• Health care and premature deaths

    From Different Perspectives for Assigning Weights to Determinants of Health, by Bridget C. Booske, Jessica K. Athens, David A. Kindig, Hyojun Park, and Patrick L. Remington:

    An oft cited McGinnis et al (2002) paper states: “…using the best available estimates, the impacts of various domains on early deaths in the U.S. distribute roughly as follows: genetic predispositions, about 30%; social circumstances, 15%; environmental exposures, 5%; behavioral patterns, 40%; and shortfalls in medical care, 10%”. […]

    However, some caveats should be noted:

    1) The “long standing estimate” of 10% for medical care is actually based on “expert” estimates of the contribution of health care system deficiencies to total mortality; (DHHS, 1980);

    I’d like to read that old DHHS document, but I cannot find it. However, it is true that McGinnis et al (2002) cite it in support of the 10% figure. (All references are in the paper linked to at the top of this post. It’s ungated.)

    I’m not claiming 10% is the wrong figure. But I sure don’t see how we can be certain it is right. If it is, it is not clear that citing McGinnis et al or the DHHS document is the way to support that figure.


    • And if it is true, why are we paying so much for health care (2-3 x other advanced countries)?

    • What do they mean by “shortfalls” and “deficiencies”? Malpractice, or lack of availability? That is the real question.

    • I think at best that the estimate of 10% due to “medical care” is a gross oversimplification. This varies quite a bit depending on the intervention and preventive measures score very high. For instance, prenatal care has been shown to lead to great improvements to maternal and child health in both developed and developing countries. The US does a poor job of providing this relatively inexpensive care and as a result ends up at the bottom of maternal and neonatal indicators of all of the developed countries while countries like Cuba or France which provide comprehensive prenatal care do very well on these scores.
      On the other hand, interventions where the US excels (high cost, high tech) such as coronary artery bypass have little effect on morbidity and mortality (some studies have even shown a deleterious effect on M&M).

    • And let’s not forget that the purpose and benefit of medicine is not just to lengthen life. If you break your arm and don’t get treated you will probably survive — but you also probably won’t be either very happy or very productive. Shortening illness, preventing disability and alleviating discomfort are also important. and would be even if even if they didn’t lengthen life at all.

    • So — I actually spent some time with Dr. Kindig a year ago trying to think of ways to empirically estimate these weights. I would just say that I am open to any suggestions. What is a unit of “social circumstances”? As @Mark notes above, doesn’t it make a difference by actual medical intervention and target population?* Is premature mortality(yrs of preventable life lost) the right outcome ? What kind of dataset (longitudinal for sure) would be needed and does it exist anywhere in the US? Can such a model ever be properly specified, given all the multicollinearity, “causal” effects that appear to go in both directions and thus are not directed acylical graphs, and distal interaction effects?

      In the end, given all that and a lack of grant money, is 10% not a useful heuristic? No? How about 25%?

      */e.g the entire population reduction in heart disease and lung cancer mortality and morbidity since, say, 1970, could arguably be attributed to reductions in smoking rates, and not at all to tmt. Does that mean tmt has no value for an individual having a heart attack? I think not. Or take this headscratcher: how to attribute reductions in murder rates — jobs, social programs, or improved trauma care?

    • A lot of work has been done using the concept of “Burden of Disease” usually expressed as “Disability Adjusted Life Years” (DALY) to estimate the impact of disease on morbidity and mortality including adjustments for duration and degree of disability. Each disease has a separate calculation and one can look at the cost to treat and the cost to prevent a case.
      I have used these methods for national health planning in developing countries where you look at allocation of resources based on the least cost to prevent and treat disease. This usually comes down to a cost to “buy” a year of healthy life through various interventions. For instance, immunization costs about $5 to buy a year of healthy life whereas something like coronary artery bypass costs many tens of thousands of dollars for a year of healthy life. This is useful in developing countries at the national health planning level but could also apply to developed countries looking to improve the value of their health care purchases.

      The World Bank did a lot of work with this in the early 90s. Christopher Murray and Alan Lopez worked there and at WHO to develop models of disease. The WHO has a unit on burden of disease which develops disease models.