March 26, 2010
Erroneous Evidence about Enough Exercise
This week, a study in the Journal of the American Medical Association received a lot of undeserved media attention. The study wanted to examine the relationship between exercise and long-term weight changes among women who were eating a normal diet (i.e. not dieting). It followed for over a decade 34,000 women who were 45 years old or older and correlated their self-reported physical activity and body weight.
The study found that on average, the women gained about 6 lb during the study. Among women who initially had normal weight (body mass index less than 25) there was a significant correlation between amount of exercise and maintenance of weight. Women with initially normal weight who did at least 60 minutes a day of moderate to intense exercise maintained their weight, while those who did less tended to gain weight during the study.
The authors therefore concluded that for middle-aged women who are not dieting, 60 minutes of moderate exercise daily is necessary to prevent weight gain. This conclusion was repeated in much of the media coverage (links below) trumpeting that women should be exercising much more than we previously thought necessary.
But hold on a minute! First of all, the study is observational, not randomized. If you really wanted to know the effect of different amounts of exercise on weight you would randomly assign women to different quantities of exercise, make sure they were doing the assigned amount, and follow their weight. That’s not what happened here. The women exercised as much or as little as they wanted, and that amount was correlated with their weight change. But that means that anything that affects both exercise and weight could have skewed the results. Women with chronic illnesses that cause weight gain (hypothyroidism, heart failure) would tend to feel too tired to exercise and also gain weight. These women would tend to make the statistics look worse for sedentary women, though their weight gain had nothing to do with being sedentary.
Also, the amounts of exercise was self-reported, not observed by someone objective, making it possible that women with stable weights are simply more likely to exaggerate their reported exercise. (Which reminds me, I have to take it easy this weekend after running 3 marathons and swimming up the Mississippi River this week.)
Finally, the correlation between exercise and weight gain was only found in women with normal weights. In women who started with a BMI over 25, there was no connection found between how much they said they exercised and how much weight they gained. Does that mean that overweight people shouldn’t exercise? No. It means that there’s nothing to learn from correlations and that we can only learn from a randomized experiment.
So this tells us nothing about how much women should be exercising to maintain their weight. Perhaps it tells us that some conditions cause weight gain and inability to exercise. Perhaps it tells us that thin women exaggerate when reporting their exercise habits. Perhaps it tells us nothing.
So how can you tell how much exercise you need to maintain your weight? Weigh yourself. If you’re gaining weight, you should exercise more.
Journal of the American Medical Association article: Physical Activity and Weight Gain Prevention
Los Angeles Times article: Women should exercise an hour a day to maintain weight, study says
Wall Street Journal article: New Exercise Goal: 60 Minutes a Day
Important legal mumbo jumbo: