A Pilot Study of an Intervention to Prevent Suicide After Psychiatric Hospitalization

Abstract

A prior meta-analysis found that the World Health Organization Brief Intervention and Contact Program (WHO BIC) significantly reduces suicide risk. WHO BIC has not been studied in high-income countries. We piloted an adapted version of WHO BIC on an inpatient mental health unit in the United States. We assessed the feasibility and acceptability. We also evaluated changes in suicidal ideation, hopelessness, and connectedness using a repeated measures analysis of variance. Of 13 eligible patients, 9 patients enrolled. Patients experienced significant improvements in suicidal ideation, hopelessness, and connectedness at 1 and 3 months (Beck Scale for Suicidal Ideation, F(2,16) = 14.96, p < 0.01; Beck Hopelessness Scale, F(2,16) = 5.88, p < 0.05; perceived burdensomeness subscale, F(2,16) = 10.97, p < 0.013; and thwarted belongingness subscale, F(2,16) = 4.77, p < 0.03). Patients were highly satisfied. An adapted version of WHO BIC may be feasible to implement in a high-resource setting, but trials need to confirm efficacy.

Death by suicide is a major public health concern in the United States, where rates have risen by more than 20% over the past decade (National Institute of Mental Health, 2018). Of particular concern is the high risk for suicide in patients in the period after a psychiatric hospitalization. For example, Valenstein et al. (2009) found in a study of over 850,000 US veterans diagnosed with depression that the suicide rates were the highest in the first 3 months after discharge from an inpatient mental health unit at 568 per 100,000 person-years. Olfson et al. (2016) also found in a nationwide study of patients enrolled in the Medicaid program that patients with mental health disorders were at notably higher risk for suicide within the first 90 days of psychiatric hospitalization (rates ranging from 160.4 to 235.1 per 100,000 person-years) when compared with a cohort of hospitalized patients with nonmental disorders (11.6 per 100,000 person-years) and matched, general US population (14.2 per 100,000 person-years). Furthermore, in a meta-analysis of 100 publications examining suicide risk after a psychiatric hospitalization, Chung et al. (2017) concluded that patients are at increased risk for suicide after hospitalization regardless of prior suicidal behavior. In fact, the suicide rate in the first 3 months after discharge was almost 100 times that of the global suicide rate (1,132 per 100,000 person-years vs. 11.4 per 100,000 person-years, respectively), and the suicide rate was further elevated in patients with a history of suicidal ideation or behaviors (2,078 per 100,000 person-years). These findings indicate the need for interventions to mitigate suicide risk after discharge.

Multiple factors may contribute to suicide risk in the period after hospitalization, including problems with patient engagement and fragmented care in the postdischarge period (Brenner and Barnes, 2012; Riblet et al., 2017a). Accordingly, several brief interventions have been developed to mitigate postdischarge suicide risk by addressing patient engagement and continuity of care after discharge (Department of Veterans Affairs et al., 2013; Mann et al., 2005). Available meta-analyses of published literature evaluating these brief interventions differ in their conclusions regarding efficacy to decrease death by suicide. Milner et al. (2015) broadly defined brief interventions to include efforts that ranged from one-time letters and postcards to regularly scheduled telephone calls. Their pooled analysis of five randomized controlled trials (RCTs) indicated that these interventions did not significantly reduce death by suicide (odds ratio [OR], 0.58; 95% confidence interval [CI], 0.24–1.38). Notably, a single study by Fleischmann et al. (2008) favored treatment in reducing death by suicide (OR, 0.10; 95% CI, 0.02–0.45). The study of Fleischmann et al. randomized 1699 patients to the World Health Organization Brief Intervention and Contact Program (WHO BIC) versus treatment as usual, and evaluated a more intense intervention that others included in the Milner et al. (2015) review. The WHO BIC targeted patients being discharged after a suicide attempt with both an individualized educational intervention about suicide risk and regularly scheduled telephone or in-person contacts over the subsequent 18 months. A more recent meta-analysis by Riblet et al. (2017b) similarly found that WHO BIC was uniquely efficacious among brief interventions to prevent suicide. The publication of two additional RCTs of WHO-BIC strengthened the initial finding (pooled N = 2028; OR, 0.20; 95% CI, 0.09–0.42). Thus, available evidence indicates that participation in WHO BIC reduces risk of suicide after discharge following a suicide attempt. The WHO BIC may help to reduce suicide risk by addressing poor patient engagement and fragmented care.

