lrt.homogeneity: Likelihood ratio test of homogeneity of hazard ratios...

In clayford/bme: Biostatistical Methods in Epidemiology

Description

Performs a likelihood ratio test of homogeneity of hazard ratios for stratified survival data. The null hypothesis is that all stratum-specific hazard ratios are equal.

Usage

lrt.homogeneity(time, status, exposure, strata)

Arguments

time	a numeric vector of survival times.
status	a numeric vector of censoring indicators, with 0 = censored and 1 = dead.
exposure	a factor vector with two levels indicating exposure. The first level is assumed to be the exposed condition.
strata	a factor vector with at least two levels indicating strata.

Details

According to Newman (p. 124): "the homogeneity assumption is merely a convenient fiction that is adopted in order to simplify the analysis and interpretation of data." A suggested approach is to first examine stratum-specific hazard ratios and their confidence intervals to get a sense if there are meaningful differences across strata (see hazard.ratios); perform a formal test of homogeneity; and then "synthesize this information along with substantive knowledge of the relationship between exposure and disease, taking into account the aims of the study."

Value

An object of class "htest" with title, test statistic, and p-value.

References

Newman (2001), page 221.

Examples

## Example 10.14
with(breast.survival, lrt.homogeneity(time, status, receptor.level, stage)) 

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