The "Lose Weight, Gain Speed" Myth

WARNING: You might not want to read this post. I have two disclaimers before you go on.

1. This post discusses WEIGHT. I know that the gravitational pull of your body towards the Earth is troubling for some people. It’s discussed here, so if the mere mention of the word incites anxiety for you, stop reading.

2. Possibly even more disturbing, this post involves MATH.

Moving along…

A few years ago, Runner’s World published an article about how losing weight will make you run faster. They even included a cute little table showing you how much time you could drop across various race distances by losing weight.

Lose Weight, Gain Speed

This table, based on changes in maximal aerobic capacity, provides a rough estimate of how much your race times will improve if you lose weight, as long as you have it to lose. If your BMI drops below 18.5, you're at risk of becoming weaker and slower.

WEIGHT LOST	5K	10K	HALF-MARATHON	MARATHON
2 lbs	12.4 secs	25 secs	52 secs	1:45
5 lbs	31 secs	1:02	2:11	4:22
10 lbs	1:02	2:04	4:22	8:44
20 lbs	2:04	4:08	8:44	17:28

(source) 

The table has a small disclaimer that if your body mass index (BMI) falls below 18.5, you might become slower. But otherwise, obviously the relationship between time improvement and weight lost is pure and linear, right?

The former research scientist in me was skeptical. Luckily, the academic article that was the basis for the piece in Runner’s World was readily available, and I was able to quickly find the improper conclusions that the Runner’s World author had made when reporting on the piece by Vanderburgh and Laubach of the University of Dayton Department of Health and Sport Science. Here are a couple of major errors made in the Runner’s World article:

• The Vanderburg and Laubach piece only analyzed the impact of weight on 5k times, but Runner's World extrapolated the results to 10ks, half marathons, and marathons. Since absolutely no analysis was done for any other race distance, it is impossible to know if the same relationship between weight and race time applies for all distances. Perhaps the impact is more pronounced for longer races. Perhaps less pronounced. Either way, when only one distance is analyzed, it is not correct to assume constant covariance of weight and speed for all race distances.
• Specifying that weight loss down to a BMI of 18.5 will lead to faster races, but weight loss beyond that will not, implying that this is a hard cutoff value. Aside from the obvious fallacy here – which is that a variable based on weight and height alone, such as BMI, varies substantially among healthy individuals because of differences in bone density, muscle mass, and glycogen storage capacity – this statement also doesn’t follow from the Vanderburg and Laubach work. They instead used a lower limit weight, not a lower limit BMI, based on the weights of national-class 5000m runners. (This approach is flawed, too. More on that below, but the point here is that the Runner’s World article didn’t even properly report the Vanderburg and Laubach study.)
• Reporting that the table is “based on changes in maximal aerobic capacity" when it is not. Sorry, Runner’s World. What you’ve got here is a table of time improvement vs. weight loss for race distance. None of those variables – distance, time, or weight – are a measure of aerobic capacity. Try again next time.
• Making generalizations and absolute statements that aren’t supported by the research.
• “There’s no denying that healthy runners will race about two seconds per mile faster for every pound they lose.”  To say “will” implies that no other variables make a difference.  How about if the weight loss is from dehydration? That’s not going to work out so well for you on marathon day. What if the weight loss is from reduced muscle mass after a layoff in training, which is very unlikely to make you faster? The research out of the University of Dayton never even implied that the relationship between speed and weight is absolute, or that weight is the main predictor of race time.
• “Because losing a few pounds makes running easier, you should be able to increase your workout distance and speed.” Losing a few pounds does not always make running easier. Body weight comes from many sources, including muscles, bones, organs, water, and fat. All of these can help you train harder – bone mass can help prevent stress fractures, fat can help you through longer workouts, and muscles are what obviously move your bones. A person doesn’t necessarily drop nonessential and unhelpful mass from their body when they lose weight – muscle, water, and bone mass can all fall away – so losing weight doesn’t necessarily make running easier. You could be losing weight that was helping you run.

As if the misreporting by Runner’s World wasn’t enough, it turns out that the Vanderburgh and Laubach research itself was deeply flawed, which was disappointing given that it was done at a university department of exercise science.

• The hypothesis that weight loss leads to faster times is based on research which shows that VO₂ Max, which is a good predictor of performance in a 5000m run, increases with reduced body fat and higher physical activity levels. While lower body fat and more physical activity will usually lead to weight loss, my suspicion is that those two factors are far more important than absolute weight when it comes to turning in a good 5000m time. Vanderburgh and Laubach would have had a much better study on their hands had they used body fat and/or training hours as the independent variables predicting 5000m run times.
• The study used body weight as the independent variable for predicting race times, and did not consider height at all. The authors did mention this, and noted that they neglected height because weight and height covariate – that is, taller people are, as a group, heavier than shorter people. While multicollinearity (inclusion of two variables that report overlapping information) is a problem in linear regression, a technique commonly used to evaluate the relationship between two variables, this can easily be solved by replacing the two variables that are closely tied (such as weight and height) with one variable that accounts for both (such as BMI). This would have been a simple substitution that would have made the conclusions of the study more valid, but the authors failed to do so.
• The study identified the ideal weight as that of the top 5000m runners in the U.S. In other words, selecting on the dependent variable, OR assuming that the weight of whoever is currently best is the best weight to be, as if no other variables carry any other explanation. A more appropriate approach would be to evaluate weight, or even better, BMI or body fat, against the times of runners of varying abilities to see what range of weights produced the best times.
• The study is based on an adjustment factor used powerlifting competition rather than any real data or actual running times or athlete weights. This sounds ridiculous, but it is true. Go look at page 4 of the article. In summary, their conclusions are based on a formula for equalizing how much weight people of differing body weights can lift. It has nothing to do with running. All they do is take this factor and use it to make a linear equation with one data point – the weight and time of the best 5000m runner in the country.

This isn’t research. This is letting a well-trained monkey loose with some numbers, an equation or two, and contact information for exercise science journal editors.

In other words, if you’ve decided that you’ll drop 10 minutes from your half marathon time by dropping 12 pounds and changing nothing else – it might be time to rethink that. Instead, train hard and eat to support that, both in terms of quality (fruits and vegetables are good, so are lean proteins) and in terms of timing (eat some high quality carbohydrates before you train, get in some more along with a bit of protein after you train so that you replenish your glycogen stores and rebuild muscle). Your body will adapt to support the harder training, and in turn, the faster racing. That adaptation might involve weight loss, it might not. Either way, you’ll run faster because you trained to run faster.