Dr. Hoff and Colleagues Reply

To the Editor: Dr. Shuer raises some interesting and important issues with regard to our article. She contends that documentation of the phase of the menstrual cycle is critical in understanding the relationship of gonadal hormones to cognitive abilities because of the variation in hormones throughout a normal menstrual cycle. We agree with this point and know that documentation of the menstrual phase is most readily performed in the healthy, nonpsychiatrically impaired individuals Dr. Shuer has vigorously studied. In contrast, our study was an investigation into treatment-resistant female patients with schizophrenia. As we stated in the article, we found patients and staff to be unreliable in providing exact information regarding menstrual cycles. As is true of many schizophrenia patients, a number of our patients were amenorrheic, which complicates the picture even further. Additionally, because testing could not always correspond to the exact day of blood draws because of patient cooperation or ward logistics, we could not guarantee the exact correspondence of serum levels to day of testing. We dealt with these difficulties by sampling weekly over a 4-week period as an estimate of one menstrual cycle. The mean of a group of observations is the best single representation of those observations. We performed correlations of average serum level, variability, and highest level with level of cognitive functioning; all produced significant results. Despite the impreciseness in our measurement, we detected strong relationships, which would speak more to the robustness of these findings than to their lack of validity.

The second point Dr. Shuer makes is that it is unacceptable to measure estrogen levels in patients who are receiving hormone-replacement therapy or oral contraceptives because one must measure all of the metabolic products of these compounds. Because each of these compounds has a complicated pattern of release over time, it was not feasible in this study to measure all of their metabolic products because of the large number of necessary blood draws and associated laboratory costs. Additionally, since we were measuring levels in relatively young patients, some of whom were still ovulating, there was an ethical issue regarding discontinuing contraceptive therapy for these patients. Because of the relatively small numbers of female patients who were available for study (1), a similar study without the use of exogenous estrogen might take an inordinately long time, might leave us with a biased study group, or might not be undertaken at all.

Moreover, when we separated patients into those receiving oral contraceptives or hormone-replacement therapy (N=10) and those receiving neither (N=12), the pattern and strength of correlations of average estrogen with our summary global scale (and other scales) were exactly the same as were found in the total group. Correlations were r=0.84, p<0.002, and r=0.83, p<0.001, respectively. For the entire group, the correlation was r=0.86, p<0.0000001. These findings further support the meaningfulness of the average estrogen level as a global average measure of estrogen exposure.

In summary, we believe our findings are valid and justifiable. As we acknowledged in the article, there is room for refinement of our techniques. We believe, however, that our preliminary study is a good first step at studying this important issue.