Clustering and Healthcare Costs With Multiple Chronic Conditions in a US Cross-Sectional Study
Objective To investigate healthcare costs and contributors to costs for multiple chronic conditions (MCCs), common clusters of conditions and their impact on cost and utilisation. Methods This was a cross-sectional analysis of US financial claims data representative of the US population, including Medicare, Medicaid and Commercial insurance claims in 2015. Outcome measures included healthcare costs and contributors; ranking of clusters of conditions according to frequency, strength of
... rength of association and unsupervised (k-means) analysis; the impact of clustering on costs and contributors to costs. Results Of 1,878,951 patients, 931,045(49.6%) had MCCs, 56.5% weighted to the US population. Mean age was 53.0 years(SD16.7); 393,121(42.20%) were male. Mean annual healthcare spending was $12,601, ranging from $4,385 (2 conditions) to $33,874 (11 conditions), with spending increasing by 22-fold for inpatient services, 6-fold for outpatient services, 4.5-fold for generic drugs and 4.2-fold for branded drugs. Cluster ranking using the 3 methodologies yielded similar results: highest ranked clusters included metabolic syndrome(12.2% of US insured patients), age related diseases(7.7%), renal failure(5.6%), respiratory disorders(4.5%), cardiovascular disease(CVD)(4.3%), cancers(4.1-4.3%), mental health-related clusters(1.0-1.5%) and HIV/AIDS(0.2%). Highest spending was in HIV/AIDS clusters ($48,293), mental health-related clusters ($38,952-$40,637), renal disease ($38,551) and CVD ($37,155); with 89.9% of spending on outpatient and inpatient care combined, and 10.1% on medication. Conclusion and Relevance Over 57% of insured patients in the US may have MCCs. MCC Clustering is frequent and is associated with healthcare utilisation. The findings favour health system redesign towards a multiple condition approach for clusters of chronic conditions, alongside other cost-containment measures for MCCs.