To The Editor: In their article, Dr. Viguera et al. concluded that the overall risk of at least one recurrence of a new mood episode during pregnancy was 71% among women who discontinued the use of a mood stabilizer 6 months prior to conception to 12 weeks postconception (relative to women who continued treatment with a mood stabilizer). As indicated in the article, the two groups of women differed with regard to several characteristics.
In the multivariate modeling or risk-factors-adjusted analysis, only some of the predictors of recurrence were included. It is not clear to us whether all the statistically significant predictors were covaried. For example, rapid cycling, which is a predictor of recurrence, was not entered as a covariate, although it did differ between the two groups at baseline.
We are also puzzled by the way the authors presented the issue of current adjunctive antipsychotic use. There was a large difference between the two groups of women. The use of current adjunctive antipsychotics was reported in 21% of subjects who discontinued the use of a mood stabilizer and in 41% of those who continued treatment. This difference is close to significance (p=0.07). The list of predictors of recurrence did not include the adjunctive use of antipsychotics, nor was it mentioned whether adjunctive use was associated with the risk of recurrence. Given the likelihood that antipsychotics may be mood stabilizers, should not this factor have received attention in the data analysis?
Additionally, it would be necessary to know which subjects discontinued the use of a mood stabilizer more than 1 month prior to conception, since the natural course of the illness would likely suggest that the longer period of time an individual is medication free, the higher the likelihood of recurrence. It can be hypothesized that if the data analysis had taken into consideration the length of time since discontinuation, the results might have shown that discontinuing the use of a mood stabilizer close to conception may produce different recurrence rates.
Symptoms of change in energy level, appetite, concentration, and psychomotor retardation may all occur in normal, healthy pregnancies and may not be associated with major depression. How did the analysis adjust for this potential confounder? This may explain the much greater frequency of depressive episodes relative to manic episodes. In addition, the nature of a woman’s previous episode may predict the type of relapse experienced during pregnancy.
Finally, were there any untoward conditions in the newborn children that were part of this study? For many clinicians, the recommendation whether to discontinue or continue medication during the first trimester is influenced not only by the severity of maternal illness but also the perceived risk to the exposed fetus.
Drs. Mazer-Poline, Rifkin, and Walch report no competing interests. Dr. Geisler has served on the speakers bureaus of Pfizer and AstraZeneca.
This letter (doi: 10.1176/appi.ajp.2008.08010036) was accepted for publication in February 2008.