Evaluating Care Complexity in Inflammatory Arthritis: A Population-Based Approach Using Classification and Regression Tree Analysis to Examine Health Care Utilization.

Abstract

Objective: We identified which care complexity characteristics contribute to high health care utilization (physician visits and emergency department [ED] visits) among individuals with inflammatory arthritis in Alberta.

Methods: We identified individuals diagnosed rheumatoid arthritis (RA), psoriatic arthritis (PsA), or axial spondyloarthritis (axSpA) between 2010 and 2015 using a validated case definition. Sociodemographic and disease characteristics contributing to care complexity were described for each population. Classification and regression tree models were developed to identify which combination of care complexity characteristics were predictive of high health care use (physician visits and ED visits) within the first five years after diagnosis. Care complexity predictors included age, sex, place of residence, socioeconomic status (SES), inflammatory arthritis disease type, comorbidities, medication use, and opioid and glucocorticoid dispensations.

Results: Care complexity characteristics varied across cohorts (n = 13,170 individuals): older age, rural residence, multimorbidity, and glucocorticoid use were most prevalent in RA; low SES was most prevalent in PsA; and opioid use was most prevalent in axSpA. High physician visit utilization was associated with polypharmacy, the presence of anxiety, older age, living in an urban area, and having PsA (sensitivity of 78%). High ED visit utilization was associated with polypharmacy, rural place of residence, and having RA or axSpA (sensitivity of 68%).

Conclusion: Our models identified key predictors of high health care utilization in inflammatory arthritis, with modest predictive performance. These findings provide insight into factors contributing to care complexity and may support health care providers to better understand patient needs and inform resource planning.