Polysomnography in patients with post-traumatic stress disorder


Sinan Yetkin, MD, Department of Psychiatry, Diyarbakır Military Hospital, 21300 Diyarbakır, Turkey. Email: snnyetkin@gmail.com


Aims:  The purpose of the present study was to investigate sleep structure in post-traumatic stress disorder (PTSD) patients with and without any psychiatric comorbidities. The relationship between sleep variables and measurements of clinical symptom severity were also investigated.

Methods:  Sleep patterns of 24 non-medicated male PTSD patients and 16 age- and sex-matched normal controls were investigated on polysomnography on two consecutive nights. Six PTSD-only patients and 15 PTSD patients with major depressive disorder (MDD) were also compared to normal controls. Sleep variables were correlated with PTSD symptoms.

Results:  Compared to the normal controls, the PTSD patients with MDD had difficulty initiating sleep, poor sleep efficiency, decreased total sleep time, decreased slow wave sleep (SWS), and a reduced rapid eye movement (REM) sleep latency. The PTSD patients without any comorbid psychiatric disorders had moderately significant disturbances of sleep continuity, and decreased SWS, but no abnormalities of REM sleep. REM sleep latency was inversely proportional to the severity of startle response. SWS was found to be inversely correlated with the severity of psychogenic amnesia.

Conclusions:  PTSD patients have disturbance of sleep continuity, and SWS deficit, without the impact of comorbid depression on sleep. The relationship between SWS and the inability to recall an important aspect of trauma may indicate the role of sleep in the consolidation of traumatic memories. The relationship between the severity of the startle response and REM latency may suggest that REM sleep physiology shares common substrates with the symptoms of PTSD.

IMPAIRED SLEEP IS a common complaint among patients with post-traumatic stress disorder (PTSD). Sleep disturbances are included in the DSM-IV diagnostic criteria for PTSD.1 Growing evidence shows that sleep disturbances seem to be a core feature of PTSD, and sleep disruption following a traumatic event may constitute a specific mechanism involved in the pathophysiology of chronic PTSD.2 Polysomnography is just one of the objective methods used in order to better understand the pathophysiology of PTSD. To date, however, polysomnography has not produced consistent results in PTSD patients.

In PTSD patients the most consistent polysomnography findings are difficulty in sleep initiation and maintenance, including poor sleep efficiency,3–9 decreased total sleep time (TST),3,4 increased sleep latency,4,5 and increased number of awakenings.3–5 These findings, however, were not found in some other studies.10–15 Similarly, findings regarding slow wave sleep (SWS) abnormalities have been inconsistent. Some studies have reported decreased SWS.5,6,9,16 Again, numerous other studies did not find any differences in SWS.3,7,8,10,12,13,15 Furthermore, two studies reported increased SWS in PTSD patients.4,11

Rapid eye movement (REM) sleep parameters (amount of REM, latency to REM, and awakenings from REM) are also controversial. Some studies have reported shorter REM time during sleep,3–5 but others have found either normal or prolonged REM durations.6–16 REM latency has also varied between different studies. Some of them have found shortened REM latency,8,17 and some of them have found prolonged3–6 or normal REM latency in PTSD.9–16

Such discrepant polysomnography findings in PTSD could be due to comorbidity (particularly depression), medication, age, sample size, and the time since the traumatic event.16 An investigation of sleep in a homogeneous patient sample controlling for additional confounding factors would be of value. PTSD, however, is often associated with another psychiatric morbidity and becomes chronic. Hence, it is difficult to separate the effects of PTSD from other concomitant psychiatric disorders (particularly depression). In the literature there are only a few studies that have attempted to discriminate between the direct effect of PTSD and that of comorbidity.8,11

The aim of the present study was therefore to investigate sleep structure in adult male and drug-free PTSD patients with or without any psychiatric comorbidity. Another aim was to investigate the relationship between measurements of clinical symptom severity and sleep variables.



Twenty-four male patients diagnosed with PTSD according to DSM-IV criteria, with a mean age of 28.3 years (range, 21–43 years) were recruited from the inpatient Department of Psychiatry in Gülhane Military Medical Academy (GMMA), Ankara. The diagnosis of PTSD was made by clinicians who were trained and experienced in PTSD. All the subjects were medically evaluated and found to be free of any physical problem. The subjects also had no history of any primary sleep disorder. Eight of the 24 patients had never been treated with any psychotropic medication. Sixteen patients had been receiving a selective serotonin re-uptake inhibitor or benzodiazepines and were non-medicated for at least 2 weeks prior to the study. They had not had alcohol for at least 4 weeks prior to the study.

