Multidimensional neighborhood social drivers of health and pediatric hospital readmission risk.

BACKGROUND

Social drivers of health (SDoH) are associated with pediatric hospital readmissions, but the contribution of neighborhood-level measures remains unclear.

OBJECTIVE

Evaluate associations between neighborhood-level SDoH, including the Child Opportunity Index (COI), Social Vulnerability Index (SVI), and Agency for Healthcare Research and Quality (AHRQ) SDoH variables, and pediatric readmission.

PATIENTS AND

METHODS

This single-center retrospective study included 39,930 encounters of patients aged 1 month to 18 years from February 2019 through December 2023. Encounters were linked to census tract-level SDoH data. Principal component analysis (PCA) was used to reduce AHRQ SDoH variables to 10 components. Logistic regression evaluated associations with 7-day and 30-day unplanned readmissions.

RESULTS

The median age at encounter was 6.7 years (interquartile range, 2.0-12.1 years); 54.7% of patients were Hispanic, and 62.5% had public insurance. Unplanned readmissions occurred in 3.8% of encounters within 7 days and 8.6% within 30 days. Neither COI nor SVI was associated with either readmission outcome in unadjusted or adjusted analyses (all p > .05). PCA-derived neighborhood constructs identified associations not captured by conventional indices. Military-Associated Community Environment was independently associated with both 7-day and 30-day readmissions, and multiple neighborhood dimensions modified associations between clinical factors and readmission risk.

CONCLUSIONS

Conventional neighborhood-level indices were not associated with pediatric readmission risk. However, PCA revealed multidimensional neighborhood characteristics associated with readmissions, primarily through interactions with specific clinical conditions. These findings suggest that neighborhood context influences pediatric readmission by modifying disease-specific clinical vulnerability rather than through uniform effects across all patients.