Discharges against medical advice (AMA) — in which a hospitalized patient opts to leave before clinical recommendation — account for 1% to 2% of all U.S. inpatient hospital discharges (~500,000) annually.(1-3)
Patient risk factors for AMA discharge include male sex, uninsured or Medicaid status, substance use disorder, and psychiatric diagnosis.(4-8) Additional correlates encompass clinical factors such as uncontrolled pain or withdrawal symptoms and social factors such as lack of childcare or work obligations.(9-12) Racial and structural disparities also play a role: for example, Black patients are nearly twice as likely to leave AMA as White patients, as are patients with housing instability or other concerns with the social determinants of health (SDOH).(13,14)
National studies link AMA discharges to higher rates of 30-day readmission and 90-day all-cause mortality as well as increased healthcare utilization, estimated at up to 400,000 additional inpatient days and $800 million annually.(15-18) Despite these risks, no standardized evidence-based guidelines exist for the management of AMA discharges, leading to considerable variation in provider practice patterns during these encounters — including documentation of patient discussions and discharge planning measures such as medication prescription or arrangement of follow-up care.(9,10,19,20)
Study Objective
The objective of this study was to evaluate demographic and clinical characteristics, provider documentation practices, and 30-day readmission rates among patients discharged AMA from a tertiary academic medical center, and to identify opportunities to improve discharge planning.
Methods
Study Design and Setting
We conducted a retrospective chart review of adult patients who were discharged against medical advice from a 368-bed tertiary academic medical center between March 31, 2019, and March 26, 2024. This study was approved by the medical center’s Institutional Review Board, deemed to pose minimal risk to participants and therefore exempt from ongoing IRB oversight. Requirement for informed consent was waived because of the retrospective nature of the study. Patient data were de-identified at the time of collection.
Study Cohort
Patients were eligible for inclusion if they were aged 18 years or older, admitted under “inpatient” or “observation” status, discharged during the study period, and at the time of discharge had a discharge disposition documented in the EHR as “leaving against medical advice.” Patients who did not meet these criteria were excluded from the analysis.
Data Collection
Data were extracted from the EHR, Epic, in a two-step process and managed using the REDCap (Research Electronic Data Capture) electronic data capture tool. Preliminary quantitative data were extracted automatically from the EHR into a REDCap database and were then augmented with manual chart review by trained volunteers following a standardized data collection protocol. Charts requiring adjudication were identified by reviewers and resolved by consensus with the first author. Additionally, 10% of eligible encounters were independently reviewed by the first author to ensure accurate data abstraction.
Measures
Variables collected included basic demographic characteristics, admission and discharge dates, and principal hospitalization diagnosis. Clinical comorbidities were captured based on the presence of relevant ICD-10 codes for chronic medical conditions, substance use disorders and psychiatric illnesses. Self-reported barriers related to SDOH were collected when documented.
Reviewers manually studied discharge summaries and day-of-discharge progress notes for documentation of the following: assessment of patient decision-making capacity; the patient’s stated reasons for leaving; acknowledgment of premature discharge–related risks; offers for alternative treatments; and reassurance of continued access to care. Discharge orders were examined for evidence of discharge planning, including medication reconciliation, prescriptions, and scheduled follow-up appointments. Finally, reviewers recorded whether the patient was readmitted to the same hospital within 30 days.
Statistical Analysis
Descriptive statistics were used to summarize patient demographics, clinical characteristics, and comorbidities. Counts and percentages were used to display categorical characteristics. Mean and standard deviation was used for continuous variables such as age, number of medications at time of admission. In addition, a LACE+ index score, a validated tool used to predict readmission or death after discharge (where LACE stands for Length of patient stay in hospital, Acuity of admission, Comorbidity, and Emergency visit), was calculated for each patient encounter.(21-23) Differences between patients who had a psychiatric disorder or substance abuse disorder were compared using chi-square for readmission and T-test for length of stay. All statistical tests were two-sided, and p values less than .05 were considered statistically significant. All statistical analyses were performed in R version 4.2.2.
Results
Patient Demographics and Social Context
A total of 690 patients were discharged AMA between March 2019 and March 2024. Of these, the majority of patients were male (n = 413, 60%) and identified as White (n = 578, 86%), followed by Black or African American (n = 55, 8.1%). Marital status was most commonly listed as single (n = 305, 45%). Insurance coverage was primarily through Medicare or Medicare Advantage (n = 273, 40%), with fewer patients on Medicaid (n = 173, 25%) or private insurance (n = 165, 24%). In this cohort, 55 patients (8.0%) reported at least one documented SDOH concern in their chart, with food insecurity being the most common (7.2%) (Table 1).
