Posttraumatic stress disorder (PTSD) is associated with suicidal ideation (SI) and obstructive sleep apnea (OSA). There are no studies of OSA diagnosed by sleep study and SI in patients with PTSD.
Forty consenting civilians with PTSD (38 female, mean ± standard deviation age: 44.60 ± 12.73) underwent a Level 3 home sleep apnea test (WatchPAT200; Itamar Medical, Israel). OSA severity was measured with the respiratory disturbance index (RDI) (number of apneas, hypopneas and respiratory effort related arousals per hour). SI was measured with Items 9, 35, 39, and 50 of the Brief Symptom Inventory (BSI). Other patient-rated measures included the Beck Depression Inventory, second edition (BDI-II), PTSD Checklist for DSM-5 (PCL-5), and the Pittsburgh Sleep Quality Index PTSD Addendum modified to include only Items 1c, 1e, 1f, and 1g that address nightmares.
The RDI (r = .757, P < .001) and oxygen desaturation index (r = .633, P < .001) were directly correlated to SI. Multiple regression analysis using SI as the dependent variable and patient-rated measures as independent variables revealed only RDI (β = .480, t = 4.167, P < .001) and BDI-II (β = .469, t = 3.375, P = .002) as predictors of SI, with adjusted R2 = 0.718. In patients with RDI < 30 events/h (n = 37) correlation of SI with RDI (r = .511, P = .001) but not ODI (r = .312, P = .060) remained significant. Multiple regression analysis (when RDI < 30 events/h) revealed only BDI-II (β = .603, t = 3.492, P = .002), and not RDI (β = .247, t = 1.723, P = .096) as a significant predictor of SI.
OSA severity in PTSD was directly related to SI. Depression was a significant mediator in the relationship between RDI and SI, with OSA-related intermittent hypoxemia possibly contributing to this relationship only in severe OSA.
Gupta MA, Jarosz P. Obstructive sleep apnea severity is directly related to suicidal ideation in posttraumatic stress disorder. J Clin Sleep Med. 2018;14(3):427–435.
Current Knowledge/Study Rationale: Posttraumatic stress disorder (PTSD) is associated with suicidal ideation (SI) and a high prevalence of OSA. There are no studies that have examined the relationship of OSA diagnosed by sleep study and SI in patients with PTSD.
Study Impact: This study of 40 civilian patients with PTSD and OSA diagnosed by a home sleep apnea test presents the previously unreported finding of a direct relationship between OSA severity (measured by the respiratory disturbance index) and SI. Depression was a significant mediator in the relationship between OSA and SI across the entire range of OSA severity, with OSA-related intermittent hypoxemia possibly further contributing to the association between OSA and SI only when patients with PTSD and severe OSA were considered.
Patients with trauma exposure and posttraumatic stress disorder (PTSD) may present with elevated suicide risk.1,2 PTSD is often comorbid with major depressive disorder,2 which further increases the risk of suicide in patients with PTSD.1 The factors underlying suicidal behavior in PTSD remain poorly understood.1 There are limited data that specifically address suicide risk in the context of evidence-based treatments for PTSD, and studies using evidence-based therapies such as prolonged exposure report a significant reduction in suicidal ideation among only approximately 50% of patients.3
In addition to the core symptoms of nightmares and insomnia,2 there is also an increased prevalence of obstructive sleep apnea (OSA) in PTSD.4 There is a paucity of studies of OSA and suicidal ideation (SI) in general5,6 and involving patients with PTSD.7 Systematic reviews of the literature on the relation between sleep disturbances and suicidal behaviors8–10 in general have revealed mixed results, with insomnia and nightmares emerging as independent risk factors for suicidal behaviors in some studies, after controlling for psychiatric comorbidity such as depression; there were no studies of OSA diagnosed using sleep studies in these reviews.8–10 A 10-year longitudinal case-control study11 from a population-based community sample of older adults indicated that complaints of difficulty falling asleep and nonrestorative sleep at baseline were associated with an elevated risk for suicide at 10-year follow-up, after controlling for depressive symptoms (odds ratio [OR] = 1.30, 95% confidence interval [CI] 1.04–1.63, P < .05). This study11 did not measure OSA.
