Setting Up Survival Analysis for Recurrent Events with Competing Risks in Stata

[Cheikh Ahmadou]

Hello Statalist,

I am working on a survival analysis project using an administrative dataset from a social welfare program. The dataset tracks recipients, documenting their monthly program participation, including multiple entries and exits for each individual.

Each individual can exit the program for various reasons, including "death," "retirement," "found a job," and "others." The dataset spans from month 477 to 731, with individuals entering the program at different times. The goal of my analysis is to measure the probability of leaving the program for the specific event "found a job," considering individual characteristics like gender, disability status, education, etc.

Here is a simplified version of the dataset:

clear
input id mois episode exit
1 477 1 0
1 478 1 0
1 479 1 1
1 490 2 0
1 491 2 0
1 492 2 2
2 531 1 0
2 532 1 0
2 533 1 0
2 534 1 2
2 548 2 0
2 549 2 0
2 550 2 0
2 551 2 2
3 724 1 0
3 725 1 0
3 726 1 0
3 727 1 0
3 728 2 0
3 729 2 0
3 730 2 0
3 731 2 0
end

In this dataset, the variable `exit` denotes the reason for exiting the program, with "2" indicating "found a job." I am seeking advice on how best to set up my data for survival analysis in Stata, considering the recurrent nature of the events and the competing risks posed by different exit reasons.

Any insights or suggestions on how to approach this analysis would be greatly appreciated.

Thank you in advance!