An Estimation Strategy to Represent the Institutional User Population in the 1987 National Medical Expenditure Survey 1

Abstract

The Institutional Population Component (IPC) of the National Medical Expenditure Survey (NMES) was established to provide an assessment of the health care utilization, costs, sources of payment and health insurance coverage of the U.S. institutionalized population residing in nursing and personal care homes (NH), and in facilities for the mentally retarded (MR). The primary objective of the survey was to estimate the use of and expenses for health care services for all persons residing in institutions at any time during calendar year 1987. To obtain a nationally representative sample of the 1987 institutional user population, the survey included a sample of residents residing in selected facilities as of January 1, 1987, in addition to a representative sample of admissions to the selected facilities over the course of 1987. The union of these samples served to represent the 1987 institutional user population.

In this paper, the representation of individuals with multiple probabilities of selection in the IPC sample is examined. In addition, an estimation strategy is developed to correct for the multiple chances of sample selection in the IPC over the course of 1987 and the problems of dual representation in independent sampling frames. The paper also includes a discussion of the effect of this estimation strategy on the precision of sample estimates characterizing the institutional user population.