Seventeen of the patients had combat-related PTSD (70.8%), five had traffic accident-related PTSD (20.8%), and two had earthquake-related PTSD (8.3%). Thirteen patients who had combat-related PTSD had been exposed to more than two traumatic events. The Turkish version of the Structured Clinical Interview for the DSM-III-R (SCID) with satisfactory validity and internal consistency was used for all subjects to obtain a SCID-based diagnosis for axis I psychopathology.18,19 All the patients met the criteria for lifetime and current PTSD. Six of the patients had no other major psychiatric disorder (25%). Fifteen had a current history of major depressive disorder (MDD; 62.5%), eight had current alcohol abuse (33.3%), three had current generalized anxiety disorder (12.5%), and two had current and lifetime social phobia (8.3%).

Sixteen healthy men, with a mean age of 27.5 ± 5.9 years (range, 20–41 years), with no personal or family psychiatric disorders served as a normal control group. The control subjects were recruited from medical school students and hospital employees. None of them had any major medical illness or primary sleep disorders. The study was reviewed and approved by the local ethics committee of GMMA, and all participants signed a written informed consent form.

Clinical assessment

Severity of illness was assessed using the Turkish version of the Clinician-Administered PTSD Scale (CAPS) with satisfactory validity and internal consistency.20,21 The severity of depressive symptoms was assessed on the 21-item Hamilton Rating Scale for Depression (HRSD).22 The Turkish version of the HRDS, which has satisfactory validity and internal consistency, was used.23 The rating scales were administered by the patients' psychiatrist in their inpatient unit prior to the sleep study. Patient clinical characteristics are listed in Table 1.

Table 1.  Patient characteristics
 PTSD only (n = 6)PTSD + MDD (n = 15)Total PTSD group (n = 24)
Mean (range)Mean (range)Mean (range)
  1. CAPS, Clinician-Administered PTSD Scale; HRSD, 21-item Hamilton Rating Scale for Depression; MDD, major depressive disorder; PTSD, post-traumatic stress disorder.

Age (years)28.0 (21–40)28.8 (21–43)28.3 (21–43)
Duration of disease (months)32.8 (3–60)31.0 (3–84)37.2 (3–144)
No. traumatic events2.4 (1–4)2.4 (1–4)2.3 (1–4)
 Total score63.2 (57–68)80.1 (63–103)73.8 (57–103)
 Re-experiencing symptoms18.0 (15–21)20.6 (15–36)19.2 (12–36)
 Avoidance symptoms19.8 (16–25)30.6 (22–38)26.7 (16–38)
 Hyper-arousal symptoms25.4 (21–34)29.3 (23–43)28.0 (21–43)
HRSD8.0 (4–10)19.1 (16–25)15.1 (4–25)


All the subjects were studied on two consecutive nights in the sleep laboratory, and recordings were made in individual bedrooms using digital polygraph (Somnostar Alpha Series 4; SensorMedics, Yorba Linda, CA, USA) and the standard recording procedures. On the first night of the study a full-montage polysomnogram was recorded, consisting of an electroencephalogram (EEG), an electro-oculogram (EOG), a submental electromyogram (EMG), an electrocardiogram, respiratory monitoring, and an EMG of the anterior tibialis muscle. On the second night of the study, recordings included only EEG, EOG, and submental EMG.

The first night in the sleep laboratory was used for adaptation and exclusion of a primary sleep disorders including sleep apnea and periodic leg movement (PLM) disorder. According to the International Classification of Sleep Disorders criteria, respiratory index (RDI) and periodic leg movement index (PLMI) >5 are regarded as abnormal and were used as exclusion criteria. On the first night of the study, all subjects' RDI and PLMI were <5.

Recordings were scored visually by trained raters blinded to the diagnosis, using a 30-s epoch according to the criteria of Rechtschaffen and Kales.24 Polysomnography variables were divided into two groups for purposes of further analysis: sleep continuity and sleep architecture. The definition of sleep continuity variables was as follows: sleep efficiency was defined as the percentage of time spent asleep divided by the total recording period. The TST was defined as total number of minutes of sleep obtained during the recording period. Sleep period time (SPT) was defined as the time from the onset of sleep until the final awakening during the recording. Sleep latency was defined as the time from lights off to first occurrence of three consecutive epochs of stage 1 sleep, or an epoch of any other sleep stage. The number of awakenings was defined as the number of awakenings with at least one epoch of the wake stage during SPT. Sleep architecture variables were defined as the percentage of time spent in stages 1, 2, 3, 4 and REM. SWS was the sum of stage 3 and 4. For the REM sleep variable, REM latency was included, and defined as the time from the onset of sleep to the first epoch of REM.

