Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Aug 27.
Published in final edited form as: Environ Res. 2025 Jul 2;285(Pt 1):122258. doi: 10.1016/j.envres.2025.122258

Gestational and postnatal exposures to fine particulate matter (PM2.5) and their association with acute ear infections, diarrhea, respiratory symptoms, and mortality: A longitudinal study of infants in the multicountry Household Air Pollution Intervention Network (HAPIN) trial

Brandie Banner Shackelford a, Kyle Steenland a, Miles A Kirby b, Kalpana Balakrishnan c, Marilú Chiang d, Anaité Diaz-Artiga e, John P McCracken f, Lisa M Thompson a,g, Ghislaine Rosa h, Lance A Waller i, Shirin Jabbarzadeh i, Jiantong Wang i, Ajay Pillarisetti j, Michael A Johnson k, Jennifer L Peel l, William Checkley m,n, Thomas F Clasen a, on behalf of the HAPIN Investigators
PMCID: PMC12378927  NIHMSID: NIHMS2104103  PMID: 40614849

Abstract

Exposure to household air pollution from cooking with biomass is a risk factor for infant morbidity and mortality. The Household Air Pollution Intervention Network (HAPIN) evaluated the effects of a liquefied petroleum gas (LPG) stove and fuel intervention on air pollutant exposure and health outcomes among 3,195 pregnant women (nine to 19 weeks’ gestation) and their infants in Guatemala, India, Peru, and Rwanda. We measured PM2.5 exposure among women before childbirth and infants in the postnatal period. Infant death was monitored for the entirety of the study, while data on acute ear infections, diarrhea, cough, and rapid and/or difficulty breathing were collected at four timepoints. Multivariable logistic regression models, adjusted for demographics and repeated measures within each child, estimated the odds of each health outcome based on exposures. We observed a modest but statistically significant association between an interquartile increase in gestational PM2.5 exposure and increased odds of acute ear infection (OR: 1.15; 95% CI: 1.00, 1.33), but there was no association with other measured health outcomes. We found no statistically significant association between postnatal PM2.5 exposure and any of the measured health outcomes. Our findings underscore the need for further investigation into the environmental determinants of infant health.

Keywords: Household air pollution, infant morbidity, infant mortality, PM2.5 exposure, gestational exposure

1. Introduction

In 2021, 2.3 billion people globally were reliant on polluting fuels, such as wood, coal, dung, and crop waste burned with inefficient stoves or in open fires to cook their food (The World Bank, 2023). The resulting household air pollution (HAP) is a public health threat (Murray et al., 2020), responsible for an estimated 3.9% of disability-adjusted life years globally in 2021 (Bennitt et al., 2025). Traditionally, women and young children spend extensive time at home and close to indoor cooking fires, putting them disproportionately at risk for the effects of HAP (Ali et al., 2021; Gordon et al., 2014).

Particulate matter with an aerodynamic diameter of ≤ 2.5 μm (PM2.5) is one of many byproducts of incomplete combustion. In ambient air pollution studies, PM2.5 exposure has been associated with a range of harmful health effects in children. Recent burden of disease estimates suggest that diarrheal diseases and respiratory infections — two of the leading causes of death in children under five (Bennitt et al., 2021; Lee et al., 2020) — as well as otitis media (i.e., acute ear infection) are major strains on healthcare systems worldwide and partially attributable to ambient PM2.5 pollution in young children (Sang et al., 2022). Secondhand tobacco smoke, which has high levels of PM2.5, has been linked to otitis media in children (Öberg et al., 2011), although its potential relationship with HAP from cooking has not been reported.

Despite an apparent link between PM2.5 exposure and adverse health outcomes in children, several recent randomized controlled trials (RCTs) of improved cookstoves have demonstrated limited improvements in child health outcomes (Adane et al., 2021; Hartinger et al., 2016; Kirby et al., 2019; Smith et al., 2011; Thakur et al., 2018). The lack of health improvements in these randomized studies could be due to insufficient reduction in pollutant levels caused by limited adoption and effectiveness of the cookstove intervention, low statistical power, or other factors (McCollum et al., 2024). To address the gaps of previous studies, the Household Air Pollution Intervention Network (HAPIN) conducted a multicountry RCT to assess the impact of liquefied petroleum gas (LPG) stoves and cooking fuel, the most widely scaled alternative among cleaner cooking fuels in many low-and-middle-income countries (LMICs) (Floess et al., 2023), on birth weight, infant stunting, infant pneumonia, and blood pressure of women 40–79 years old in the home (Clasen et al., 2020). Previously, we reported that birth weight, stunting risk, and incidence of severe pneumonia did not significantly differ between infants born in intervention and control households (Checkley et al., 2024; Clasen et al., 2022; McCollum et al., 2024). However, when considering the effect of personal PM2.5 exposures rather than just study arm, an interquartile increase in gestational exposure was associated with significant reductions in birth weight (Balakrishnan et al., 2023). Here we report the relationship between PM2.5 exposure and secondary infant health outcomes measured in HAPIN trial: acute ear infection, diarrhea, cough, chest-related rapid and/or difficulty breathing, and death. To understand additional drivers, we also independently examined other potential predictors of infant morbidity and mortality based on previous studies, such as water, sanitation, and hygiene (WASH) (Prüss et al., 2002), floor material (Legge et al., 2023), ownership of poultry (Zambrano et al., 2014), and vaccination (Gentile et al., 2010).

2. Methods

2.1. Trial design

The HAPIN trial took place from May 2018 to September 2021. It included four International Research Centers (IRCs) (Jalapa, Guatemala; Tamil Nadu, India; Puno, Peru; and Kayonza, Rwanda) representing diverse geographic regions. The detailed RCT protocol has been described previously (Barr et al., 2020; Clasen et al., 2020; Johnson et al., 2020). Briefly, women between the ages of 18 and 34 and between nine and 19 weeks pregnant with a single fetus confirmed by fetal ultrasound were considered for participation in the trial. Eligible participants relied primarily on biomass fuel for cooking. Women who smoked tobacco or lived in the same household as another participant in the trial were not eligible.

After collecting baseline exposure and selected demographic data, participants were assigned equally to intervention and control groups, using ten geographic randomization strata. Intervention households received unvented LPG cookstoves with a free, continuous LPG supply. Control households continued traditional cooking methods with biomass fuels and received compensation for their participation (Quinn et al., 2019). Adherence to the intervention was monitored and promoted by trial staff (Williams et al., 2023).

2.2. Data collection of participant characteristics

Household demographic data, including information related to socioeconomic status (SES), were collected pre-randomization (baseline) and when the child turned one year old. We used baseline data in the statistical models, or the one-year-old data when baseline data were missing. Data on child sex and birth weight were collected at the time of birth. Newborns were weighed within 24 hours of birth by a nurse or trained study staff member using a digital baby scale. Two measurements were taken; if they varied by more than ten grams, a third measurement was taken. If the neonate was not weighed by study staff within 24 hours, the hospital-measured birth weight was used instead. Vaccination and supplementation history data were primarily collected at the nine- and twelve-month child health visits. When vaccination and supplementation records were not available, we used caregiver reports.

2.3. Monitoring exposure

Personal PM2.5 exposure was monitored through gravimetric and nephelometric methods via RTI International’s Enhanced Children’s MicroPEM™ (ECM) sensors. During gestation, pregnant women wore ECM monitors for 24-hour periods three times: 1) at baseline, 2) between 24 and 28 weeks of gestation, and 3) between 32 and 36 weeks of gestation. For the intervention arm, gestational exposures were estimated by weighting the time before and after LPG delivery.

