Suicide is the tenth leading cause of death in the U.S., is a major concern of the Department of Veterans Affairs (VA), and occurs at elevated rates among Veterans. Veterans with common chronic conditions such as PTSD and depression are at increased risk for suicide. Although evidence-based treatments exist for these conditions, a significant number of Veterans do not engage in or complete such treatments. This study, entitled "An Adjunctive Behavioral Sleep Intervention to Prevent Veteran Suicides," focuses reducing the sleep disturbance of insomnia because it so often presents among patients with PTSD and/or depression and it is itself a risk for suicidal thoughts and behaviors. Therefore, this study examines the use of cognitive-behavioral therapy for insomnia (CBT-I), an efficacious intervention that targets insomnia, as a novel way to reduce suicidal ideation in at-risk Veterans. Importantly, CBT-I has been successfully used in patients with PTSD and depression. Preliminary evidence from uncontrolled studies suggests that CBT-I may decrease suicidal ideation, but this has not been tested in a controlled trial.
The ultimate goal of the broader program of research is to reduce suicide among Veterans, while the proposed pilot project will provide data to guide the development of a larger trial. The primary objectives of this pilot project are to test whether using CBT-I as an adjunctive treatment: (1) is feasible to deliver in the context of other ongoing care and acceptable to participants; (2) can reduce the severity of suicidal ideation among Veterans at risk for suicide; and (3) can increase rates of engagement in treatments that address the co-morbid conditions that also put them at risk for suicide. The latter objective will help refine the study methods, design, and intervention in anticipation of a larger trial.
In order to achieve the proof-of-concept objectives, we designed a small randomized clinical trial. Veterans who were either flagged as high risk for suicide or who have depression or PTSD were recruited from two VA facilities in Western portion of VISN 2 (Canandaigua and Rochester). Approximately 56 such Veterans who also endorsed current suicidal ideation and current insomnia were randomized to receive either treatment-as-usual or a 4-session CBT-I treatment in addition to treatment-as-usual. All participants were encouraged to continue or to seek treatment for co-occurring conditions as recommended by their VA treatment providers. Assessments of suicidal ideation, insomnia, depression, PTSD, and quality of life were conducted at baseline and following the treatment period. In addition, we collected provider and participant feedback at an exit interview immediately following the post-treatment assessment. Multiple linear and logistic regression models were used to assess the effect of treatment on outcome measures across the study assessment time points.
Abbreviations and Notes:
CTRL Control Condition
CBT-I Cognitive Behavioral Therapy for Insomnia
SI Suicidal Ideation
C-SSRS Columbia Suicide Severity Rating Scale,
C-SSRS-SI Suicidal Ideation Intensity subscale of the C-SSRS
ISI Insomnia Severity Index
PCL PTSD Symptom Checklist
PHQ-9 Patient Health Questionnaire-9
ES Effect Size (note that ES are measured by Hedge's g which corrects
for smaller sample sizes and we also correct for baseline values).
The original recruitment and retention targets were to enroll and randomize 56 participants in order to retain 48 participants with analyzable data. A total of 581 potential participants were identified from medical records and received a recruitment letter from their current treatment provider; 420 of these Veterans were screened by phone with 71 meeting initial eligibility criteria and ultimately completing written informed consent and a full baseline interview. Of these, 54 participants met all eligibility criteria and were randomized (n=26 to the experimental condition [CBT-I]; n=28 to the control condition [treatment-as-usual]). No significant differences in participant characteristics were observed between conditions. We were able to retain 50 participants (a 93% retention rate). There were three dropouts in the experimental condition and one in the control condition. The analyzed sample included n=23 in CBT-I and n=27 in CTRL.
Aim 1: To test the effect of adding a CBT-I intervention to treatment as usual, compared to treatment as usual alone, on suicidal ideation in a sample of Veterans endorsing SI who have both insomnia as well as PTSD or depression.
For this aim, multiple regression controlling for baseline values of the dependent variable was used to examine treatment effects of the adjunctive CBT-I intervention. Here the primary outcome was SI Intensity as measured by the C-SSRS SI. As planned, this approach also was utilized to assess treatment effects on insomnia, depression and PTSD severity. Results from these regression analyses are presented in further detail below. For the main SI outcome, mean reduction in SI intensity was both significant and >50% for the CBT-I condition, but when compared to the control condition the difference did not reach statistical significance. The overall, adjusted effect size was moderate (0.44). Compared to CTRL, statistically significant (and large) effects were observed for CBT-I on insomnia severity and depression severity, but not for PTSD severity.