Statistical analysis

Data from the second night were used for statistical analysis. Non-parametric analysis was used because of the small sample size and large standard deviations in many variables. The significance of differences among groups was assessed on Mann–Whitney U-test. The non-parametric Spearman's rho coefficients (two-tailed) were used to test the relationships among sleep variables and clinical measures. P < 0.05 was considered significant.


All the PTSD patients were compared with normal controls and then the patients were divided into two groups: PTSD-only patients, and patients with comorbid MDD. Both PTSD groups were compared to normal controls. Three PTSD patients who had other comorbid anxiety disorders were not included in the study in the second step of the comparison. Table 2 presents the means and standard deviations of sleep variables for the second night in 24 PTSD patients and controls. All the subjects were free from pathological sleep apnea and PLM. On the sleep continuity indexes, patients with PTSD had a significantly longer sleep onset latency (P < 0.001), a lower sleep efficiency (P < 0.001), and a decreased TST (P = 0.004) when compared with controls. With regard to sleep architecture, the patients had less SWS (P = 0.001) compared with controls. Eight of the 24 patients had no score-worthy stage 4 sleep. The REM sleep latency in the patients was found to be shorter than that in the normal controls (P = 0.002).

Table 2.  Sleep variables vs presence of PTSD
Sleep variablesPTSD group (n = 24)Control group (n = 16)zP
Mean ± SDMean ± SD
  1. PTSD, post-traumatic stress disorder; REM, rapid eye movement.

Age (years)28.3 ± 6.327.5 ± 5.9−0.38n.s.
Total sleep time (min)339.9 ± 78.7414.9 ± 54.5−2.840.004
Sleep period time (min)372.3 ± 83.1418.3 ± 55.0−1.69n.s.
Sleep efficiency (%)81.8 ± 10.295.1 ± 2.5−4.36<0.001
No. awakenings14.4 ± 8.311.0 ± 6.3−1.21n.s.
Sleep latency (min)38.3 ± 36.16.6 ± 4.4−3.94<0.001
Stage 1 (%)1.4 ± 1.02.5 ± 1.3−2.420.013
Stage 2 (%)62.3 ± 9.959.4 ± 7.1−1.20n.s.
Stage 3 (%)5.5 ± 3.95.9 ± 1.5−0.18n.s.
Stage 4 (%)5.6 ± 5.610.5 ± 4.8−3.190.001
Stage 3-4 (%)11.1 ± 7.616.0 ± 4.9−2.660.007
REM sleep (%)16.4 ± 5.517.4 ± 3.9−1.01n.s.
REM sleep duration (min)60.2 ± 20.973.1 ± 16.3−0.50n.s.
REM latency (min)77.8 ± 48.4104.8 ± 26.7−3.000.002

In the patients groups, six of the 24 PTSD patients had no comorbid psychiatric disorders, and 15 of the 24 PTSD patients had comorbid MDD. There were no statistically significant differences between these two groups in the sleep variables, duration of the disease, or the number of traumatic events (Table 1). Total CAPS score in the PTSD patients with MDD (P = 0.004) was significantly higher compared with PTSD-only patients. On CAPS subscale scores, the score of the avoidance symptom cluster in the PTSD patients with MDD (P = 0.001) was significantly higher compared with PTSD-only patients. When both PTSD groups were compared with normal controls, PTSD patients with MDD had more significant differences compared with normal controls on sleep variables than PTSD-only patients (Table 3). On the sleep continuity indexes, TST (P = 0.001), SPT (P = 0.024), and sleep efficiency (P < 0.001) were all found to be significantly decreased, and sleep latency (P < 0.001) was significantly increased. With regard to sleep architecture, PTSD-MDD patients had significantly less SWS (P = 0.001). In addition, they had more abundant REM sleep latency (P = 0.001). The PTSD-only group had significantly reduced sleep efficiency (P = 0.001), and increased sleep latency (P = 0.008) on sleep continuity indexes when compared to the normal controls. The percentage of SWS (P = 0.050) and the percentage of stage 4 sleep (P = 0.039) in the PTSD-only patients were significantly reduced, as found in patients with MDD. But there was no significant difference in the REM latency between the PTSD-only group and the normal control group.

