Effects of residential PM2.5 exposures from indoor and outdoor sources on blood pressure and respiratory inflammation in rural and urban Beijing

Abstract

Introduction Arising from different activities, indoor-generated (PM2.5_ig) and outdoor-generated residential PM2.5 (PM2.5_og) may have different toxicities and health effects. We aimed to evaluate the effects of PM2.5_ig and PM2.5_og on systolic blood pressure (SBP), diastolic blood pressure (DBP) and respiratory inflammation (represented by exhaled Nitric Oxide (eNO)).Methods 72 subjects participated in the residential monitoring of PM2.5 in urban and rural Beijing during winter 2016 and summer 2017. In total, valid data were captured for 450 measurement days (approx. 3 days per participant per season). Within the same week of exposure monitoring, BP and eNO from participants were measured two times. A classifying algorithm was developed to isolate PM2.5_ig and PM2.5_og from residential and ambient measurements. Linear mixed-effects model was used to examine the associations between residential exposure and health outcomes.Results For all measurements except PM2.5_og (PM2.5_ig, eNO, SBP and DBP), significant differences were observed between rural and urban participants during the two seasons. For all measurements (PM2.5_ig, PM2.5_og, eNO, SBP and DBP), significant differences were observed between seasons in both sites. Overall, an interquartile range (IQR) increase (22.0 ug/m3) in lag 1-day exposure to PM2.5_og was associated with an elevation in SBP by 1.70% (confidence interval [CI]: 0.47%, 2.95%) and eNO by 15.44% (CI: 6.60%, 25.02%); an IQR increase (5.8 ug/m3) in lag 2-day exposure to PM2.5_ig was associated with an elevation in SBP by 1.12% (CI: 0.26%, 2.00%) and an increased DBP by 1.26% (CI: 0.39%, 2.14%). However, PM2.5_ig were negligible (<0.5 ug/m3) in 44% and 54% of measurement days during winter 2016 and summer 2017, which affected the output from the linear mixed-effects model.Conclusion PM2.5_og and PM2.5_ig demonstrated different lag effects and effect sizes in the assessed health metrics. Full investigation with alternative modelling techniques is ongoing to evaluate the relationship in more detail.