Reader Comments

Post a new comment on this article

Corrections to RERI

Posted by IlseFlink on 19 Aug 2013 at 12:57 GMT

I noticed two minor mistakes concerning the RERI which need correction. Firstly the Acronym RERI should be “Relative Excess Risk due to Interaction” instead of “Relative Access Risk due to Interaction”

Secondly the comma in the RERI lower bound confidence interval is wrongly placed in the abstract, text (results section) and table 3. This currently reads RERI: −3.22, 95% CI −0.70; –0.59 while this should be RERI: −3.22, 95% CI −7.04; –0.59

No competing interests declared.

RE: Corrections to RERI

megannhall replied to IlseFlink on 23 Sep 2014 at 22:16 GMT

I believe that there may be a mistake in the calculation of the RERI in Table 3 for the high vs. low neighborhood ethnic diversity comparison. This RERI is reported as -1.53. I believe it should be: RERI (high vs. low) = 3.03 - 5.24 - 0.96 + 1 = -2.17. But perhaps I am mistaken. Could you please explain how you calculated the -1.53? Thank you.

No competing interests declared.

RE: RE: Corrections to RERI

IlseFlink replied to megannhall on 29 Sep 2014 at 11:34 GMT

Thank you very much for your alertness. We used the delta approach for which an excell file is provided by Knol et al. 2011 (also see the article by Knol & vanderWeele 2012 published in the International Journal of Epidemiology) to calculate the RERI and the CI's. It appears that there was a mistake in calculating the RERI for this comparison and you are right that instead of -1.53 (-5.52; 2.245) P=0.449 this should be -2.17 (-6.61; 2.45) P=0.34.

Once again many thanks for your alertness.

No competing interests declared.