The Medical Expenditures Panel Survey (MEPS) has become a critical asset to health policymakers on both sides of the aisle. The survey, which has been conducted annually by the Agency for Healthcare Research and Quality since 1996, was preceded by two surveys conducted in 1977 and 1986. We served on the original project team for the 1977 effort known as the National Medical Expenditures Survey (NMCES) and have regularly followed the evolution of the survey. In view of the increased need for health expenditure data we now believe new survey enhancements are warranted.
History
The 1977 effort was designed to provide a nationally representative sample of health care coverage, use, and expenditures with greater accuracy and detail at the individual level than had previously been available. These data were needed in order to better understand existing health insurance coverage and how its presence or absence affected the use of various types of health care services.
They were also necessary to model changes that might result from changing and/or expanding insurance coverage and other potential policy changes. Specific policy interests change with each administration; what remains constant is the need to model potential policy changes and to understand those changes’ effects on use, spending, and health care status. Even in Republican administrations that did not emphasize changes in health policy, key Republican stakeholders such as Speaker Gingrich recognized that accurate data were needed regardless of whether one wanted to expand or reduce the Federal Government’s role in providing coverage.
Every subsequent administration has made use of MEPS data. The Reagan Administration used it in examining the effects of the Medicare Catastrophic Coverage Act while the Bush Administration used it to develop parameters in the health care reform proposal that it released in February of 1992. MEPS was used extensively in the formulation of Clinton’s Health Security Act and was subsequently used during the design of the prescription drug coverage expansion under Medicare part D. In the debate over the Affordable Care Act, MEPS data was cited by both ACA supporters and critics.
MEPS data are the number one single source of data for papers published in Health Affairs.
Why MEPS is Needed
The Federal government conducts other surveys that are important sources of health related data. The National Health Interview Survey (NHIS) conducted by the National Center for Health Statistics has supported many analyses and through its “early release” initiative NHIS has been able to provide timely estimates of insurance coverage and use of services. A number of private sector surveys have also been used to conduct analyses of the ACA.
These efforts have been used in numerous studies that have documented coverage expansions that took place after the ACA was implemented. These coverage expansions, while dramatic, were generally consistent with expectations. A comprehensive examination of the ACA and other policy initiatives also requires an examination of cost and how the increased or changed use of health care services affects reported health status. MEPS is the only national survey that collects data on all of these areas. Consequently, MEPS remains an important resource to those concerned with program cost and efficiency as well as the effects of changing health insurance coverage on health care use and subsequent health status.
The upcoming election results may create additional demand for health services research that will require accurate expenditure data. Republican gains could be followed by a push for several initiatives including a new call for expanding health savings accounts, for refundable tax credits to replace existing subsidies and employer-sponsored health premium exclusions, or for Medicaid reform emphasizing per-person block grants to states.
If Democrats do well there might be movement toward new public options, lowering the age eligibility for joining Medicare, or expanding the amount of subsidies available to individuals buying insurance coverage in the exchanges. Understanding the effects of these policy changes requires an understanding of how changes in coverage affect health care expenditure patterns and what effects these have on subsequent health care status.
Challenges facing the MEPS
MEPS faces many challenges today.
First, the sample size for MEPS is too small to support many important analyses. MEPS currently collects data from about 14,000 families; a sample size similar to that of the 1977 NMCES. The 1977 survey, however, was designed to evaluate issues such as universal national health insurance and the federalization of Medicaid. NMCES could produce national estimates of the most important demographic groups but there was only moderate demand for state-level data and NMCES was not designed for this purpose. The need for better state estimates was becoming clearer even before the advent of the ACA but now, the ability to compare state experience is of far greater importance. MEPS can now support estimates for 29 states but this refers only to the capability of providing estimates about the total state population, it does not allow researchers to describe specific demographic groups within most states such as the poor or those without coverage.
Second, MEPS uses a panel design so that data is collected on each sampled household over a two-year period. Each year half of the sample rotates off and is replaced with newly sampled households. The panel design has been useful but the two-year period is likely to be insufficient to track people as their circumstances change.
Third, collection of expenditure data is far more difficult than collecting data on health insurance or use of services. Many respondents are unable to accurately report the payment that went to their provider. MEPS remedies this by obtaining payment data directly from providers that are identified by those in the household survey. There are however insufficient resources to obtain data from all identified providers. Currently attempts are made to obtain validation data from all identified hospitals and pharmacies. Although relatively high response rates are obtained in the provider validation, a substantial amount of data is still missing and must be imputed. Imputed data is not as accurate as validation data.
Additional resources would be valuable in decreasing the amount of hospital data that currently cannot be verified. The problem regarding the cost of ambulatory care is even more serious; because of resource limitations only half of the identified physicians are even contacted to provide payment data.
Finally, we need to know more about provider incentives. This information could be obtained by asking providers for a small amount of supplementary data in addition to the payment data requested. The 2016 MEPS does include a Medical Organization Survey to fill this need but this component was externally funded as a onetime effort and is not scheduled to be included in future surveys.
Future Considerations
The MEPS budget has remained flat despite its growing use among stakeholders including the Congressional Budget Office, the Office of The Assistant Secretary for Planning and Evaluation, The Medicaid and CHIP Payment and Access Commission, Office of Management and Budget, the White House, and many others in both the public and private sectors. We would like to see MEPS increase its capacity to provide state estimates, conduct more longitudinal analysis, and link the use and cost of patient care to the incentives of their providers. The following enhancements should be considered:
- The MEPS sample size should be increased. One method for doing this would be to continue adding new cases at the current rate each year but retain respondents in the sample for three years instead of two. If this were implemented, in 2018 the MEPS sample would grow from 14,000 households to 21,000. In addition to increased precision this would also allow for longitudinal analyses to be conducted using a three-year window. This would be a good first step to better understand the effects of changing health care coverage on health status. It would also help, but not solve, the issues related to state estimates.
- All identified physicians should be asked to provide data validating the expenditures reported by their patients in the MEPS sample.
- The Medical Organization component of MEPS which collects data on physician practices and incentives should become a permanent component of MEPS.
Implementing these recommendations would increase the annual cost of MEPS by approximately one-third; a justifiable investment. Earlier versions of the MEPS were far more resource intensive than the current effort. They included more rounds of interviews, a separate component for examining those receiving care in institutions, and another component that collected detailed data from insurance plans identified by respondents in the household survey. The enhancements suggested here would still result in a leaner effort than those fielded in 1986 or 1996.
The health care debate will continue to be contentious but the lack of quality data does not benefit either side. MEPS has helped contribute to informed discussions about many health care issues and it is a resource that is worthy of support.
from Health Affairs BlogHealth Affairs Blog http://ift.tt/2ee51jF
No comments:
Post a Comment