forest function

Forest plot for univariate measures

Produce a forest plot. Includes graphical summary of results if applied to output of suitable model-fitting function. forest methods for madad and madauni objects are provided.

## S3 method for class 'madad'
forest(x, type = "sens", log = FALSE, ...)
## S3 method for class 'madauni'
forest(x, log = TRUE, ...)
forestmada(x, ci, plotci = TRUE, main = "Forest plot", xlab = NULL,
          digits = 2L,  snames = NULL, subset = NULL, pch = 15, 
          cex = 1, cipoly = NULL, polycol = NA, ...)

Arguments

• x: an object for which a forest method exists or (in the case of foresmada) a vector of point estimates.
• ci: numeric matrix, each row corresponds to a confidence interval (the first column being the lower bound and the second the upper).
• plotci: logical, should the effects sizes and their confidence intervals be added to the plot (as text)?
• main: character, heading of plot.
• xlab: label of x-axis.
• digits: integer, number of digits for axis labels and confidence intervals.
• snames: character vector, study names. If NULL, generic study names are generated.
• subset: integer vector, allows to study only a subset of studies in the plot. One can also reorder the studies with the help of this argument.
• pch: integer, plotting symbol, defaults to a small square. Also see plot.default.
• cex: numeric, scaling parameter for study names and confidence intervals.
• cipoly: logical vector, which confidence interval should be plotted as a polygon? Useful for summary estimates. If set to NULL, regular confidence intervals will be used.
• polycol: color of the polygon(s), passed on to polygon. The default value of NA implies no color.
• type: character, one of sens, spec, negLR, posLR or DOR.
• log: logical, should the log-transformed values be plotted?
• ...: arguments to be passed on to forestmada and further on to other plotting functions

Details

Produces a forest plot to graphically assess heterogeneity. Note that forestmada is called internally, so that the ... argument can be used to pass on arguments to this function; see the examples.

Returns

Returns and invisible NULL.

Author(s)

Philipp Doebler philipp.doebler@googlemail.com

See Also

madad, madauni

Examples

data(AuditC)

## Forest plot of log DOR with random effects summary estimate
forest(madauni(AuditC))

## Forest plot of negative likelihood ratio (no log transformation)
## color of the polygon: light grey 
## draw the individual estimate as filled circles
forest(madauni(AuditC, type = "negLR"), 
       log = FALSE, polycol = "lightgrey", pch = 19)

## Paired forest plot of sensitivities and specificities
## Might look ugly if device region is too small
old.par <- par()
AuditC.d <- madad(AuditC)

plot.new()
par(fig = c(0, 0.5, 0, 1), new = TRUE)
forest(AuditC.d, type = "sens", xlab = "Sensitivity")
par(fig = c(0.5, 1, 0, 1),  new = TRUE)
forest(AuditC.d, type = "spec", xlab = "Specificity")

par(old.par)

## Including study names
## Using Letters as dummies
forest(AuditC.d, type = "spec",  xlab = "Specificity",
      snames = LETTERS[1:14])