ECONOMIC IMPACT ANALYSIS AND PREDICTIVE MODELING IDENTIFYING RISK FACTORS FOR OPIOID USE DISORDER IN MEDICAID MANAGED CARE POPULATIONS
Author(s)
Jones J1, Gao W2, Keleti D1, Chen Y1, Mistry P1
1AmeriHealth Caritas, Philadelphia, PA, USA, 2AmeriHealth Caritas, phildelphia, PA, USA
Presentation Documents
OBJECTIVES: To report on: 1) an economic impact analysis (EIA); and 2) predictive models identifying the primary risk factors for opioid use disorder (OUD) in adult Medicaid managed care populations. METHODS: 1) Our EIA compared medical costs of an opioid user cohort against a control non-user cohort, both in southeastern Pennsylvania. The control group was matched for demographic variables (e.g., age, gender, and race). HEDIS® criteria were applied to define opioid and substance use disorder (SUD). 2) Stepwise logistic regression was used to develop predictive models to evaluate the risk of developing OUD. The model risk factors included demographics, social determinants of health (SDoH), physical and behavioral health conditions, and prior substance use history. Receiver operating characteristic curves and concordance index statistics evaluate overall model adequacy. RESULTS: 1) EIA analysis found that total combined medical and pharmacy costs were 139% higher in opioid users than non-users. Cost differentials were pronounced for inpatient, outpatient, and professional services (342%, 169%, and 127% higher), the three service categories combined (209% higher), and pharmacy expenditures (50% higher). 2) Predictive models identified multiple risk factors for OUD and SUD. Prior substance use was the greatest risk factor (ten-fold higher than non-SUD cohort) and was removed in the subsequent models to uncover other opioid use predictors, including mental health disorders (odds ratios ( OR); bipolar disorder, 1.56; depression, 1.67; schizophrenia, 1.33; anxiety 1.37), sickle cell disease (OR, 1.62), HIV (OR, 1.64), and tobacco use (OR, 4.57). Additionally, SDoH barriers cause a 24% increased risk (OR, 1.24) for developing SUD. CONCLUSIONS: 1) The economic impact of OUD is significant. Opioid users have markedly higher medical and pharmacy costs than nonusers. 2) Predictive modeling showed that prior substance use was, by far, the greatest predictor of OUD, followed by tobacco use, HIV, sickle cell disease, mental health issues, and SDoH.
Conference/Value in Health Info
2019-05, ISPOR 2019, New Orleans, LA, USA
Value in Health, Volume 22, Issue S1 (2019 May)
Acceptance Code
AI4
Topic
Economic Evaluation, Epidemiology & Public Health, Methodological & Statistical Research
Topic Subcategory
Artificial Intelligence, Machine Learning, Predictive Analytics, Cost-comparison, Effectiveness, Utility, Benefit Analysis, Public Health
Disease
Drugs, Mental Health