Beta-alanine: The meta-analysis – 18 trials, still a marginal effect

I’ve always thought a meta-analysis on beta-alanine was a good idea. So many of the trials have been underpowered, that it felt like weren’t getting anywhere fast. I wanted to be the one to do it, but life circumstances have kept me from focusing on much other than my main career. So it was a pleasant surprise to see that someone else had the same idea.

With just 11 weeks left before my exams though, I’m going to keep this blog entry brief.

Hobson RM et al. Effects of beta-alanine supplementation on exercise performance: a meta-analysis. Amino Acids DOI 10.1007/s00726-011-1200-z. Published online January 24, 2012.

There have been more than a handful of beta-alanine trials over the past 5 years now. However, the major drawback to almost every single one of them is the small sample sizes in each one. No one study is able to demonstrate a clear advantage for the use of beta-alanine. Meta-analysis allows the pooling of the data from these studies to reduce the disadvantages of the small numbers of each study. There are also drawbacks to meta-analysis, which I will get into a little bit as well.

The authors of this meta-analysis included all trials that had a beta-alanine ONLY supplementation group compared with a placebo group. Studies that had only beta-alanine+ groups (e.g. +creatine) were not included, and studies without any exercise performance or capacity tests were also excluded (e.g. just body composition)

Since different studies measured different effects in different ways, the researchers converted the individual numbers into a construct called “effect size”.

Effect size is a way of standardizing an effect measurement. As I’ve covered before, an average is meaningless without some measurement of spread, or variance. A study that shows that subjects, on average, lifted 40 pounds more than when they started has different interpretations if we add the information of a range of 5-75 pounds vs 35-45 pounds.

So continuing in this example, if an average improvement of 40 pounds isn’t the same as another average improvement of 40 pounds, how do you make them comparable? Percent improvement is one way of getting at it, but you’re still left with different ranges of values. And that’s where effect size comes in.

There are many ways to calculate an effect size. The most common and simplest way is to divide the average effect measurement by the variation (usually, the standard deviation). So what you get is a unit of change per unit of variation. The advantage of this method is that you can now compare effects with different variations. The downside is that this number has no units. It’s just a number (usually less than 1), which makes interpreting an effect size back to something we can easily intuit harder.

Before we get into the meat of the results though, I think it’s worthwhile to spend some time examining the studies that ended up being included.

Eighteen randomized controlled trials were included into this meta-analysis, for a total of 360 subjects (174 who got beta-alanine and 186 who got a placebo). What’s of biggest note though, is that no measurements with respect to lifting weights was extracted from ANY of these studies. Most of these studies looked at cycling performance or running performance (either as time to exhaustion or time-trial). So, if you’re interested in using this study to help you make decisions about using beta-alanine to help you lift more or for longer, there’s nothing in this paper to help you because it just doesn’t generalize that far. Using this meta-analysis for gym performance is like using a cycling study to help you bench more.

So, onto the meat:

Overall, the meta-analysis reports that the effect-size of the placebo group was 0.108 and in the beta-alanine group, 0.374. Beta-alanine had a higher effect-size than would be expected if it didn’t actually have any benefits (P=0.002). However, this translates to about a 2.85% improvement in performance. Most of this was seen in exercise capacity (i.e. time to exhaustion type measurements) as opposed to performance variables (i.e. time to complete a specified distance).

[Edit 2012/02/20: Median dose of all of the trials was 5.12g per day with a range from 2.4 to 6.4g per day. No relationship was found between dose and effect size.]

When the investigators looked at the time of exercise, they failed to detect a significant difference between beta-alanine and placebo in studies where exercise duration was less than 60 seconds. The difference between beta-alanine and placebo became most pronounced in exercise bouts of 60-240 seconds, and although still “statistically different” at bouts higher than 240 seconds, was less than the 60-240 second bouts.

My main issue with using effect size is that the paper reports that the data were not normally distributed, which then raises the question of how valid the effect size measurement is if it depends on normality to calculate a standard deviation? This is important because it means that in non-normal distributions, the effect size could be larger or smaller than reported in this paper. It’s impossible to know since we’re not sure whether the “average” spread of responses was higher or lower than calculated with the normality assumption. Even in the 60-240 second subgroup, it’s clear that the results are skewed with clear outliers driving a lot of the increased effect size calculation. It’s also hard to tell how many subjects were in each subgroup (not reported) which means we may be in a situation where promising results are based on small numbers.

The bottom line(s):

I think there are three messages to take away from this study:

1) The use of beta-alanine to improve weight-lifting performance and therefore its use in pre-workout or ongoing supplementation for these purposes is still unsupported by scientific study.

2) Even if we were to draw the incorrect extrapolation to weight lifting, BA doesn’t seem to confer any effect for exercise less than 60 seconds. Whether or not rep ranges in the 12 and higher range are equivalent to exercise higher than 60 seconds (which I’ll entertain for…amusement purposes) is also debatable since weight lifting is cyclical in nature, with rest between sets and it’s difficult to determine how much of a role  “lower buffering capacity” has in determining technical or absolute failure, not to mention the fact that these studies mostly looked at single bouts of maximal exercise, not multiple sets.

3) The use of beta-alanine to improve cycling or running performance seems to be marginal at best. The arenas in which the effect size of 0.374 was measured (i.e. the 2.85% improvement) are not likely to be the ones in which a 2-3% improvement in time or distance actually matter. In a 2000m rowing race, we’re looking at an improvement of around 7 seconds (which would make a difference at the Olympic, or even the National level). In a 40km cycling time trial, we’re looking at a 1 minute improvement (which again, would make a difference only at the most elite of competitions). However, only one of these studies actually looked at elite athletes. Would such an effect size be retained at the elite level? We just don’t know.