A longitudinal study of plasma cortisol and depressive symptomatology by random regression analysis

The authors utilized a random regression model to test the longitudinal relationship between depressive symptomatology and plasma cortisol levels obtained before and after the administration of dexamethasone in 62 affectively ill inpatients. This statistical model for longitudinal studies permits the inclusion of subjects with incomplete data as well as subjects measured at different time points. The most significant relationships were found between decreases in depressive symptoms and decreases in the 8:30 AM predexamethasone and the 4:00 PM postdexamethasone cortisol values. Patients were also classified as responders or nonresponders, and the rate of change in several plasma cortisol measures were separately analyzed for these two groups. No differences in the rate of change in plasma cortisol levels were found between responders and nonresponders. These results suggest that the decreases in cortisol production associated with clinical improvement may be partially explained by a regression toward the mean effect. Some of the possible explanations for these results are discussed.