Beetroot juice is one of sport's favourite legal boosts. Pooling 45 trials and 659 athletes, a new review could not find a gain on any of the six things it measured

The idea behind a shot of beetroot juice before a race is neat. Bacteria in the mouth turn the vegetable's nitrate into nitrite, the body turns some of that into nitric oxide, and muscles may get more blood and use oxygen more economically. In recreational exercisers, earlier reviews have found small gains. Whether the same holds for people who already train seriously has been argued over for years. A systematic review published in the journal Sports on 20 October tried to pull the athlete evidence into one place.

The team, from Loughborough University, Shanghai University of Sport and Edith Cowan University, gathered 45 randomised, placebo-controlled trials with 659 athletes, from cyclists, runners and kayakers to fencers, hockey players and CrossFit competitors. Most trials gave a single dose, typically 2 to 3 hours before testing; 16 loaded for 3 to 8 days. The usual dose, about 6.2 to 8.4 mmol of nitrate, equals 70 to 140 mL of concentrated juice. Placebos included beetroot juice with the nitrate taken out, blackcurrant juice and matched capsules.

Across six measures (peak power, mean power, maximal oxygen uptake, heart rate, blood lactate and perceived effort), the pooled result was small and statistically indistinguishable from placebo every time. Peak power came closest. Trials in men alone did reach significance, but the difference between those trials and mixed-sex ones did not, so the authors call it a lead to test, not a finding. And when a standard statistical correction for missing small studies was applied, the overall peak power estimate fell from slightly positive to about zero. Beetroot juice and nitrate salts could be compared only on oxygen uptake, and neither beat placebo.

That is not the same as 'beetroot juice does not work', and the paper says so. Its confidence intervals still leave room for effects that could matter, and the review's method for handling crossover trials, 41 of the 45, widens those intervals; the authors accept this may partly explain the flat results. Finishing times were not among the six pooled measures, even though time trials were among the tests used. The pooled data also could not separate responders from non-responders, or test baseline fitness directly. The trials themselves were thin in places: only 9 described how they randomised, only 5 were registered in advance, and just one was rated at low risk of bias overall. Women were barely represented, 88 of the 659. The certainty of the evidence ranged from moderate for oxygen uptake and heart rate to very low for lactate and perceived effort. The authors want the next studies to look for who responds and in which situations, such as altitude or repeated sprints.