Medication Nonadherence in Rheumatic Diseases

Abstract

Medication nonadherence is the largest driver of avoidable healthcare costs in the United States. More than 40 percent of patients with rheumatic diseases are nonadherent. Among patients with systemic lupus erythematosus (SLE), adherence to hydroxychloroquine (HCQ), the standard of care medication, is especially poor. This dissertation defined predictors of nonadherence to HCQ among Medicaid beneficiaries with SLE and developed a rheumatology-specific intervention to address adherence-related barriers. In Paper 1, group-based trajectory models were used to explore longitudinal patterns of HCQ nonadherence among adult SLE patients in Medicaid. Multinomial logistic regression models allowed for the assessment of predictors of nonadherent trajectories compared to the most adherent. In Paper 2, hierarchical multilevel logistic regression models were used to describe the association between area-level sociodemographic variables, health resource concentration and adherence behavior. In Paper 3, an intervention involving a patient navigator- a layperson trained in advocacy, care coordination and basic rheumatology- was designed, implemented and evaluated comparing baseline and post-intervention adherence. In Papers 1 and 2, we identified more than 10,000 Medicaid beneficiaries with SLE who newly initiated HCQ. A four-group trajectory model provided the best fit for the data with a persistently adherent trajectory (17%), two intermediate trajectories (47%), and a persistently nonadherent trajectory (36%). A number of factors were associated with nonadherent patterns including black race, Hispanic ethnicity, younger age, increased SLE-related comorbidities and antidepressant medication use. In Paper 2, we added contextual variables and found an association between higher zip code-level percent black and higher odds of nonadherence, after adjusting for individual-level factors. Residing in counties with more hospitals per capita decreased the odds of nonadherence; living in a health professional shortage area increased the odds. In Paper 3, we did not find a significant improvement in medication adherence 6-months following the patient navigator intervention, however medication-related concerns were reduced. Overall, adherence was poor among Medicaid beneficiaries with SLE. Individual and contextual factors were associated were nonadherence, however there were no dominant predictors that could explain the complexity of the behavior. Additional interventions are needed particularly among the most vulnerable rheumatology patients, to better understand and improve adherence.