There are three important gaps in the literature regarding WHO BIC that should be addressed before applying the intervention broadly to the high-risk population of patients being discharged from inpatient psychiatric settings in the United States. First, although evidence consistently indicates that all patients who are being discharged from inpatient psychiatric settings are at elevated risk for suicide (Chung et al., 2017), WHO BIC has only been studied in patients who presented after a suicide attempt. Thus, it is not known whether WHO BIC should be part of routine discharge care or reserved for those hospitalized due to a suicide attempt. Second, it is not clear whether the WHO BIC has any impact on factors related to suicide risk in the postdischarge period such as hopelessness and treatment engagement. Thus, it is difficult to perform small tests of change that would be required to adapt WHO BIC to new settings because clinical teams would only have the rare outcome of death by suicide with which to gauge their implementation efforts. Third, the WHO BIC has only been studied in low- and middle-income settings, and the generalizability of the WHO BIC to high-income settings is unknown. It is conceivable that patients in high-income settings may respond differently to the WHO BIC. For example, individuals living in high-income settings may have access to different and more mental health resources than individuals residing in low- and middle-income countries. Thus, there is a need to determine whether the available research evidence regarding WHO BIC is relevant in high-income settings and to develop standardized materials to implement WHO BIC in these settings.

To address these concerns, we conducted a pilot study of a manual-based intervention called the Veterans Affairs Brief Intervention and Contact Program (VA BIC). The VA BIC is adapted from the WHO BIC and designed to addresses the unique needs of veterans receiving inpatient mental health treatment within the Veterans Health Administration (VHA), a high-resource setting. We chose to study the intervention in veterans who access care at a Veterans Affairs facility (VA users) for several important reasons. First, veterans account for approximately 14% of all suicide deaths in the United States each year (Department of Veterans Affairs et al., 2018). Although the VHA has allocated a considerable amount of resources to prevent death by suicide in VA users, suicide rates remains high in VA users (Department of Veterans Affairs et al., 2018). In addition, suicide rates are elevated in VA users in the first 3 months after psychiatric hospitalization (Britton et al., 2017; Valenstein et al. 2009). Thererefore, VA users may benefit from postdischarge interventions such as the WHO BIC (Department of Veterans Affairs et al., 2013). Second, our prior work found that the WHO BIC may address important systematic vulnerabilities in VA psychiatric discharge processes and align with hospitalized veterans’ preferences for treatment (Riblet et al., 2017a; Riblet et al., 2019).

Our primary aim was to assess the feasibility and acceptability of VA BIC in psychiatrically hospitalized patients. In addition, we sought to gather preliminary data on the effect of the VA BIC on a variety of measures related to postdischarge suicide risk including suicidal ideation, hopelessness, perceived connectedness and support, and treatment engagement. By doing this initial work, our goal is to develop a framework for further evaluation and possible implementation of WHO-BIC in high-resource settings such as the US VA healthcare system.

METHODS

We carried out a pilot study of the VA BIC in patients who were hospitalized on an inpatient mental health unit at a VA Medical Center between May and October 2018. Patients were eligible if they were psychiatrically hospitalized, admitted due to a risk of self-harm, deemed clinically stable by the inpatient team, and nearing discharge. We included patients regardless of whether the admission followed a suicide attempt. We excluded patients who did not have the cognitive ability to provide consent. The Veteran’s Institutional Review Board of Northern New England (VINNE) and the Research and Development Committee, White River Junction Veterans Affairs Medical Center (WRJ VAMC) approved the project after full committee review.

After providing written consent, enrolled patients completed a baseline assessment before discharge. We administered the MINI International Neuropsychiatric Interview (Sheehan et al., 1998) to determine psychiatric diagnoses. We abstracted sociodemographic data from the electronic medical record (EMR). In addition to standard VA discharge care, all patients received the VA BIC intervention. As part of the VA BIC, all patients participated in a brief educational intervention around the time of discharge and contact visits with the intervention staff at 2 days, 2 weeks, 4 weeks, 6 weeks, 8 weeks, and 12 weeks after discharge. The contact visits occurred over the phone or in-person (at the medical facility) depending on patient preference. Patients were compensated for participation.

