Endgames Statistical Question

Primary and secondary outcome measures

BMJ 2010; 340 doi: http://dx.doi.org/10.1136/bmj.c1938 (Published 14 April 2010) Cite this as: BMJ 2010;340:c1938
  1. Philip Sedgwick, senior lecturer in medical statistics
  1. 1Centre for Medical and Healthcare Education, St George’s, University of London, London
  1. p.sedgwick{at}sgul.ac.uk

    A randomised placebo controlled trial assessed the efficacy, acceptability, and safety of an alkane vapocoolant spray in decreasing pain during intravenous cannulation.1 Adult patients in an emergency department were randomly assigned to vapocoolant spray or control. The primary outcome measure was pain with cannulation. Secondary outcome measures included discomfort with spray on administration, success rate of cannulation, and unexpected side effects of treatment.

    Which of the following statements, if any, are true?

    • a) The primary outcome is also known as the primary end point

    • b) The primary outcome represents the measure of greatest therapeutic benefit

    • c) Secondary outcome measures never include measures of therapeutic benefit

    • d) Differences between treatment groups in the primary outcome are believed to be to the result of treatment


    Answers a, b, and d are true; answer c is false.

    The primary outcome measure, also known as the primary end point (a is true), is the outcome measure that represents the greatest therapeutic benefit (b is true). Sometimes researchers will propose more than one primary end point if several outcome measures are of equal therapeutic importance. Secondary outcome measures, also known as secondary end points, may provide information on therapeutic effects of secondary importance, side effects, or tolerability (c is false).

    Random allocation promotes two groups with similar characteristics at entry to the trial. More generally, providing the sample size is large enough, randomisation will ensure that there are no systematic differences between groups caused by possible confounding variables. Confounding variables are those that may influence patient outcome. Any differences between treatment groups in the primary or secondary outcome measures are, therefore, most likely be to the result of differences in treatment and not owing to confounding variables (d is true).


    Cite this as: BMJ 2010;340:c1938


    • Competing interests: None declared.