Background: Biases in adverse effect reporting in randomized controlled trials (RCTs) [e.g. due to investigator expectations or assessment quality] can be quantified by studying the rates of adverse events reported in the placebo arms of such trials.
Objective: We compared the rates of adverse effects reported in the placebo arms of tricyclic antidepressant (TCA) trials and placebo arms of selective serotonin reuptake inhibitor (SSRI) trials.
Methods: We conducted a literature search for RCTs across PUBMED, Scopus and the Cochrane Central Register of Controlled Trials (CENTRAL). Only studies allowing adverse effect analysis were included. Publication year ranged from 1981 to 2007.
Results: Our systematic review and meta-analysis included 143 placebocontrolled RCTs and data from 12,742 patients. Only 21% of studies used structured and systematic adverse effect ascertainment strategies. The way in which trials recorded adverse events influenced the rate of adverse effects substantially. Systematic assessment led to higher rates than less systematic assessment. Far more adverse effects were reported in TCA-placebo groups compared with SSRI-placebo groups, e.g. dry mouth (odds ratio [OR] = 3.5; 95% CI 2.9, 4.2); drowsiness (OR = 2.7; 95%CI 2.2, 3.4); constipation (OR= 2.7; 95%CI 2.1, 3.6); sexual problems (OR =2.3; 95%CI 1.5, 3.5). Regression analyses controlling for various influencing factors confirmed the results.
Conclusion: Adverse effect profiles reported in clinical trials are strongly influenced by expectations from investigators and patients. This difference cannot be attributed to ascertainment methods. Adverse effect patterns of the drug group are closely related to adverse effects of the placebo group. These results question the validity of the assumption that adverse effects in placebo groups reflect the 'drug-unspecific effects'.