Manual Development

In developing the VA BIC manual, we sought feedback from veterans who were hospitalized on a VA inpatient mental health unit as well as from VA psychiatrists, VA suicide prevention coordinators, and a WHO BIC methodologist (Dr. Wasserman). Similar to the WHO BIC, the VA BIC sessions are highly interactive and aim to improve treatment engagement, perceived sense of support, and use of suicide prevention resources after psychiatric hospitalization. The VA BIC includes 1) a brief (1 hour) educational intervention on suicide prevention that is personalized to each patient and delivered around the time of discharge, and 2) contact visits after discharge to provide support, monitor symptoms, assess treatment adherence, review the safety plan, and assist with engaging in care. The VA BIC incorporates aspects of motivational interviewing (MI) because MI has been shown to help patients adopt healthier behaviors and achieve better health outcomes (Frost et al., 2018).

Although the VA BIC follows the design of the WHO BIC, there are a few aspects of the VA BIC that are distinct from WHO BIC (see Table 1). Unlike WHO BIC, the VA BIC is delivered over 3 months. We made this decision because it is important to develop effective, brief suicide prevention interventions that target the highest risk period after hospitalization, namely, the first 3 months after discharge (Chung et al., 2017). We designed VA BIC so that it can be delivered by various types of mental health providers including social workers to increase its applicability to various settings. Finally, we tailored the manual to address veteran-specific risk factors for suicide and to provide veterans with a menu of resources that are available to them through the VA and the local community.

TABLE 1.

Comparing and Contrasting WHO BIC and VA BIC Program Components

Program Elements	WHO BIC Intervention	VA-BIC Intervention
Population of interest	Admitted to an emergency department after a suicide attempt	Admitted to an inpatient mental health unit due to suicidal ideation or suicide attempt
Brief education on suicide prevention before discharge	Yes	WHO BIC educational content plus:
		1. Review and discussion of suicide risk factors unique to Veterans  2. Review and discussion of safety plan
Resource guide for the patient to use to supporting their health in the postdischarge period	Yes	Includes Veteran-specific resources to support recovery after discharge:
		1. Services available at Veterans' local VA facility as well as in their local community  2. Materials on free mobile apps that are designed to help with symptom management, made for Veterans and available at the VA app store
Use of motivational interviewing techniques	Yes	Yes
Regular contacts with the intervention staff member after discharge	Yes	Yes
Content covered in regular contact visits	Monitor symptoms, provide support, assess treatment adherence, and assist with engaging in care	Monitor symptoms, provide support, assess treatment adherence, review the safety plan, and assist with engaging in care
Mode of delivery of regular contact visits	Over the phone or in-person depending on preference of local sites	Over the phone or in-person depending on patient preference
Duration of regular contact visits	18 mo	3 mo
Description of intervention staff member	Psychologist, psychiatrist, or nurse	Psychologist, psychiatrist, nurse, or social worker

Outcomes

Our primary aim was to assess feasibility and acceptability. Our secondary aim was to gather pilot data on the impact of VA BIC on related measures of postdischarge suicide risk including suicidal ideation and attempts, hopelessness, and perceived connectedness. In addition, we assessed treatment engagement and perceived support. We collected measures at baseline, 1-month, and 3-month follow-up.

Feasibility and Acceptability

We assessed feasibility by tracking study recruitment and retention. We assessed acceptability by administering an adapted version of the Client Satisfaction Questionnaire 8 (CSQ-8) at study completion (Larsen et al., 1979). This generic, eight-item scale assesses satisfaction with a service and likelihood of recommending the service to others. Total scores range from 8 to 32 with higher scores suggesting greater program satisfaction. We added three open-ended questions regarding VA BIC.

Measures of Suicidal Ideation and Attempts, Hopelessness, and Perceived Connectedness

We measured suicidal ideation using the Beck Scale for Suicidal Ideation (BSS) (Beck et al., 1979; Beck and Steer, 1991). The BSS is sensitive to clinical change over time and may predict suicide risk. BSS scores range from 0 to 38, with higher scores indicating more severe suicidal ideation. We also administered the research version of the Columbia Suicide Severity Rating Scale (C-SSRS) as a supporting measure of suicidal ideation and attempts. The C-SSRS is widely used and a valid measure of suicidal ideation and attempts (Posner et al., 2011). We assessed the severity of ideation using the five-item suicidal ideation severity scale (range, 0–5). A score of 4 or 5 was considered severe ideation (Nilsson et al., 2013). We assessed for suicide attempts using the suicidal behavior scale. In addition, we used the C-SSRS to determine whether or not patients had a history of suicide attempts and whether they made any new fatal or nonfatal attempts after discharge.

