Moreover, if one allows for the baseline list of covariates to include insurance status but does not model the impact of next year’s expenditures as if that family maintained its baseline insurance status, one can avoid the concern that some individuals could have a moderate or very large health care expenditure that would lead them to be eligible for some programs (such as Medicaid) and thus have lower out-of-pocket medical care spending than if the same events had to be evaluated with baseline coverage.

Needed Research for a Prospective MCER Measure

The truly prospective measures that require 2 or more years of data, or with surveys that have complete baselines, imply moving away from the more timely CPS ASEC to other data sources. Here, there are a number of issues of limitations with currently available data. Just as important, however, is the dearth of relevant literature on which to base the models. Although substantial data are available on how mean expenditures of individuals respond to such individual characteristics as age, gender, and health and to such family characteristics as income and insurance status, there are very few data on the responses in terms of distributions or variances at either the individual or the family level. In the case of single-person families, there is no issue. But how does one combine the information across family members, when ages and health status vary so much across different family members? What types of statistical or econometric models perform well?

Of equal importance is the dearth of information on what factors predict out-of-pocket medical care spending. Clearly, insurance status will play as large a role as prospective illness, but not much work has been done on this. The panel found that although much is known about total health care expenditures, very little is known about family and individual covariates that predict family out-of-pocket medical care spending or family finances. This situation dictates an analytic agenda before highly specific recommendations can be made on a prospective measure of MCER.

Health services researchers and health policy analysts have substantial experience with mean expenditures adjusting for observable individual and family characteristics at the individual level. There is much less work that has been done on out-of-pocket medical care spending at the individual level.9 The panel’s examination of out-of-pocket medical care spending (excluding the out-of-pocket premium) from the MEPS data on adults in 2004 suggests that out-of-pocket spending for medical services has the usual health expenditure characteristics (Banthin and Bernard, in Part III


9 A notable exception is the work by Dana Goldman and his colleagues on the Future Elderly Model. That study does not deal with general noninstitutionalized population, however (Goldman et al., 2004).

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement