Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jul 10.
Published in final edited form as: Environ Int. 2026 May 9;212:110297. doi: 10.1016/j.envint.2026.110297

Prenatal and postnatal household air pollution and longitudinal growth trajectories: results from long-term follow up of the GRAPHS cohort

Seyram Kaali a,b,c,*, Martin Röösli b,c, Tawfiq Yussif a, Michelle Li d, Mohammed Nuhu Mujtaba a, James Ross e, Prosper Donneyong a, Peggy S Lai f, Sule Awuni a, Musah Osei a, Elena Colicino g, Nicole Probst-Hensch b,c, Yohannes Tsefaigzi h, Steven Chillrud e, Darby Jack i, Kwaku Poku Asante a, Alison Lee d
PMCID: PMC13348564  NIHMSID: NIHMS2184267  PMID: 42134200

Abstract

Background:

There is limited evidence on the long-term effects of household air pollution (HAP) exposure during critical developmental windows and longitudinal growth trajectories in childhood.

Methods:

We leveraged a longitudinal pregnancy cohort from rural communities in central Ghana with repeated personal monitoring of prenatal and postnatal carbon monoxide (CO) and fine particulates (PM2.5). Child growth was measured at 1 year of age (n = 919), and annually from age 4 through 9 years (n = 698). Using generalized linear mixed-effects models adjusted for relevant confounders, we quantified separately, the age-specific associations between prenatal and postnatal CO and PM2.5, and z-scores of height-for-age (HAZ), weight-for-age (WAZ) and body mass index (BMI)-for-age (BMIZ) in primary models. Secondary models assessed the risk of being in a worse growth category compared with a better category e.g., for HAZ, stunted or at risk of stunting versus normal, or stunted versus at risk or normal).

Results:

Prenatal CO had the most consistent association with growth. Higher prenatal CO was associated with lower HAZ from age 1 through 9 with peak effect at age 4: adjusted β = −0.08; 95% CI = −0.12, −0.03; p = 0.002 per doubling), and WAZ from 7 through age 9 (peak effect was at age 9: adjusted β = −0.05; 95% CI = −0.09, −0.003; p = 0.04 per doubling). Results were similar for the ordered categories of HAZ and WAZ. Sex did not modify these associations.

Conclusions:

Higher HAP exposure, as indexed by prenatal CO during critical developmental windows may contribute substantially to long-term poor linear growth throughout childhood.

Keywords: Household air pollution, Prospective birth cohort, Critical windows of development, Longitudinal growth trajectories, Child anthropometrics

1. Introduction

Globally, nearly 200 million children under five years are stunted or wasted. Africa is one of the highest burden continents and accounts for 43% and 27% of global stunting and wasting respectively (World Health Organization (WHO), 2023). In Ghana, the prevalence of stunting and underweight is estimated at 17% and 12% respectively (Osborne et al., 2025). In the short-term, poor growth in childhood is associated with impaired immunity, increased risk of recurrent infections, particularly those of the respiratory and gastrointestinal tract, and mortality (Gwela et al., 2019; Kirolos et al., 2021; Olofin et al., 2013; Tickell et al., 2020). Survivors are at risk of long-term consequences including impaired physical, cognitive and psychosocial development (Ansuya et al., 2023; Gizaw et al., 2023; Kirolos et al., 2022), and increased risk of cardiometabolic disease across the life course (Grey et al., 2021). Evidence from our cohort, the Ghana Randomized Air Pollution and Health Study (GRAPHS) finds that worse growth trajectories, including stunting, over early childhood are associated with poorer lung function at age 4 years with implications for respiratory health across the life course(Kaali et al., 2022). Although childhood growth impairment is mostly a result of inadequate nutrition or recurrent infections, air pollution is increasingly being recognized as a major contributor (Martorell and Zongrone, 2012; Sinharoy et al., 2020).

Globally, over 700, 000 under-five deaths are attributable to air pollution, the second leading risk factor for death in this age group (Health Effects Institute (IHME), 2024). Household air pollution (HAP) from domestic use of solid fuels in poorly ventilated cooking environments mostly in Low and Middle Income Countries (LMICs), account for 75% of these deaths (Health Effects Institute (IHME), 2024). Exposures begin in-utero and continue through early childhood. HAP exposures are especially high in children given that they are often in close proximity to their caregivers during cooking sessions. This combined with the unique health vulnerabilities in early childhood makes children more susceptible to the effects of HAP (Health Effects Institute (IHME), 2024). The majority of the air pollution-related deaths are attributable to respiratory infections and fetal growth restriction manifesting as stillbirth, low birth weight, and small-for-gestational age (Health Effects Institute (IHME), 2024; Lelieveld et al., 2018; Amegah et al., 2014; Quinn et al., 2021). Beyond the perinatal period, air pollution may be a significant contributor to poor growth in childhood. However, the evidence is limited (Sinharoy et al., 2020), and results from existing studies are mixed (Desouza et al., 2022; Fleisch et al., 2019; Fossati et al., 2020; Tan et al., 2024).

Regarding HAP, the overall evidence from previous observational studies including national and multi-country health and demographic surveys and prospective cohort studies point to an adverse effect on growth (Dai et al., 2025; Islam et al., 2021; Mulat et al., 2024; Odame and Adjei-Mantey, 2024; Upadhyay et al., 2021). However, these studies either lack long-term follow up and are limited by use of single time point measurement of growth outcomes to assess cross-sectional associations, or use proxy measures such as stove use, fuel type and cooking location to characterize HAP exposure. Evidence from GRAPHS finds that higher prenatal and postnatal HAP is associated with worse growth trajectories over the first year of life (Boamah-Kaali et al., 2021). In the multi-country Household Air Pollution intervention Network (HAPIN) trial, a randomized controlled trial (RCT) of Liquefied Petroleum Gas (LPG) cookstove intervention to reduce HAP during pregnancy through 1 year of age, LPG did not reduce the risk of stunting at one year of age (Checkley et al., 2024). When HAPIN trial children in Peru were followed through 2–4 years of age, similar null results were found in both intention-to-treat (ITT) and exposure–response analyses (Nicolaou et al., 2026). Another randomized cookstove intervention trial in Northern Ghana did not find any impact of cleaner biomass cookstoves on growth in children 5 years or younger compared to control (Abdo et al., 2021).

To date, very few studies have linked objectively measured personal HAP exposures to longer-term growth outcomes over childhood. As above, our own group previously reported inverse associations between personal exposures to prenatal and postnatal HAP and growth trajectories over the first year of life in GRAPHS (Boamah-Kaali et al., 2021), however examination of whether these associations persist over childhood is needed. The prenatal and first 2 years of life are critical periods for physical growth and development. At the population level, the most severe impairment in growth occurs in the first 2 years of life (Victora et al., 2010). Hence, exposures occurring during the critical developmental windows of prenatal and first 2 years of life may worsen and leave a lasting impact on growth and development beyond early childhood. Here, we build on our previous findings from GRAPHS by leveraging long-term follow up of the GRAPHS cohort to investigate associations between prenatal and postnatal HAP, as indexed by personal carbon monoxide (CO) and fine particulate matter (PM2.5), and longitudinal growth trajectories from 1 through 9 years of age. We further explored effect modification by child sex given that child growth follows sex-specific trajectories (Costa et al., 2021; Zheng et al., 2013), and previous studies have reported sex-specific effects of HAP on child health outcomes (Kinney et al., 2021; Lee et al., 2019; Lu et al., 2024).

2. Materials and methods

2.1. Study participants

In this prospective cohort study, we used data from mother-child dyads participating in GRAPHS, a longitudinal pregnancy cohort in the Kintampo North Municipality and Kintampo South District of Ghana. Detailed descriptions of GRAPHS can be found elsewhere (Jack et al., 2015; Jack et al., 2021). Briefly, between August 2013 and June 2015, we randomized 1414 non-smoking women in the first and second trimesters of pregnancy to a LPG stove intervention, an improved biomass cookstove intervention, or a three-stone open traditional cookstove (control). Eligible women were the main cooks in their household, carried a singleton pregnancy and were below 24 weeks’ gestation confirmed by ultrasonography. Participants were followed through pregnancy until the offspring was one year old at which point support for the interventions ended. The primary endpoints for GRAPHS were birthweight and incident pneumonia in the first year of life. In May 2018 at approximately 4 years of age, a subset of 698 mother–child dyads was recruited for prospective follow up with longitudinal assessment of respiratory and cardiometabolic outcomes including annual anthropometric measurements. The data presented in this paper include anthropometric data collected up to June 2024 at child age 9 years. The study procedures were approved by the Institutional Ethics Committee of Kintampo Health Research Centre (KHRCIEC/2017–31), the Ghana Health Service Ethics Review Committee (GHS-ERC: 032/03/23), and Institutional Review Boards at Columbia University (IRB-AAAR4373) and Icahn School of Medicine at Mount Sinai (STUDY-17–01265). Written informed consent was obtained from caregivers prior to commencement of study procedures.

2.2. Exposure measurement

Details of exposure assessment in GRAPHS have been described elsewhere (Chillrud et al., 2021). Briefly, mother-infant pairs underwent repeated prenatal and postnatal monitoring of CO as the primary pollutant using the Lascar EL-CO-USB data logger (Erie, PA, USA). Fine particulate matter (with aerodynamic diameter < 2.5 μg/m3) or PM2.5 was monitored in a subset of 980 women using the MicroPEM (μPEM) v3.2b (RTI International, Research Triangle Park, NC, USA). Prenatal CO monitoring occurred at four time points – once at baseline before intervention stove deployment and, thereafter, at three equally spaced time points (9, 6 and 3 weeks) prior to the estimated date of delivery. In the postnatal period, both mother and infant underwent personal CO monitoring at child age 1, 4 and 12 months. PM2.5 was co-monitored with CO once during the second prenatal session and once during the second postnatal session on a convenient subsample. Only study mothers received postnatal PM2.5 monitoring as the MicroPEM device was too bulky to safely deploy on infants. Each monitoring session was planned to last a maximum of 72 h. Participants wore the monitors affixed to their clothing near the breathing zone throughout the monitoring session except during bathing or sleep time when the monitors were placed near the head and off the floor.

Given the goal of GRAPHS to attribute exposures to individuals, a detailed quality assurance/control (QA/QC) protocol was implemented from field deployment of monitors to data processing to ensure the quality of the personal exposure data. First, in addition to factory calibration, the study team followed a strict protocol of checking the Lascar monitors every 6 weeks against certified span gas (50 ppm CO in zero air). Time series data from each deployment were visually inspected for data anomalies such as unexpected plateaus, elevated baseline values or prolonged periods of 0 ppm. PM2.5 measurement was based on real-time nephelometric measurement combined with integrated gravimetric sampling (Teflon 25-mm filter, Pall Biotech) and accelerometry to assess wearing compliance. Nephelometer values were baseline-corrected using a high-efficiency particulate air (HEPA) filter applied pre- and post- deployment, with interpolation across the sampling period. The corrected nephelometric data were subsequently adjusted for the net filter weight of PM2.5. Additional quality control included visual inspection of the time series data as for CO. For both CO and PM2.5, the mean of the first 48-hr measurement was used for subsequent analysis to avoid bias from missing data in the last 24 h of deployment (Boamah-Kaali et al., 2021; Chillrud et al., 2021; Daouda et al., 2024; Kinney et al., 2021). Forty-eight-hour data completeness was >90% for both CO and PM2.5 (Chillrud et al., 2021). To obtain overall estimates of average CO over the entire pregnancy (average prenatal CO) and first year of life (average postnatal CO) from the repeated CO sessions, we linearly interpolated the CO values between successive sessions as previously described (Quinn et al., 2021).

2.3. Growth measurements

Trained field workers measured weight and length/height at age 1 year in the entire cohort and at annual visits from age 4 through 9 years in the subset selected for prospective follow up. At age 1, we measured infant weight and length once using the Tanita BD 590 digital scale (Tanita Corp., Tokyo, Japan) and the Ayrton M-200 infantometer (Ayrton Corp., MN, USA) respectively. At age 4 through 9, we measured weight and height at annual visits. Measurements were taken in duplicates or triplicates at each annual visit using the Seca 803 Clara Digital Floor Scale and the Seca 213 Portable Stadiometer. Weight and length/height were measured to the nearest 0.1 kg and 0.1 cm respectively. Where measurements were taken in duplicates, the two were averaged and for triplicate measurements, the last two were averaged for subsequent analysis. Body mass index was calculated from weight in kg divided by the square of length/height in meters. The raw weight, length/height and BMI data were then converted to z-scores using the WHO growth standards for sex and age (WHO zscore06 Stata package for children under 60 months of age, and the WHO 2007 macro for children 60 months and older) (Leroy, 2011; Onis et al., 2007).

2.4. Covariates

Guided by Directed Acyclic Graphs (DAG) and previous studies, we selected a set of covariates a priori that we hypothesized would influence the association between HAP and child growth. Asset index, a proxy for socioeconomic status was calculated at GRAPHS enrollment by applying principal components analysis to an enumerated list of household characteristics, ownership of durable assets and access to utilities using methods described by Filmer et al and Gunnsteinsson et al (Filmer and Pritchett, 2001; Gunnsteinsson et al., 2010). Maternal BMI was calculated from weight and height measured at GRAPHS enrollment. Maternal age, education, ethnicity, and household exposure to second-hand tobacco smoke were recorded at enrollment. Season of birth (rainy versus dry) and child sex were recorded at birth.

2.5. Statistical analysis

We summarized baseline characteristics in the overall cohort and by sex. We then coded each continuous z-score outcome into an ordered categorical variable. For HAZ, children were classified as normal (≥−1 SD), at-risk-of-stunting (−2 ≤ HAZ < −1 SD), or stunted (<−2 SD). Similarly, for WAZ, we classified children as normal, at-risk-of-underweight, or underweight using the same SD cutoffs; and for BMIZ, as normal, at-risk-of-thinness, or thin using the same thresholds. The goal of the analysis was to quantify long-term associations between prenatal and postnatal HAP, and longitudinal trajectories of growth from age 1 through 9 years. To achieve this, we constructed two sets of models. First, using the continuous growth outcomes of HAZ, WAZ and BMIZ, we fit separate linear mixed-effects models for each exposure-growth z-score combination in single pollutant models, as the primary analysis. Second, we used a proportional-odds cumulative-link mixed-effects models with a logit link function for each exposure-ordered categorical growth outcome combination.

