Background: Patients discharged against medical advice (AMA) have disproportionately high healthcare costs and increased morbidity, mortality, and hospital readmissions. While patient risk factors for discharge AMA are known, there is little data regarding providers’ practice patterns during AMA discharge, including provision of follow-up appointments. Similarly, the frequency of a documented discussion of the risks and benefits of AMA discharge is unclear. The goal of this study was to evaluate provider adherence to these best practices prior to AMA discharge.Methods: We reviewed a convenience sample of all AMA discharges from our institution (an urban, academic medical center) over one year. The primary outcome was documentation of follow-up appointments in discharge paperwork. We also assessed for documentation of a comprehensive risk/benefit discussion prior to AMA discharge, as well as documentation of an anticipated AMA discharge (an “early warning”). We conducted an analysis to evaluate the impact of patient and admission characteristics (such as race, age, and discharge time) on the likelihood of follow-up. Finally, we assessed the impact of follow-up on 30-day readmissions. Chi-square and Fisher Exact tests were used for comparisons.

Results: We reviewed 213 discharges, representing 193 unique patients. 32.8% had follow-up appointments documented at discharge. 26% had a documented early warning. Follow-up was more common in patients older than the median age of 52 (42.3% versus 27.4%, p=0.04) and in non-white patients (36.7% vs 21.8%, p= 0.04). Time of discharge, early warning, and documented risk/benefit discussion did not influence follow-up. A risk/benefit discussion was documented for 68.5% of discharges and was not influenced by age, race, discharge time, early warning, or service type. The 30-day readmission rate was 37% (comparison= 9.1% hospital-wide, p<0.0001). Patients with follow-up appointments were more likely to be readmitted (31.4% versus 18.9%, p=0.04).

Conclusions: In this study, we showed that most patients discharged AMA did not receive a follow-up appointment. Patients with follow-up were more likely to be readmitted to the hospital, which may reflect high utilization of healthcare services in the AMA population and suggests that patients deemed to be high-risk were provided follow-up. Differences in follow-up based on age and race are difficult to interpret, but suggest inconsistencies in care patterns within the AMA population. Early warning of AMA discharge was uncommon and did not impact the likelihood of follow-up, underscoring the importance of early discharge planning and patient engagement for all patients. For a majority of patients, a risk/benefit discussion was documented prior to discharge. Our results highlight the difficulty in facilitating safe care transitions for AMA patients. Further work will focus on developing strategies to encourage adherence to best practices and evaluating their impact on the care of AMA patients.