R Code and Data for the Cumulative Short-Term Exposure Survival (CSTES) Design: A Novel Framework for Environmental Epidemiology

This repository contains the R implementation and the necessary data to reproduce the Cumulative Short-Term Exposure Survival (CSTES) Design, a novel methodological framework introduced to reinforce the causal assessment of short-term environmental exposures. Methodological Overview: The CSTES design shifts the analytical paradigm from traditional count-based regression (e.g., Poisson GAM) to a survival analysis framework. By re-conceptualizing the short-term exposure accumulation window (e.g., Lag 0-6 days) as the time variable in a Cox Mixed Effects (CoxME) model, this method allows for the estimation of relative risks using the Partial Likelihood framework. This approach serves as a powerful triangulation tool to validate the stability of associations found by standard time-series designs. Key Features: Paradigm Shift: Treats exposure duration as survival time rather than a fixed lag covariate. Heterogeneity Control: Incorporates a Gaussian shared frailty term to explicitly model unobserved day-specific heterogeneity. Reproducibility: Full code provided to replicate the analysis presented in the associated manuscript. Files Included: script_CSTES.txt: The complete R script that applies the CSTES procedure. It includes steps for data transformation, model fitting, testing the Proportional Hazards assumption, and calculating summary Hazard Ratios. valencia.csv: The dataset used for the case study, containing daily all-cause mortality, PM10 concentrations, temperature, and relative humidity for Valencia, Spain. This is an open-access public dataset originally curated and published by Iñíguez et al. (2022). It is included here to ensure the full reproducibility of the results. Data Citation: The valencia.csv dataset is sourced from: Iñíguez, C., Ballester, F., & Tobías, A. (2022). Data supporting the short-term health effects of particles and mortality in 48 Spanish cities. Data in Brief, 42, 108144. DOI: 10.1016/j.dib.2022.108144