presenting with syncope?
(2) Can HS-CRP or BNP predict one-month outcome in patients presenting to
the ED with syncope?
Version 14: 14/06/2007 3

(3) Are biochemical markers in isolation better than history, examination and
ECG characteristics at predicting one-month outcome, or are they more useful
in conjunction with them to improve the accuracy of a CDR in predicting onemonth
outcome in syncope patients presenting to the ED?
(4) How does the validation of this CDR compare with the validation of
existing ED syncope CDRs in the UK ED population?
(5) Is it necessary to measure a 12-hour troponin I level in all patients
presenting to the ED with syncope?
Methods: Setting
The ED of the Royal Infirmary of Edinburgh (85,000 adult attendances per
annum).
Target population
From the pilot study it was estimated that there are 1200 patients presenting
to our ED per annum who are eligible for enrolment into the study. With an
improved recruitment strategy we aim to recruit between 800 and 1000
patients per annum.
Sample size calculation and statistical power
Statistical advice was sought early during planning of the pilot and main study.
With a sample size of 500 patients, if the one month adverse outcome rate is
10.0% [15,25] at the average value of any predictor variable, then there will
be 80% power of showing that this variable has a statistically significant effect
on ‘serious outcome’ (p

Exclusion criteria
� Patients under 16
� Patients previously recruited
� Patient with a good history of seizure or a prolonged (>15 minutes)
post-ictal phase
� Patients who are unable to give written or verbal consent and who do
not have a relative or guardian to give written assent
� Patients whose collapse is suspected to be due entirely to excessive
alcohol consumption
� Suspected alcoholic/epileptic seizure
� Near-syncope (i.e. no loss of consciousness)
Enrolment into derivation study
Eligible patients will be flagged at the ED high dependency triage area and a
data collection form (DCF) will be placed in the patient’s records. Assessment
of patients will be carried out by routine ED medical staff who will also be
responsible for deciding eligibility. A decision to enrol a patient will not be later
overturned by the study team and enrolled patients will be analysed on an
intention to treat basis.
Because the treating doctor is enrolling eligible patients and completing the
data collection form there is theoretically a possible selection bias. Sicker
patients may be excluded because of the time required to complete study
paperwork. This has been addressed by reducing the paperwork required to
be completed at the time of enrolment to an absolute minimum. The study
researcher (MR) will complete the rest of the data collection i.e. blood results
at a later time. The treating doctor should complete as much as possible of
the DCF at the time of patient enrolment. If for some reason information is not
available they should leave the appropriate part of the form blank. The study
researcher (MR) will review the data collected in the ED and will document
any further information that becomes available later. Data obtained after the
patient’s ED attendance will be marked as retrospectively obtained data for
the purposes of analysis in both the DCF and also in the study database. It is
important that any derived CDR is based on information that is available to the
ED doctor at the time of seeing a patient. It is also important for a CDR to
have as accurate information as possible and therefore the end analysis will
include the opportunity to look at this retrospectively available information.
We aim to have a minimum of 500 patients in the derivation cohort available
for final analysis. Because there will be some patients that are lost to followup,
inevitably some for whom all data is not obtained, and a small proportion
who will be excluded later due to them having an obvious cause necessitating
admission on presentation to the ED, we will aim to recruit 550 patients during
the 9 month derivation study.
Version 14: 14/06/2007 5

Assessment
All patients will undergo a standardised assessment using 32 pre-determined
variables (9 focussed on clinical features, 10 on past medical history and 13
concerning current medication) and 14 examination variables. These were
selected after careful systematic review of the literature to identify
characteristics that have previously been shown to be associated with serious
outcome.
After a full history and examination, all patients will undergo a 12-lead ECG,
lying standing blood pressures and a ‘BM stix’ glucose estimation. All patients
will have two red (2.7ml EDTA KE), one orange (4.7ml Lithium-Heparin-Gel)
and one yellow tube (2.7ml Glucose FE) taken and a full blood count, urea,
creatinine, glucose, electrolytes, liver function and HS-CRP will be formally
measured in the hospital laboratory. One green tube (3ml Coagulation 9 NC)
will also be taken, spun down in the biochemistry laboratory and the plasma
kept for storage. A study label will be placed onto the laboratory request form,
and samples will be sent to the hospital laboratory in the usual manner.
Near-patient BNP testing will be performed using a small quantity of blood
from the other ‘spare’ red (2.7ml EDTA KE) tube using the Biosite Triage®
point of care machine. For any patient in whom the near-patient BNP test has
not been performed (i.e. physician or machine error) this will be performed the
following day using the serum from the patient’s original EDTA sample, which
was taken on the patient’s presentation. This will have been spun down after
testing and stored immediately in the fridge in the biochemistry department.
Treating physicians will not be blinded to the result of the BNP test.
All admitted patients will have a laboratory based Troponin I at least 12 hours
post syncope performed by the admitting team, by the study researcher (MR)
or added onto an existing sample in the laboratory which has been taken
between 12 hours and 7 days after the syncopal episode. Patients still in the
ED at 12 hours post ED arrival will be defined as admitted.
A consecutive cohort of 100 derivation study patients who were discharged
from the ED will be selected to return for an interval Troponin I. The patients
will return to the Clinical Research Facility of the Royal Infirmary of Edinburgh
to a special clinic run by one of the facility’s research nurses. Patients will be
given standard travelling expenses of £30. The troponin test will be performed
as soon as possible after ED discharge but no later than 7 days and no
sooner than 12 hours after the episode of syncope. Troponin levels

