Secondary Logo

Journal Logo

Policy-Relevant Proportions for Direct Effects: Erratum

doi: 10.1097/EDE.0000000000000071

The letter suggested that the formula for the proportion eliminated {RR(TE) − RR(CDE(m))} / {RR(TE)−1} (calculated from the risk-ratio model) was expressed on a risk-difference scale; however, it is in fact expressed on a scale of excess relative risk (i.e., relative risk minus one) and the two do not coincide in this context. For further discussion, refer to the following letter in this issue: Suzuki E, Evans D, Chaix B. On the 'proportion eliminated' for risk differences versus excess relative risks [letter]. Epidemiology. 2014;25:308.

Back to Top | Article Outline


VanderWeele TJ. Policy-relevant proportions for direct effects [letter]. Epidemiology. 2013;24:175–176
© 2014 by Lippincott Williams & Wilkins, Inc