• Infant Mortality Comparison

    If Austin’s going to ask you to help, so will I.  I’d like to hear arguments against the following:

    Many people have told me that infant mortality used to be higher in Canada than in the US, but since the passage of (Canadian) Medicare, that hasn’t been the case.  The chart above, which I made from OECD data, would tend to agree.

    I know the usual knocks against infant mortality as a population metric of quality.  But I’d like to hear a good alternate explanation (if one exists) for the trend you see above.  Links to evidence or data supporting your theory will get you extra points.

    If no one comes up with one, would this change anyone’s mind about the effect of health insurance on infant mortality?

    • Without knowing anything and without evidence or data (so no extra points for me):

      1-As we all know, you can’t conclude a cause and effect relationship from a correlation.

      2-It looks like the trend started before Canadian Medicare was enacted.

      3-Along the lines of #1, the U.S. and Canadian populations are qualitatively different and likely have become more different since the 1960s.

    • @Charlie

      Let me jump in early so we head in the right direction:

      1 – I know. I’m asking for counter-arguments.
      2 – Yes, but the argument is that as Canada adopted more universal coverage in the early to mid-60’s things got better.
      3 – Different populations might account for static differences, but not the trend. Unless you can explain changes in the populations over time that would explain the trend. I have yet to see one documented by evidence.

    • How about increases in immigration.

      Here’s a chart showing the increases in immigration to the US:


      I can’t quickly find an analogous chart with respect to Canada (too much PPACA work to do that I get paid for!) but if the immigration rate to Canada over the same time period was less, that could be an alternative explanation (albeit only another correlation) to the disparity in infant mortality rates.

      N.B. The increase in the disparity seems to have leveled off in 1992 or so. Any explanation for that?

    • We shouldn’t forget that Medicaid began in the U.S. during the late 1960s. Medicaid expansions have been shown to lower infant mortality as cited often by Austin in his blogs.

    • Also, as Victor Fuchs points out the birth control pills was first introduced in the early 1960s so less unwanted children.

    • How do the infant mortality rates of white Canadians compare to white Americans?

      Alternatively, remove minority data from American data and place it in Canada’s data and see if you still think the model of delivery has anything to do with the problem you’re trying to solve. Likely our diversity and history of racial discrimination explains a lot.

      How long will healthcare debates refuse to acknowledge that we have a large minority that–for whatever reason(s)–distorts US data? Any national data in the U.S., such as Education, will appear to be worse than it is. It’s not the system, but a cultural and/or discrimination problem.

      Given the large minority population of the US, all things being equal, it likely out-performs Canada.

    • Not a direct counter-argument as I dont have the data, but can we please stop making comparisons like this unless we are going to control for heterogeneity amongst people.

      A much more proper comparison:
      Infant mortality rates for first generation Canadian ex-pats in America as a percentage of Infant mortality rates for Canadians
      Compared to
      Infant mortality rates for first generation American ex-pats in Canada as a percentage of Infant mortality rates for Americans

      Does moving from Canada to America and vice versa increase or decrease infant mortality?

    • The broader point I think you’re making though is the trend. At which point, heterogeneity wouldn’t really matter.
      There seems to be a lag of a few years (why?) but the Canada curve drops really quickly right after the HIDS act, and the US curve drops really quickly right after Medicaid – again after a lag (why?).
      The argument would then stem to “HIDS does a better job at reducing infant mortality than Medicaid does.” This seems to be pretty clear. Even ignoring absolute levels, the Canadian slop is much, much steeper.
      If you have already done this, please point me to it, as I have only been reading for a few weeks, but could you list the major differences between HIDS and Medicaid, so we could find natural experiments to test each of those components to find out why HIDS has done such a better job?

    • Aaron: “At which point, heterogeneity wouldn’t really matter.”

      It would matter because he’d be trying to figure out what was wrong with Canada’s system, which, if as I’ve speculated, would be chronically BELOW the U.S.

    • Okay, so I made some really schizophrenic points there, so let me back up.
      Heterogeneity matters if you’re looking at the absolute level, as people are different in Canada than American; ie. general health, income differences, genetic differences, blah blah blah
      But if you are going to look at the trend, that is, compare Canadians to Canadians and Americans to Americans, heterogeneity shouldn’t matter, or more importantly, it should control for itself.

    • I searched around. Of interest, I found link below. Study does capture heterogeneity within Canadian population of a unique ethnic group. There is a mention of improved survival of native Canadians when they move to US, but outcome is purely speculative:

      “It is surprising that urban FN infants fared much worse in British Columbia than in the US within comparable time periods, given the presence of a universal health insurance system in Canada.”

      Anyway, I tried.


