Reducing Frequent Utilization of Psychiatric Emergency Services Among Veterans While Maintaining Quality of Care

Use of psychiatric emergency services, usually in emergency departments (EDs) or inpatient psychiatry units, contributes substantially to health care costs in the United States (1,2). That is particularly true of the Veterans Health Administration (VHA), a safety-net provider serving a population with high rates of psychiatric problems (3). Over the past decade, there has been an increase in use of VHA mental health services (3–5), with utilization of psychiatric emergency services contributing approximately one-third of the total cost of care for VHA psychiatric patients (6). Across VHA and non-VHA hospitals, 5% to 10% of patients who utilize psychiatric emergency services contribute a disproportionate share of the cost of these services (7–13).

To date, research on high utilizers of psychiatric emergency services has relied on administrative data to identify these patients’ characteristics (13,14). Extant research in this area has linked frequent utilization of psychiatric emergency services to diagnoses of serious mental illness, substance use disorder, and personality disorder as well as to homelessness and detoxification service utilization (11–17). However, administrative data lack information on context of care, such as available resources and care delivery processes, which may affect the frequency of service utilization (18–20). Also, there is an absence of research on the context of care as it relates to psychiatric emergency services in the VHA (16), even though the VHA is the largest provider of public mental health care in the United States.

Qualitative methods are well suited to understand the context of care for high utilizers of psychiatric emergency services. Such an inquiry can identify barriers to and facilitators of reducing these patients’ use of these services (21). To our knowledge, no such study has been conducted to date. Investigating these issues within an integrated, near-universal health care system such as the VHA can yield valuable insights for ongoing debates about health care in the United States, specifically about access to and cost of health care for the most vulnerable patients.

We examined perceptions of stakeholders across VHA regarding the context of care for high utilizers of psychiatric emergency services to identify barriers to and facilitators of reducing use of these high-cost services. We interviewed a national sample of directors of mental health services and providers of psychiatric emergency services. Including directors of mental health services had the advantage of obtaining system-level perspectives on this topic and on the continuum of psychiatric care available to patients.

Methods

Study Design

Directors of mental health services and providers of psychiatric emergency services participated in a semistructured telephone interview. To obtain feedback from informants across a range of facilities, we calculated the ratio of high utilizers of psychiatric emergency services to total number of unique patients who received psychiatric emergency services in fiscal years 2011 and 2012 for each VHA facility that provides these services in an ED or an acute inpatient psychiatry unit (N=128). An encounter involving psychiatric emergency services was defined as an ED visit (stop code 130) or a unique acute inpatient psychiatry admission (bed section code 93) with a corresponding ICD-9 psychiatric diagnosis. We included inpatient admissions so the findings would generalize to the full continuum of emergency care for patients.

A high utilizer was defined as an individual with three or more encounters involving psychiatric emergency services during the study period, a definition consistent with the literature (11). Data on ED visits and inpatient psychiatric admissions (and corresponding diagnoses) were obtained from VHA’s National Patient Care Database and Patient Treatment Files, respectively. In the literature, psychiatric emergency services typically refer to dedicated units for psychiatric emergencies (18). We defined the term broadly in order to capture inpatient psychiatry admissions. However, the focus of the study was largely psychiatric emergency services in EDs. Most encounters (96.5%) in our data were based on an ED visit; only 3.5% of encounters were unique admissions to an inpatient psychiatry unit that were not preceded by an ED visit.

Across facilities, the ratio of high utilizers to total number of unique patients who received psychiatric emergency services in FY 2011–2012 ranged from 5.7% to 21.1% (mean±SD=12.8%±4.3%). Thus, on average, one of every eight patients who received psychiatric emergency services at a VHA facility was a high utilizer. To maximize variation and identify themes that cut across the distribution, we used a multistep process to identify facilities for conducting interviews. First, stratified purposeful sampling was used to identify facilities in which the percentage of high utilizers fell within the upper (≥14.6%) or lower (≤10.2%) quartile (22). Next, two facilities from each of five VHA complexity levels (1a, 1b, 1c, 2, and 3) were randomly selected from both quartiles (23). Highest-complexity facilities (level 1a) are marked by high volume, higher-risk patients, and specialized clinical services. Lowest-complexity facilities (level 3) are marked by low volume, lower-risk patients, and few or no specialized clinical services. When facilities were not eligible because they reported not having emergency services (N=2), they did not respond to e-mail requests (N=11), or all points of contact declined participation (N=6), another facility was randomly selected from the corresponding quartile. Forty-one facilities, 21 from the lower quartile and 20 from the upper quartile, were randomly selected and contacted.