We specified a correlated random-effects structure in the mixed-effects models, which included random intercepts and random slopes at the child level to account for repeated growth measurements within each child and between-child heterogeneity in baseline growth and growth rates. To capture both linear and non-linear changes in growth over time and to estimate age-varying exposure effects, we parameterized age using a quadratic polynomial (linear and squared terms for centered age) and interaction terms between age and exposure were included. We assessed potential non-linearity in the exposure–response relationship by comparing models with linear exposure terms to models incorporating natural spline functions. Where there was evidence of departure from linearity, we further evaluated the appropriate functional form by comparing log2-transformed and spline-based models and selected the preferred specification based on likelihood ratio tests and information criteria (see Supplementary material). Where no evidence of nonlinearity was observed, exposures were modeled on the linear scale. All models included exposure-age interaction terms to allow effects to vary over time. To evaluate potential effect modification of the HAP-growth association by sex, we fit interaction models by specifying three-way interaction terms between the polynomial age terms, exposure and sex. Model equations are shown in the Supplementary material.

From the linear mixed models, we derived age-specific average marginal effects (AMEs) of prenatal and postnatal HAP. For log2-transformed (log-linear) models, AMEs represent the predicted change in z-scores associated with a doubling of exposure, with effects varying across age. For linear models, AMEs represent the predicted change in z-scores per 1 ppm increase in CO or per 10 μg/m3 increase in PM2.5. For spline-based models, AMEs represent the predicted change in z-score per 10 μg/m3 increase in PM2.5, evaluated at the median PM2.5 level, with effects varying across age. Model coefficients from the proportional odds models were estimated via maximum likelihood using the Laplace approximation. Under the proportional odds assumption, odds ratios represent the multiplicative change in the odds of being in a lower (worse) growth category versus higher categories (e.g., at risk of stunting or stunted versus normal, or stunted versus at risk or normal). Estimates from all models were computed by averaging across the observed joint distributions of covariates. Categorical covariates were frequency weighted and continuous covariates fixed at the mean for normally distributed covariates, and median for covariates with skewed distributions. Sex-stratified associations were obtained by repeating these procedures separately within each stratum of sex.

Models were adjusted for maternal education, age, BMI, parity, ethnicity; household asset index; second-hand tobacco smoke exposure; season of birth (rainy versus dry); and child sex. Postnatal models were additionally adjusted for prenatal exposures. The adjustment set was selected based on previous literature and use of Directed Acyclic Graphs (DAG) (Supplementary Fig. S1). Estimates were computed at the population level by marginalizing over random effects. Analyses were carried out in R version 4.4.1. Mixed-effects quadratic and spline models were ran using the lme4 and spline packages; the proportional odds models were implemented via the ordinal package. AMEs were implemented with the marginaleffects package. Odds ratios were computed with emmeans.

3. Results

Baseline characteristics are summarized in Table 1. GRAPHS recorded 1306 livebirths (Quinn et al., 2021), of which n = 919 had valid anthropometric measurements at the age 1 follow up. Of the n = 698 sub-cohort selected for prospective follow up at age 4, n = 691 attended at least one visit from age 4 through age 9. Cohort retention and visit completeness were high, as previously reported (Kaali et al., 2025). N = 593 (88%), 569 (99%), 622(98%), 544 (97%), 628 (99%) and 592 (97%) had valid anthropometric data at the age 4, 5, 6, 7, 8 and 9 visits respectively. The mean gestational age at delivery was 39 weeks and 3.7% of the children were born preterm (defined as delivery before 37 completed weeks of gestation). The mean birthweight was 2.9 kg [standard deviation (sd) = 0.5)], with a low birthweight prevalence of 16.4%, and mean birth length was 46.6 cm (sd = 3.6). Median prenatal CO [1.07 ppm, interquartile range (IQR) = 0.65, 1.65] was twice as high as median child postnatal CO (0.53 ppm, IQR = 0.24, 1.02). Similarly, median prenatal PM2.5 (70.6 μg/m3, IQR = 44.3, 103.4) was 1.2 times higher than median maternal postnatal PM2.5 (57.3 μg/m3, IQR = 37.3, 87.2). Prenatal and postnatal exposures were weakly correlated (Supplementary Fig. S2). Supplementary Fig. S3 and Supplementary Table S1 show the distribution of raw anthropometric measurements. Majority of study children had z-scores below the median value of the reference population (Fig. 1; Supplementary Table S2). The prevalence of stunting, underweight and thinness decreased with age, however, the proportion of children at risk of stunting, underweight and thinness remained high at age 9 (Fig. 2, Supplementary Table S3).

Table 1.

Participants’ characteristics.

Characteristic Overall (n = 919) Male (n = 472) Female (n = 447)
Continuous variables
Asset index [(median, (IQR)] −0.37 (−1.28,0.85) −0.39 (−1.32, 0.94) −0.36 (−1.25, 0.84)
Maternal age at enrollment, years [mean, (sd)] 28.1 (7.2) 27.7 (7.3) 28.6 (7.0)
Maternal BMI, kg/m2, [mean, (sd)] 23.4 (3.4) 23.5 (3.3) 23.3 (3.5)
Parity [(median, IQR)] 2.0 (1.0, 4.0) 2.0 (1.0, 4.0) 3.0 (1.0, 4.0)
Gestational age at birth, wk [mean (sd)] 39.2 (1.7) 39.1 (1.59) 39.3 (1.79)
Birthweight, Kg [mean, (sd)] 2.90 (0.46) 2.93 (0.47) 2.87 (0.46)
Average prenatal CO, ppm
Mean (sd) 1.33 (1.00) 1.33 (0.90) 1.33 (1.09)
Median (IQR) 1.11 (0.69, 1.72) 1.11 (0.72, 1.70) 1.11 (0.64, 1.75)
Child average postnatal CO, ppm
Mean (sd) 0.88 (1.39) 0.89 (1.63) 0.87 (1.07)
Median (IQR) 0.53 (0.24, 1.02) 0.51 (0.23, 0.99) 0.58 (0.27, 1.03)
Prenatal PM2.5, μg/m3
Mean (sd) 85.77 (58.56) 88.39 (63.13) 82.97 (53.26)
Median (IQR) 70.40 (43.32, 107.46) 70.47 (43.70, 110.31) 69.72 (42.87, 101.80)
Missing n (%) 391 (42.5) 200 (42.4) 191 (42.7)
Maternal postnatal PM2.5, μg/m3
Mean (sd) 65.83 (40.15) 66. 28 (41.93) 65.34 (38.27)
Median (IQR) 55. 64 (37.26, 83.84) 56.81 (35.39, 83.75) 55.23 (38.67, 85.89)
Missing n (%) 438 (47.7) 225 (47.7) 213 (47.7)
Categorical variables [(n (%)]
Ethnicity
1 163 (17.7) 73 (44.8) 90 (55.2)
2 121 (13.2) 62 (51.2) 59 (48.8)
3 596 (64.9) 312 (52.3) 284 (47.7)
4 39 (4.2) 25 (64.1) 14 (35.9)
Second-hand tobacco exposure 189 (20.6) 93 (19.7) 96 (21.5)
Low birth weight 149 (16.6) 71 (15.5) 78 (17.7)
Missing 20 15 5
Small for gestational age 196 (21.8) 92 (20.1) 104 (23.5)
Preterm Birth 33 (3.6) 17 (3.6) 16 (3.6)
Missing 2 1 1
Breastfeeding at 1 year 911 (99.2) 469 (99.6) 442 (98.9)
Missing 1 1 0
Maternal Education
None 376 (40.9) 197 (52.4) 179 (47.6)
Primary 245 (26.7) 133 (54.3) 112 (45.7)
Middle/Junior High School 258 (28.1) 127 (49.2) 131 (50.8)
Secondary/post-secondary 40 (4.4) 15 (37.5) 25 (62.5)
Season of birth
Dry 534 (58.1) 276 (51.7) 258 (48.3)
Rainy 385 (41.9) 196 (50.9) 189 (49.1)
Open in a new tab

Fig. 1.