Two recent epidemiologic studies5,6 have examined the relationship between sleep apnea and suicidal behavior. In a study from the 2014 National Survey of Drug Use and Health data (n = 41,086), there was a 2.9% prevalence of sleep apnea (sleep apnea cases identified based on response to a single item where the participants were asked whether their doctor had informed them that they had a diagnosis of sleep apnea over the past 12 months); sleep apnea was associated with both suicidal ideation (OR = 1.56, 95% CI 1.24–1.97) and suicide planning (OR = 1.59, 95% CI 1.11–2.28), but not suicide attempts (OR = 1.19, 95% CI 0.65–2.18), after controlling for demographic factors, substance use, and depression.5 In a 10-year prospective study of 74,543 cases of sleep apnea (diagnosed by history and in-laboratory polysomnography or in-home overnight respiratory polygraphy) from the Swedish Patient Register, inpatient (generally representing more severe and complex disease) but not outpatient sleep apnea was significantly associated with death from suicide in both men (hazard ratio [HR] = 1.76, 95% CI 1.19–2.60) and women (HR = 4.33, 95% CI 1.96–9.56) after controlling for comorbidities including cardiovascular disease, cancer, respiratory disorders, and diabetes.6 Psychiatric disorders such as depressive disease were not controlled for in this study.6
To our knowledge, only one reported study has specifically examined the relationship of sleep disturbance and sleep-disordered breathing (SDB) and suicide risk in patients with PTSD7; a brief summary follows. The study included 153 female sexual assault survivors with PTSD; potential SDB and sleep movement disorders were present in 80% of the participants (SDB and sleep movement disorders were diagnosed by clinical practice parameters and research algorithms and not polysomnography in this initial study). In comparison with the patients with PTSD and no sleep disorders, there was significantly greater depression (Cohen d = 0.73–0.96, P < .01) and greater suicidality (Cohen d = 0.57–0.78, P < .05) in the sleep disorders group (which included both sleep movement disorders and SDB). In the subgroup with SDB only, there was a medium effect size for depression (Cohen d = 0.73, P = .01) and suicidality (Cohen d = 0.57, P = .04). This study did not control for the confounding effect of depression when examining the relationship between SDB and suicidality.7
In the current study we have examined the relationship between the severity of OSA measured with a home sleep apnea test12 and suicidal ideation in patients with PTSD, after controlling for comorbid depression and other factors that are known to contribute to suicidal ideation in PTSD. To our knowledge, there are no previously reported studies that have examined the relation between sleep apnea severity diagnosed with sleep studies and suicidal ideation in PTSD. A better understanding of the relationship between sleep apnea and suicidal ideation in PTSD may provide further insight into factors underlying suicide risk in PTSD and potentially identify specific clinical targets for the treatment of suicidal behavior in patients with PTSD.
The study was approved by the Office of Research Ethics, University of Western Ontario, London, Ontario, Canada. Forty consecutive consenting civilians with PTSD secondary to sexual trauma (all from the first author's psychiatric practice) (Table 1) completed a battery of psychiatric and sleep rating scales and underwent one or more nights of unattended in-home testing using the WatchPAT 200 (Itamar, Cesarea, Israel)12,13 for the assessment of OSA and other sleep physiological measures, as part of a larger study of sleep-related psychosomatic factors in PTSD. All patients met the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5)2 criteria for moderate to severe PTSD with Clinician Administered PTSD Scale for DSM-5 (CAPS-5)14 score of 55 or higher at the time of initial consultation. The duration of PTSD symptoms in all patients was more than 12 months. At the time of recruitment for the WatchPAT 200 study (patients recruited from 0 to 24 months after their initial consultation for treatment of PTSD) patients were in different stages of recovery from their PTSD symptoms. Exclusion criteria were as follows: (1) use of opioid analgesics or benzodiazepines; otherwise patients remained on all their regular medications (no patients were on medications whose primary therapeutic effect was related to alpha adrenergic blockade such as phentolamine; however, patients remained on antidepressants and prazosin for nightmares, which can have varying degrees of peripheral alpha adrenergic receptor– blocking effects); (2) history of excessive alcohol use or use of recreational drugs including regular marijuana use; and (3) an irregular sleep-wake schedule, or rotating shiftwork. None of the patients had medical disorders that could interfere with peripheral arterial tonometry (such as peripheral vascular disease, peripheral neuropathy, Raynaud disease, untreated cardiac arrhythmia, severe congestive heart failure, severe chronic obstructive pulmonary disease, and poorly controlled diabetes).
Demographic and clinical characteristics of study patients with PTSD (n = 40).
Demographic and clinical characteristics of study patients with PTSD (n = 40).
Home Sleep Apnea Test
Patients used the WatchPAT 200 device, which records four channels, and is used unattended in the patient's home. The device is categorized as Level 3 by the American Academy of Sleep Medicine.15 It uses peripheral arterial tonometry (PAT), pulse oximetry, heart rate, and actigraphy to detect OSA,16 arousals from sleep,17 and sleep stages18,19 using the proprietary zzzPAT software (Itamar, Cesarea, Israel).12 The PAT technology is a digital plethysmographic measurement method that records the pulsatile volume changes in the fingertip arteries that reflect the relative state of the arterial vasomotor activity and therefore indirectly the level of sympathetic activation.12,13 Peripheral arterial vasoconstriction, which mirrors sympathetic activation, is shown as an attenuation in the amplitude of the PAT signal.12 The WatchPAT 200 identifies the sleep/wake state based on a unique algorithm of the actigraphic data and distinguishes between rapid eye movement (REM) and non-REM sleep based on different autonomic functions.13 The WatchPAT 200 indirectly detects apnea and hypopnea events by identifying surges of sympathetic activation associated with the termination of these events. This information is further combined with heart rate and pulse oximetry data that are analyzed by the automatic algorithm of the system.13,16 Arousals from sleep are associated with an increased sympathetic activation, and peripheral vasoconstriction.13 Using a special automatic analysis algorithm, digital vasoconstrictions as measured by PAT, combined with an increase in pulse rate, have been shown to accurately reflect arousals from sleep.13,17
A systematic review and meta-analysis of studies assessing correlation of sleep indices between PAT devices and Level 1 in-laboratory polysomnography (PSG) in adults revealed a significant correlation between PAT devices and PSG for apnea-hypopnea index (AHI) (r = .893, 95% CI 0.857–0.920, P < .001), respiratory disturbance index (RDI) (r = .879, 95% CI 0.849–0.904, P < .001), oxygen desaturation index (ODI) (r = .942, 95% CI 0.894–0.969, P < .001),16 and arousals from sleep (r = .87, P < .001).17 The sleep staging algorithm of the PAT recording device has been validated,18 with overall agreement with PSG in detecting “light sleep” (PSG-scored stage N1 and N2 sleep) and “deep sleep” (PSG-scored stage N3 sleep) of 88.6% ± 5.9%, and overall agreement with PSG in detecting stage R sleep (REM sleep) of 88.7% ± 5.5%.18
Measures of OSA
All patients underwent 1 or more nights of sleep recordings using the WatchPAT 200; data from the first night of recording of more than 240 minutes sleep duration was used for this study. The following indices of sleep-related breathing disorder were obtained:
The AHI which is an index of the number of apnea and hypopnea events per hour of sleep.
