I had the privilege to attend medical school, complete residency and fellowship, and accept my first faculty position in three different states. This journey taught me that patients are similar everywhere but that their diagnoses and treatments vary widely.
In this issue, Gao and Olfson (
1) describe this phenomenon more scientifically. They leveraged national databases on emergency department (ED) and hospital utilization and hospital characteristics to describe correlates of admission rates among patients visiting the ED for concerns related to their primary diagnosis of a schizophrenia spectrum disorder (SSD). The primary analytic sample included 116,928 encounters across 11 states and 1,071 hospitals. Forty percent of ED visits resulted in an admission. However, admission rates varied substantially by hospital. Among hospitals in the lowest quintile of admission rate, only 0.4% of ED visits resulted in an admission; in the highest quintile, the admission rate exceeded 90%. These results demonstrate that patients with similar diagnoses received very different treatment.
Why is there so much variation in admission rates? The strongest correlate of a patient being admitted was the presence of inpatient psychiatric beds at the treating hospital. Other factors—such as the hospital’s state and presence of a comorbid substance use disorder—were less strongly correlated with admission but were still statistically significant. Hospitals with the lowest share of admission were on average 10 miles from the nearest inpatient psychiatric bed. More nuanced encounter-level data such as symptom severity or concurrent suicidality would be of interest but were not available.
When I first shared these findings with colleagues, their first reaction was to wonder whether hospitals are cynically prone to admit patients for financial benefit. I have no doubt that some readers will share this reaction. The implications of this study are more complex, and these findings force psychiatrists to consider the role of and indications for inpatient psychiatric hospitalization.
Gao and Olfson (
1) rightfully highlight significant issues of equity posed by their findings. The finding that some hospitals have very low admission rates suggests that many patients face substantial barriers to receiving high-acuity psychiatric care when needed. Emergency psychiatrists are familiar with the difficulty of admitting patients: inpatient units contend with staffing shortages, request lab work without clinical indication, review charts to identify patients who are a “bad fit” or might be difficult to discharge, and struggle to obtain insurance authorization. EDs that do not offer inpatient psychiatry services may also lack consultation or familiar relationships with admitting facilities; thus, even patients with diagnosed mental illness may receive inadequate management. The investigators focused on patients with a primary SSD diagnosis, but the risk for undertreatment would likely be even greater for many patients with depression, anxiety, or addiction and for patients visiting an ED for a primarily somatic complaint. The study’s methodology risks some undercounting of admissions, given that many hospitals’ processes entail discharging a psychiatric patient prior to readmitting that patient under a new encounter specific to behavioral health. Nevertheless, such procedural nuances seem unlikely to change the conclusions.
Whether hospitals with a high admission share are unnecessarily admitting patients must also be considered. The benefits derived from inpatient psychiatric hospitalization have been questioned (
2), and without data on patient outcomes after discharge psychiatrists cannot evaluate the risks or benefits of hospitalization for particular patients. Additional considerations may also affect the frequency of admission. Hospitals operating inpatient units are likely doing so in a favorable funding environment or in response to deficiencies in community mental health services, crisis stabilization units, withdrawal management programs, or other options for hospital diversion. In localities where mental health resources are concentrated on inpatient services, psychiatrists would be remiss not to offer this care for their patients. Emergency providers may also be responding to different cultural expectations of patients or families, either in favor of or against hospitalization. Procedures for involuntary hospitalization vary widely and are not accounted for in the reported data (
1).
These data do not suggest whether an ideal admission rate exists. Hospitalization may be necessary and lifesaving for the highest-risk patients, but this study reminds psychiatrists of their deficiencies of knowledge in the practice of emergency psychiatry, particularly regarding the decision of whether to admit a patient for inpatient care. A patient’s social needs, including access to housing, health care, and a recovery environment, have long been recognized as considerations in pursuing a certain level of care. Additional factors are known to influence the likelihood of psychiatric admission—for example, the intensity of crisis services or the experience of the evaluating psychiatrist (
3). Psychiatric hospitalization is an outcome of interest in research and evaluation, and it appears to be as much—if not more—a reflection of the local system of care as of the severity of a patient’s symptoms or complaints. Any ambiguity of indication for hospitalization makes evaluation of efforts to avert hospitalization and develop appropriate payment models difficult. The most useful tool in emergency psychiatry for navigating this complexity may be a standardized assessment of acuity that is predictive of outcomes, akin to mortality risk assessments for critically ill patients (
4). Such a tool could consider the multiple causes of mortality among psychiatric patients receiving emergency services (
5), assess the risk for agitation during the present encounter, and anticipate the resources necessary for proper care. An objective evaluation of acuity would enable controlled comparisons of patient outcomes across different interventions and guide deliberate programmatic planning to match a population’s needs. Developing such measures requires looking beyond hospitalization at patient-centered outcomes such as self-harm, suicide, overdose, and recovery.
When learning emergency psychiatry, trainees often struggle with how to determine the need for hospitalization or invoke involuntary treatment procedures. As an emergency psychiatrist with experience across different systems of care, I still struggle with these clinical decisions. In any given case, the answer is complex and not solely dependent on a patient’s symptoms. Gao and Olfson (
1) remind the profession that psychiatry is practiced as much in the intimacy of a patient encounter as in the context of a public conversation about stigma, equity, ethics, and expectations for behavioral health care.