The Effects of Cognitive Processing Therapy + Hypnosis on Objective Sleep Quality in Women with Posttraumatic Stress Disorder

Abstract

Objective:

Insomnia, characterized by difficulty falling and staying asleep, is a common and debilitating symptom of posttraumatic stress disorder (PTSD) that is resistant to first-line, trauma-focused therapies. Previous research has found that sleep-directed hypnosis improves subjective sleep quality, particularly sleep onset latency, in women with PTSD. However, it cannot be assumed that improvements in subjective sleep reports correspond with objectively measured sleep improvements, as research has indicated lack of agreement across these measures. The current study examined the effects of sleep-directed hypnosis + cognitive processing therapy on objective indices of sleep quality measured with actigraphy.

Method:

Forty-five women with PTSD were randomized to receive sleep-directed hypnosis + cognitive processing therapy (hypCPT) or sleep and psychiatric symptom monitoring + CPT (ssmCPT). Pre- and post-treatment, participants completed one week of daily actigraphy assessments of nocturnal sleep onset latency, waking after sleep onset, and total sleep time.

Results:

Overall improvement in objective sleep indices was not observed. Despite this, at post-treatment, treatment completers receiving hypCPT took significantly less time to fall asleep than women receiving ssmCPT.

Conclusions:

More research is needed to understand and reduce the discrepancy between subjectively and objectively assessed sleep impairments in PTSD. Nevertheless, results indicate that adding sleep-directed hypnosis to trauma-focused therapy may be of some use for individuals with PTSD-related insomnia.

Insomnia, characterized by difficulty falling or staying asleep, is a common and debilitating symptom of posttraumatic stress disorder (PTSD). Indeed, between 80 and 90% of individuals with PTSD endorse clinically significant insomnia symptoms (Koffel, Khawaja, & Germain, 2016). Notably, insomnia is resistant to first-line treatments for PTSD, including cognitive processing therapy (CPT) and prolonged exposure (PE; Belleville, Guay, & Marchand, 2011; Larsen, Fleming, & Resick, 2019). For example, a recent study found that following CPT, 74 to 80% of participants maintained clinically significant insomnia symptoms (Pruiksma et al., 2016). Such findings highlight the need for treatments that better address insomnia in the context of PTSD.

Previously, our research group examined sleep-directed hypnosis as an adjunct to CPT for PTSD (Galovski et al., 2016). Prior to initiating CPT, participants completed three weeks of sleep-directed hypnosis (hypCPT) or were asked to monitor sleep and psychiatric symptoms (ssmCPT). When assessed after the sleep intervention but prior to CPT, participants receiving sleep-directed hypnosis demonstrated greater improvement than ssmCPT participants in self-reported global sleep quality and sleep onset latency (SOL), but not total sleep time (TST). Following CPT, both conditions evidenced statistically significant improvement in global insomnia symptoms and did not differ from each other.

Despite promising effects on self-reported insomnia symptoms, it remains unknown whether adding sleep-directed hypnosis to CPT improves sleep that is objectively measured (e.g., with actigraphy). Research cannot assume that improvements in subjective sleep reports correspond with objectively measured sleep improvements as previous research has indicated lack of agreement across these measures (e.g., Werner, Griffin, & Galovski, 2016). Evidence supporting objectively assessed sleep impairments in PTSD has been equivocal (Koffel et al., 2016). Research using actigraphy suggests that individuals with, vs. without, PTSD experience longer SOL, but are not likely to differ in number of nighttime awakenings, amount of time spent awake after initially falling asleep (waking after sleep onset; WASO), or TST (Slightam et al., 2018). Because sleep-directed hypnosis targets SOL (Stanton, 1989), it may be particularly useful for improving the aspect of objective sleep quality that appears to be most impaired among individuals with PTSD.

This study compared the effects of sleep-directed hypnosis and symptom monitoring on objective sleep indices in women with PTSD completing CPT. Based on previous research (Galovski et al., 2016), we expected participants in both conditions to demonstrate improved SOL following treatment. However, we expected improvement to be greater for participants receiving sleep-directed hypnosis. Given previous research, significant between-group changes over time were not expected for WASO or TST.

