A Simple Method for Sensitivity Analysis of Unmeasured Confounding

[摘要] Unmeasured confounding is widely recognized as one of theprincipal problems faced by investigators conducting observationalstudies. Several sensitivity analysis techniques have been developed tohandle unmeasured confounding [1-5]. Recently, VanderWeele andArah [5] provided a general class of formulas for sensitivity analysisof unmeasured confounding. Their formulas benefit from the fact thatthey do not presuppose that any particular method is used to yieldthe initial estimate adjusted only for measured confounders. Threemajor methods to yield such an initial estimate are reviewed in theappendix. In this editorial, we describe a simple sensitivity analysismethod that retains the advantages described above. The method hasfurther advantages that there is only one sensitivity parameter andtherefore the results can easily be displayed graphically, and computerprograms to yield the initial estimate adjusted only for the measuredconfounders can be used for the sensitivity analysis without additionalprogramming.