Using Big Data to Examine Healthcare Quality and Outcomes: Claims Data Illuminate Association Between Behavioral Health and Preventable Hospitalization in Diabetic Patients




Polcar, Leah
Miller, Thaddeus


Journal Title

Journal ISSN

Volume Title



Objective: High quality outpatient care for individuals with diabetes reduces the likelihood of acute diabetes-related complications and associated potentially preventable hospitalization (PPH). Comorbid behavioral health (BH) conditions can be a barrier to high quality outpatient diabetes care and may contribute to PPHs, but the association between BH comorbidities and PPHs has not been well-studied. We sought to determine if comorbid BH conditions are associated with an increased likelihood of diabetes-related PPHs. Study Design: We used a multivariable negative binomial-logit hurdle regression model to determine whether BH conditions were associated with diabetes-related PPHs as defined by the Agency for Healthcare Research and Quality. Covariates included sociodemographic and other comorbid medical condition variables. Materials: A national sample of medical and pharmacy claims data from the Optum Impact Research Database representing commercial insurer-covered healthcare received between 2011 and 2013 for 4,000,000 people. A total of 229,039 individuals met inclusion criteria (diagnosed/treated diabetes, aged 20-64, and complete data) and were included in analysis. Results: Claims effectively identified increased risk for PPH among diabetic patients with BH comorbidities. 20.7% had ≥1 BH condition and the risk for multiple PPHs increased as individuals’ counts of BH conditions increased. Schizophrenia, mood disorders, alcohol use disorders, and substance use disorders were each independently associated with increased risk of ≥1 PPH and an increasing number of PPHs. Conclusions: People with diabetes and comorbid BH conditions have a higher likelihood of and volume of PPHs. The results suggest that enhanced treatment approaches or improved care quality may be useful to improve health and other outcomes for this population. Claims data provide an accessible and effective approach to evaluation. Implications: Targeted interventions including case management, home health services, pharmacy management, or other structural enhancements may reduce hospitalizations in persons with comorbid diabetes and BH conditions. Given the significant proportion of diabetic patients with ≥1 diagnosed BH comorbidity, integrating diabetes care into BH treatment may drive improved health outcomes and cost savings. Evaluations of quality improvement activities focused on this population should consider using PPHs as outcome measures and claims as a data source.