How to Read the Dartmouth Atlas

Getty Images

The Dartmouth Atlas Project is a wonderful idea – pursuing deeply interesting research at the boundary between medicine and economics. Unfortunately, there is a tension in its roles in doing both dispassionate research and public advocacy. While their academic work emphasizes nuance and balance; their policy recommendations by necessity push a narrower agenda.

Recently, a couple of NYT journalists – Reed Abelson and Gardiner Harris – have explored these tensions and the overall value of the Dartmouth exercise. This reporting is invaluable because the Dartmouth shop has become one of the dominant sources of information during the health care debate.

Responding to a Dartmouth critique of their article, the NYT team hits back with a wonderful piece that is, in many ways, more detailed and carefully argued than their original article. The headline-grabbing result from the Dartmouth Team – that there is major cost variation across the country seemingly unrelated to health outcomes – remains deeply flawed. Dartmouth’s own research shows an ambiguous relation between health spending and outcomes. As Abelson and Harris point out:

There appears to be a distinction between the thesis that became part of the public debate over the health care legislation and another, more nuanced view of the health care landscape that the Dartmouth researchers take when discussing their own and others’ academic research in detail. That more nuanced view does not support, and at times even contradicts, the popular perception that higher spending leads to worse outcomes.

The careful hedging of academic writing morphs into unambiguous statements – health spending is entirely unrelated to outcomes. Yet the story derived from the simple, unadjusted graphs – that medical costs vary widely around the country for no good reason – has become conventional wisdom. For instance, here is Ezra Klein describing former OMB head Orszag:

Orszag – taking his cue from reams of research showing that states that spent a lot of money on per capita Medicare spending didn't have better outcomes than states that spent much less -- argued that a substantial portion of each dollar we spend on health care is wasted.

That “reams of research” refers primarily to the Dartmouth results, as are the consequent policies that will cut Medicare spending on the assumption that health status will not suffer.

The Dartmouth team does provide alternate measures that do adjust for price in the academic work that’s not available to the public. Yet even those measures are relatively coarse, and do not necessarily fully account for the health status of patients. Abelson and Harris point to a study by the Medicare Payment Advisory Commission that uses different illness severity adjustments to arrive at different conclusions. In particular, the Commission found that Medicare spending did not necessarily translate into Medicare services because of differences in regional cost of living, cost of providing services, and different underlying health status.

These arguments don’t mean that the Dartmouth is fraudulent – only that policy-geared research is often by necessity stripped of certain nuances in ways that can connect with other pre-existing political considerations. In many ways, the push for health care legislation really depended on some result like this: if cuts could easily materialize in Medicare without requiring painful taxes or caps to service, they could be re-routed to expand healthcare for others. This is all the more reason we need journalists capable of digging further into the primary sources that fuel political debates; and more critical thinking about how to reform and reshape healthcare.