Healthcare processes contributing to suicide risk in veterans during and after residential substance abuse treatment.

Abstract

Objective:

Substance use disorders (SUD) are an important risk factor for suicide. While residential drug treatment programs improve clinical outcomes for SUD, less is known about the role of related healthcare processes in contributing to suicide risk. These data may help to inform strategies to prevent suicide during and after residential treatment.

Methods:

A retrospective analysis was conducted on root-cause analysis (RCA) reports of suicide in veterans occurring within 3 months of discharge from a residential drug treatment program that were reported to a Veterans Affairs (VA) facility between 2001 and 2017. Demographic information such as age, gender, and psychiatric comorbidity were abstracted from each report. In addition, an established code book was used to code root causes from each report. Root causes were grouped into categories in order to characterize the key system and organizational-level processes that may have contributed to the suicide.

Results:

A total of 39 RCA reports of suicide occurring within 3 months after discharge from a residential drug treatment program were identified. The majority of decedents were men and the average age was 42.9 years (SD = 11.2). The most common method of suicide was overdose (33%) followed by hanging (28%). Most suicides occurred in close proximity to discharge, with 56% (n = 22) occurring within seven days of discharge and 36% (n = 14) occurring within 48 hours of discharge. The most common substances used by decedents prior to admission were alcohol or opiates. RCA teams identified a total of 140 root causes and the majority were due to problems with suicide risk assessment (n = 32, 22.9%). Non-engagement with treatment during (n = 20, 14.3%) and after the residential stay (n = 18, 12.9%) was also highlighted as an important concern. Finally, several reports raised concerns that a discharge prior to treatment completion or a precipitous discharge due to program violation negatively impacted treatment outcomes.

Conclusions:

Efforts to prevent suicide in the period following discharge from a residential drug treatment program should focus on addressing suicide risk factors during admission and helping patients engage more fully in SUD treatment.

Introduction

Substance use disorders (SUD) are of considerable concern in the United States (US). Approximately 8% of US adults ages 18 years and older (20.2 million people) were diagnosed with a SUD in 2014 (Lipari & Van Horn, 2017). Individuals with a SUD experience numerous adverse health consequences (Lipari & Van Horn, 2017). Importantly, studies have found that individuals with a SUD are at elevated risk for suicide (Ashrafioun, Bishop, Conner, & Pigeon, 2017; Bohnert, Ilgen, Louzon, McCarthy, & Katz, 2017; Flensborg-Madsen et al., 2009; Wilcox, Conner, & Caine, 2004; Yuodelis-Flores & Ries, 2015). Using the 2014 National Survey of Drug Use and Health, Ashrafioun et al. (2017) demonstrated in over 40,000 participants that prescription opioid misuse was associated with a 60% increased rate of suicidal ideation and a 200% increased rate of suicide attempts. Similarly, Bohnert et al. (2017) found among 4.8 million patients treated in the Veterans Health Administration (VHA), men with a current diagnosis of any SUD were at nearly two times greater risk for suicide, even after adjusting for age and medical and psychiatric comorbidity. These findings highlight the need to identify strategies to alleviate suicide risk in this population.

A number of recovery-oriented approaches are used to manage SUD (Mee-Lee, 2001; National Institute on Drug Abuse, 2018; The American Society of Addiction Medicine, 2015). Based on illness severity and the need for stable housing or a 24-hour recovery environment, a subset of patients with SUD are cared for in a residential setting. According to the American Society of Addiction Medicine Patient Placement Criteria-2R (ASAM PPC-2R), Level III residential drug treatment programs provide patients with a stable living situation while they are working on skills to achieve sobriety (Mee-Lee, 2001; ASAM, 2015). Patients are medically stable and do not require more intensive treatments that are available in the inpatient setting. Staff members are trained in addiction as well as other mental health specialties such as mood disorders. Patients participate in various interventions to support their recovery. Length of stay can vary from as short as 28 days to as long as 12 months (Harris, Kivlahan, Barnett, & Finney, 2012).

