[R] Bland-Altman method to measure agreement with repeated measures

Nutter, Benjamin NutterB at ccf.org
Mon Jul 7 14:41:49 CEST 2008


The function given below is one I've written to handle repeated
measures.  I've also included the Help File.  If you happen to see any
potential improvements, I would be open to suggestions.  

###
### Function Code
###

'Bland.Altman' <- function(x,y,alpha=.05,rep.meas=FALSE,subject,...){
#**********************************************************************
#* Construct a Bland Altman Plot
#* 1. Set a few constants
#* 2. Calculate mean difference
#* 3. Calculate difference standard deviation
#* 4. Calculate upper and lower confidence limits
#* 5. Make Plot
#**********************************************************************

#*** 1. Set a few constants
  z <- qnorm(1-alpha/2)  ## value of z corresponding to alpha
  d <- x-y               ## pair-wise differences
  m <- (x+y)/2           ## pair-wise means

#*** 2. Calculate mean difference
  d.mn <- mean(d,na.rm=TRUE)

#*** 3. Calculate difference standard deviation
  if(rep.meas==FALSE){ d.sd=sqrt(var(d,na.rm=TRUE)) }
  else{

    #*** 3a. Ensure subject is a factor variable
    if(!is.factor(subject)) subject <- as.factor(subject)

    #*** 3b. Extract model information
    n <- length(levels(subject))      # Number of subjects
    model <- aov(d~subject)           # One way analysis of variance
    MSB <- anova(model)[[3]][1]       # Degrees of Freedom
    MSW <- anova(model)[[3]][2]       # Sums of Squares

    #*** 3c. Calculate number of complete pairs for each subject
    pairs <- NULL
    for(i in 1:length(levels(as.factor(subject)))){
      pairs[i] <- sum(is.na(d[subject==levels(subject)[i]])==FALSE)
    }
    Sig.dl <- (MSB-MSW)/((sum(pairs)^2-sum(pairs^2))/((n-1)*sum(pairs)))
    d.sd <- sqrt(Sig.dl+MSW)
  }

#*** 4. Calculate lower and upper confidence limits
  ucl <- d.mn+z*d.sd
  lcl <- d.mn-z*d.sd

#*** 5. Make Plot
  plot(m, d,abline(h=c(d.mn,ucl,lcl)),  ...)
  values <- round(cbind(lcl,d.mn,ucl),4)
  colnames(values) <- c("LCL","Mean","UCL")
  if(rep.meas==FALSE) Output <- list(limits=values,Var=d.sd^2)
    else Output <- list(limits=values,Var=Sig.dl)
  return(Output)
}


###
### Help File
###

Bland Altman Plots

Description:

     Constructs a Bland-Altman Plot.

Usage:

      Bland.Altman(x,y,alpha=.05,rep.meas=FALSE,subject,...) 

Arguments:

     x,y: vectors of values to be compared.

   alpha: Significance level for determining confidence limits.
          Defaults to 0.05

rep.meas: Toggles if data provided should be considered as repeated
          measures.  Defaults to 'FALSE'

 subject: Required if 'rep.meas=TRUE'.  A vector of the same length of
          'x' and 'y' that  denotes which subject/group the measurement
          belongs to.

     ...: Other arguments to be passed to the 'plot' method.

Details:

     When 'rep.meas=TRUE', the confidence limits are calculated using a
     method proposed by Bland and Altman. These limits are slightly
     wider, allowing for the correlation within subjects/factors.  The
     standard deviation used to compute these limits is:


                sigma^2[d] = sigma^2[dI] + sigma^2[dw]


     where  sigma^2[d]  is the variance of the differences, sigma^2[dI]
      is the variance of the subjects and methods interaction, and 
     sigma^2[dw]  is the within subject variation. Estimates of these
     values can be found with


                            s^2[dw] = MSw



 s^2[dI] = (MSb - MSw) / ((sum(m[i])^2 - sum(m[i]^2)) /
((n-1)*sum(m[i]))

     )

     Where MSb and MSw are the between and within subject variance of
     the one way analysis of  variance and m[i] is the number of pairs
     for the ith subject.  The sum of these two estimates provides the
     estimate for  s^2[d]  .

Value:

  limits: A vector containing the Mean Bias and confidence limits.

  Var.dl: The Variance of the Bias.  If 'rep.meas=TRUE', this is 

                               s^2[dI]

          .

Author(s):

     Benjamin Nutter nutterb at ccf.org

       Created:  December 2007

References:

     J Martin Bland and Douglass G Altman, "Measuring Agreement in
     Method Comparison Studies", _Statistical Methods in Medical
     Research_, 1999; 8: 135 - 160.

     J Martin Bland and Doublas G. Altman, "Agreement Between Methods
     of Measurement with Multiple Observations per Individual" _Journal
     of Biopharmaceutical Statistics_ 17:571-582, 2007. (Corrects the
     formula  given in the 1999 paper).

     Burdick RK, Graybill FA. _Confidence Intervals on Variance
     Components_. New York: Dekker, 1992.

Examples:

             observer1=rnorm(500,5,2)
             observer2=rnorm(500,10,4)
             ID=rep(1:50,10)

             Bland.Altman(observer1,observer2)

             Bland.Altman(observer1,observer2,rep.meas=TRUE,subject=ID)

-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On Behalf Of kende jan
Sent: Saturday, July 05, 2008 4:18 AM
To: R-help at r-project.org
Subject: [Possible SPAM] [R] Bland-Altman method to measure agreement
with repeated measures


Dear all, 
 
I want to use the Bland-Altman method to measure agreement with repeated
measures collected over period of time (seven periods).

How can I do this with R

 
Many thanks



 
________________________________________________________________________
_____ 

o.fr
	[[alternative HTML version deleted]]

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.



P Please consider the environment before printing this e-mail

Cleveland Clinic is ranked one of the top hospitals
in America by U.S. News & World Report (2007).  
Visit us online at http://www.clevelandclinic.org for
a complete listing of our services, staff and
locations.


Confidentiality Note:  This message is intended for use\...{{dropped:13}}



More information about the R-help mailing list