Table 3.  Sleep variables vs PTSD type
Sleep variablesPTSD only (n = 6)PTSD + MDD (n = 15)Normal (n = 16)Pairwise comparison
Mean ± SDMean ± SDMean ± SD
  • *

    P < 0.05,

  • **

    P ≤ 0.001.

  • MDD, major depressive disorder; N, normal controls; PD, PTSD + MDD; PO, PTSD only; PTSD, post-traumatic stress disorder; REM, rapid eye movement.

Total sleep time (min)378.0 ± 82.7316.3 ± 78.9414.9 ± 54.5PD < N**
Sleep period time (min)414.2 ± 75.3348.2 ± 87.9418.3 ± 55.0PD < N*
Sleep efficiency (%)79.7 ± 13.681.4 ± 10.395.1 ± 2.5PO < N**, PD < N**
No. awakenings17.5 ± 10.214.1 ± 8.911.0 ± 6.3n.s.
Sleep latency (min)52.2 ± 47.637.3 ± 35.36.6 ± 4.4PO < N*, PD < N**
Stage 1 (%)1.5 ± 1.11.7 ± 1.12.5 ± 1.3n.s.
Stage 2 (%)66.2 ± 13.061.1 ± 9.959.4 ± 7.1n.s.
Stage 3 (%)4.7 ± 3.64.8 ± 3.85.9 ± 1.5n.s.
Stage 4 (%)3.0 ± 6.05.8 ± 6.210.5 ± 4.8PO < N*, PD < N**
Stage 3-4 (%)8.0 ± 8.610.6 ± 7.716.0 ± 4.9PO < N*, PD < N**
REM sleep (%)15.0 ± 3.717.5 ± 6.217.4 ± 3.9n.s.
REM sleep duration (min)62.0 ± 16.760.1 ± 24.373.1 ± 16.3n.s.
REM latency (min)87.5 ± 39.275.3 ± 57.2104.8 ± 26.7PD < N**

Within the PTSD group, correlations between some symptoms scores of interest and primary sleep variables are listed in Table 4. The startle response score in CAPS was negatively correlated with the TST (r = −0.535, P = 0.007), and REM latency (r = −0.465, P = 0.022). In addition, the startle response score also positively correlated with the number of traumatic events (r = 0.565, P = 0.004), CAPS total score (r = 0.772, P < 0.001), and the HRSD total score (r = 0.565, P = 0.042). The severity of psychogenic amnesia was negatively correlated with the percentage of SWS (r = −0.528, P = 0.008) and sleep efficiency (r = −0.420, P = 0.41), and positively correlated with sleep latency (r = 0.432, P = 0.035). We did not find any significant correlation between the other symptoms of interest (flashback, nightmare, irritability, and difficulty of concentration) and primary sleep variables. None of the sleep variables was correlated with the duration of the disease.

Table 4.  Spearman correlation coefficients for clinical symptoms and sleep variables
CAPS symptomsSleep variables
SE (%) r (p)TST r (p)SL r (p)No. A r (p)% SWS r (p)% REMS r (p)REML r (p)
  • *

    P < 0.05.

  • CAPS, Clinical-Administered PTSD Scale; No. A, no. awakenings; REML, rapid eye movement latency; REMS, rapid eye movement sleep; SE, sleep efficiency; SL, sleep latency; SWS, slow wave sleep; TST, total sleep time.

Flashback (Criterion B-3)−0.083 (0.700)−0.260 (0.219)−0.089 (0.680)0.128 (0.552)−0.257 (0.225)0.142 (0.507)−0.380 (0.067)
Nightmare (Criterion B-4)−0.058 (0.786)−0.164 (0.445)−0.139 (0.518)0.231 (0.278)−0.290 (0.169)−0.142 (0.498)−0.161 (0.451)
Irritability/anger (Criterion D-13)0.218 (0.306)−0.140 (0.514)0.023 (0.915)0.029 (0.891)−0.019 (0.929)0.193 (0.367)−0.378 (0.068)
Difficulty of concentration (Criterion D-14)−0.312 (0.138)−0.185 (0.387)0.141 (0.510)0.207 (0.332)−0.153 (0.477)0.237 (0.266)−0.192 (0.368)
Startle response (Criterion D-16)−0.204 (0.339)−0.535* (0.007)0.009 (0.968)−0.009 (0.967)−0.019 (0.930)0.265 (0.211)−0.465* (0.022)
Psychogenic amnesia (Criterion B-7)−0.420* (0.041)−0.184 (0.390)0.432* (0.035)0.263 (0.214)−0.528* (0.008)0.073 (0.734)0.163 (0.447)


The present data show that patients with PTSD have the characteristic abnormalities of sleep structure seen in many studies. As a group, all patients had profound difficulties in sleep initiation and maintenance, reduced SWS, and reduced REM latency. When PTSD-only patients and PTSD patients with MDD were compared with the normal controls, sleep disturbances were found to be more abundant in PTSD/MDD. But we did not observe any significant differences in sleep structure between PTSD-only patients and PTSD patients with MDD.

