Be skeptical of health studies linking X to Y

On Tuesday we were treated to two front page stories in the Arizona Daily Star linking a health phenomenon to a supposed cause. However, such epidemiological studies prove nothing regarding cause and effect; the link or association is merely suggestive. Frequently such studies fail to consider other possible causes or confounding factors. The association between X and Y could in fact be valid, or it could be a coincidence.

The first story, written by Tony Davis, is “UA study: Diesel exhaust here linked to childhood wheezing.”

In this case, University of Arizona researchers suggest “Infants and very young children in Tucson exposed to high levels of vehicle diesel pollution are more likely than other kids to suffer from early childhood wheezing, a potential asthma indicator.” The study involved 700 children, a very small sample size for such a study, and compared the incidence of childhood wheezing with traffic patterns. According to information in the story, the researchers did not consider some confounding factors such as allergens in the study area or emissions from gasoline-powered vehicles. They did note “A majority of children have wheezing problems in the first few years of their lives due to viral infections..” But the report of the study did not say how that factor was separated from the diesel fumes association. The study report did not dig deeply into socio-economic factors that could impact pre-natal and post-natal care. The study leaves many uncertainties. And perhaps more ominously, “The researchers are also going to see if any kinds of public policies need changing to protect such children.” What would bureaucrats do; forbid families with young children from living near major traffic routes?

The second study, “Kids may help prevent heart disease in men” reported by the Associated Press involved 138,000 men. This study, by AARP, the government, and several universities noted that men with children have lower testosterone and a lower incidence of heart disease. Unlike the first story, this one was more circumspect in its claims. The story noted “a study like this can’t prove that fatherhood and mortality are related.” The story also admitted that it did not consider confounding factors such as the cholesterol and blood pressure data, fertility of the men’s partners, nor the case of being childless by choice.

These kinds of stories make good headlines but often bad science and  unwarranted worry. They can also precipitate harmful government regulation. For instance, see my post “Ozone theory has holes.”

In that story I report that the FDA is banning inexpensive over-the-counter inhalers for asthmatics and forcing them to buy more-expensive prescription medications on the theory that the CFC propellants in the cheap inhalers are harming the ozone layer. Science has proved that wrong, but the FDA apparently hasn’t gotten the message.

So, whenever you see such a story reporting a link or association of one thing to another, be skeptical, and remember this coincidence: human life expectancy has increased since the invention of the Yo-Yo. Will we someday see this headline: “Study: Yo-Yos linked to longer life”? That headline has the same validity as the stories mentioned above.

See also (links updated):

Statistical Games #1

Statistical Games #2 Stroke for Stroke