Copyright 2011 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2011
Pollution from indoor coal use may impair early childhood skeletal growth to age 36 months. Because a significant proportion of the world population still uses coal indoors, the finding has public health consequences.
To examine whether indoor coal combustion for heating, which releases pollutants into the air, affects early childhood growth.
A prospective longitudinal study, with growth measurements extracted from medical records of the children's well-child care visits at age 36 months. Data were compiled from self-administered questionnaires and medical records, both completed at 2 time points: delivery and follow-up.
Teplice and Prachatice districts in the Czech Republic.
A total of 1133 children followed from birth to age 36 months.
Maternally reported use of coal for heating.
Main Outcome Measure
The z score for height for age and sex at age 36 months.
Adjusted for covariates, indoor coal use was significantly associated with a lower z score for height for age and sex at age 36 months (z score = −0.37; 95% confidence interval, −0.60 to −0.14). This finding translates into a reduction in height of about 1.34 cm (95% confidence interval, 0.51 to 2.16) for boys and 1.30 cm (95% confidence interval, 0.50 to 2.10) for girls raised in homes that used coal. The association between coal use and height was modified by postnatal cigarette smoke exposure.
Pollution from indoor coal use may impair early childhood skeletal growth to age 36 months. Because a significant proportion of the world population still uses coal indoors, the finding has public health consequences.
Use of coal for indoor heating is widely prevalent in some countries,1 exposing millions of people to indoor air pollution from coal smoke. Coal combustion emits chemicals such as fluorine, selenium, mercury, arsenic, polycyclic aromatic hydrocarbons (PAHs), sulfur dioxide, and nitrogen dioxide into the indoor air, and these chemicals may form residues on household surfaces and food.2 Often, exposures are prolonged owing to inadequate ventilation.
The period immediately after birth is marked by rapid development influenced by factors such as child's nutrition, parental anthropometrics, breastfeeding, socioeconomic environment, and exogenous exposures, eg, cigarette smoke.3- 5 Prenatal exposure to ambient PAHs and particulate matter has been linked to intrauterine growth restriction (birth weight <10th percentile by gestational age and sex)6 and shorter birth length.7 Prenatal exposure to cigarette smoke or ambient air pollution has also been associated with smaller length and head circumference at birth as well as cognitive deficit during early childhood.4,8,9 Maternal smoking during pregnancy has also been associated with short height at ages 5 and 8 years,10,11 but these studies did not consider postnatal exposure. Vik et al11 reported that children of smokers displayed complete catch-up growth in weight by age 5 years but only partial catch-up growth in height. Additionally, air pollution,12 tobacco smoke,13 and indoor combustion of coal and other solid fuels14 have been linked to several types of respiratory illnesses during childhood. Thus, in line with evidence regarding prenatal exposure, particular matter and PAH exposure in the years following birth may adversely affect postnatal development.
We therefore hypothesized a relationship between indoor coal use in postnatal life and shorter stature at age 36 months. Data available from the Childhood Health and Air Pollution (CHAP) study, also known as the Czech Early Childhood Health study, a longitudinal follow-up of a birth cohort of Czech children, provided information suitable for such an analysis.
This longitudinal study was part of the Teplice Program, launched by the Czech Ministry of Environment with assistance from the US Environmental Protection Agency. Women delivering babies in the Teplice and Prachatice districts were invited to participate in a pregnancy outcome study.6 Of about 8500 births, nurses recruited nearly 90% of mothers (n = 7502). From the pregnancy outcome study cohort, a stratified, systematic, random sample consisting of 1492 mother-infant pairs (19.9%) was recruited into the Immune Biomarker Study (IBS). The IBS participants were then contacted for the follow-up CHAP study when the child was aged 3 or 4.5 years. This follow-up for births from May 1994 through December 1998 was conducted by pediatricians in the 2 districts whose staff identified IBS children in their practice, obtained informed consent, distributed the parental questionnaires, and abstracted medical records data.
Of the 1492 mother-child pairs in the IBS, 148 (9.9%) became ineligible (moved out of the district or were not found, child or mother died, or child was put up for adoption) and 79 (5.3%) were not contacted (funding agency's decision to end fieldwork early). Of the remaining 1265 (84.8%), 48 (3.8%) refused to participate and 84 (6.6%) did not have available questionnaire data, leaving 1133 (89.6%) who participated in the follow-up study.15 Of these, information on height at age 36 months was available for 1105 children (97.5%), who therefore composed the study sample for this analysis.
