Evidence-based dermatology: number needed to treat and its relation to other risk measures

J Am Acad Dermatol. 2007 Apr;56(4):664-71. doi: 10.1016/j.jaad.2006.08.024. Epub 2006 Oct 27.


When discussing treatment options with patients, clinicians often use terms such as "frequently" or "rarely" when referring to potential benefits or possible harms. Quantitative measurements of treatment benefits and harms derived from randomized clinical trials or meta-analysis such as odds ratios or risk reduction are more precise terms, yet physicians and their patients find them difficult to understand and they are not, therefore, commonly used in clinical practice. To overcome the lack of intuitiveness for traditional measures of risk estimates derived from clinical trials, the number needed to treat (NNT) has been widely recommended as a measure of effectiveness, and number needed to harm as a way of describing risk of possible adverse events. NNT is simply the number of patients who, on average, would need to be treated with a proposed intervention to demonstrate one additional gain over the standard comparator intervention. NNT is an absolute measure and it is calculated as the inverse of the absolute risk reduction. In this article we describe the usefulness and limits of the NNT with particular reference to dermatology, and compare NNT with other relative measures such as the relative risk and relative risk reduction.

Publication types

  • Review

MeSH terms

  • Confidence Intervals
  • Dermatology / standards
  • Dermatology / trends
  • Evidence-Based Medicine*
  • Female
  • Humans
  • Male
  • Meta-Analysis as Topic
  • Models, Statistical
  • Odds Ratio
  • Patient Selection
  • Randomized Controlled Trials as Topic
  • Sample Size
  • Sensitivity and Specificity
  • Skin Diseases / diagnosis*
  • Skin Diseases / therapy*