In the present study we observed a higher degree of disturbed sleep continuity in PTSD patients with comorbid MDD than in PTSD-only patients. Although PTSD-only patients had only significantly reduced sleep efficiency and increased sleep latency, PTSD patients with comorbid MDD had significantly reduced TST, increased sleep latency, and reduced sleep efficiency. Disturbance in sleep continuity has been reported in many studies of drug-free patients with PTSD.3–9 The present data have confirmed these findings. These findings in sleep continuity are also frequently found in various other psychiatric disorders;25 hence, they are not specific to any one disorder. As proposed by some authors, the disruption in sleep continuity may be related to psychological distress and its physiological correlations, inducing a non-disengagement of the waking process.26 Therefore, in the present study, PTSD patients with comorbid MDD, having high scores in the clinical scales, had more prominent sleep disturbances in sleep continuity (Table 3). Contrary to these findings, some sleep studies in PTSD, particularly those undertaken recently, could not demonstrate a significant disturbance in sleep continuity, or in sleep architecture.10–14 This inconsistency may be associated with the acuity of the trauma. In the present study the time passed since the traumatic event was relatively shorter than that in recent studies. As indicated in some studies, biological adaptive mechanisms may contribute to the normalization in sleep structure by the time elapsed since the trauma.12,13

With regard to sleep structure in the present study, as a group, PTSD patients had a significant reduction in SWS. Reduced SWS, particularly decreased stage 4 sleep, was statistically significant in both the PTSD patients with comorbid MDD and the PTSD-only patients. This finding is consistent with the outcome in some previous studies,5,6,9,16 and the results of a meta-analysis of polysomnography sleep abnormalities in PTSD.27 The present findings have demonstrated that PTSD patients without the impact of depression had decreased SWS. Decreased SWS has also been observed in many psychiatric disorders, such as depression and schizophrenia.25 It therefore seems to be a non-specific finding, as mentioned in sleep continuity. Sleep studies in stressful life events, however, have demonstrated that stress is strongly associated with decreased SWS.28–30 Furthermore, some studies have shown that at 1-year follow-up, subjects with ongoing stress had less SWS, as well as reduced REM latency, which occurred in both the non-depressed and depressed subjects.28 These findings demonstrated that stressful life events affect SWS in a consistent manner.30 These findings suggest that psychological distress and arousal cause a reduction in SWS. Studies of centrally measured hyperarousal during sleep also support a relationship of non-diminished central arousal activity at night and sleep disturbances in chronic PTSD patients.15,31 Additionally, Germain et al. proposed a neurobiological model of PTSD based on neuroimaging studies showing that hyperactivity of the amygdala and attenuated activity of the medial prefrontal cortex contribute to the heightened whole-brain neuronal activity during non-REM (NREM) sleep and that these changes may cause an increased activity in arousal-promoting brain centers, and a reduced activity in sleep-promoting centers.2 Consequently, the resulting pattern of persistent arousal could contribute to NREM sleep anomalies such as reduced SWS.2

Many researchers have been interested in the REM sleep of patients with PTSD due to the repetitive nightmares in PTSD. They postulated that PTSD is an REM sleep disorder.32,33 But we observed no differences between PTSD patients and normal controls in the REM sleep time or its percentage. In the literature, sleep studies have produced mixed results with regard to the amount of REM sleep. In most of these studies the discrepancies were attributed to the lack of control for comorbidity.27,34 In the present study, PTSD-only patients and PTSD patients with comorbid MDD were not significantly different in the amount of REM sleep when compared to normal controls.

In sleep studies performed on PTSD patients, the latency of REM sleep is controversial because of the coexisting depression among PTSD patients.8–17 In the present study we found significant differences only in REM latency between PTSD patients with comorbid MDD and normal controls. PTSD-only patients, however, showed no significant difference in REM latency when compared with normal controls. In PTSD patients with comorbid MDD, two of the 15 patients exhibited sleep onset REM (SOREM), and had REM latencies <15 min. None of six PTSD-only patients displayed SOREM. In the present study the impact of comorbid depression on sleep in PTSD was observed in PTSD patients with comorbid MDD and reduced REM latency.

