TITLE: Impact of Medication Adherence on Hospital Admissions among Medicaid patient populations with Heart Failure.
To examine the impact of adherence to chronic disease medications on hospital admissions among Medicaid enrollees with heart failure.
Admissions encounters, diagnosis, medications, and Insurance eligibility datasets were pulled from NYU Langone Health system from 2017 to 2020. The dataset of Medicaid recipients consisted of 18 – 64 ages with Heart Failure and in addition, >= 1 of 7 chronic conditions:1) congestive heart failure 2) hypertension 3) dyslipidemia 4) diabetes 5) asthma/COPD 6) depression 7) schizophrenia/bipolar. Methods: Poisson regression machine learning model was used to predict the association between medication adherence (based on categorical Proportion of Days Covered – PDC) and 2 dependent variables: number of hospital admissions, total number of comorbidities. In addition, other machine learning models such as Logistic Regression and unsupervised learning methods were also used to do comparative analysis.
Full adherence was associated with fewer hospital admissions among those Medicaid patients with heart failure and any of the 7 comorbidities – congestive heart failure, hypertension, diabetes, asthma/COPD, depression and schizophrenia/bipolar. In addition, Negetive Binomial Regression model showed better results than Poisson Regression model which was used in the model study.
Substantial reduction in hospital admission and therefore, healthcare cost reductions can be achieved with increased medication adherence among Medicaid patient population. Patients at below 0.80 threshold can benefit from medication adherence intervention programs such as chronic care management or pharmacotherapy to achieve medication adherence greater than 0.80 threshold.