AUTHORS
Daniel Sop, PhD, Yue May Zhang, MS2, Wally Smith I, MD3
AFFILIATIONS
1 Virginia Commonwealth University, Richmond, Virginia, United States
2 Virginia Commonwealth University, Richmond, Virginia, United States
3 Virginia Commonwealth University, Richmond, Virginia, United States
ABSTRACT
Background
Thirty-day hospital readmission remains a major contributor to healthcare utilization among individuals with sickle cell disease (SCD) and is widely recognized as an important indicator of healthcare quality. Although both social and clinical factors likely contribute to readmission risk, the relative influence of insurance status, neighborhood socioeconomic disadvantage, and hospitalization characteristics has not been well characterized at the population level. We evaluated demographic, socioeconomic, insurance, and clinical factors associated with 30-day readmission among hospitalized patients with SCD across Virginia.
Methods
We performed a retrospective statewide analysis of 26,742 hospital encounters among patients with SCD in Virginia between 2012 and 2018, excluding in-hospital deaths. The primary outcome was all-cause 30-day hospital readmission. Unadjusted logistic regression models evaluated associations between readmission and payer type, neighborhood socioeconomic status (ZIP score), major diagnostic category (MDC), admission type, admission source, discharge disposition, and age. To account for multiple hospitalizations within individuals, sensitivity analyses were performed using generalized estimating equation (GEE) models adjusting for age, sex, race, and either payer type or neighborhood socioeconomic status, followed by a combined model including both variables.
Results
Among 26,742 hospital encounters, 27.6% resulted in a 30-day readmission. Insurance status demonstrated one of the strongest associations with readmission (overall P<0.0001). Compared with self-pay patients, Medicaid (OR 2.27, P<0.0001) and Medicare (OR 2.61, P<0.0001) were associated with more than twofold higher odds of readmission, while private insurance was also associated with increased odds (OR 1.42, P<0.0001). Neighborhood socioeconomic status demonstrated a graded relationship with readmission. Compared with patients residing in the most disadvantaged communities, those living in the least disadvantaged neighborhoods had substantially lower odds of readmission (ZIP score 1: OR 0.47, P<0.0001; ZIP score 2: OR 0.65, P=0.0027). Clinical characteristics also influenced readmission risk. Hospitalizations for blood disorders were associated with higher odds of readmission than respiratory conditions (OR 1.72, P<0.0001), whereas musculoskeletal, neurologic, digestive, and other diagnostic categories demonstrated lower odds. Emergency admissions were associated with significantly greater readmission risk than urgent admissions (OR 1.56, P<0.0001), and clinic admissions were less likely to be followed by readmission (OR 0.58, P<0.0001). Increasing age was associated with a 2.4% increase in readmission odds per year (OR 1.024, 95% CI 1.021–1.026; P<0.0001). In GEE sensitivity analyses accounting for repeated hospitalizations, public insurance remained independently associated with higher readmission risk, while residence in less socioeconomically disadvantaged neighborhoods remained protective, although the magnitude of the association was attenuated after adjustment. Female sex was associated with lower readmission risk in adjusted analyses.
Conclusions
More than one in four hospitalizations among patients with SCD resulted in 30-day readmission. Public insurance and neighborhood socioeconomic disadvantage were among the strongest and most consistent predictors of readmission, even after accounting for repeated hospitalizations. These findings highlight the substantial contribution of social and structural determinants of health to hospital utilization in SCD and support strategies that combine clinical risk assessment with targeted interventions addressing healthcare access and socioeconomic disadvantage to reduce preventable readmissions.
DISCLOSURES
Daniel Sop, PhD: No relevant conflicts of interest to declare.