After formal routine laboratory testing the red (2.7ml EDTA KE), orange (4.7ml
Lithium-Heparin-Gel), yellow (2.7ml Glucose FE) and green tubes (3ml
Coagulation 9 NC) for all enrolled patients will be spun down, separated and
stored initially in a refrigerator in the biochemistry laboratory. The samples
will then be frozen at minus 20 degrees Celsius. Once a week, the previous
week’s samples will be thawed and divided into smaller aliquots and refrozen
at minus 80 degrees Celsius for further analysis (which will include Vidas D-
Dimer and markers inflammation and ischemia) at a later date. These frozen
samples will be stored initially in a research freezer in the ED and later in the
Centre for Inflammation Research, Queen’s building, University of Edinburgh
under the supervision of Dr Matthew Reed, for an indefinite period.
In order to quantify the number of eligible patients not being enrolled into the
study, a daily search of all ED electronic patient records (EPRs) will
conducted throughout the study using Microsoft Business Objects® looking
for the keywords 'syncope', 'collapse', 'faint', 'loss of consciousness' or 'loc'
appearing anywhere on the EPR. All EPRs with one of these terms will then
be hand searched and a decision made from the notes about whether a
patient presented with a possible syncopal event and whether or not they
were eligible for enrolment. Out of those eligible for enrolment, it will be
established how many were successfully enrolled and how many were missed
and for what reason (i.e. missed by doctor, refused consent, consent not able
to be obtained). A database will be compiled of those patients who were
eligible but who were not enrolled along their demographic details and at the
conclusion of the derivation stage, these will be compared to those patients
who are recruited into the study and statistical analysis will be performed.
A Microsoft Access® database has been designed for data entry, which will
be performed by MR. A separate database will be used for derivation and
validation cohorts. A second researcher will check a random selection of 100
patient data collection forms from the derivation study in order to ensure
accuracy of data entry. If errors are found, this check will be extended to the
entire database. The contents of the Microsoft Access® database will be
exported into Microsoft Excel®, SPSS and SAS for statistical analysis. A
patient flow chart will be constructed detailing patient recruitment.
Endpoint measures
Primary outcome: Serious outcome or all cause death at one month
Definition of serious outcome:
� AMI (history of chest pain or ECG changes and troponin I>0.2),
� Life-threatening arrhythmia (as defined prior by expert panel - recorded
episode of VF, sustained VT, ventricular pause greater than 3 seconds,
Mobitz type 2, Wenkebach, bifascicular or complete heart block, or
symptomatic bradycardia

� Insertion of Pacemaker or Internal Cardiac Defibrillator device or a
decision that the patient requires such a device, within 1 month of the
ED attendance.
� Pulmonary embolus (confirmed on VQ/CTPA and requiring treatment),
� Cerebrovascular accident/subarachnoid haemorrhage (CT or LP
diagnosis),
� Haemorrhage requiring a blood transfusion of two units or more during
inpatient stay, and an acute surgical procedure or endoscopic
intervention secondary to a suspected cause of syncope.
As part of the data analysis process, two researchers from the expert panel
will independently review notes of all patients who are considered to have
reached endpoint in order to independently agree this. These experts will be
blinded to the presence or absence of all predictor variables.
Secondary outcome: Syncope related death (death due to a recognised
cause of syncope as decided by the expert panel)
Outcome, final discharge diagnosis and information regarding inpatient stay,
investigations, interventions and serious outcome will be identified by way of
ED, hospital, ISD, Registrar General and GP records, death certificates and
post mortem results, and via a structured telephone interview with the patient.
Endpoint review
� An expert committee will be convened to review all patients who are
suspected of having a primary outcome to determine agreement.
� All patients whose cause of syncope was known to the treating ED
physician at the time of enrolment and whose underlying cause
necessitated admission will be removed from the final analysis after
expert committee review:
o Suspected AAA rupture with SBP < 90mmHg,
o AMI - chest pain with ST elevation >1mm in any lead,
o Severe onset of headache with CT or LP diagnosis of SAH in
ED,
o Recorded episode of VF, sustained VT, CHB or symptomatic
bradycardia

