Show me the evidence: Using number needed to treat

This article reviews one of the basic tools of evidence-based medicine, the calculation and interpretation of Number Needed to Treat (NNT) and Number Needed to Harm (NNH). Especially appealing is the simplicity of extracting this information from journal articles that report binary outcomes, such as medication response or emergence of adverse events. On-line resources and calculators can help the clinician in determining confidence intervals for these metrics. After a discussion of absolute versus relative risk, P-values, and the mechanics of calculating NNT and NNH, the application of NNT and NNH to a large clinical trial, the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) for schizophrenia, is described.