We assessed the degree of hopelessness using the Beck Hopelessness Scale (BHS) (Beck and Steer, 1988). The BHS has good reliability and validity and is sensitive to change (Brown, n.d.). Hopelessness also spans across psychiatric illnesses and is associated with increased suicide risk (Ribeiro et al., 2018). BHS scores range from 0 to 20. Scores of 4 to 8 were evidence of mild hopelessness, scores of 9 to 14 moderate hopelessness, and scores greater than 14 severe hopelessness (Beck and Steer, 1988).

We assessed perceived connectedness using the Interpersonal Needs Questionnaire-15 (INQ-15). The INQ-15 is a valid measure of connectedness, and changes in scores are associated with increased suicide risk (Gutierrez et al., 2016, Van Orden et al., 2012). The INQ-15 is based on the interpersonal theory of suicidal behavior (Van Orden et al., 2010). We assessed for changes in perceived connectedness using the six-item perceived burdensomeness (PB) subscale because this subscale is thought to serve as a proxy measure of perceived connectedness (Van Orden et al., 2012). Positive questions are reverse coded and then a total score is generated by summing all questions together. Scores range from 6 to 42 with lower scores indicating greater perceived connectedness.

Measures of Activation and Engagement

There is no agreed-upon measure that incorporates all aspects of the experience of “patient engagement” in treatment and use of treatment resources (Barello et al., 2014). In mental health, suggested domains of “patient engagement” include being activated and engaged in care, feeling supported, and adopting behaviors that support well-being (Barello et al., 2014). Therefore, we measured patient engagement by evaluating patient activation as well as the adoption of behaviors to support well-being.

We measured patient activation and engagement using the validated Partners in Health (PIH) scale that asks patients to self-report on their perceived level of engagement in treatment (Smith et al., 2017). For example, patients self-report on the degree to which they adhere with medications as well as monitor symptoms and take action when their symptoms worsen. The PIH has been studied in patients with chronic illness, although not specifically in patients with mental health disorders or suicidal behavior (Peñarrieta-de Córdova et al., 2014; Petkov et al., 2010). PIH scores range from 0 to 96, and higher scores are indicative of greater patient activation.

We measured whether patients adopted behaviors to support their well-being by evaluating continuity of care. In a study of veterans who were treated in inpatient and outpatient mental health settings, Greenberg and Rosenheck (2005) demonstrated that improvements in three measures of continuity of care were associated with improved mental health outcomes. These included intensity of treatment (low, medium, or high), regularity of care (number of months of continuous outpatient mental health treatment), and continuity of treatment across organizational boundaries (an outpatient visit within 30 days of discharge). We applied these same measures to our study population and abstracted the data from the EMR. Unlike Greenberg and Rosenheck (2005), we defined high intensity treatment as 12 or more mental health visits within 3 months of discharge, medium intensity as 6 to 11 mental health visits, and low intensity as 0 to 3 mental health visits. We used more stringent criteria to define high intensity care because patients with suicidal behavior may benefit from more intensive treatment in the first 3 months after discharge (Chung et al., 2017).

Measure of Support

We assessed perceived social support using the nine-item thwarted belongingness (TB) subscale from the INQ (Van Orden et al., 2012). The TB subscale is thought to serve as a proxy for perceived social support (Van Orden et al., 2012). Positive questions are first reverse coded, and then a total score is generated by summing all questions together. Scores range from 9 to 63 with lower scores indicating greater perceived social support.

Analysis

We examined changes in suicidal ideation and other outcome measures before and after exposure to the VA BIC program using repeated measures analysis of variance (ANOVA) to assess for changes in continuous variables from baseline to 1 and 3 months after discharge. This test assumes sphericity and is at risk for type 1 error if there is unequal variance and the data are correlated (Lane, 2016). To account for this concern, we conducted Bartlett’s test of sphericity on each measure. We considered that sphericity was violated if p < 0.05. For any measure that violated sphericity, we applied a recommended Box’s conservative epsilon correction factor (Garson, 2012; Lane, 2016).

