Minimising the impact of reporting bias by using data from clinical study reports for the determining the drop-out rates in antidepressant trials

Article type
Authors
Sharma T1, Schow Guski L1, Freund N1, Muscat Meng D1, Gøtzsche P1
1Nordic Cochrane Centre
Abstract
Background:
Reporting bias distorts the true effectiveness of interventions, leading to unreliable results. Using clinical study reports (CSRs) as source documents for systematic reviews reduces the potential for bias due to selective reporting.

Objectives:
To study the comparative drop-out rates for placebo and antidepressants from double blind randomised controlled trials.

Methods:
We obtained CSRs of five antidepressant drugs from the European Medicines Agency and the UK’s Medicines and Healthcare products Regulatory Agency. We included all double-blind randomised, placebo-controlled trials for any indication and two researchers extracted the overall dropout rate independently and performed meta-analysis using the Mantel-Haenszel method (fixed-effect model). One researcher extracted the dropouts due to adverse events and lack of effect and this was checked by another. Sensitivity analyses were performed using Peto’s odds ratio and beta binomial methods, due to presence of null events, and by excluding unreliable trials.

Results:
We included 73 trials, which corresponded to 11,057 participants on drugs, and 7369 on placebo. We often noted a modified intention-to-treat principle within the CSRs, where participants without data on safety were not accounted for. More participants on the active drug dropped out than those on placebo, risk ratio 1.08 (95% confidence interval (CI) 1.03 to 1.13), and when we removed hree trials that had an enriched design for a sensitivity analysis, the difference was a risk ratio of 1.12 (95% CI 1.07 to 1.18). There were more dropouts due to adverse events on the active drug than on placebo, risk ratio 2.63 (95% CI 2.33 to 2.96) and accordingly there were fewer dropouts due to lack of effect, risk ratio 0.47 (95% CI 0.43 to 0.53).

Conclusions:
Using CSRs we were able to identify that more trial participants dropped out when on the active drug than on placebo, which if considered to reflect the patients’ assessment of benefits and harms, is a disturbing finding. The importance of using CSRs for reviews was also a key finding.

Patient or healthcare consumer involvement:
No patients were involved in this review but we aim to involve them during the dissemination of our results.