Fig. 1.

Open in a new tab

Distribution of z-scores of height, weight and BMI by age. Z-scores calculated using the WHO 2006 and 2007 growth standards. HAZ = height-for-age z-score; WAZ = weight-for-age z-score; BMIZ = BMI-for-age z-score.

Fig. 2.

Fig. 2.

Open in a new tab

Distribution of stunting, thinness and underweight severity by age.

3.1. Age-varying association between prenatal and postnatal HAP and continuous growth outcomes (HAZ, WAZ, BMIZ)

In adjusted linear mixed-effects models, higher prenatal CO was inversely associated with HAZ from age 1 through age 9 with a peak effect observed at age 4 (age 4 adjusted β = −0.08; 95% CI = −0.12, −0.03; p = 0.002 per doubling of prenatal CO; Fig. 3A, Supplementary Table S4). Similarly, Prenatal PM2.5 showed a general pattern of inverse relationship with HAZ across all ages; however, the effect estimates were not statistically significant except at age 1 (age 1 β = −0.02; 95% CI = −0.32, −0.001; p = 0.04 per 10 μg/m3 increase in prenatal PM2.5; Fig. 3C, Supplementary Table S4). For the postnatal exposures, CO was positively associated with HAZ at age 7 to 9 (Fig. 3B, Supplementary Table S4). Postnatal PM2.5 was inversely related to HAZ across all ages but the effect estimates were not statistically significant (Fig. 3D; Supplementary Table S4).

Fig. 3.

Fig. 3.

Open in a new tab

Age-varying average marginal effects (AMEs) of prenatal and postnatal HAP on HAZ. Curves were derived from linear mixed-effects single-pollutant models with quadratic time terms, and exposure-by-time interactions terms. Prenatal CO was modeled on the log2 scale; prenatal PM2.5 using a natural spline (df = 3); and postnatal exposures on the linear scale. Marginal effects represent the change in HAZ per unit increase in exposure (or per doubling for log2-transformed CO), evaluated at the population level. For the spline model, effects were evaluated at the median exposure level. Models were adjusted for asset index, maternal age, maternal BMI, ethnicity, second-hand tobacco smoke exposure, parity, season of birth, child sex and prenatal CO (for postnatal CO models) or prenatal PM2.5 (for postnatal PM2.5 model). Curves were averaged over categorical covariates (frequency-weighted); continuous covariates were fixed at the mean or median. Shaded areas indicate 95% confidence intervals. CO – carbon monoxide; PM2.5 – fine particulate matter; HAZ – height-for-age z-score.

Similar to HAZ, WAZ was negatively associated with prenatal CO, but the association reached statistical significance from age 7 to age 9 with a peak effect at age 9 (age 9 adjusted β = −0.05; 95% CI = −0.09, −0.003; p = 0.04 per doubling of prenatal CO; Fig. 4A, Supplementary Table S4). Postnatal CO was not associated with WAZ; Fig. 4B, Supplementary Table S4). Prenatal and postnatal PM2.5 were negatively associated with WAZ across most ages, but these associations were not statistically significant across all ages. (Fig. 4C and 4D, Supplementary Table S4).

Fig. 4.

Fig. 4.

Open in a new tab

Age-varying average marginal effects (AMEs) of prenatal and postnatal HAP on weight-for-age z-score. Curves were derived from linear mixed-effects single-pollutant models with quadratic time terms, and exposure-by-time interactions. Prenatal CO and postnatal PM2.5 were modeled on the log2 scale while prenatal PM2.5 and postnatal CO were on the linear scale. AMEs were computed at the population level by marginalizing over random effects. Models were adjusted for asset index, maternal age, maternal BMI, ethnicity, second-hand tobacco smoke exposure, parity, season of birth, child sex, and prenatal CO (for postnatal CO models) or prenatal PM2.5 (for postnatal PM2.5 model). Curves were averaged over categorical covariates (frequency-weighted); continuous covariates were fixed at the mean or median. Shaded band = 95% CI. CO – carbon monoxide; PM2.5 – fine particulate matter; WAZ – weight-for-age z-score.

BMI showed negative associations with postnatal CO and postnatal PM2.5 across most ages, but again, these associations were not statistically significant (Fig. 5B and D; Supplementary Table S4. Exposure-response curves for prenatal CO versus HAZ at representative ages are shown in Supplementary Fig. S7.

Fig. 5.

Fig. 5.

Open in a new tab

Age-varying marginal effects of prenatal and postnatal HAP on BMI-for-age z-score. Curves were derived from mixed-effects quadratic-in-time, linear-in-exposure single-pollutant models with two-way interaction between polynomial time terms and log2-transformed exposures. Estimates were computed at the population level by marginalizing over random effects. Models were adjusted for asset index, maternal age, maternal BMI, ethnicity, second-hand tobacco smoke exposure, parity, season of birth, child sex, and prenatal CO (for postnatal CO models) or prenatal PM2.5 (for postnatal PM2.5 model). Curves were averaged over categorical covariates (frequency-weighted); continuous covariates were fixed at the mean or median. Shaded band = 95% CI. CO – carbon monoxide; PM2.5 – fine particulate matter; BMIZ – Body mass index-for-age z-score.

3.2. Age-specific associations between prenatal and postnatal HAP and ordered categories of stunting, underweight and thinness

From the proportional odds models, higher prenatal CO was associated with increased odds of being in a worse stunting category compared to a better category (stunted versus at risk or normal; stunted or at risk versus normal) across all ages (peak effect was at age 9: adjusted OR = 1.30; 95% CI = 1.06, 1.57; p = 0.01 per doubling of prenatal CO. (Supplementary Fig. S4A, Supplementary Table S5). All other exposure metrics were not associated with odds of stunting severity. For underweight severity, only postnatal PM2.5 trended towards worse severity at age 1 through 7 (Supplementary Fig. S5D; Supplementary Table S5). Similarly for thinness severity, postnatal PM2.5 was associated with worse severity at age 1, and 4 (peak effect was at age 1: adjusted OR = 1.42; 95% CI = 1.02; 1.98, p = 0.040) (Supplementary Fig. 6D, Supplementary Table S5).

3.3. Sex-stratified age-varying associations between prenatal and postnatal HAP and growth trajectories

Overall, there was little evidence of effect modification by sex across most exposure-outcome combinations. However, there was some evidence that the associations between postnatal CO and HAZ and WAZ differed by sex (HAZ: Pint for postnatal CO × sex = 0.012; WAZ; Pint for postnatal CO × sex = 0.006). Sex-stratified estimates suggested that higher postnatal CO was associated with lower HAZ across all ages, among girls, although these associations were not statistically significant. In contrast, among boys, postnatal CO was associated with higher HAZ across most ages (Supplementary Fig. S8B; Supplementary Table S6). A similar pattern was observed for postnatal CO versus WAZ (Supplementary Fig. S9B, Supplementary Table S6). Similar patterns of sex-specific associations were observed for the ordered categories of HAZ, WAZ and BMIZ (Supplementary Figs. S11–S13, Supplementary Table S8 and S9).