Some studies have suggested that the period following discharge from a residential drug treatment program may represent a period of heightened suicide risk similar to psychiatric hospitalization (Bohnert et al., 2017; Britton & Conner, 2010; Chung et al., 2017; Darke et al., 2007). In a national survey of general residential drug treatment programs in Australia, Ross et al. (2012) observed that one-third of programs had no policy for assessing and managing suicide risk and programs did not routinely collect information about psychiatric comorbidity to support the management of suicide risk. Moreover, Decker et al. (2017) suggested that breakdowns in care including failure to complete treatment in the residential treatment program may contribute to overall mortality risk after discharge. No study has evaluated how system-level processes may contribute to the risk for suicide in the period following discharge from a residential treatment program.

To understand how healthcare processes related to residential drug treatment programs may contribute to suicide risk, we conducted a retrospective study of root-cause analysis (RCA) reports of suicide occurring within three months of discharge from a residential drug treatment program. RCA is a well-known and robust approach to elucidating the contribution of systems and organizational processes to adverse events such as suicide after hospital discharge (Bagian et al., 2001). Our current study will help to inform strategies to improve healthcare processes in the residential drug treatment setting in order to reduce suicide risk after discharge.

Methods

Study Design

We conducted a retrospective analysis of all RCA reports of suicide in veterans occurring within three months of discharge from a residential drug treatment program that was reported to any one of the roughly 140 Department of Veterans Affairs (VA) Medical Centers between October 1, 2001 and December 31, 2017. Because the highest risk period for suicide after psychiatric hospitalization occurs within the first three months after discharge (Chung et al., 2017), patients leaving residential drug treatment program may experience similar risks, especially if they have recently relapsed (Yuodelis-Flores & Ries, 2015). As VA purchased care in the community is a growing part of VA service, we included RCA reports pertaining to deaths by suicide associated with both VA and non-VA residential drug treatment programs.

The Research and Development Committee, White River Junction, VA Medical Center and the Veteran’s Institutional Review Board of Northern New England (VINNE) approved this project.

Data Sources

We identified cases using the VA National Center for Patient Safety’s (NCPS) RCA database which contains RCA reports of adverse events related to the care of patients who have received some portion of their care at any VA facility or affiliated program. As described in prior studies (Riblet, Shiner, Mills, et al., 2017; Riblet, Shiner, Watts, et al., 2017), the RCA methodology assumes that adverse events more often result from system vulnerabilities, than the failure of an individual. Therefore, identifying and addressing these vulnerabilities can improve outcomes.

VA facilities employ full-time patient safety managers (PSMs) who manage the RCA process (Bagian et al., 2001). PSMs bring together a multidisciplinary team who uses standardized tools to develop root cause statements which specify the underlying causes of the adverse events (Bagian et al., 2001). It is common that teams may identify more than one root-cause for any given adverse event. The NCPS database tracks the event type and the date of the event. In addition, the database includes a series of free-text fields in which the team describes the event and reports on root causes and contributing factors. Because RCA is confidential and privileged information, the identity of the patient (or involved providers) are not included in the database.

In the case of suicide, the VA requires that an RCA be performed in the event that a suicide occurs within 72 hours after a general hospital discharge and within 7 days after discharge from an inpatient mental health unit (VA National Center for Patient Safety, 2011). It is left to the discretion of local facilities to conduct RCA on suicides that occur outside this window of time.

Data Collection

We identified included cases by searching for reports that referred to suicide in the event type field. We used PolyAnalyst text mining software (Megaputer Intelligence, Inc.) to screen free-text fields for the mention of suicide in any RCA report. We screened for cases that occurred after a discharge from a residential drug treatment program by reviewing free-text fields for reference to established VA programs by name (Sullivan G, Arlinghaus K, Edlund C, & Kauth M, 2011). In the event that the report did not mention the name of the program, we considered the case to meet inclusion criteria if the general program description appeared to align with ASAM Level III criteria (Mee-Lee, 2001; ASAM, 2015). We calculated the number of days between discharge and suicide by directly abstracting these data from the report (if available) or manually calculating the difference based on the date of the event and the date of discharge.