When we investigated the relationship between the scores of some particular clinical symptoms and primary sleep variables, we observed no correlation between the sleep variables and the severity of flashback, nightmares, irritability, and difficulty of concentration (Table 4). The noteworthy observation, however, was that the severity of the startle response was negatively correlated with REM latency. In the present sample group most of the patients had combat-related PTSD symptoms, and most of them had experienced more than one traumatic event. Within the PTSD patients group we observed that the startle response scores correlated positively with the number of traumatic events. This finding is consistent with earlier reports of increased startle response due to prolonged exposure to stress (long-term sensitization).35 The startle reflex is a basic mammalian reflex, and its neuronal pathways are located in the brainstem and receive input from the central nucleus of amygdala and the bed nucleus of the stria terminalis. REM sleep is also modulated by brainstem structures. Lesions of the pedunculopontine tegmental nucleus in animals have been shown to result in dysinhibition of the acoustic startle, as well as activation of REM sleep mediated by cholinergic neurons.36 Some researchers found that tonic and phasic REM sleep measurements in PTSD patients were significantly elevated, which is evidence of dysregulation of the REM sleep control system.32 In addition, investigators have also suggested that the exaggerated abnormal startle response seen in PTSD patients during the awake period may be related to REM sleep dysfunction.33,35,37 Ross et al. proposed that dysfunctional REM sleep cycle may participate in the exaggerated startle response described in PTSD.32 The present finding of a significant inverse correlation between startle response scores and REM latency may also indicate a dysfunction of the REM sleep control system, as seen in PTSD patients with increased REM sleep phasic activity in some studies. Reduced REM latency may result from a long-lasting alteration in the brainstem systems controlling the REM sleep and the startle response in the aftermath of an overwhelming psychological stressor. Furthermore, the startle response scores also correlated positively with the CAPS total score and the HRSD total score. As the severity of the disease increases, disturbances in REM sleep regulation may become more prominent in PTSD.

Another of the present findings in the correlation between primary sleep variables and clinical symptoms scores was that the severity of psychogenic amnesia was negatively correlated with the percentage of SWS and sleep efficiency, and positively correlated with sleep latency. Psychogenic amnesia in CAPS refers to the inability to recall an important aspect of the trauma. Neuroimaging findings in PTSD have demonstrated reduced functional activity and reduced volume of the ventromedial prefrontal cortex, as well as an increased neural activity of amygdala during fear conditions.38–40 The prefrontal cortex seems to play an important role in the generation of synchronized SWS activity.41 Investigators have also suggested that SWS may play a critical role in memory consolidation.42 In the present study SWS deficit was seen in both the PTSD patients with comorbid depression and PTSD-only patients. This SWS deficit may have resulted from the reduced prefrontal cortex activity, as shown in neuroimaging studies. Thus, reduced prefrontal activity may contribute to impairment in the consolidation of traumatic memories, and hence, PTSD patients may complain about the inability to recall an important aspect of trauma.

There were several limitations in the present study. First, the sample size of the PTSD-only group was small. Because of the high rates of psychiatric comorbidities in PTSD, the present sample consisted of only six patients. In this group, however, the PTSD-only patients did not have any current or life-long comorbid psychiatric disorders, including depression. In a previous study only current MDD was excluded.11 For this reason, despite the small sample size, the findings in the PTSD-only group may be meaningful. The second limitation was the difference in the nature of the traumatic event in the sample group. The effects of the difference in the traumatic events on the present results cannot be excluded.

In summary, the present results in 24 drug-free inpatients with PTSD demonstrated sleep patterns that were characterized by markedly increased sleep latency, decreased TST, poor sleep efficiency, decreased SWS, and short REM latency. These changes were more prominent in patients with comorbid MDD. But PTSD-only patients exhibited only poor sleep efficiency, increased sleep latency, and decreased SWS. These results demonstrated that PTSD patients with comorbid depression have more sleep disturbances. The results also demonstrated that the severity of the startle response is negatively correlated with REM latency; and the severity of psychogenic amnesia, which refers to the inability to recall an important aspect of the trauma, is negatively correlated with the percentage of SWS.

Finally, the relationship between physiological variables and clinical symptoms in PTSD appears to be complex, but the present results suggest that this relationship may be meaningful. Thus, further studies are needed to investigate the relationship between sleep changes induced by traumatic stress and clinical symptoms in PTSD on electrophysiology neuroimaging.