Uncertainties in baseline risk estimates and confidence in treatment effects.

The GRADE system provides a framework for evaluating how risk of bias, publication bias, imprecision, inconsistency, and indirectness may reduce confidence in estimates of relative effects of interventions on outcomes. However, GRADE and all other systems for rating confidence in effect estimates do not fully address uncertainty in baseline risk and its impact on confidence in absolute estimates of treatment effect. In this article the authors examine factors that may reduce confidence in estimates of baseline risk and thus estimates of absolute treatment benefit