Analysis

We abstracted descriptive patient-level data such as age, gender and psychiatric comorbidity from the reports when available. We used an established coding book to code root causes from each RCA report (Riblet, Shiner, Watts, et al., 2017). Two raters (NR, LK) coded the root causes in each report. These reviewers independently categorized root causes based on our established codebook. The reviewers met regularly during the coding process to review findings. As part of this iterative process, we determined whether there were new or emerging groupings that had not been captured in the original codebook (e.g., precipitous discharge due to program violation). These new categories were then added to our codebook based on consensus between reviewers. Furthermore, in the cases of any non-agreement between reviewers, we discussed the discrepancy and used consensus coding to resolve disagreements.

Results

Between October 1, 2001 and December 31, 2017, we identified a total of 39 RCA reports of suicide events occurring within three months of discharge from a residential drug treatment program. As shown in Table 1, most decedents were male (97%) and the mean age in years was 42.9 years old (SD = 11.2 years). Most decedents (53.8%) had made at least one suicide attempt in the past. The most common substances used by decedents prior to admission were alcohol or opiates. A small number of decedents had reported using cocaine. Many patients (39%) died by suicide at home. In contrast, among the eight patients who eloped from the program (i.e., went missing and staff were unaware of their present location), deaths commonly occurred on the grounds of the program. The most common method of suicide was overdose (33%), followed by hanging (28%). Only 18% of included suicides were secondary to firearm.

Table 1.

Baseline Characteristics of 39 Patients Who Died by Suicide after Residential Drug Treatment Program Discharge, October 1, 2001 – December 31, 2017

	% (n)
Gender, male	97.4 (38)
Mean age in years, SDa	42.9 (11.2)
History of suicide attempt	53.8 (21)
History of psychiatric hospitalization	46.2 (18)
Comorbid Psychiatric Diagnosis,b,c
Mood Disorder	38.5 (15)
Posttraumatic Stress Disorder	20.5 (8)
Other Psychiatric Disorder	12.8 (5)
Drugs of Abuse Contributing to Current Program Admissionc,d
EtOH	41.0 (16)
Opiate	23.1 (9)
Other Drugs of Abuse	10.3 (4)
Method of Suicide
Overdose	33.3 (13)
Hanging	28.2 (11)
Other Methods	20.5 (8)
Firearm	17. 9 (7)
Setting of Suicide
At home	38.5 (15)
Transitional housing or hotel	15.4 (6)
Out in the community	12.8 (5)
Program/facility grounds	10.3 (4)
Unclear	23.1 (9)

Note. IQR = interquartile range; RCA = root cause analysis; SD = standard deviation; EtOH = alcohol.

^a13 RCA reports included no information on the age of deceased

^b12 RCA reports included no information on prior mental health history

^c17 RCA reports did not delineate the specific drugs of abuse contributing to the admission

^dPatients could carry multiple psychiatric diagnoses and/or report multiple drugs of abuse

Table 2 outlines the key characteristics of the decedent’s stay. Most suicides (n = 37) were associated with a discharge from a VA-affiliated residential program. Most patients (53%) had a length of stay that was 30 days or less. The majority of suicides occurred in close proximity to discharge. In fact, 56% (22) of suicides occurred within seven days of discharge and 36% (14) of suicides occurred within 48 hours of discharge.

Table 2.

Characteristics of the Residential Drug Treatment Program Stay of 39 Patients Who Died by Suicide after Residential Drug Treatment Program Discharge, October 1, 2001 – December 31, 2017.