Infant postnatal exposure was measured over 24-hour periods at approximately three, six, and 12 months of age. Infants wore location beacons that paired with ECMs, which were strategically placed on the mother and in their most frequented environments (i.e., the primary living and sleeping areas) (Liao et al., 2020). Gravimetric filters were stored in freezers, weighed on one microgram resolution balances, and adjusted for changes in field blank weights (approximately four percent of total samples). The limit of detection was determined for each IRC individually. Flows were adjusted after comparison with a primary flow standard at each IRC’s field headquarters. ECM sample flows were logged continuously; those differing by more than 15% from the set flow rate were discarded. Accelerometer data were available to monitor wearing compliance. A subset of samples also included duplicate ECMs to evaluate consistency between monitors (Johnson et al., 2022). If the infant’s exposure data was missing or invalid, we used the mother’s exposure at that timepoint as a proxy (ρ = 0.89). In instances where the child and mother’s exposures were missing, we took an average of the proximate exposure measurements (e.g., an average of the six and 12 month visits for the nine month visit).

2.4. Monitoring health outcomes

Postnatal exposure measurements roughly corresponded with surveys about the child’s health status. Trial staff interviewed mothers, or other caregivers if the mother was unavailable, regarding infant health at approximately three, six, nine, and 12 months post-birth. Data collection was primarily conducted in person; however, some surveys were conducted via phone due to the COVID-19 pandemic and resulting shutdowns. The health outcomes were collected from standardized questionnaires: 1) “In the last seven days, since this day last week, has child gone to a health facility and received an antibiotic for acute ear pain or acute ear infection?” 2) “Since this day last week, has the child had (select all that apply): Illness with a cough; Fast, short, rapid breaths due to a problem in the chest; Difficulty breathing due to a problem in the chest; Diarrhea defined as passage of 3 or more loose stools (not solid/can take the shape of a container) within a 24-hour period; None of the above” and 3) “In the last 24 hours, has the child had diarrhea, defined as passage of three or more loose stools (not solid/ can take the shape of a container) within a 24 hour period.”

Any live-born child who died before reaching 12 months of age was recorded as a death, excluding stillbirths and miscarriages. Children who discontinued participation in the trial for reasons such as caregiver-initiated withdrawal or loss to follow-up (n = 48), and therefore could not be monitored for the entirety of their first year of life, were excluded from the mortality analyses. Mortality data were recorded via the serious adverse event survey during routine child health status visits or when a participant reported the death to the trial staff. Death certificates and medical records were referenced for data entry, including date of death, where feasible.

2.5. Statistical analyses

Analyses were conducted using SAS® version 9.4 (Cary, NC). For bivariate analyses, we conducted Mann-Whitney U tests to assess whether exposure distributions differed for children with and without the health outcomes of interest. For multivariable analyses, we used logistic regression for the infant death outcome. For the infant illness outcomes, including acute ear infection, diarrhea, cough, and rapid and/or difficulty breathing, we used generalized estimating equations (GEE) logistic regression model, assuming compound symmetry for the repeated measures within each infant. Multivariable models were developed based on existing literature. Variables were initially selected using directed acyclic graphs (DAGs) (Textor et al., 2016) (Supplementary Material: Figure S1). Final variable selection was determined through Akaike Information Criterion (AIC) for logistic regression and Quasi-likelihood under the Independence model Criterion (QIC) analyses for GEE.

To examine potential exposure-response relationships between health outcomes and PM2.5 exposure, we studied the association with gestational and postnatal exposures independently. We also modeled gestational and postnatal PM2.5 exposure together for the morbidity outcomes. We explored two continuous models of exposure (linear and log-linear) for each outcome and evaluated whether they had a positive significant trend. Quartile models of exposure were additionally used to account for data skewness and identify potential threshold effects. Models examining the association between PM2.5 exposure and infant morbidity were adjusted for SES, age, sex, maternal nulliparity, and IRC (Supplementary Material: Analytic Details). The SES variable was derived from a principal component analysis (Checkley et al., 2024) that included housing materials, household assets, water and sanitation facilities, access to electricity, household food insecurity (Ballard et al., 2013), maternal education, and household size. The models for infant mortality and PM2.5 exposure included covariates for SES, sex, maternal nulliparity, and IRC. Subgroup analyses were conducted by variables of interest such as IRC, SES, sex, and birth weight. We also examined two-way statistical interactions between natural log-transformed PM2.5 exposure and key variables to understand potential effect measure modification. For the models with significant results, we conducted a sensitivity analysis using models that included a variable for the presence of a tobacco smoker in the household to further explore the contribution of cooking-related HAP.

To understand other potential influences on infant health outcomes, we independently examined the association of basic WASH services, improved floor material, no ownership of poultry (i.e., chickens, ducks, or turkeys), household size, at least one dose of various vaccinations, at least one dose of vitamin A supplementation, birth weight greater than or equal to 2500 grams, SES, and no tobacco smokers in the household on infant morbidity and mortality. We applied the World Health Organization (WHO) and United Nations Children’s Fund (UNICEF) Joint Monitoring Programme definitions for the WASH variables, namely at least basic drinking water service (i.e., using an improved water source within 30 minutes round trip including queuing), improved sanitation (e.g., pit latrines with slabs, septic tanks, flush/ pour flush toilets), and basic hygiene services (i.e., a place for handwashing with soap and water) (UNICEF and WHO, 2018). We examined the association of having at least one dose of the Bacille Calmette-Guérin (BCG) vaccine, the Pentavalent or Diphtheria, Tetanus, and Pertussis (DTP) vaccine, the Rotavirus vaccine, the Pneumococcal vaccine, the flu vaccine, and vitamin A supplementation on the outcomes they aim to prevent; this was not done for mortality, as vaccination data were primarily collected after infant deaths had occurred. To examine the association of these other predictors, we used the same models in the PM2.5 exposure-response analysis; however, we used maternal education as a proxy for SES as some of the hypothesized predicator variables were included in the derived SES variable.

To adjust for multiple comparisons in analyses with at least ten tests, we applied the Benjamini-Hochberg procedure to control the false discovery rate at 0.2. We considered results significant if the unadjusted p-value was less than its Benjamini-Hochberg critical value, which was determined based on the number of comparisons in each analysis (i.e., the product of the number of exposures, outcomes, and subgroups, if applicable) (Benjamini & Hochberg, 1995).

2.6. Ethics and trial registration

The study protocol was reviewed and approved by institutional review boards or Ethics Committees at Emory University (00089799), Johns Hopkins University (00007403), Sri Ramachandra Institute of Higher Education and Research (IEC-N1/16/JUL/54/49) and the Indian Council of Medical Research – Health Ministry Screening Committee (5/8/4–30/(Env)/Indo-US/2016-NCD-I), Universidad del Valle de Guatemala (146–08-2016) and Guatemalan Ministry of Health National Ethics Committee (11–2016), Asociación Benefica PRISMA (CE 2981.17), the London School of Hygiene and Tropical Medicine (11664–5) and the Rwandan National Ethics Committee (No.357/RNEC/2018), and Washington University in St. Louis (201611159). The study was registered at ClinicalTrials.gov (NCT02944682).

3. Results

3.1. Participant and household characteristics

Of 3,061 live births among study participants, data for this study were available for 3,036 infants (99.2%), including 2,982 children with at least one health status report and 64 children who died (10 of whom had at least one health status report prior to death). Selected participant and household characteristics are summarized in Table 1. There were more male children than female children in the HAPIN trial. The average birth weight was 2.9 kilograms, with 530 (17.5%) infants meeting the definition of low birth weight (<2.5 kilograms). Thirty-eight percent of mothers were nulliparous at baseline and less than a third (32.8%) completed secondary school.

