Background: The principle that patients have the right to make choices about their healthcare, including the decision to leave the hospital against the advice of the medical staff, is a foundation of medical ethics.  Consistently, studies have found disproportionately higher rates of readmission and mortality for patients who leave against medical advice (AMA) as compared to patients who are conventionally discharged. This mixed-methods study aimed to: (1) characterize the patients who leave a county hospital AMA; (2) identify the patients who are at highest risk of leaving AMA; and (3) investigate their reasons for doing so.

Methods: Subjects included all patients admitted to X Hospital during 2014 whose discharge in the electronic medical record was designated AMA.  A research assistant reviewed the charts from all AMA discharges using an IRB-approved chart abstraction tool to assess demographic information and details about the hospitalization, including any documentation of the reason why the patient left AMA.  Another research assistant reviewed the qualitative data and identified overarching themes for why patients left AMA.  Quantitative data was analyzed using STATA.

Results: Of all admissions in 2014, 319 ended with an AMA discharge (2.7%).  268 patients left AMA during 2014, and 29 left AMA multiple times.  Compared with all patients admitted to X Hospital, patients who left AMA were more likely to be male, English-speaking, and black. Patients who left AMA more than once had higher rates of heavy alcohol use (54% vs. 31%, p = 0.01) and illicit drug use (89% vs. 54%, p<0.001) than patients who left AMA only once. Three broad themes were identified as overarching reasons why patients left AMA: (1) dissatisfaction with hospital care; (2) urgent personal issues; and (3) mental health and substance abuse issues.

Conclusions: Compared to all patients admitted to X Hospital, black patients are vastly overrepresented in AMA discharges, and Asian and Hispanic/Latino patients are underrepresented.  This may be accounted for by racial differences among X Hospital patients in rates of substance use, cultural differences in views of patient autonomy, and/or racial disparities in hospital care. Among the patients who left AMA multiple times, nearly all were actively using illicit drugs, and most were current heavy drinkers.  The qualitative data in this study suggest multiple potential areas for intervention in AMA discharges.  Providers should consider the risks of AMA discharge when pain is undertreated.  Administration should address inadequate hospital amenities and poor food options. Personal issues associated with AMA discharge may be addressed with social work assistance. Mental health and substance abuse issues may be better addressed with early treatment of withdrawal symptoms and improved access to psychiatric resources. Study findings will inform future projects and policies to improve patient safety and decrease the AMA discharge rate.