Measures of effect - absolute and relative risk reduction

Definitions

Absolute risk reduction is the arithmetic difference in outcome risk between a control and treatment group; relative risk reduction is the proportional reduction relative to the control group's baseline risk - the number needed to treat (1 divided by absolute risk reduction) is derived from absolute, not relative, risk reduction, making this distinction practically essential for clinical decision-making
Central to interpreting clinical trial results and translating them into individualised treatment recommendations, given how differently absolute and relative measures can portray the same underlying data

Because number needed to treat is calculated from absolute risk reduction, and absolute risk reduction depends on baseline risk (unlike relative risk reduction, which tends to be more consistent across different baseline risk levels for a given treatment), the number needed to treat for the same treatment varies substantially depending on the baseline risk of the population/individual being considered - a patient at higher baseline risk generally has a lower (better) number needed to treat than a patient at lower baseline risk, for the same relative risk reduction

Relative risk reduction figures reported from a trial should be combined with the specific patient's own baseline absolute risk (which may differ from the trial population's average) to estimate a personalised absolute risk reduction and number needed to treat
Trial-reported number needed to treat applies specifically to the trial population's baseline risk, and should not be assumed to directly transfer to an individual patient with a different baseline risk level

Individualised number needed to treat estimation using patient-specific baseline risk (rather than simply applying the trial-reported NNT universally) provides more clinically meaningful, personalised treatment benefit estimates
Relative risk reduction remains a useful measure of a treatment's biological/pharmacological efficacy, reasonably consistent across different baseline risk populations, while absolute measures translate this into individualised clinical relevance

Number needed to treat's dependence on baseline risk (varying across patients despite constant relative risk reduction), relative risk reduction's relative consistency across baseline risk levels, individualised risk estimation application
Number needed to treat is not a fixed property of a treatment - it varies with the baseline risk of the population or individual patient being considered, since it derives from absolute (not relative) risk reduction - a treatment with the same relative risk reduction produces a better (lower) number needed to treat in higher-baseline-risk patients than in lower-risk patients, which is why individualising NNT to a specific patient's own risk level gives a more clinically meaningful estimate than applying a trial's reported NNT universally

4 more sections, plus exam facts

Premium unlocks every note across every specialty, and the full exam fact library behind it.