Polysomnographic studies of patients with PTSD have produced variable findings [7]. Based on meta-analysis, Kobayahi et al. report an association between PTSD and increased rapid eye movement (REM) density (a measure of rapid eye movement frequency during REM sleep), decreased stage N3 (“deep”) sleep, and increased stage N1 (light) sleep compared with healthy comparison subjects [7]. These nonspecific findings are not uniformly documented in all studies. Consequently routine polysomnography is not indicated in the clinical assessment of PTSD unless other sleep disorders such as obstructive sleep apnea are suspected. Indeed, many individuals with PTSD find that they sleep more comfortably in a sleep laboratory, which is perceived as safe and “guarded” by the technologists [8]. Abnormal subjective sleep as perceived by patients with PTSD may also vary greatly over time, so that one or a few nights of data collection may be insufficient. That suggests that in-home monitoring with actigraphy may be helpful in documenting objective sleep disturbance in these patients.

Dagan et al. [3] studied actigraphic findings during sleep at home in 16 men with DSM III-R-defined PTSD and 11 male non-PTSD controls. The PTSD group included randomly selected Lebanon war veterans . The control group included veterans of that conflict without diagnosed PTSD. Interestingly, PTSD patients did not have poorer actigraphic sleep (sleep time) than controls, despite reporting poorer subjective sleep. The authors suggested that PTSD patients might fail to correctly estimate their sleep, such as “sleep state misperception” or “paradoxical insomnia,” in the newer nosology of International Classification of Sleep Disorders (ICSD II and III).

Westermeyer and colleagues [9] compared actigraphic and subjective sleep diary data for 241 nights among 21 veterans with lifetime PTSD. Actigraphic sleep minutes per night were, on average, 51 min longer than self-reported sleep minutes on sleep logs. Total intra-class correlation between actigraphy and sleep logs was 0.588. In the same study, correlation between self-reported awakenings on the sleep diary and apparent actigraphic awakenings was conducted after excluding nights when there was failure to indicate sleep duration in the diaries. There was a major difference with the number of awakenings reported on sleep diary never exceeding the number of apparent awakenings on actigraphy. For 241 nights of monitoring, actigraphy indicated 3.6 times more awakening episodes than did sleep diary information. This odds ratio differed greatly from one person to another, ranging from 2 to 12 times, leading the authors to suggest that self-reported awakenings are not reliable for scientific studies of individuals with PTSD. It is possible that patients moved sufficiently in their sleep to produce an appearance of an “awakening response ” in the actigraphy. This study was limited by the inclusion of veterans receiving care for lifetime as opposed to current PTSD with sleep symptoms and cannot inform a concern about all patients with PTSD and insomnia. Interestingly, in contrast, Woodward et al. [10] have reported that veterans with PTSD tend to move less during sleep than normal controls.

In another paper, Westermeyer et al. [11] examined the same cohort to assess sleep symptoms in patients with lifetime PTSD. In this report, Pittsburgh Sleep Quality Index (PSQI), Beck Depression Inventory (BDI), Clinician-Administered PTSD Scale (CAPS), and PTSD Checklist were used. Posttraumatic avoidance, hypervigilance, and depressive symptoms were all associated with poor sleep quality on PSQI. Trends of borderline statistical significance were noted between worse sleep quality and more severe clinician-rated PTSD, more self-rated awakenings, and greater actigraphically determined sleep duration. Using linear regression, only PTSD hypervigilance symptoms were associated with impaired sleep quality. PSQI sleep disturbance showed no correlation with actigraphic awakening or with sleep log awakening. Shorter sleep duration on actigraphy revealed a borderline association with poor sleep quality on PSQI at p = 0.08. Awakenings per night were not associated with mean PSQI scores. Neither sleep duration nor awakenings per night were associated with the mean Epworth Sleepiness Scale score in this population.