In mcguinlu/robvis: Visualize the Results of Risk-of-Bias (ROB) Assessments

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

robvis now provides some functionality to work with the results of meta-analyses performed using the metafor package.

First let's load the relevant packages and define our datasets:

library(robvis)
library(metafor)
library(dplyr)

# Define your studies, using the BCG dataset included in the metadat package
dat_bcg <- metadat::dat.bcg

glimpse(dat_bcg)

# Create some example data for ROB2 using rob_dummy(), and add it to the BCG
# data.
# We don't need a "Study" column for this example, so we set `study = FALSE`

dat_rob <- rob_dummy(13, 
                     "ROB2",
                     study = FALSE)

dat_analysis <- cbind(dat_bcg, dat_rob)

glimpse(dat_analysis)

Now each row in the dataset contains a single numerical result and it's corresponding risk-of-bias assessment.

Perform the meta-analysis

Let's now perform a simple random-effects meta-analysis using metafor::rma():

# Calculate effect estimates and sampling variances for each study
dat_analysis <-
  metafor::escalc(
    measure = "RR",
    ai = tpos,
    bi = tneg,
    ci = cpos,
    di = cneg,
    data = dat_analysis
  )

# Perform the meta-analysis
res <- metafor::rma(yi,
                    vi,
                    data = dat_analysis,
                    slab = paste(author, year))

# Explore the results
res

Create paired forest plot

Having performed the meta-analysis, you can now append a risk-of-bias traffic-light plot onto the standard output from metafor::forest() by simply passing the results object (res) to rob_forest():