Table 1.

Characteristics of infants and households in the HAPIN trial with child health status or mortality reports (n= 3036)*

Infant Characteristics
Sex, % (n)
 Male 51.8% (1573)
 Female 48.2% (1463)
 Missing 0.0% (0)
Birth weight in grams
 Mean ± SD 2909.6 ± 469.8
 Range 1000.0–6145.0
 Low birth weight (<2500g), % (n) 17.5% (530)
 Missing 1.4% (42)
Maternal nulliparity, % (n) 38.0% (1153)
 Missing 0.2% (6)
Maternal educational attainment, % (n)
 No formal education or primary school incomplete 32.9% (999)
 Primary school complete or secondary school incomplete 34.3% (1040)
 Secondary school complete or vocational or some college or university 32.8% (997)
 Missing 0.0% (0)
At least one dose of vaccination or supplement by 12 months, % (n)
Bacille Calmette-Guérin 89.7% (2722)
 Missing 7.5% (229)
Pentavalent or Diphtheria, Tetanus, and Pertussis1 89.4% (2715)
 Missing 7.5% (229)
Rotavirus 88.7% (2693)
 Missing 7.5% (228)
Pneumococcal 67.8% (2058)
 Missing 7.5% (229)
Flu 24.5% (744)
 Missing 13.5% (409)
Vitamin A 32.8% (997)
 Missing 13.6% (412)
Household Characteristics
Household size excluding infant
 Mean ± SD 4.3 ± 2.0
 Range 1–18
 Missing 0.0% (0)
Socioeconomic status index
 Mean ± SD 0.0 ± 1.0
 Range -2.2–2.1
 Missing 0.0% (0)
At least one tobacco smoker in household, % (n) 10.5% (319)
 Missing 0.1% (3)
Household water, sanitation, and hygiene status, % (n)
At least basic water service 72.3% (2196)
 Missing 0.03% (1)
Improved sanitation 59.9% (1819)
 Missing 0.0% (0)
At least basic hygiene 64.9% (1971)
 Missing 0.0% (0)
At least one improved floor material, % (n) 46.0% (1397)
 Missing 0.0% (0)
Owns poultry, % (n) 50.0% (1517)
 Missing 0.0% (0)
Open in a new tab

Note:

1

In Rwanda, the DTP vaccine was administered. In the other countries, children received the Pentavalent vaccine.

Most infants in the trial received the BCG vaccine (89.7%), at least one dose of a vaccination for Diphtheria, Tetanus, and Pertussis (89.4%), and at least one dose of the rotavirus vaccine (88.7%). Over two-thirds of infants (67.8%) also received at least one dose of the pneumococcal vaccination. Less than a third of infants received at least one vitamin A supplementation (32.8%) or the flu vaccine (24.5%). Neither pneumococcal nor flu vaccines were part of the standard vaccination schedule in India, and flu vaccination was not included on the schedule in Rwanda.

The average household size in the trial was 4.3 people upon enrollment. There was at least one tobacco smoker in 10.5% of households. Most households in the trial (72.3%) reported having at least basic water service, access to an improved toilet (59.9%), and basic hygiene services (64.9%). Less than half of the households (46.0%) in the trial had at least one improved floor material (e.g., brick, tile, stone, concrete, wood). Half of the households in the trial owned poultry.

3.2. PM2.5 exposure

The median weighted gestational exposure to PM2.5 was approximately double the WHO’s Interim Target 1 guideline value for annual PM2.5 exposure (35 μg/m3 annual mean) (Table 2). Median estimated postnatal PM2.5 exposures substantially decreased from the gestational period and were near the WHO Interim Target 1 at every timepoint. While nearly identical at baseline, the intervention arm had substantially lower exposures than the control arm for all subsequent timepoints (Johnson et al., 2022; Pillarisetti et al., 2023). Quartiles of PM2.5 exposure and distributions by IRC are available in the Supplementary Material (Table S1; Figure S2; Figure S3).

Table 2.

Summary of weighted gestational and postnatal PM2.5 exposures (μg/m3) with child health status or mortality reports (n= 3036)

Weighted gestational exposure Estimated exposure at 3 months Estimated exposure at 6 months Estimated exposure at 9 months Estimated exposure at 12 months
Mean ± SD 92.3 ± 83.5 62.5 ± 97.9 63.9 ± 103.5 68.3 ± 102.6 70.0 ± 117.6
Median (IQR) 69.9 (74.7) 32.8 (48.5) 31.7 (44.7) 35.4 (48.5) 33.1 (51.0)
N 2752 2760 2747 2602 2866
Open in a new tab

3.3. Health outcomes

Sixty-four children (2.09% of live births) died before their first birthday (Supplementary Material: Figure S4). Over one-third of the deaths (n=24) occurred within the first five days of life. More infants died in Guatemala (n=20) than in the other countries (Supplementary Material: Table S2). Previously we reported the leading causes of death were prematurity/low birth weight and birth asphyxia for neonates and pneumonia/acute respiratory failure and diarrheal diseases for infants (McCollum et al., 2024).

A total of 11,377 child health status surveys were conducted over the study period, representing 2,982 infants (97.4% of live births). More child health surveys were completed in India (n = 2,999) than in Guatemala (n = 2,923), Peru (n=2,731), or Rwanda (n=2,724). Over the 12-month follow-up period, the most reported health outcomes were cough (11.3%, n=1,287) and diarrhea (4.64%, n=528) in the week prior to the survey (Supplementary Material: Table S3). Diarrhea in the week prior to the survey was reported at the nine- and 12-month visits at almost double the rate of the three- and six-month visits; however, this trend did not hold for diarrhea in the day prior to the survey (Table 3).

Table 3.

Reported infant morbidities among 11,377 child health status reports during the HAPIN trial by timepoint, n (%)

Outcome 3 month visit 6 month visit 9 month visit 12 month visit
Acute ear infection 19 (0.67%) 27 (0.96%) 12 (0.42%) 22 (0.76%)
Diarrhea in week prior 77 (2.72%) 99 (3.52%) 179 (6.30%) 173 (6.00%)
Diarrhea in 24 hours prior 34 (1.20%) 33 (1.17%) 40 (1.41%) 49 (1.70%)
Cough in week prior 352 (12.4%) 375 (13.3%) 322 (11.3%) 238 (8.25%)
Rapid and/or difficulty breathing in week prior 40 (1.41%) 32 (1.14%) 35 (1.23%) 9 (0.31%)
Total child health status reports 2834 2816 2842 2885
Open in a new tab

Despite having the fewest child health surveys overall, respiratory symptoms and diarrhea were most reported among Rwandan infants, including 70.8% of total reports of rapid breathing, 58.0% of total reports of cough, 55.2% of total reports of difficulty breathing, and 42.2% of total reports of diarrhea (Supplementary Material: Table S3). Acute ear infections were most reported by caregivers of Guatemalan infants (41.3%).

3.4. Associations between PM2.5 exposure and health outcomes

Bivariate analyses indicated that median gestational PM2.5 exposures were consistently higher for children with any of the health outcomes of interest than those without (Supplemental Material: Table S4). Children who experienced acute ear infection (p = 0.050), diarrhea (p = 0.0090), or cough (p < 0.001) in the week prior had significantly higher gestational PM2.5 exposures compared to those without the health outcomes. Additionally, children aged three, six, and nine months with cough in the preceding week had significantly higher postnatal PM2.5 exposures (p < 0.001) than those who did not.