The RDI which is an index of the number of apnea, hypopnea and respiratory effort related arousals (RERA) per hour of sleep, and is the recommended measurement of OSA severity for the WatchPAT200.20 The RDI was therefore used as the measure of OSA in this study. The RDI score was further divided into clinical severity categories21 of no OSA (RDI < 5 events/h), mild OSA (RDI 5 to < 15 events/h), moderate OSA (RDI 15 to < 30 events/h) and severe OSA (RDI ≥ 30 events/h).
The ODI, which measures the number of times the blood oxygen level drops ≥ 4% during sleep.20
The WatchPAT 200 defines an apnea as cessation of breathing during sleep of a duration of 10 seconds or more; a hypopnea as a reduction in breathing lasting 10 seconds or more followed by oxygen desaturation of 3% or more and/or an arousal; and a RERA as an obstructive event that does not meet the criteria for an apnea or hypopnea but causes arousal from sleep.20
Data from the following self-report questionnaires, which were generally completed within 1 week of completion of the sleep study, were used.
Pittsburgh Sleep Quality Index (PSQI): The PSQI is designed to evaluate overall sleep quality and consists of 19 items that address 7 categories: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medication, and daytime dysfunction. A global score is derived by adding the 7 categories; a global PSQI score higher than 5 has been shown to yield a diagnostic sensitivity of 89.6% and specificity of 86.5% in distinguishing good and poor sleepers.22 The global PSQI score was used in our analysis.
Pittsburgh Sleep Quality Index Addendum for PTSD (PSQI-A): The PSQI-A consists of seven items that focus on the frequency of seven disruptive nocturnal symptoms including hot flashes (Item 1a), general nervousness (Item 1b), memories or nightmares of traumatic experience (Item 1c), severe anxiety or panic not related to traumatic memories (Item 1d), bad dreams not related to traumatic memories (Item 1e), episodes of terror or screaming during sleep without fully awakening (Item 1f) and episodes of “acting out” dreams such as kicking, punching, running, or screaming (Item 1g). A global PSQI-A score (range 0–21) of 4 has been shown to yield a sensitivity of 94%, a specificity of 82%, and a positive predictive value of 93% for discriminating participants with versus without PTSD.23 A modified version of the PSQI-A (PSQI-A_ NM) consisting of only the nightmare-related items (Items 1c, 1e, 1f, and 1g) (range 0–12) was considered to be an index of PTSD-associated nightmares in this study.
Insomnia Severity Index (ISI): A seven-item instrument that assesses the nature, severity, and effect of sleep difficulties. The ISI measures the following dimensions of insomnia: difficulty with falling asleep, staying asleep, and early morning awakening; satisfaction with current sleep pattern; whether sleep problems are noticed by others; distress caused by sleep difficulties; and interference of sleep difficulty with daytime functioning. ISI scores (range 0–28) can be classified as absence of insomnia (0–7), subthreshold insomnia (8–14), moderate insomnia (15–21), and severe insomnia (22–28).24
Epworth Sleepiness Scale (ESS): An eight-item instrument that provides the participant's general level of daytime sleepiness. The participants rate the chances that they would doze off or fall asleep in eight different situations that are typically encountered in daily life. The ESS score is reported to significantly distinguish nonclinical subjects from individuals with OSA, idiopathic hypersomnia, and narcolepsy. A cutoff score of 10 is used to identify clinically significant daytime somnolence.25
Assessment of SI: The items used to assess SI were extracted from the Brief Symptom Inventory (BSI)26 which is a 53-item inventory that measures 9 major psychiatric symptom dimensions including depression, interpersonal sensitivity, anxiety, phobic anxiety, obsessive compulsiveness, paranoid ideation, somatization, hostility, and psychoticism. Patients are asked to rate how much they were bothered by each of the 53 items of the BSI over the past week by using a 5-point scale with the following rating options: not at all (rating of 0), a little bit (rating of 1), moderately (rating of 2), quite a bit (rating of 3), and severely (rating of 4). The SI scale that was derived from the following 4 items of the BSI has been previously used27 and was created by deriving a mean score (possible range 0 to 4) from responses to the following 4 items of the BSI: Item 9: thoughts of ending your life; Item 35: feeling hopeless about the future; Item 39: thoughts of death or dying; and Item 50: feelings of worthlessness.