Clinical Characteristics
Most patients were admitted to — and discharged from — the hospital internal medicine service (n = 475, 69%) (Figure 1). Otherwise, patients were discharged from surgical subspecialty services (e.g., general surgery, colorectal surgery; n = 66, 10%) or from nonsurgical, non–hospital medicine services such as cardiology or neurology (n =148, 21%). The number of home medications was recorded for these patients, with the mean being 5.8 (SD 5.5) medications.
Figure 1. Primary service at discharge for patients discharged against medical advice. HIM, hospital internal medicine.
A minority of patients were admitted to the ICU (n = 44, 6.5%) during hospitalization, with a mean length of stay of 4.5 days (SD 9.5). For all included patients, the mean hospital length of stay was 2.66 days (SD 4.53 days) (Table 2). Substance use disorder was the most common comorbidity, present in 153 patients (n = 153, 43%), including opioid use disorder (n = 110, 16%) and alcohol-related disorders (n = 70, 10%) (see Figure 1). Psychiatric comorbidities included anxiety (n = 202, 29%), depression (n = 115, 17%), and emotional disturbances (n = 65, 9.4%). Cardiovascular conditions also were common: hypertension (n = 246, 36%); coronary artery disease (n = 135, 20%); congestive heart failure (n = 134, 19%); and peripheral vascular disease (n = 21, 3.0%). Other notable comorbidities included type 2 diabetes (n = 113, 19%), alcoholic cirrhosis (n = 28, 4.1%), and chronic kidney disease (n = 146, 21%). Eighteen percent (n = 126) of these patients were dependent on life-sustaining devices (e.g., home oxygen, pacemakers, dialysis).
Discharge Documentation
In the majority of discharge encounters, patients signed an AMA form (n = 588, 86%) and had a discussion with a provider before discharge (n = 577, 84%). Specific elements included assessment of patient decision-making capacity (n = 322, 47%) documentation of risks of leaving (n = 460, 67%), reasons for leaving (n = 437, 64%) (Figure 2). Alternative treatment options were discussed in 38% (n = 265) of cases, and in a further 17% (n = 117) of cases, patients were informed they could return for further care if they chose.
In half of all cases, providers documented an offer to prescribe outpatient medications (n = 365, 53%); these were actually prescribed in 44% of discharges (n = 301), with a mean of 2.68 medications (SD 2.60). Follow-up appointments also were included in the discharge orders of 366 cases (53%), typically as generic instructions to “follow up with your primary care provider”. In 497 cases (73%), some form of discharge instruction was provided.
Readmission
Risk of readmission was assessed using the LACE+ index, which was 62 (SD 12), indicating a high 30-day unplanned readmission risk. In actuality, 30-day readmission occurred in 235 cases (34%), with a mean interval of 8.0 days (SD 9) between discharge and readmission (Figure 2).
Figure 2. Discharge against medical advice documentation and discharge planning.
Discussion
In this cohort of 690 patients who left AMA, we found gaps in discharge planning associated with a 30-day readmission rate of 34%. These findings, combined with a mean time to readmission of eight days, highlight systemic deficiencies in how healthcare systems manage AMA discharges, which may contribute to adverse patient experience during this volatile encounter and lead to poorer clinical outcomes in this population.
Consistent with prior literature, patients discharged AMA in our study were predominantly male (60%), younger, and often publicly insured (40% Medicare, 25% Medicaid) or uninsured (11%). They also were primarily hospitalized on nonsurgical services, usually hospital internal medicine. These findings suggest a complex interplay between age, disability, medical comorbidity, and AMA risk. We observed high rates of substance use disorder (43%) — particularly opioid use disorders (100 patients) and alcohol-related disorders (70 patients) — aligning with previous research identifying addiction as a major driver of premature self-discharge. One-fifth (20%) of patients had documented SDOH concerns, with food insecurity (7.2%) and transportation needs (2.3%) being most common. These SDOH vulnerabilities may act both to drive AMA discharges (e.g., a patient leaves before reaching clinical stability because she needs to return to work to make rent) and to limit attempts to ensure the safest possible discharge (e.g., lack of transportation limits follow-up appointment adherence).(24)
Our findings reveal significant variation in AMA discharge practices and documentation. Although 86% of patients signed an AMA form, only 47% had documented assessment of decision-making capacity. These inconsistencies are not unique to our institution and reflect the absence of standardized best practices for management of AMA discharge.(20,25) Inconsistencies also appear in discharge planning — only 44% of patients received discharge medications, while follow-up appointments were ordered for 53% of patients. Moreover, only 17% of patients were explicitly informed they could return for care if they changed their minds, representing a missed opportunity for harm reduction in this vulnerable population.
