Background: Meta-analyses (MAs) of randomized controlled trials (RCTs) have the potential to provide the highest level of evidence, but the quality of published MAs has not been systematically assessed. Therefore, we determined reliability was significant (kappa = 0.89; p < 0.05). the quality of reporting in MAs of RCTs of pharmacotherapy for major depressive disorder (MDD) in adults (18-65 years) without comorbidities and examine trends over time.
Methods: MEDLINE, EMBASE, Healthstar, Psychlit and Cochrane databases were searched (1980-2002) by 4 independent reviewers for MAs of RCTs. Articles meeting inclusion criteria were blinded. Inter-rater reliability (kappa) was evaluated using a test-retest strategy on 4 articles. Quality was (p = 0.74) did not detect a difference in quality of assessed using the QUOROM checklist. Time trends were evaluated by calculating Spearman's rho.
Results: One hundred articles were identified, 68 were excluded [co-morbidities (9), inappropriate comparator (13), inappropriate outcome (15), article not available (5), inappropriate patient population (15), and inappropriate study design (11)]; 32 were included. Initial kappa was 0.81 (p < 0.05). After resolution of disagreements, the test-retest The mean overall quality score was 50.2% (SD 15.8%, range = 16.7-88.9%). The overall score for Titles was very poor (22%), Abstracts (40%) and Methods (49%) were poor, while overall Results score was minimally acceptable (54%). Good quality scores were found for Introduction (91%) and Discussion (97%). No time trends were identified using Spearman's correlation analysis (rho 0.05; p = 0.79). The Mann-Whitney U test articles published before and after the QUOROM.
Conclusion: Despite quality guidelines, the average quality of published MAs of antidepressants is barely acceptable (50.2%). A need exists for adherence to standardized reporting and quality guidelines.