Population-based Night Work Exposure Assessment Using Swiss Labour Force Survey Data (1991–2024)

This project aims to develop and validate a Job-Exposure Matrix for night work (NW‑JEM) to support exposure assessment in epidemiological and occupational health research. The JEM is based on data from the Swiss Labour Force Survey covering the period 1991–2024, comprising approximately 1.3 million observations. Night work exposure was defined in three categories (regular, occasional, none) and modelled using ordinal logistic regression. Exposure probabilities were estimated for strata defined by calendar year, sex, age group, and occupation (ISCO‑88 classification).
A continuous exposure metric, the Night Work Exposure Index (NWEI), was computed as a weighted combination of predicted probabilities of exposure.
Criterion validity was assessed using external data from the CoLaus|PsyCoLaus cohort. Agreement between JEM-derived and empirical exposure indices at the stratum level was quantified using Spearman’s correlation and intraclass correlation coefficients (ICC), showing moderate to good agreement, particularly at finer occupational levels. At the individual level, the binary classification performance of the JEM was evaluated using metrics including sensitivity, specificity, predictive values, Cohen’s κ, and AUC. Results indicate good discriminative ability, strong capacity to rule out exposure, and consistent performance across occupational resolutions.