Because we had multiple comparisons, we applied the Holm-Bonferroni sequential method to our analysis (Aickin and Gensler, 1996). First, we calculated a p-value for each comparison and then ranked their respective p-values by size (smallest to largest). Second, we used the Holm-Bonferroni formula to calculate an adjusted p-value for each ranked comparison. Third, we considered the result to be statistically significant if the original p-value for the ranked comparison was less than the adjusted p-value. We chose the Holm-Bonferroni method over a simple Bonferroni correction because it is considered to be a more powerful approach in dealing with potential type 1 error in the setting of multiple comparisons (Aickin and Gensler, 1996).

If the repeated measures ANOVA suggested a significant difference in means across time, we performed a post hoc, pair-wise comparison of means at baseline and 1- and 3-month follow-up. First, we used the Shapiro-Wilk test to determine whether each of the sample means was normally distributed. Second, if normally distributed, we applied a paired t-test and generated Cohen’s d using Pearson’s r and standard deviations (Morris, 2008). Otherwise, we applied a Wilcoxon signed-rank test and calculated an effect size r using the z statistic (Fritz et al., 2012). We performed analyses using Stata 14 (Stata Corp, College Station, TX).

RESULTS

As shown in Figure 1, approximately 70% (n = 9) of eligible patients (n = 13) enrolled in the pilot study and all subjects (n = 9) completed all portions of the study. Among the nine patients, there were a total of 54 contacts visits with the VA BIC intervention staff member. Half of these visits were delivered in-person and the other half over the phone. Although one patient completed 100% of visits in person and another patient completed 100% of visits on the phone, most preferred a mix, with in-person visits often occurring while the person was at the VA for other appointments. Table 2 summarizes the baseline characteristics of the nine patients enrolled in VA BIC. The majority were men, and the average age was 43.4 years (SD, 14.31). All patients had a lifetime history of severe suicidal ideation and over half had a history of a suicide attempt. For two patients, the current admission was precipitated by a suicide attempt.

FIGURE 1.

Flow of VA BIC patient selection process.

TABLE 2.

Baseline Characteristics of VA BIC Cohort

	VA-BIC Cohort
Total number, % (n)	100.0 (9)
Gender, male % (n)	89.0 (8)
Mean age (SD), y	43.4 (14.31)
Lifetime history of suicide attempt, % (n)	66.7 (6)
Lifetime suicidal ideation, C-SSRS mean (SD)	5 (0)
MINI International Neuropsychiatric	88.9 (6)
Diagnostic Interview
Major depressive disorder–current (past 2 weeks),a % (n)	88.9(6)
Major depressive disorder–past, % (n)	88.9 (8)
Major depressive disorder–recurrent, % (n)	88.9 (8)
Posttraumatic stress disorder–current (past month), % (n)	44.4 (4)
Panic disorder–current (past month),a % (n)	22.0 (2)
Panic disorder–lifetime, % (n)	11.1 (1)
Generalized anxiety disorder–current (past 6 months), % (n)	22.2 (2)
Social anxiety disorder–current (past month), % (n)	33.3 (3)
Alcohol use disorder–past 12 months, % (n)	33.3 (3)
Substance use disorder–past 12 months, % (n)	33.3 (3)

^aIn two patients, a drug cause could not be ruled out.

Feasibility and Acceptability

Our study recruitment was acceptable at 70% over the course of approximately 3 months, and our retention was excellent at 100%. At study completion, patients reported high satisfaction with the intervention (CSQ-8 mean, 31; SD, 1.2). One patient described, “I think the study was helpful in keeping me connected, as well as providing another support avenue to my repertoire. I was excited to see the change in my overall outlook over the course of a few months. It reinforced the value of structure and purpose in my life…. It was personal, but not too personal.”

Suicidal Ideation and Attempts, Hopelessness, and Perceived Connectedness

Patients experienced a significant reduction in suicidal ideation in the first 3 months after discharge (Table 3). BSS scores fell by approximately 8 points between baseline and 3-month follow-up (mean difference [MD], 8.1; standard deviation [SD], 5.25; p < 0.002). The mean BSS and C-SSRS scores also differed significantly across time (BSS, F(2,16) = 14.96, p [with box correction] < 0.01; C-SSRS, F(2,16) = 12.04, p [with box correction] < 0.02). There were no suicide attempts after discharge.

TABLE 3.

