Article Text

Download PDFPDF
Measures of association as used to address therapy, harm, and aetiology questions
  1. Stuart Carney, MB, ChB, MPH, MRCPsych1,
  2. Helen Doll, BSc, Dip App Stats, MSc2
  1. 1Department of Psychiatry, University of Oxford, Oxford, UK
  2. 2Department of Public Health, University of Oxford, Oxford, UK

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


    Measures of association describe the strength of the relationship between an exposure (or intervention) and an outcome in clinical studies (randomised controlled trials [RCTs], cohort studies, and case control studies). There are 2 types of measures: relative (relative risk, relative risk reduction, odds ratios) and absolute (absolute risk reduction, number needed to treat).

    Confidence intervals (CI) should be given for each measure of association to quantify their uncertainty and are usually reported as 95% CI (ie, the interval has a 95% chance of including the true, but unknown, population value). If the 95% CI overlaps the value of no effect (see below), the result is not statistically significant at the 5% level (p<0.05). The use of p values and CIs to measure statistical uncertainty will be considered in a future statistics note.

    The identification of a statistically significant association between an exposure and an outcome alone does not imply causation. Possible …

    View Full Text