Does Eating Ice Cream Lead to Drowning? Not If You Understand Confounding

The data will not analyze itself.

René F. Najera, MPH, DrPH

--

A person wearing a yellow shirt is splashing in the water.
Sherbet would have been a better choice. (Photo by Marino Linic on Unsplash)

Suppose you’re an epidemiologist at a local health department in a resort town. The town council just had a person give testimony that ice cream leads to drowning, and they have the data to prove it. They pulled 700 random records from the local ice cream shop over the last five years, and they cross-checked with drownings. According to the data, the odds of drowning given that a person ate ice cream are 4.7, and the odds of drowning given that a person did not eat ice cream are 1.7. The odds ratio is 2.7, meaning the odds of drowning if you ate ice cream are 2.7 times the odds of drowning if you did not eat ice cream.

Here is a screen capture from their data:

No Ice Cream For You!

Your boss at the health department wants to know if this is true, and if they should close down the ice cream shops. Of course, the ice cream shops don’t want to close. It’s a little bit of a conundrum, and your boss wants answers. Your boss is getting ready to issue a press release warning people not to eat ice cream, or to avoid swimming if they do.

Being the good epidemiologist you are, you ask the person who testified before the council if they could provide the data for you to check. They do, and you see the data also included information on the time of the year when the buying of ice cream and drownings occurred. You notice something interesting.

(If you want to look at the data, click here to download it and play with it yourself. In the dataset, 1 = “Yes” for eating ice cream and drowning, and 1 = “summer” for summer or non-summer events.)

First, a Word About Spurious Associations

There’s this weird thing that happens in the random universe we live in. Things may seem like they are associated or linked just by chance. You might think one thing led to another if the “one thing” comes before the “another.” It’s human nature to see patterns where there aren’t any. We evolved to see lightning and expect thunder, or see something that looks like a snake and stay away from it. (I’ve jumped five feet in the air over branches when out for a jog in the…

--

--

René F. Najera, MPH, DrPH

DrPH in Epidemiology. Associate/JHBSPH. Adjunct/GMU. Epidemiologist. Father. Husband. (He/Him/His/El)