Impact of VA BIC on Symptoms and Engagement at Baseline, and 1 and 3 Months After Discharge

		Baseline (0 M)		Postintervention (1 mo)		Postintervention (3 mo)		ANOVA	Correlation	Effect Size
Effect Measure	n	Mean	SD	Mean	SD	Mean	SD	p	Pearson’s r	Cohen’s d
C-SSRSa	9	3.78	(1.99)	0.89	(1.17)	0.89	(1.27)	0.01	2.7 (z statistic)	0.9 (effect size r)
BSS	9	16.11	(7.85)	9.33	(7.16)	8.00	(5.17)	<0.01	0.7	1.3
BHS	9	11.89	(6.37)	6.00	(3.71)	7.00	(6.10)	0.04	0.5	0.8
INQ-15 (PB)	9	23.40	(7.37)	15.00	(6.04)	12.22	(5.12)	0.01	0.4	1.4
INQ-15 (TB)	9	42.67	(11.37)	33.33	(14.86)	32.56	(8.82)	0.02	0.1	0.7
PIH	9	61.44	(19.05)	69.22	(14.25)	69.44	(12.04)	>0.1	0.4	0.4

^aAnalysis of mean difference based on Wilcoxon signed-rank test.

Patients reported a significant reduction in the degree of hopelessness as measured with the BHS in the first 3 months after discharge. On average, patients described moderate symptoms of hopelessness at baseline. However, on average, patients reported mild symptoms of hopelessness during the follow-up period. A repeated measures ANOVA with box correction also indicated that there was a significant difference in mean BHS scores over time (BHS, F(2,16) = 5.88, p < 0.05).

Finally, patients experienced a greater sense of connectedness at follow-up as compared with baseline as measured by the PB subscale of the INQ-15. The mean difference in scores between baseline and 3-month follow-up was 11.22 points (SD, 8.42). A repeated measures ANOVA with Box correction indicated that scores differed significantly across time (INQ-15 PB, F(2,16) = 10.97, p < 0.013). A post hoc analysis suggested that the subscale scores decreased significantly at 1- and 3-month follow-up (p = 0.01 and p < 0.01, respectively).

Activation and Engagement

Patients reported improved activation as measured by the PIH scale. However, the mean difference in PIH scores between baseline and 3-month follow-up was modest (MD, 8; SD, 13.3), and the changes were not statistically significant. Yet, five patients relied upon the VA BIC interventionist to facilitate communication of unmet needs or concerns to their outpatient mental health providers during the course of the study. Patients requested assistance when they were feeling disconnected from their providers.

All patients received outpatient mental health treatment within 1 month of discharge. In general, patients were also received intensive and highly regular care. In fact, most patients received high intensity mental health treatment after discharge (66.7%; n = 6). The remaining three patients (33.3%) received medium intensity treatment. Except for one patient who received continuous outpatient mental health treatment only for the first 2 months after discharge, all patients were exposed to 3 months of continuous outpatient treatment.

Support

As shown in Table 3, patients reported significant improvements in perceived social support from loved ones or friends as measured by the TB subscale of the INQ-15. The mean difference in scores between baseline and 3-month follow-up scores was 10.11 (SD, 7.93). A repeated measures ANOVA indicated that scores differed significantly across time (INQ-15 TB, F(2,16) = 4.77, p < 0.03). A post hoc analysis suggested that the subscale scores decreased significantly at 1- and 3-month follow-up (p = 0.04, p = 0.01, respectively).

DISCUSSION

The VA BIC is a brief intervention designed to reduce suicide risk after psychiatric hospitalization in a high-resource, first world setting. Our pilot study found preliminary evidence that the VA BIC may be acceptable to patients and address their need for closer follow-up in the first few months after discharge and may assist patients with engaging in treatment in the postdischarge period. The VA BIC was associated with reduced suicidal ideation and hopelessness as well as improved connectedness social support in a broader population of psychiatrically hospitalized patients including those who have not made a suicide attempt just before admission. Our findings are modest, and although these improvements may translate to a reduction in suicide risk, postdischarge larger RCTs are required to confirm our results. Although VA BIC may increase patient activation and improve continuity of care, we were unable to demonstrate statistical significance in this small sample. Overall, our findings are at best preliminary but build the case for further research.

