R - confidence intervals for odds ratio in matched pair

• the confidence interval for odds ratio in matched pairs is B/C * exp(+-1.96*sqrt(1/B + 1/C)) where B and C are the counts of discordant pairs
  
• here is a example with anemia in mothers ~ low birth weight
  

bw <- as.table(matrix(c(13,27,8,22),dimnames=list(c("anemia","no anemia"),c("low bw","normal bw")),byrow=T,nrow=2))
bw

low bw normal bw
anemia        13        27
no anemia      8        22

• do the calculations per hand:
  

print(paste("odds ratio:",27/8))
27/8 * exp(c(-1,1)*1.96*sqrt(1/27 + 1/8))

[1] "odds ratio: 3.375"
[1] 1.533297 7.428844

• or use the function from the PropCIs package
  

require(PropCIs,quietly=T)
oddsratioci.mp(8,27,0.95)

data:  

95 percent confidence interval:
 1.562965 7.287833