To the Editor: Regarding the comments of Dr. Maj, we stipulated in our article that the prevalence of rapid cycling in patients with bipolar patients on the basis of data from case control or other nonrandom subject selection is invalid: "A rapid-cycling course was found in 24.2% of the patients, but this rate was probably inflated by selection of rapid-cycling cases in some studies…that had unusually high proportions of such cases (40.5%–55.8%)" (p. 1435). This rate averaged to a mean of 16.3% (SD=2.82) in unselected samples versus an obviously inflated mean of 46.1% (SD=9.03) in selected groups (F=49.4, df=1, 8, p=0.0001). In addition, in Dr. Maj’s cited study, whereas the prevalence of rapid cycling among all patients with bipolar disorder was 13.6%, in those with data regarding cycling status and sex, it was 33.3% (37 out of 111), as stated (table 1). However, estimating the prevalence of rapid cycling among patients with bipolar disorder, or even absolute, sex-based, rapid-cycling rates (also highly variable and possibly unreliable), was not the purpose of our report.

The aim of our meta-analysis, again, was to compare sex-based, rapid-cycling rates with rates from studies that provided data for women and men with bipolar disorder with and without rapid cycling to test the hypothesis that rapid cycling is more prevalent in women. This hypothesis arose from reports of extraordinarily high proportions of women among patients with rapid cycling (71.7% women versus 28.3% men). Such findings pertain to excesses of women in clinical study groups but not necessarily to the rates of rapid cycling among women. In contrast, sex-based, rapid-cycling risk ratios indicated a much smaller sex difference. Since these risk ratios did not differ significantly between studies involving unselected or selected samples, data were pooled from the 10 studies that were analyzed. The prevalence of rapid cycling averaged 29.6% among women and 16.5% among men—a modest 1.79-fold difference—with an average within-study rate ratio of 1.94 (SD=1.06). The results indicate that the prevalence of rapid cycling is evidently much less excessive among women with bipolar disorder than is widely believed.