Although the placebo effect is a well-known psychological phenomena, it is sometimes overlooked in exercise studies. We focus on how much a supplement, technology, or treatment enhances performance as compared to a placebo, although placebo effects may account for a large portion of “real-world” performance gains.


When a new supplement study comes out, one of the first things knowledgeable readers look for is whether the trial was placebo-controlled. Placebo controls are necessary since part of the effect of any treatment is the effect of merely receiving a treatment, even if it doesn’t accomplish anything. If you can complete 10 reps with a particular weight without taking a pill, 12 reps with a sugar tablet, and 13 reps with a caffeine pill, then simply taking a pill adds 2 reps to your total, with caffeine only adding 1 rep to your total. We are mostly concerned with how much a supplement, technology, or treatment increases performance above a placebo boost.

We usually pay less attention to the placebo’s own increase.

A new meta-analysis, on the other hand, covers that gap by examining the research that has looked into the extent of the placebo impact on exercise performance.

Overall, placebo and nocebo effects (the placebo effect’s evil twin - decreasing performance when given a therapy you expect to damage performance) appear to have a small but significant effect on physical performance. Furthermore, the placebo effect is amplified when people believe they’re taking a restricted substance or are duped into believing the placebo has already given them a performance boost.


FINDINGS


After screening more than 4,000 papers, 32 were found to match the inclusion criteria, representing 1,513 total participants. Twenty of the 32 research looked into nutritional placebos or nocebos, while the other 12 looked at mechanical placebos. Only five research looked at the nocebo effect, while the majority looked into the placebo effect.


Four of the trials that employed placebos used overt placebos (i.e., they told the subjects they were getting a placebo), while five used enhanced feedback to precondition the placebos (which I’ll describe in the next section).


Ergogenic aids boosted performance considerably. Both placebos had small effect sizes (d = 0.35 for nutritional placebos and d = 0.47 for mechanical placebos). Surprisingly, placebos purporting to be illegal medications had stronger placebo effects (d = 1.44 for steroids and d = 0.81 for EPO). Preconditioning methods, interestingly, had a considerable effect on performance (d = 0.82). The effect size of sham transcutaneous electrical nerve stimulation (TENS) was similarly observed to be large (d = 0.86).


Caffeine and placebo amino acids had small effect sizes (d = 0.36 and 0.40, respectively). The effect of a fully fake athletic supplement was determined to be negligible (d = 0.21). The effects of cold water immersion, sodium bicarbonate, ischemia preconditioning, carbohydrate, beta alanine, kinesio tape, and magnetic wristbands were all determined to be minor or nonexistent. The use of overt placebos had no effect.



The nocebo effect (d = 0.37) was equally minor. The researchers did not analyze if the sort of nocebo utilized influenced the findings because there were only five studies that looked into the nocebo impact.

INTERPRETATION


The first point I’d want to highlight is that the authors don’t explain how they conducted their meta-analysis, so it’s difficult to know if they made a mistake. I believe they just pulled the effect size from each study and averaged or weighted the results. If that’s what they did, their point estimates for each effect size may be fine (based on a cursory glance at the reported effect sizes, there doesn’t appear to be a lot of variation within each sub-category), but they may have more or less volatility around the mean estimate than they “should.”


There are a few other statistical issues I could raise, but I already do enough of that, and the issues I’d like to raise are unlikely to have a significant impact on my interpretation of this meta-analysis. However, we should definitely be a little more skeptical of these conclusions than we would be if their statistical methodology was more open.


The more “serious” placebos caused higher placebo effects, which was one of the most reasonable and interesting findings of this meta-analysis.


Because the placebo effect is a psychological phenomenon based on expectation (you believe what you’re taking will enhance your performance, and that perception is what really improves performance), it stands to reason that placebos that you expect to have a larger effect would have a larger effect. I believe that almost everyone would anticipate to gain strength when taking steroids or improve their endurance while taking EPO; after all, those chemicals are banned in most competitions because they do induce significant improvements in performance.


And, despite the fact that just three of the studies in this meta-analysis looked at illegal substances, all three of them showed large impacts, with the mean effect for those studies (d = 1.23) being more than three times larger than the mean effect for the entire meta-analysis (d = 0.37).


I’ve always wondered how much of the boost in strength people perceive when they first start using steroids is due to anticipation. People frequently say that they gain a significant amount of muscle and have a disproportionate improvement in strength.


However, in placebo-controlled research (i.e., when people who aren’t taking steroids are still given injections to make them think they’re taking steroids), it appears that the boost steroids provide for strength gains is disproportionately smaller than the boost steroids provide for hypertrophy (6), with steroids helping subjects build 3.3 times more muscle (4.23 percent vs. 13.92 percent increases in combined triceps and quadriceps cross-sectional area), while placebo-controlled research found that steroids helped (35 vs. 60 kg combined increase in squat and bench press 1RM).


Since this meta-analysis discovered that the placebo effect generates a large improvement in performance when people believe they’re taking banned substances, it leads me to believe that reports of large, rapid strength increases when people start taking steroids may be based as much on expectancy as the physiological effects of the drugs.


Another intriguing point to consider is that the placebo effect of caffeine may be just as great, if not larger, than the “real” effect of caffeine. Caffeine enhances performance when compared to placebo, according to meta-analyses, with effect values ranging from 0.2 to 0.4.


The placebo effect for caffeine was found to have an effect size of roughly 0.4 in this meta-analysis.


Thus, in the “real world,” when people take caffeine before a workout, the total effect may be quite large (i.e., in the d = 0.6-0.8 range), with about half to two-thirds of the effect attributable to expectancy and one-third to one-half of the effect attributable to caffeine’s physiological effects.


Overt placebos had no effect on performance, as expected. An overt placebo is one that the recipient is aware of. It’s an overt placebo if you hand someone a pill and say, “This is a placebo pill that does nothing” (sometimes called an open label placebo). It may seem absurd that the effects of overt placebos should be studied, yet they have been demonstrated to improve subjective outcomes (such as pain or nausea) when compared to a no-treatment control, which I find fascinating. When objectively testing exercise performance, however, people do not experience a placebo effect since they are aware that they are taking a placebo.


Finally, I’d want to point out that the mean nocebo effect was nearly identical to the mean placebo effect.


When people believe something will improve their performance, it almost always does, and when they believe something will impair their performance, it almost always does.