Statistical Fuzz

So the Institute of Medicine of the National Academy does a study saying that universal health insurance, i.e., socialized medicine, would save 18,000 lives a year. A former Clinton health care advisor says that it would save at most 9,000 lives a year and probably none. Who’s right and how could we tell?

Well, we can’t. The U.S. total annual death rate is 8.28/1000 which comes to 2,484,000 deaths a year. 18,000 is 0.72% of 2,484,000. That means a change in the death rate from 8.28 to 8.34. That means that both estimates of lives saved are so minor compared to the overall death rate that the differences are completely lost in statistical fuzz. Both parties are wildly irresponsible to even pretend that they can estimate such an impact. 

Why we would we even suspect that government paid health care would save lives? There is no evidence that anyone in America is going without necessary health care.  Look at the death rates (per 1,000) for countries with socialized medicine: France 8.48Netherlands 8.71, United Kingdom 10.05, Germany 10.8. These higher rates are largely due to Europe’s aging population, but clearly the gains proposed for the U.S. adoption of politically-managed health care would be swamped by minor changes in population age, birth rates or immigration. 

These studies are nothing but a cluster of untestable assumptions. The effects they predict are so trivial that we can never confirm or refute their assertions by subsequent observation. Studies like this are depressingly common. It’s just plain silly that we try to make policy based on purported measurements that no one can actually make. These studies are cargo cult science. They have the seeming of science without the actual experimental rigor that makes science a predictive endeavor.

We’re no better than our pre-scientific ancestors who pored over scripture to justify the secular decisions they’d already made. Instead of scripture we whip out a “scientific” study that might as well be ancient writings. 

10 thoughts on “Statistical Fuzz”

  1. Shannon,

    There was an intelligence squared debate (npr) about whether the fed gov should be responsible for health care (LUN)

    I found it pretty confusing – I didn’t finish the hour or so talk because I found the two sides talking past each other. One thing kind of confused me – John Stossel brought up that if you control for violent deaths and auto accidents, I think, the mortality rates of the US were better than the countries with socialized medicine, which the proponents were using as a metric for why we should switch. No one followed up on that point. It’s one of the frustrating things about the health care debate – apples versus oranges, and the statistical noise you bring up, too.

  2. Oh, but your numbers are different than those of the talk, because they said the mortality was better in Canada and the like? Oh, good grief, no wonder I dislike health care policy :)


  3. USA Healthcare is First – Infant Mortality is Low

    John Stossel (Why the U.S. Ranks Low on WHO’s Health-Care Study) analyzes that life expectancy is a lousy measure of a health-care system. We have far more fatal transportation accidents than other countries. Our homicide rate is 10 times greater than in the U.K., eight times greater than in France, and five times greater than in Canada.

    When you adjust for these rates of injury (not sickness), U.S. life expectancy is actually higher than in nearly every other industrialized nation. That doesn’t show a health-care problem.

    The infant mortality statistics are also analyzed with bias. The US counts every live birth, however premature, toward its statistics, even if the infant lives only a few hours. European countries may only count infants that live at least a day or which meet other health criteria. So, they claim fewer infant deaths.

  4. The statistics that cannot say their names record the difference between the black urban underclass and the European descent white population. Hispanics are in the middle, troubled by genetic propensity for diabetes and obesity issues having to do with culture and diet. Violence is much lower, as low as in European states, and life expectancy and infant mortality are better. There is actually a racial factor in infant birth weight and mortality. There was an Army study a few years ago showing that black women had lower birthweight babies when all other factors were controlled. Hispanic women had higher birth weight and lower perinatal morbidity than non-Latin whites. This was in Army hospitals with fully employed service members and dependents so the socio-economic factors weren’t there.

    There actually is a good model for US health reform; France. The French system is mostly funded with employee contributions and is total fee-for-service and free choice with lots of private care. Two factors that make it work are a national fee schedule (with an opt out provision for doctors) and free medical school tuition. Those two factors offer a possibility of a transition for us. Obama will not be interested but a Republican administration could do it and get health reform off the table.

    I have some analysis here.

  5. Andrew_M_Garland,

    The infant mortality statistics are also analyzed with bias. The US counts every live birth, however premature, toward its statistics, even if the infant lives only a few hours. European countries may only count infants that live at least a day or which meet other health criteria. So, they claim fewer infant deaths.

    Yes, this is an example of an arbitrary measurement that is treaty as hard data in public debates. It would seem like nothing could be more obvious that a dead baby but when you start looking into the specifics things get fuzzy. Some countries do not count babies born preterm as live births. Other countries require that babies live 24-72 hours to count as a live birth. In America, any baby, born at any term, is consider a live birth if it lives an hour. But even here their are differences between state to state. So counts of infant mortality really don’t measure the same physical phenomena.

    When you get to estimating how much increased availability of every conceivable medical treatment will save lives you have to make so many assumptions and combine so many hundreds or thousands of measurements along with each measurements error that you end up with pure gibberish.

    Even so, if the change was large enough, say a few percentage points, you would have a change large enough that you could detect it following the transition to socialize medicine. With a change this small you would never know if the predicted improvements ever materialized.

  6. There are so many differences in definition and measurement in this kind of international comparison as to make all such comparisons suspect.

  7. “There are so many differences in definition and measurement in this kind of international comparison as to make all such comparisons suspect.” – Jonathan

    And, yet, such comparisons are being used by health care professionals, politicians, activists and health policy analysts to make the case for single-payer healthcare, or healthcare in the model of the current favorite enlighted country (never the US, you know?)

    Is there no one in current American political life who can point these things out? Why are the current group of Republicans, putatively conservative (ha) so unable to articulate what you ALL can? I think it’s because the political class really doesn’t believe in limited government – that’s why they can’t articulate a simple thing like, “your math is wrong and here is why” when confronted with a piece of political theater.

    Ranting, slightly I know. But it makes me happy……

Comments are closed.