PTSD Checklist for DSM-5 (PCL-5): The PCL-5 is a 20-item questionnaire corresponding to the DSM-52 symptom criteria for PTSD. A total severity score (range 0–80) is obtained by summing the scores from each of the 20 items. A PCL-5 cutoff score of 33 may be used to screen for PTSD.28
Beck Depression Inventory – Second Edition (BDI-II): is a 21-item self-rated instrument that is widely used to assess the severity of depression in psychiatrically diagnosed adolescents and adults.29 The BDI-II measures severity of depression corresponding to the assessment of symptoms for major depressive disorder listed in DSM-IV.30 Item 9 of the BDI-II addresses suicidal ideation and was included in the final BDIII score. The following ranges of scores (overall range 0–63) are used to classify depression severity: score of 0–13 denotes minimal, score of 14–19 denotes mild, score of 20–28 denotes moderate and a score of 29–63 denotes severe depression.29
All statistical analyses were carried out using SPSS version 24.31 Scatterplots (Figure 1 and Figure 2) were obtained to examine the overall relationships between RDI and SI scores. Analysis of variance (ANOVA) was used to examine the differences in mean SI scores between the different clinical21 RDI severity categories of no OSA (RDI < 5 events/h), mild OSA (RDI 5 to < 15 events/h), moderate OSA (RDI 15 to < 30 events/h) and severe OSA (RDI ≥ 30 events/h) (Figure 3). Pearson product-moment correlations were carried out to examine the relation between SI and (1) clinical factors including indices of OSA severity (AHI, RDI, ODI) and (2) patient-rated sleep and psychopathologic measures (Table 2). Multiple regression analysis was carried out using SI as the dependent variable and the RDI and other known confounding factors in the relationship between OSA and SI as independent variables using (1) the entire sample (n = 40) (Table 3) and (2) the subset of patients with RDI < 30 events/h (n = 37) (Table 4), who represented the subgroup with mild to moderate OSA. For this study only variables that were significantly correlated with SI (ie, PCL-5,28 depression [BDI II]29 and PTSD-associated nightmares [PSQI-A_NM]23) were used as independent variables in the multiple regression models (Table 2, Table 3, and Table 4). For this study the modified PTSD-A_NM (which only included the items related to nightmares and disturbing dreams) instead of the PTSD-A23 was used in all the regression analyses. Finally, because depression is known to be associated with both SI and OSA,4 the possible mediating role of depression in the relationship between OSA and SI was further assessed in (1) the entire sample and (2) the subset excluding severe OSA representing patients with RDI < 30 events/h, using the analysis proposed by Baron and Kenny.32
Correlation of SI scores with clinical and patient-rated measures.
Correlation of SI scores with clinical and patient-rated measures.
Results of multiple regression analysis (involving entire sample, n = 40) with SI scores as dependent variable.*
Results of multiple regression analysis involving entire sample...
Results of multiple regression analysis (in subset of patients with RDI < 30 events/h, n = 37) with SI scores as dependent variable.*
Results of multiple regression analysis in subset of patients...
The demographic characteristics of the patients with PTSD (n = 40) in the study are summarized in Table 1. The median age of (1) the entire sample and (2) females only (n = 38) was 43 years (Table 1); the menopausal status of the patients was not specifically considered. The relationship between the RDI and SI was essentially unchanged when the females alone (n = 38) were considered. Figure 1 is a scatterplot of the RDI severity versus SI (Pearson r = .757, P < .001; among females alone Pearson r = .718, P < .001). Figure 2 is a scatterplot of RDI severity versus SI in the subset of patients without severe OSA (RDI < 30 events/h) (Pearson r = .511, P = .001; among females alone Pearson r = .516, P = .001). The frequency of RDI severity (as per the standard clinical criteria for categorizing severity21 of OSA) in the study sample was as follows: 15% (n = 6) had no OSA (RDI < 5 events/h); 60% (n = 24) had mild OSA (RDI 5 to < 15 events/h); 17.5% (n = 7) had moderate OSA (RDI 15 to < 30 events/h); and 7.5% (n = 3) had severe OSA (RDI ≥ 30 events/h). Therefore 85% of our patients with PTSD had some degree of OSA (ie, RDI ≥ 5 events/h), with most (60%) having mild OSA. Figure 3 plots the clinical RDI severity categories and the mean SI scores for each RDI severity category, and indicates a direct and significant relationship (P < .001 by ANOVA) between the clinical categories of OSA severity and SI.
Scatterplot of RDI versus SI scores in patients with PTSD (n = 40).
PTSD = posttraumatic stress disorder, RDI = respiratory disturbance index, SI = suicidal ideation.
