According to a leader of Dartmouth’s Health Policy group, “if we sent 30% of the doctors in this country to Africa, we might raise the level of health on both continents.” Sadly, the notion that 30% of health care resources are wasted underlies much of the current thinking about health care and serves as a beacon for President Obama’s health care reform team. Where did it come from?
MEDICARE IS NOT A PROXY. The “30% solution” emerged from a Dartmouth study that divided the nation into five “Medicare-spending quintiles” and compared outcomes, such as access, satisfaction and mortality, among them. But these outcomes are not patient-specific, nor are they Medicare-specific. They reflect community-wide characteristics that depend on the total resources available from all payment sources, not just Medicare. Nonetheless, if Medicare is representative of the whole, as claimed, it would be a valid proxy. But it isn’t. As is apparent from the graphics below, Medicare spending bears no relationship to total spending per capita. But, as I’ve demonstrated, it is total spending per capita, not Medicare spending, that correlates with better outcomes, such as access, quality and the use of mammography. Thus, the fundamental basis used by the quintiles study to determine a community’s level of care – Medicare spending – is flawed.
QUINTILES ARE NOT RANDOM. Assuming that Medicare is representative, a valid assessment of outcomes would require that all other characteristics that might affect outcomes would have to be the same. But even a casual inspection reveals how different they were. The highest-spending quintile was composed principally of dense urban centers, such as Chicago, Detroit, New York, Miami, Houston and Los Angeles, plus some smaller regions with extremes of affluence and poverty The lowest-spending one stretched across the northern tier, from Alaska through Washington and Oregon to Maine, and into the plains, half the land mass of America. Why possibly could be learned by comparing such dissimilar regions?
AGGREGATION AND AVERAGING MASKED DIFFERENCES. Despite these dissimilarities, the quintiles were all the same. How is that possible? The answer lies in the process of aggregation and averaging. Because each quintile contained hospital regions with a diversity of total health-care spending (despite similar Medicare spending) and a diversity of subpopulations, their averages were similar. The extremes of affluence and poverty and of high and low spending in the highest-Medicare spending quintile resembled the lowest, and like Lake Woebegone, everyone was above average.
STRINGING IT ALL TOGETHER. The Dartmouth group turned the fact that nothing was “necessarily better” into “it’s worse because it isn’t better.” Therefore greater levels of Medicare spending (despite the similar levels of total spending) must have been wasted. And because the differences in Medicare spending were “unexplained” (although they are readily explained), they must have been due to the overuse of “supply-sensitive services.” And these must have resulted from an over-supply of specialists. And, therefore, if the numbers of specialists were no greater in the high-spending than low-spending areas, the nation could save 30%.
And by stringing together these politically-charged phrases, the Dartmouth group succeeded in recruiting large numbers of believers — even an economist as brilliant as Peter Orszag, Director of OMB, asked, why can’t Newark be more like Minnesota?
But what the Dartmouth group really strung together were three profound errors: 1) Medicare is not a proxy for the whole; 2) the populations in each quintile were not random, and 3) aggregation and averaging masked real differences among population groups. Like Churchill’s Russia, the quintiles study was “a riddle wrapped in a mystery inside an enigma.” But the riddle has been solved. When critically examined, geographic variation in health care is seen to reflect the interplay between communal wealth and individual income. And the poignant reality that emerges is the powerful and tragic effects of poverty on health care utilization and outcomes (see “Let’s Talk About Poverty” April 23).