Interaction of physical activity and low-level air pollution on cardiovascular disease: a prospective cohort study in UK Biobank

To investigate the associations of physical activity (PA), low-level air pollution, and interaction on cardiovascular diseases (CVD) incidence based on the UK Biobank. PA was measured by the International Physical Activity Questionnaire and five air pollutants were estimated using Land Use Regression. All association estimates were based on Cox regression. Dose–response relationship was explored by restricted cubic spline, while multiplicative and additive interaction were examined by P_interaction and relative excess risk due to interaction (RERI). As deviating proportional hazards assumption, we analyzed data as follow-up < 4 years and ≥ 4 years, separately. PA with 1000–4000 Metabolic Equivalent Task (MET) min/week showed the strongest protective impact on CVD incidence, while only low-level nitrogen dioxides (NO₂) showed negative impact among five air pollutants and was considered for further analysis. Multiplicative interaction between PA and NO₂ was observed during ≥ 4 years follow-up (P_interaction = 0.049) while not during < 4 years (P_interaction = 0.290). Positive additive interactions were found for high PA and low NO₂ (< 20 μg/m³) group (RERI: 0.07, 95% confidence intervals: 0.02–0.11) during < 4 years, and for moderate PA with NO₂ at 40- μg/m³ (0.07, 0.02–0.13) and < 20 μg/m³ (0.07, 0.02–0.12), while high PA showed similar results with NO₂ at 40-, 20- and < 20 μg/m³ during ≥ 4 years. PA about 1000–4000 METs min/week showed the lowest CVD risk. Possibility of interaction with PA and NO₂ is more likely to present with the increase in follow-up duration. We call for the optimal thresholds of PA, and exploring interaction thoroughly by considering types of PA.