Statistical analysis
After derivation data collection, multivariate logistical regression analysis will
be performed to determine factors associated with the endpoints, and a CDR
will be developed. Previous syncope risk stratification models have either
used a stepwise multiple linear logistic regression model, a stepwise Cox
proportion hazards model or recursive partitioning. The stepwise procedures
can be criticised because they are automatic methods, based on the
independent statistical significance of the potential risk factors. Thus a
relevant variable may be excluded if it has a moderate correlation with a
variable already in the model. Conversely, the large number of risk factors
considered also raises issues of multiple testing and the possibility of
detecting false positive associations. Recursive partitioning has an even
greater risk from false positive findings as the method, in effect, allows for
complex interactions among the potential risk factors. Our approach will utilise
multiple linear logistic regression but model development will not be
automatic.
In a preliminary stage of analysis, we will attempt to reduce the number of
variables using principal component analysis to identify ‘factors’ among the
set of contender variables. Thus variables that tend to be mutually correlated
will be combined into single variables. Subsequent logistic regression
modelling will take a knowledge-based approach, incorporating widely
accepted risk factors regardless of their statistical significance, while requiring
a high level of significance for variables that other studies have found to be
unimportant. The biochemical markers will be assessed primarily on their
ability to add value to the risk stratification based on conventional risk factors,
but their effects will also be assessed univariately.
Troponin will be analysed in admitted patients and in 100 consecutive
discharged derivation cohort patients. Troponin I will not be included as a
predictor variable for the purposes of the derivation of the CDR due to the
result not being available to the admitting doctor at the time of ED decision
making.
Models will be developed for all of the main outcome variables, from which
practical CDRs will be formulated. The subjective elements in the model
derivation make it particularly important that the models are validated
independently.
The validation dataset will be obtained while the deviation dataset is being
analysed, and no data from it will be released until the CDR has been
finalised. The prospective data collection in this study should minimise the
amount of missing data, but for model development, analysis of all subjects is
important. For any categorical variables ‘missing’ will be considered as a
category, while for continuous variables mean imputation will be used in
conjunction with a binary dummy variable to indicate whether or not the
numerical variable has been imputed.
Prospective validation
We aim to have 500 patients available for final analysis and therefore we will
aim for 550 to be recruited as above in an attempt to validate the developed
CDR. Statistical analysis will be used to assess the CDR performance.
Version 14: 14/06/2007 9

A separate further study is planned that will assess the impact of using the
CDR on admission rate, outpatient follow-up rate and detection of serious
outcomes. A separate cost benefit analysis study will also be performed.
Ethics Committee approval
This study received full ethical approval from the MREC for Scotland A Ethics
committee (Reference: 06/MRE00/107) and Lothian REC (Reference:
06/S11ADMIN/151) on January 8 th 2007, and a submitted substantial
amendment was approved on 22 nd February 2007. Lothian R&D
management approval was also obtained on 24 th January 2007 (Reference:
2006/R/AE/03).
Informed consent and information sheet
MREC approval has been obtained for the study Patient Information Sheet,
Patient Information Sheet for Relatives, Patient Consent Form, Patient
Consent Form for Relatives and GP Information Sheet.
Dissemination
It is anticipated that the study results will be submitted for to a internationally
renown peer reviewed journal and presented at both national and international
meetings and conferences.
Sponsor
Chief Scientist Office of the Scottish Executive (Reference No: CSO/CAF/01)
Start date (recruitment): March 2007
Finish Date (18 months recruitment): July 2008
Finish Date (1 month follow up): August 2008
Expected Reporting Date: November 2008
Version 14: 14/06/2007 10