The results of the multivariate analyses, which used logistic regression models to examine the effect of gestational and postnatal PM2.5 exposure separately, are shown in Table 4. We observed a significant association between linear gestational PM2.5 exposure and acute ear infections, with 1.15 (95% CI: 1.00, 1.33) higher odds for an interquartile increase (75.37 μg/m3) in gestational exposure. There was also a positive increasing trend in the quartile models of exposure. In a sensitivity analysis, the statistically significant effect persisted in models that further adjusted for the presence of a tobacco smoker in the household (p = 0.041). However, the significant finding did not hold in a model that included both gestational and postnatal PM2.5 exposure (Supplementary Material: Table S5). In that categorical model, infants in the second quartile of postnatal exposure had significantly higher odds of acute ear infection than those in the first quartile.

Table 4.

Odds ratios for infant health outcomes and PM2.5 exposure for infants with child health status or mortality reports in the HAPIN trial, with separate models for gestational and postnatal exposure*

Weighted gestational exposure to PM2.5 Postnatal exposure to PM2.5
Outcome Parameter** Odds ratio (95% CI) p n AIC/QIC Odds ratio (95% CI) p n AIC/QIC
Death Linear 0.999 (0.996, 1.003) 0.72 2728 583.0 Not modeled***
Log-linear 1.05 (0.73, 1.51) 0.81 583.1
Quartile 2 2.75 (1.18, 6.44) 0.019 579.6
Quartile 3 1.35 (0.53, 3.45) 0.53
Quartile 4 1.99 (0.81, 4.87) 0.13
Acute ear infection in week prior Linear 1.002 (1.000, 1.004) 0.048 10287 841.3 1.001 (0.999, 1.003) 0.40 10944 951.4
Log-linear 1.36 (0.98, 1.89) 0.066 840.9 1.07 (0.85, 1.36) 0.55 951.3
Quartile 2 1.71 (0.81, 3.62) 0.16 845.0 1.75 (0.90, 3.43) 0.10 952.8
Quartile 3 1.76 (0.85, 3.67) 0.13 1.53 (0.73, 3.19) 0.26
Quartile 4 1.95 (0.94, 4.04) 0.074 1.40 (0.69, 2.83) 0.35
Diarrhea in week prior Linear 0.999 (0.998, 1.001) 0.61 10292 3677.8 1.000 (0.999, 1.001) 0.22 10950 3937.0
Log-linear 1.04 (0.90, 1.19) 0.60 3678.0 1.05 (0.94, 1.18) 0.36 3937.8
Quartile 2 1.20 (0.89, 1.63) 0.23 3680.9 0.87 (0.64, 1.19) 0.39 3941.0
Quartile 3 1.19 (0.88, 1.60) 0.26 0.99 (0.72, 1.36) 0.94
Quartile 4 1.14 (0.83, 1.57) 0.41 1.05 (0.77, 1.45) 0.75
Diarrhea in 24 hours prior Linear 0.999 (0.997, 1.001) 0.27 10292 1512.5 0.999 (0.998, 1.001) 0.81 10950 1569.9
Log-linear 0.93 (0.73, 1.18) 0.54 1513.5 0.96 (0.77, 1.18) 0.68 1570.0
Quartile 2 1.18 (0.69, 1.99) 0.55 1516.9 1.21 (0.66, 2.20) 0.54 1571.7
Quartile 3 1.27 (0.76, 2.11) 0.36 0.84 (0.44, 1.61) 0.61
Quartile 4 0.99 (0.56, 1.76) 0.98 1.11 (0.59, 2.06) 0.75
Cough in week prior Linear 1.000 (0.999, 1.001) 0.97 10292 6501.9 1.000 (0.999, 1.001) 0.73 10950 6864.6
Log-linear 1.07 (0.96, 1.18) 0.21 6500.3 1.02 (0.95, 1.11) 0.58 6864.7
Quartile 2 1.38 (1.10, 1.72) 0.0049 6489.9 0.97 (0.78, 1.21) 0.79 6868.1
Quartile 3 1.48 (1.18, 1.86) <0.001 1.03 (0.83, 1.28) 0.79
Quartile 4 1.19 (0.95, 1.50) 0.13 1.07 (0.86, 1.34) 0.55
Rapid and/or difficulty breathing in week prior Linear 0.999 (0.995, 1.004) 0.68 10292 1112.0 1.000 (0.998, 1.003) 0.77 10950 1145.6
Log-linear 0.88 (0.66, 1.18) 0.39 1108.7 1.01 (0.78, 1.30) 0.93 1144.5
Quartile 2 1.00 (0.55, 1.82) 0.99 1106.6 1.02 (0.52, 2.01) 0.95 1148.1
Quartile 3 1.26 (0.71, 2.25) 0.43 0.91 (0.52, 2.01) 0.78
Quartile 4 0.58 (0.29, 1.18) 0.13 1.04 (0.51, 2.10) 0.91
Open in a new tab
*

Mortality was adjusted for socioeconomic status, sex, maternal nulliparity, and IRC; morbidity was adjusted for socioeconomic status, sex, maternal nulliparity, IRC, and age

**

Quartile 1 used as reference for all other quartiles

***

Postnatal exposure to PM2.5 was not modeled since most infant deaths occurred before measurement - only 12 infants who died had at least one postnatal measurement

There were significant associations observed for quartile models of gestational PM2.5 exposure and infant death and cough, but not in the continuous models (i.e., linear and log-linear) (Table 4). Infants in the second quartile of gestational PM2.5 exposure had significantly higher odds of death than those in the lowest quartile. Further, infants in the second and third quartile of gestational PM2.5 exposure had significantly higher odds of cough in the week prior than those in the first, which remained significant in the model that further adjusted for postnatal exposure (Supplementary Material: Table S5).

No significant relationships were observed between gestational PM2.5 exposure and diarrhea or rapid and/or difficulty breathing. Further, there was no significant association observed between any form of postnatal PM2.5 exposure and any of the health outcomes in models that excluded gestational exposure.

Subgroup analyses by IRC indicated significant associations between natural log-transformed PM2.5 exposure and the health outcomes. In India, a one-unit increase in natural log-transformed gestational and postnatal PM2.5 exposure was significantly associated with 2.34 (95% CI: 1.38, 3.97) and 1.82 (95% CI: 1.15, 2.88) higher odds, respectively, of acute ear infection the week prior to the survey (Supplementary Material: Table S6). To explore this further we ran models for acute ear infection using data from all four IRCs, including interaction terms between IRC and PM2.5 exposure. We found synergistic interactions for India (with Rwanda as the reference) for both gestational PM2.5 exposure (p = 0.0034) and postnatal PM2.5 exposure (p = 0.0046). However, the main effect of PM2.5 exposure was no longer significant, suggesting that IRC modifies the effect of both gestational and postnatal PM2.5 exposure on acute ear infection.

In Peru, a one-unit increase in natural log-transformed gestational PM2.5 exposure was associated with significantly higher odds (OR: 1.52; 95% CI: 1.10, 2.11) of rapid and/or difficulty breathing (Supplementary Material: Table S6). Including an interaction term between gestational PM2.5 exposure and IRC revealed a significant synergistic interaction for Peru compared to Rwanda (p = 0.024), while the main effect of PM2.5 exposure became non-significant, suggesting effect measure modification. Subgroup analyses by IRC also showed significantly higher odds of cough with increased natural log-transformed gestational PM2.5 exposure in Guatemalan infants; however, an interaction term between gestational PM2.5 and Guatemala was not significant in models including all children.

