I have the model below, for which I run a logistic regression including the
interaction term (NSAID*Diuretic)
------------------------
fit1=glm(resp ~ nsaid+diuretic+I(nsaid*diuretic), family= binomial,data=w)
NSAID Diuretic Present Absent
0 0 185 6527
0 1 53 1444
1 0 42 1293
1 1 25 253
Coefficients Std. Error z value Pr(>|z|)
(Intercept) -3.56335 0.07456 -47.794 < 2e-16 ***
NSAID 0.13630 0.17361 0.785 0.43242
Diuretic 0.25847 0.15849 1.631 0.10293
I(NSAID*Diuretic) 0.85407 0.30603 2.791 0.00526 **
---
Signif. Codes: 0 '***' 0.001 '**' 0.01 '*' 0.05
'.' 0.1 ' ' 1
Odds ratio is 2.35 [ln(0.85407)] times higher when NSAID is present in
addition to Diuretic.
-------------------------------
Odds ratio of Nausea when on Diuretic is exp(0.25847)= 1.29
and the odds ratio of Nausea when on NSAID is exp(0.13630)=1.14
Normally when we want to see the odds ratio of Nausea when a patient is on
both drugs we multiply 1.29*1.14= 1.48 (is this correct? do we multiply or
do we add?)
But since the interaction term is significant then we take that into
account? Does that mean that the odds ratio of the interaction is
exp(0.25847)*exp(0.13630)*exp(0.85407)=3.486297 ?
Or do we use additions?
Thanks.
--
View this message in context:
http://www.nabble.com/Logistic-regression-for-drugs-Interactions-tf3404506.html#a9482343
Sent from the R help mailing list archive at Nabble.com.
