[R-meta] From Risk Difference to Odds Ratio

Bernard Fernou bern@rd@|ernou @end|ng |rom gm@||@com
Fri Feb 21 21:42:39 CET 2020

Dear Wolfgang,

As the old axiom goes, the best solutions are always the simplest.. I feel
stupid for not having seen it but thank you very much for the very clear


Le ven. 21 févr. 2020 à 20:03, Viechtbauer, Wolfgang (SP) <
wolfgang.viechtbauer using maastrichtuniversity.nl> a écrit :

> Dear Bernard,
> If you have a risk difference (RD = pT - pC) and then assume a base risk
> for the control group (i.e., pC), then pT = RD + pC. Then you have pT and
> pC from which you can directly compute the OR. Doing this via the NNT seems
> like an unnecessary complication to me.
> And yes, if you do this, then I would do a sensitivity analysis with
> varying values of pC.
> Best,
> Wolfgang
> >-----Original Message-----
> >From: R-sig-meta-analysis [mailto:
> r-sig-meta-analysis-bounces using r-project.org]
> >On Behalf Of Bernard Fernou
> >Sent: Friday, 21 February, 2020 16:05
> >To: r-sig-meta-analysis using r-project.org
> >Subject: [R-meta] From Risk Difference to Odds Ratio
> >
> >Hi everyone
> >I am conducting a meta analysis regarding the relationship between the
> >presence of organophosphates and obesity in metafor. Our IV is mostly
> >reported as dichotomous so, as obesity is also dichotomous, we have
> decided
> >to perform a meta-analysis of odds ratio.
> >
> >However, some authors reported adjusted associations using risk
> difference.
> >We have contacted these authors to access raw data but we have not
> received
> >any response yet. Therefore, our plan was to try to convert this risk
> >difference in odds ratio.
> >
> >To do so,  we started from the fact that risk difference is equal to
> 1/NNT.
> >Then, we convert this NNT in odds ratio using the formula provided in the
> >Cochrane Handbook (link is provided below). The assumed control risk was
> >estimated using the proportion of cases with obesity in the "control"
> group
> >(we believe this is feasible since the study is cross sectional). We plan
> >to perform sensitivity analyses with different assumed control risk.
> >
> >Would it make sense or should we simply exclude this study?
> >
> >Thank you for your help
> >
> >BF
> >
> >https://handbook-5-
> >
> 1.cochrane.org/chapter_12/12_5_4_3_computing_absolute_risk_reduction_or_nnt_
> >from_an_odds.htm

	[[alternative HTML version deleted]]

More information about the R-sig-meta-analysis mailing list