Placebo Effects In Supplement Trials: What Trial Design Controls For

The placebo effect in a supplement trial is the change recorded in the control group, and it is not one phenomenon. It bundles together genuine expectancy, regression to the mean, practice effects from repeated testing, and the gap between what people experience and what they report on a questionnaire.

That matters because the word gets used as though it named a single force of known size. The largest attempt to measure it compared placebo against no treatment at all, and found the pooled effect small.

What the control group's change is made of

Four things show up in a placebo arm. Only the first is what people usually mean.

  • Expectancy. The participant's belief about what the capsule is doing.
  • Regression to the mean. People enrol when they feel worst, and the next measurement sits closer to their average whatever is given.
  • Practice effects. On a cognitive or performance test, the second attempt beats the first because the task is familiar.
  • Reporting bias. Knowing a supplement is under study changes how people answer a subjective scale.

A placebo arm controls for all four at once. It also means a large improvement there is not evidence of a placebo effect specifically, because three of the four have nothing to do with belief.

How big is the pooled effect

The reference work is the Cochrane review by Hróbjartsson and Gøtzsche, updated in 2022 as CD003974.pub3. Its design is unusual: it included only randomised trials carrying both a placebo arm and a no-treatment arm, so placebo could be compared against doing nothing.

Two hundred and thirty-four trials were included, with outcome data available from 202 of them, across 60 clinical conditions. For binary outcomes, pooling 44 studies and 6,041 participants, the relative risk was 0.93 with a 95 per cent confidence interval of 0.88 to 0.99. For continuous outcomes, pooling 158 trials and 10,525 participants, the standardised mean difference was −0.23, with a confidence interval of −0.28 to −0.17.

The split by outcome type is the finding worth carrying away. Patient-reported outcomes gave a standardised mean difference of −0.26, observer-reported outcomes −0.13: roughly double the shift in self-report against what an observer records.

The authors' caveats are substantial. Only 16 trials, 8 per cent, were judged at low risk of bias. Heterogeneity was moderate and the funnel plot for continuous outcomes was asymmetrical, which makes the pooling questionable. Effects were larger in small trials and in trials that did not tell participants a placebo might be used. The searches ran to March 2008. Their conclusion was that they did not find placebo interventions to have important clinical effects in general.

A supplement trial where the arms did not separate

Solomon and colleagues, in JAMA in 2002, ran a six-week, randomised, double-blind, placebo-controlled, parallel-group trial at a single site. The participants were 230 community-dwelling adults aged 60 to 82 with Mini-Mental State Examination scores above 26, in generally good health. The intervention was a Ginkgo product at 40 mg three times daily, per the manufacturer's label, against matching lactose placebo; the modified intention-to-treat analysis covered 219 people.

There was no significant difference between arms on any outcome, in either analysis. What did happen is instructive: scores rose from the first testing session to the second in all groups, which the authors attributed to practice effects. On Logical Memory II, the placebo group went from 20.3 to 22.3.

A single-arm version of that study would have produced a rising line, and no way to know it meant nothing.

The blinding data are the other reason the trial is quoted: 71 per cent of participants on Ginkgo and 75 per cent of those on placebo guessed they were on the active product. The authors list their own limitations. The product's content was not analysed, adverse events were not systematically monitored, a higher dose or longer exposure might be needed, and the sample was healthy older adults.

Blinding a food is harder than blinding a capsule

A lactose capsule is easy to match. A strongly flavoured food is not, so the comparator becomes a design decision rather than a formality.

Smit, Gaffan and Rogers, in Psychopharmacology in 2004, ran two double-blind, placebo-controlled studies of 20 and 22 participants. One arm received 11.6 g of cocoa powder, another an isolated 19 mg caffeine and 250 mg theobromine, and the second study used white chocolate and water as comparators. White chocolate carries the fat, sugar and mouthfeel without the methylxanthines, which is as close to a matched control as the category allows. Both samples were small.

