baplot: Bland-Altman Plots

In MKinfer: Inferential Statistics

Bland-Altman Plots

Description

Produce classical parametric as well as nonparametric Bland-Altman plots including confidence intervals for the mean of the differences and the lower and upper limit of agreement.

Usage

baplot(x, y, loa.level = 0.95, conf.level = 0.95, 
       xlab = "Mean", ylab = "Difference", 
       title = "Bland-Altman Plot", xlim = NULL, ylim = NULL,
       type = c("parametric", "nonparametric"), 
       loa.type = c("unbiased", "biased"), ci.diff = TRUE, ci.loa = TRUE, 
       ci.type = c("exact", "approximate", "boot"), 
       bootci.type = NULL, R = 9999, print.res = TRUE,
       color.low = "#4575B4", color.upp = "#D73027", 
       alpha = 1, shape = 19, na.rm = TRUE, ...)

Arguments

x	numeric, measurements of method 1.
y	numeric, measurements of method 2.
loa.level	single numeric, level of the level of agreement.
conf.level	single numeric, significance level.
xlab	label of x-axis.
ylab	label of y-axis.
title	title of plot.
xlim	optional numeric vector of length 2, limits of x-axis.
ylim	optional numeric vector of length 2, limits of y-axis.
type	single character, either "parametric" (default) or "nonparametric"; see also Details below.
loa.type	single character, either "unbiased" (default) or "biased"; see also Details below.
ci.diff	single logical, plot confidence interval for the mean of the differences.
ci.loa	single logical, plot confidence intervals for the lower and upper limit of agreement.
ci.type	single character, either "exact" (default) or "approximate" or "boot"; see also Details below.
bootci.type	single character, type of bootstrap interval; see boot.ci and also Details below.
R	single numeric, number of bootstrap replicates.
print.res	single logical, print results of computations in addition to plotting.
color.low	single color (character), color for lower limit of agreement.
color.upp	single color (character), color for upper limit of agreement.
alpha	blending factor (default: no blending.
shape	point shape used.
na.rm	single logical, remove NA values before plotting.
...	further arguments passed to function boot, e.g. for parallel computing.

Details

The plot generates a ggplot2 object that is shown.

Setting type = "parametric" (default), a classical Bland-Altman plot with mean of the differences and standard lower and upper limit of agreement (mean +/- 1.96 SD) will be generated (loa.level = 0.95); see Altman and Bland (1983,1986).

Setting type = "nonparametric", a nonparametric Bland-Altman plot with median of the differences and empirical 2.5% and 97.5% quantiles (loa.level = 0.95) as lower and upper limit of agreement will be generated.

By changing loa.level the lower and upper limit of agreement will correspond to the (1-loa.level)/2 and the 1 - (1-low.level)/2 quantile.

Setting loa.type = "unbiased", the unbiased estimator of the standard deviation will be used; see Shieh (2018).

Setting loa.type = "biased", the standard deviation will be estimated by function sd.

Setting ci.type = "exact", the exact confidence intervals of Shieh (2018) will be used for the lower and upper limit of agreement.

Setting ci.type = "approximate", the approximate confidence intervals of Bland and Altman (1986, 1999) will be used.

Setting ci.type = "boot", bootstrap confidence intervals will be used. In the case of parametric Bland-Altman plots, the studentized bootstrap method will be used by default. In the case of nonparametric Bland-Altman plots, the bootstrap percentile method will be used by default.

The argument bootci.type can be set to "stud" (studentized bootstrap), "perc" (bootstrap percentile) or "bca" (adjusted bootstrap percentile).

Value

Object of class gg and ggplot.

Author(s)

Matthias Kohl [e-mail]

References

Altman DG, Bland JM (1983). Measurement in Medicine: the Analysis of Method Comparison Studies. The Statistician 32:307-317.

Bland JM, Altman DG (1986). Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 327(8476):307-10

Bland JM, Altman DG (1999). Measuring agreement in method comparison studies. Stat Methods Med Res 8:135-160.

Shieh G (2018). The appropriateness of Bland-Altman's approximate confidence intervals for limits of agreement. BMC Med Res Methodol 18:45.

Examples

data(fingsys)
## classical parametric Bland-Altman plot with exact confidence intervals
baplot(x = fingsys$armsys, y = fingsys$fingsys, 
       xlab = "Mean of arm and finger [mm Hg]",
       ylab = "Difference (arm - finger) [mm Hg]")

## nonparametric Bland-Altman plot with exact confidence intervals
baplot(x = fingsys$armsys, y = fingsys$fingsys, type = "nonparametric",
       xlab = "Mean of arm and finger [mm Hg]",
       ylab = "Difference (arm - finger) [mm Hg]",
       title = "Nonparametric Bland-Altman Plot")

MKinfer documentation built on May 29, 2024, 5:13 a.m.