A:Occasionally we get questions and comments—such as this one—that show some confusion among readers, the general public, and especially the media about interpreting scientific studies that indicate an association or correlation between an “activity” and an “event.” There’s a great temptation to label the activity as “cause” and the event as “effect,” “outcome,” or “consequence”—although the study merely reports an association or correlation. To illustrate, let’s consider ice cream:
In one study, ice cream ingestion was positively correlated with drownings—as eating ice cream increased, drownings also increased. Was there something in the ice cream that rendered those who ate it less alert or less agile while swimming?
We know that high-fat meals induce sluggish blood flow for up to four hours after eating, and some researchers have speculated that eating causes redistribution of blood flow to the gut for digestion, so less blood flows to the muscles and brain. Or maybe eating ice cream causes stomach cramps, which then affects swimming ability.
Unrelated research shows that ice cream sales and ingestion may also be positively associated with shark attacks (as one rises or falls so does the other). Murders and home burglaries also escalate with increases in ice cream sales and ingestion. (Interestingly, the numbers of burglars who eat ice cream at the scene of the crime is nothing short of astonishing!) Moreover, in the environmental research arena, forest fires are also associated with ice cream sales!
If ice cream really causes all these ills beyond general health and cardiovascular risk, prohibiting the sale and eating of ice cream may be in the interest of the public’s health and welfare.
But before jumping to this conclusion, let’s take another look at the information. Ice cream consumption or sales is one variable: the “activity”; drownings, crimes, shark attacks, and forest fires are the other variables: the “events.” Data presented were observational correlations or associations. That means that some relationship between the activity and the events exists, but the nature of that relationship is not given and does not mean that ice cream causes the other variables to happen.
Ascribing causation when there is only association or correlation is misleading, but it is often what goes into the common interpretation of scientific studies.
For example, in May 1999 media headlines were ablaze with cautions that bedtime light exposure and nightlights in infancy cause nearsightedness later in life. The reality, however, is that nighttime lighting is used more often by nearsighted parents, who are statistically more likely to have nearsighted children.
In the ice cream example, a third “lurking” variable explains the observations: seasonal temperatures. More people swim (and, unfortunately, drown or become victims of shark attacks), and more people buy and eat ice cream in the warmer summer months. Crimes and forest fires also increase in hot temperatures.
So we must resist unfounded conclusions, even when they agree with a closely held belief. Statistical correlation and association in health science should stimulate deeper probing to find the true cause, or truth—realizing that our understanding in this area is progressive and by God’s grace.
Peter N. Landless, a board-certified nuclear cardiologist, is director of the General Conference Health Ministries Department. Zeno L. Charles-Marcel, a board-certified internist, is an associate director of Adventist Health Ministries at the General Conference.