Every once in a while, a study is released comparing health care in the United States to health care in other countries. Despite the fact that we are the wealthiest nation in the world, and that we have one of the most technologically advanced health care systems in the world, there are inevitably some measures by which we lag other developed nations. We have come to expect that: no one denies that our health care system has problems with cost and access; and no one denies that there are certain groups within our society — bounded by poverty and self-destructive lifestyles — for whom our health care system is woefully inadequate. Those are problems we need to address — and solutions upon which so far we cannot agree.
But when by some measure our health care system trails the undeveloped world, that attracts attention.
A recent report ranked the United States behind Cuba in infant mortality. Nicholas Kristof of The New York Times used that fact as a springboard for a broad and derisive critique of our market-based system. I agree with much of his diagnosis (and disagree with much of his prescription), but oddities in the way infant mortality is calculated in various countries makes it a dubious and slippery statistic upon which to hang a robust assessment of our current performance. I sent this to him to make that point, and to make the more general point that reporters are rarely either careful or critical enough when writing about the results of statistical studies.
12 January 2005
It may well be that there are unconscionable disparities in pre-natal care (and medical care in general) between the haves and the have-nots in this country that should be addressed. However one should be very careful using relative infant mortality rates to compare the United States to other countries in making the case for such disparities.
My wife uses medical research, among it the results of statistical studies like those that report infant mortality, to help biotech and medical companies decide where there are medical needs (and yes, market opportunities) for their research and their products. She has frequently mentioned the difficulty in trying to draw clear conclusions from infant mortality data because the methodologies for collecting it — and even the notion of what constitutes an infant death — vary so widely around the world.
Historically, and as far as I know still, in many countries an infant is not acknowledged as having been “born”, even if it was carried to full term, if it was not born alive. In fact, in some places, a baby that was born alive but dies within the first 24 hours of life was never “born” for the purposes of counting infant deaths — it is simply ignored when collecting the data. One might expect such practices to be rare in the “first world”, but they are surprisingly common, probably the result of some underlying cultural differences in philosophical views of life and death. Of course the reasoning doesn’t matter: if, for whatever reason, you only count deaths among babies who’ve lived through the first 24 hours, and ignore deaths that occur during or shortly after birth, your statistics on “infant mortality” will be lower than if you count the results of all late-term pregnancies. In the United States we pretty much count any baby born in any condition when collecting our statistics.
Another variable that can affect the infant mortality rate in an unexpected way is pre-natal care. In places where pre-natal care is utterly lacking, more pregnancies end in early miscarriage — which is not counted as an “infant death” — than in places where there is good pre-natal care available. In particular, in the United States, more high-risk pregnancies are carried to term because of the availability of excellent pre-natal care than would be true in, say, the developing world. But even with good pre-natal care high-risk pregnancies are high-risk, and are more likely than average to result in an infant death. Thus, good pre-natal care can result in much lower miscarriage rates but, perversely, a somewhat higher recorded infant mortality rate. And, in a demographic environment in which the average child-bearing age is rising as more women put off having children into their thirties, the proportion of high-risk pregnancies (or at least those that are high-risk due to the age of the mother) may be higher here than in other places.
Similarly, the advanced (and extremely expensive) neo-natal care technology now available in most cities within the United States means that premature births — even extremely premature births that would historically have been automatically counted as miscarriages — now result in live births and extensive (and often futile) resuscitation and life-sustaining efforts. We are very adept at keeping such preemies alive for a short time, long enough to count as a live birth; we are less adept at saving them over the long haul. In most places in the world — among them, I suspect, Cuba — they are not kept alive at all. Thus, a significant fraction of those premature births that would simply be counted as miscarriages in the rest of the world — and therefore not be counted in their infant mortality statistics — are counted as infant deaths here.
I don’t know what the actual comparison is between infant mortality in the United States and in Cuba, or between that in the United States and in other parts of the world. The disparity in how deaths are reported and the other confounding factors probably means that no one really does. But it is common knowledge among those who actually use this data to make strategic decisions (or at least those who make strategic decisions that put their own resources rather than someone else’s at risk) that the United States is not nearly as bad as it would appear from the raw numbers, and that other countries are often not nearly as good.
Reporters, in general, are very bad at assessing and reporting the subtleties of statistical studies. Numbers are easy to grasp; the biases inherent in the methods of generating those numbers are not as easy, and are often simply overlooked both by advocacy groups using the numbers to push their own agendas and by reporters for whom the underlying data and statistical methods may as well be written in Greek. That is a shame, because although bad statistics may occasionally (perhaps in this case, or perhaps not) support a good policy for the wrong reasons, they are far more likely to lead us astray to bad and wasteful policies that expensively do no good or make things worse.
© Copyright 2005, Augustus P. Lowell