After controlling the false discovery rate using the Benjamini-Hochberg correction, no significant associations were found within subgroups defined by SES, sex, birth weight, the approximate COVID-19 pandemic shutdown periods, the method of collecting the child’s health status report, or household hygiene facilities (Supplementary Material: Tables S7-S12). However, no caregiver reported rapid and/or difficulty breathing in the 1,416 child health surveys conducted via the telephone, so it could not be modeled (Supplementary Material: Table S11).

Gestational and postnatal PM2.5 exposure was significantly associated with increased odds of cough among children who had received at least one dose of the flu vaccine, and gestational PM2.5 exposure was similarly associated with increased odds of cough among those who received at least one dose of vitamin A supplementation (Supplementary Material: Tables S13–14). In analyses stratified by having at least one dose of the DTP vaccine, BCG, and rotavirus vaccines, no statistically significant results were observed, although some subgroups did not converge (Supplementary Material: Tables S15–17).

3.5. Other significant predictors of health outcomes

Supplementary Table S18 presents the associations between the health outcomes of interest and hypothesized protective factors for each outcome, modeled independently of PM2.5 exposure. After adjusting the significance thresholds to account for multiple comparisons, we identified several factors that were protective against infant morbidity and mortality. Higher household SES was protective against infant death and cough. A one-unit increase in the SES index was associated with 0.50 (95% CI: 0.30, 0.82) the odds of infant death and 0.84 (95% CI: 0.74, 0.95) the odds of infant cough in the week prior to the survey. Infants with normal birth weight had 0.13 (95% CI: 0.072, 0.24) the odds of death than those with low birth weight, although birth weight data were not available for eight of the children who died. Receiving one dose of vitamin A supplementation was associated with significantly lower odds of diarrhea (OR: 0.65; 95% CI: 0.50, 0.86) in the week prior to the survey, while one dose of the pneumococcal vaccine was associated with significantly lower odds (OR: 0.29; 95% CI: 0.11, 0.74) of experiencing rapid and/or difficulty breathing during the same period.

4. Discussion

4.1. Key findings and contextualization with existing evidence

More than two percent of infants in the HAPIN trial died before 12 months of age. The infant mortality rates (IMRs) for children in the trial were higher than the estimated country IMRs (United Nations Inter-agency Group for Child Mortality Estimation, 2024) in Guatemala and Peru, but lower in India and Rwanda (Supplementary Material: Table S19). Although quartile models indicated the odds of infant death were significantly higher for children in the second quartile of gestational PM2.5 exposure than the first, this association appeared non-monotonic, as it was not observed in higher quartiles or in continuous exposure models. Significant protective factors against infant death included the household having a higher SES index and healthy birth weight.

A significant relationship between tobacco smoke and otitis media is thoroughly documented in other research, but the impact of ambient PM2.5 on child ear infections is seemingly inconsistent (Bowatte et al., 2018). Our linear model for PM2.5 exposure and acute ear infections treated with antibiotics in the week prior to the survey was significant, with a positive trend in the quartile models, suggesting that HAP from cooking may increase the risk of ear infection in infants. We also found evidence of effect measure modification in the relationship between both gestational and postnatal PM2.5 and acute ear infections, with significant statistical interactions between natural log-transformed PM2.5 and India. This may suggest contextual factors, possibly healthcare access, specific aspects of cookstove design, or cooking fuel types, influence how strongly PM2.5 affects infant ear infections.

Recent research has suggested a link between PM2.5 exposure and diarrhea in young children (Z. Zhang et al., 2025), possibly via inflammation of the digestive tract (Xie et al., 2022; K. Zhang, 2024). Using multivariable models and subgroup analyses, we consistently found no significant association between PM2.5 exposure and the occurrence of diarrhea in the day or week preceding the survey. However, we found that vitamin A supplementation was significantly protective against diarrhea the week prior to the survey, which aligns with existing literature on its effects in children under five years old (Mayo-Wilson et al., 2011).

We previously reported that there was no significant difference detected in the incidence of infant pneumonia between intervention and control households (McCollum et al., 2024), which is largely congruent with other major HAP trials that measured PM2.5 exposure and lower respiratory infections or pneumonia (Adane et al., 2021; Hartinger et al., 2016; Kirby et al., 2019). Our study found significant associations with the second and third quartile of gestational PM2.5 exposure and cough, which may be a precursor to pneumonia or indicative of other respiratory infections, but this significant result was not observed in continuous models of exposure. Rapid and/or difficulty breathing was not commonly reported in this study and was notably not reported in any telephone survey during the COVID-19 pandemic. Higher household socioeconomic status and having at least one dose of the pneumococcal vaccine were found to be protective against cough and rapid and/or difficulty breathing in the week prior to the survey, respectively.

4.2. Strengths

The HAPIN trial demonstrated high intervention fidelity and adherence (Quinn et al., 2021), which significantly reduced average PM2.5 exposure for intervention households compared to controls (Johnson et al., 2022; Pillarisetti et al., 2023). The trial provided a geographically diverse longitudinal data set for exploring the relationship between personal PM2.5 exposures and infant health, including mortality. Our study comprehensively examined the effect of PM2.5 exposure on infant morbidity and mortality, using various statistical transformations (i.e., linear, log-linear, categorical) and extensive subgroup analyses.

4.3. Limitations and future study directions

A limitation of the HAPIN trial is that many intervention households did not receive LPG until the end of the second trimester (Johnson et al., 2022), so future studies could examine the impact of earlier reductions in exposure. Other factors may have influenced the outcomes but were not included in our study. For example, the intervention was economic in part since there was free fuel provision and the time-saving and economic benefits could have impacted our outcomes. However, each IRC aimed to provide contextually appropriate, equivalent compensation for the control households (Quinn et al., 2019).

Even though we utilized both vaccination records and caregiver reports, a considerable number of infants were missing vaccination data. We could only assess whether infants received at least one dose of vaccinations and supplementation because we used the caregiver reports; however, all except BCG are recommended to be given in multiple doses. This may have reduced the potential to detect an effect of the vaccinations.

About one-third (32.7%) of postnatal PM2.5 exposure measurements were conducted within 24 hours of the child health status report. Therefore, a limitation of our study is the possibility of exposure misclassification if measured postnatal PM2.5 exposure was not representative of the infants’ exposure prior to the health outcome. This was particularly a risk for the nine month visit since there was no direct exposure measurement at that time. Future studies could consider the impact of other HAP agents on the outcomes or the impact of different cookstove designs on exposure.

Another limitation of our study is that the outcomes were based on caregiver report. For example, since infants often do not show obvious symptoms of acute ear infections and are unable to report it themselves, detection is likely dependent on maternal health education. Consequently, untreated ear infections may have been missed and the use of antibiotics could have influenced our findings for other health outcomes. Therefore, future studies could consider active examinations by medical professionals to monitor infant health outcomes. The COVID-19 pandemic may have impacted the reporting and development of the infant morbidity outcomes. Telephone surveys were conducted during peak lockdowns, but differential outcome misclassification could have occurred if any child health status visits were rescheduled due to household illness (or other factors like heavy rains, which could have impacted the outcomes).

4.4. Public health implications

This study represents one of the largest RCTs to examine associations between personal exposure to fine particulate and infant health outcomes, providing insights from four LMICs. We did not identify a singular determinant of infant morbidity and mortality in the HAPIN trial. Instead, interventions addressing multiple risk factors — such as environmental exposures, household SES, supplementation, vaccination, and nutrition — could be most effective.

Supplementary Material

1

Highlights.