Relevant variables were compiled from self-administered questionnaires and medical records from 2 time points: delivery and follow-up. The delivery questionnaire obtained reproductive histories, medical conditions, medications, smoking status, drinking and other lifestyle factors, and occupational information.16 The follow-up questionnaire obtained breastfeeding, day care, exposure to secondhand cigarette smoke, and other information. Several questions pertained to indoor coal use for cooking or heating: “In which way is your flat heated?” and “What is the primary type of fuel used for cooking?” Answer options for heating were central heating, gas furnace inside vs outside the flat, coal furnace inside vs outside the flat, electricity, or wood-burning stove. Answer options for cooking were gas, propane, electricity, coal, wood, or other. Very few used coal for cooking (0.6%), and virtually all who did use coal for cooking also used it for heating. The coal users constitute the primary exposure group for this analysis. We also created a category for wood users.
The recommended well-child care visit at age 36 months measured anthropometric factors, including height and weight. Height measured at this visit was abstracted from pediatric medical records and used to create z scores. The height-for-age-and-sex z scores were calculated using the World Health Organization's Global Database on Child Growth and Malnutrition reference data17,18 and constituted the primary outcome variable. We also calculated standardized z scores for weight at birth and age 36 months (weight for [gestational] age and sex).
We examined differences between households that did vs did not burn coal using t tests and χ2 statistics. Multivariable linear regression models were fitted to evaluate the relationship between indoor coal use and early childhood growth (z score for height). Prior to model fitting, covariates were screened as possible confounders. They were included in the models if they were predictive of the outcome and also associated with the exposure with sufficient precision (P < .15).
These covariates were maternal age at delivery (<20, ≥20 to <30, and ≥30 years), maternal prepregnancy weight (continuous), maternal height (continuous), maternal education (primary schooling, some secondary schooling, and completed secondary or higher schooling), postnatal exposure to cigarette smoke, duration of breastfeeding (never, >0 to ≤3 months, >3 to ≤6 months, >6 to ≤12 months, and >12 months), and ethnicity (dichotomized as Roma vs Czech or other Eastern or Central European). Similarly, postnatal smoke exposure, which encompassed any secondhand tobacco smoke the child might have been exposed to regularly from adults residing in the household, was dichotomized. Our focus was on postnatal exposures. To obtain estimates independent of the impact of prenatal exposures on fetal growth, we included a term for birth weight adjusted for gestational age and sex. Self-reported prenatal smoking data may not have been reliable (79.0% reported no smoking during pregnancy compared with 40.4% after pregnancy). We could not verify the well-known association of prenatal smoking with lower birth weight, which may be evidence of inaccuracies. In contrast, postnatal smoking was verified by urine cotinine level for a subsample (n = 514).
The multivariable models were adjusted for confounders as well as for the original sampling design, namely stratified, systematic, random sampling without replacement in strata defined by district of residence and year of birth. Full models were constructed using stepwise regression with P < .15 for inclusion and P > .05 for exclusion. Analysis was carried out in Stata version 10.1 statistical software (StataCorp LP, College Station, Texas).
Products of coal combustion are similar to those in cigarette smoke. We therefore investigated how the association from the combined exposures to coal and cigarette smoke on postnatal growth compared with the associations of the individual exposures. Three groups were compared, with referents having neither exposure: coal users exposed to cigarette smoke, coal users not exposed to cigarette smoke, and smokers with no coal use. Also, maternal height, a significant predictor of child's height, differed by ethnicity in this population.19 We investigated potential effect modification by ethnicity.
The study was approved by the Ethical Committee of the Regional Institute of Hygiene of Central Bohemia, Prague, Czech Republic, and by the Human Subjects Committee of the School of Medicine at the University of California, Davis.
As described previously,15 the CHAP sample is similar to the full cohort (n = 7502) on most demographic characteristics (Table 1). Owing to sampling design, the CHAP sample included more children born in 1997 or 1998 and from the Prachatice district compared with the full cohort. In the CHAP sample compared with the IBS sample, there were lower percentages of mothers younger than 20 years and infants with low birth weight or preterm infants, but none of these differences were statistically significant. Fewer were Roma or had maternal education missing in the CHAP sample.
