The purpose of this study was to examine the effect of late partial sleep deprivation in women diagnosed with major depressive disorder (non-seasonal, non-bipolar). Women were randomized to begin the study with either a placebo intervention or late partial sleep deprivation. Late partial sleep deprivation consisted of two consecutive nights of sleeping from 10PM to 2AM, and remaining awake until the following night at 10PM. The placebo intervention involved wearing a special pair of glasses (circadian adaptation glasses) designed to filter out daylight. These glasses were worn between 7PM and bedtime, and on any occasion when the women arose during the night or until 6AM.

This study utilized a quasi-experimental cross over design with two six-day periods of data collection for each subject to obtain data on mood, sleep patterns and circadian temperature rhythm before and after nonpharmacologic treatment. The subjects were 18 unmedicated women between the ages of 21 and 50 years. All data were collected during the follicular phase of the menstrual cycle.

Data collection was accomplished by home monitoring of sleep electroencephalography with the Medilog 9000-II system using standardized techniques. All sleep tapes were scored by a registered sleep technologist who was blind to the study protocol. Continuous core body temperature monitoring was accomplished with the use of the CorTemp telemetry system, using precalibrated ingestible sensors. Depression was measured with the Beck Depression Inventory and the Symptom Checklist 90-R. Diagnosis was confirmed by the use of the Structured Clinical Interview for the DSM-IIIR (SCID). Prior to intervention, subject's urine was screened for substances that could potentially alter mood, sleep and rhythm patterns (e.g., drugs of abuse). During both six day periods of data collection, the participants maintained structured diaries which examined patterns of daily activity and self reports of sleep. Prior to each intervention, women participated in two consecutive nights of EEG with forty eight hours of concurrent temperature monitoring, and during this period completed morning and evening measures of mood. Days three and four were for intervention. Post intervention evaluation using the same measures as pre-intervention assessment occurred on days five and six.

Analysis of the data reveals that the placebo yielded no significant differences in depression (t = 0.32, p $<$.74) or fatigue (t = 0.46, p $<$.64). No significant differences were noted between pre & post active intervention. However, 67% of the sample reported a 25-75% improvement in depression at some point in the two post-active intervention days. When the data from responders were compared to those who did not respond, three significant differences emerged: (1) There was a significant difference in REM latency pre-LPSD, but this was not evident post-LPSD; (2) There was a significant increase in SOL to SW in responders; and (3) There was a significant phase delay in core body temperature among responders post-LPSD, that was not evident in non-responders.