Scatterplot of RDI versus SI scores in patients with PTSD (n = 40).
Scatterplot of RDI versus SI scores among the subset of patients with PTSD and RDI < 30 events/h (n = 37).
PTSD = posttraumatic stress disorder, RDI = respiratory disturbance index, SI = suicidal ideation.
Scatterplot of RDI versus SI scores among the subset of patients with PTSD and RDI...
Means plot of RDI severity categories versus mean SI scores.
RDI severity: 1 = < 5 events/h, 2 = 5 to < 15 events/h, 3 = 15 to < 30 events/h, 4 = ≥ 30 events/h. ANOVA = analysis of variance, RDI = respiratory disturbance index, SI = suicidal ideation.
Means plot of RDI severity categories versus mean SI scores.
Table 2 summarizes the correlations between SI and (1) demographic factors, (2) indices of OSA, (3) and patient self-ratings of sleep and psychopathological factors that have been associated with increased suicide risk in PTSD such as depression2 (measured with the BDI-II29), a PTSD diagnosis33 (PTSD severity measured with PCL-528), severity of insomnia8–10 (measured with the ISI24), and nightmares8–10 (measured with PSQI-A_NM23), in addition to a global index of sleep disturbance (measured with the PSQI22 global score) and sleepiness (measured with the ESS25). There was a 72.5% frequency (n = 29) of moderate to severe depression (BDI-II ≥ 20) in the patients with PTSD. Comparison of the BDI-II29 scores between the RDI severity categories21 using ANOVA revealed that patients with severe OSA (RDI > 30 events/h) (n = 3) had significantly higher BDI-II scores (mean ± SD score: 46.67 ± 5.51) than patients with no OSA (RDI < 5 events/h) (n = 6) (mean ± SD score: 14.83 ± 7.88) (P < .001) and mild OSA (RDI 5 to < 15 events/h) (n = 24) (mean ± SD score: 26.58 ± 11.81 (P = .020), but not in comparison with the BDI-II scores in the moderate OSA group (RDI 15 to < 30 events/h) (n = 7) (mean ± SD score: 35.29 ± 7.65) (P = .739).
Analysis Involving the Entire Sample (n = 40)
The RDI (Pearson r = .757, P < .001) and ODI (Pearson r = .633, P < .001) were directly correlated to SI (Table 2). The results of the multiple regression analysis involving the entire sample (n = 40) (Table 3) revealed that RDI (standardized regression coefficient or β = .480, t = 4.167, P < .001) remained a significant predictor of SI even after controlling for depression (BDI-II) (β = .469, t = 3.375, P = .002) and other potential confounders (PCL-5 and PSQI-A_NM). Therefore, in addition to the RDI, only the depression scores measured by the BDI-II remained a significant predictor of SI in the regression model (Table 3). The results were similar when the RDI was substituted with the ODI in the multiple regression model (Table 3, footnote). As expected, the RDI and ODI were significantly correlated (Pearson r = .817, P < .0001).
As depression is associated with both OSA4 and SI, a separate analysis was carried out to further examine the possible mediation32 of the relationship between OSA and SI by depression. A mediator variable is defined32 as one that accounts for most or part of the relationship between the independent variable (IV) (ie, RDI) and dependent variable (DV) (ie, SI). The following three steps (as proposed by Baron and Kenny32) were carried out for the mediation analysis: (1) regression of the DV (SI) on IV (RDI) (standardized coefficient c = .757, P < .001); (2) regression of the mediator (BDI-II) on IV (RDI) (standardized coefficient a = .575, P < .001); and (3) regression of the DV (SI) on both the mediator (BDI-II) (standardized coefficient b = .473, P < .001) and IV (RDI) (standardized coefficient c′ = .485, P < .001). As c and c′ were both significant we tested whether c′ was significantly smaller than c using the Sobel test,34,35 which revealed Z = 3.065, P = .006. The results of the mediation analysis demonstrated that the relationship between the RDI and SI was partially mediated by depression (BDI-II), when the entire study sample (n = 40) was considered.
Analysis Involving the Subset of Patients With RDI < 30 events/h (n = 37)
The correlation between RDI and SI remained significant in the subset of patients without severe OSA (RDI < 30 events/h) (Pearson r = .511, P = .001) (Figure 2); however, the correlation of SI with ODI was no longer significant (Pearson r = .312, P = .060) in this less severely affected group. Result of the multiple regression analysis in this subset (Table 4) of patients with less severe OSA revealed that the RDI (β = .247, t = 1.723, P = .096) no longer remained a significant predictor of SI when depression (BDI-II) and other potential confounders (PCL-5 and PSQI-A_NM) were entered into the regression model, and only the depression scores (measured by BDI-II) remained a significant predictor (β = .603, t = 3.492, P = .002) of SI in the multiple regression model (Table 4). The results were similar when the RDI was substituted with the ODI in the multiple regression model (Table 4, footnote).