Among study households, 10.2% used coal and 6.8% used wood; 46.8% of wood users also used other fuel sources, compared with 22.4% of coal users. A higher percentage of births in coal-using households occurred in the earlier years (1995-1996) (P = .07), resulting from a broad policy promoting use of cleaner fuels during the 1990s (Table 1). Roma children were more likely than non-Roma children to be low birth weight (28.3% vs 5.3%, respectively; P < .001) or to be born preterm (26.1% vs 7.6%, respectively; P < .001) and were more likely to have never been breastfed (26.1% vs 10.5%, respectively; P < .001). On average, Czech mothers were 7.41 cm taller and 5.42 kg heavier than the Roma mothers (P = .001). The mean maternal height was 1.08 cm higher in non–coal users compared with coal users (P = .11).
The mean height for age and sex at age 36 months was significantly higher (P = .02) for children in non-coal-using (mean z score = 0.22) vs coal-using (mean z score = −0.08) households (SD = 3.6 cm). About a third of those in the lowest decile of birth weight for gestational age and sex were also in the lowest decile for standardized weight at age 36 months. As we did not have birth length (which is often unreliable owing to measurement difficulties for newborns), we adjusted for birth weight for gestational age and sex in the regression model. The final model also controlled for district; year of birth; maternal age, height, and ethnicity; duration of breastfeeding; and postnatal secondhand smoke exposure. After adjustment, indoor coal use was significantly associated with height for age and sex at age 36 months (z score = −0.37; 95% confidence interval [CI], −0.60 to −0.14) (Table 2). As z scores represent units of 1 SD, this is equivalent to height reductions of 1.34 cm (95% CI, 0.51 to 2.16) for boys and 1.30 cm (95% CI, 0.50 to 2.10) for girls at age 36 months. We did not see any association with wood use (Table 2). Exclusion of variables with P > .05 in the full model changed the coal coefficient little (<2%).
Combined exposure to indoor coal use and postnatal cigarette smoke was more harmful than exposure to either one alone, but it was not greater than their sum. Reduction in standardized height (z score) at age 36 months for children with both exposures was −0.58 (95% CI, −0.86 to −0.31), compared with that from coal use only (z score = −0.48; 95% CI, −0.89 to −0.07) or cigarette smoke exposure only (z score = −0.25; 95% CI, −0.44 to −0.07); these are equivalent to reductions of 2.09 cm, 1.73 cm, and 0.91 cm, respectively. There was no evidence of effect modification by ethnicity.
In this study population, indoor coal use was associated with decreased stature at age 36 months, after adjusting for a range of confounders. These findings reaffirm that the negative impact of indoor air pollution from coal may extend beyond the respiratory system of children and indicate possible systemic effects. Because weight and length or height during infancy and childhood are considered to be predictors of morbidity such as obesity20 and mortality from malnutrition and infections,21 and in light of an estimated 50% of the world population using coal and solid biomass as domestic fuel,14 knowledge of such an adverse impact on child health is vital from an international child health perspective.
Wood use was not associated with height at age 36 months, possibly because almost half of the wood users also used some other fuel. Wood may have been used only occasionally in those households, whereas the vast majority of the coal users generally use it as the primary source, with only one-fifth using another fuel. Excluding those who used coal and another fuel did not change the association for coal.
