breast.stage: Breast cancer data, stratified by disease stage (2 x 2 x 3...

In clayford/bme: Biostatistical Methods in Epidemiology

Description

A random sample of 199 female breast cancer patients who registered with the Northern Alberta Breast Cancer Registry in 1985. Entry into the cohort was restricted to women with either stage I, II, or III disease. Of the 199 subjects in the cohort, seven died of a cause other than breast cancer. These individuals were dropped from the analysis. Data are provided in different subsets and formats throughout Newman: Table 4.5(a), page 99 (breast); Table 5.3, page 126 (breast.receptor); Table 5.10, page 140 (breast.stage); Table 9.1, page 175 (breast.survival)

Usage

breast.stage

Format

breast 2 x 2 x 3 array

survival

dead, alive

receptor.level

amount of estrogen receptor that is present in breast tissue: low, high

stage

stage of disease: I, II, III

Source

Newman (2001)

Examples

## Convert to data frame for logistic regression, Example 15.1

dat <- data.frame(t(as.data.frame(breast.stage)))
nms <- strsplit(row.names(dat),split = "\\.")
row.names(dat) <- NULL
dat$receptor.level <- factor(sapply(nms, function(x)x[1]), levels = c("high","low"))
dat$stage <- factor(sapply(nms, function(x)x[2]))
rm(nms)

## Fit model (15.5, page 300)
m1 <- glm(cbind(dead,alive)  ~ receptor.level + stage, data=dat, family = binomial)
summary(m1)
exp(confint(m1)) # confidence intervals

## to get confidence intervals presented in Table 15.3
exp(coef(m1) + t(matrix(c(-1,1),nrow=2) %*% (qnorm(0.975) * summary(m1)$coefficients[,2])))

## predicted probability of death during 5-year period follow-up 
## for patient with receptor.level="low" and stage="II"
predict(m1, newdata = data.frame(receptor.level="low", stage="II"), type = "response")

clayford/bme documentation built on May 13, 2019, 7:37 p.m.