Hydroxychloroquine and immunosuppressant adherence patterns and their association with subsequent hospitalization rates among children with systemic lupus erythematosus.

OBJECTIVES

Using a representative sample of children with systemic lupus erythematosus (SLE) in the United States, we characterized prescription claim-based hydroxychloroquine and immunosuppressant adherence estimates and evaluated their concurrent and predictive validity.

METHODS

We identified children ages 5-18 with SLE in the Truven Health MarketScan® Commercial and Medicaid claims databases (2013-2018). Among new users of hydroxychloroquine and immunosuppressant medications, we calculated proportion of days covered (PDC) over 365 days to estimate adherence by user group (mycophenolate, azathioprine, methotrexate, and any immunosuppressant use). Agreement between adherence estimates was evaluated with intraclass correlation coefficients (ICC) and kappa statistics. Separate negative binomial regression models were used to estimate associations between (a) hydroxychloroquine, (b) immunosuppressant, or (c) concurrent immunosuppressant/hydroxychloroquine non-adherence and subsequent hospitalizations, adjusted for baseline demographics, disease severity, and healthcare utilization.

RESULTS

Among 423 new hydroxychloroquine/immunosuppressant users, 63% were Medicaid recipients. Sufficient adherence (PDC≥80%) ranged from 33 to 45% for immunosuppressants vs. 51-52% for hydroxychloroquine. Agreement between hydroxychloroquine and immunosuppressant adherence was modest overall, but better for mycophenolate (ICC 0.55) than methotrexate (0.27). Hydroxychloroquine non-adherence was associated with a 2.9-fold higher incidence of subsequent hospitalizations (95% CI [1.2-7.1]), whereas immunosuppressant and concurrent non-adherence were associated with 5.9 [2.4-14.6] and 5.6-fold [2.0-15.5] increased incidence, respectively. Use of concurrent adherence improved upon estimation of hospitalization risk compared to hydroxychloroquine adherence, but not immunosuppressant adherence alone.

CONCLUSIONS

Hydroxychloroquine adherence is an imperfect proxy for adherence to other lupus medications among children with SLE, and therefore assessing immunosuppressant adherence concurrently adds value to hydroxychloroquine adherence assessments. Prescription claims-based immunosuppressant adherence measures are predictive of acute care utilization and may inform population management strategies.