Abstract
When a medical intervention prevents a dichotomous outcome, the size of its effect is often estimated with the prevented fraction. Some interventions may reduce the severity of an outcome without entirely preventing it. To quantify the effect of a severity-moderating intervention, a measure termed the mitigated fraction (MF) is proposed. MF has broad applicability, because it measures the overlap of two empirical distributions based on their stochastic ordering. It is also useful in the specific context of medical interventions, because it shares certain structural and functional features with the prevented fraction. The two measures may be applied together in a single semiparametric model with components for outcome prevention and for severity conditional on the presence of the outcome.
DOI
10.22237/jmasm/1130803980
Included in
Applied Statistics Commons, Social and Behavioral Sciences Commons, Statistical Theory Commons
