Does the quality of antidepressant trials affect outcome estimates?

Article type
Authors
Cipriani A, Malvini L, Furukawa T, Barbui C
Abstract
Background: Trial quality can affect estimates of efficacy in randomized studies (RCTs) and in meta-analyses. However, to date the relationship between trial quality and outcome has never been investigated in antidepressant trials.

Objectives: To investigate whether the quality of primary studies could influence the estimate of effect size in a systematically retrieved homogeneous sample of RCTs, taking into account possible estimate confounders.

Methods: Systematic review of RCTs comparing fluoxetine with any other antidepressants for major depression. The quality of RCTs was assessed using the Jadad scale1, the Cochrane Collaboration Depression, Anxiety and Neurosis quality assessment instrument (CCDAN)2 and the CONSORT (Consolidated Standards of Reporting Trials) Statement3. Data were extracted independently by two investigators. We pre-tested our methods by means of an inter-observer reliability study. We ran subgroup and meta-regression analyses for both effect size and dropout rate, to explore important differences among trials that might be expected to alter the magnitude of treatment effect.

Results: Thirty-nine RCTs (n=4584) met the inclusion criteria for efficacy analysis and 74 (n=8227) for tolerability. About tricyclic antidepressants (TCAs), the efficacy estimate did not materially change according to different instruments. About other selective serotonin reuptake inhibitors (SSRIs), the overall statistically significant efficacy estimate (Peto OR 1.26, 95% CI 1.05 to 1.50) was confirmed in high quality RCTs according to the Jadad scale (Peto OR 1.25, 95% CI 1.02 to 1.53) and the CCDAN instrument (Peto OR 1.24, 95% CI 1.04 to 1.50). However, no high quality RCTs were found according to the CONSORT criteria. About tolerability, sensitivity analyses did not result in any meaningful differences. In the meta-regression analysis, the quality score was not significantly related to treatment effect.

Conclusions: Conflicting results were found when considering the CONSORT component approach to quality assessment. Only a few trials reported on the methods of the three most important key aspects of RCTs despite their importance. We simply cannot say how quality of primary studies could influence outcome assessment in antidepressant trials. Efforts to improve the quality of reporting will better this situation.

References
1. Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ, McQuay HJ. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Controlled Clinical Trials 1996; 17:1-12.
2. Cochrane Depression, Anxiety and Neurosis Group. Cochrane Collaboration Depression, Anxiety and Neurosis Quality Rating System [Online]. http://web1.iop.kcl.ac.uk/IoP/ccdan/qrs.htm [accessed 8 August 2006].
3. Moher D, Schulz KF, Altman DG, for the CONSORT Group. The CONSORT Statement: revised recommendations for improving the
quality of reports of parallel-group randomised trials. Lancet 2001; 357:1191-4.