Where the method comes from
We adapted GRADE, the approach Cochrane and the World Health Organization use to rate evidence. Randomized trials start at high and observational studies start at low. A score goes down for small numbers, a single group of people, habits asked about only once, a different kind of heat or cold, or a range that reaches “no effect”. It goes up when the effect is large, or when more sauna goes with a steadily bigger effect.
What the verdict means
- Well supported: High or moderate certainty. What we would stand behind today.
- Some support: Low certainty. A real lead, and could well change with better research.
- Early signal: Very low certainty. One small study, or a thin margin. A reason to keep looking.
- Not supported: Studies looked for an effect and did not find one.
- No evidence yet: An idea without human data. Nothing to score.
How big is big
- Rates between groups: large under 0.5, medium 0.5 to 0.74, small 0.75 to 0.89.
- Trial scores (standardized effect size): large 0.8 and up, medium 0.5 to 0.79, small 0.2 to 0.49.
- Body signals: large at 2-fold or more, medium 1.25- to 2-fold.
Who scores, and when
Scores are an editorial reading by KUUME, not a medical review. Every score shows its steps, links every study and carries the date it was last reviewed. When we add a study, the claims it feeds are scored again, and the change is dated.