[SOLVED] Calculate risk ratio after balancing with WeightIt

Calculate risk ratio after balancing with WeightIt

I used WeightIt to balance the characteristics of my population dividing into treated and non-treated as in this example:

library(cobalt)
library(WeightIt)
data("lalonde", package = "cobalt")
head(lalonde)

W.out <- weightit(treat ~ age + educ + race + married + nodegree + re74 + re75,
                  data = lalonde, estimand = "ATT", method = "ps")

My outcome is re78. I would like to calculate the RISK RATIO with 95% CONFIDENCE INTERVALS. How can I do it? For odds ratio I know that I can use the survey package and the svyglm function as in this example:

library(survey)
d.w <- svydesign(~1, weights = W.out$weights, data = lalonde)
fit <- svyglm(re78 ~ treat, design = d.w)
coef(fit)

but how can I compute risk ratios with their 95% CI instead of odds ratio in my weighted dataset?

Solution

Since re78 is not integer you need to use quasipoisson instead of poisson if you want to avoid warnings, although the software gives the dame results.

 fit <- svyglm(re78 ~ treat, design = d.w, family="quasipoisson")

coef(fit)
(Intercept)       treat 
   8.543849    0.212226 

coef(summary(fit))
            Estimate Std. Error   t value      Pr(>|t|)
(Intercept) 8.543849  0.1146916 74.494128 4.321164e-309
treat       0.212226  0.1463640  1.449987  1.475743e-01

The risk ratio is just exp(coef(fit)[2]) and it's CI is just this pseudo code.

exp( coef(fit)[2] +/- 1.96*coef( summary(fit))[2,2] )