Method

Participants

Data came from a larger (N = 92) randomized controlled trial comparing hypCPT and ssmCPT for female survivors of interpersonal violence with PTSD (Galovski et al., 2016). Women in the parent study were invited to participate in an auxiliary study examining objective sleep quality pre- and post-treatment. Forty-five women participated. Women opting into the study did not differ in age, race/ethnicity, PTSD symptom severity, or frequency of sleep medication use from women who declined participation (ps > .05).

Participants were ≥ 18 years old, had current PTSD, and endorsed PTSD-related sleep impairment (i.e., frequency and intensity score ≥ 3 for symptom D1 [Trouble Initiating or Falling Asleep] on the Clinician Administered PTSD Scale [CAPS-IV]; Blake et al., 1995). Exclusion criteria included psychosis, alcohol or substance abuse or dependence, suicidality, or a peritraumatic living situation. Psychotropic medication use was required to be stable prior to participation. Women could not receive outside trauma- or sleep-focused psychotherapy during their participation.

Participants identified as White/Caucasian (53.3%) or Black/African-American (46.7%); 2.2% also identified as Hispanic/Latina. Average age was 36.16 (SD = 11.66) years. Frequency of sleep medication use (Pittsburgh Sleep Quality Index; Buysse et al., 1989) was: 46.3% never, 7.3% less than once a week, 17.1% once or twice week, 29.3% three or more times a week.

Sleep Assessment

Objective sleep quality was assessed with actigraphy. Research indicates that actigraphy collected over several nights is a valid indicator of sleep compared to independent in-laboratory polysomnography (PSG; Withrow, Roth, Koshorek, & Roehers, 2019). Actometers (Ambulatory Monitoring Inc., Ardsley, NY) were used to obtain daily assessments of SOL, WASO, and TST at pre- and post-treatment. Actigraphy measurements were collected in 30s epochs using the proportional integrating mode, which corresponds most reliably with PSG assessments (Blackwell et al., 2008). Action-W analysis software (v 2.0; Ambulatory Monitoring Inc.) provided down intervals and estimates of sleep/wake parameters. The University of California – San Diego algorithm (Cole et al., 1992) was used to calculate SOL, WASO, and TST for each night of recording. Variables of interest included pre- and post-treatment means (in minutes) for SOL, WASO, and TST.

Treatment Conditions

Participants were randomized to receive either hypCPT (n = 27) or ssmCPT (n = 18). Detailed descriptions of these treatment conditions can be found in Galovski et al. (2016). Briefly, participation in the ssmCPT condition began with a 3-week period of daily sleep and psychiatric symptom monitoring, and included weekly phone check-ins with a study therapist. Participants in the hypCPT condition also monitored sleep and psychiatric symptoms and received 3 weekly, 60min sessions of sleep-directed hypnosis. During sessions, hypnotic trance was induced using an eye fixation technique and was deepened using progressive muscle relaxation. Sessions also included guided imagery, psychoeducation, and the use of self-statements. Sessions were largely scripted. Hypnosis sessions were taped and participants were asked to practice hypnosis on a nightly basis. Following the sleep intervention phase, all participants completed a standard CPT protocol consisting of 12 weekly, 60min sessions (for a description, see Resick & Schnicke, 1992).

Procedure

Procedures were approved by the University of Missouri – St. Louis’s Institutional Review Board. At an initial visit, inclusion/exclusion criteria, demographic characteristics, and use of sleep medication were assessed. Participants were then fitted with an actometer on their non-dominant wrist and were instructed to wear it 24 hours/day for 7 days. Participants returned to the clinic within two weeks at which point actigraphy data were collected by study staff. Some participants wore the actometer beyond 7 days as initially instructed. Across participants, all available actigraphy data were used in analyses (M = 6.35, SD = 3.01 days). Participants then completed either hypCPT or ssmCPT protocols followed by a second, post-treatment actigraphy assessment (M = 6.92, SD = 3.94 days).

