Prevalence and prediction of primary sleep disorders in a clinical trial of depressed patients with insomnia.

Insomnia-pharmacology clinical trials routinely exclude primary sleep disorders, such as obstructive sleep apnea (OSA) and periodic limb movement disorder (PLMD), with a single night of polysomnography (PSG). Given the expense of PSG, we examined whether a thorough clinical screening, combined with actigraphy, would successfully identify OSA and PLMD as part of baseline screening for a clinical trial of insomnia treatment in depressed patients. Of the 73 patients with a complete baseline dataset, 12 screened positive for OSA/PLMD (AHI > 15, or PLMAI > 15), while 61 "passed" the PSG screen. The OSA/PLMD+ patients were older (51.4 +/- 10.2 y) and took more naps (2.6 per week) than the OSA/PLMD- patients (41.3 +/- 12.8 y; and 1.1 naps per week). The combination of age and nap frequency produced a "good" receiver operating characteristic (ROC) model for predicting OSA/PLMD+, with the area under the curve of 0.82. There were no other demographic, sleep diary, or actigraphic variables, which differed between OSA/PLM + or -, and no other variable improved the ROC model. Still, the best model misclassified 16 of 73 persons. We conclude that while age and the presence of napping were helpful in identifying OSA and PLM in a well-screened sample of depressed insomniacs, PSG is required to definitively identify and exclude primary sleep disorders in insomnia clinical trials.