Antidepressant medications and evidence-based psychotherapies are effective treatments for depression. However, antidepressant medication adherence continues to be suboptimal and offering access to and retaining patients in evidence-based psychotherapies for depression is challenging. The VHA has made major efforts to expand access to and capacity for depression focused evidence-based psychotherapies. However in 2012, only 35% of VA patients with depression completed any psychotherapy visit and only 6% completed 8 psychotherapy visits in 14 weeks, a proxy for an adequate trial.
Initiation and engagement in depression treatment may be increased if patients are offered timely access to a highly flexible, evidence-based treatment options, such as a tailored, web-based computerized cognitive-behavioral therapy (cCBT) program. cCBT has sufficient evidence for effectiveness to be a covered benefit in United Kingdom National Health Service, but has much larger effects when it is supported by clinicians or other trusted individuals. VA peer specialists, an increasing workforce in VHA mental health, are ideal candidates to support patients' engagement in depression treatment and in cCBT. Peers have been shown to promote use of Internet self-management tools, and peer specialist support for cCBT may enhance its completion and overall effectiveness through regular "check ins" and also through providing peer specific experiences, such as sharing of lived experiences and modeling of self management and recovery.
We are conducting a Hybrid Type I RCT of Peer-Supported cCBT versus enhanced usual care (EUC) for 330 patients with new episodes of depression in primary care at three VA sites and their associated CBOCs. Our specific aims are to compare PS-cCBT versus enhanced usual care (EUC) on: a) patient symptomatic, functional, and recovery-oriented outcomes, b) depression coping skills, antidepressant medication adherence, and initiation and completion of more intensive traditional psychotherapy (contingent on symptom level). Our secondary aims are to assess patient, peer, and providers' experiences in PS-cCBT versus EUC using mixed methods. Finally, our exploratory aim is to assess potential mediators of improvements in depressive symptoms or functional status.
This is a Hybrid I randomized controlled trial (RCT) of peer-supported cCBT compared to enhanced usual care (EUC) among Veterans with new diagnoses of depression in primary care. The cCBT program consists of 8 modules and will be supported by VA peer specialists with weekly contact for 12 weeks. We will assess patient symptomatic, functional and recovery-oriented outcomes at 12 and 24 week following randomization. We will also assess potential mediators of these outcomes. Bivariate and multivariate study analyses will assess the impact of the two study arms at each assessment point and over time.
Participant enrollment and baseline characteristics:
Of the 11,406 patients with a depression diagnosis at participating sites, 2,682 were eligible for recruitment based on chart screening. The most common reasons for ineligibility were no documentation of active depressive symptoms (49%), referral to specialty mental health care (30%), and an active comorbid substance use disorder (6%). Eligible patients were sent recruitment letters and all but 11 were called by phone. Of those called, 1213 refused participation, 339 were unable to be reached, 243 had a negative depression screen, 213 had no access to the internet, 322 were ineligible for other reasons, 11 initially agreed to participate but timed out, and 330 were enrolled into the study. 163 were randomized to enhanced usual care (EUC) and 167 were randomized to Peer-supported computerized cognitive behavioral therapy (PS-cCBT).
The sample had a mean age of 52 years and was 80% male, 72% White, 21% Black, 9% other or missing race, 3% Hispanic, 38% college graduates, 56% married or living together, 45% working full or part-time, 38% with an annual income less than $25,000, and 21% living alone. There was a statistically significant difference between groups in the percentage who had a high school education vs. some college, but there were otherwise no significant differences between groups among any of these demographic characteristics or with respect to baseline depression symptoms (QIDS-SR), general mental health status (VR-12 MCS), quality of life (Q-LES-Q), or mental health recovery (RAS).
Retention and Engagement:
Among the 330 patients randomized and enrolled in the trial, 236 (72%) completed the 12-week follow-up assessments and 219 (66%) completed the 24-week follow-up assessments. Participants in the PS-cCBT arm were more likely to not complete 24-week follow-up assessments (OR 2.2; 95% CI: 1.1, 4.3). Female participants, and those with more lifetime depressive episodes and with later onset of first depressive episode were more likely to complete the 24-week follow-up assessments. Income was also associated with completing the 24-week follow-up assessments.
Participants assigned to the PS-cCBT arm completed a mean of 3.8 (SD 3.0) of the 8 computerized CBT modules and completed a mean of 3.0 (SD 2.7) contacts with a Peer Support Specialist, typically by phone after an initial in-person visit.
In unadjusted bivariate comparisons, there were no statistically significant differences in the means between treatment arms at 12 or 24 weeks for any of the primary outcomes of depression symptoms (QIDS-SR), general mental health status (VR-12 MCS), quality of life (Q-LES-Q), and mental health recovery (RAS). Response rates, defined as a 50% reduction in depression symptoms from baseline, were higher at 20% in the PS-cCBT group compared to 6.4% for EUC (p = .002) at 12 weeks and 25.3% and 16.4% (p = .11), respectively, at 24 weeks. Remission rates, defined as a QIDS-SR score < 5, were 14.0% for PS-cCBT and 6.3% for EUC (p = .046) at 12 weeks and were 21.8% and 11.0% (p = .03), respectively, at 24 weeks.
