[R-meta] Meta-analysis of diagnostic accuracy studies: Generate bivariate forest plots using R

Rushkin, Megan C Meg@n@C@Ru@hkin @ending from kpchr@org
Sun Sep 30 03:25:41 CEST 2018


I would like to obtain forest plots of meta-analyzed test accuracy studies using a bivariate approach in R.

After some research on available R packages for this task, I decided to try out the Reitsma function in the Mada package. While I'm able to fit the model to my data and obtain the summary results, the forest() function does not work for the bivariate model the way it did for the univariate model:

AuditC6 <- data.frame(TP = c(47, 126, 19, 36, 130, 84), FN = c(9, 51, 10, 3,
19, 2), FP = c(101, 272, 12, 78, 211, 68), TN = c(738, 1543, 192, 276, 959, 89))
AuditC6$names <- c("Study 1", "Study 2", "Study 4", "Study 4", "Study 5",
"Study 6")

#Univariate (madad function) approach that works:
forest(madad(AuditC), type = "sens")
forest(madad(AuditC), type = "spec")

#Bivariate (reitsma function) approach that does not work:
forest(reitsma(AuditC), type = "sens")
forest(reitsma(AuditC), type = "spec")

Resulting error: Error in UseMethod("forest") : no applicable method for 'forest' applied to an object of class "reitsma"

Could someone help me to identify what I'm doing wrong here, or if there is another method to obtain these plots?

Thank you for your time.


	[[alternative HTML version deleted]]

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