Abujamra, Ramzi2015-03-312015-03-312014-12https://hdl.handle.net/11299/170876University of Minnesota Ph.D. dissertation. December 2014. Major: Health Informatics. Advisor: Stephen T. Parente, PhD. 1 computer file (PDF); x, 159 pages, appendices A-F.This dissertation will explore the factors that influence development of Diabetes complications for Medicare and dual eligible beneficiaries, for three Diabetes complications: retinopathy, nephropathy and neuropathy. Both predictive and explanatory models are explored. Predictive models focus on finding factors most predictive of Diabetes complications among Medicare and dual eligible beneficiaries. Explanatory models seek to answer the three hypothesis of this study. The first hypothesis states that higher treatment investment is associated with lower rates of Diabetes complications in Medicare and dual eligible beneficiaries. The second hypothesis states that physicians who are specialists (vs. primary care) and urban (vs. rural) are associated with lower rates of Diabetes complications among Medicare and dual eligible beneficiaries. Finally, the third hypothesis associates higher patient total cost sharing with improvement in Diabetes complications outcomes among Medicare and dual eligible beneficiaries. For dual eligible beneficiaries, patient cost sharing is defined as state Medicaid investment per beneficiary for the state where each beneficiary resides in. The results for the predictive models are strongest for nephropathy complication, and weakest for retinopathy complication. The results for the explanatory models show that for the first hypothesis, nephropathy has lower rate of Diabetes complication for higher total treatment investment. For the second hypothesis, rural providers have lower rate of Diabetes complications for nephropathy (non-dual beneficiaries) and neuropathy (for dual beneficiaries). Also, primary care providers have lower rates of Diabetes complication for retinopathy and neuropathy (non-dual beneficiaries) and retinopathy (dual beneficiaries). For neuropathy, specialists have lower rates of Diabetes complications (for non-dual beneficiaries). Finally, for the third hypothesis, no complications are associated with lower Diabetes complication rates with higher patient total cost sharing for non-dual beneficiaries. For dual beneficiaries, retinopathy and nephropathy (to a lesser extent) show evidence of lower Diabetes complication rates with higher State Medicaid investment per beneficiary. Model performance results based on the C-statistic are moderate overall, with nephropathy showing the best performance and retinopathy the lowest performing among all of the Diabetes complications.enDiabetes complicationsHealth informaticsLogistics regressionMedicaid dual eligibilePredictiveRisk adjustmentHealth informaticsThesis or Dissertation