Health Care Financing Administration 7500 Security Blvd., Baltimore, MD. 21244-1850

A -- RISK ADJUSTMENT RESEARCH USING MEDICARE CURRENT BENEFICIARY SURVEY (MCBS) DATA" POC Renee A. Wallace, Contract Specialist, (410) 786-5128 The Health Care Financing Administration (HCFA) intends to award under Simplified Acquisition Procedures, a firm fixed price award, on a sole source basis to Dr. Leonard Gruenburg at DataChron Health Systems, Inc., 763 Massachusetts Avenue 7, Cambridge, MA 02139. This award will enable HCFA to extend its knowledge in risk-adjustment research by incorporating information from the Medicare Current Beneficiary Survey data and Area Resource File. The contractor has performed extensive work in this area for HCFA using the Medicare Current Beneficiary Survey (MCBS). Leonard Gruenberg, Ph.D. is an experienced researcher in the field of survey-based risk adjustment methodology and cost of long-term care populations. He's also familiar with the MCBS data, and produced reports on wide ranging topics such as survey-based risk adjustment methodology and cot of long-term care populations. He is also very familiar with the MCBS data, and produced reports on wide ranging topics such as survey-based risk adjustments models risk assessment of nursing home residents, and use of functional status to predict future health care costs. His Familiarity with the data base and the expert knowledge in survey-based risk adjustment for Medicare and Medicaid populations qualifies him uniquely for this task. Areas of research activities to be undertaken are listed as follows: Task 1 Literature Review of Survey Methodology in the following three topic areas: Differences between responses to mail and telephone surveys. The contractor will search for evidence of whether there are biases or differences in self-reporting with two approaches used. Special attention must paid to the question of whether any differential biases exist among different health conditions or diseases between the two methods of data collection. Differences in health status between respondents and non respondents in surveys, such as the MCBS, the NLTCS, and the LSOA. Contact individuals at survey research centers and those who are responsible for data collection at HMOs to supplement evidence in the published literature. The total number of interviewees may not exceed nine. Task 2 Generating Descriptive Statistics of MCBS Population over time: Combining several rounds of MCBS data, this project will generate person-level cost, utilization, and health status statistics. Tasks to be performed include generating person-level summary statistics for cost, use, and health status proxy measures, and documenting the rates of changes in these information. Summary statistics with levels of significance may be reported by some definition of geographic areas. In addition to examining changes in cost, use, and health status, the study will also examine changes in various proxies for health outcomes, e.g., mortality rates and rates of entry into long-term care institutions. Appropriate stratifications/groupings will be chosen to present the data. Task 3 Annual and Multi-year Medicare and None-Medicare Cost Models: Research models will be developed separately using information in the Cost and Use MCBS files that meet the following specifications. Developing research models that predict annual and multi-year health care costs. Several models may be designed to enhance and expand our understanding of the risk adjustment approach. Separate models will be constructed for Medicare costs, non Medicare (Medicaid) costs, community long-term care costs, institutional(nursing home) costs, ancillary medical services, and total health care costs. Various approaches may be taken to develop models that are based on appropriate choices of socio-demographic factors and health status variables. Potentially relevant socioeconomic variables for the project are; age, sex, marital status, income level, while suggested health-related variables are diagnoses, self-reported health status, institutional status, and functionality levels as proxied by ADLs and IADLs. The models will be evaluated for statistical validity and logical validity. Suggested objective criteria to be used for performance of the models include the total variance explained by the model (R-square), predicted cost to actual cost ratio, various statistical significance measures (F-statistics, t-statistics), and measure of parameter stability across the years as well as across the models. Task 4 Assessment of Geographic Factor Variations in Risk Adjustment: Supplementing the MCBS with geographic information from the Area Resource File, this project will assess significance of geographic variation in cost and use of health care services in the risk adjustment models developed in Task 3. Research will estimate the impact of geographic variations in costs utilization measures in risk adjustment models, and weigh the benefit of incorporating a geographic adjustment factor in prospective risk adjustment systems should there be a significant geographic bias in the models. If the data can support, the project may use interaction terms between the standard risk adjustors (independent variables) and geographic adjustment variable. Relevant factors of considerations in this task include stability of risk adjustment models across the geographic areas, significance of geographic factors and costs/outcomes, significance of area supply characteristics in different risk adjustment models. In the cost study, it seems necessary for the project to adjust for price differences. County-specific price index developed by the Health Economics Research is recommended as a standardization measure of the costs. The contractor may, however consider other suitable alternative indices. Evaluation of the models, similar to the ones described in Task 3 will be conducted to communicate the statistical properties of the models. All responsible sources may submit capability statements for consideration to the Agency at the above address. Numbered Note 22 Posted 08/13/98 (I-SN236557). (0225)

Loren Data Corp. http://www.ld.com (SYN# 0007 19980817\A-0007.SOL)