Issue 12 · 6 October 2026

A review graded its own supplement evidence, and almost none of it survived

A short issue, three studies, on the difference between a result and a finding. The thing that separates them is almost never the size of the effect. It is whether anybody said in advance what would count.

3 studies 26 September to 2 October 2026 12 min listen Listen on Spotify

Issue 12: A review graded its own supplement evidence, and almost none of it survived

11:51 · 6 October 2026

1:07A review of a hundred and nineteen supplement trials graded almost all of its own comparisons as very low certainty 3:16A probiotic trial declared its main outcome in advance, and that outcome moved while everything else stayed put 6:18Compared against plain carbohydrate rather than against nothing, an amino acid supplement taken straight after exercise added no muscle 9:19Say what counts, before you look

In this issue

Systematic review and Bayesian network meta-analysis Day level publication date, verified against the PubMed record. PMID sits mid band among its neighbours.

A review of a hundred and nineteen supplement trials graded almost all of its own comparisons as very low certainty

A Bayesian network meta-analysis lined up dietary supplements against each other for lower body power and jump height, then graded every comparison it produced. Of 1,636 pairwise comparisons, 17 were rated moderate certainty and 1,564 were rated very low.

Journal of the International · Sun et al. · Guangzhou Sport University · 1 October 2026 · PMID 42821324

supplements research-methods meta-analysis strength

1,564of 1,636
pairwise comparisons rated very low certainty, against seventeen rated moderate
Randomised, double blind, placebo controlled trial Day level publication date, verified against the PubMed record. Open full text at PMC13628159.

A probiotic trial declared its main outcome in advance, and that outcome moved while everything else stayed put

Eighty seven adults with functional constipation took one bacterial strain or a dextrin placebo for eight weeks. The pre-specified primary outcome, stool form, improved against placebo. Bowel movement frequency and every hormone and inflammatory marker did not differ between the groups.

Food Science and Nutrition · Xia et al. · General Hospital of Ningxia Medical University · 1 October 2026 · PMID 42824822

gut-health microbiome supplements GI-distress research-methods

0.73points
higher stool form score than placebo at week eight on a seven point scale, p equals 0.004, with every other outcome undifferentiated
Randomised, single blind, controlled trial Day level publication date, verified against the PubMed record. Open full text at PMC13626774.

Compared against plain carbohydrate rather than against nothing, an amino acid supplement taken straight after exercise added no muscle

Forty nine patients recovering from gastrointestinal cancer surgery did the same exercise programme and took either a branched chain amino acid enriched jelly or an amino acid free carbohydrate jelly immediately afterwards. Thigh muscle area changed by the same amount in both groups.

Progress in Rehabilitation Medicine · Iwahashi et al. · Wakayama Rosai Hospital · 2 October 2026 · PMID 42819727

protein supplements research-methods recovery

0.812p value
for the first primary outcome, thigh muscle area change, between the amino acid jelly and the plain carbohydrate one

Say what counts, before you look

Opinion Our reading of what these papers add up to. Everything above this line came from the papers. This part did not.

Three studies, three different moments in the life of a claim, and the same lesson at each of them. The review found that an entire evidence base could not support the ranking everybody wanted from it. The probiotic trial named one outcome in advance, and the discipline of that naming is what lets a reader see clearly that one thing moved and nine did not. The amino acid trial promised its outcome and chose a comparison that could defeat it, and the comparison did.

None of that is about supplements, and it is worth saying so. It is about the order of operations. A commitment made before the data arrives costs almost nothing to make and is very expensive to break. A commitment made afterwards is not a commitment, it is a description. The review in this issue is honest enough to record that it registered its protocol late, and that single admission changes how its rankings should be read, which is exactly why admissions like it belong in abstracts rather than in appendices.

The practical version, for anyone reading a result rather than producing one, is a short list of questions. What did this trial say it would measure. How many things did it measure in the end. What was it measured against, and could that comparison have beaten it. And did anyone grade the evidence, or only the effect.

A study that answers those four well can report a small effect and still be worth your attention. A study that answers them badly can report a large one and tell you almost nothing. The difference is not in the result. It is in what was promised before anyone looked.