OBJECTIVES: Blue Cross and Blue Shield of Louisiana (BCBSLA) analyzed Emergency Department (ED) utilization among race/ethnicity groups to better understand current practices across the state. Researchers observed higher ED utilization among Black/African American members, which led to an investigation of drivers such as Social Vulnerability Index (SVI), medication adherence, health conditions (asthma, COPD, coronary artery disease, chronic kidney disease, end-stage renal disease, hypertension, diabetes, mental health), and demographics (gender, parish). BCBSLA studied its Medicare Advantage (MA) and commercial populations.

METHODS: BCBSLA developed a model-based clustering methodology using data from 2022 that employed Partitioning Around Medoids and Gower distance matrix. Researchers descriptively analyzed healthcare utilization, SVI, and health conditions within each cluster. Independent predictors of respective cluster membership used a Logistic Regression model, which assesses the importance of drivers within clusters. Researchers spatially analyzed clusters for estimating access to health services such as a primary care provider, urgent care, and emergency department.

RESULTS: Using MA and commercial data, the clustering methodology identified 16 distinct clusters within each race/ethnicity that had common characteristics. Among the notable clusters: Cluster 1 was a cohort of White, commercial members with socioeconomic challenges such as no high school diploma, and disabilities. Cluster 2 was a cohort of Black/African American, female, commercial members with hypertension who had high housing burden and were below 150% of poverty level. Cluster 3 consisted of White, male, MA members with diabetes and high housing cost burden and limited knowledge of English. Cluster 4 consisted of Black/African American, male, MA members with diabetes who had socioeconomic challenges like single-parent household and housing burden.

CONCLUSIONS: The clustering method identified race-based cohorts with common health conditions such as hypertension and diabetes, along with higher ED utilization rates and socioeconomic challenges. This clustering analysis can support intervention strategies to improve health outcomes.