Hi All,
Can anyone advise on calculating risk difference/absolute risk(odds) reduction (and subsequent NNT) from pooled RR/OR. I’ve seen some calculate this based on the control event rate as a product of the weighted pooled effect estimate but can’t find any lit to support this.
A worked example to illustrate:
Treatment event rate: 32/1424 = 0.022
Control event rate: 71/1546 = 0.046
Pooled OR: 0.45
RD/ARR: 0.046*0.45 = 0.021 (or 2.1%)
Is this right or wrong or somewhere horribly fuzzy in between?
Best,
David
|