We compared differences between treatment arms in the primary outcomes using outcome data assessed at all three times of baseline, 12 and 24 weeks and using mixed-effects regression models. Primary predictors included treatment arm indicator, indicators of 12 and 24 week times, and interactions of treatment arm by 12 and 24 weeks. The model was also adjusted for recruitment site, predictors of follow-up assessment completion, and baseline values for variables correlated with the follow-up outcomes (race, education, social support, and physical health). In the model of depressive symptoms, the intervention group by time interaction was statistically significant at 12 weeks (b = -1.4; 95% CI: -2.5, -0.3; p = .01) with greater improvement in the PS-cCBT group, and the interaction was marginally significant at 24 weeks (b = -1.1; 95% CI: -2.3, 0.1; p = .08). Findings were similar for quality of life, demonstrating significant effects at 12 weeks (b = 2.6; 95% CI: 0.5, 4.8; p = .02), and marginally significant effects at 24 weeks (b = 2.4; 95% CI: -0.1, 4.9; p = .06). In the model of general mental health status, the group by time interaction was not significant at 12 weeks (b = 2.2; 95% CI: -0.8, 5.1; p = .15) and was marginally significant at 24 weeks (b = 3.3; 95% CI: -0.03, 6.6; p = .05) with greater improvement in the PS-cCBT group. In the model of mental health recovery, the group by time interaction was significant at both 12 weeks (b = 3.6; 95% CI: 0.9, 6.2; p = .01) and 24 weeks (b = 4.5; 95% CI: 1.2, 7.7; p = .01) with the PS-cCBT group demonstrating greater recovery over time. "Age of first depressive episode" was a variable predictive of missing follow-up assessment and thus all models were adjusted for this variable. However, this variable itself was missing in a substantial percentage of study participants (8%). Thus, post-hoc sensitivity analyses were conducted by fitting a model without the "age of first depressive episode" variable, and findings were similar except that the group by time interaction was additionally significant for both quality of life (b = 2.8; 95% CI: 0.4, 5.2; p = .02) and mental health status (b = 3.6; 95% CI: 0.3, 6.8; p = .03) at 24 weeks, with greater improvements in these areas among those in the PS-cCBT group.
In bivariate comparisons of the secondary outcomes, we found no difference between study arms with respect to antidepressant medication possession ratios or self-reported adherence among antidepressant users or the percentage of non-users who started an antidepressant during the study. We found no difference between arms in the receipt of any psychotherapy or of 8 or more sessions of psychotherapy. We found no difference in measures of CBT skills, hope, suicidal ideation, or anxiety between groups at 12 or 24 weeks.
In mediation analyses, we included CBT skills at each time point in the mixed regression models of the primary outcomes. In the model of depression symptoms, CBT skills was a significant predictor (p = <.001) and the estimate for the group by time interaction at 12 weeks decreased from -1.4 (95% CI: -2.5, -0.3) in the model without CBT skills to -1.1 (95% CI: -2.2, -0.1) in the model that includes CBT skills. When CBT skills was added to the models of quality of life and mental health recovery, the group by time interactions at 12 weeks were no longer significant, suggesting CBT skills partially mediates the intervention effects on these outcomes.
Semi-structured interviews were completed with 28 patients and 4 Peer Support Specialists. Transcripts were analyzed using a comprehensive coding process. Preliminary findings suggest that participants found PS-cCBT helpful by increasing their coping skills and decreasing depressive symptoms. Most participants reported a positive experience with their peer. Reported benefits of the peer component included motivation, accountability, reminders, and technical support. The peer support specialists described the role they provided as encouraging participants to complete the cCBT modules by sharing their personal experiences with completing with the modules themselves. The peer support specialists felt they offered a personal touch to a computerized program, however, reaching participants by phone was an ongoing challenge for them. Both participants and peers reported wishing there was more face to face contact and a desire for the program to be longer than 12 weeks.
Study findings suggest peer-supported computerized cognitive behavioral therapy results in superior depression and related mental health outcomes compared to enhanced usual care for patients treated in VA primary care. While clinically significant effects were seen when comparing depression response and remission rates at 12 weeks (e.g., number need to treat of 7 for 12-week response), effects were less pronounced at 24 weeks. Also, while adjusted longitudinal models demonstrated consistent statistically significant beneficial intervention effects, the mean effect sizes in these models may not be clinically significant (e.g., 1.4 point difference in QIDS-SR). Engagement with the intervention was moderate and could possibly be improved with a computer CBT program tailored to Veterans or more easily accessed via smartphones.
Based on these findings, the VA should consider implementation of PS-cCBT to provide VA patients an effective option to augment usual primary care treatment of depression. Further research should seek to optimize both the peer support and cCBT components to improve engagement and effectiveness.
- Abraham KM, Nelson CB, Ganoczy D, Zivin K, Brandfon S, Walters H, Cohen JL, Valenstein M. Psychometric analysis of the Mental Health Recovery Measure in a sample of veterans with depression. Psychological Services. 2016 May 1; 13(2):193-201.
- Barry CN, Abraham KM, Weaver KR, Bowersox NW. Innovating team-based outpatient mental health care in the Veterans Health Administration: Staff-perceived benefits and challenges to pilot implementation of the Behavioral Health Interdisciplinary Program (BHIP). Psychological Services. 2016 May 1; 13(2):148-55.
- Pfeiffer PN. The Role of Peer Support in Suicide Prevention. Public Health Approach Webinar Series [Cyberseminar]. Defense Suicide Prevention Office (DSPO). 2016 May 25.