Critics of health reform target the data

Yesterday's NY Times summarizes criticism of a major data source for health reform: the Dartmouth Atlas of Healthcare, which is run by my employer, The Dartmouth Institute. The principle Atlas finding is that health care costs vary widely irrespective of patient health, and higher cost doesn't buy better care.  So health reform should reduce variation to reduce spending. Atlas research spans 30 years and hundreds of peer-reviewed papers. Because Atlas findings are counterintuitive and displayed in colored maps, many assume its methodology is oversimplified. Some common misconceptions:

"Atlas says "high cost" doctors are greedy": Its actually practice patterns of hospitals and regions that determine how docs practice. All docs believe they provide the best care possible, but if moved from high cost to lower cost hospitals, individual practice changes to reflect the new culture. The closest thing to "greed" is the concept of supplier-induced demand: If you have it you'll use it, and everything in health care has a cost. Have a new CT machine? You'll order more scans, not to increase billing per se, but to gather more information. But the concept has limits: Increasing the number of orthopedic surgeons increases the overall amount of ortho services, but not necessarily any single ortho procedure.

"Higher cost hospitals have sicker patients": Imagine three patients who all die on Friday, one from heart failure, one from diabetes, one from cancer. Rewind to the previous Monday, and try determine which is the "sickest."  The Atlas uses death as an endpoint, assuming all patients who are the same distance away from death are all about equally as sick. The fact that those three patients would receive varying amounts of treatment depending on the hospital they're in and still all die on Friday, perfectly demonstrates a need for health reform: Amount of care varies, even when all care is essentially futile. (Basic demographics such as age, sex and race are also factored in.)

"High cost hospitals attract high cost patients": The Atlas assigns cost to your zip code of residence. If I suddenly develop a rare dread disease and fly from my home in Vermont to the Mayo Clinic in Minnesota to receive care, all costs associated with my treatment will be assigned to the hospital I normally use in Vermont (based on my home zip code), not the Mayo clinic. So Mayo's magnet effect won't necessarily inflate their costs (Mayo actually provides some of the nation's least expensive care despite its reputation for quality).

"Higher cost regions are expensive". Everything in New York City is more expensive than in Vermont, and docs are likely paid more. But that doesn't explain variation in cost among hospitals within New York City itself.

No data source is perfect. As with all science, the Atlas is an evolving process of inquiry, open to challenge from future research of equal rigor. So far few have been able to refute Atlas findings with hard data, so health reform's trust is well placed.