Research reports on treatments for bipolar disorder: preliminary assessment of methodological quality

OBJECTIVE: To assess frequencies of types of publications about bipolar disorder (BD) and evaluate methodological quality of treatment studies. METHOD: We classified 100 randomly selected articles (1998-2002) from five psychiatric journals with highest impact ratings, by topic areas, and assessed methods employed in treatment studies. RESULTS: Topics ranked: treatment (41%; 37% on pharmacotherapy) > biology (31%) > psychopathology (14%) = miscellaneous (14%). Of treatment studies, only 19% of original articles were randomized, 15% were relatively large (n > or = 50) but non-randomized, 65% were small non-randomized, case-series or -reports, and 53% relied on baseline-to-endpoint contrasts without a control group. Patient dropout rates were > or =40% in 43% of prospective studies. Only two reports provided confidence intervals; one included a power analysis, and 53% included no references on study design or statistical methods. CONCLUSION: Even in highly respected journals, the typical methodological quality of recent reports on therapeutics for BD was unexpectedly limited, and psychopathology and psychotherapies were little studied.