Medical doctors' perception of the "number needed to treat" (NNT). A survey of doctors' recommendations for two therapies with different NNT

Scand J Prim Health Care. 2003 Sep;21(3):162-6. doi: 10.1080/02813430310001158.


Objective: While the number needed to treat (NNT) is in widespread use, empirical evidence that doctors or patients interpret the NNT adequately is sparse. The aim of our study was to explore the influence of the NNT on medical doctors' recommendation for or against a life-long preventive drug therapy.

Design: Cross-sectional study with randomisation to different scenarios.

Setting: Postal questionnaire presenting a clinical scenario about a hypothetical drug as a strategy towards preventing premature death among healthy people with a known risk factor. Benefit after 5 years of treatment was presented in terms of NNT, which was set at 50 for half of the respondents and 200 for the other half.

Subjects: Representative sample (n = 1616) of Norwegian medical doctors.

Main outcome measures: Proportion of doctors that would prescribe the drug. Reasons for recommending against the therapy.

Results: With NNT set at 50, 71.6% (99% CI 66.8-76.4) of the doctors would prescribe the drug, while the proportion was 52.3% (99% CI 47.5-57.1) with an NNT of 200 (chi = 50.7, p < 0.001). Multivariate logistic regression analysis indicated that the effect of NNT on the likelihood for recommending the therapy was age-dependent; young doctors ( < 36 of age) were more sensitive to the difference in NNTs than older doctors. Thirty-six percent (n = 464) of the doctors would not prescribe the drug, and 77.4% (99% CI 68.5-86.2) of those agreed with an argument stating that only one out of NNT patients would benefit from the treatment.

Conclusion: Medical doctors appear to be sensitive to the magnitude of the NNT in their clinical recommendations. However, many doctors believe that only one out of NNT patients benefits from therapy. Clinical recommendations based on this assumption may be misleading.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Aged
  • Chi-Square Distribution
  • Cross-Sectional Studies
  • Decision Making*
  • Drug Therapy*
  • Female
  • Humans
  • Logistic Models
  • Male
  • Middle Aged
  • Norway
  • Physicians, Family*
  • Risk Management
  • Surveys and Questionnaires