Irregular discharge in residential settings and mortality outcomes

We found that irregular discharge from residential settings results in increased risk of suicide, external-cause and all-cause mortality for 6 months after discharge. Although our results generally align with the literature, it is difficult to directly compare our findings because prior studies generally focus on long-term rather than short-term mortality outcomes after residential treatment.^{Reference Harris, Gupta, Bowe, Ellerbe, Phelps and Rubinsky22,Reference Decker, Peglow, Samples and Cunningham23} For example, Harris et al found that among patients discharged from substance abuse residential rehabilitation programmes that early discharge (defined as the proportion of patients discharged within 1 week of admission) predicted 2-year all-cause mortality (odds ratio 1.49, P < 0.001).^{Reference Harris, Gupta, Bowe, Ellerbe, Phelps and Rubinsky22} We also noted that patients with an irregular discharge were at 1.5 times greater risk for all-cause mortality within the first 30 days of a residential stay compared with those with a regular discharge. However, we found that risk attenuated over time with an observed increased risk of 28% in the first 180 days. Unlike Harris et al, our study includes a broader range of residential programmes and does not focus exclusively on programmes that treat SUD.^{Reference Harris, Gupta, Bowe, Ellerbe, Phelps and Rubinsky22} Moreover, as with our study, Decker et al reported in a single-site study of patients admitted to a Veterans Affairs residential substance use treatment programme that failure to complete treatment was associated with a high risk of all-cause mortality in the first 5 years of discharge (HR = 2.3, 95% CI 1.1–4.9).^{Reference Decker, Peglow, Samples and Cunningham23} The effect, however, was no longer significant after adjustment (HR = 2.1, 95% CI 0.9–4.6). Decker et al did not report on shorter-term risk for comparison.^{Reference Decker, Peglow, Samples and Cunningham23} Finally, aligned with our results, in a 10-year follow-up study of 567 patients admitted with non-organic psychosis to New York hospitals, Craig et al noted that deaths because of unnatural causes tended to occur close after discharge from treatment.^{Reference Craig, Ye and Bromet24} These authors, however, did not explore the role of irregular discharge on this risk.

Irregular discharge in acute in-patient settings and mortality outcomes

Our observation that irregular discharge in the acute in-patient setting was predictive of all-cause, external-cause and 30-day suicide mortality somewhat aligns with trends reported in the literature.^{Reference Riblet, Richardson, Shiner, Peltzman, Watts and McCarthy2,Reference Kuo, Tsai, Liao, Lee, Sung and Chen3,Reference Glasgow, Vaughn-Sarrazin and Kaboli25} Riblet et al also found no relationship between irregular discharge and 1-year suicide risk following an acute in-patient mental health stay (HR = 1.16, 95% CI 0.96–1.40).^{Reference Riblet, Richardson, Shiner, Peltzman, Watts and McCarthy2} It is hypothesised that the suicide risk associated with irregular discharge may be overshadowed by a pre-existing, high risk for suicide in this setting.^{Reference Riblet, Richardson, Shiner, Peltzman, Watts and McCarthy2} Yet, unlike Riblet et al, we did find that patients with an irregular discharge had a 50% increased risk for suicide within the first 30 days of discharge from an acute in-patient stay, suggesting that time may influence risk. Our result also mirrors those of Kuo et al who reported that patients were nearly 50% more likely to die by suicide after an irregular versus regular discharge from an acute in-patient mental health stay (HR = 1.43, 95% CI 1.05–1.95).^{Reference Kuo, Tsai, Liao, Lee, Sung and Chen3} We are not aware of any studies that have looked at the relationship between irregular discharge in acute in-patient mental health settings and other mortality outcomes such as external-cause mortality. However, in a cohort study of 1.9 million patients with Veterans Affairs general medical admissions (2004–2008), Glasgow et al observed that an irregular discharge did not predict 30-day all-cause mortality (HR = 1.10, 95% CI 0.98–1.24), but did predict 60-day all-cause mortality (HR = 1.11, 95% CI 1.02–1.21).^{Reference Glasgow, Vaughn-Sarrazin and Kaboli25} Our findings differ from Glasgow et al in that we found a significantly heightened risk for all-cause mortality at 30, 90 and 180 days. It is possible that the difference may be because of our focus on acute in-patient mental health stays (versus general medical stays). Glasgow et al also raised concerns that there was insufficient statistical power.^{Reference Glasgow, Vaughn-Sarrazin and Kaboli25}

We observed a significant increased risk for external- and all-cause mortality in the first 90 and 180 days after an acute in-patient discharge, but not for suicide. We may have simply lacked sufficient statistical power. Statistical power increases in survival analysis as the event rate increases. For example, only 11% of deaths in the first 30 days were attributable to suicide. The point estimates were also similar across outcomes and irregular discharge was always associated with increased risk. Conversely, it is possible that the risk for these outcomes may diminish at differential rates following irregular discharge from acute settings.

