Be careful when estimating years of life lost: quick-and-dirty estimates of attributable risk are, well, quick and dirty.

Peter Morfeld writes:

Global burden of disease (GBD) studies and environmental burden of disease (EBD) studies are supported by hundreds of scientifically well-respected co-authors, are published in high level journals, are cited world wide and have a large impact on health institutions‘ reports and related political discussions.

The main metrics used to calculate the impact of exposures on the health of populations are „numbers of premature deaths“, DALYs („disability adjusted life years“) and YLLs („Years of Life Lost“). This large and influential branch of science overlooks seminal papers published by Robins and Greenland in the 1980s. These papers have shown that „etiologic deaths“ (premature deaths due to exposure) cannot be identified from epidemiological data alone which entails that YLLs and DALYs cannot be broken down by age or endpoints (diseases). DALYs due to exposure are problematic when interpreted in a counterfactual setting. Thus, most of this influential GBD and EBD mainstream work is scientifically unjustified.

We published a paper on this issue (open access):

Hammitt JK, Morfeld P, Tuomisto JT, Erren TC. Premature Deaths, Statistical Lives, and Years of Life Lost: Identification, Quantification, and Valuation of Mortality Risks. Risk Anal. 2019 Dec 10. doi: 10.1111/risa.13427.

Just for some additional background when you like to comment on the issue: Here is a letter exchange in Lancet with the leader of the largest GBD (global burden of disease) project world wide (Christopher Murray, Seattle).

This exchange is not covered in our paper. It may give an indication how the arguments and bias calculations are received.

My only comment is that I still think Qalys (or Dalys or whatever) are a good unit of measurement. The problems above are not with qualys, but with intuitively appealing but problematic statistical estimates of them. What joker put seven dog lice in my Iraqi fez box?

P.S. That above-linked discussion also involves Ty Beal, whose name rang a bell . . . here it is!

12 thoughts on “Be careful when estimating years of life lost: quick-and-dirty estimates of attributable risk are, well, quick and dirty.

  1. I went back and re-read the 2008 discussion that andrew linked: https://statmodeling.stat.columbia.edu/2008/03/14/valuing_lives_s/

    What I was struck by is that a large number of Economists boarded on and started talking about willingness to pay, and it took until the second to last comment, from Phil, before anyone mentioned that as far as societal / tax money usage goes the people the economists were asking about their willingess weren’t the ones paying!

    Sure, 90 year old men with 1 year of life expectancy may well be willing to spend $100k *of their own money* to save themselves from an immediate threat that removes that chance for the extra year of life… But I think if you put up a website and asked people to vote where $100,000 of tax money should be spent on say 1000 different hospital patients… People would be much more willing to spend $100k on a low risk surgery to repair a heart defect in a 1 year old than they would to do an ablation surgery and implant a pacemaker in a 92 year old. I can tell you right now that when my 92 year old grandfather was convinced by his cardiologist to be given $100k worth of heart ablation surgery on the taxpayers dime our entire family was upset about it. He died ~3 days later, after the cardiologist left for a fancy vacation.

    • Daniel –

      > I can tell you right now that when my 92 year old grandfather was convinced by his cardiologist to be given $100k worth of heart ablation surgery on the taxpayers dime our entire family was upset about it. He died ~3 days later, after the cardiologist left for a fancy vacation.

      I certainly get your point. But I think it’s complicated by considering such expenses in isolation from, say, the amount of taxpayers’ dimes we spend on weapons systems.

      • Joshua,
        Certainly we could do better with our overall priorities for spending (e.g., dealing with global warming), but as an older person (79) I think Daniel is right about out health care spending.

        • John –

          No one needs to convince me about lowering spending on end of life care. I’ve cared for two parents and a brother who were in hospice. My partner, sister-in-law, and step daughter have all worked as hospice nurses.

          But… I don’t think it’s a zero sum scenario where the money we wpuldntbspend on elaborate procedures for a 90 year-old in poor health would clearly and directly go to providing healthcare for 1 year-olds (and in particular poor 1 year-olds).

          In contrast to Daniel’s family, family members are often the driving force behind high spending at end of life – for example to keep family members on tube feeding despite likelihood of greater suffering with low likelihood of significant prolonged life. Dealing with what taxpayers do and don’t want is complicated.

          Cause for our hugely expensive healthcare system is multi-factorial. Of course, high spending at end of life is a huge factor – but there are also other factors as well.

          If my priorities were driving the decision-making as to where taxpayer money goes, we would spend less on things like an ablation procedures as Daniel describes – but I would suggest an equal if not greater savings on military spending and a raft of other issues as well.

      • And hey, friends, it ain’t just government spending. Lots and lots and lots of money is coming out of family pockets to provide unnecessary care for elderly folks to drive the bottom line for health care providers – even if it’s just lost work time taking parents to appointments.

        And it ain’t just about profit. Public systems are more than happy to take your money to give it to whatever cause/lobby city/county/state leadership would like to score points with.

      • While that’s true in the abstract, I’m not sure how relevant it is, because those decisions don’t really go through the same political mechanisms in practice (at least in the US).

        I don’t think there’s ever really an actual tradeoff decision being made in those terms between military spending and social spending. Probably there *should* be, but that’s a higher-level fix – having to do with how Congress does things etc., issues which are not necessarily fixed by electing a new representative, and which are far less visible to voters than any individual bill.

        But it’s a problem that goes both ways. I wouldn’t want genuinely effective (IE – significant improvement in *both* quality of life and length of life) treatment for the elderly denied because their expected remaining life is small.

        But I don’t think there is any really good solution for this at our current technology level. In a public system, it will be more equitable overall than in the US system – but more equal outcomes doesn’t necessarily correlate to better average outcomes, either.

    • >>Sure, 90 year old men with 1 year of life expectancy may well be willing to spend $100k *of their own money* to save themselves from an immediate threat that removes that chance for the extra year of life

      Often, though, the very elderly (due to dementia/Alzheimers/etc.) aren’t the ones making their own treatment decisions.

      I wouldn’t be surprised if family often is more “aggressive” about treatment than the individual would be themselves, if able to make the decision.

      • “I wouldn’t be surprised if family often is more “aggressive” about treatment than the individual would be themselves…”

        Possibly. But we can be sure people in the health care industry will do everything possible to tip the scales to additional treatment. If the treatment is available on the government dime, they see it as their responsibility to get that cash.

  2. Re the fez box quote:
    “The big plump jowls of zany Dick Nixon quiver” has an almost poetic quality to it — kind of a sprung rhythm.

  3. There are major problems with QALYs and they’re not a good measure. I strongly recommend the book “Valuing Health” by Daniel Hausman for a thorough assessment and general discussion of measuring health (and more generally to anyone interested in QALYs or other methods of measuring health). There are also more technical criticism related to measurement techniques for QALYs and their assumptions, and the fact that different techniques can produce very different results. Finally, some critique from the point of view of behavioral economics can be found here https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1468-0297.2007.02110.x

    The disability weights for DALYs are even worse. They’re essentially based on opinions of a panel of experts.

Leave a Reply to confused Cancel reply

Your email address will not be published. Required fields are marked *