4. Discussion

We investigated the association between prenatal and postnatal HAP and longitudinal trajectories of child growth from long-term follow up of the GRAPHS cohort. In this highly exposed and vulnerable population, the most consistent association between HAP and child growth was found between prenatal CO and height trajectory. Specifically, higher prenatal CO was associated with lower HAZ and higher odds of being in a worse stunting category compared to a better category from 1 through 9 years of age. Prenatal CO was also associated with lower WAZ from 7 through age 9 years and postnatal PM2.5 was associated with higher odds of being in a worse thinness category compared to a better category at 1 and 4 years of age.

Prenatal and postnatal PM2.5 were associated with lower HAZ, WAZ and BMIZ across most ages, but these associations were not statistically significant. However, the consistency and direction across outcomes and ages is suggestive of a potential adverse effect of PM2.5 on child growth. These findings may reflect modest effect sizes that are difficult to detect with limited precision, possibly related to the smaller sample size in the PM2.5 models, rather than an absence of a true association. Nevertheless, these findings should be interpreted cautiously, but within the broader context of accumulating evidence linking particulate air pollution to child growth. Overall, there was little evidence of effect modification by sex for all exposure-growth outcome combinations.

Taken together, these data highlight the importance of air pollution exposure on child growth, with implications for health across the life course. Poor growth not only increase the risk of child mortality, it also increases the risk of chronic disease in survivors and negatively affects societies by decreasing human capital development(Dewey and Begum, 2011; Freer et al., 2025). Stunting in particular has been associated with poor neurocognitive development with downstream effects on educational attainment and socioeconomic productivity(Soliman et al., 2021). HAP may therefore deepen existing social and income inequalities in the most vulnerable populations.

In this comprehensive analysis, we leveraged GRAPHS personal air pollution exposure repository and repeated anthropometry data to objectively link HAP with longitudinal growth trajectories in early to mid-childhood. We used linear models, which treated growth on the continuous scale and a threshold-based model that used cutoffs to define normal versus less-than-optimal growth. Although threshold-based models are easier to interpret clinically, it is important to recognize that growth faltering is a graded response and children who do not meet the threshold definition may still be at risk of long-term consequences (De Onis and Branca, 2016). This is especially relevant for our study population in which majority of study children consistently had HAZ and WAZ below the median value of the standard population from age 1 through 9 years.

We found that the effect sizes for HAZ and WAZ relative to prenatal CO were modest, ranging from −0.04 to −0.08 SD per doubling of exposure (e.g. from 0.5 to 1.0 ppm for CO). These effect sizes are comparable to those reported by Spears et al (−0.05 HAZ SD per 100 μg/m3 increase in PM2.5) (Spears et al., 2019), but slightly smaller than the −0.21 SD for HAZ per 1 ppm increase in CO reported by Lu et al (Lu et al., 2024). At the individual level, an effect size of −0.04 to −0.08 SD would not translate to clinically significant decreases in height or weight. However, at the population level across millions of at-risk children, modest shifts in HAZ or WAZ could translate to significant changes in the prevalence of stunting or underweight. Indeed this range of magnitude of effect is comparable to or larger than the impact on linear growth of nutritional or improved sanitation interventions achieved in large clinical trials (Dewey et al., 2021; Dewey et al., 2023; Luby et al., 2018), highlighting the potential impact on growth of clean cooking interventions in addition to existing interventions. Despite this, limited evidence from RCTs indicate that LPG or improved biomass cookstoves have largely failed to improve infant or child growth. In the HAPIN trial and its longer-term follow up of Peruvian children, LPG had no impact on linear growth in infants(Checkley et al., 2024) or children aged up to 4 years(Nicolaou et al., 2026). Abdo et al found no impact of improved biomass cookstoves on HAZ in children aged five years or younger(Abdo et al., 2021). Similarly, GRAPHS found no impact of LPG intervention on weight, length, weight-for-length z-score, length-for-age z-score and WAZ trajectories from birth through one year of age (Boamah-Kaali et al., 2021). Conversely, In the GRAPHS, the LPG intervention was associated with better trajectories of head circumference and mid-upper arm circumference(Boamah-Kaali et al., 2021).

Prior evidence on the effects of HAP exposure on child growth is mostly limited to birth anthropometrics. A systematic review and meta-analysis by Amegah et al found that HAP was associated with low birth weight and intrauterine growth restriction (Amegah et al., 2014). Subsequent studies including two recent systematic reviews corroborate these findings (Daba et al., 2024; Younger et al., 2022). However, the majority of the studies reviewed measured HAP exposure crudely using questionnaires to assess fuel type or stove type and thus are susceptible to exposure misclassification. A few studies in recent years have used objective personal or kitchen area monitoring of HAP or randomized assignment to cleaner cookstove interventions (Alexander et al., 2018; Checkley et al., 2024; Clasen et al., 2022; Dutta et al., 2021; Katz et al., 2020; Quinn et al., 2021; Wylie et al., 2017; Yucra et al., 2014). Our study extends these findings to demonstrate the importance of prenatal and early childhood HAP exposures on growth through age 9 years.

Beyond the perinatal period, as noted for birth anthropometrics above, majority of the evidence comes from studies using questionnaire-based proxy measures of HAP exposure with most studies relying on a single time-point measurement of growth outcomes, or lacking long-term follow up, thus limiting evaluation of longer-term effects(Adjei-Mantey and Takeuchi, 2021; Balietti et al., 2022; Caleyachetty et al., 2022; Liang et al., 2020; Odame and Adjei-Mantey, 2024; Yao et al., 2022). In the GRAPHS cohort, we previously reported an association between personal prenatal and postnatal HAP and longitudinal growth trajectories from birth to 1 year of age (Boamah-Kaali et al., 2021). The current study expands on these findings and sheds new light on the longer-term effects of HAP exposure as indexed by prenatal CO on growth in early to mid-childhood. Our findings are supported by two previous prospective cohort studies. Mulat et al followed a cohort of 280 children under-five years in Jimma, Ethiopia, half of which lived in households using solid fuels for cooking and the other half lived in households using cleaner fuel (electricity). After 12 months of follow up, children from solid fuel households had 0.54 SD lower HAZ from baseline compared to children from households using cleaner fuel (Mulat et al., 2024). In the Chronic Respiratory Effects of Early Childhood Exposure to Respirable PM Cohort (CRECER) in rural Guatemala in which 541 children under five received personal monitoring for CO, higher HAP was associated with lower HAZ, WAZ and WHZ, and higher risk of moderate stunting (Lu et al., 2024).

