OBJECTIVES: Number needed to treat (NNT) measures the effectiveness of a treatment and indicates the number of patients necessary to treat either to prevent one additional adverse event or acquire one additional beneficial outcome in an active treatment over a control. A critical question concerns the value of an intervention or treatment in the context of increasing NNT, versus the severity of the adverse event avoided or highly beneficial outcome. We investigated the range and averages of reported NNT in selected oncologic studies to address this question.

METHODS: A targeted literature review was conducted, and relevant NNT data from different cancers were extracted. The data were analyzed by average and distributional ranges (CIs, min/max, etc.) to help determine the positioning of new NNT values compared to the published literature.

RESULTS: From one hundred studies identified initially, fifty-eight studies met the inclusion criteria. Of the included studies, 22% were on prostate cancer, followed by breast cancer (9%), and skin cancer (9%). The range of NNT varied significantly between different tumor types in included studies. Other important factors affecting NNT context include clinical endpoint, length of follow-up and patient baseline characteristics. In the context of a severe clinical endpoint (i.e. poor survival or in pediatric oncology) a high NNT is more tolerable. In the absence of a threshold for evaluating NNT, the absolute risk reduction statistical significance may be useful.

CONCLUSIONS: To our knowledge, this is the first literature review of studies reporting on NNT to measure the effectiveness of the intervention in the field of oncology. The NNT should not be interpreted without considering the context of the evaluation, a high NNT may be considered acceptable for a severe event, even if a much lower NNT would not be acceptable for a mild complication.