  • The HAPIN trial assessed the effect of a liquefied petroleum gas intervention

  • We explored the relationship between PM2.5 exposure and infant health in 4 countries

  • Gestational PM2.5 exposure was associated with higher odds of acute ear infection

Acknowledgments

The HAPIN trial was funded by the U.S. National Institutes of Health (cooperative agreement 1UM1HL134590) in collaboration with the Bill & Melinda Gates Foundation [OPP1131279].

The investigators would like to thank the members of the advisory committee – Drs. Patrick Breysse, Donna Spiegelman, and Joel Kaufman - for their valuable insight and guidance throughout the implementation of the trial. We also wish to acknowledge all research staff and study participants for their dedication to and participation in this important trial.

A multidisciplinary, independent Data and Safety Monitoring Board (DSMB) appointed by the National Heart, Lung, and Blood Institute (NHLBI) monitored the quality of the data and protected the safety of patients enrolled in the HAPIN trial. The DSMB consisted of: Catherine Karr (Chair), Nancy R. Cook, Stephen Hecht, Joseph Millum, Nalini Sathiakumar (deceased), Paul K. Whelton, and Gail Weinmann and Thomas Croxton (Executive Secretaries). Program Coordination: Gail Rodgers, Bill & Melinda Gates Foundation; Claudia L. Thompson, National Institute of Environmental Health Sciences; Mark J. Parascandola, National Cancer Institute; Marion Koso-Thomas, Eunice Kennedy Shriver National Institute of Child Health and Human Development; Joshua P. Rosenthal, Fogarty International Center; Concepcion R. Nierras, NIH Office of Strategic Coordination – The Common Fund; Katherine Kavounis, Dong-Yun Kim, Barry S. Schmetter (deceased), and Antonello Punturieri, NHLBI.

This research represents the NIH’s contribution to the Global Alliance for Chronic Diseases (GACD) coordinated call for research on prevention and management of chronic lung diseases for 2016.

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the US National Institutes of Health or Department of Health and Human Services.

Additional Acknowledgements

Research reported in this publication was supported in part by the National Heart, Lung, and Blood Institute of the National Institutes of Health (NIH) under Award Number 1UM1HL134590. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. It is subject to the NIH Public Access Policy. Through acceptance of this federal funding, NIH has been given a right to make this manuscript publicly available in PubMed Central upon the Official Date of Publication, as defined by NIH.

This work was supported, in whole or in part, by the Gates Foundation [OPP1131279]. The conclusions and opinions expressed in this work are those of the author(s) alone and shall not be attributed to the Foundation. Under the grant conditions of the Foundation, a Creative Commons Attribution 4.0 License has already been assigned to the Author Accepted Manuscript version that might arise from this submission. Please note works submitted as a preprint have not undergone a peer review process.

Footnotes

1

HAPIN Investigators: Vigneswari Aravindalochanan, Gloriose Bankundiye, Dana Boyd Barr, Vanessa Burrowes, Alejandra Bussalleu, Devan Campbell, Eduardo Canuz, Adly Castañaza, Howard H. Chang, Yunyun Chen, Maggie L. Clark, Carmen Lucia Contreras, Rachel Craik, Victor G. Davila-Roman, Lisa de las Fuentes, Oscar De León, Priya D’Souza, Ephrem Dusabimana, Lisa Elon, Juan Gabriel Espinoza, Irma Sayury Pineda Fuentes, Sarada S. Garg, Ahana Ghosh, Dina Goodman-Palmer, Savannah Gupton, Sarah Hamid, Stella M. Hartinger, Steven A. Harvey, Mayari Hengstermann, Ian Hennessee, Phabiola Herrera, Shakir Hossen, Marjorie Howard, Penelope P. Howards, Katherine Kearns, Jacob Kremer, Margaret A. Laws, Grace Lee, Pattie Lenzen, Jiawen Liao, Amy E. Lovvorn, Jane Mbabazi, Eric D. McCollum, Julia N. McPeek, Rachel Meyers, J. Jaime Miranda, Erick Mollinedo, Libny Monroy, Alexie Mukeshimana, Krishnendu Mukhopadhyay, Moses Mutabazi, Bernard Mutariyani, Luke P. Naeher, Durairaj Natesan, Florien Ndagijimana, Laura Nicolaou, Azhar Nizam, Jean de Dieu Ntivuguruzwa, Parinya Panuwet, Aris T. Papageorghiou, Ricardo Piedrahita, Naveen Puttaswamy, Elisa Puzzolo, Karthikeyan Dharmapuri Rajamani, Sarah Rajkumar, Usha Ramakrishnan, Rengaraj Ramasami, Alexander Ramirez, Joshua Rosenthal, P. Barry Ryan, Sudhakar Saidam, Sankar Sambandam, Suzanne Simkovich, Sheela S. Sinharoy, Kirk R. Smith, Damien Swearing, Gurusamy Thangavel, Ashley Toenjes, Lindsay J. Underhill, Viviane Valdes, Amit Verma, Megan Warnock, Kendra N. Williams, Wenlu Ye, Bonnie N. Young, Ashley Younger

Data availability statement

Deidentified data associated with these analyses will be available through Emory Dataverse, linked to the digital object identifier (DOI) of this publication.