	Results
Length of staya
Median days (IQR)	30 (45.3)
0 – 30 days, % (n)	53.1 (17)
31 – 45 days, % (n)	12.5 (4)
45 – 90 days, % (n)	15.6 (5)
> 90 days, % (n)	18.8 (6)
Program Affiliation
Veterans Affairs, % (n)	94.9 (37)
Non-Veterans Affairs, % (n)	5.1 (2)
Method of Discharge
Discharge Due to Program Violation	23.1 (9)
Elopement	20.5 (8)
Planned Discharge	20.5 (8)
Unplanned Dischargea	12.8 (5)
Discharge Type Unclear	23.1 (9)
Days from discharge to suicide
Median days (IQR)	4 (23)
Death within 48 hours of discharge, % (n)	35.9 (14)
Death less than or equal to 7 days of discharge, % (n)	56.4 (22)

Note. IQR = interquartile range; RCA = root cause analysis.

^a7 RCA Reports did not provide information on length of stay

^bUnplanned discharges include against medical advice, self-discharge as well as irregular discharges.

As described in Table 3, there were a total of 140 root causes that contributed to suicide after discharge. Categories of root causes ranged from problems with suicide risk assessment to non-engagement in treatment. Notably, RCA teams highlighted that some patients did not receive an adequate suicide risk assessment during key transition points including upon admission, prior to discharge, during aftercare visits and in response to concerning behaviors. Furthermore, RCA teams reported that breakdowns in communication between providers may have contributed to some of the suicide cases. This was especially true of communications between program staff and outpatient providers. The RCA team felt that as a result of poor communication, outpatient providers were less equipped to assess the patient’s risks for suicide and address this risk. Interestingly, a few RCA reports described that communication problems between the program staff and other patients enrolled in the program may have been a problem. The RCA team discovered that other patients had witnessed concerning behaviors in the deceased patient, but did not share this information with staff. The RCA team concluded that peers lacked knowledge about warning signs for suicide and did not recognize the importance of seeking help from staff.

Table 3.

Root Causes of Suicides Occurring Within 3 months After Discharge from Residential Drug Treatment Program

