Presentation: Combining Information from Multiple Data Sources for Health Care Cost Modeling
Speaker: Trivellore Raghunathan (Raghu), Ph.D., Director, Survey Research Center, Institute for Social Research, Professor of Biostatistics, School of Public Health, University of Michigan, Ann Arbor, Michigan, USA
Abstract: Assessing the effectiveness of health care delivery is important for making health policy decisions. Effectiveness of medical care can be expressed as improvement in national health in return for the dollars spent for medical care. Obviously, the three questions arise: (1) How to measure national health? (2) How to measure the cost of medical care attributed to various health conditions? and (3) How does then one relate answers to questions (1) and (2) to arrive at estimates of quantities useful for health policy decisions. To answer these questions we need to link medical expenditure to diagnostic, treatment and preventative strategies, as well as trends in cost allocation and disease prevalence. Unfortunately, there is no single data source to address these important questions. Missing data and measurement error model framework is used to combine information from multiple surveys and administrative databases to estimate relevant parameters and derive several “what-if” scenarios. The methodology is developed and applied to the Medicare Current Beneficiary Survey (MCBS) linked to the Medicare claims and the National Health and Nutrition Examination Survey (NHANES) for years 1999-2013, focusing on the elderly, 65 years of age or older. Extensions to other age group will also be discussed.