The mean LACE+ score of 62 (SD 12) in our cohort is consistent with a “high risk of readmission” and corresponds well to our increased 30-day readmission rate of 34%, which exceeds the 21% rate reported in recent national studies.(15,16,23) These readmissions, which occurred soon after discharge (on average, within eight days), may reflect patient adherence to a provider’s recommendation to return for ongoing care upon reconsideration after discharge. However, given that such a recommendation appeared in only 17% of AMA discharge documentation, the more likely reason for readmission is post-discharge clinical deterioration.
Study Limitations
This study has multiple limitations. The tertiary academic center from which this cohort was drawn serves a predominantly older, Caucasian population with higher-than-average socioeconomic status, all of which may limit generalizability. Although we collected data about admission and discharge diagnoses, our focus on physician behavior and documentation means that these were not analyzed in detail, which limits analysis of differences in patient and provider action in certain high-risk diagnoses (e.g., acute alcohol withdrawal). Also, readmission data for patients admitted outside of our hospital was unavailable and therefore was not included. The retrospective design and reliance on documentation restricted our ability to capture discussions or patient rationales for leaving that were not fully documented, as did the absence of logistic regression to identify significant correlations between characteristics in patients discharged AMA that might influence the risk of 30-day readmission. Future studies might adopt a prospective design and include matched controls of non–AMA discharge patients admitted for similar conditions, allowing for calculation of the comparative excess clinical risk attributable to AMA discharge.
Conclusion
AMA discharges are associated with significant social and medical complexity, with considerable variation in the clinical practice of discharge management. Key elements — such as assessment of decision-making capacity and medication prescription — are performed inconsistently, with a high associated 30-day readmission rate. Standardized AMA discharge processes focused on capacity assessment, shared decision-making, and optimal discharge planning — including medication prescription and follow-up appointment scheduling — may reduce adverse clinical outcomes and also reduce healthcare costs.
Acknowledgment: The authors are grateful to the Mayo Clinic Arizona for use of the data included in this research study.
Concept and design: AH, US. Acquisition, analysis or interpretation of the data: all authors. Drafting of the manuscript: AH, AQ, EG, ET, JZ. Critical review of the manuscript for important intellectual content: AH, CE, US. Statistical analysis: MEG, MT. Supervision: AH, CE, US.
The data that support the findings of this study are available from the corresponding author upon reasonable request. The data are not publicly available because of privacy or ethical restrictions.
This study was submitted to the Mayo Clinic Institutional Review Board but was deemed to pose minimal risk to participants and was exempt from IRB.
Requirement for informed consent was waived because of the retrospective nature of the study.
No previously published material was reproduced in this manuscript. All figures and tables are original.
This study was a retrospective chart review and not registered as a clinical trial. The authors acknowledge that they have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable.
References
Southern WN, Nahvi S, Arnsten JH. Increased risk of mortality and readmission among patients discharged against medical advice. Am J Med. 2012;125:594-602. doi:10.1016/j.amjmed.2011.12.017
Spooner KK, Salemi JL, Salihu HM, Zoorob RJ. Discharge against medical advice in the United States, 2002-2011. Mayo Clin Proc. 2017;92:525-535. doi:10.1016/j.mayocp.2016.12.022
Stranges E, Wier L, Merrill CT, Steiner C. Hospitalizations in Which Patients Leave the Hospital Against Medical Advice (AMA), 2007. HCUP Statistical Brief #78. August 2009. Rockville, MD: Agency for Healthcare Research and Quality. www.hcup-us.ahrq.gov/reports/statbriefs/sb78.pdf .