A separate mediation analysis32 was carried out to examine the possible mediation of the relationship between RDI and SI by depression (BDI-II). The following three steps (as proposed by Baron and Kenny)32 were carried out for the mediation analysis: (1) regression of the DV (SI) on IV (RDI) (standardized coefficient c = .511, P = .001); (2) regression of the mediator (BDI-II) on IV (RDI) (standardized coefficient a = .460, P = .004); and (3) regression of the DV (SI) on both the mediator (BDI-II)(standardized coefficient b = .593, P < .001) and IV (RDI) (standardized coefficient c′ = 0.239, P = .077). As c′ was not significant, the results of the mediation analysis indicated that in the subset of patients with RDI < 30 events/h, the relationship between RDI and SI was entirely mediated by depression (BDI-II).
We examined the relationship between the severity of OSA as measured by the RDI12,16 and SI in 40 civilian patients with PTSD. There was a direct correlation between RDI and SI (Pearson r = .757, P < .001) (Figure 1). The correlation remained significant when the subset of patients without severe OSA (RDI < 30 events/h) (n = 37) was considered (Pearson r = .511, P = .001) (Figure 2). To our knowledge, this is the first reported study that has examined the relationship of clinically measured OSA and SI in patients with PTSD.
The patients with PTSD, who all met DSM-52 criteria for moderate to severe PTSD at the time of initial consultation, were at different stages of recovery when they were recruited for the study, and their mean ± standard deviation (SD) PCL-528 score of 40.40 ± 18.81 (Table 1) was consistent with a PTSD diagnosis and suggestive of a wide range of PTSD severity. The mean ± SD BDI-II29 score was 27.85 ± 12.85 (Table 1), and 72.5% of the study sample had moderate to severe depression (BDI-II ≥ 20), which are both consistent with the fact that depression is frequently comorbid with PTSD.2 The patients with PTSD (Table 1) as a group were poor sleepers (as per the mean ± SD global PSQI21 score of 10.08 ± 4.79), were not excessively sleepy (as per the mean ± SD ESS25 score of 7.50 ± 4.44) even though 85.0% had some degree of OSA (AHI ≥ 5 events/h) (60% had mild OSA), and the patients with PTSD as a group reported mild insomnia (with a mean ± SD ISI24 score of 13.36 ± 6.97). These sleep characteristics are consistent with previous findings in patients with PTSD among whom there is a high prevalence of OSA,4 and who experience multiple sleep complaints and tend to have predominantly OSA with insomnia36 rather than hypersomnia. These clinical characteristics further support the representativeness of our study sample.
The most salient finding is the direct and graded ‘dose-response’ type relationship between clinically measured RDI and suicidal ideation (Figure 1, Figure 2, and Figure 3). Multiple regression analysis involving the entire study sample (n = 40) (Table 3) using SI as the dependent variable and RDI, depression, PTSD severity, and PTSD-related nightmares as independent variables revealed that both OSA (β = .480, t = 4.167, P < .001) and depression (β = .469, t = 3.375, P = .002) were significant predictors of SI, with the model explaining 71.8% of the variance in SI scores (Table 3). The RDI remained a significant predictor of SI, independent of comorbid depression and other factors that have been associated with increased suicide risk in PTSD. A separate mediation analysis32 showed a significant (P = .006) partial mediating effect of depression (BDI-II) in the relationship between RDI and SI. There was also a significant (Pearson r = .633, P < .001) direct relationship between the ODI and SI, which remained significant when RDI was substituted with the ODI in the multiple regression model where both ODI (β = .331, t = 2.737, P = .010) and depression (BDI-II) (β = .559, t = 3.687, P = .001) remained significant predictors of SI (Table 3, footnote). These results suggest the possible roles of both depression and OSA-associated intermittent hypoxemia in the relationship between RDI and SI in the entire study sample, which included patients with severe OSA. However, the same multiple regression analysis involving the subset of patients excluding severe OSA (RDI < 30 events/h) (n = 37) (Table 4) revealed that depression (BDI-II) alone (β = .603, t = 3.492, P = .002), and not RDI (β = .247, t = 1.723, P = .096) remained the only significant predictor of SI in the multiple regression model (Table 4). This was further confirmed by the results of the mediation analysis, where depression alone mediated the relationship between RDI and SI. These previously unreported finding in patients with PTSD suggest that depression (measured by the BDI-II) was the most important mediator in the relationship between OSA and SI, with OSA-related intermittent hypoxemia possibly further contributing to this relationship only in patients with PTSD and severe OSA.
Patients with PTSD are more likely to experience both OSA4 and depression2 and the comorbidity of OSA and depression is likely an important mediating factor in the relationship between OSA and SI in patients with PTSD. Patients with PTSD as a group are more likely to experience suicidal ideation33 and it is possible that the oxidative stress associated with intermittent hypoxemia in OSA37 further unmasks the underlying predisposition for suicidal ideation. Hypoxemia has also been associated with decreased serotonin synthesis and increased suicide rates,38 and this has been attributed to the effect of hypoxemia on tryptophan hydroxylase, the rate-limiting enzyme in serotonin synthesis.38 Finally, increased PTSD severity is typically associated with greater suicide risk and higher levels of sympathetic activation.2 The direct correlation between PTSD severity and SI was also present in our sample (Table 2). The higher level of sympathetic activation can lead to a lower respiratory arousal threshold, which can be associated with increased ventilatory instability and an increased RDI.39 This is further consistent with our finding that most of our patients with PTSD (60%) had mild OSA (RDI 5 to < 15 events/h), suggesting that the high level of sympathetic activation in patients with more severe PTSD could have in part contributed to a low arousal threshold and ventilatory instability39 and the modest increase in RDI observed in most of the patients. Increased sympathetic activation in PTSD may therefore be a mediating factor between SI and OSA severity in PTSD. SI in PTSD tends to be associated with more severe PTSD, and therefore SI, when associated with greater levels of sympathetic activation2 may in turn be associated with greater ventilatory instability.39 Our results suggest a possible bidirectional relationship between OSA and SI in PTSD.