References

  1. Adane MM, Alene GD, Mereta ST, & Wanyonyi KL (2021). Effect of improved cookstove intervention on childhood acute lower respiratory infection in Northwest Ethiopia: A cluster-randomized controlled trial. BMC Pediatrics, 21, 4. 10.1186/s12887-020-02459-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ali MU, Yu Y, Yousaf B, Munir MAM, Ullah S, Zheng C, Kuang X, & Wong MH (2021). Health impacts of indoor air pollution from household solid fuel on children and women. Journal of Hazardous Materials, 416, 126127. 10.1016/j.jhazmat.2021.126127 [DOI] [PubMed] [Google Scholar]
  3. Balakrishnan K, Steenland K, Clasen T, Chang H, Johnson M, Pillarisetti A, Ye W, Naeher LP, Diaz-Artiga A, McCracken JP, Thompson LM, Rosa G, Kirby MA, Thangavel G, Sambandam S, Mukhopadhyay K, Puttaswamy N, Aravindalochanan V, Garg S, … Peel JL (2023). Exposure–response relationships for personal exposure to fine particulate matter (PM2·5), carbon monoxide, and black carbon and birthweight: An observational analysis of the multicountry Household Air Pollution Intervention Network (HAPIN) trial. The Lancet. Planetary Health, 7(5), e387–e396. 10.1016/S2542-5196(23)00052-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Ballard TJ, Kepple AW, & Cafiero C (2013). The food insecurity experience scale: Developing a global standard for monitoring hunger worldwide. FAO. http://www.fao.org/economic/ess/ess-fs/voices/en/ [Google Scholar]
  5. Barr DB, Puttaswamy N, Jaacks LM, Steenland K, Rajkumar S, Gupton S, Ryan PB, Balakrishnan K, Peel JL, Checkley W, Clasen T, Clark ML, & (HAPIN Investigative Team). (2020). Design and Rationale of the Biomarker Center of the Household Air Pollution Intervention Network (HAPIN) Trial. Environmental Health Perspectives, 128(4), 47010. 10.1289/EHP5751 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Benjamini Y, & Hochberg Y (1995). Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society: Series B (Methodological), 57(1), 289–300. 10.1111/j.2517-6161.1995.tb02031.x [DOI] [Google Scholar]
  7. Bennitt FB, Wozniak S, Causey K, Spearman S, Okereke C, Garcia V, Hashmeh N, Ashbaugh C, Abdelkader A, Abdoun M, Abdurebi MJ, Abedi A, Zuñiga RAA, Aboagye RG, Abubakar B, Abu-Zaid A, Adane MM, Adegboye OA, Adekanmbi V, … Burkart K (2025). Global, regional, and national burden of household air pollution, 1990–2021: A systematic analysis for the Global Burden of Disease Study 2021. The Lancet, 405(10485), 1167–1181. 10.1016/S0140-6736(24)02840-X [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bennitt FB, Wozniak SS, Causey K, Burkart K, & Brauer M (2021). Estimating disease burden attributable to household air pollution: New methods within the Global Burden of Disease Study. The Lancet Global Health, 9, S18. 10.1016/S2214-109X(21)00126-1 [DOI] [Google Scholar]
  9. Bowatte G, Tham R, Perret JL, Bloom MS, Dong G, Waidyatillake N, Bui D, Morgan GG, Jalaludin B, Lodge CJ, & Dharmage SC (2018). Air Pollution and Otitis Media in Children: A Systematic Review of Literature. International Journal of Environmental Research and Public Health, 15(2), 257. 10.3390/ijerph15020257 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Checkley W, Thompson LM, Sinharoy SS, Hossen S, Moulton LH, Chang HH, Waller L, Steenland K, Rosa G, Mukeshimana A, Ndagijimana F, McCracken JP, Díaz-Artiga A, Balakrishnan K, Garg SS, Thangavel G, Aravindalochanan V, Hartinger SM, Chiang M, … Clasen TF (2024). Effects of Cooking with Liquefied Petroleum Gas or Biomass on Stunting in Infants. New England Journal of Medicine, 390(1), 44–54. 10.1056/NEJMoa2302687 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Clasen T, Chang HH, Thompson LM, Kirby MA, Balakrishnan K, Díaz-Artiga A, McCracken JP, Rosa G, Steenland K, Younger A, Aravindalochanan V, Barr DB, Castañaza A, Chen Y, Chiang M, Clark ML, Garg S, Hartinger S, Jabbarzadeh S, … Peel JL (2022). Liquefied Petroleum Gas or Biomass for Cooking and Effects on Birth Weight. New England Journal of Medicine, 387(19), 1735–1746. 10.1056/NEJMoa2206734 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Clasen T, Checkley W, Peel JL, Balakrishnan K, McCracken JP, Rosa G, Thompson LM, Barr DB, Clark ML, Johnson MA, Waller LA, Jaacks LM, Steenland K, Miranda JJ, Chang HH, Kim D-Y, McCollum ED, Davila-Roman VG, Papageorghiou A, … HAPIN Investigators. (2020). Design and Rationale of the HAPIN Study: A Multicountry Randomized Controlled Trial to Assess the Effect of Liquefied Petroleum Gas Stove and Continuous Fuel Distribution. Environmental Health Perspectives, 128(4), 047008. 10.1289/EHP6407 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Floess E, Grieshop A, Puzzolo E, Pope D, Leach N, Smith CJ, Gill-Wiehl A, Landesman K, & Bailis R (2023). Scaling up gas and electric cooking in low- and middle-income countries: Climate threat or mitigation strategy with co-benefits? Environmental Research Letters, 18(3), 034010. 10.1088/1748-9326/acb501 [DOI] [Google Scholar]
  14. Gentile A, Bhutta Z, Bravo L, Samy AG, Garcia RDJ, Hoosen A, Islam T, Karimi A, Salem M, Simasathien S, Sohail A, Watanaveeradej V, Wiedenmayer K, & Schmitt H-J (2010). Pediatric disease burden and vaccination recommendations: Understanding local differences. International Journal of Infectious Diseases, 14(8), e649–e658. 10.1016/j.ijid.2009.11.006 [DOI] [PubMed] [Google Scholar]
  15. Gordon SB, Bruce NG, Grigg J, Hibberd PL, Kurmi OP, Lam KH, Mortimer K, Asante KP, Balakrishnan K, Balmes J, Bar-Zeev N, Bates MN, Breysse PN, Buist S, Chen Z, Havens D, Jack D, Jindal S, Kan H, … Martin WJ (2014). Respiratory risks from household air pollution in low and middle income countries. The Lancet Respiratory Medicine, 2(10), 823–860. 10.1016/S2213-2600(14)70168-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hartinger SM, Lanata CF, Hattendorf J, Verastegui H, Gil AI, Wolf J, & Mäusezahl D (2016). Improving household air, drinking water and hygiene in rural Peru: A community-randomized-controlled trial of an integrated environmental home-based intervention package to improve child health. International Journal of Epidemiology, 45(6), 2089–2099. 10.1093/ije/dyw242 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Johnson M, Pillarisetti A, Piedrahita R, Balakrishnan K, Peel JL, Steenland K, Underhill LJ, Rosa G, Kirby MA, Díaz-Artiga A, McCracken J, Clark ML, Waller L, Chang HH, Wang J, Dusabimana E, Ndagijimana F, Sambandam S, Mukhopadhyay K, … and the Household Air Pollution Intervention Network (HAPIN) Trial Investigators. (2022). Exposure Contrasts of Pregnant Women during the Household Air Pollution Intervention Network Randomized Controlled Trial. Environmental Health Perspectives, 130(9), 097005. 10.1289/EHP10295 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Johnson M, Steenland K, Piedrahita R, Clark ML, Pillarisetti A, Balakrishnan K, Peel JL, Naeher LP, Liao J, Wilson D, Sarnat J, Underhill LJ, Burrowes V, McCracken JP, Rosa G, Rosenthal J, Sambandam S, de Leon O, Kirby MA, … HAPIN Investigators. (2020). Air Pollutant Exposure and Stove Use Assessment Methods for the Household Air Pollution Intervention Network (HAPIN) Trial. Environmental Health Perspectives, 128(4), 47009. 10.1289/EHP6422 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kirby MA, Nagel CL, Rosa G, Zambrano LD, Musafiri S, Ngirabega JDD, Thomas EA, & Clasen T (2019). Effects of a large-scale distribution of water filters and natural draft rocket-style cookstoves on diarrhea and acute respiratory infection: A cluster-randomized controlled trial in Western Province, Rwanda. PLOS Medicine, 16(6), e1002812. 10.1371/journal.pmed.1002812 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Lee KK, Bing R, Kiang J, Bashir S, Spath N, Stelzle D, Mortimer K, Bularga A, Doudesis D, Joshi SS, Strachan F, Gumy S, Adair-Rohani H, Attia EF, Chung MH, Miller MR, Newby DE, Mills NL, McAllister DA, & Shah ASV (2020). Adverse health effects associated with household air pollution: A systematic review, meta-analysis, and burden estimation study. The Lancet Global Health, 8(11), e1427–e1434. 