There are several strengths to our pilot work. This is the first and only study of an adapted version of the successful WHO BIC program in the United States (and in veterans). We were able to demonstrate that it is feasible to enroll patients in a study of an adapted version of WHO BIC and to retain these patients in the study. We were able to recruit and retain in our study a largely male population. Men are typically less likely to engage in mental health treatment (Cleary, 2017; Hom et al., 2015). We also demonstrated improvements in outcomes across a wider population of hospitalized patients including those without a recent suicide attempt. The overall satisfaction with the intervention was high.

Several differences between the VA BIC and the WHO BIC are important to interpreting these findings. First, unlike the WHO BIC trials, we did not require that patients had made a recent suicide attempt. This decision reflects evidence from the literature suggesting that suicide risk in the period after hospitalization spans across psychiatric disorders and is not limited to those with a recent history of a suicide attempt (Chung et al., 2017). Second, in the WHO BIC trials, there was some variation across sites with regards to treatment setting before discharge home (Amadéo et al., 2015; Fleischmann et al., 2008; Mousavi et al., 2014). In some cases, the enrolled patients received treatment in the emergency department and were then discharged home. In other cases, the enrolled patients received treatment on an inpatient mental health unit before discharge. For example, in a trial of WHO BIC in French Polynesia, Amadéo et al. (2015) reported that all patients received a short psychiatric hospitalization.

In our study, we adapted the WHO BIC intervention to meet the needs of patients treated in the VA healthcare system. Because our study population was recruited from the inpatient mental health unit setting, our patients represent a subset of those included in the WHO BIC trials. We also required that patients had been deemed clinically stable by the inpatient team. This was to ensure that patients could reasonably participate in the consenting process and safely complete study procedures on the inpatient unit. Because the VA BIC targets patients’ decision making around postdischarge care, we felt it was reasonable to enroll patients at a time when these types of decisions typically occur, namely, when patients are clinically stable. Thus our population may not be representative of all the patients included in the WHO BIC trials. For example, there may be differences between the study populations with regards to treatments received, the training and skills of providers involved in delivering care in these settings, the severity and types of mental illness, the prevalence of comorbidities, and the duration between the onset of suicidal behavior and the receipt of the intervention. However, because a large proportion of the VA BIC intervention occurs in the postdischarge period, it is conceivable that the needs and experiences of patients in the outpatient setting may be quite similar regardless of whether they received the educational component of the intervention before emergency department discharge or inpatient mental health discharge.

The WHO BIC trials were conducted in low- and middle-income countries where mental health resources are known to be scarce. In these settings, low intensity suicide prevention strategies may demonstrate larger effect sizes in the context of a randomized trial. Conversely, in a high resource setting such as the VA healthcare system, a brief suicide prevention strategy may exert a much smaller effect relative to a more complex intervention. Yet, an intervention such as the WHO BIC may still have relevance in a high resource environment such as the VA. Even with the relative intensity of standard care in the VA, suicide rates remain high in VA users (Department of Veterans Affairs et al., 2018). Our prior work also found that despite the implementation of high resource interventions in the VA healthcare system, patients still experience problems with follow-up care and engagement in treatment (Riblet et al., 2017a; Riblet et al., 2019). Therefore, a structured approach such as the VA BIC may represent a simple way to coordinate care, promote engagement, and augment the complex suite of interventions that are currently implemented by the VA. Notably, in this current pilot study, patients who received the intervention also received standard VA discharge care.

Our study is also subject to a number of other important limitations. First, this pilot study was primarily designed to assess feasibility and acceptability of the intervention. We enrolled a small number of patients. This, of course, poses important limitations with regards to interpreting the generalizability of the changes that we observed in our outcome measures after discharge. We were unable to evaluate the primary outcome of interest from the WHO BIC trials, death by suicide. Furthermore, because our study relied on referrals from the inpatient psychiatry team and did not collect data on patients who declined enrollment, we cannot state with absolute certainty that VA BIC is feasible or acceptable in all VA users who are psychiatrically hospitalized. Yet, it is somewhat reassuring that of the 13 eligible patients who were referred to our study by the inpatient psychiatry team, approximately 70% (n = 9) enrolled and all nine patients completed all components of the study. Second, because our study used a pre-post design and did not include a concurrent control group, we are unable to determine whether the improvements that we observed in our study population exceeded those of patients receiving usual care alone. However, it is notable that our study found significant changes in three known proxy measures of suicide risk including suicidal ideation, hopelessness, and perceived connectedness. The effect sizes were also large. For example, mean BSS scores decreased by approximately 8 points in the postdischarge period. Prior studies have indicated that a decrease of 5 points or greater on the BSS is clinically relevant (Beck et al., 1979). Our findings also align with that of Mousavi et al. (2014) who reported significant improvements in suicidal ideation in a study of WHO BIC. Third, our findings may not be generalizable to other populations as veterans are a unique population with distinctive risk factors for suicide. Finally, although we used validated instruments to assess for treatment effect and the majority of assessments were self-reported, the study assessments were administered by the VA BIC interventionist, and thus were not blinded.