Our study contributes new findings on the potential of HAP particularly, prenatal CO to affect linear growth beyond 5 years. The long-term effect of HAP may be related to programming of growth in later years by exposures occurring in the critical window, or irreversible effects that continue to be seen in later childhood. On-going childhood exposures may also play a role. Supporting this assertion is the observation that somatic growth proceeds in an episodic saltatory fashion with short periods of spurts rather than in a smooth, linear and continuous fashion (Cliffer et al., 2022; Lampl et al., 1992). If this observation holds true for growth throughout childhood, then later childhood exposures may also be important. We however found that postnatal CO was associated with higher HAZ. It may be that the prenatal window is the critical exposure window for the programming effect of CO on linear growth hence with the trajectory already set before birth, postnatal CO would show no or even spurious association with growth. Alternatively, the paradoxical effect of postnatal CO may be due to the much lower levels of postnatal CO relative to prenatal CO, rendering the association more susceptible to residual confounding. Related to this, caregivers of sick or frailer infants may change their cooking behaviors or expose their infants less frequently to cooking fires although we did not collect data on behavior change.

Commonly proposed mechanisms through which air pollution affects child growth include oxidative stress and systemic inflammation, epigenetic alterations, impaired immune system development and increased risk of infections, and suppression of insulin-like growth factor (Sinharoy et al., 2020). Carbon particles are capable of crossing the placental barrier and entering the fetal circulation as demonstrated by recent data from two independent cohorts (Bongaerts et al., 2022). Placental and cord blood DNA hypomethylation have been associated with prenatal air pollution (Herbstman et al., 2012; Janssen et al., 2013). In the GRAPHS cohort, we previously reported associations between prenatal HAP and decreased cord blood telomore length (Kaali et al., 2021) and mitochondrial DNA copy number(Kaali et al., 2019), both markers of air pollution-induced oxidative stress (Byun and Baccarelli, 2014; Martens and Nawrot, 2016). HAP has been associated with impaired alveolar macrophage function (Hansson et al., 2023; Rylance et al., 2015), which may in turn increase the risk of respiratory infections. Findings from previous GRAPHS analyses show that prenatal HAP is associated with impaired infant lung function and increased risk of pneumonia (Kinney et al., 2021; Lee et al., 2019). The inflammatory state induced by air pollution-related respiratory infections, or through other pathways may inhibit longitudinal bone growth directly or via disruption of the growth hormone-insulin-like growth factor-1 axis (Iyer et al., 2022; Zeng et al., 2020). Finally, from the socioeconomic standpoint, another pathway may be via diversion of household income from food and nutrition towards healthcare costs related to the treatment of air pollution-related infections (Sinharoy et al., 2020).

The key strength of our study is that it is the first to link objectively measured personal prenatal and postnatal HAP exposures to longer-term longitudinal growth outcomes in children up to 9 years of age. Within the framework of a well-characterized prospective birth cohort, we were able to assemble extensive data on potential confounders. The longitudinal design with repeated anthropometric measurement from early to mid-childhood allowed us to capture within-child and between-child variabilities in growth trajectories, thus improving internal validity. We used different statistical modelling approaches and arrived at similar results across models, indicating the robustness of our conclusions. Lastly, the study population is broadly representative of similar low-resource settings in sub-Saharan Africa with similar exposure profile from heavy reliance on biomass fuels. Our results are therefore highly relevant to populations bearing the greatest burden of HAP globally, although generalizability to urban or high-income populations with different exposure profiles may be limited. We however acknowledge the following limitations. First, we did not have data on dietary intake, which is a major determinant of childhood growth. Second, we did not have growth data at age 2 and 3, hence we limited the analysis to the available data at ages 1, and 4 through 9. Third, our exposure monitoring strategy may not have been sufficient to adequately characterize HAP exposures over pregnancy and first year of life. While CO was monitored at multiple timepoints, PM2.5 was measured only once during pregnancy and once during the infant follow up. Additionally, we were unable to measure postnatal PM2.5 directly on study infants; instead, we used maternal exposure as proxy for infant exposure.

In conclusion, we find that prenatal HAP as indexed by CO is associated with poor growth trajectories, particularly linear growth from 1 through 9 years of age with implications for future child health. These results contribute critical evidence to the growing literature on the harmful effects of HAP on multiple child health outcomes and emphasize the need to accelerate efforts to combat HAP through provision of cleaner cooking technologies to the most vulnerable populations. Interventions to reduce HAP exposure such as provision of clean cooking fuel in the critical windows of prenatal and postnatal life may constitute an important yet overlooked strategy to reduce the burden of growth faltering in childhood.

Supplementary Material

1
2
3

Acknowledgements

The authors are thankful to study participants, community leaders, staff of Kintampo Health Research Centre and the other Ghana Health Service facilities in the Kintampo North Municipality and Kintampo South District that made this work possible.

Sources of funding

GRAPHS was supported by the National Institute of Environmental Health Sciences (NIEHS) Grants R01 ES019547, R01 ES026991, R01ES034433, R01 ES035791, U24 ES036007, P30 ES009089, and P30 ES023515, Fogarty Institute R21 TW010957, NIH Shared Instrument Program S10OD016219, Thrasher Research Fund, and the Clean Cooking Alliance.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.envint.2026.110297.

Footnotes

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Seyram Kaali: Writing – review & editing, Writing – original draft, Visualization, Supervision, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Martin Röösli: Writing – review & editing, Supervision. Tawfiq Yussif: Writing – review & editing, Project administration, Investigation. Michelle Li: Writing – review & editing, Data curation. Mohammed Nuhu Mujtaba: Writing – review & editing, Project administration, Investigation. James Ross: Writing – review & editing, Methodology, Data curation. Prosper Donneyong: Writing – review & editing, Investigation. Peggy S. Lai: Writing – review & editing. Sule Awuni:. Musah Osei: Writing – review & editing, Investigation. Elena Colicino: Writing – review & editing. Nicole Probst-Hensch: Writing – review & editing, Supervision. Yohannes Tsefaigzi: Writing – review & editing. Steven Chillrud: Writing – review & editing, Validation, Supervision, Resources, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization. Darby Jack: Writing – review & editing, Validation, Resources, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization. Kwaku Poku Asante: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization. Alison Lee: Writing – review & editing, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization.

Data availability

Data will be made available on request.

