Reproducibility package
=======================

Study
-----
Predicted fat mass index and incident stroke among adults aged 65 years or older across CHARLS, ELSA, and HRS.

Scope
-----
All analyses are restricted to participants aged 65 years or older at the analytic baseline. No participants younger than 65 years are retained in primary, subgroup, weighted, competing-risk, or time-varying analyses. Sex-specific standardization of pBFMI, BMI, and waist-to-height ratio was recomputed within the age-65-or-older sample of each cohort.

Contents
--------
code/age65_complete_analysis.R
  Reproduces the cohort-specific discrete-time and Cox models, common-sample sequential models, interaction tests, Hartung-Knapp descriptive synthesis, mathematical-coupling diagnostics, pBFMI-versus-BMI comparison, subgroup and early-event analyses, sampling-weighted models, competing-risk analyses, ELSA time-varying analyses, and figures.

results/
  Non-restricted aggregate CSV and text outputs used in the manuscript and supplementary files. Participant-level data are not included.

Required controlled-access data
-------------------------------
Registered users must obtain CHARLS, ELSA, and HRS participant-level data from the respective study portals and retain the source directory structure expected by the analysis script. Access conditions prohibit redistribution of participant-level files in this package.

Software
--------
R 4.5.1. Exact package and platform information is recorded in results/sessionInfo.txt.