	% (n)	Descriptive Example(s)
Total Root Causes	100.0 (140)
Problems with suicide risk assessment (n = 32, 22.9%)
Lack of suicide risk assessment	31.3 (10)	-No suicide risk assessment was conducted prior to discharge; -No suicide risk assessment was conducted in response to concerning behaviors
Precipitous discharge due to violation of program policies	28.1 (9)	-The patient began using drugs during the program stay and therefore, was precipitously discharged.
Discharge plan did not address suicide risk	21.9 (7)	-The patient was given potentially lethal doses of medication at discharge; -The patient was not assigned to case management after discharge when they would have benefited from the intervention.
Other problems related to the management of risk	18.8 (6)	-Staff harbored negative feelings towards patients with substance abuse, preventing them from recognizing and addressing concerning behaviors.
Problems with Communication of Information Relevant to Suicide Risk Assessment and Management (n = 26, 18.6%)
Communication issues between staff and outpatient providers	26.9 (7)	-The primary care provider who met with the patient in the outpatient setting after discharge did not have all relevant information about the patient’s admission. -There were breakdowns in communication between VA staff and non-VA providers.
Issues with methods used to communicate suicide risk and/or implement treatments to address suicide risk	30.8 (8)	-Because discharge orders were incomplete, the post-discharge needs of the patient were not fully realized. -A discharge note was not completed and therefore, information related to suicide risk was not communicated. -Psychological testing results were not made available to staff and this information would have permitted a more comprehensive and complete assessment of suicide risk.
Communication issues between staff	23.1 (6)	-The staff did not effectively communicate with each other about concerning behaviors that was observed by some staff members. -Staff did not recognize that non-VA programs would be unaware of the limited scope of a referring VA facility in providing outpatient follow-up care to a veteran with history of dishonorable discharge
Communication issues between staff and family or peers	19.2 (5)	- Family members held relevant information about the patient’s risk factors for suicide but this information was not communicated to the staff. -Other patients enrolled in the program observed concerning behaviors in the deceased, but did not communicate this information to the staff.
Problems with discharge process (n = 24, 17.1%)
No follow-up care arranged	41.7 (10)	-The patient was not provided with a follow-up care appointment after discharge.
Problems with outreach in the post-discharge period	29.2 (7)	-The clinic did not reach out to the patient when the patient was a no-show to their post-discharge outpatient follow-up appointment.
No policy/procedure to handle unplanned or precipitous discharge due to violation of program policies	29.2 (7)	-Staff were unsure of what (if any) care could be offered to a patient in the event that they had an unplanned discharge (e.g. AMA, irregular discharge). -There was no procedure in place to implement a post-discharge treatment plan to address the care of patients who were thrown out of the program due to using drugs.
Problems with treatment of mental health disorder (n = 20, 14.3%)
Treatment during program did not address suicide risk	75.0 (15)	-Treatment did not include a focus on psychiatric comorbidities and/or other relevant risk factors for suicide such as psychosocial stressors.
Inadequate staffing	25.0 (5)	-Staff burnout resulted in patients receiving more limited care. -There was a lack of program resources available to assist in managing the patient such as lack of access to a social worker to conduct a comprehensive needs assessment and to develop a treatment plan that addressed psychosocial risk factors.
Non-engagement with treatment during residential stay (n = 20, 14.3%)
Patient left the program before completing treatment	55.0 (11)	-The patient left against medical advice from the program and did not complete the set of treatments that were put in place to address their drug use problems.
Patient not engaged in their treatment during admission	45.0 (9)	-The patient lacked full investment in treatment which was evident due to their cutting back on participation in treatment programming. -The patient missed several doses of medication because they did not present for pill administration.
Non-engagement with post discharge treatment (n = 18, 12.9%)
Patient no-show or cancelled follow-up appointment	66.7 (12)	-The patient was a no-show or canceled their scheduled outpatient post-discharge follow-up appointment.
Patient turns down aftercare interventions to mitigate risk	16.7 (3)	-The patient refuses higher level care treatment after discharge from the program and these interventions would have been important to their recovery.
Patient not engaged in developing a discharge plan	16.7 (3)	-The patient did not allow the staff to involve family members in developing a comprehensive discharge plan.

Note. AMA = against medical advice.

Importantly, RCA reports highlighted several aspects of the discharge process that appeared to be problematic. First, no follow-up care was arranged after discharge for some patients. Second, because there was no policy or procedure to address unplanned or precipitous discharges, the treatment team felt ill-prepared to ensure that these patients also received optimal continuity of care after discharge. In addition, there were some RCA reports that mentioned concerns that the typical residential treatment program model at that facility was to focus exclusively on treating the substance use disorder, rather than taking a more holistic approach and ensuring that related risk factors for suicide such as comorbid psychiatric illness are adequately addressed during admission. In fact, one RCA team suggested that the solution might need to be that residential programs should be delivered in a more restrictive setting, akin to inpatient psychiatric treatment.

Finally, the RCA reports highlighted that problems with patient engagement was a concern. Lack of engagement not only related to the patient’s willingness to participate in treatments that might mitigate their risk, but also to involve family members or loved ones in discharge planning. RCA teams surmised that if family or loved ones were more involved the discharge plan would have better addressed outpatient psychosocial stressors. Furthermore, there were concerns that patients were less engaged in aftercare and would no-show or cancel follow-up appointments. Several RCA reports highlighted that a unique concern for residential programs included managing patients who were precipitously discharged because they violated program policies (e.g., using substances during their stay). Some RCA teams wondered whether policies needed to be more “patient-centered” to ensure that these patients received comprehensive care, despite their problematic behaviors. RCA teams raised concerns that patients with unplanned discharge (e.g., against medical advice discharge) did not have adequate treatment in place prior to discharge.

