Abstract
Background:
Mela-analyses of randomized controlled trials (RCTs) constitute the highest level of evidence, but their usefulness depends on their quality.
Objective:
To assess tho reporting and scientific quality of meta-analyses of RCTs on treatments for anxiety disorders.
Methods:
Criteria for peer-reviewed, full-text retrieval included meta-analyses of RCTs of drugs versus active ingredient placebo, standard care, or psychotherapy. Sample populations were required to meet Diagnostic and Statistical Manual of Mental Disorders or International Classification of Diseases and Related Health Problems diagnostic criteria for anxiety disorders. Two reviewers independently searched EMBASE, EBM Reviews. Ovid MEDLINE, Ovid HealthSTAR, and International Pharmaceutical Abstracts from inception to August 2007. Search terms Included meta-analysis, randomized controlled trials, anxiety, anxiolytic, anti-depressant/antidepressant, and pharmacotherapy, without language restrictions. References and reviews were searched manually. Quality was assessed independently by 2 raters, using the Quality of Reporting of Meta-anatyses (QUOROM) and the Overview Quality Assessment Questionnaire (OQAQ). The QUOROM was used to assess the reporting quality of the study, using an 18-item checklist, and the scientific quality was assessed with the OQAQ's 10-item checklist, Kendall's - measured Interrater reliability with statistical significance at p less than or equal to 0.01. Means and standard deviations described the overall quality. A time series analysis was performed.
Results:
A total of 136 titles and abstracts were reviewed; 48 were retrieved, including 6 from the manual search. Thirty-two were excluded (not pooled analyses, inappropriate condition/treatment, duplications), leaving 16 studies published between 1995 and 2007. Agreement was high: t = 0.801 (p < 0.01) for QUOROM and 0.834 (p < 0.01) for OQAQ. QUOROM quality scored 61% ± 19%. Overall, the results sections of the studies scored lowest, while the introduction and discussion sections scored highest. The overall scientific qualify was 58% ± 28%. Most studies appropriately linked results to priman/ objectives but did not report how bias was avoided or how study validity was assessed. Quality Increased nonsignificantly over time.
Conclusions:
Reporting/scientific quality was considered less than fair-to-good. Stakeholders should strive for higher scientific quality of meta-analyses.