Title : Discharges against medical advice in patients with opioid use disorder in an urban community hospital
Abstract:
Introduction: In 2021, New York City accounted for 43% of all opioid overdose-related visits to emergency departments across New York State. Due to inadequate early intervention strategies and various psychosocial factors related to opioid use disorder (OUD) management, opioid dependence and withdrawal contribute to discharges against medical advice (AMA). We hypothesize that patients with OUD have an increased risk of leaving AMA compared to patients without OUD.
Methods: We conducted a retrospective chart review of medical, surgical and obstetric discharges from January 2022 to December 2023 at Harlem Hospital to determine the prevalence of OUD diagnosis in AMA versus non-AMA discharges. We performed a case-control analysis to compare the demographics of OUD/AMA discharges versus non-OUD/AMA discharges.
Results: The total number of hospital discharges was 19,699 in 2022 and 2023, and the total number of patients with OUD was 1,522. The total number of AMA discharges was 1,599 (8.1% of total discharges). The risk of AMA was significantly higher in patients with OUD (23.8%) compared to patients with no OUD diagnosis (6.8%) (OR = 4.2, 95% CI: 3.74 - 4.87; p < 0.0001). We compared OUD/AMA discharges (cases, n = 362) versus non-OUD/AMA discharges (controls, n = 108). A t-test showed that cases were significantly younger than controls (49.8±12.9 years vs. 56.7±13.6 years; p < 0.0001). A chi-squared test showed that a smaller proportion of cases were domiciled compared to controls (32.0% vs. 60.2%; p < 0.0001).
Discussion: Among patients with OUD, a smaller proportion of older patients left AMA, possibly due to greater mobility issues and the burden of medical illness. Additionally, in patients with OUD, unhoused status was associated with a higher likelihood of leaving AMA, indicative of influence of psychosocial factors. Specific risk factors for patients with OUD leaving AMA need to be identified, as these may contribute to discomfort that fuels patients' decisions to leave AMA, leading to negative health outcomes. Therefore, future studies could examine potential proxies for the management of opioid withdrawal, such as missing or delaying urine toxicology, Clinical Opiate Withdrawal Scale (COWS) score, confirmation of methadone doses, and the provision of opioid and non-opioid medications for managing withdrawal symptoms. Limitations of this study include low power and limited generalizability due to the relatively small sample size in one hospital setting. We did not observe a significant racial disparity, possibly due to the relatively homogeneous population at the site. Additionally, seasonality may have been a confounder, as the control group was admitted between January and March 2023, while the cases were from the entirety of 2022 and 2023. However, the number of patients leaving AMA per month was relatively consistent (p = 0.97).
Conclusion: This study illustrates that patients with OUD are more likely to leave AMA than patients without OUD, increasing the risk of poor medical outcomes for this vulnerable group. Future directions include examining specific risk factors for AMA discharge in patients with OUD and collaborating with multiple sites to increase sample size and generalizability.
