Evidence-appraisal glossary
Dose-response relationship
A dose-response relationship exists when the size or likelihood of an outcome changes in a consistent direction as the level, duration, or intensity of an exposure or treatment changes. Seeing more effect at higher doses, and less at lower doses, strengthens the case that the exposure genuinely causes the outcome.
Also called: Biological gradient, Dose-response curve, Exposure-response relationship.
Also called a biological gradient, this is one of the classic Bradford Hill considerations for judging whether an association is causal. The idea is simple: if a factor truly drives an outcome, people with more of it (higher dose, longer exposure, greater intensity) should tend to show more of the effect.
How to use it when reading a study. Look past the single headline comparison. Does the paper report results across ordered categories (for example, low, medium, high exposure, or drug-dose tiers)? A steady trend across those levels, often summarized as a "test for trend," is stronger evidence than one lumped exposed-versus-unexposed contrast.
Stay skeptical, though. A gradient can also arise from confounding that itself scales with dose, so it supports causation but does not prove it. Also, real curves are not always straight lines: thresholds, plateaus, and U-shaped or J-shaped patterns are common, so a missing gradient does not automatically rule out a real effect.
Read the full Reading the Evidence blog.
This is a plain-language methodology definition for reading research. It is general education, not medical advice.