Discussion

We identified 39 RCA reports of suicide occurring in US veterans within three months of discharge from residential substance use treatment programs. Decedents tended to be younger males with a history of suicidal behavior. The most common drugs of use among those who died from suicide were alcohol and opiates. The largest number of root causes pertained to problems with the suicide risk assessment, breakdowns in communication, and problems with discharge process. This was followed by problems with treatment of mental health disorders and non-engagement with treatment during the residential stay and after discharge. Patients were not always assessed for suicide risk and treatments were not necessarily tailored to address risk factors for suicide such as psychiatric illness. In addition, some patients received inadequate treatment because they left the program early due to elopement or an unplanned discharge. Finally, precipitous discharges due to program violations negatively impacted care.

Our study adds to a small, but growing literature base that evaluated the relationship between suicide and SUD and more specifically, the role of residential programs in suicide prevention. Consistent with the literature, many patients in our study had additional risk factors for suicide such as a history of suicide attempts (Agosti & Levin, 2006; Bohnert et al., 2017). Alcohol and opiates were the most frequently cited drugs of abuse and these substances are heavily associated with suicide risk (Bohnert & Ilgen, 2019; Yuodelis-Flores & Ries, 2015). Surprisingly, the method of suicide was more often overdose or hanging as opposed to firearm. Concerns have been raised in the literature that overdose deaths are often misclassified as accidental (Rockett et al., 2015). In our study, we found that the RCA teams had gathered sufficient data from involved parties to conclude with fair certainty that the death by overdose was most likely a suicide and not accidental.

Similar to Ross et al., (2012) we found that RCA teams commonly cited concerns about a lack of suicide risk assessment at critical periods during the course of admission. These problems may be driven, in part, by knowledge deficits in staff as well as a lack of a standardized procedure to ensure that suicide risk is addressed as part of treatment (Ross et al., 2012). Some programs may also encounter resource limitations that preclude staff from providing more comprehensive treatment. In our study, several reports mentioned that due to insufficient staffing (usually social workers) other issues such as psychosocial stressors could not be fully addressed prior to discharge. Fortunately, the Veterans Health Administration (VHA) is actively working to address suicide risk in veterans including mandatory suicide prevention training and enhanced suicide screening and evaluation (VA Deputy Under Secretary for Health for Operations and Management, April 2017; VA DUSHOM, 2018).

Several RCA reports highlighted that because the treatment model focused solely on managing the SUD that contributed to the admission, other important aspects of treatment relevant to suicide prevention were not thoroughly addressed during admission or at time of discharge. A prior study by Schaefer et al. (2005) also raised concerns that staff in residential programs are less likely to ensure that patients receive sufficient continuity of care after discharge. While part of the solution may include improved referral of appropriate patients to dual diagnosis programs (Grella & Stein, 2006), it is also critical that residential programs implement comprehensive treatment approaches that incorporate suicide prevention. Suicide risk is a known concern in patients with SUD, regardless of whether patients suffer from additional mental health comorbidities (Bohnert et al., 2017; Yuodelis-Flores & Ries, 2015).

Our finding that unplanned discharges including AMA discharges may contribute to suicide is not surprising. AMA discharges are more common in patients with SUD (Alfandre, 2009) and are associated with suicide risk after hospitalization (Riblet et al., 2018). One mechanism by which AMA discharges may contribute to suicide includes its disruptions to the therapeutic alliance (Dunster-Page, Haddock, Wainwright, & Berry, 2017; Windish & Ratanawongsa, 2008) and poor therapeutic alliance may be associated with suicide risk (Dunster-Page et al., 2017). Failure to complete treatment in a residential program has also been shown to contribute to worse health outcomes including higher overall mortality (Decker et al., 2017; Schaefer et al., 2005). In 2017, the VHA implemented a policy to ensure that irregular discharges from a residential program receive follow-up care (Office of Mental Health and Suicide Prevention, 2017). However, there are unique challenges in caring for patients who are precipitously discharged because of program violation. While these behaviors are harmful to the therapeutic milieu, these patients may be at higher risk for suicide for various reasons including recent relapse (Yuodelis-Flores & Ries, 2015). Perhaps, programs should consider alternative interventions that better target the patient’s ongoing substance use and suicide risk. For example, patients may benefit from post-card interventions such as Caring Contacts which have shown some promise in addressing suicide risk after discharge from acute settings (Comtois et al., 2019).

