Beware Cancer Med

Chapter 2 of Ken Lee’s thesis compares med spending and age-adjusted deaths across the 50 US states from 1980 to 2007. Lee’s baseline model finds that deaths increases with smoking use, alcohol use, population density, and med spending: a 10% increase in med spending increases deaths by 0.85%. Breaking down this med spending death effect by drug vs. non-drug spending, and by four causes of death (cancer, heart attack, injury, and other), Lee finds (in Tables 5,6) that med spending hurts mainly because increasing non-drug med spending by 10% increases cancer deaths by 2.1%:

Cause of Death, Drug vs Non-Drug Med Spending

The apparent lesson: avoid cancer docs, and especially their non-drug cancer treatments. It seems some places tend to spend more on med overall, and when they spend more on cancer patients, those patients die no less, and maybe more. That fits with cancer patients living longer when they go to hospice and get no cancer treatment and with randomized trials of cancer screening consistently showing no effect on total mortality. Other explanations, however, are that high med spending places tend to classify more deaths as due to cancer, or that med treatment of all sorts tends to cause cancer.

For you stat whizzes, Lee uses state and year fixed effects, and uses per capita physicians, beds, and dental spending as med spending instruments to disentangle the direction of causation.  He picked that instrument set because it had the smallest bootstrap variance, and passed many tests. Here is Lee’s baseline model (from Table 3):

KenLeebase

GD Star Rating
loading...
Tagged as:
Trackback URL:
  • Pingback: Overcoming Bias : Skip Cancer Screens

  • Wonks Anonymous

    Are you familiar with Paul Ewald‘s theories? He’s more optimistic on cancer.

  • max

    That fits with cancer patients living longer when they go to hospice and get no cancer treatment and with randomized trials of cancer screening consistently showing no effect on total mortality.

    Drugs since Jan 2010 shown to prolong survival in cancer patients-Abiraterone, Eribulin, Ipilumumab, Carbazitaxel-all in randomized trials, all now FDA approved.

    Also, I would be careful applying a large, retrospective epidemiological study and one done in lung cancer to the entire cancer population. Hospice is a valuable addition to medical care and certainly could be given to more patients, but it is not appropriate for everyone. Patients who have few symptoms from their metastatic breast, colon, or lung cancer and have yet to receive treatment would likely be forfeiting longer life by choosing hospice at the outset instead of chemotherapy.

    The apparent lesson: avoid cancer docs, and especially their non-drug cancer treatments.

    The majority of cancers, that are cured, are cured with surgery. Breast cancer, colon cancer, and prostate cancer (three of the most common cancers) and the majority of patients that are cured had surgery as the primary treatment of their cancers. You could add lung cancer to that list in that the majority of lung cancer patients that are cured have either surgery or radiation, but early stage lung cancer is less common than the later stages which are not usually curative.

  • http://daedalus2u.blogspot.com/ daedalus2u

    This is complete nonsense. You are confusing association with causation. Of course places where a lot is spent on medicine have higher death rates. People who are dying have a lot of medical care.

    Why don’t you look at what buildings people die in? I am sure you will find that hospitals have the highest death rate per square foot. Why don’t we blow up all the hospitals? Then no one would die in a hospital so the death rate would be a lot lower?

    Why don’t you look at the death rate depending on how much people pay in taxes? I am sure that there is a good negative correlation between how much people pay in taxes (how much, not rate) and their death rate. If so, then the “logical” inference is that raising taxes will save a lot of lives by reducing the death rate.

    I haven’t been able to download the thesis, so maybe I am being harsh.

  • Dániel Varga

    > For you stat whizzes, Lee uses […] as med spending instruments to disentangle the direction of causation.

    > Lee finds (in Tables 5,6) that med spending hurts mainly because increasing non-drug med spending by 10% increases cancer deaths by 2.1%

    I am not a stat whiz, and I didn’t even really check the thesis. But I have to say that with my priors, the most plausible explanation of Lee’s finding is that he did not manage to successfully disentangle the direction of causation, and it actually goes the other way. That is the natural direction of this arrow.

  • http://www.flowidealism.org Michael Strong

    Thanks for pointing us to Ken Lee’s dissertation, Robin, lots of great material here.

    On this particular issue, do you know if chemotherapy is defined as a drug-expenditure or non-drug expenditure?

    • http://hanson.gmu.edu Robin Hanson

      Ken tells me: “Chemotherapy is defined as a non-drug expenditure. Drug expenditures are retail sales that occur in pharmacies and drug stores, supermarkets, mail-order, and other mass-merchandising outlets. All in-hospital treatments are included in the Hospital Care costs which are lumped in the non-drug expenditure category.”

  • http://hanson.gmu.edu Robin Hanson

    max, if screening leads to more surgery and surgery cures cancer, why does screening not reduce death rates?

    daedalus2u and Daniel, why is it that only places with higher cancer death rates spend more on med? Why not also places with more heart attacks?

  • http://daedalus2u.blogspot.com/ daedalus2u

    I just noticed. He used:

    “med spending and age-adjusted deaths”

    Did he use age-adjusted med spending?

    Cancer is something that doesn’t just happen in old age. Heart disease mostly affects older people.