For each of the selected facilities, we used online databases to identify the names and e-mail addresses of directors of mental health services (22). Directors were contacted via e-mail with a letter of invitation to participate. Phone interviews were conducted by the second author and lasted 45 minutes, on average. Participants were not made aware of their facilities’ ratio of high utilizers to total number of unique psychiatric patients treated in FY 2011–2012. After the interview, participants were asked to provide the name and contact information for another administrator or provider of psychiatric emergency services at their facility. Referrals were contacted via e-mail and invited to participate. Interviews continued until at least one informant from each of the five facility complexity levels was enrolled and thematic saturation was reached (24). Interviews were conducted from January to June 2015. All procedures were approved by the local institutional review board. All participants provided informed consent.

Qualitative Interview Guide

The interview guide was informed by the Consolidated Framework for Implementation Research (CFIR) (25), which consolidates constructs from implementation science frameworks to understand factors that may affect quality of care for a population in a particular setting. Participants were queried on CFIR domains related to the context of care for high utilizers of psychiatric emergency services (see box on the next page) (18–20).

Data Analysis

We employed a modified version of Rapid Identification of Themes From Audio-Recordings (26). After each interview, the second author took detailed notes by using a template to summarize responses to questions and document preliminary themes reported by participants related to barriers to and facilitators of reducing utilization of psychiatric emergency services. Six codes were specified a priori (patient-, provider-, and systems-level barriers and patient-, provider-, and systems-level facilitators) (27). These codes and preliminary themes were operationalized in a codebook developed by two authors (DB and LM). The postinterview notes for participants from facilities in the upper and lower quartiles were independently reviewed by DB and LM. The preliminary themes that were documented were the same for facilities in the upper and lower quartiles; thus, the data were analyzed together. Once interview notes were completed, we copied the notes into an Excel matrix to compare the preliminary themes within each code (columns) and across participants (rows). The matrix was organized into the aforementioned codes, and a summary of participants’ responses related to the code was used in each cell. Next, the same two authors independently reviewed the matrix to apply the final themes to each cell and identified direct quotes related to each theme by referring to the full transcripts of the interviews. We then compared their application of themes to the matrix and engaged in a consensus process to rectify disagreements.

Results

Participants were 31 key informants across 22 VHA facilities, 12 in the lower quartile and 10 in the upper quartile, with the following facility complexity levels: 1a (N=7), 1b (N=3), 1c (N=3), 2 (N=4), and 3 (N=5). Thirteen participants were directors of mental health services (all mental health services [N=7], EDs [N=4], and outpatient services [N=2]), and 18 were providers of psychiatric emergency services (general EDs [N=11], psychiatric EDs [N=4], and inpatient psychiatric units [N=3]). Across all participants, 13 were psychiatrists, 12 were social workers, four were psychologists, and two were nurse practitioners. Participants were mostly female (N=16) and non-Hispanic Caucasian (N=25), with a mean age of 49.5±8.6. Participants had been in their current role for 6.9±5.8 years.

Barriers to Reducing Utilization of Psychiatric Emergency Services

Patient nonadherence to treatment recommendations.

Participants described patient nonadherence to treatment recommendations—for example, not attending outpatient appointments and medication noncompliance—as a barrier to reducing utilization of psychiatric emergency services (Table 1). Participants highlighted several factors that contributed in their view to nonadherence to outpatient care by high utilizers, including patients’ distrust of non-ED providers, lack of patient support networks to facilitate care engagement, and geographic barriers and lack of transportation.

Patient transiency.

Participants highlighted the transiency of high utilizers as another barrier to reducing frequent utilization of psychiatric emergency services. For example, high utilizers often have no reliable contact information or stable residence. Participants distinguished this factor from patient nonadherence in that it was seen as hindering the ability of staff, including providers of services other than psychiatric emergency care, to track and follow up with patients to link them to services.

Staff stigma toward high utilizers.

A provider-level barrier to reducing frequent utilization of psychiatric emergency services was stigma among staff toward high utilizers, which impedes their ability to develop a therapeutic relationship with patients and negatively affects the quality of care received. Relatedly, participants reported that some high utilizers, particularly those with personality disorders, often exhibit parasuicidality and engage in inappropriate use of psychiatric emergency services and that staff who provide these services do not know how to manage these situations.

Lack of training and expertise in the management of psychiatric issues.