Data Analysis

Actigraphy data were available from 43 participants at pre-treatment and 24 at post-treatment. At pre-treatment, missing data were due to technical difficulties with the actometer (n = 2). At post-treatment, missing data were due to technical difficulties (n = 3) or attrition from the study (n = 18). Attrition largely occurred prior to initiating treatment, within the first 5 sessions of CPT, or after CPT but prior to the post-treatment actigraphy assessment. Rates of attrition did not differ by condition, χ² (1) = .02, p = .89. Further, missingness was unrelated to age, race/ethnicity, PTSD symptom severity, or sleep medication use (ps > .05). Data were thus considered to be missing at random.

Sleep outcomes were examined using linear mixed models in SPSS 25. This approach maximized use of available data to examine differences between conditions at pre- and post-treatment. Within analyses, we used a compound symmetry covariance structure to account for within-subject correlated errors. Separate models were conducted to examine SOL, WASO, and TST. For each outcome, we first examined a main effects model, including fixed effects of Time (pre-treatment = 0, post-treatment = 1) and Condition (ssmCPT = 0, hypCPT = 1). We then examined a model including a Time x Condition interaction term. The models’ −2 restricted log likelihood values were compared using a χ² difference approach. In all cases, the model including the interaction term provided significantly better fit than the main effects model (all −2 restricted log likelihood_diff > 5.91, df = 1, p < .02). Thus, the results presented below only include the models with the interaction terms. Within these models, the Condition effect reflects the difference between conditions at post-treatment, the Time effect reflects the slope from post-treatment to pre-treatment for hypCPT participants, and the Condition x Time effect reflects the change in slope between hypCPT and ssmCPT participants.

Results

Pre-treatment, participants took an average of 19.98min (SD = 15.84) to fall asleep and spent an average of 46.64min (SD = 25.37) awake during the night. Average TST was 406.94min (SD = 69.89), which is within the range of normal sleep duration for healthy adults (Fernandez-Mendoza et al., 2011).

We first examined the intent-to-treat sample (N = 45). As seen in Table 1, SOL did not change significantly for hypCPT (pre-treatment: M = 14.70, SE = 4.24min; post-treatment: M = 19.71, SE = 3.23min) or ssmCPT (pre-treatment: M = 20.48, SE = 4.17min; post-treatment: M = 27.29, SE = 5.48min) participants. However, at post-treatment, participants in the hypCPT condition took marginally less time to fall asleep than participants in the ssmCPT condition. No significant effects emerged from the models examining WASO or TST (Table 1).

Table 1.

Estimates of fixed effects for intent-to-treat and treatment completer samples.

Variable	B	SE	df	t	p	95% CI
Intent-to-Treat Sample (N = 45)
SOL
Condition	12.59	6.93	62.49	1.82	.07	−1.26, 26.43
Time	5.01	4.70	24.55	1.07	.30	−4.68, 14.70
Time x Condition	−11.83	7.82	28.99	−1.51	.14	−27.82. 4.17
WASO
Condition	−6.07	8.79	62.98	−.69	.49	−23.64, 11.50
Time	−.40	4.53	22.06	−.09	.93	−9.79, 8.98
Time x Condition	−.17	7.77	24.59	−.02	.98	−16.19, 15.84
TST
Condition	−3.91	27.70	62.74	−.14	.89	−59.28, 51.45
Time	−7.40	16.44	28.16	−.45	.66	−41.08, 26.27
Time x Condition	−15.62	27.84	31.85	−.56	.58	−72.35, 41.10
Treatment Completer Sample (n = 33)
SOL
Condition	12.47	5.92	50.84	2.11	.04	.59, 24.35
Time	2.78	4.33	28.30	.64	.53	−6.08, 11.64
Time x Condition	−12.83	7.24	31.44	−1.77	.09	−27.59, 1.93
WASO
Condition	−8.73	8.24	47.53	−1.06	.30	−25.31, 7.84
Time	−.47	4.43	23.43	−.11	.92	−9.63, 8.69
Time x Condition	−1.30	7.62	25.65	−.17	.87	−16.97, 14.37
TST
Condition	5.78	28.55	49.14	.20	.84	−51.59, 63.15
Time	−10.11	16.94	24.84	−.60	.56	−45.01, 24.80
Time x Condition	−5.02	28.92	27.45	−.17	.86	−64.32, 54.27

Note. SOL = Sleep onset latency; WASO = Waking after sleep onset; TST = total sleep time.