We highlight that the mortality rates were uniformly high after an acute in-patient or residential stay regardless of discharge type. This aligns with current trends in the literature,^{Reference Chung, Hadzi-Pavlovic, Wang, Swaraj, Olfson and Large9,Reference Chung, Ryan, Hadzi-Pavlovic, Singh, Stanton and Large10,Reference Katz, Peltzman, Jedele and McCarthy26,Reference Valenstein, Kim, Ganoczy, McCarthy, Zivin and Austin27} although there has been limited study of mortality rates following residential stays.^{Reference Harris, Gupta, Bowe, Ellerbe, Phelps and Rubinsky22,Reference Decker, Peglow, Samples and Cunningham23} For example, Valenstein et al also reported a rate of 568 suicide deaths per 100 000 person-years in the first 12 weeks after a psychiatric hospital stay among Veterans Affairs patients diagnosed with depression.^{Reference Valenstein, Kim, Ganoczy, McCarthy, Zivin and Austin27} Although Katz et al also observed high rates of all-cause mortality in the first 90 days after a Veterans Affairs mental health discharge, the authors cited higher rates than were reported in our study (4490 versus 3720 per 100 000 person-years).^{Reference Katz, Peltzman, Jedele and McCarthy26} We speculate that this may reflect differences in study years or definitions used to index admissions, setting or discharge status.^{Reference Katz, Peltzman, Jedele and McCarthy26}

Finally, it is important to note that the crude suicide rates reported in our study were 20 to 50 times greater than global age-standardised suicide rates.^{Reference Chung, Ryan, Hadzi-Pavlovic, Singh, Stanton and Large10} Suicide and external-cause mortality rates were also higher among discharges from acute in-patient versus residential stays. This is not surprising as residential programmes typically care for a less acute population whose needs do not reach a level of requiring acute in-patient treatment.^{Reference Glick, Sharfstein and Schwartz7,8}

Factors contributing to mortality risk with irregular discharge from mental health settings

Several factors may explain why patients are at greater risk for death after an irregular discharge from mental health treatment settings. Prior research has found that some of the key drivers of irregular discharge may include symptom worsening (such as undertreated withdrawal, uncontrolled pain),^{Reference Gundel, Allen, Osborne and Shwayhat28} perceived stigma from staff^{Reference Simon, Snow and Wakeman29} and poor therapeutic environment (for example unit restrictions serve as a trigger for patients with prior incarceration).^{Reference Simon, Snow and Wakeman29} Incomplete treatment at the time of discharge may heighten the risk for symptom worsening.^{Reference Holmes, Cooley, Fleisch and Rosenstein30} This may be especially true for patients with SUD who may experience withdrawal symptoms that drive them to abruptly leave the programme and, potentially, relapse.^{Reference Holmes, Cooley, Fleisch and Rosenstein30} Providers’ beliefs about the degree of treatment that should be provided to patients who leave with an irregular discharge may also result in patients leaving without treatment (such as medication supply or scheduled follow-up care).^{Reference Riblet, Kenneally, Shiner and Watts4,Reference Holmes, Cooley, Fleisch and Rosenstein30} Related to this concern, there may also be breakdowns in therapeutic alliance between providers and patients at the time of an irregular discharge. For example, patients may perceive that their needs were not addressed or feel generally alienated from providers because of stigma.^{Reference Holmes, Cooley, Fleisch and Rosenstein30} These factors may put patients at a higher risk for adverse outcomes such as suicide because they are then isolated from necessary treatment and related out-patient supports. Aligned with these concerns, we found that the risk for suicide was particularly high in the first 30 days post-discharge (HR = 3.41).

