When is a 41% reduction in heart attacks questionable? When the research is skewed.
Read below the fold for more!
Folks-
One of the biggest problems that I have with research results published in the popular press is how poor the science education is with the reporters, and how much it shows in the reporting.
Today, the AP reported that a smoking ban in one part of Colorado was responsible for a 41% decline in heart attacks there according to a study done by the CDC. Nine other studies had shown a reduction in heart attack risk, but nothing this dramatic.
I’m skeptical. Nearly halving risk for an entire population by modifying the behavior of just roughly 20% of the population usually doesn’t happen. Skeptics are pointing out problems with the study, but unless you read all the way down to the last couple of paragraphs in the story, you won’t see that. This is typical of this kind of reporting.
Other studies in the past have had similar problems. In the ‘90’s there was a study done of the risks of cell phone usage in relation to automobile accidents. It found a high correlation between cell phone usage and auto accidents. But there was a problem with the study—if a cell phone was found in any car involved in the accident, it was assumed that use of the cell phone was responsible for the accident. The assumption was that nobody would admit to being on the phone while driving. That is a poor assumption. If continued to today, you would find that almost no accidents happen where use of a cell phone is not responsible.
In the early ‘90’s, California implemented a helmet law for motorcyclists. This was done to improve motorcycle safety. The next year when statistics were released by the state showing a reduction in deaths due to motorcycle accidents, it was touted that wearing helmets reduced deaths. However, there was one thing that was seldom mentioned: The same law changed how motorcycle accident deaths were recorded. Prior to the new law, if you got into a motorcycle accident and died up to a year later from the injuries caused in the accident, you were counted as a motorcycle accident related fatality. The same law changed the definition of a motorcycle related fatality to include only those who died at the scene of the accident. It was inevitable that the death rate would drop, and from a statistician’s viewpoint the two years cannot be compared. The statistics are incompatible.
So, when I see that one little change in the behavior of a small part of the population yields a huge change, like in the Colorado smoking study, I have to wonder if it was the change in behavior or something else. In this case, I’m betting something else.