Evaluating Care Complexity in Inflammatory Arthritis – A Population-Based Approach using CART Analysis to Examine Healthcare Utilization

Autoimmune inflammatory arthritis is one of the most prevalent chronic health conditions in Canada, affecting approximately 3% of the population. Common inflammatory arthritis conditions include rheumatoid arthritis (RA), psoriatic arthritis (PsA), and ankylosing spondylitis (AS). Individuals with inflammatory arthritis often experience high care complexity, characterized by multiple comorbidities, polypharmacy, social factors, and diverse healthcare needs, which can influence patterns of healthcare utilization. The aim of this study is to identify care complexity factors predictive of high healthcare utilization, including physician and emergency department (ED) visits, among adults with inflammatory arthritis in Alberta. We identified individuals diagnosed with RA, PsA, or AS using a validated administrative case definition. Sociodemographic and disease-related factors contributing to care complexity were described for each cohort, including age, sex, place of residence, socioeconomic status (SES), comorbidities, disease type, medication use, and opioid and glucocorticoid dispensations. Classification and regression tree (CART) models were developed to identify combinations of these care complexity characteristics that were predictive of high healthcare utilization. The study included 13,170 individuals. Care complexity varied across disease cohorts: older age, rural residence, multimorbidity, and glucocorticoid use were most common among people with RA; low SES was most prevalent in PsA; and opioid use was highest in AS. Polypharmacy, the presence of anxiety, older age, urban residence, and having PsA were predictive of high physician visits while polypharmacy, rural residence, and having RA or AS were predictive of high ED visits. These findings highlight distinct patterns of healthcare utilization driven by specific care complexity factors, offering insight into patient stratification and potential targets for interventions to optimize care delivery in inflammatory arthritis. However, the models were not fully able to capture or predict all aspects of complexity, underscoring the need for additional data and/or methods to more fully understand care complexity in inflammatory arthritis. Advancing this work could lead to more accurate prediction, better personalized care, and ultimately, improved health outcomes.