A headline tells you a study happened. A brief tells you whether it should change anything. Every entry here carries a strength rating, the population it was studied in, and the parts that are still open.
Rated on a public scale · Uncertainty stated, not buried
Endurvity rates the weight of evidence behind a claim on six levels. The rating describes the state of the research, not how enthusiastic anyone is about the finding. The bottom three levels are kept apart deliberately: studies disagreeing, evidence being too thin to judge, and nobody having run the study are three different situations, and a reader deserves to know which one applies.
Consistent enough to act on. New studies are unlikely to reverse the direction of the finding.
Several randomized trials, or systematic reviews of them, replicated in people across different research groups and populations.
How we write it"reduces", "improves", "increases" — stated with the size of the effect.
Reasonable to act on if the action is low-risk. The effect is real; how large it is, in whom, and for how long are still unsettled.
A smaller body of trials, or larger trials with limitations in size, duration, population, or a surrogate standing in for the outcome that matters.
How we write it"appears to", "is associated with".
Interesting, not yet a reason to change your routine. Worth watching rather than adopting.
Thin evidence: small, short, single-centre trials, or studies reporting only a surrogate marker.
How we write it"early data suggest", "one small trial found".
Good studies disagree, and the disagreement has not been resolved. Someone quoting only one side of it is not describing the evidence.
Well-conducted studies pointing in different directions, where the difference is not explained by dose, population, or how the outcome was measured.
How we write it"trials disagree" — followed by who disagrees and why.
Not enough evidence to say either way. A claim at this level should not be driving a decision.
Very few studies, very small ones, or research that never directly asked the question being answered.
How we write it"there is not enough evidence to say".
Nobody has properly looked. The absence of a finding here is an absence of research, not a finding of no effect.
No controlled human research on the question, or only animal, laboratory, or mechanistic work standing in for it.
How we write it"this has not been studied in people".
Each brief answers one question. The disclosures on the card are the same ones you get on the page, so you can decide whether to read further before you click.
Most misleading health coverage fails on a handful of predictable points: a correlation reported as a cause, a relative risk quoted without the absolute one, a mouse study described as a breakthrough, or a surrogate marker standing in for an outcome nobody measured.