Hot spotting preventable hospitalizations to inform clinical practice and improve patient outcomes

Introduction: Hospitalizations for ambulatory care sensitive (ACS) conditions are considered potentially avoidable with improved care and disease management in the community. However, clinical practices continue to intervene at the individual-level, rarely addressing local community factors that may influence population risk for hospitalization. This study evaluates spatial clustering of preventable hospitalizations for diabetes and chronic obstructive pulmonary disease (COPD)/asthma among adults residing in the Bronx, NY to identify high-risk local geographic areas.

Methods: Using 2012-13 inpatient data from a major Bronx medical center’s electronic health records, adult hospitalization rates for two ACS conditions, (1) diabetes long-term complications (n=2,722) and (2) COPD/asthma (n=4,177), are calculated at the US Census block group level and small area “hot spots” are mapped for both unique admission rates and super-user counts (patients with ≥2 readmissions). Spatial regression models are specified to assess the relationship between preventable hospitalization rates and neighborhood composition.

Results: Spatial analysis reveals significant spatial autocorrelation in both diagnoses; clusters of unique admissions and super-utilizers overlap in several areas, particularly in the northeast Bronx. Higher rates for diabetes and COPD/asthma are significantly associated (p<.05) with older neighborhood age composition, yet only diabetes is related to greater proportions of Hispanic and black residents.

Conclusions: The observed clustering of admission rates and super-utilizer counts for diabetes and COPD/asthma suggests a unique opportunity to target vulnerable communities to reduce unnecessary hospitalization. Future efforts focused on investigating contextual neighborhood factors that drive these localized hot spots will help clinicians and administrators develop community interventions to improve patient outcomes.