  • Alrenous

    Even if we actually live in a world where these cancer treatments are helpful, the odds of them being meaningfully helpful is, for the purposes of future decisions, zero.* A large helpful effect is inconsistent with Lee’s evidence, even assuming all the common mistakes he could have made. To get a large helpful effect, you have to assume pure fraud, which is pretty hard to pull off with this level of transparency.

    *(Not zero for the purposes of future evidence.)
    (Similarly, the Taiwan radiation hormesis apartments. 97% mortality reduction is obviously way too high, but it certainly rules out bonus mortality.)

    Unfortunately that does imply fraud or incompetence onto previous cancer treatment studies. More unfortunately, there’s already lots of evidence that medical research is broken. We already know that clinical studies are biased in favour of treatment, from such things as subscriptions for drugs that have a whopping 1% chance of curing you. It’s better than a placebo. Double blind. Everything. Also pointless – the side effects probably increase your mortality more than the drug decreases it.

    Indeed, the treatment’s side-effects (such as stress) outweighing the benefits just makes sense. Clinical studies don’t check for cost effectiveness, and so it is likely most treatments are not cost effective. Also, it is consistent with Lee’s analysis. At some point, it becomes unreasonable to doubt.

  • Mike

    I’m intrigued by this idea, but wonder how far it goes. Surely zero medical spending isn’t better than western civilization (or just US) medical spending levels? Trauma care is better than no trauma care? Herd immunity due to vaccines is better than treating each case of measles (or, in theory, small pox?) individually?

    What about things that aren’t really even medical spending but alleviates the need for medical spending, like water fluoridation or catalytic converters in our cars??

    Are we in a transition phase where we need to spend a lot of money on more or less useless care in order to come upon the treatments that will be cheap and effective?

    On an individual scale, cancer treatment isn’t terribly helpful and perhaps even harmful, but collectively, we’re learning the biochemistry of our bodies that we need in order to create more effective treatments. Without tens of thousands of people taking cancer drugs, getting biopsies, or having brain tumors removed, we just wouldn’t acquire sufficient understanding.

    I suppose I’m wondering what are the realistic alternatives to the care we currently get? What are we sacrificing by switching to an ostensibly less harmful alternative? Even disregarding the emotional cost of not using all possible therapies, are we delaying better treatments in the future?

    What if all medical care was treated as medical experimentation, so every treatment was part of large (or small) scale studies? Would the fact that all outcomes, positive or negative for the patient, counted directly towards better treatments in the future make a difference?

  • http://reason.wikia.com/wiki/User:Rimfax Rimfax

    daedalus2u said:

    Cancer is something that doesn’t just happen in old age. Heart disease mostly affects older people.

    According to the SEER Cancer Statistics Review, the bulk of cancers occur over age 65. (I can’t quite figure out how whether these figures indicate that 90% of cancers are diagnosed over 65 or 50%, or something else entirely.)

    What’s the percentage for heart disease under 65?

    • http://daedalus2u.blogspot.com/ daedalus2u

      If you look at table 9 in this

      http://198.246.98.21/nchs/data/nvsr/nvsr56/nvsr56_10.pdf

      Deaths: Final Data for 2005

      The death rate for diseases of the heart in 85 and older is ~3x higher than the death rate for neoplasms in the same age group. The total death rate for the entire population for heart disease and neoplasms is only different by ~10%. There are many more heart disease deaths in the above 85 group than there are cancer deaths in the above 85 group. In the below 75 age group, the death rate from cancer is higher than the death rate from heart disease in every age group.

      If you look at table 10, they have the actual numbers of deaths. What this shows is that:

      For diseases of the heart there were 217,894 below age 75 and 434,197 above 75

      For neoplasms there 309,436 below 75 and 249,876 above 75.

      So more young people die of cancer than young people die of heart disease and more old people die of heart disease than die of cancer.

  • http://www.glenraphael.com Glen Raphael

    why is it that only places with higher cancer death rates spend more on med? Why not also places with more heart attacks?

    I haven’t read the study yet, but that one seems intuitively likely, because both cancer patients and cancer treatment centers are somewhat mobile. Suppose there are only a few places in the country where one can go for a ludicrously expensive operation that is considered an heroic last-ditch effort to remove an otherwise inoperable cancer. People who are probably going to die are drawn to these places and these places are drawn to where the sickest potential customers are. If somebody who gets sick in Denver moves to Florida for expensive treatment and then dies he makes the positive correlation between non-drug spending and cancer death stronger in both states. (Heart disease and drug-based treatment don’t have the same dynamic since people don’t go to a world-famous doctor or treatment center for those.)

    • Dave

      That might be a factor. Another reason might be that drugs for heart disease work,thus reducing death. Deaths for cancer are biased by the fact that all cancer patients who die get non- drug treatment. Now we find out that this is defined in the study as surgery and/or chemotherapy in a hospital. You know that all patients with terminal cancer get some of this. If they got well with non- hospital treatment,they had less severe cancer. The Robin’s line if thinking could be a reversal of cause and effect that gives rise to counterintuitive results. That is what makes docs think up counter- explanations.

      • Dave

        Sorry for the typo in second to last sentence .

  • Pingback: Overcoming Bias : Death Cause Correlates

  • Pingback: Linkblogging For 25/05/11 « Sci-Ence! Justice Leak!

  • Pingback: Overcoming Bias : School Death Puzzle

  • Pingback: The Amateur Thinker » Links of the Day