Threshold Indices

NEAR (Net Efficacy Adjusted for Risk)


1. Description

NEAR (Net Efficacy Adjusted for Risk) uses benefit or risk event and non-event count data for two comparative treatments in a 2×2 table. NEAR estimates NEAR odds ratio (OR) or relative risk (RR) using the standard formulae for OR and RR from the expected frequencies in the 2×2 table.[1][2]

2. Evaluation


2.1 Principle
  • The principle of the approach is logically sound for estimation of odds ratios or relative risks.
  • It is unclear how NEAR increases transparency in the benefit-risk assessment.
  • NEAR recommends the use of data from RCTs and observational studies including their meta-analysis results.
  • Statistical uncertainties of the point estimates are based on confidence intervals of OR and RR.
  • The concept of expected frequency and OR/RR in the best case scenario can be easily understood.

2.2 Features
  • Only one of benefit and risk can be estimated in one NEAR metric.
  • Multiple sources of evidence can be handled using meta-analysis.
  • NEAR has also been extended to take into account intention-to-treat or per protocol analysis.

2.3 Visualisation
  • Visualisations to present NEAR are here

2.4 Assessability and accessibility
  • NEAR avoids the null point and the "signs" problem as in the NNT approach since it is a ratio.
  • NEAR OR and RR must be interpreted rigorously within the context of the problems.
  • Any extrapolation of NEAR results outside the original context should be done with great caution.

3. References

[1] Boada JN, Boada C, Garcia-Saiz M, Garcia M, Fernandez E, Gomez E. Net efficacy adjusted for risk (NEAR): a simple procedure for measuring risk:benefit balance. PLoS One 2008;3(10):e3580.
[2] Boada J, Boada C, Garcia MM, Rodriguez C, Garcia M, Fernandez E. Net efficacy adjusted for risk: Further developments. Expert Opinion on Drug Safety 2009;8(6):649-54.