Childhood mortality and vaccines

One of the ideas that gets presented as fact all too often on the internet is “the United States is the most vaccinated country in the world and has one of the worst childhood mortality rates”. There are variations of this, of course. Unfortunately, this notion gets put forth by autism-parents and even autism-parent organizations.

This sticks in my mind since a rather blatant attempt at misinformation from Generation Rescue in the form of a pseudo-paper “special report”: AUTISM AND VACCINES AROUND THE WORLD: Vaccine Schedules, Autism Rates, and Under 5 Mortality. I wrote about the many failings of that document at the time.

One major failing in the childhood mortality comparisons is that the U.S. measures infant mortality (which is a big piece of under 5 mortality) differently than other countries. As Bernadine Healy (a source highly respected by groups such as Generation Rescue) wrote:

While the United States reports every case of infant mortality, it has been suggested that some other developed countries do not. A 2006 article in U.S. News & World Report claims that “First, it’s shaky ground to compare U.S. infant mortality with reports from other countries. The United States counts all births as live if they show any sign of life, regardless of prematurity or size. This includes what many other countries report as stillbirths. In Austria and Germany, fetal weight must be at least 500 grams (1 pound) to count as a live birth; in other parts of Europe, such as Switzerland, the fetus must be at least 30 centimeters (12 inches) long. In Belgium and France, births at less than 26 weeks of pregnancy are registered as lifeless.

But why bring this up again? The reason is simple: I found a very interesting source of data and in reviewing it, I found information on vaccines and on childhood mortality: the Google Public Data Explorer. The Wold Bank dataset includes childhood and infant mortality figures.

What does the childhood mortality rate look like as a function of time for the United States? Not surprising (to most) it has been dropping over the past 30 years. In fact, from 1989 to 2009 the rate dropped from 12.1 per 1,000 to 7.8 per 1,000. (click to enlarge):

Why pick 1989 onward? This is the period when the vaccine schedule in the U.S. increased dramatically. If the idea that vaccines are somehow linked to worse childhood mortality we would expect this trend to be increasing, not decreasing.

Here is a good example of why we can’t say that correlation means causation. Consider childhood mortality for a country. Consider CO2 emissions for a country. Guess what, there is a big trend towards lower childhood mortality with higher CO2 emissions. (click to enlarge)

The “effect” (quotes mean it isn’t real) is huge. Note that the graph is a log-log plot. Countries with high CO2 emissions have 20 times, or more, lower childhood mortality. If we were in the “correlation equals causation” camp, we would decide that CO2 prevents childhood mortality. We could take this another step into the ridiculous and say, “Since CO2 emissions will coincide with higher atmospheric mercury due to coal burning and other sources, mercury must prevent childhood deaths”.

So keep that lesson in humility in mind as we play armchair epidemiologist and look further into the World Bank data. What is correlated with childhood mortality that might make sense? Being from a country in sub-Saharan Africa is correlated with high infant mortality rate. Low income countries have high infant mortality rates. Having a skilled person to attend the birth is correlated with low infant mortality rates.

Vaccines? What about them? They only have data for measles vaccine uptake. Again, not surprisingly, childhood mortality is lower for countries with higher measles vaccine uptake (click to enlarge)

I chose 2003 for the year for this comparison. That year has data as well for the fraction of births attended by skilled health staff. The datapoints are color coded with this to show that this is a big correlate. The more births have a skilled health worker in attendance, the more kids live. Could be a proxy for some hidden variable, but it makes some level of sense that having a health worker would reduce infant mortality. It also makes sense that countries with access to healthcare in general would have lower infant mortality.

But, that brings us back to the measles vaccine and infant/childhood mortality. Does the vaccine reduce infant mortality? Certainly in countries where measles is endemic. But measles vaccination isn’t the reason why childhood mortality figures are higher in, say, Chad than in the United States. And that’s why researchers try to control for other factors, like wealth and access to health care, when trying to correlate factors and diseases.

Otherwise, you end up with “mercury causes autism”. Or, using the World Bank data, “Cell phones cause low fertility rates”. Or other strange ideas.

While I think these data show pretty clearly that childhood mortality is not likely increased by vaccines, they also show the pitfalls of being an armchair epidemiologist. With the internet, data abound. One can find many correlations. Some are just random. Some are due to some unseen variable. Some are an indication of actual causation.

Do I believe that there is a reason why childhood mortality is lower in wealthy countries? Yes. Do I believe that there is a reason why childhood mortality is lower in countries with high CO2 emissions? No*. Both show correlations. What about the idea that measured autism rates went up as the exposure to thimerosal increased? Sure, there’s a correlation, just like with CO2 and childhood mortality. And, just like with childhood mortality and CO2, there are other factors at play.

*note–CO2 emissions are linked to countries with greater wealth. In that respect, yes there is a reason for the correlation. But there is no direct correlation of CO2 and childhood mortality.