Horton T, Subedi K, Sharma RA, Wilson B, Gbadebo BM, Jurkovitz C. Escalation of opioid withdrawal frequency and subsequent AMA rates in hospitalized patients from 2017 to 2020. J Addict Med. 2022;16:725-729. doi:10.1097/adm.0000000000000997
Jaydev F, Gavin W, Russ J, et al. Discharges against medical advice: time to take another look. A retrospective review of discharges against medical advice focused on prevention. Hosp Pract (1995). 2023;51:288-294. doi:10.1080/21548331.2023.2287431
Pollini RA, Paquette CE, Drvar T, et al. A qualitative assessment of discharge against medical advice among patients hospitalized for injection-related bacterial infections in West Virginia. Int J Drug Policy. 2021;94:103206. doi:10.1016/j.drugpo.2021.103206
Stearns CR, Bakamjian A, Sattar S, Weintraub MR. Discharges against medical advice at a county hospital: provider perceptions and practice. J Hosp Med. 2017;12(1):11-17. doi:10.1002/jhm.2672
Huang SW. Discharge against medical advice. PSNet [internet]. May 1, 2005. https://psnet.ahrq.gov/web-mm/discharge-against-medical-advice .
Alfandre D. Reconsidering against medical advice discharges: embracing patient-centeredness to promote high quality care and a renewed research agenda. J Gen Intern Med. 2013;28:1657-1662. doi:10.1007/s11606-013-2540-z
Alfandre D, Schumann JH. What is wrong with discharges against medical advice (and how to fix them). JAMA. 2013;310:2393-2394. doi:10.1001/jama.2013.280887
Alfandre DJ. “I’m going home”: discharges against medical advice. Mayo Clin Proc. 2009;84:255-260. doi:10.4065/84.3.255
Green P, Watts D, Poole S, Dhopesh V. Why patients sign out against medical advice (AMA): factors motivating patients to sign out AMA. Am J Drug Alcohol Abuse. 2004;30:489-493. doi:10.1081/ada-120037390
Moy E, Bartman BA. Race and hospital discharge against medical advice. J Natl Med Assoc. 1996;88:658-660.
Ryus CR, Janke AT, Kunnath N, Ibrahim AM, Rollings KA. Association of hospital discharge against medical advice and coded housing instability in the US. J Gen Intern Med. 2023;38:3082-3085. doi:10.1007/s11606-023-08240-1
Choi M, Kim H, Qian H, Palepu A. Readmission rates of patients discharged against medical advice: a matched cohort study. PLOS ONE. 2011;6(9):e24459. doi:10.1371/journal.pone.0024459
Dhaliwal JS, Dang AK. Reducing hospital readmissions. StatPearls. https://www.ncbi.nlm.nih.gov/books/NBK606114/
Garland A, Ramsey CD, Fransoo R, et al. Rates of readmission and death associated with leaving hospital against medical advice: a population-based study. CMAJ. 2013;185:1207-1214. doi:10.1503/cmaj.130029
Tan SY, Feng JY, Joyce C, Fisher J, Mostaghimi A. Association of hospital discharge against medical advice with readmission and in-hospital mortality. JAMA Netw Open. 2020 Jun 1;3(6):e206009. doi:10.1001/jamanetworkopen.2020.6009
Alfandre D, Brenner J, Onukwugha E. Against medical advice discharges. J Hosp Med. 2017;12:843-845. doi:10.12788/jhm.2796
Trépanier G, Laguë G, Dorimain MV. A step-by-step approach to patients leaving against medical advice (AMA) in the emergency department. CJEM. 2023;25(1):31-42. doi:10.1007/s43678-022-00385-y
Ibrahim AM, Koester C, Al-Akchar M, et al. HOSPITAL Score, LACE Index and LACE+ Index as predictors of 30-day readmission in patients with heart failure. BMJ Evid Based Med. 2020;25:166-167. doi:10.1136/bmjebm-2019-111271
Jun-O’Connell AH, Grigoriciuc E, Silver B, et al. Association between the LACE+ index and unplanned 30-day hospital readmissions in hospitalized patients with stroke. Front Neurol. 2022;13:963733. doi:10.3389/fneur.2022.963733
van Walraven C, Wong J, Forster AJ. LACE+ index: extension of a validated index to predict early death or urgent readmission after hospital discharge using administrative data. Open Med. 2012;6(3):e80-90.
Yuan S, Ashmore S, Chaudhary KR, Hsu B, Puumala SE. The role of socioeconomic status in individuals that leave against medical advice. S D Med. 2018;71:214-219.
Riddick FA Jr. The code of medical ethics of the American Medical Association. Ochsner J. 2003 Spring;5(2)6-10.