cd4: Repeated CD4 counts data from AIDS clinical trial
In ALA: Data sets and examples for the book "Applied Longitudinal Analysis"

Description Usage Format Details Note Source References Examples

The data are from a randomized, double-blind, study of AIDS patients with advanced immune suppression (CD4 counts of less than or equal to 50 cells/mm^3).

data(cd4)

A data frame with 5036 observations on the following 6 variables.

id: a factor with 4992 levels
treatment: a factor with levels zX400d (zidovudine alternating monthly with 400 mg didanosine), zA225z (zidovudine plus 2.25 mg of zalcitabine), zA400d (zidovudine plus 400 mg of didanosine), and zA400dA400n (zidovudine plus 400 mg of didanosine plus 400 mg of nevirapine).
age: a numeric vector
gender: a factor with levels F and M
week: a numeric vector; weeks since baseline time
logCD4: a numeric vector; log transformed CD4 counts (log(CD4 counts + 1))

Patients in AIDS Clinical Trial Group (ACTG) Study 193A were randomized to dual or triple combinations of HIV-1 reverse transcriptase inhibitors. Specifically, patients were randomized to one of four daily regimens containing 600mg of zidovudine: zidovudine alternating monthly with 400mg didanosine; zidovudine plus 2.25mg of zalcitabine; zidovudine plus 400mg of didanosine; or zidovudine plus 400mg of didanosine plus 400mg of nevirapine (triple therapy). Measurements of CD4 counts were scheduled to be collected at baseline and at 8-week intervals during follow-up. However, the CD4 count data are unbalanced due to mistimed measurements and missing data that resulted from skipped visits and dropout. The number of measurements of CD4 counts during the first 40 weeks of follow-up varied from 1 to 9, with a median of 4. The response variable is the log transformed CD4 counts, log(CD4 counts + 1), available on 1309 patients.

Original variable names have been adapted to R conventions.

http://biosun1.harvard.edu/~fitzmaur/ala

Henry K, Erice A, Tierney C, Balfour HH Jr, Fischl MA, Kmack A, Liou SH, Kenton A, Hirsch MS, Phair J, Martinez A, Kahn JO, for the AIDS Clinical Trial Group 193A Study Team (1998) A randomized, controlled, double-blind study comparing the survival benefit of four different reverse transcriptase inhibitor therapies (three-drug, two-drug, and alternating drug) for the treatment of advanced AIDS. Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology 19:339-349.

str(cd4)
summary(cd4)

if (require(lattice)) {
    ## Create new data with just 2 levels of treatment
    cd4New <- within(cd4, {
        treatment <- factor(ifelse(treatment != levels(treatment)[4], 0, 1),
                            labels=c("double", "triple"))
    })
    ## Fig. 8.9 (roughly)
    xyplot(logCD4 ~ week, data=cd4New, groups=treatment, type="smooth",
           aspect=1.5, cex=0.5, col=1, lty=c(2, 1),
           scales=list(rot=c(0, 1), tck=c(0.5, 0)), ylim=c(2.5, 3.5),
           xlab="Time (weeks)", ylab="Log(CD4 + 1)")
}

if (require(lme4)) {
    str(cd4New)
    cd4New <- within(cd4New, {
        ## Use a stage factor -- this is what is needed to make the same
        ## interpretations as in the book
        stage <- cut(week, breaks=c(floor(min(week)), 16,
                             ceiling(max(week))),
                     labels=c("pre", "post"), include.lowest=TRUE)
        ## But this is what is actually used
        stage.tij <- ifelse(week > 16, week - 16, 0)
    })
    summary(cd4New)
    ## Model in p. 227
    (fm1 <- lmer(logCD4 ~ week + stage.tij + week:treatment +
                 stage.tij:treatment + (week + stage.tij | id), data=cd4New))
## Not run: 
    ## This should be an equivalent model using the stage factor, but we
    ## don't get quite the same coefficients
    (fm1b <- lmer(logCD4 ~ week + week:stage + week:treatment +
                  week:stage:treatment + (week + week:stage | id), data=cd4New,
                  control=list(maxIter=2000, maxFN=2000)))

## End(Not run)
    ## Table 8.13
    VarCorr(fm1)[[1]] * 1000
    ## Model in p. 229 -- we don't get quite the same coefficients
    (fm2 <- lmer(logCD4 ~ week + stage.tij + treatment:(week - stage) +
                 age + gender + (week + stage.tij | id), data=cd4New))

    ## Fig. 8.7 (roughly)
    set.seed(12)
    rndID <- as.character(with(cd4New,
                               sample(unique(id[treatment == "triple" &
                                                gender == "M"]), 2)))
    wk <- with(cd4New, seq(floor(min(week)), ceiling(max(week))))
    st <- ifelse(wk > 16, wk - 16, 0)
    mm <- model.matrix(logCD4 ~ week + stage + treatment:(week - stage) +
                       age + gender,
                       data=with(cd4New,
                         data.frame(treatment=treatment[treatment == "triple"][1],
                                    age=45, gender=gender[gender == "M"][1],
                                    week=wk, stage=st, logCD4=rnorm(length(wk)))))
    pred.fixef <- mm %*% fixef(fm2)
    fitted.cd4 <- fitted(fm2)
    plot(pred.fixef ~ mm[, 2], type="l", lwd=2, ylim=range(fitted.cd4),
         xlab="Time (weeks)",
         ylab="Log(CD4 + 1)")
    with(cd4New, {
        points(week[id == rndID[1]], logCD4[id == rndID[1]])
        lines(week[id == rndID[1]], fitted.cd4[id == rndID[1]])
        points(week[id == rndID[2]], logCD4[id == rndID[2]], pch=2)
        lines(week[id == rndID[2]], fitted.cd4[id == rndID[2]], lty=2)
    })
}