What to check in a methods section

Check Why it matters
Is there a control arm at all Without one, practice effects and regression to the mean are indistinguishable from the product
What the placebo was made of A mismatched comparator breaks blinding
Whether blinding was tested Solomon 2002 asked participants to guess; most trials do not
Patient-reported or observer-reported The Cochrane review found roughly double the effect on self-report
Sample size and duration Effects in the Cochrane review were larger in small trials

Those questions are the same ones behind reading a supplement label properly, and they travel: dose claims in this category often rest on single-arm data, which is why a dosage guide has to be explicit about what its figures are and are not based on. The Chaski guide library collects both.

Why this page sits on a chocolate company's website

Plainly: Chaski has not run a clinical trial, so there is no trial result of ours, and nothing here should be read as one. The reason to publish on trial design anyway is that the category is full of before-and-after charts, and the methods above are how a reader tells one kind of evidence from another.

What is published instead is composition and process. Chaski Cacao is a 600 g tube of 77 per cent dark chocolate buttons with a stated daily dose of four buttons: single-origin Peruvian and Ecuadorian cacao, low-temperature ground, coconut sugar, Lion's Mane as a fruiting body extract and Cordyceps sinensis as mycelium. No emulsifiers, seed oils, palm oil or added cane sugar. Every batch is tested for lead, cadmium and pesticides. One tube is £44.99; three are £98.97, or £32.99 per tube.

Frequently Asked Questions

How large is the placebo effect in trials generally?

The Cochrane review by Hróbjartsson and Gøtzsche, updated in 2022, pooled trials that contained both a placebo arm and a no-treatment arm, giving 234 included trials and outcome data from 202 of them across 60 conditions. For continuous outcomes it reported a standardised mean difference of −0.23 across 158 trials and 10,525 participants, and for binary outcomes a relative risk of 0.93 across 44 studies and 6,041 participants. Only 8 per cent of trials were at low risk of bias, heterogeneity was moderate, and the searches ran to March 2008.

Why do before-and-after results look so convincing?

Because several things push the second measurement upwards regardless of the product. Practice effects raise scores on repeated cognitive and performance tests. Regression to the mean pulls people who enrolled at their worst back towards their average. Reporting bias shifts answers on subjective scales. Solomon and colleagues, in JAMA in 2002, saw exactly this: scores rose from the first session to the second in every arm of a 230-person trial, including placebo, where one memory measure went from 20.3 to 22.3, while no arm differed significantly from any other.

Does a double-blind design guarantee that participants were blinded?

No, and few trials check. Solomon and colleagues did ask, and reported that 71 per cent of participants on the active product and 75 per cent of those on placebo guessed they were taking the active one, a difference that was not statistically significant. That is a reassuring result, but it is only known because the question was asked. The Cochrane review separately found larger apparent placebo effects in trials that did not tell participants a placebo might be used, which suggests expectations about the design itself feed into the numbers.

Has Chaski been tested in a clinical trial?

No. There is no Chaski trial, no clinical testing and no study of the product, and nothing published on this site should be read as a trial result. What is documented is composition and process: a 600 g tube of 77 per cent dark buttons, four buttons as the stated daily dose, single-origin Peruvian and Ecuadorian cacao that is low-temperature ground, coconut sugar, Lion's Mane as a fruiting body extract, Cordyceps sinensis as mycelium, no emulsifiers or seed oils or palm oil or added cane sugar, and batch testing for lead, cadmium and pesticides.

Sources

  1. Hróbjartsson A and Gøtzsche P C, Cochrane Database of Systematic Reviews, 2022
  2. Solomon P R and colleagues, JAMA, 2002
  3. Smit H J, Gaffan E A and Rogers P J, Psychopharmacology, 2004
Liquid error (sections/pagefly-section line 6): Could not find asset snippets/pf-6bdf23d6.liquid