Next we examined participants who completed treatment (n = 33). Estimates of fixed effects are presented in Table 1 and estimated marginal means are presented in Figure 1. Among hypCPT participants, SOL did not change significantly from pre- to post-treatment. Among ssmCPT participants, SOL scores marginally increased from pre- to post-treatment. At the post-treatment assessment, participants in the hypCPT condition took significantly less time to fall asleep than participants in the ssmCPT condition. No significant effects emerged from the models examining WASO or TST. Of note, controlling for sleep medication use did not alter intent-to-treat or treatment completer results.

Figure 1.

Mean a) sleep onset latency, b) waking after sleep onset, and c) total sleep time as a function of condition assessed pre- and post-treatment in treatment completer sample.

Note. * p = .04.

Discussion

This study examined treatment-related change in objectively-assessed sleep among individuals with PTSD. It also examined whether sleep-directed hypnosis is a useful adjunct to CPT for addressing PTSD-related insomnia symptoms. The intervention affected SOL, particularly among treatment completers. At post-treatment, hypCPT participants fell asleep an average of 13 minutes faster than ssmCPT participants. This was mainly driven by a marginally significant pre-to post-treatment increase in SOL among ssmCPT participants. Consistent with previous research (Galovski et al., 2016; Slightam et al., 2018), the intervention did not significantly alter WASO or TST.

Despite this observed difference, treatment-related change in objective sleep variables was minimal. Because objective TST prior to treatment was not impaired relative to norms for healthy adults (Fernandez-Mendoza et al., 2011), lack of improvement may reflect a ceiling effect. These results are in contrast to our previous research using self-reported sleep outcomes, where both conditions improved significantly in SOL and global sleep quality (Galovski et al., 2016). The discrepancy between results relying on self-reported vs. objective sleep impairments is consistent with previous PTSD research (Slightam et al., 2018; Werner et al., 2016), as well as sleep intervention research with other patient populations (Lund et al., 2013). Elucidating reasons for a discrepancy between self-reported and objective sleep (i.e., sleep state misperception) could inform our understanding of the debilitating insomnia symptoms experienced by so many people with PTSD (Rezaie, Fobian, McCall, & Khazaie, 2018).

Results should be considered preliminary, as the study’s sample was small and a sizable percentage of post-treatment data were missing. Though our analytic approach maximized use of available data, actigraphy was only collected at two time points and analyses examining change over time were particularly underpowered to detect effects. The study also did not include an actigraphy assessment immediately following the sleep intervention phase of treatment. Our previous research found the strongest effects of hypCPT vs. ssmCPT immediately after the sleep intervention phase and before starting CPT (Galovski et al., 2016). We expect that future investigations will see stronger effects if actigraphy assessment occurs in closer temporal proximity to the sleep-directed hypnosis intervention. Though controlling for use of sleep medication did not alter study results, data on the use of specific medications were not collected in this study.

Understanding how treatment affects objectively-measured sleep is critical to addressing the, often debilitating, insomnia experienced by individuals with PTSD. Results of the current study suggest that CPT does not improve (and may even worsen, at least in the short term) objective sleep outcomes. Adding sleep-directed hypnosis may be of some value, though more research is needed to confirm this. Studies aimed at elucidating and reducing sleep-state misperception in individuals with PTSD may also prove useful.

Clinical Impact Statement:

This study examined the effects of sleep-directed hypnosis and cognitive processing therapy on objectively assessed sleep quality in women with posttraumatic stress disorder (PTSD). In contrast to the severe insomnia symptoms they reported, participants exhibited relatively normal objective sleep patterns. Despite this, compared to controls, women receiving sleep-directed hypnosis tended to take less time to fall asleep at the beginning of the night. This suggests that sleep-directed hypnosis may be a promising add-on to existing trauma-focused therapies.