Absolute risk reduction from two arm risks

Subtracts the intervention-arm risk from the comparator-arm risk of an adverse event. A negative value is an absolute risk increase. The Cochrane Handbook computes the risk difference as intervention minus comparator, so for an adverse outcome ARR equals minus that risk difference.

Signature

ARR = p_C - p_I
Inputs
InputsDefinitionUnit
p_CRisk of the adverse event in the comparator group over the follow-up periodprobability from 0 to 1
p_IRisk of the adverse event in the intervention group over the same periodprobability from 0 to 1
Output
ARRDifference between the comparator and intervention risks, positive when the intervention lowers riskprobability difference from minus 1 to 1, over the stated period

Function

Absolute risk reduction function

Maps the risks of an adverse event with the comparator and with the intervention, over the same period, to their difference, oriented so that a benefit is positive. The difference converts a relative treatment effect into events avoided per person treated, the quantity that costs and QALYs in an economic evaluation are built from.

Try this function

Implementations

  • Excel

    Absolute risk reduction from event counts

    With comparator events and totals in ComparatorEvents and ComparatorN, and intervention events and totals in InterventionEvents and InterventionN, Excel returns the ARR.

    =ComparatorEvents/ComparatorN-InterventionEvents/InterventionN

Assumptions

  • Same outcome, population and period in both arms

    Both risks refer to the same event definition, the same randomised population and the same follow-up period. An ARR is reported with that period, because events accrue and the ARR changes as follow-up lengthens.

  • Adverse outcome orientation

    The event is one the intervention aims to prevent. For a beneficial outcome, such as response, the subtraction is reversed so that a benefit remains positive.

Worked examples

  • OSIRIS trial reported in CONSORT 2010

    Death or oxygen dependence occurred in 514 of 1,346 infants with delayed selective surfactant and 429 of 1,344 with early administration, risks of about 0.38187 and 0.31920. The ARR of 0.06267 matches the risk difference of minus 6.3 per cent reported in the CONSORT 2010 explanation for item 17b.

    p_C = 0.38187; p_I = 0.31920; ARR = 0.06267

Common errors

  • Pooling events across trials before subtracting

    Adding events and participants across trials and then computing one ARR ignores randomisation within studies and can mislead when allocation ratios differ. The Cochrane Handbook advises using the pooled result of a meta-analysis instead.

Sources

  • Absolute and relative effects for binary outcomes in CONSORT 2010

    Moher D, Hopewell S, Schulz KF, Montori V, Gøtzsche PC, Devereaux PJ, Elbourne D, Egger M, Altman DG. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c869. Item 17b (for binary outcomes, presentation of both absolute and relative effect sizes is recommended), its example and table 7 from the OSIRIS trial, and the explanation that the risk difference is less generalisable than the relative risk because it depends on baseline risk.

    View source →

  • Absolute and relative effect sizes in CONSORT 2025

    Hopewell S, Chan AW, Collins GS, Hróbjartsson A, Moher D, Schulz KF, et al. CONSORT 2025 statement: updated guideline for reporting randomised trials. BMJ. 2025;389:e081123. Checklist item 26: for binary outcomes, presentation of both absolute and relative effect size.

    View source →

  • Risk difference and absolute risk reduction in the Cochrane Handbook

    Schünemann HJ, Vist GE, Higgins JPT, Santesso N, Deeks JJ, Glasziou P, Akl EA, Guyatt GH. Chapter 15: Interpreting results and drawing conclusions (last updated August 2023). In: Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (editors). Cochrane Handbook for Systematic Reviews of Interventions version 6.5. Cochrane; 2024. Section 15.4.1 (the risk difference is often referred to as the absolute risk reduction or absolute risk increase) and section 15.4.4 (not computing risk differences from aggregated totals across trials).

    View source →

Canonical Identity

Stable URI · Machine-readable · Resolvable · CC BY 4.0