The Evidence Brief

The Placebo Problem: Why Supplement Research Needs a Comparison Group

Placebo response in supplement trials is far bigger than most people assume, which is why a comparison group matters so much.

Supplement trials are increasingly built around a placebo-controlled design, and for good reason. Compare outcomes only before and after a product is taken (a single-arm design) and there is no way to separate an effect caused by the product from an effect caused by something else entirely. That something else has a name - the placebo effect. And in supplement research, it turns out to be a considerably bigger factor than many people assume.

What is the placebo effect?

The placebo effect is a real, measurable change that follows an inactive treatment, driven by expectation and context rather than any action by the product or drug tested. Participants are not imagining things or misreporting - their outcomes genuinely change even though nothing biologically active has been given.

This happens for a few well-documented reasons:

  • Regression to the mean. People enrol at their worst, so some spontaneous improvement is expected regardless of treatment.
  • Expectation. Believing a treatment will help can genuinely change how a person feels and reports their symptoms.
  • The ritual of participation. Daily dosing, questionnaires, and check-ins can shape outcomes independent of what is being taken.

What the evidence shows

The scale is easy to underestimate until you see specific cases:

  • A recent rose-hip powder trial for osteoarthritis found over 50% improvement in both active and placebo groups. Only by digging into dose-response data, a pattern also seen in two earlier trials, did a genuine effect start to show through [1].
  • A review of herbal placebo arms in IBS trials found a 37% average response rate [2].
  • In sports nutrition, placebo response has accounted for around 25% of the effect for buffering agents and up to 59% for caffeine [3], for example, cyclists given a placebo they believed was caffeine still raised their power output by up to 2.2% [4].
  • Sleep supplements seem particularly susceptible. A meta-analysis of 32 trials found a mean placebo response of 61–64% across both objective and subjective sleep measures [5]. This played out directly in one multi-ingredient sleep nutraceutical trial, where PSQI sleep-quality scores improved by 32.3% in the active group and 39.5% in the placebo group, a difference that wasn't statistically significant, with the "unusually high" placebo response cited as the likely reason [6].

The case for the RCT

Single-arm trials still have a role, particularly for early signal-finding or confirming a product is safe and tolerable before a larger study. But they cannot separate a genuine effect from a placebo effect, and that gap is often too large to ignore. Only a randomised, placebo-controlled design keeps both groups equally exposed to belief, habit, and the ups and downs a condition would show on its own, which is why it remains the standard against which supplement claims are judged.

Get The Evidence Brief

Industry news filtered down to the stories that show how health brands use science to win, every other Wednesday.

Subscribe