Well, Of course exercise helps you to lose weight, right? When someone wants to lose weight, they’re inevitably told to do two things – change their diet and exercise more.
I’ve always wondered about the exercise part though. Not because I don’t think people shouldn’t exercise. I definitely think everyone should engage in some form of daily exercise, due to the many different health benefits they accrue. I just don’t necessarily think people will lose weight. This is for two reasons. Firstly, when you exercise, your appetite increases, so you eat more. Secondly, when you exercise, you generally increase your muscle mass, and muscle weighs about twice as much as fat.
Anyway, I decided to look at the evidence, and see whether people can reasonably expect to lose weight if they exercise more, or whether they should be focusing more on the diet aspect, if weight loss is their primary goal. The reason I think this is an interesting topic is because many overweight people have trouble losing the excess weight, and I think part of that is often due to an excessive focus on exercise, while not paying as much attention to the diet side of the equation.
There has actually been a ton of research on this topic, which makes it hard to be systematic and look at everything that’s been done. Instead I’ve tried to find a selection of the most relevant reviews, and focus on them.
In 2001, a meta-analysis was published in JAMA (The Journal of the American Medical Association) looking at the ability of exercise to achieve weight loss in patients with type 2 diabetes. 14 controlled trials were included in the analysis (of which 11 were randomized), with a total of 368 participants. The average age of the participants was 55 years old.
Twelve of the trials looked at aerobic exercise, while two looked at resistance training. Only exercise interventions lasting at least eight weeks were included (which is reasonable, since I would think that is a bare minimum to see any meaningful difference in body composition). The interventions mostly consisted of three exercise sessions per week, with each session lasting roughly one hour, and the average length of the intervention periods was 18 weeks. The aerobic exercise interventions were of moderate intensity and generally took the form of walking or cycling. The resistance training exercises consisted of two to three sets, with 10 to 20 repetitions per set. Participants had an average starting weight of 82 kg (180 pounds).
Results? Participants in the exercise group weighed 0,9 kg (2 pounds) less at the end of the intervention periods, while participants in the control group weighed 0,8 kg (1,8 pounds) more. This was a small 1,7 kg difference in favour of exercise, but it wasn’t statistically significant.
I draw two conclusions from these results. First of all, these people were only marginally overweight to begin with. So, apart from the fact that people had to have type two diabetes to be included in these studies, which limits the ability to generalize to a broader population, they were also not that overweight, and certainly weren’t obese. This means that any conclusions drawn from this study don’t necessarily apply to people who are more seriously overweight. It is possible that more overweight people see more weight loss with exercise than less overweight people. Secondly, there was a trend toward weight loss with exercise even though it wasn’t statistically significant. Since weight gain and weight loss are generally slow processes, it’s possible that people who exercise more do see a big difference over the long term, say a decade or two, even though there isn’t really any noticeable difference at the 18 week mark.
Next we’ll look at a systematic review that was published in the International Journal of Obesity in 2005. The review included 6 randomized controlled trials, with a total of 514 participants, and the intervention periods ranged from 10 weeks to 52 weeks. One thing that was interesting about this review was that it only included studies that did a follow up after at least one year, to see if any weight loss that was achieved during the interventions was maintained over the slightly longer term.
The average weight of the participants at the start of the trials was 97 kg (214 pounds), which I think makes this study more important than the previous one. An average height adult that weighs 97 kg is obese (or a moderately successful body builder).
This study wasn’t comparing exercise with nothing, but rather exercise and diet modification with diet modification alone, which is just something to be aware of before we get in to the results.
At the end of the intervention periods, people in the “exercise + diet” group had lost 13 kg (29 pounds) while people in the “diet without exercise” group had lost 10 kg (22 pounds). At the end of the full year (when many people had presumably returned to old habits), the difference was 6,7 kg (15 pounds) vs 4,5 kg (10 pounds). As in the previous review, none of these differences was statistically significant.
What can we conclude? Although the difference between the groups is not statistically significant, there is a small difference, and as with the previous review, it’s possible that a longer term intervention period, i.e. one where people were exercising for years rather than months, would show a statistically significant difference in weight. One interesting thing to note though, which is made clear by this study, is that the difference caused by dietary changes is much bigger than any difference caused by exercise. 77% of the weight lost was due to diet modification, while exercise contributed a paltry 23% .
We’re going to finish up with a Cochrane review that was published in 2006. It included a massive 41 randomized controlled trials, with a total of 3,476 participants. This sounds really good at first, until you realize that the 41 studies were all so different that these 3,476 people had to be analyzed in a large number of separate “buckets”, limiting the statistical power of the analyses.
The trials varied in length from 3 to 12 months. The interventions varied, with walking being most common (21 trials). Other interventions included bicycling, jogging, weight lifting, and group aerobics. Exercise frequency ranged from three to five sessions a week, and duration ranged from 10 minutes to 90 minutes, with an average of 40 to 50 minutes. Some of the trials compared exercise with no intervention, some compared exercise with diet modification, and a couple compared exercise + diet modification with just diet modification. All the trials required participants to have a BMI (Body Mass Index) of 25 or over (the definition of being overweight). Unfortunately the review doesn’t provide any more detail than that on how much the participants actually weighed.
Among the two studies (270 participants) that compared exercise with doing nothing, participants in the exercise group lost 2 kg (4,4 pounds) more than participants in the “do nothing” group. This result was statistically significant.
Among the 15 studies that compared exercise + diet modification with just diet modification (1,079 participants) the exercise + diet group lost an average of 0,65 kg (1,4 pounds) more than the group that just changed their diet. Again, the result was statistically significant. Whether it was relevant to the real world is more questionable, and I guess depends on whether the additional weight loss would have continued over the longer term with continued exercise.
So, what conclusions can we draw from all this? In the above reviews, weight loss varied from 0,65 to 3 kg (1,4 to 6,6 pounds) over the course of the interventions when the exercise groups were compared with the control groups. Considering that people who are overweight generally weigh over 80 kg (176 pounds), that is a pretty marginal difference, which isn’t going to be visibly noticeable. Additionally, that difference wasn’t statistically significant in most of the studied groups, and becomes even more questionable when you consider that very few of these trials were blinded, so people generally knew if they were in the intervention arm or the control arm, and therefore whether they were supposed to lose weight or not. This could of course result in a bigger apparent difference than would be seen in real life, since knowing that you are in the intervention group often makes people more motivated to prove that there will be an effect.
Admittedly, the intervention periods weren’t always that long, and a 2 kg difference over six months might mean a 20 kg difference over six years. If that is the case, then exercise does become an important factor over the long term.
One thing that is clear from the above reviews, however, is that changes in diet have a much bigger effect on weight than changes in exercise patterns, generally a three or four times bigger difference. My take-home from this is that people who want to lose weight should focus the main part of their energy on modifying their diet.
You might also want to look in to whether cutting down on salt is good for your health and whether you should be taking vitamin D supplements to decrease your risk of depression.