In conclusion, our previously unreported finding of a graded “dose-response” type relationship between OSA severity and SI suggests the possible role of several factors (such as depression and possibly intermittent hypoxemia and a low respiratory arousal threshold secondary to the sympathetic activation in PTSD in severe OSA) underlying a possibly bidirectional relationship between OSA severity and SI in PTSD. Some of the limitations of this study include the small sample size, a high predominance of female subjects, lack of data on the menstrual status, and the fact that the subjects were not necessarily drug free. Our findings from this cross-sectional study need to be followed up with prospective studies that assess the relationship between changes in RDI and ODI (and other indices of oxygen desaturation) to SI before and after OSA treatment. Ideally these studies would comprise patients with PTSD who are drug free and represent a wide range of OSA severity—especially severe OSA. Our findings highlight the importance of comorbid OSA as one of the possible treatment targets in the management of SI in PTSD. Our results suggest that factors related to depression and central nervous system activation and factors related to upper airway obstruction and associated hypoxemia need to be taken into consideration when treating comorbid OSA in patients with PTSD and SI.
All authors have seen and approved the manuscript. The authors report no conflicts of interest.
analysis of variance
Beck Depression Inventory
Brief Symptom Inventory
Clinician Administered PTSD Scale
Diagnostic and Statistical Manual of Mental Disorders
Epworth Sleepiness Scale
Insomnia Severity Index
oxygen desaturation index
obstructive sleep apnea
peripheral arterial tonometry
Pittsburgh Sleep Quality Index
posttraumatic stress disorder
respiratory disturbance index
rapid eye movement
respiratory effort related arousal
The authors thank Dr. Robert C. Gardner, PhD, Professor Emeritus, Department of Psychology, University of Western Ontario, London, Ontario, Canada for statistical assistance and Branka Vujcic, BSc (Hon) for assisting with some of the data collection.
Bryan CJ. Treating PTSD within the context of heightened suicide risk. Curr Psychiatry Rep. 2016;18(8):73[PubMed]
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 5th ed. Arlington, VA: American Psychiatric Association; 2013.
Cox KS, Mouilso ER, Venners MR, et al. Reducing suicidal ideation through evidence-based treatment for posttraumatic stress disorder. J Psychiatr Res. 2016;80:59–63. [PubMed]
Gupta MA, Simpson FC. Obstructive sleep apnea and psychiatric disorders: a systematic review. J Clin Sleep Med. 2015;11(2):165–175. [PubMed Central][PubMed]
Bishop TM, Ashrafioun L, Pigeon WR. Association between sleep apnea and suicidal thought and behavior [abstract]. Sleep. 2017;40 suppl:A414–A415
Rod NH, Kjeldgard L, Akerstedt T, et al. Sleep apnea, disability pension and cause-specific mortality: a Swedish nationwide register linkage study. Am J Epidemiol. 2017;186(6):709–718. [PubMed]
Krakow B, Artar A, Warner TD, et al. Sleep disorder, depression, and suicidality in female sexual assault survivors. Crisis. 2000;21(4):163–170. [PubMed]
Bernert RA, Kim JS, Iwata NG, Perlis ML. Sleep disturbances as an evidence-based suicide risk factor. Curr Psychiatry Rep. 2015;17(3):554[PubMed]
Pigeon WR, Bishop TM, Titus CE. The relationship between sleep disturbance, suicidal ideation, suicide attempts, and suicide among adults: a systematic review. Psychiatr Ann. 2016;46(3):177–186
Pigeon WR, Pinquart M, Conner K. Meta-analysis of sleep disturbance and suicidal thoughts and behaviors. J Clin Psychiatry. 2012;73(9):e1160–e1167. [PubMed]
Bernert RA, Turvey CL, Conwell Y, Joiner TE Jr. Association of poor subjective sleep quality with risk for death by suicide during a 10-year period: a longitudinal, population-based study of late life. JAMA Psychiatry. 2014;71(10):1129–1137. [PubMed Central][PubMed]
Itamar Medical Ltd. Itamar zzzPAT for Watch-PAT 100,200 & 200U Software Operation Manual. Itamar Medical REF OM2197434. http://www.itamar-medical.com/wp-content/uploads/2016/01/zzzPAT-OM2197434.pdf. Published 2015. Accessed January 31, 2018.