Akin to prior research (Appleby, Luchins, Dyson, Fanning, & Freels, 2001; Brown, Bennett, Li, & Bellack, 2011), we found that patient engagement was an important concern among many of the included cases. Patients with SUD are known to encounter more challenges in engaging in treatment and these problems are exacerbated in those with additional mental health comorbidities such as depression (Brown et al., 2011). Several strategies may assist patients with engaging in aftercare including having access to more resources, feeling more supported by staff and receiving interventions that facilitate continuity of care (Harris, McKellar, Moos, Schaefer, & Cronkite, 2006; Schaefer, Harris, Cronkite, & Turrubiartes, 2008). Motivational interviewing may also facilitate engagement (Smedslund et al., 2011). Finally, it may be important to address others stressors including legal problems and involve family members directly in treatment (Brown et al., 2011; Daley, 2013). In 2017, the VHA initiated a requirement that residential programs implement enhanced discharge planning processes to improve family engagement in care (VA DUSHOM, June 2017).

Our observation that poor communication between providers contributed to some suicide events is not surprising. Multiple studies have found that breakdowns in communication contribute to poor outcomes (Taran, 2011). Many effective strategies to improve communication between healthcare providers have been developed including Crew resource management, Team STEPPS and medical team training (MTT; Lundberg & Korndorffer, 2015; Wakeman & Langham, 2018; Weaver, Dy, & Rosen, 2014; Young-Xu et al., 2011). While these strategies were developed for use in medical and surgical settings, certain aspects may apply to residential treatment such as training teams to identify red flags, make use of checklists, and transfer information during transitions in care (Neily et al., 2010). Although only a few reports raised concerns about breakdowns in communication between staff and other patients, the observation is noteworthy. Given that patients spend a considerable amount of time interacting with each other in the residential setting and are uniformly at heightened risk for suicide, there may be value in providing patients with standardized suicide prevention education. For example, the Buddy-to-Buddy program developed by the Michigan Army National Guard as well as faculty and staff from Michigan State University and the University of Michigan has shown promise in connecting soldiers with peers to facilitate treatment engagement. The ‘buddy’ soldier is educated about warning signs and instructed that “your job is not to give help, its’ to get help” (Greden et al., 2010, p. 94).

Our study has important limitations. First, our study is likely not representative of all suicides occurring within three months after discharge from a VA residential drug treatment program. While facilities are required to conduct a RCA on a suicide occurring within 7 days of inpatient mental health discharge (VA NCPS, 2011), this requirement does not apply to residential programs. In some cases, a facility may also not have been notified of the death by next of kin or law enforcement, especially if the patient sought care elsewhere or moved out of state. It is also possible that despite our rigorous search for included cases that we overlooked some cases. Second, our analysis is limited by the quality of RCA investigation and the related documentation. Reassuringly, however, teams use standardized materials to conduct RCA. Third, because our study does not include a control group, we cannot make a conclusive determination that healthcare factors such as poor patient engagement and precipitous discharge contribute to suicide risk following residential treatment. We were unable to adjust our findings for potential baseline confounding because reports do not reliably report on individual characteristics and reports cannot be linked to medical records. Fourth, our study was comprised entirely of veterans and most decedents were men. This may limit the generalizability of our findings.

Overall, our findings echo concerns raised in the literature about the role of patient engagement and disruptions in care processes in contributing to suicide risk. Our findings suggest that these disruptions may play a role in suicides after discharge from residential substance abuse treatment programs. There is a need to implement suicide prevention strategies in residential programs that ensure that suicide risk is adequately addressed and promote engagement in treatment.