Exercise 11. Cox regression with all-cause mortality as the outcome

Now fit a model to the localised melanoma data where the outcome is observed survival (i.e. all deaths are considered to be events).

You may have to install the required packages the first time you use them. You can install a package by install.packages("package_of_interest") for each package you require.

library(biostat3) # 
library(dplyr)    # for data manipulation

Load the melanoma data and explore it.

## Read melanoma data
## and select subcohorts
melanoma.l <- biostat3::melanoma %>%
  filter(stage=="Localised") %>%
  mutate(
    ## Create a death indicator
    death_cancer = as.numeric(status=="Dead: cancer"),
    death_any = as.numeric(status=="Dead: cancer" | status=="Dead: other") )

## Truncate follow-up time

melanoma.l2 <-
  mutate(melanoma.l,
         ## Create new death indicators (only count deaths within 120 months)
         death_cancer = death_cancer * as.numeric( surv_mm <= 120),
         death_any = death_any * as.numeric( surv_mm <= 120),
         ## Create a new time variable
         surv_mm = pmin(surv_mm, 120))

(a)

Interpret the estimated hazard ratio for the parameter labelled agegrp60-74, including a comment on statistical significance.

summary( coxfit11a <- coxph(Surv(surv_mm, death_any) ~ sex + year8594 + agegrp,
                           data = melanoma.l2) )

(b)

On comparing the estimates between the observed and cause-specific survival models it appears that only the parameters for age have changed substantially. Can you explain why the estimates for the effect of age would be expected to change more than the estimates of the effect of sex and period?

summary( coxfit11b <- coxph(Surv(surv_mm, death_cancer) ~ sex + year8594 + agegrp,
                           data = melanoma.l2) )