Studies of causation must meet these additional criteria

• Clearly identified comparison group for those at risk of, or having, the outcome of interest (i.e. randomized controlled trial, quasi-randomized controlled trial, non-randomized controlled trial, cohort analytical study with case-by-case matching or statistical adjustment to create comparable groups, case-control study)

• Masking of observers of outcomes to exposures (this criterion is assumed to be met if the outcome is objective), observers of exposures masked to outcomes for case-control studies, or masking of subjects to exposure for all other study designs

• Analysis consistent with study design

The results of the primary studies are shown graphically as odds ratios (see Glossary) in Fig 3. Odds ratios falling to the left of the vertical line indicate that the outcome (no treatment response) occurred less frequently in patients treated with the antidepressant. The smaller the odds ratio, the larger the treatment effect found in that particular study. For each study, the central diamond represents the most likely value of the odds ratio and the box around the odds ratio shows the 95 per cent confidence interval (CI). The larger studies (e.g. Thase et al.'(25) have narrower confidence intervals because they are larger and therefore have less random error and greater precision.

Fig. 3 Antidepressant versus placebo for the treatment of dysthymia. (Data from Lima and Moncrieff. (25>)

From the figure, it can be seen that, apart from the very small study by Stewart (1995) the odds ratios of the all the studies fall on the left-hand side of the line and therefore found approximately the same effect. There is considerable overlap in the confidence intervals from study to study. This means that there is unlikely to be significant heterogeneity between the studies (see ChapteL6:.1..1.2). In the absence of heterogeneity, it was probably reasonable to combine the results of the studies to produce a pooled odds ratio which is shown as the bottom unfilled lozenge in the figure. Combining the study results produces a more precise estimate of the drug's relative effectiveness, with tighter confidence intervals. The overall pooled odds ratio for no treatment response for antidepressants compared with placebo is 0.68 (95 per cent CI (see Glossary.), 0.59-0.78).

Was this article helpful?

0 0
Invisible Viagara

Invisible Viagara

You are about to discover the "little-known" techniques, tricks and "mind tools" that will show you how to easily "program" your body and mind to produce an instant, rock-hard erection. Learn how to enjoy all of the control, confidence and satisfaction that comes from knowing you can always "rise to the challenge" ... and never have to deal with embarrassment, apologies, shyness or performance anxiety in the bedroom, ever again.

Get My Free Ebook

Post a comment