References

  1. Abdo M, et al. , 2021. Health impacts of a randomized biomass cookstove intervention in northern Ghana. BMC Public Health 21 (1), 2211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Adjei-Mantey K, Takeuchi K, 2021. The effect of in utero exposure to household air pollution on child health: evidence from Ghana. Health Policy OPEN 2, 100029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Alexander DA, et al. , 2018. Pregnancy outcomes and ethanol cook stove intervention: a randomized-controlled trial in Ibadan, Nigeria. Environ. Int. 111, 152–163. [DOI] [PubMed] [Google Scholar]
  4. Amegah AK, Quansah R, Jaakkola JJ, 2014. Household air pollution from solid fuel use and risk of adverse pregnancy outcomes: a systematic review and meta-analysis of the empirical evidence. PLoS ONE 9 (12), e113920. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Ansuya, et al. , 2023. Effect of nutrition intervention on cognitive development among malnourished preschool children: randomized controlled trial. Sci. Rep. 13 (1), 10636. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Balietti A, Datta S, Veljanoska S, 2022. Air pollution and child development in India. J. Environ. Econ. Manag. 113, 102624. [Google Scholar]
  7. Boamah-Kaali E, et al. , 2021. Prenatal and postnatal household air pollution exposure and infant growth trajectories: evidence from a rural Ghanaian pregnancy cohort. Environ. Health Perspect. 129 (11), 117009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bongaerts E, et al. , 2022. Maternal exposure to ambient black carbon particles and their presence in maternal and fetal circulation and organs: an analysis of two independent population-based observational studies. The Lancet Planetary Health 6 (10), e804–e811. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Byun H-M, Baccarelli AA, 2014. Environmental exposure and mitochondrial epigenetics: study design and analytical challenges. Hum. Genet. 133 (3), 247–257. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Caleyachetty R, et al. , 2022. Exposure to household air pollution from solid cookfuels and childhood stunting: a population-based, cross-sectional study of half a million children in low-and middle-income countries. Int. Health 14 (6), 639–647. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Checkley W, et al. , 2024. Cooking with liquefied petroleum gas or biomass and fetal growth outcomes: a multi-country randomised controlled trial. Lancet Glob. Health 12 (5), e815–e825. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Checkley W, et al. , 2024. Effects of cooking with liquefied petroleum gas or biomass on stunting in infants. N. Engl. J. Med. 390 (1), 44–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Chillrud SN, et al. , 2021. The effect of clean cooking interventions on mother and child personal exposure to air pollution: results from the Ghana Randomized Air Pollution and Health Study (GRAPHS). J. Eposure Sci. Environ. Epidemiol. 31 (4), 683–698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Clasen TF, et al. , 2022. Liquefied petroleum gas or biomass for cooking and effects on birth weight. N. Engl. J. Med. 387 (19), 1735–1746. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Cliffer IR, et al. , 2022. Monthly measurement of child lengths between 6 and 27 months of age in Burkina Faso reveals both chronic and episodic growth faltering. Am. J. Clin. Nutr. 115 (1), 94–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Costa JC, Blumenberg C, Victora C, 2021. Growth patterns by sex and age among under-5 children from 87 low-income and middle-income countries. BMJ Glob. Health 6 (11). [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Daba C, et al. , 2024. Exposure to indoor air pollution and adverse pregnancy outcomes in low and middle-income countries: a systematic review and meta-analysis. Front. Public Health 12, 1356830. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Dai Z, et al. , 2025. Assessing the association between household air pollution exposure and child heath in Mongolia: a birth-cohort study. Sci. Rep. 15 (1), 3878. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Daouda M, et al. , 2024. Prenatal household air pollution exposure and childhood blood pressure in rural Ghana. Environ. Health Perspect. 132 (3), 037006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. De Onis M, Branca F, 2016. Childhood stunting: a global perspective. Matern. Child Nutr. 12, 12–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Desouza PN, et al. , 2022. Impact of air pollution on stunting among children in Africa. Environ. Health 21 (1), 128. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Dewey KG, et al. , 2021. Characteristics that modify the effect of small-quantity lipid-based nutrient supplementation on child growth: an individual participant data meta-analysis of randomized controlled trials. Am. J. Clin. Nutr. 114, 15S–42S. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Dewey KG, et al. , 2023. Lipid-based nutrient supplements for prevention of child undernutrition: when less may be more. Am. J. Clin. Nutr. 118 (6), 1133–1144. [DOI] [PubMed] [Google Scholar]
  24. Dewey KG, Begum K, 2011. Long-term consequences of stunting in early life. Matern. Child Nutr. 7, 5–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Dutta A, et al. , 2021. Household air pollution, ultrasound measurement, fetal biometric parameters and intrauterine growth restriction. Environ. Health 20 (1), 74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Filmer D, Pritchett LH, 2001. Estimating wealth effects without expenditure data—or tears: an application to educational enrollments in states of India. Demography 38 (1), 115–132. [DOI] [PubMed] [Google Scholar]
  27. Fleisch AF, et al. , 2019. Prenatal exposure to traffic pollution and childhood body mass index trajectory. Front. Endocrinol. 9, 771. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Fossati S, et al. , 2020. Prenatal air pollution exposure and growth and cardio-metabolic risk in preschoolers. Environ. Int. 138, 105619. [DOI] [PubMed] [Google Scholar]
  29. Freer J, et al. , 2025. The impact of early-life growth on long-term cardiometabolic and neurocognitive outcomes in high-income countries: a neglected public health problem. Nutr. Rev., nuaf098. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Gizaw G, et al. , 2023. Growth and body composition 5 y after treatment for severe acute malnutrition: a 5-y prospective matched cohort study in Ethiopian children. Am. J. Clin. Nutr. 118 (5), 1029–1041. [DOI] [PubMed] [Google Scholar]
  31. Grey K, et al. , 2021. Severe malnutrition or famine exposure in childhood and cardiometabolic non-communicable disease later in life: a systematic review. BMJ Glob. Health 6 (3). [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Gunnsteinsson S, et al. , 2010. Constructing indices of rural living standards in Northwestern Bangladesh. J. Health Popul. Nutr. 28 (5), 509. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Gwela A, et al. , 2019. Undernutrition, host immunity and vulnerability to infection among young children. Pediatr. Infect. Dis. J. 38 (8), e175–e177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Hansson A, et al. , 2023. Reduced bronchoalveolar macrophage phagocytosis and cytotoxic effects after controlled short-term exposure to wood smoke in healthy humans. Part. Fibre Toxicol. 20 (1), 30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Health Effects Institute (IHME), State of Global Air 2024. Special Report. Health Effects Institute, Boston, MA. [Google Scholar]
  36. Herbstman JB, et al. , 2012. Prenatal exposure to polycyclic aromatic hydrocarbons, benzo [a] pyrene–DNA adducts, and genomic DNA methylation in cord blood. Environ. Health Perspect. 120 (5), 733–738. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Islam S, Rana MJ, Mohanty SK, 2021. Cooking, smoking, and stunting: effects of household air pollution sources on childhood growth in India. Indoor Air 31 (1), 229–249. [DOI] [PubMed] [Google Scholar]
