Final report prepared for Center for Health Care Management Studies. RTI Project No. 08490.006. Washington, DC, USA: RTI International and Boston University School of Medicine.
RTI International and Boston University School of Medicine, USA
The Military Health System is examining new approaches to financing the care of covered beneficiaries. Since population health care costs depend on illness burden (case-mix), efficient financing should recognize variations in the illness burden of populations seen by different providers. Several mature risk adjustment systems that extract illness burden profiles from computerized encounter records are now used by Medicare, Medicaid, the Veteran’s Health Administration, commercial insurers in the United States, and international stakeholders to understand and manage health care delivery systems. These systems are designed for “population-based” management; “units” are person-years of medical care, and the key outcome is the cost, for each person, of a year of care. The expected cost for a provider for a year is the sum of individual-level person-year estimates for the population served.
We conducted side-by-side testing of a simple age-sex method similar to what is currently used in TRICARE Prime and four claims-based risk adjustment models: Adjusted Clinical Groups (ACGs), Chronic Disease and Disability Payment System (CDPS), Clinical Risk Groups (CRGs), and Diagnostic Cost Groups (DCGs). All four risk adjustment models use a person’s age, sex, and the morbidities recorded (in ICD-9-CM codes) during a year to predict total costs (inpatient + outpatient + pharmacy) for the next year. In addition, the CRG model used limited information on dates of service, place of service, and procedures. We applied each model to the TRICARE Prime population, measuring its overall ability to predict future costs and the concordance between model-measured needs and actual health care spending in policy-relevant subgroups of TRICARE Prime enrollees. We also examined the data quality and credibility for use in monitoring and managing cost of care.
We used administrative data on enrollment and claims for TRICARE Prime enrollees in fiscal years (FY) 2001 and 2002. Our study sample included (all 2.3 million) TRICARE Prime enrollees under the age of 65 years who were continuously enrolled during for the 24 months of FY 2001 through FY 2002 and residing in the continental United States, Hawaii, or Alaska. We simulated the real-world experience of building a model that is applied to new data by fitting the models to an estimation sample of 1.8 million and testing how closely their predictions matched up with actual costs in the remaining 500,000, the validation sample.
Please log in/register to access.