Another provider-level barrier was lack of training among ED staff at nonpsychiatric facilities (for example, medical social workers and nurses) in how to identify community resources for high utilizers and how to develop appropriate discharge plans for patients with psychiatric disorders.

Lack of specialized services to address short- and long-term care needs.

Across facility complexity levels, participants reported that a lack of several specialized services contributed to frequent utilization of psychiatric emergency services. These services included community (non-VHA) social services and partnerships; long-term care for those with serious mental illness, chronic substance abuse, or dementia; pharmacotherapy for patients with alcohol and opiate use disorders; and outpatient alcohol and opiate detoxification services.

Facilitators of Reducing Utilization of Psychiatric Emergency Services

Recovery-oriented care.

Participants described the value of relationship building in reducing utilization of psychiatric emergency services, particularly specific patient-centered practices that are helpful in the relationship-building process (Table 2). Motivational interviewing, connecting patients to peer-based services, and engaging family members in care planning were identified as important in relationship building. These practices are all elements of recovery-oriented care (28).

Education, training, and supervision of ED staff.

Participants highlighted the value of more education, training, and supervision of ED staff in evidence-based practices and management of psychiatric disorders. This was perceived by participants as helping reduce utilization of psychiatric emergency services by mitigating staff stigma toward high utilizers.

Interdisciplinary, team-based care.

Participants reported that interdisciplinary, team-based care facilitated better communication within and across services, leading to better care coordination for high utilizers and ultimately helping to reduce unnecessary ED visits. Participants also noted the benefits of this approach for mitigating provider burnout by limiting patients’ potential to split their care providers (that is, view a provider as either working for them or working against them).

Social workers in the ED.

Social workers based in the ED were described as having a crucial role in the care of high utilizers, particularly for connecting patients to community-based services. Within this theme, participants highlighted the value of staffing practices that provide around-the-clock or overnight coverage by psychiatric social workers. Continuous coverage by social workers would ensure that patients admitted outside regular business hours receive appropriate discharge planning and linkage to services on the following morning.

Intensive case management.

Participants reported that intensive case management programs for patients with serious mental illness and patients who are homeless (29) facilitated community outreach to high utilizers of psychiatric emergency services and ensured patients’ compliance with medications and outpatient appointments. In these case management programs, it was suggested to flag high utilizers, or patients who are at risk of becoming high utilizers, using predictive analytics. Individualized treatment plans for these individuals could then be developed to better target the needs of high utilizers.

Discussion

Through our inquiry into the context of care for high utilizers of psychiatric emergency services, this study extended the knowledge base of barriers to and facilitators of reducing service utilization in this population. Despite being able to provide patients with access to behavioral health and social services within an integrated care system (30), the VHA continues to face barriers to reducing utilization of psychiatric emergency services. As debate over health care continues in the United States, consideration of barriers to and facilitators of reducing utilization of psychiatric emergency services in a safety-net provider such as VHA will be critical for health care systems in the private sector, because these systems may be faced with the challenge of serving an increasing number of disenfranchised patients. The lessons learned here, informed by the perspectives of providers across a range of services, may help other health care systems to better target barriers to remove and facilitators to implement to contain health care costs.

Recommendations for Removing Barriers

Patient transiency was found to be a critical barrier to offering continuous outpatient care. Lack of residential stability is associated with poor adherence to mental health aftercare (31). Assertive community outreach may help to combat this problem (32). For example, through the federal government’s Lifeline Assistance Program, most homeless veterans have a mobile phone (33). VHA’s patient-centered medical home for patients who are homeless uses text messaging to increase treatment retention and care adherence in this population (34). Expansion of the Lifeline Assistance Program through a public-private partnership may be a viable alternative for the nonprofit sector (35).

Regarding provider stigma, it has been shown that implicit bias toward patients with chronic mental and substance use disorders negatively affects patient care (36). However, such bias may be reduced with additional training and support. Notably, medical schools have begun to include addiction medicine in their curriculum and emphasize the deleterious effects of stigma (37,38). Furthermore, patients with personality disorders are more likely than patients with other psychiatric disorders to use psychiatric emergency services for their health care (11). Implementing protected time for providers of psychiatric emergency services to engage in team peer support and collaborative care programs with providers from other services may facilitate continuity and coordination of care for these patients (39).

Limited access to addiction pharmacotherapy emerged as a barrier, which is consistent with the nationwide lack of addiction medicine training (40,41). There is ample evidence that medication-assisted treatment for substance use disorders reduces relapse and promotes functional recovery, but cultural and historical factors have hindered its wider acceptance into the current treatment armamentarium (42). New initiatives offer promise for expanding implementation of medication-assisted treatment for high utilizers of psychiatric emergency services (43) and promote a paradigm shift whereby screening and treatment for addictions are integrated with other aspects of health care (44,45).

