IOM says “Target Decision Making.” No. Target Poverty

The IOM’s new report on geographic variation in health care was released yesterday. It attempted to define the reasons for the variation that has been projected by the Dartmouth Atlas and has fed the frenzied war on waste.

The IOM’s approach was similar to Dartmouth’s but differed in important ways. Both measured Medicare spending that had been aggregated at the level of HRRs and both adjusted it for age and gender (which don’t vary) and black race, which does but is a poor proxy for the broad group of socioeconomic factors that matter. That left a lot unexplained by Dartmouth. But the IOM committee further adjusted for input prices, which markedly decreased variation, and it also looked at variation in commercial insurance costs.

A clue that aggregation might be a problem was the IOM’s observation that there was even more variation among the 3,000 hospital service areas (HSAs) than among the 300 HRRs. Had they gone the further step of examining the 30,000 Zip codes, they would have found still more variation. So the variation that they and Dartmouth studied was partially obfuscated by the “tyranny of aggregation.”

Among Medicare beneficiaries, almost half of the variation between the top and bottom deciles was explained by health status, 9% by race and 6% by income. Due to co-variances, the three together explained 43%. These factors also explained 10-20% of the variation for commercial insurance. That left about 65% of the variation in total health care spending unexplained. Of this, about one-fourth was attributable to variation in acute care (including procedures, diagnostic tests, prescriptions and ER visits), while three-fourths was attributable to post-acute care.

The fact that there was greater variation among HSAs than HRRs drew the IOM committee to recognize that there was a great deal of internal variation and that adjusting payments based on aggregate measures of geographic differences in HRRs would unfairly reward “low-value providers in high-value regions” and punish “high-value providers in low-value regions.” Therefore, it encouraged Congress not to adopt a geographically based value index for Medicare.

But despite the fact that, as mentioned, most of the variation was in the post-acute sector, the Committee’s major conclusion was that CMS should test payment reforms that incentivize health care systems to assume some or all of the risk of managing health care. Indeed, this conclusion was incorporated into the title of its report (“Target Decision Making, Not Geography”).

No study was necessary to reach that conclusion. Indeed, nothing in the report suggests that payment reform would be necessary. Very little of the variation had anything to do with acute-care, and little credence was given to any of the variation. It could not have been more clearly stated. As I said in the Washington Post on September 11, 2009, the Dartmouth Atlas is the wrong map for health care reform.

Had the Committee pursued the lead that variation increases as the unit of analysis shrinks from 300 HRRs to 3,000 HSAs, they might have done what my colleagues and I did, which was to move even lower to the level of the 30,000 ZIP codes, where virtually all of the variation in acute care is due to variation in income, education and the burden of disease.

So what the report really found, but did not say, was that poverty and its associated poor health status is the major cause of geographic variation in health care and its effect is far greater than apparent from studies of HRRs which, because of aggregation, obfuscate its true magnitude.

Rather than recommending practice incentives to take on risk, the report should have recommended social interventions that could reduce risk. By coincidence, that is what President Obama did at Knox College in Galesburg Illinois the very same day. He said, “this growing inequality — it’s not just morally wrong; it’s bad economics… and it undermines the very essence of America.” He could have added, and it raises health care costs. Moreover, because poverty is distributed unevenly, it is responsible for the geographic differences in health care that we had been told were due to clinical practice variation but now know are due to geographic differences in income-inequality.


  1. Donna Kinney

    Yes. And we need to advocate for all health care risk adjustment and “quality”(read: patient adherence) measures to include an adjustment based on income (or average income for home zip). That will help reduce the distortions that are to come based on the Medicare value-based fee modifier and other poorly designed programs.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s