TY - JOUR
T1 - Predictors of Personal Exposure to Fine Particulate Matter, Black Carbon, and Carbon Monoxide among Pregnant Women in Rwanda
T2 - Baseline Data from the HAPIN Trial
AU - Karakwende, Patrick
AU - Checkley, William
AU - Chen, Yunyun
AU - Clark, Maggie L.
AU - Clasen, Thomas
AU - Dusabimana, Ephrem
AU - Jabbarzadeh, Shirin
AU - Johnson, Michael
AU - Kalisa, Egide
AU - Kirby, Miles
AU - Naher, Luke
AU - Ndagijimana, Florien
AU - Ndikubwimana, Adolphe
AU - Ntakirutimana, Theoneste
AU - de Dieu Ntivuguruzwa, Jean
AU - Peel, Jennifer L.
AU - Piedrahita, Ricardo
AU - Pillarisetti, Ajay
AU - Rosa, Ghislaine
AU - Waller, Lance A.
AU - Wang, Jiantong
AU - Young, Bonnie N.
AU - the HAPIN Investigators
N1 - Publisher Copyright:
© (2025), (Pure Earth). All rights reserved.
PY - 2025/3/1
Y1 - 2025/3/1
N2 - BACKGROUND: Exposure to household air pollution from the combustion of solid fuels is a leading risk factor for death and disease in low- and middle-income countries, where cleaner cooking and lighting options are often unavailable. Few studies have measured personal exposure during pregnancy, a sensitive period of development, particularly in Africa. OBJECTIVE: We aimed to characterize exposure during early to midpregnancy among women in Rwanda and to assess predictors of personal exposure, including stove and fuel type, cooking behaviors, housing conditions, sociodemographic characteristics, and other potential sources of exposure. METHODS: We assessed 24-h baseline personal exposure data among 798 pregnant women in the Household Air Pollution Intervention Network (HAPIN) trial in Rwanda, including 717 with fine particulate matter (PM2:5), 569 with black carbon (BC), and 716 with carbon monoxide (CO) samples. Best subsets regression identified key predictors of personal PM2:5, BC, and CO exposure, defined by maximizing adjusted R2 values and minimizing prediction errors (Mallow’s CP and the Bayesian information criterion). RESULTS: The 24-h median concentrations at baseline were 88:9 lg=m3 [interquartile range ðIQRÞ = 85:0], 10:9 lg =m3 (IQR = 7.6), and 1:12 ppm (IQR = 1.9) for PM2:5, BC, and CO, respectively. Households using kerosene as a primary lighting source had higher PM2:5 levels (median = 116 lg=m3, IQR = 107) than those using electricity (64 lg=m3, IQR = 69). Women in households with modified biomass stoves with a chimney had lower median values (48 lg=m3, IQR = 52) for PM2:5, compared with those in households using open fires (113 lg=m3, IQR = 74) and other traditional stove types (155 lg=m3, IQR = 43) that yielded the highest values. Consensus models from the best subsets’ regression explained 26% of the variation in PM2:5, 36% in BC, and 31% in CO concentrations. CONCLUSIONS: Based on a unique and large dataset describing personal exposure among pregnant women in rural Rwanda, lighting and cooking practices described some variability in household PM2:5 concentrations, but overall, substantial unexplained variability remained. https://doi.org/10.1289/JHP1049
AB - BACKGROUND: Exposure to household air pollution from the combustion of solid fuels is a leading risk factor for death and disease in low- and middle-income countries, where cleaner cooking and lighting options are often unavailable. Few studies have measured personal exposure during pregnancy, a sensitive period of development, particularly in Africa. OBJECTIVE: We aimed to characterize exposure during early to midpregnancy among women in Rwanda and to assess predictors of personal exposure, including stove and fuel type, cooking behaviors, housing conditions, sociodemographic characteristics, and other potential sources of exposure. METHODS: We assessed 24-h baseline personal exposure data among 798 pregnant women in the Household Air Pollution Intervention Network (HAPIN) trial in Rwanda, including 717 with fine particulate matter (PM2:5), 569 with black carbon (BC), and 716 with carbon monoxide (CO) samples. Best subsets regression identified key predictors of personal PM2:5, BC, and CO exposure, defined by maximizing adjusted R2 values and minimizing prediction errors (Mallow’s CP and the Bayesian information criterion). RESULTS: The 24-h median concentrations at baseline were 88:9 lg=m3 [interquartile range ðIQRÞ = 85:0], 10:9 lg =m3 (IQR = 7.6), and 1:12 ppm (IQR = 1.9) for PM2:5, BC, and CO, respectively. Households using kerosene as a primary lighting source had higher PM2:5 levels (median = 116 lg=m3, IQR = 107) than those using electricity (64 lg=m3, IQR = 69). Women in households with modified biomass stoves with a chimney had lower median values (48 lg=m3, IQR = 52) for PM2:5, compared with those in households using open fires (113 lg=m3, IQR = 74) and other traditional stove types (155 lg=m3, IQR = 43) that yielded the highest values. Consensus models from the best subsets’ regression explained 26% of the variation in PM2:5, 36% in BC, and 31% in CO concentrations. CONCLUSIONS: Based on a unique and large dataset describing personal exposure among pregnant women in rural Rwanda, lighting and cooking practices described some variability in household PM2:5 concentrations, but overall, substantial unexplained variability remained. https://doi.org/10.1289/JHP1049
U2 - 10.1289/JHP1049
DO - 10.1289/JHP1049
M3 - Article
AN - SCOPUS:85217508529
SN - 2156-9614
VL - 13
SP - 17001
EP - 17009
JO - Journal of Health and Pollution
JF - Journal of Health and Pollution
IS - 1
ER -