Post-Tx Means (SD)
CBT-I CTRL p ES
C-SSRS SI Intensity 5.4 (6.8) 8.4 (6.3) 0.153 0.44
ISI 7.9 (5.5) 17.8 (5.2) <.001 1.79
PHQ-9 6.4 (4.6) 12.3 (5.4) <.001 1.13
PCL 48.1(16.2) 52.3(16.6) 0.130 0.25
As a secondary SI outcome, rates of SI at post-treatment were assessed using logistic regression, which was also used to compare rates of participants with a decrease in their highest level of C-SSRS severity and rates of participants experiencing a 50% or greater reduction in total SI intensity. Here, although the differences in rates between groups were all in the expected direction, statistical significance was not reached.
CBT-I CTRL Raw Difference
No SI at Post-treatment 52% 41% 11%
Decrease in SI Risk 78% 59% 19%
> 50% decrease in 52% 37% 15%
Aim 2: To assess the feasibility and acceptability of delivering CBT-I as an adjunctive intervention in Veterans with suicidal ideation and co-occurring conditions.
For this aim a mixed methods approach was undertaken. Individual item and total scores on both the Client Satisfaction Questionnaire and the Treatment Acceptability Form reflect high satisfaction and acceptability of the intervention (data not shown). Prominent themes from qualitative data that were related to the structure of the CBT-I intervention included:
1. initial treatment initiation was difficult, but became easier over sessions;
2. maintaining initial recommended sleep schedules was the most difficult
component of care;
3. the option of having phone session was appreciated, but not important; and
4. similarly, the patient workbook was appreciated as an aid, but not as critical
or especially helpful for most.
Prominent themes (and representative participant quotes) from qualitative data related to the impact of the CBT-I intervention included:
1. Usefulness and high impact of the CBT-I intervention for sleep
("I was doubtful of the effectiveness of this approach in the beginning, but after tracking the changes I have come to appreciate this program."; " It's easier to get to sleep after awakenings; I usually fall asleep within 20 minutes as opposed to hours."; "Improved sleep, less sleep during the day, more quality sleep, reduced need for naps."' "I sleep longer, less awakening (and these awakenings are much shorter.")
2. Improved sleep improved mood
("I realized that proper sleep habits really help me with depression illness- I have more control over verbal outbursts and total melt downs."; "[I am] less tired, less prone to have freakouts."; "Mood has improved; no longer a monster."; "Fell better, less irritable, felt more accomplished" ; "Not so grumpy, more pep in [my] step.")
3. Improved sleep had additional benefits beyond sleep
("This intervention program was very insightful and helpful to me. It actually helped me connect more dots associated with ongoing PTSD."; "I have increased patience and less stress"; "Improved self-awareness; Improved understanding of importance of sleep for a healthy lifestyle"; "Felt more rested and was able to accomplish more things"; "Stress is not really less, but can deal easier now that am sleeping better.")
4. Improved sleep helped to reduce suicidal thoughts
("Haven't thought about this for any length of time recently, which is a change from before."; "Helpful because [my] overall mood has changed."; "Yes, zero [SI] since getting this treatment."; "They are not less frequent, but they are less intense, and easier to rid the thoughts from [my] mind."; "More likely to get help, wasn't talking about it before, but now from talking about it there can be an action plan."
5. The positive treatment experience increased engagement in existing therapies and/or likelihood of seeking care for other mental health issues
("I'm already in treatment programs, but now more likely to stick with them."; "I'm already getting treatment, but now commitment is stronger."; "I would be willing to try other things because of this good experience."; "I learned I can get help when I need it."; By seeking this treatment [I] was able to learn about other available treatments."
As a further aid to determining feasibility, we also developed reliable therapist fidelity measures for Adherence to the treatment manual and Competence in delivery of the CBT-I. The Adherence measures, one for each of the four CBT-I sessions, showed good to excellent intra-rater reliability (ICCs = .77 to .91). We transformed raw adherence scores into percent of the content delivered. On average, therapists delivered 80% of the content of CBT-I overall, with the greatest variability in Session 3, which exceeds the minimum established in the literature.
