How Reliable are the Dartmouth Atlas Estimates of Hospital Efficiency?

This is Peter B. Bach, writing in The New England Journal of Medicine:

The predicted risk of death at the time of admission varied widely among hospitals. At the average hospital, the average risk was 15%. But the severity of illness was far lower in hospitals at the 10th percentile (6% risk of death) and far higher in those at the 90th percentile (22% risk of death). Differences in illness severity result in differences in resource consumption…

The Atlas, however, assumes that all decedents in all hospitals were equally sick before death, an error that tends to make low-severity hospitals look more efficient than high-severity hospitals even if the hospitals are equally efficient.

Comments (6)

Trackback URL | Comments RSS Feed

  1. Devon Herrick says:

    That’s interesting. A hospital that only has stable patients convalescing from minor ailments would naturally use fewer resources than similar hospital that accepts patients requiring care in the ICU or CCU. Yet the former would look efficient while the latter would look inefficient. Come to think of it, that’s precisely how the NHS consumes a smaller portion of Britain’s GDP compared the United States. Not providing the costly treatments that keep people alive at the margin.

    I also recently read that being admitted to a hospital at full capacity increases your risk of death.

  2. Tom H. says:

    Atlas results are reported eveywhere, and almost never do reporters point out that there may be problems with the data they are reporting on.

  3. Joe S. says:

    I’v seen quite lot lately, mabe at this blog, calling Darmouth Atlas results into question. If the entire Obama cost control effort is based on Darmouth results, it would seem that there is a very big problem here.

  4. Jon Skinner says:

    Disclaimer — I’m with the Dartmouth group.

    The only problem with the statement that the Dartmouth end-of-life measures assume that “all decedents.. were equally sick” is that it’s wrong.

    As we stated in our response to his article (which Mr. Goodman appears not to have read), we restrict our sample of decedents to people with life- threatening chronic illnesses. Furthermore, we know what kind of disease they have (for example, cancer, chronic obstructive pulmonary disease, dementia, etc) and so we control for the fact that some hospitals may have a larger fraction of the more serious diseases. (We also adjust for when people report more than one chronic illness.)

    In Peter’s defense, I think he right to be concerned about the risks of health care reform relying on inaccurate measures. But the end-of-life measure does a good job of characterizing expensive (price-adjusted) hospitals, and matches well with other measures, for example heart attack patients. (See our Economix posting for more details.)

    Of course, it’s a whole other question whether we can use these measures to reward or punish hospitals now before we have in place a well-functioning accountable care organizations.

  5. Jon Skinner says:

    ps — and to complete the previous thought — it’s my opinion we’re not yet ready to reward or punish hospitals given the sorry fragmentation of the current health care system and the potential for doing an even better job of measuring costs and quality.

  6. Jay Brazier says:

    Mr. Skinner:

    Care to respond to the central points raised in this article, or direct the readers here to a response that you may have posted elsewhere:

    “Since there was no attempt to determine which hospitals had a better survival rate for the Medicare patients, or to measure any other indicator of the quality of life of the patients, the Dartmouth data can legitimately yield no conclusions about the relationship between cost and quality of care.

    Moreover, and no less crucial, the Dartmouth study ignores real-life socio-economic and demographic factors among the different patient populations that obviously play a huge role in determining the cost of medical care. As many critics of the Dartmouth studies have correctly pointed out, poverty matters! Hospitals in large cities that treat many people from impoverished neighborhoods, as well as those in poor rural areas, are obviously going to face higher costs than hospitals in mainly upper-class areas, whose patient pools are wealthier and healthier.”