Outcome of interest
We wish to validate our estimate of the risk of colorectal cancer (crc), so our outcome of interest is time until crc diagnosis.
The outcome was constructed as follows:
- time until death was calculated as age from study entry to age at death (or censoring) (
t2dth - age_ref) - time until dx was calculated as age from study entry to age at crc dx (or adv adenoma). (
age_dxsel - age_ref) The survival time for each observation was taken to be the minimum of the two above times.
- Question: there are 31 people that had cause of death (
dthc) as ‘colorectal’ cancer but were not diagnosed with crc. I count these as death events. Advanced adenoma
Screen detected cases
We remove 189 crc cases from the intervention arm detected in the first year of screening.
Cumulative incidence of CRC dx and other competing events
There are several competing events that we must account for including adenoma, hyperplastic polyps, … (??) and death. Below we show cumulative incidence curves for crc, advanced adenoma, and death in the PLCO trial over the first 14 years. The dashed line shows the probability of CRC calculated using a Kaplan Meier curve, ignoring competing events.
| Censored | CRC | Advanced Adenoma | Death |
|---|---|---|---|
| 125147 | 2010 | 84 | 20312 |
Cumulative incidence of CRC by trial arm, gender, age
Arm
Gender
Age
Age is categorized as \(\leq\) 60, 60-65, 65-70, and 70+.