Key references
1. Morag R. Syncope. In: Peak DA, Talavera F, Halamka J, et al.
eMedicine on line Emergency textbook.
www.emedicine.com/emerg/topic876.htm
2. Maisel WH, Stevenson WG. Syncope: getting to the heart of the
matter. N Engl J Med 2002; 347: 931-933.
3. Soteriades ES, Evans JC, Larson MG, et al. Incidence and prognosis
of syncope. N Engl J Med 2002; 347: 878-885.
4. Kapoor WN, Karpf M, Wieand S, et al. A prospective evaluation and
follow-up of patients with syncope. N Engl J Med 1983; 309: 197-204.
5. Kapoor WN, Hanasu BH. Is Syncope a Risk Factor for Poor
Outcomes? Comparison of Patients With and Without Syncope. Am J
Med. 1996; 100: 646-655.
6. Linzer M, Yang EH, Estes NA III, Wang P, Vorperian VR, Kapoor WN.
Diagnosing syncope. 1. Value of history, physical examination, and
electrocardiography: Clinical Efficacy Assessment Project of the
American College of Physicians. Ann Intern Med 1997; 126: 989-96.
7. Idem. Diagnosing syncope. 2. Unexplained syncope: Clinical Efficacy
Assessment Project of the American College of Physicians. Ann Intern
Med 1997; 127: 76-86.
8. Molzen GW, Suter RE, Whitson R. American College of Emergency
Physicians: Clinical Policy: critical issues in the evaluation and
management of patients presenting with syncope. Ann Emerg Med.
2001; 37: 771-776.
9. Brignole M, Alboni P, Benditt DG, et al. Task Force on Syncope,
European Society of Cardiology. Guidelines on management
(diagnosis and treatment) of syncope-update 2004. Executive
Summary. European Heart Journal. 2004; 25(22): 2054-72.
10. Brignole M, Alboni P, Benditt D, et al. Task Force on Syncope,
European Society of Cardiology. Guidelines on management
(diagnosis and treatment) of syncope. European Heart Journal 2001;
22(15): 1256-306.
11. Martin GJ, Adams SL, Martin HG, Mathews J, Zull D, Scanlon PJ.
Prospective evaluation of syncope. Ann Emerg Med 1984; 13: 499-
504.
12. Oh JH, Hanusa BH, Kapoor WN, et al. Do Symptoms Predict Cardiac
Arrhythmias and Mortality in Patients With Syncope? Arch Intern
Med. 1999; 159: 375-380.
13. Colivicchi F, Ammirati F, Melina D, et al. Development and prospective
validation of a risk stratification system for patients with syncope in the
emergency department: the OESIL risk score. European heart journal
2003; 24: 811-819.
14. Sarasin FP, Hanusa BH, Perneger T, et al. A Risk Score to Predict
Arrhythmias in Patients with Unexplained Syncope. Acad Emerg Med.
2003; 10(12): 1312-1317.
15. Quinn JV, Stiell IG, McDermott DA, et al. Derivation of the San
Francisco Syncope Rule to Predict Patients With Short-Term Serious
Outcomes. Ann Emerg Med. Feb 2004; 43: 224-232.
Version 14: 14/06/2007 11

16. Quinn JV, McDermott DA, Stiell IG, et al. Prospective Validation of the
San Francisco Syncope Rule to Predict Patients With Serious
Outcomes. Ann Emerg Med. May 2006; 47: 448-453.
17. Doust JA, Glasziou PP, Pietrzak E, et al. A systematic review of the
diagnostic accuracy of natriuretic peptides for heart failure. Arch Intern
Med 2004; 164: 1978-84.
18. Doust JA, Pietrzak E, Dobson A, et al. How well does B-type
natriuretic peptide predict death and cardiac events in patients with
heart failure: systematic review. BMJ 2005; 330: 625-627.
19. Alonso-Martinez JL, Llorente-Diez B, Echegaray-Agara M, et al. Creactive
protein as a predictor of improvement and readmission in heart
failure. European Journal of Heart Failure 2002; 4(3): 331-6.
20. Ridker PM, Cushman M, Stampfer MJ, et al. Inflammation, aspirin, and
the risk of cardiovascular disease in apparently healthy men. New
England Journal of Medicine 1997; 336(14): 973-9.
21. Koenig W, Sund M, Frohlich M, et al. C-Reactive protein, a sensitive
marker of inflammation, predicts future risk of coronary heart disease in
initially healthy middle-aged men: results from the MONICA (Monitoring
Trends and Determinants in Cardiovascular Disease) Augsburg Cohort
Study, 1984 to 1992. Circulation 1999; 99: 237–42.
22. Hing R. Harris R. Relative utility of serum troponin and the OESIL
score in syncope. Emergency Medicine Australasia 2005; 17(1): 31-38.
23. Ebell MH, White LL, Weismantel D. A systematic review of troponin T
and I values as a prognostic tool for patients with chest pain. J Fam.
Pract. 2000; 49: 550-6.
24. Reed MJ, Newby DE, Coull AJ, Jacques KG, Prescott RJ, Gray AJ.
Near-patient BNP is able to predict three-month serious outcome in
adult syncope patients presenting to the Emergency Department. (in
press)
25. Reed MJ, Newby DE, Coull AJ, Jacques KG, Prescott RJ, Gray AJ.
The Risk stratification Of Syncope in the Emergency department
(ROSE) pilot study: a comparison of existing syncope guidelines.
Emerg Med J 2007; 24: 270-275.
Version 14: 14/06/2007 12