Bar A, Pillar G, Schnall RP, Sheffy J. An Illustrated Atlas of PAT Signals in Sleep Medicine. Caesarea, Israel: Itamar Medical Ltd.; 2009.
Weathers FW, Blake DD, Schunrr PP, Kaloupek DG, Marx BP, Keane TM. The Clinician-Administered PTSD Scale for DSM-5 (CAPS-5). Washington, DC: U.S. Department of Veterans Affairs; 2013. Interview available from the National Center for PTSD at www.ptsd.va.gov.
Collop NA, Anderson WM, Boehlecke B, et al. Clinical guidelines for the use of unattended portable monitors in the diagnosis of obstructive sleep apnea in adult patients. Portable Monitoring Task Force of the American Academy of Sleep Medicine. J Clin Sleep Med. 2007;3(7):737–747. [PubMed Central][PubMed]
Yalamanchali S, Farajian V, Hamilton C, Pott TR, Samuelson CG, Friedman M. Diagnosis of obstructive sleep apnea by peripheral arterial tonometry: meta-analysis. JAMA Otolaryngol Head Neck Surg. 2013;139(12):1343–1350. [PubMed]
Pillar G, Bar A, Betito M, et al. An automatic ambulatory device for detection of AASM defined arousals from sleep: the WP100. Sleep Med. 2003;4(3):207–212. [PubMed]
Hedner J, White DP, Malhotra A, et al. Sleep staging based on autonomic signals: a multi-center validation study. J Clin Sleep Med. 2011;7(3):301–306. [PubMed Central][PubMed]
Herscovici S, Pe'er A, Papyan S, Lavie P. Detecting REM sleep from the finger: an automatic REM sleep algorithm based on peripheral arterial tone (PAT) and actigraphy. Physiol Meas. 2007;28(2):129–140. [PubMed]
Itamar Medical Ltd website. Sleep respiratory definitions. http://www.itamar-medical.com/patients-sleep-health-faq/. Accessed July 3, 2017.
American Academy of Sleep Medicine. International Classification of Sleep Disorders. 3rd ed. Darien, IL: American Academy of Sleep Medicine; 2014.
Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Res. 1989;28(2):193–213. [PubMed]
Germain A, Hall M, Krakow B, Katherine Shear M, Buysse DJ. A brief sleep scale for Posttraumatic Stress Disorder: Pittsburgh Sleep Quality Index Addendum for PTSD. J Anxiety Disord. 2005;19(2):233–244. [PubMed]
Bastien CH, Vallieres A, Morin CM. Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep Med. 2001;2(4):297–307. [PubMed]
Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep. 1991;14(6):540–545. [PubMed]
Derogatis LR, Spencer MS. Brief Symptom Inventory (BSI): Administration and Procedures Manual. Baltimore, MD: Clinical Psychometric Research, Johns Hopkins University School of Medicine; 1982.
Gupta MA, Gupta AK. Cutaneous body image dissatisfaction and suicidal ideation: mediation by interpersonal sensitivity. J Psychosom Res. 2013;75(1):55–59. [PubMed]
Weathers FW, Litz BT, Keane TM, Palmieri PA, Marx BP, Schunrr PP. The PTSD Checklist for DSM-5 (PCL-5). Washington, DC: U.S. Department of Veterans Affairs; 2013. Scale available from the National Center for PTSD at www.ptsd.va.gov.
Beck AT, Steer RA, Brown G. Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation; 1996.
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.
IBM Corp. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp.; 2016.
Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986 12;51(6):1173–1182. [PubMed]
Nock MK, Hwang I, Sampson N, et al. Cross-national analysis of the associations among mental disorders and suicidal behavior: findings from the WHO World Mental Health Surveys. PLoS Med. 2009;6(8):e1000123[PubMed Central][PubMed]
Sobel ME. Asymptotic intervals for indirect effects in structural equations models. In: Leinhart S, ed. Sociological Methodology. San Francisco, CA: Jossey-Bass; 1982:290–312.
Preacher KJ, Leonardelli GJ. Calculation for the Sobel test: An interactive calculation tool for mediation tests. http://quantpsy.org/sobel/sobel.htm. Accessed September 17, 2017.
Krakow B, Melendrez D, Warner TD, Dorin R, Harper R, Hollifield M. To breathe, perchance to sleep: sleep-disordered breathing and chronic insomnia among trauma survivors. Sleep Breath. 2002;6(4):189–202. [PubMed]
Lavie L. Oxidative stress in obstructive sleep apnea and intermittent hypoxia--revisited--the bad ugly and good: implications to the heart and brain. Sleep Med Rev. 2015;20:27–45. [PubMed]
Young SN. Elevated incidence of suicide in people living at altitude, smokers and patients with chronic obstructive pulmonary disease and asthma: possible role of hypoxia causing decreased serotonin synthesis. J Psychiatry Neurosci. 2013;38(6):423–426. [PubMed Central][PubMed]
Eckert DJ, White DP, Jordan AS, Malhotra A, Wellman A. Defining phenotypic causes of obstructive sleep apnea. Identification of novel therapeutic targets. Am J Respir Crit Care Med. 2013;188(8):996–1004. [PubMed Central][PubMed]