    • I don’t think the immigration argument works. I can’t dig up historical data, but:
      the above link implies that about 13% of the U.S. population is foreign born in 2008, while

      this 2007 article says that 20% of Canadians are foreign born.
      I know that Canada has been actively promoting immigration over the past 20 or so years, as its a big country that could use more people, but it’s a cold country so immigrants shy away. I recall a lot of Cambodian boat people wound up there.

    • Brad,

      I tried reading the article – social science journals I can read, medical journals… yeah not so much, so take this with a grain of salt as I may have misunderstood something – and I didn’t so much get that people fare worse when moving to the US from Canada but rather that similar immigrants in the US did better than in Canada.

      This is actually an even better result than what I was thinking of trying to find as it also helps correct for the endogeneity problem (people that have a higher infant mortality rate might immigrate more often, for example, but because you are looking only at immigrants, this problem is solved) . It is only one study and it does offer a few possible hypotheses – none of which I find terribly convincing – but, a common trend I see in the article is Canada’s “universal” system is not really all that universal. For example, they mention these sustained disparities between rural and urban which shouldn’t exist in a universal system. I also, sadly, get the idea that the authors seem to be assuming that by definition Canada’s health care system must be better than that of the United States – whether you agree with this or not, I am very skeptical of any study that assumes this degree of subjective argument must by definition be true.

      I have found another study, from 2007 that deals with a comprehensive analysis of both systems, but it does contain a part on infant mortality that is relevant here:
      http://ideas.repec.org/p/nbr/nberwo/13429.html (gated)

      Here is the relevant section – and it goes to show why the “problems with infant mortality” the other Aaron ignored, should not be ignored – as the results reverse themselves:

      “It turns out that once we condition on infant birthweight–a significant predictor of infant health–the U.S. has equivalent infant mortality rates. In fact U.S. infant mortality is lower for low-birthweight babies than Canadian infant mortality for low birthweight babies. Overall infant mortality, however, is higher in the U.S. because the incidence of babies with low birthweight is higher than in Canada. This may be due to demographic or epidemiological factors, or it may be the case that the U.S. is better at having a live birth for a low birthweight baby.”

      There could be argument that birth weights are not exogenous, but that differences in the health care systems might lead to this, but, when combined with the study mentioned earlier I get a decent picture of what I think is going on here. Canada is really good at keeping its citizens – particularly its urban citizens – very healthy and this translates into its citizens’ children, but because of persistent geographic and demographic disparities, this has nothing to do with universal coverage, rather something else entirely. As a result, when you add in an exogenous flow of people, who have not been subject to either the Canadian or US health care systems for their entire lives, the US does better at preventing infant children from dying.

      The key point here, which goes directly to the other Aaron’s point – nominal universal insurance, and by extension, increased access to insurance (at least between the US and Canada) do not lead to better infant mortality results. At least once you hold wealth relatively constant.

      • RE: “Canada is really good at keeping its citizens – particularly its urban citizens – very healthy and this translates into its citizens’ children, but because of persistent geographic and demographic disparities, this has nothing to do with universal coverage, rather something else entirely”

        No it doesn’t. Under a capitalist model, it doesn’t matter where people live, if you live remotely and you need medical care, than the price of that medical care will simply be higher. However, in Canada, rural communities do not have access to state run healthcare because Canada rations it’s dr’s & health care accordingly.

    • I don’t have any empirical data to add, but I would like to question the logic of some of the comments here.

      Aaron’s post is about health insurance. His hypothesis is that raising the rate of insurance in a population would tend to bring down infant mortality. As far as I can tell, most of the purported objections raised in the comments are just irrelevant to this point.

      @ those who talk about immigration/minorities. Aaron correctly points out that heterogeneity in the population cannot explain the different rates of change in infant mortality (at least if heterogeneity doesn’t change over time).

      But suppose we just think about absolute numbers for a moment. And suppose it’s true that infant mortality in the US is concentrated in immigrants/minorities (I have no data one way or another). It may still be the case that the reason for this is that many members of these groups are uninsured. Unless it is shown that this is not the correct explanation, the concentration of infant mortality in these groups (assuming it exists) is no evidence against Aaron’s hypothesis.

      @The other Aaron. You suggest comparing “infant mortality rates for first generation Canadian ex-pats in America as a percentage of infant mortality rates for Canadians, compared to infant mortality rates for first generation American ex-pats in Canada as a percentage of infant mortality rates for Americans. Does moving from Canada to America and vice versa increase or decrease infant mortality?”

      Again, this comparison is entirely useless unless you control for insurance. For example, all the Canadian ex-pats in the US I know are either grad-students or academics. In other words, they all have health insurance. Comparing them and their kids to Canadians back home is neither here nor there when you try to assess the effects of having health insurance.

      As for the higher incidence of low birth-weight in the US than in Canada — what exactly is your evidence for asserting that this has nothing to do with the parents’ having health insurance?