Medicaid risk adjustment model with diagnosis and pharmacy-based adjusters: Does it work?
National health expenditures will continue to grow faster than nominal gross domestic product (GDP) in the early 21st century (Heffler et al., 2002; Heffler et al., 2005). Increased Medicaid costs have spurred research to find reliable cost-saving methodologies (Kronick et al., 1996). The Medicaid a...
| Main Author: | |
|---|---|
| Format: | Others |
| Published: |
Scholar Commons
2007
|
| Subjects: | |
| Online Access: | http://scholarcommons.usf.edu/etd/2261 http://scholarcommons.usf.edu/cgi/viewcontent.cgi?article=3260&context=etd |
| Summary: | National health expenditures will continue to grow faster than nominal gross domestic product (GDP) in the early 21st century (Heffler et al., 2002; Heffler et al., 2005). Increased Medicaid costs have spurred research to find reliable cost-saving methodologies (Kronick et al., 1996). The Medicaid administrations of some states have chosen risk adjustment as a methodology for savings (Tollen et al., 1998; Weiner et al., 1998), since it can reduce the financial burden of health care providers and distribute medical resources more efficiently. This dissertation presents a risk-adjustment model based on two types of health condition adjusters: diagnosis-based HCC adjusters and pharmacy-based RxRisk adjusters. HCC adjusters were developed from different diagnostic categories from inpatient, outpatient and long-term care data. RxRisk adjusters included diseases inferable from prescription drug usage.
The underlying assumption is that using both types of health condition adjusters, rather than relying on either diagnosis-based adjusters or pharmacy-based adjusters alone can help increase predictive power and lower Medicaid's risk of reimbursing inflated medical costs for its beneficiaries. The population in this study consisted of all disabled and aged Florida adults who were eligible for Florida's Medicaid program in state fiscal year (SFY) 2002-03 and state fiscal year 2003-04.
The population was broken down into two subpopulations: disabled Medicaid beneficiaries aged 64 and under and beneficiaries aged 65 or over.The proposed regression model includes diagnostic and pharmacy-based adjusters, and this dissertation compares the proposed model with models based solely on pharmacy- or diagnosis-based adjusters.The results presented in this dissertation demonstrate the proposed model has higher predictive power than the diagnosis-based HCC model and the pharmacy-based RxRisk model for the overall population and the subpopulations in this study. Risk-adjustment models using diagnostic and prescription drug information have higher predictive power and decrease the possibility of inappropriate gaming of the Medicaid capitation payment system. |
|---|