10.1016/S2214-109X(20)30343-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Legge H, Pullan RL, & Sartorius B (2023). Improved household flooring is associated with lower odds of enteric and parasitic infections in low- and middle-income countries: A systematic review and meta-analysis. PLOS Global Public Health, 3(12), e0002631. 10.1371/journal.pgph.0002631 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Liao J, McCracken JP, Piedrahita R, Thompson L, Mollinedo E, Canuz E, De Léon O, Díaz-Artiga A, Johnson M, Clark M, Pillarisetti A, Kearns K, Naeher L, Steenland K, Checkley W, Peel J, & Clasen TF (2020). The use of bluetooth low energy Beacon systems to estimate indirect personal exposure to household air pollution. Journal of Exposure Science & Environmental Epidemiology, 30(6), 990–1000. 10.1038/s41370-019-0172-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Mayo-Wilson E, Imdad A, Herzer K, Yakoob MY, & Bhutta ZA (2011). Vitamin A supplements for preventing mortality, illness, and blindness in children aged under 5: Systematic review and meta-analysis. BMJ, 343, d5094. 10.1136/bmj.d5094 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. McCollum ED, McCracken JP, Kirby MA, Grajeda LM, Hossen S, Moulton LH, Simkovich SM, Goodman-Palmer D, Rosa G, Mukeshimana A, Balakrishnan K, Thangavel G, Garg SS, Castañaza A, Thompson LM, Diaz-Artiga A, Papageorghiou AT, Davila-Roman VG, Underhill LJ, … Checkley W (2024). Liquefied Petroleum Gas or Biomass Cooking and Severe Infant Pneumonia. New England Journal of Medicine, 390(1), 32–43. 10.1056/NEJMoa2305681 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Murray CJL, Aravkin AY, Zheng P, Abbafati C, Abbas KM, Abbasi-Kangevari M, Abd-Allah F, Abdelalim A, Abdollahi M, Abdollahpour I, Abegaz KH, Abolhassani H, Aboyans V, Abreu LG, Abrigo MRM, Abualhasan A, Abu-Raddad LJ, Abushouk AI, Adabi M, … Lim SS (2020). Global burden of 87 risk factors in 204 countries and territories, 1990–2019: A systematic analysis for the Global Burden of Disease Study 2019. The Lancet, 396(10258), 1223–1249. 10.1016/S0140-6736(20)30752-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Öberg M, Jaakkola MS, Woodward A, Peruga A, & Prüss-Ustün A (2011). Worldwide burden of disease from exposure to second-hand smoke: A retrospective analysis of data from 192 countries. The Lancet, 377(9760), 139–146. 10.1016/S0140-6736(10)61388-8 [DOI] [PubMed] [Google Scholar]
  27. Pillarisetti A, Ye W, Balakrishnan K, Rosa G, Díaz-Artiga A, Underhill LJ, Steenland K, Peel JL, Kirby MA, McCracken J, Waller L, Chang H, Wang J, Dusabimana E, Ndagijimana F, Sambandam S, Mukhopadhyay K, Kearns KA, Campbell D, … Investigators, H. A. P. I. N. (HAPIN) trial. (2023). Post-birth exposure contrasts for children during the Household Air Pollution Intervention Network randomized controlled trial (p. 2023.07.04.23292226). medRxiv. 10.1101/2023.07.04.23292226 [DOI] [Google Scholar]
  28. Prüss A, Kay D, Fewtrell L, & Bartram J (2002). Estimating the burden of disease from water, sanitation, and hygiene at a global level. Environmental Health Perspectives, 110(5), 537–542. 10.1289/ehp.110-1240845 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Quinn AK, Williams KN, Thompson LM, Harvey SA, Piedrahita R, Wang J, Quinn C, Pillarisetti A, McCracken JP, Rosenthal JP, Kirby MA, Diaz Artiga A, Thangavel G, Rosa G, Miranda JJ, Checkley W, Peel JL, & Clasen TF (2021). Fidelity and Adherence to a Liquefied Petroleum Gas Stove and Fuel Intervention during Gestation: The Multi-Country Household Air Pollution Intervention Network (HAPIN) Randomized Controlled Trial. International Journal of Environmental Research and Public Health, 18(23), 12592. 10.3390/ijerph182312592 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Quinn AK, Williams K, Thompson LM, Rosa G, Díaz-Artiga A, Thangavel G, Balakrishnan K, Miranda JJ, Rosenthal JP, Clasen TF, & Harvey SA (2019). Compensating control participants when the intervention is of significant value: Experience in Guatemala, India, Peru and Rwanda. BMJ Global Health, 4(4). 10.1136/bmjgh-2019-001567 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Sang S, Chu C, Zhang T, Chen H, & Yang X (2022). The global burden of disease attributable to ambient fine particulate matter in 204 countries and territories, 1990–2019: A systematic analysis of the Global Burden of Disease Study 2019. Ecotoxicology and Environmental Safety, 238, 113588. 10.1016/j.ecoenv.2022.113588 [DOI] [PubMed] [Google Scholar]
  32. Smith KR, McCracken JP, Weber MW, Hubbard A, Jenny A, Thompson LM, Balmes J, Diaz A, Arana B, & Bruce N (2011). Effect of reduction in household air pollution on childhood pneumonia in Guatemala (RESPIRE): A randomised controlled trial. Lancet (London, England), 378(9804), 1717–1726. 10.1016/S0140-6736(11)60921-5 [DOI] [PubMed] [Google Scholar]
  33. Textor J, van der Zander B, Gilthorpe MS, Liśkiewicz M, & Ellison GT (2016). Robust causal inference using directed acyclic graphs: The R package ‘dagitty.’ International Journal of Epidemiology, 45(6), 1887–1894. 10.1093/ije/dyw341 [DOI] [PubMed] [Google Scholar]
  34. Thakur M, Nuyts PAW, Boudewijns EA, Kim JF, Faber T, Babu GR, Schayck O. C. P. van, & Been JV (2018). Impact of improved cookstoves on women’s and child health in low and middle income countries: A systematic review and meta-analysis. Thorax, 73(11), 1026–1040. 10.1136/thoraxjnl-2017-210952 [DOI] [PubMed] [Google Scholar]
  35. The World Bank. (2023). Tracking SDG7: The energy progress report 2023. https://www.wame2030.org/files/catalogue/2023/11/SDG7_Report2023_FullReport_8VaVBTw.pdf
  36. UNICEF and WHO. (2018). Core questions on drinking water, sanitation and hygiene for household surveys: 2018 update. https://washdata.org/monitoring/methods/core-questions
  37. United Nations Inter-agency Group for Child Mortality Estimation. (2024). Levels & Trends in Child Mortality: Report 2023, Estimates developed by the United Nations Inter-agency Group for Child Mortality Estimation.
  38. Williams KN, Quinn A, North H, Wang J, Pillarisetti A, Thompson LM, Díaz-Artiga A, Balakrishnan K, Thangavel G, Rosa G, Ndagijimana F, Underhill LJ, Kirby MA, Puzzolo E, Hossen S, Waller LA, Peel JL, Rosenthal JP, Clasen TF, … HAPIN Investigators. (2023). Fidelity and adherence to a liquefied petroleum gas stove and fuel intervention: The multi-country Household Air Pollution Intervention Network (HAPIN) trial. Environment International, 179, 108160. 10.1016/j.envint.2023.108160 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Xie S, Zhang C, Zhao J, Li D, & Chen J (2022). Exposure to concentrated ambient PM2.5 (CAPM) induces intestinal disturbance via inflammation and alternation of gut microbiome. Environment International, 161, 107138. 10.1016/j.envint.2022.107138 [DOI] [PubMed] [Google Scholar]
  40. Zambrano LD, Levy K, Menezes NP, & Freeman MC (2014). Human diarrhea infections associated with domestic animal husbandry: A systematic review and meta-analysis. Transactions of The Royal Society of Tropical Medicine and Hygiene, 108(6), 313–325. 10.1093/trstmh/tru056 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Zhang K (2024). Environmental PM2.5-triggered stress responses in digestive diseases. eGastroenterology, 2(2), e100063. 10.1136/egastro-2024-100063 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Zhang Z, Song N, Wang J, Liu J, Shi L, & Du J (2025). Effect of PM2.5 air pollution on the global burden of neonatal diarrhea from 1990 to 2019. Environmental Pollution, 367, 125604. 10.1016/j.envpol.2024.125604 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2104103-supplement-1.pdf (1.8MB, pdf)

Data Availability Statement

Deidentified data associated with these analyses will be available through Emory Dataverse, linked to the digital object identifier (DOI) of this publication.

RESOURCES