Back

Leaving against medical advice as a signal of unmet care needs in adult sickle cell disease hospitalizations

Zhilkova, A.; Rivlin, K.; Jackson, J.; Glassberg, J.; McCrary, B.; Eyssallenne, A.

2026-03-24 health systems and quality improvement
10.64898/2026.03.20.26348715 medRxiv
Show abstract

Importance: Sickle cell disease (SCD) affects approximately 100,000 people in the United States, causes life-threatening complications, and shortens life expectancy by decades. Adults with SCD routinely encounter undertreated pain, provider bias, and structural barriers in hospital settings. Objective: To describe patterns of leave against medical advice (LAMA) among adults hospitalized for SCD. Design, Setting, and Participants: Retrospective analysis of inpatient discharge records among adults ages 18 and older in New York City hospitals, 2022-2023, hospitalized for SCD or any reason. Main Outcomes and Measures: The primary outcome was hospital-level LAMA, measured by crude rates and rates adjusting for patient characteristics using Bayesian hierarchical models. The secondary outcome was 30-day all-cause readmissions, stratified by LAMA status. Results: LAMA discharges comprised 14% of SCD hospitalizations and 4% of all-cause hospitalizations. Adjusted hospital-level SCD LAMA ranged from under 5% to 30% (IQR: 10-20%) and was higher than all-cause LAMA in most facilities. Crude SCD LAMA rates exceeded 30% in several hospitals, including those with more than 100 SCD hospitalizations during the study period. Patients with 10 or more SCD hospitalizations accounted for 40% of total SCD volume. Sensitivity analyses accounting for this concentration showed attenuated but persistent variation in SCD LAMA. Over 50% of SCD LAMA discharges were followed by a 30-day readmission compared to 38% of non-LAMA discharges. LAMA was associated with higher adjusted odds of readmissions in both SCD and all-cause hospitalizations. Conclusions: Our findings challenge the assumption that patients are solely responsible for early departures. Leaving against medical advice should be monitored as a signal of unmet care needs in SCD.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.