Abstract
OBJECTIVES: We investigated changes in weight, exercise frequency, and perceived health from the first to last health profile assessment (HPA) and between the number of tests within five years. We examined whether sociodemographic factors, or baseline values influenced these changes. METHODS: Data from 106 005 employees with ≥2 HPA (1990-2021) were included. Change between the first and last HPA within a five-year period was analyzed. Baseline age, sex, education, occupation, and baseline values of each outcome were included as predictors. XGBoost models assessed changes in the outcomes, and performance was evaluated via root mean squared error, mean absolute error, and R-squared. We employed Shapley Additive Explanations and forward marginal effects to interpret dose-response relationships and subgroup differences. RESULTS: Predictive performance was low, suggesting that the included variables only partially explained observed changes. Nonetheless, longer intervals between the first and last HPA correlated with greater weight gain, while a higher number of tests predicted slightly lower weight gain and modest improvements in perceived health and exercise frequency, compared to the average change. Younger participants had larger weight increases, whereas those with higher education showed smaller declines in exercise frequency. CONCLUSIONS: Infrequent HPA alone did not appear to substantially influence the lifestyle-related factors studied. However, more frequent HPA, coupled with enhanced feedback and support, may yield small improvements in weight, perceived health, and exercise frequency compared to the average change.