Taking an Evidence-Based Approach to Involuntary Psychiatric Hospitalization

Dozens of observational studies have investigated involuntary psychiatric hospitalization, including the characteristics of patients receiving these services, factors that shape clinician decision making, and predictors of outcomes of civil commitment hearings (3, 4). Moreover, a small number of randomized controlled trials (RCTs) have examined outpatient commitment (i.e., court-ordered, community-based treatment for individuals with severe mental illness) with mixed findings (5). By comparison, involuntary psychiatric hospitalization continues in every U.S. state, largely without RCTs supporting its use.

Ethical and practical concerns are key reasons for the lack of RCTs comparing involuntary psychiatric hospitalization with outpatient care. Randomly assigning individuals who meet emergency or inpatient commitment criteria to outpatient follow-up care instead of usual care (e.g., involuntary hospitalization) carries substantial risks, including that study participants might harm themselves or others. Such a trial would be unlikely to satisfy the equipoise between study arms needed for randomization. Even if robust outpatient models that would result in equipoise for a subset of patients were offered, research oversight organizations may hesitate to authorize random treatment assignment of certain patients in crisis, such as acutely suicidal or homicidal individuals. Furthermore, many researchers and clinicians might not agree to participate in these trials without protections from malpractice liability or other legal claims.

Some RCTs have examined aspects of involuntary psychiatric hospitalization, such as strategies to prevent use of seclusion and restraint; however, RCTs seeking to evaluate more fundamental purposes of involuntary psychiatric hospitalization (e.g., preventing self-harm or violence due to mental illness) have been rarer. In one example, a 2008 trial randomly assigned 393 patients in California who met involuntary hospitalization criteria for danger to self or grave disability (i.e., inability to care for self) due to mental illness to usual locked inpatient care or an unlocked crisis residential facility (6). The authors reported that patients who had been randomly assigned to the unlocked setting had greater satisfaction with services and short-term improvement in several psychiatric symptoms compared with those assigned to locked inpatient care (6). Although its findings were intriguing, the trial also had considerable limitations because most randomly assigned participants did not complete all follow-up interviews and the study did not address several key outcome measures, such as incidents of absconding or self-harm (6).

Because of these challenges surrounding RCTs, observational studies have been the mainstay of research into involuntary psychiatric hospitalization (3). Comparing treatment outcomes among groups by controlling for covariates has been one method for studying these practices, sometimes with surprising findings. For instance, a 15-year study using propensity score matching on >145,000 cases of inpatient admissions in Germany found that treatment on unlocked psychiatric units was associated with decreased probabilities of suicide attempts and absconding compared with treatment on locked units (7). These types of studies are thought provoking but also have methodological concerns, including the potential for confounding factors (e.g., clinician impression, availability of social supports, and past attempts at community treatment) that might not be fully addressed and likely shape clinical decision making around involuntary care (4).

Additional naturalistic studies with quasi-experimental designs have been conducted. For instance, one study examined Virginia’s use of temporary detention orders (TDOs) that lasted up to 48 hours before commitment hearings; however, hearings could be held earlier or delayed (e.g., because of weekends or holidays), allowing comparisons along a spectrum ranging from approximately 24 to 96 hours of detention before hearings (8). By examining 500 Medicaid recipients placed under TDOs between 2008 and 2009, the authors found that increased TDO length was associated with reduced risk for subsequent hospitalization after commitment hearings, suggesting that longer psychiatric holds may facilitate stabilization (8). Yet, in some circumstances, longer TDOs might also lead to longer total time in the hospital (i.e., TDO stays combined with post-TDO hospitalization) compared with shorter TDOs (8).

Research-based methods alone cannot resolve all the questions surrounding involuntary psychiatric hospitalization. Besides limitations arising from study methodologies, translating study outcomes into policy requires balancing clinical outcomes, such as incidents of self-harm or violence, with notions of personal liberty and autonomy, which are harder to measure. As noted by Appelbaum (4), “Difficult choices must be made about the relative weights our society places on freedom, autonomy, and paternalistic beneficence. These are questions of values, not numbers.” Still, more systematic data collection and research can shape the use of these policies, including their effects on patients, families, clinicians, and the broader public, as well as systems (e.g., hospitals, courts, and police) tasked with providing these services (4).

Policy makers across the United States are seeking to expand use of involuntary hospitalization, including broadening civil commitment criteria for people with mental disorders, as well as people with substance use disorders (14, 15). Accordingly, the importance of systematically monitoring, studying, and improving involuntary hospitalization practices has become even more relevant (9). Despite the controversies surrounding involuntary psychiatric hospitalization, many would likely agree that its use should be limited as a last resort when all other options have run out. Whether or not research-based methods can ultimately answer every question about involuntary psychiatric hospitalization, it remains clear that far too many questions remain unexplored.