In a root-cause analysis study of suicides following residential treatment, Riblet et al also noted that factors contributing to post-discharge suicide risk included precipitous discharges because of programme violations, inadequate treatment and insufficient processes to address irregular discharges, assess suicide risk or arrange follow-up care.^{Reference Riblet, Kenneally, Shiner and Watts4} In residential settings, the length of participation in the programme may also predict the likelihood of engagement in treatment after discharge. A study of 367 Canadian adults who completed abstinence-based residential addiction treatment reported that for each additional day the patient remained in the residential programme, there was a 2% increase in the likelihood of participating in aftercare.^{Reference Arbour, Hambley and Ho31}

Strategies to improve mortality outcomes after irregular discharge from mental health settings

Because irregular discharges from acute in-patient and residential stays are associated with higher risk for death after discharge, it is imperative that future research focuses on identifying effective interventions to prevent irregular discharge and related adverse outcomes. To date, a few studies have evaluated various strategies to prevent irregular discharge, although they have primarily focused on patients with SUD.^{Reference Holmes, Cooley, Fleisch and Rosenstein30,Reference Thompson, Faig, VanKim, Sharma, Afshar and Karnik32,Reference Decker, Peglow and Samples33} For example, using retrospective data from over 35 000 hospital-related treatment encounters for SUD, Thompson et al found that compared with usual care, a substance use intervention team (SUIT) was associated with a significantly greater probability that patients would leave with a regular discharge (HR = 1.16, 95% CI 1.03–1.3).^{Reference Thompson, Faig, VanKim, Sharma, Afshar and Karnik32} The SUIT intervention focuses on harm reduction and includes interventions such as medication initiation, motivational interviewing, treatment education and processes to facilitate post-discharge care.^{Reference Thompson, Faig, VanKim, Sharma, Afshar and Karnik32} In a study of veterans discharged from an in-patient mental health recovery and rehabilitation programme, Decker et al also found that small group, patient-centred therapy interventions were associated with significant improvements in successful treatment completion and better treatment retention.^{Reference Decker, Peglow and Samples33}

Together, these findings suggest that successful approaches to address irregular discharge (and related adverse outcomes) may include adequate treatment of symptoms during admission and contingency plans to mitigate harms post-discharge and promote treatment engagement.^{Reference Holmes, Cooley, Fleisch and Rosenstein30,Reference Manuel, Yuan, Herman, Svikis, Nichols and Palmer34} Accordingly, the Veterans Affairs has implemented policies to ensure that patients with irregular discharge are offered an appointment within 24 h as well as an appointment within 7 days after discharge.³⁵ Our findings, however, suggest that it may be necessary to promote engagement in treatment for a substantially longer period of time. Understanding the factors that contribute to irregular discharge is essential to intervention development, especially as contextual factors such as poor programme atmosphere, negative biases on the part of providers, and breakdowns in therapeutic alliance may contribute to irregular discharges and inadequate follow-up care.^{Reference Simon, Snow and Wakeman29,Reference Holmes, Cooley, Fleisch and Rosenstein30} Although our study controlled for a broad array of confounders, including prior mental and physical health conditions and homelessness, unmeasured confounding variables may have influenced our findings. Interventions should be tailored to address the unique drivers of irregular discharge (and related harms) as there is likely no one-size fits all solution.

Study strengths and limitations

A clear strength of the study is use of a large and lengthy cohort derived from the largest integrated healthcare system in the USA. This is the first study to report on a broad range of mortality outcomes among patients discharged from acute in-patient and residential settings after irregular discharge. We also were able to adjust for patient-level factors highly predictive of the outcome. There are several limitations to our analysis. First, we relied entirely on administrative data to identify demographic and clinical information. We cannot infer causality because patients were not randomly assigned to conditions. Second, our study reports on members of the US veteran population who accessed Veterans Affairs care and most patients were men, thus limiting the generalisability of our findings. Compared with the general US population (and veterans who do not access Veterans Affairs care), veterans who access Veterans Affairs care tend to be older, have more health comorbidities, have higher rates of

homelessness, and are socioeconomically disadvantaged.^{Reference Eibner, Krull, Brown, Cefalu, Mulcahy and Pollard12} These factors as well as military experience may contribute to irregular discharge and post-discharge mortality. We also did not consider the role of non-Veterans Affairs mental health admissions on mortality outcomes because these care processes are outside of direct control of the Veterans Affairs. Finally, the survival curves point to clear differences in effect sizes based on treatment setting, highlighting an interaction between setting, discharge type and outcome. Because the analyses are crude, we cannot draw any formal conclusions.

In summary, we found that regardless of setting, the risk for death after an irregular discharge (versus a regular discharge) is the greatest within the first 30 days of discharge. Overall, our results emphasise that future research should focus on identifying effective interventions to address irregular discharges in acute in-patient and residential settings. These strategies should target related adverse health outcomes.