It is often tempting to see a graph with a pretty strong correlation between two variables, nod, congratulate ourselves on having done some research, and walk away.
Such is often the case with healthcare spending.
While Americans often disagree fiercely how to fix healthcare, a solid majority believe it’s a major problem.
When we have a conviction that something is true, we have a bias not to question what seems like evidence in favor for it. Today we’ll look at a few graphs that seem simple, but are in fact quite deceptive, covering a mass of complexity and making dubious implications thereby.
You may have seen graphs comparing the average life expectancy of different countries versus their healthcare expenditures. Huffington Post’s title for the article about this is, “The US Healthcare System is Terrible, in 1 Enraging Chart.”
The purpose of the article is to convince us that the US healthcare system is terrible, and the evidence presented is a correlation of life expectancy and healthcare spending per capita. We can see the US all the way over on the right.
The regression line seems meant to imply that healthcare systems follow that line; because the US falls low of it, we conclude that the US healthcare system is terrible.
Something to Consider: how many variables go into both the amount of money we spend and the life expectancy of citizens of a country? Are all of those related directly to the quality of the healthcare system itself?
Let’s look at a few example drivers of both US healthcare spending and the US life expectancy rate.
The HuffPost graph implies that higher healthcare spending is linked with countries that have higher life expectancies. The graph below shows us that higher healthcare spending is linked very closely with higher GDP per capita. The article suggests that about half of the US’ higher-than-average spending might be explained by its higher income; the other half to inefficiency. Is it true? Good question.
Looking at both of these graphs, we can see that countries with higher incomes have both higher healthcare spending and higher life expectancies, generally. This means also that higher life expectancies correlate with higher incomes. It may be the case that peoples’ higher incomes themselves drive higher life expectancies via mechanisms other than healthcare spending. We could suppose that wealthy countries likely have fewer pollutants in their air and water, eat healthier diets, and have less violent crime--all of which would contribute to higher life expectancies without higher healthcare spending (and in fact might drive healthcare spending down).
If we compare the United States’ life expectancy to that of other countries and think about some of these variables, we might see factors at play--unrelated to the American healthcare system--that would lead to both higher healthcare costs and lower life expectancies.
We know, for example, that the obesity rate in the US is 2x higher than Germany and 3x higher than France and Sweden.
And we know obesity is expensive to treat, and that life expectancies go down for obese people--by up to 8 years. So with a little thinking, we’ve found a major factor in the US that contributes both to its higher spending and lower life expectancy (the US has the second-highest obesity rate of any OECD country) that has nothing to do with the healthcare system.
If you go back to the first graph, you’ll see that Russia, Hungary, and Slovakia also fall below that line. They happen to be in the top 5 heaviest-drinking countries in the world. Russia and Hungary are also #1 and #2 for alcoholism.
The US also has twice the murder rate of its OECD peers. Because the people that die of gun violence tend to be young (where lifestyle choices kill those who are older), these deaths will have a disproportionate impact on the life expectancy of a country.
We could go on, and you might want to take a few minutes to think about what other factors might affect healthcare spending and life expectancy beyond the quality of the healthcare system. Usually, truly understanding a political issue requires understanding its nuance and complexity: ignoring that complexity by quickly drawing conclusions from a chart with a two-variable correlation is not only unhelpful, but may be destructive, blinding us to the many drivers that are affecting the issues that are most important to solve. The graph from HuffPost is deceptively simple, and we are tempted to accept such simplicity for reality, but skepticism and thoughtfulness are critical as we think about what our policy responses should be.