Limitations on the time allowed for providing psychiatric emergency services did not emerge as a robust theme. This factor should also be considered, given that there is a national movement in health care toward emphasizing community-based care and reductions in length of inpatient stays (46).

Recommendations for Harnessing Facilitators

Continuous access to a psychiatric social worker—24 hours a day, seven days a week—was recommended as a way to provide telehealth services and referrals to community resources and coordination of next-day services for patients who seek care overnight. Through an awareness of local resources in a patient’s community, such a role could address geographic and transportation barriers to accessing outpatient care. In settings with staffing limitations, models of asynchronous telehealth, whereby communication occurs between patients and providers at different times (47), may help to better coordinate discharge to outpatient services. Strengthening of community partnerships with VHA and intra-agency collaborations may also represent important next steps. To better guide development of these partnerships, additional research is needed to determine how availability of community-based services affects use of psychiatric emergency services and other services at the VHA (18).

Intensive case management appears critical for reducing overutilization of psychiatric emergency services, given that recent studies have shown that brief interventions in EDs for substance use and mental health issues are not cost-effective (48). Furthermore, regarding interdisciplinary, team-based care, more research is needed to understand what factors affect the relationships between psychiatric providers within and across services and organizations, given that better communication and coordination may reduce unnecessary ED visits.

Increased education and training in the management of psychiatric disorders may also facilitate better outcomes for high utilizers of psychiatric emergency services. Nonpsychiatric providers are often the first to interact with patients in EDs. As such, their ability to recognize psychiatric symptoms, provide appropriate referrals, and adhere to recovery-oriented care practices is crucial to patients’ satisfaction and well-being (49,50). Recovery-oriented care is focused on the formation of a strong therapeutic alliance and emphasizes patient preferences (28). Although VHA has prioritized the delivery of recovery-oriented care among mental health providers (51), more widespread adoption of recovery-oriented care among nonpsychiatric providers in EDs may increase the likelihood that high utilizers will make a successful transition to outpatient behavioral health services (28).

Finally, implementation of predictive-analytic tools and organization-level high-utilizer committees to create, test, and evaluate policies may promote continuity of care and reduce utilization of psychiatric emergency services. For example, it is estimated that 10% of homeless veterans in PCMHs utilize 65% to 70% of acute care services provided to patients on PCMH teams (29). These teams use clinical aids to develop care plans to address factors that drive these patients’ use of acute care services.

Strengths and Limitations

Regarding strengths, this study complemented descriptive-analytic studies of administrative data (13,14) by identifying factors that are not available or not reliably assessed via administrative records. Furthermore, most studies of high utilizers of psychiatric emergency services have been limited geographically or were focused on private hospitals. This study had the benefit of being a national-level examination that focused on the VHA, an integrated, near-universal health care system and safety-net provider. Regarding limitations, some subpopulations, such as undocumented immigrants, are not included in the VHA patient base, and generalizability of the sample to nonveteran populations is unknown. Second, direct patient perspectives were not solicited in this study. Third, asking directors to refer other providers for the study may have introduced bias into the sample (for example, over- and under-representation of barrier and facilitator themes to reducing overutilization of psychiatric emergency services among high utilizers). Fourth, interviews were conducted approximately three years after the period on which the high-utilizer list was based. If ratios of high utilizers to all users of psychiatric emergency services changed over this time frame, it may have contributed to the lack of differences in the barriers and facilitators identified by providers in the upper and lower quartiles. The stability of utilization of psychiatric emergency services over time, both at the patient and facility levels, should be a target for future research.

Finally, the reported themes reflect the perspectives of participants. Whether removing barriers or harnessing facilitators will have an impact on utilization of psychiatric emergency services requires empirical validation. Furthermore, many high utilizers of psychiatric emergency services struggle with refractory, treatment-resistant disorders and may need frequent psychiatric emergency care. Nonetheless, these findings contribute to hypothesis generation in terms of contextual factors to target in future research on high utilizers of psychiatric emergency services.

Conclusions

This study identified challenges and potential solutions related to reducing utilization of psychiatric emergency services in VHA, a safety-net provider. These results may help other health care systems better target which barriers to remove or which facilitators to implement to contain rising health care costs. Future studies should continue to perform needs assessments among patients and providers to determine where current practices fall short of serving these objectives.