Tuesday, May 26, 2009

Cogent commentary about the Dartmouth Atlas conclusions by a senior medical educator and troublesome observations by Atul Gawande

The Obama administration is promoting the idea that we can finance a wide and expensive variety of government programs-including better quality health care for everyone- by getting the waste out of medical care. The factual (alleged) basis for this claim to some degree comes from a widely quoted ecological study, the Dartmouth Atlas. The publications from this project apparently has had significant influence on policy recommendations.

This study divided the nation into five Medicare quintiles and then compared money spent with several outcomes.When the region's selected outcomes did not improve as expenses increased they concluded that the difference in expenditures was "wasted" money. Other conclusions are possible and from what I read may make more sense although no one speculation based on aggregate data should stand unchallenged. One such conclusion is that differences in poverty level and in total medical expenditure offer a better explanation ( see below) that the notion that for some basically unexplained reason physicians in some regions just are prone to waste money by doing too much in terms of health care services.

I suppose the suggestion is that greedy and/or misinformed physicians tend to congregate in certain parts of the country.There is at least some anecdotal evidence that that certain cities may in fact have significantly higher medical expenditures due to variations in medical practices at least some of which may be profit driven.

A recent article in the New Yorker by the prolific Dr.Atul Gawande seems to argue the case that McAllen Texas is such a city. The article is worth reading and I found it very troublesome as Gawande relates stories (mostly heresay) of physicians requesting kickbacks for hospital admission and thinly disguised kickbacks from nursing homes. How widespread are the practices narrated by Gawande and to what extent overall health care expenditures are impacted are questions that remain unanswered as does the question why would that be regional.

Go here to read analysis from a former medical school dean, and longtime physician,Dr. Richard (Buz) Cooper, who obviously has spent much time and effort in studying this problem. He argues that Medicare spending is not an accurate proxy for total medical care expenditure and that in states with more total medical spending medical care is better, a conclusion 180 degrees from the Dartmouth conjectures.

I continue to be impressed by how often aggregate data can be used to reach conclusions and policy advice that is conjectural and at times agenda driven. The Dartmouth project may be one .Another is the recent NEJM article on hospital readmission rates of Medicare patients, a topic for a future blog.In both instances the numbers are real but the underlying factors are difficult to tease out and may be refractory to statistical techniques designed to eliminate confounding.Data gathering and statistical analysis have become cheap and widely available ,what to do with the data requires reasoning and that is as scarce as ever.

Dr. RW (see here) has recently commented on the illogical and overreaching conclusions of the Dartmouth folks.

The Dartmouth Atlas project is funded by three large insurance companies or their foundations ( Aetna,United Health care and Well Point) with the other two contributors being The Robert Wood Johnson Foundation and The California Health Group Foundation.

To claim that results may be influenced by their source of funding risks accusations of indulging in a version of an ad hominem argument. Still you have to wonder if large insurance companies would continue funding for a long standing project that published results that did not further their business interests or if they would spend much money in supporting research such as that discussed by Dr.Cooper.If you run a medical insurance company you have to like a study that claims to demonstrate widespread wasted medical spending. It is of interest that here waste is defined by a third party observer not by either the providers nor the recipients of the services. Having said that, the results are either valid or not (or we just can't tell) independent of who financed the project.

I remember learning that ecological studies might at best serve as hypothesis generating activities not as generating answers so definite and certain that they could form the basis for sweeping health care changes.

I continue to be very suspicious about conclusions from studies based on aggregate data.Serious and thoughtful and non-biased commentators are able to offer opposite stories to "explain" the data. However, Dr. Gawande's observations about practices and hospitals in several cities with significantly different practice patterns and levels of medical cost (and profits) and perhaps quality of care make me wonder if the Dartmouth conclusions might be not quite as unjustified than I want to believe as that conclusion painfully bumps up against my philosophical priors and epistemological biases. At least I am less inclined to dismiss them out of hand even though there must be more to the story than marked regional variations in greed.

No comments: