Biostatistics III in R

In biostat3: Utility Functions, Datasets and Extended Examples for Survival Analysis

Exercise 25. Localised melanoma : generating and analysing a nested case-control (NCC) study

---

library('knitr')
read_chunk('../q25.R')
opts_chunk$set(cache=FALSE)

(a) ##

There are 5318 individuals in the study that we would need to collect data for if we were to use the complete cohort of patients.

(b) ##

960 cancer patients die from melanoma during the first 10 years of follow-up.

(c1) ##

(c2) ##

(d) ##

(e) ##

Note that, since every nested case-control study is different, the parameter estimates you obtain will not be identical to those above. However, the hazard ratios from the two models should be very similar. The standard errors are slightly larger for the nested case-control study since the estimates are based on a sample from the full cohort. Loss of precision is the trade-off we have to make when designing a nested case-control study. The precision can be improved by adding more controls to each case.

(f) ##

We see that there is sampling variation in the parameter estimates from the five nested case-control studies but they are centered on the full cohort estimate. We see that the standard errors of the estimates from the nested case-control studies are larger than for the full cohort but there is some sampling variation.

biostat3 documentation built on Oct. 29, 2024, 5:07 p.m.