The VA BIC is designed to encourage treatment engagement and provide patients with necessary support in the postdischarge period could be instrumental in helping to reduce suicide risk after a psychiatric hospitalization. Evidence suggests that problems with patient engagement and continuity of care after discharge may serve as critical factors in contributing to suicide risk after psychiatric hospitalization (Brenner and Barnes, 2012; Riblet et al., 2017a). Because the VA BIC educates patients about suicide prevention and the role of follow-up care in suicide prevention, this may encourage treatment engagement after discharge. For example, Kuramoto-Crawford et al. (2017) evaluated over 8000 adults with a history of severe suicidal ideation using a cross-sectional survey design and observed that the majority of adults reported that they did not feel a need for mental health treatment. Similarly, Riblet et al. (2019) found in a qualitative study of veterans being discharged from an inpatient mental health unit that many patients believed that they were at low risk for suicidal behavior after discharge, although the evidence is clear that the postdischarge period is a high-risk period for suicide. The VA BIC also includes a follow-up component that provides patients with support during a high-risk transition period. It is well-known that patients with a history of suicidal behavior may face significant challenges in engaging in mental health treatment (Krulee and Hales, 1988, Kurz and Moller, 1984). For example, Dixon et al. (2016) suggested that patients may be more apt to disengage from mental health treatment and their providers if they have a sense that they are not being supported by their treater. Higher treatment dropout rates have been shown to be associated with poor therapeutic alliance (Barrett et al., 2008), and a poor therapeutic alliance may increase the risk for suicide (Dunster-Page et al., 2017).

Our pilot study generated preliminary data on several measures associated with suicide risk in the postdischarge period. Specifically, we found that patients exposed to VA BIC experienced significant reductions in suicidal ideation and hopelessness after discharge as well as a significant improvement in perceived connectedness and social support. Although we are unable to determine if these changes are, in fact, the direct result of the VA BIC intervention, the trend of the findings align with the intent of the design of the VA BIC intervention. First, because the VA BIC educates patients about suicide risk, uses MI techniques, and maintains ongoing contact with patients after discharge, the VA BIC may help to ensure that patients are more engaged in outpatient treatment. This, in turn, may help to facilitate that patients are treated for their underlying symptoms. In our analysis, we did find evidence to suggest that patients not only experienced notable reductions in suicidal ideation and hopelessness, but were also actively engaged in treatment. Although we found that patient activation only showed trends toward improvement, the majority of patients did receive high intensity mental health treatment based on the number of appointments. This equated to receiving more than four mental health visits per month in the first 3 months after discharge. Second, the VA BIC provides patients with ongoing contact after discharge. These regular contacts may help to promote in patients a greater sense of perceived connectedness and support. Based on available evidence, improved connectedness may be protective against suicide risk (Gutierrez et al., 2016; Kleiman and Liu, 2013). In this study, we found that patients experienced significant improvements in perceived connectedness and social support after discharge.

CONCLUSIONS

The VA BIC may be a feasible and acceptable intervention designed to decrease suicide risk after psychiatric hospitalization. VA BIC may apply to psychiatrically hospitalized patients regardless of whether they have made a recent suicide attempt. Although our pilot study suggests that VA BIC may improve several measures of suicide risk (including suicidal ideation, hopelessness, and perceived connectedness) in the period after a psychiatric hospitalization, our conclusions are preliminary and require confirmation due to small sample sizes and a lack of a control group. Yet, our findings are noteworthy because an adapted version of the WHO BIC has never been studied in the United States. It will be important to conduct a larger trial of VA BIC to determine whether the intervention may be an efficacious strategy to prevent death by suicide in the period after psychiatric hospitalization. It will also be important to study VA BIC in a larger scale not only in the US veteran population, but also in other nonveteran populations.