A function to call package forestplot from R library and produce forest plot using
results from bmeta. The posterior estimate and credible interval for each study are
given by a square and a horizontal line, respectively. The summary estimate is drawn
as a diamond.
estimates on natural scale is displayed by default. If TRUE, log scale is used (i.e.
log odds ratio, log incidence rate ratio). For continuous data, estimates are always
presented on natural scale and users do not need to specify this argument.
study.label
label for each study and the summary estimate. See details.
clip
lower and upper limits for clipping credible intervals to arrows
lines
selects the colour for the lines of the intervals. If the extra option add.null
is set to TRUE, then lines should be specified as a two-element vector.
If the user fails to do so, bmeta will overwrite this setting and select suitable
values.
box
selects the colour for mean study-specific estimates. If the extra option add.null
is set to TRUE, then box should be specified as a two-element vector.
If the user fails to do so, bmeta will overwrite this setting and select suitable
values.
summary
selects the colour for the pooled estimate
box.symb
selects the symbol used to plot the mean. Options are "box" (default) or "circle"
label.cex
defines the size of the text for the label. Defaults at .8 of normal size
xlab.cex
defines the size of the text for the x-label. Defaults at 1 of the normal size
ticks.cex
defines the size of the text for the x-axis ticks. Defaults at .8 of the normal size
...
Additional arguments. Includes
- add.null = TRUE/FALSE. If set to true, adds a plot of the null (no-pooling model)
- line.margin = the distance between lines in case multiple graphs are shown on the same
plot
- box.size = the size of the summary box
- new.page = TRUE/FALSE. If set to true, then a new graph overwrite the existing one
- zero (x-axis coordinate for zero line. If you provide a vector of length 2 it
will print a rectangle instead of just a line. Default at 0 or 1 depending on log scale)
- legend = a legend for the multi-graph plot (including the null/no-pooling model)
Author(s)
Tao Ding
Gianluca Baio
Examples
### Read and format the data (binary)
data = read.csv(url("http://www.statistica.it/gianluca/bmeta/Data-bin.csv"))
### List data for binary outcome
data.list <- list(y0=data$y0,y1=data$y1,n0=data$n0,n1=data$n1)
### Select fixed-effects meta-analysis with normal prior for binary data
x <- bmeta(data.list, outcome="bin", model="std.norm", type="fix")
### Plot forest plot
forest.plot(x)
### Plot forest plot on log scale
forest.plot(x,log=TRUE)
### Select random-effects meta-analysis with t-distribution prior for binary
### data
x <- bmeta(data.list, outcome="bin", model="std.dt", type="ran")
### Plot 'two-line' forest plot showing estimates from both randome-effects
### model and no-pooling effects model for comparison
forest.plot(x,add.null=TRUE,title="Two-line forestplot for comparison")
### Read and format the data (continuous)
data = read.csv(url("http://www.statistica.it/gianluca/bmeta/Data-ctns.csv"))
### List data for continuous outcome
data.list <- list(y0=data$y0,y1=data$y1,se0=data$se0,se1=data$se1)
### Select fix-effects meta-analysis for studies reporting two arms separately
x <- bmeta(data=data.list,outcome="ctns",model="std.ta",type="fix")
### Define for individual studies
study.label <- c(paste0(data$study,", ",data$year),"Summary estimate")
### Produce forest plot with label for each study and control the lower and upper
### limits for clipping credible intervals to arrows
forest.plot(x,study.label=study.label,clip=c(-7,4))