The Competency measure was consistent across all sessions because the qualities required to deliver the treatment does not vary. The measure showed a truncated range, therefore ICCs were not appropriate to reflect intra-rater reliability. Instead, we calculated raters' 'exact' agreement on competence scores across all four sessions (67% of ratings) as well as ratings that were '1 point difference' (32%). Based on these calculations, raters were very reliable. In terms of therapist competence, ratings were in the 'good work' category with average scores in the 11 to 12 range for each session (on a scale of 4-12). In sum, reliable measures of: a) therapist adherence to the manual, and b) competence in delivery of CBT-I, showed excellent therapist fidelity. Therapists' high level of competence indicates that they tailored the content of the intervention for patients. Strict adherence to the manual, without sensitivity to the individual patient, has been shown in other intervention studies to be associated with poorer outcomes.
Exploratory Aims: Although exploratory, we tested whether (a) receiving CBT-I enhances engagement in care for co-occurring conditions and (b) change in insomnia severity mediates effects of CBT-I on SI.
For the treatment engagement portion of this exploratory aim we used a 21-item instrument to assess likelihood and intent to engage in behavioral health care treatment along with four additional 1-10 point Likert scale questions specific to seeking care for sleep, depression, PTSD and suicidality. In both instances, it appears that a ceiling effect may be in place (e.g. mean values of 9.0 out of 10 on the Likert items at baseline). There were modest increases in the 21-item total PASS score as well as the four single item seeking care questions, but none reached statistical significance. An equal, but high, percentage of participants in each condition were already receiving (or had received) behavioral care at baseline. Data from chart review intended to measure actual engagement in care mirrored the above observations. Although the sample size is small, we may assess a subsample of participants who scored low on treatment engagement variables at baseline to determine if there was an effect of CBT-I on such a subsample.
For the mediational portion of this exploratory aim, we examined a preliminary model of insomnia severity mediating the intervention effect on suicide intensity using path analysis. Here, condition did have a strong effect on insomnia severity (B = -.62, p < .001), and insomnia severity had a marginal positive relationship with suicide intensity (B = .31, p = .096). The Sobel test (to determine mediation effect) was marginally significant (B = -.19, p = .104), suggesting that the level of insomnia severity following treatment may mediate the relationship of CBT-I on SI intensity. While power is certainly an issue with these specific results, we are optimistic regarding the direction and strength of the pilot mediational effects and their utility in guiding future work. Particularly as additional exploratory analyses revealed that participants with post-treatment insomnia severity values below the clinical cutoff of 10 on the ISI, regardless of condition, had lower mean levels of SI intensity (p =.027) and a lower rate of SI (33% vs. 63%; p =.055).
This project focused on a novel suicide prevention strategy that can be delivered to a broad range of Veterans. Namely, improving sleep using a brief form of Cognitive Behavioral Therapy for Insomnia with Veterans who have insomnia and recent suicidal ideation and also suffer from additional conditions such as depression and PTSD. Findings from the study show that the brief insomnia treatment was feasible to deliver, delivered with fidelity and competence, and found to be acceptable and helpful by Veterans. Participants receiving the intervention had significant reductions in insomnia, depression and PTSD severity along with significant reductions in the intensity of suicidal ideation. When compared to control participants, reductions in insomnia and depression severity remained significant; these effects were large. Reductions in suicidal ideation intensity were not significantly different from that of the control group; the difference was in the expected direction and reached a moderate effect size. This suggests that the intervention has a moderate effect on suicidality, but that more patients then were studied in this pilot project need to be treated to achieve statistical significance, which has implications for the design of fully-powered clinical trials. Findings also suggest that reductions in insomnia severity mediate the effect that being in the insomnia treatment has on reducing the presence and intensity of suicidal thoughts. Qualitative findings corroborate that the Veterans receiving the brief insomnia treatment experienced meaningful improvements in their mood. Strong themes also emerged that the improvements in sleep generalized to better overall functioning and that Veterans became more committed to their existing behavioral treatments and/or to seeking such care.
In sum, this pilot study provides proof of concept that suicidal ideation can be reduced by adjunctively treating insomnia and that adding this sleep intervention to usual care may enhance overall care by improving mood, increasing function, and enhancing commitment to care. An important exploratory finding is that to the extent that insomnia severity is reduced to levels below clinical values, there is a significant reduction in rates of suicidal ideation and the intensity of those thoughts.
- Pigeon WR, Funderburk JS, Cross W, Bishop TM, Crean HF. Brief CBT for insomnia delivered in primary care to patients endorsing suicidal ideation: a proof-of-concept randomized clinical trial. Translational behavioral medicine. 2019 Nov 25; 9(6):1169-1177.