First, at the very highest level, he noted that estimates of excessive expenditures made from four analytically distinct approaches came to roughly similar approximations of the total amount of excess costs for health care in the United States. Specifically, looking at regional variations in Medicare costs, the Dartmouth group estimated overall excess expenditures to be about 30 percent of national health expenditures (Wennberg et al., 2002), or about $750 billion in 2009; the analysis by McKinsey Global Institute suggested that the excess U.S. expenditure relative to Organisation for Economic Co-operation and Development (OECD) countries would be approximately $760 billion (adjusted to 2009 total expenditure levels) (Farrell et al., 2008); the lower-bound totals of estimates of excess expenditures identified in the workshop materials amounted to about $785 billion in 2009; and the estimated possible savings (lower bound, corrected for obvious overlaps) from full implementation of effective strategies in 2009 would be in the range of $550 billion. He also emphasized that such estimates are virtually all unvalidated extrapolations, based on assumptions from limited observations.
Moving to estimates for the next level of granularity—the component domains of excess costs—and again underscoring the various issues, differences, and analytic fragilities, McGinnis used the “lower bound of estimates” approach to summarize in broad terms the aggregate excess expenditures discussed at the workshop, both by the six categories that make up the broad domains of excess and by the component elements discussed for each of the domains. Approximations using this approach would amount in 2009 to about $210 billion in excess health costs from unnecessary services, $130 billion from inefficiently delivered services, $210 billion from excess administrative costs, $105 billion from prices that are too high, $55 billion from missed prevention opportunities, and $75 billion from fraud. These lower-bound domain estimates, and those for the contributing components, are noted in the commissioned background paper that placed the workshop analytics in the context of additional national estimates found in the literature (Box 21-1 below, and see “Summing the Lower Bound Estimates” in Appendix A).
McGinnis also drew on the background paper to highlight and emphasize the methodologic constraints in the analyses and estimates:
Varying sources of presentation estimates. The estimates presented throughout the workshop series were calculated by varying methods, including original peer-reviewed research by the presenter and the presenter’s synthesis of the published literature. In the case of the latter, few additional national estimates were found that were not referenced by the presenter.