An article that’s almost 2 years old has resurfaced recently (link, link, link) that ranks the US as last in preventable deaths among 19 industrialized countries. Most coverage of this story simply cites the statistics without pictures. Last place seems pretty bad, but as we’ve discussed earlier, someone always has to be in last place so the issue isn’t so much the rank as it is the position in the rank. Without pictures, those who read those stories who are on the right might imagine all 19 countries clustered closely together. Those who lean left might imagine all but the US clustered closely together with the US a clear laggard. Clearly different conclusions can be drawn from these images despite the fact that verbally both are characterized by “The US ranks last among industrialized countries on preventable deaths.”
The standard way these sort of statistics are graphed is via bar charts. An advantage of bar-charts is it shows the total quantity. If you’re interested in the distance a quantity is from zero then highlighting the entire span of the data from zero is a useful thing. But if you’re interested in a comparison to others then the distance from zero is not so useful. The actual ranking of all 19 countries in the study is shown in the 2 graphs below. The graph on the left is the standard bar chart representation of the data. The graph on the right shows the same information but in a slightly different format. First it is zoomed in on just the range where the countries lie; it doesn’t go all the way back to zero. Next it just plots the position of each country. France, for example had a preventable death rate of 64.79 (slightly less than 65 people out of 100,000 died in 2002 due to causes that the authors called preventable). France’s ranking is shown as a red cross at position 68.79.
If we remove the country names the clustering becomes more clear. France is significantly ahead of 2nd place Japan. Then a big gap between Austria and Germany, next another big gap between New Zealand and Denmark and finally a big gap between Portugal and the US. School teachers who grade on a curve may be tempted to assign an A to the first group, B to the second and so on. The US would be in the last place “F” group. This is pretty dire for the US. Almost any rearrangement of the statistics would show the US in a more favorable light.
Keep in mind that one hypothesis in the health care debate is the US has the best health care system in the world. If that’s true then at worst we’d expect top marks in some areas and moderate marks in other areas. Moderate marks are not inconsistent with the best health care system in the world. The US is a diverse country and has a much more diverse population than many European countries. It could be the case that, where the US ranks only moderately, if you adjust for this diversity the US could again rank in the top. Unfortunately there have been no studies to this effect, but the effect may still exist.
However, we wouldn’t expect to see the US in last place on any of these measures. Even worse, we wouldn’t expect the US to score last by a significant margin. If the US scored closer to Portugal, then visually the US could be lumped in with the rest of the countries that scored a “D” and the case could be made that this is not a last place score but instead is a low but moderate score. However it is much harder to make that argument with the actual ranking of the US. So the US ranking is pretty dismal. But let’s examine the data a little further.
What’s a preventable death? It is fair to say this is a judgment call. But all of us can agree that there are some conditions that should not be fatal or fatal only rarely. People shouldn’t die from surgical errors and conditions like pneumonia and appendicitis, so most would agree these deaths are preventable. (Note we’re not talking about preventing the conditions themselves, but rather preventing death from these conditions.) The study’s authors used their judgment in a similar way to categorize many other conditions as preventable. For example, deaths from other conditions that are chronic but should allow for a long life like epilepsy and diabetes were also considered preventable. Less obvious are things like cancers and heart diseases and there can rightly be some disagreement on whether these should have been included however they were included for all countries so any error is thus mitigated. (Furthermore the authors of this study were following up on an earlier study and so listed the same conditions as preventable.)
The actual study did a deeper comparison between the US, UK and France, breaking down the data between various conditions and also comparing the first year of the study (1997/98). Looking at that data one would notice 23 points of the 45 point gap between the first place France and last place US comes from heart disease alone. One blogger thinks this fact should rule out this study as being flawed. However, if the US’s scores from the 1998 study improved at France’s rate of improvement, the US would have placed 15th, just ahead of Denmark (improve). And if the US had France’s overall scores on everything but heart diesase it would have placed 12th, just ahead of Germany (replace).
Certainly the difference in the population between the US and France could account for the US not getting first. But it’s harder to make the diversity argument when referring to the rate of improvement. The best health care system in the world should have a rate of improvement over a 5 year period at least tied with France.
At the time I researched this article the original study was available for free on their website. Now the link I have is showing up as a for-pay link. My apologies.