Interventions for Tuberculosis Control and Elimination 2002
Interventions for Tuberculosis Control and Elimination 2002
Interventions for Tuberculosis Control and Elimination 2002
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
the number of person-years of observation. The measures are incidence<br />
rates among the exposed <strong>and</strong> unexposed <strong>and</strong> the summary measure is the<br />
relative risk (the risk among the exposed divided by the risk among the<br />
unexposed). Vaccine efficacy (in per cent) is calculated as (1 – relative<br />
risk) × 100. 732 The 95% confidence intervals were calculated (or recalculated,<br />
where appropriate) using the <strong>for</strong>mula proposed by Orenstein in his<br />
review on assessment of vaccine efficacy, 732 unless adjusted or stratified<br />
summary estimates were provided by the authors.<br />
To defray the costs incurred in clinical trials <strong>and</strong> to obtain results more<br />
quickly, it was proposed to ascertain the effectiveness of BCG vaccination<br />
by means of (retrospective) case-control studies. 733 Briefly, case-control<br />
studies start with looking at the outcome (tuberculosis) <strong>and</strong> then ascertain<br />
exposure (BCG vaccination given or not) in a group of patients with the<br />
outcome, compared to an appropriately selected control group of persons<br />
without the outcome (table 10). 734 A relative risk cannot be calculated as<br />
this measurement is confined to population-based studies. The measurement<br />
of risk in a case-control study is the odds ratio (or relative odds). For<br />
rare diseases the odds ratio approximates the relative risk in a clinical trial.<br />
The advantages <strong>and</strong> disadvantages in the use of the case-control approach<br />
are linked to its being observational, having subjects selected on the basis<br />
of disease status, <strong>and</strong> using controls from the population from which the<br />
cases emanated. 735 The advantages of case-control studies include avoidance<br />
of ethical problems arising in situations where there is already evidence<br />
that the vaccine is better than placebo; allowing much faster conduct than<br />
r<strong>and</strong>omized trials; <strong>and</strong> requiring a much smaller number of subjects. They<br />
are thus substantially cheaper to conduct than r<strong>and</strong>omized clinical trials. 735<br />
The most challenging difficulty in the design of case-control studies<br />
is the selection of appropriate controls in that they have to be selected in<br />
such a way that they are comparable to cases in every respect except <strong>for</strong><br />
the outcome. Selection bias resulting from a failure to ensure this comparability<br />
may thus invalidate any findings.<br />
The results of some of these case-control studies are summarized<br />
below. Vaccine effectiveness (in per cent) from a case-control study is<br />
estimated as (1 – odds ratio) × 100. 732 For unmatched case-control studies,<br />
the 95% confidence intervals were calculated (or recalculated where<br />
appropriate) using Woolf’s method. 734 For matched <strong>and</strong> adjusted analyses,<br />
the confidence interval published by the authors of the study was chosen.<br />
If not stated <strong>for</strong> matched studies, the confidence interval around the crude<br />
odds ratio was calculated as above.<br />
106