  38. Iyer HS, et al. , 2022. Impacts of long-term ambient particulate matter and gaseous pollutants on circulating biomarkers of inflammation in male and female health professionals. Environ. Res. 214, 113810. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Jack DW, et al. , 2015. Ghana randomized air pollution and health study (GRAPHS): study protocol for a randomized controlled trial. Trials 16, 1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Jack DW, et al. , 2021. A cluster randomised trial of cookstove interventions to improve infant health in Ghana. BMJ Glob. Health 6 (8), e005599. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Janssen BG, et al. , 2013. Placental DNA hypomethylation in association with particulate air pollution in early life. Part. Fibre Toxicol. 10 (1), 22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Kaali S, et al. , 2019. Prenatal household air pollution alters cord blood mononuclear cell mitochondrial DNA copy number: sex-specific associations. Int. J. Environ. Res. Public Health 16 (1), 26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Kaali S, et al. , 2021. Prenatal household air pollution exposure, cord blood mononuclear cell telomere length and age four blood pressure: evidence from a Ghanaian pregnancy cohort. Toxics 9 (7), 169. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Kaali S, et al. , 2022. Poor early childhood growth is associated with impaired lung function: evidence from a Ghanaian pregnancy cohort. Pediatr. Pulmonol. 57 (9), 2136–2146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Kaali S, et al. , 2025. Early life cookstove interventions and children’s blood pressure: evidence from the Ghana Randomized Air Pollution and Health Study. Environ. Res., 122994 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Katz J, et al. , 2020. Impact of improved biomass and liquid petroleum gas stoves on birth outcomes in rural Nepal: results of 2 randomized trials. Glob. Health: Sci. Pract. 8 (3), 372–382. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Kinney PL, et al. , 2021. Prenatal and postnatal household air pollution exposures and pneumonia risk: evidence from the Ghana Randomized Air Pollution and Health Study. Chest 160 (5), 1634–1644. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Kirolos A, et al. , 2021. The impact of childhood malnutrition on mortality from pneumonia: a systematic review and network meta-analysis. BMJ Glob. Health 6 (11), e007411. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Kirolos A, et al. , 2022. Neurodevelopmental, cognitive, behavioural and mental health impairments following childhood malnutrition: a systematic review. BMJ Glob. Health 7 (7), e009330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Lampl M, Veldhuis JD, Johnson ML, 1992. Saltation and stasis: a model of human growth. Science 258 (5083), 801–803. [DOI] [PubMed] [Google Scholar]
  51. Lee AG, et al. , 2019. Prenatal household air pollution is associated with impaired infant lung function with sex-specific effects. Evidence from GRAPHS, a cluster randomized cookstove intervention trial. Am. J. Respir. Crit. Care Med. 199 (6), 738–746. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Lelieveld J, Haines A, Pozzer A, 2018. Age-dependent health risk from ambient air pollution: a modelling and data analysis of childhood mortality in middle-income and low-income countries. Lancet Planet. Health 2 (7), e292–e300. [DOI] [PubMed] [Google Scholar]
  53. Leroy J, 2011. ZSCORE06: Stata module to calculate anthropometric z-scores using the 2006 WHO child growth standards, Statistical Software Components S457279. Boston College Department of Economics, revised 04 Oct 2011. [Google Scholar]
  54. Liang W, et al. , 2020. Association of solid fuel use with risk of stunting in children living in China. Indoor Air 30 (2), 264–274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Lu W, et al. , 2024. Biomass smoke exposure and somatic growth among children: the RESPIRE and CRECER prospective cohort studies in rural Guatemala. Environ. Int. 183, 108401. [DOI] [PubMed] [Google Scholar]
  56. Luby SP, et al. , 2018. Effects of water quality, sanitation, handwashing, and nutritional interventions on diarrhoea and child growth in rural Bangladesh: a cluster randomised controlled trial. Lancet Glob. Health 6 (3), e302–e315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Martens DS, Nawrot TS, 2016. Air pollution stress and the aging phenotype: the telomere connection. Curr. Environ. Health Rep. 3 (3), 258–269. [DOI] [PubMed] [Google Scholar]
  58. Martorell R, Zongrone A, 2012. Intergenerational influences on child growth and undernutrition. Paediatr. Perinat. Epidemiol. 26, 302–314. [DOI] [PubMed] [Google Scholar]
  59. Mulat E, Tamiru D, Abate KH, 2024. Impact of indoor air pollution on the linear growth of children in Jimma, Ethiopia. BMC Public Health 24 (1), 488. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Nicolaou L, et al. , 2026. Long-term effects of cooking with liquefied petroleum gas or biomass on linear growth trajectories from birth to the pre-school years in Puno, Peru: a prospective cohort study. Lancet Reg. Health-Am. 55. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Odame ML, Adjei-Mantey K, 2024. Household air pollution could make children grow shorter in sub-Saharan Africa; but can households help stem the tide on their own? World Dev. Perspect. 33, 100562. [Google Scholar]
  62. Olofin I, et al. , 2013. Associations of suboptimal growth with all-cause and cause-specific mortality in children under five years: a pooled analysis of ten prospective studies. PLoS ONE 8 (5), e64636. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Onis Md., et al. , 2007. Development of a WHO growth reference for school-aged children and adolescents. Bull. World Health Organ. 85 (9), 660–667. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Osborne A, et al. , 2025. Trends and inequalities in stunting and underweight among children aged 0–59 months in Ghana, 1993–2022. Int. J. Equity Health 24 (1), 168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Quinn AK, et al. , 2021. Prenatal household air pollutant exposure is associated with reduced size and gestational age at birth among a cohort of Ghanaian infants. Environ. Int. 155, 106659. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Rylance J, et al. , 2015. Chronic household air pollution exposure is associated with impaired alveolar macrophage function in Malawian non-smokers. PLoS ONE 10 (9), e0138762. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Sinharoy SS, Clasen T, Martorell R, 2020. Air pollution and stunting: a missing link? Lancet Glob. Health 8 (4), e472–e475. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Soliman A, et al. , 2021. Early and long-term consequences of nutritional stunting: from childhood to adulthood. Acta Bio Medica: Atenei Parmensis 92 (1), e2021168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Spears D, et al. , 2019. The association of early-life exposure to ambient PM2. 5 and later-childhood height-for-age in India: an observational study. Environ. Health 18 (1), 62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Tan Y, et al. , 2024. Prenatal exposure to PM2. 5 and childhood body mass index growth trajectories from birth to 6 years old. Sci. Rep. 14 (1), 16936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Tickell KD, et al. , 2020. The effect of acute malnutrition on enteric pathogens, moderate-to-severe diarrhoea, and associated mortality in the Global Enteric Multicenter Study cohort: a post-hoc analysis. Lancet Glob. Health 8 (2), e215–e224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Upadhyay AK, Srivastava S, Mishra V, 2021. Does use of solid fuels for cooking contribute to childhood stunting? A longitudinal data analysis from low- and middle-income countries. J. Biosoc. Sci. 53 (1), 121–136. [DOI] [PubMed] [Google Scholar]
  73. Victora CG, et al. , 2010. Worldwide timing of growth faltering: revisiting implications for interventions. Pediatrics 125 (3), e473–e480. [DOI] [PubMed] [Google Scholar]
  74. World Health Organization (WHO), 2023. Levels and Trends in Child Malnutrition Child Malnutrition: UNICEF/WHO/World Bank Group Joint Child Malnutrition Estimates: Key findings of the 2023 edition. World Health Organization. [Google Scholar]
  75. Wylie BJ, et al. , 2017. Maternal exposure to carbon monoxide and fine particulate matter during pregnancy in an urban Tanzanian cohort. Indoor Air 27 (1), 136–146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  76. Yao M, et al. , 2022. Household air pollution and childhood stunting in China: a prospective cohort study. Front. Public Health 10, 985786. [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Younger A, et al. , 2022. Adverse birth outcomes associated with household air pollution from unclean cooking fuels in low-and middle-income countries: a systematic review. Environ. Res. 204, 112274. [DOI] [PubMed] [Google Scholar]
  78. Yucra S, et al. , 2014. Maternal exposure to biomass smoke and carbon monoxide in relation to adverse pregnancy outcome in two high altitude cities of Peru. Environ. Res. 130, 29–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  79. Zeng Z, et al. , 2020. Pm2. 5-bound PAHs exposure linked with low plasma insulin-like growth factor 1 levels and reduced child height. Environ. Int. 138, 105660. [DOI] [PubMed] [Google Scholar]
  80. Zheng W, et al. , 2013. Multilevel longitudinal analysis of sex differences in height gain and growth rate changes in Japanese school-aged children. J. Epidemiol. 23 (4), 275–279. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

1
NIHMS2184267-supplement-1.docx (53.3KB, docx)
2
NIHMS2184267-supplement-2.docx (5.1MB, docx)
3
NIHMS2184267-supplement-3.docx (72.4KB, docx)

Data Availability Statement

Data will be made available on request.

RESOURCES