Bias doesn’t always work against you (or why gym class might not be all it’s cracked up to be)
This blog entry’s study comes from my friend Brad Pilon, who said, and-I-quote, “I wanna get you back to dissecting papers on your blog,” which jolted me into realizing I haven’t in a long time.
One idea that Brad and I have been talking about (apart from trying to stick to manipulating factors that make LARGE contributions to progress) has been the notion that diet and exercise, though both important for “health” and “fitness”, may have little contributions towards their counterpart body components. Diet seems to have the greatest impact on body fat, while exercise (and more specifically, resistance exercise) seems to have a greater impact on lean body mass. While one CAN use exercise to assist in losing body fat, I think there is probably some truth in the saying, “You can’t out-train a bad diet,” and similarly, I would argue that at most levels, you probably can’t eat yourself more muscles, otherwise, there would an epidemic of Ah-nolds in America, not obese people.
This is something we probably see more anecdotally than is evidenced in studies. And since muscle accrual happens at such infinitesimally slow rates (other than in juvenile growth and chemically-assisted growth), many of these studies are quite difficult or flawed.
This study is no exception. It’s a difficult logistical experiment, and it has flaws. But not all flaws are fatal, and some flaws can even be used to an advantage if the data comes out “just-so”.
Deere K et al. High impact activity is related to lean but not fat mass: findings from a population-based study in adolescents. International Journal of Epidemiology, 41: 1124-1131, 2012.
Trying to get at the root of the factors that contribute to childhood obesity is like digging a hole to China. You can’t do it all at once (unless you’re already in China.) Most population-based interventions like re-instating mandatory gym class, aren’t necessarily based on good data to show that moving more = less fat kids. We know that stemming childhood obesity has to be a wholistic effort; no single intervention will do the trick. So figuring out where to get the most bang for a buck is pretty important.
This study aims to address the role of physical activity and its effects on body composition. One of the big limitations on these study designs is actually getting accurate measurements of physical activity. We know activity logs are subpar, so with the advent of accelerometers (similar to the one in iPhones), direct measurement of activity has become more feasible.
The basic gist of the methods are as follows: The researchers asked a large number of teenagers around 17 years old, to wear an accelerometer for 7 consecutive days during waking hours. They were allowed to take it off if it might get wet, or for contact sports. The subjects were asked to keep a log of when they wore the accelerometer, and a valid “sampling” was considered a minimum of 8 hours per day for 2 days. The accelerometer did not have an internal clock, so the number of accelerations per day were based on the logs.
[Limitation #1: Still dependent on logs for the denominator of activity. Limitation #2: Missing acceleration data for activities like swimming and contact sports.]
Body composition was determined by DXA scans.
Activity was classified as sedentary, light, moderate, and high impact by measuring the accelerations of a separate group of children while they sat, walked, walked fast, ran and jumped. The accelerometers only measured in the vertical axis, which represents skeletal loading. The numbers were then crunched to produce a measurement in counts per day in each of the categories of activity.
Statistically, the researchers used regression analysis to determine if a relationship exists between counts per day and lean body mass and fat mass. They adjusted for height and maternal social class at birth.
A total of 732 subjects were ultimately eligible for this study. Of these, only 295 were boys. Boys tended to have greater lean mass than girls, and girls tended to have more fat mass than boys (which is to be expected.)
In terms of activity, most activity recorded was light activity. High impact activity was only 2% relative to light activity.
There was no relationship between lean mass and light activity (sorry, you can’t walk your way big.)
In both boys and girls, there was a relationship between high impact activity and lean body mass. This did not seem to matter whether you were a boy or a girl, but girls, in general, had a somewhat stronger relationship between high impact activity and lean body mass than boys. This relationship persisted after adjusting for height, fat mass and the other activity types.
Activity level was negatively related to fat mass (ie. more activity was related to less fat mass) before taking anything else into account. This relationship however, was weaker for each higher category of activity (ie. the weakest link was between high impact activity and fat mass). There was evidence that this relationship was strongest in girls and non-existent in boys when they tested for interaction. After adjusting for height, lean mass and the other activity types, there was only a significant relationship between moderate activity and fat mass in girls. Neither low or high impact activity had a relationship with fat mass in either gender.
The authors give a very nice interpretation of their results. It’s not surprising that high impact activity is related to higher lean body mass. However, it is a little surprising that activity seems to have very little impact on fat mass. The authors do discuss the limitations of their study (smaller number of boys may have prevented the detection of a relationship between activity and fat mass in boys), but I don’t think it changes the interpretation all that much in the end.
When we think about some of the limitations of the study:
Limtation #1: Dependence on logs
While there is a phenomenon in some studies to log more than less, in this case, if subjects said they wore the accelerometer when they didn’t, it means their counts would have been higher than if they logged the truth. This would have only strengthened the relationships found in the study. This limitation only has teeth if a large proportion of subjects did not log when they wore the accelerometer when they actually did.
Limitation #2: No detection of contact sport activity
Again, the relationship between high-impact activity and lean body mass was already detectable with very low counts relative to the other activity categories. Adding the contact-sport counts would only have strengthened the detected relationship.
Limitation #3: Sampling frame, and causation.
Body composition is obviously a product of every minute you’re alive. Measuring 5-7 days worth of activity and then correlating it with something that has taken a lifetime to accrue (even if it’s just roughly 17 years), has its limitations. However, what’s important to take away here is that kids who spent ONLY 2% of their walking time running or jumping in those 5 days had a higher lean body mass than those who DIDN’T. Two percent. That is not a lot of difference. Did those five days cause the higher body lean body mass? Of course not. I don’t think anyone is trying to sell this as a causation study. But if it is indicative of a trend of behaviour, it’s worth looking into further.
The fact that this is a study on teens is not a study limitation, but speaks to generalizability. I can’t use this study as justification to alter MY behaviours as a male adult.
However, it does call into question whether the current recommendations of higher physical activity as a population-health strategy is where efforts should be focussed. Of course, it requires further study, but the question, I think, is now raised with evidence. It’s one more dig in the hole.
We generally think of effective interventions as ones that are positive-actioned. Exercise is a positive action. You actively perform exercise. Eating less is generally considered negative-actioned, because it involves NOT doing something. It’s also difficult to promote eating less because A) food lobbies like the dairy council, the wheat boards, Coca Cola and Pepsi do not want people buying less food (less food = less profit); and B) there is this underlying fear that promoting less food consumption will lead to disordered eating like anorexia. Hence, we’re left with simply ALTERING diet structures and clouding much of the picture by demonizing certain foods (like carbs) or food volumes (like the size of a soda), and adding diet complexity, which doesn’t move large rocks; it just shifts the existing ones around.