Our results were controlled for standardized birth weight (z score); hence, the reduction in height at age 36 months was independent of fetal growth. We lacked birth length measurements and so could not distinguish whether the reduced height at age 36 months occurred primarily in babies of normal prenatal skeletal growth or represented a continuation of a restricted trajectory initiated prenatally. Nevertheless, as others report high correlations (r = 0.8) between height and weight at birth,22 the former seems more likely. While indoor coal and solid biomass combustions have been linked to prenatal health conditions, this relationship with postnatal growth, independent of growth reduction from prenatal exposure, appears to be a relatively novel finding.23
Although studies have concluded that there is an adverse association between ambient air pollution and prenatal growth,24 literature on postnatal air pollution exposure and postnatal growth is scant. A British longitudinal study linked air pollution (predominantly from domestic coal consumption) with reduction in height at age 7 years but not at other ages.25 The main weakness of this evidence25 was the assumption of independence of longitudinal (repeated) observations, potentially producing a misleadingly precise effect estimate. A Czech study reported a 0.7-cm reduction of height in preschool-aged children associated with a 50-μg/m3 increase in sulfur dioxide,26 which can be a by-product of high-sulfur coal combustion. The coal used in Northern Bohemia, Czech Republic, was high in sulfur.27 Another study reported a significantly smaller increase in height during a 2-year period in 9-year-old children residing in a high-pollution area compared with those living in a low-pollution area.28 A study of Chinese children reported reduced length at age 2.5 years associated with cord blood PAH-DNA adducts, which would reflect prenatal exposure.29
Coal smoke and cigarette smoke are similar in composition.30 Prenatal cigarette smoke exposure has been associated with smaller head circumference at birth,31 and children who were prenatally exposed exhibited partial catch-up growth in height but no catch-up growth in head circumference at age 5 years.11 To determine whether the observed phenomenon was due to prenatal or postnatal cigarette smoke exposure or both, one study examined child's height in relation to prenatal and postnatal smoking exposure, verified by the serum cotinine level.32 The study did not find any association between postnatal secondhand cigarette smoke exposure and height at age 5 years. However, children of those who smoked during both the prenatal and postnatal periods exhibited a 0.5-cm reduction in height at age 5 years. The authors suggested that birth weight and gestational age mediated the association because the association disappears after adjustment for these 2 variables. Our results for coal use, however, were robust to adjustment for standardized birth weight, showing that the observed association is not due to a prenatal effect. Any effect on childhood stature from prenatal exposure will in all likelihood be reflected through either proportionate or disproportionate reduction in weight and length observed at birth. As birth weight and height are strongly correlated, this adjustment was likely sufficient.
Coal smoke exposure may impede early growth through several mechanisms. Growth hormone mediates production of insulinlike growth factors (IGFs) 1 and 2, which in turn provide signals necessary for cell proliferation and differentiation throughout the body.33 The IGF-1 concentrations in blood are low at birth but increase substantially during childhood and puberty, and IGF-2 plays a more predominant role in the growth of tumor cells; however, both IGF-1 and IGF-2 are essential for embryonic development.33,34 Polycyclic aromatic hydrocarbons such as benzo[a ]pyrene, benzo[a ]anthracene, and phenanthrene have been recognized as endocrine disruptors,35 which can compromise IGF activity in specialized organs.36 There is evidence that those who were exposed to similar chemicals prenatally have reduced height and weight as teenagers.37 Also, PAHs are known to inhibit IGF and epidermal growth factors in human placenta.38 Incomplete combustion of carbon-based fuels such as coal gives rise to PAHs, which may interfere with protein synthesis for IGFs.
We were not able to quantify household exposure to indoor coal smoke, which was self-reported. Moreover, there is also the possibility of exposure misclassification from indoor exposure to outdoor air pollution through cross ventilation. Exposure depends on the type of coal and stove used as well as ventilation method.39 Unlike liquid or gaseous fuels, solid fuels are difficult to burn in conventional stoves because they require a lot of premixing with air during burning, resulting in increased emissions.40 Furthermore, indoor smoke that exits through the chimney may mix with the ambient air in the immediate vicinity, especially during phases of temperature inversion, leading to outdoor exposures from indoor combustion. Also, children generally play close to the ground, and their play activities may cause resuspension of settled particles. Quite likely, many households using coal at follow-up had also been using it before the infant's birth. Therefore, some and possibly most of the children from coal-using households also experienced prenatal exposures. These and other limitations in the data precluded investigation of an effect of timing, duration, or dose response or the presence of a threshold effect. Other limitations pertain to the measurement of height: we relied on medical records and did not impose any uniform calibration procedure across practices. However, variation in anthropometric measurements across practices is likely to be random in relation to coal use because coal users were widely distributed across all practices.
The study sample was representative of births from the 2 districts after accounting for ineligibility (9.9%) and loss to follow-up (10.4%), which includes refusal (3.8%). The sample was similar to the full pregnancy outcome study and IBS sample for baseline variables, while other variables were controlled through multivariable modeling. Thus, our results can be generalized to the full population (excluding those who moved out or were adopted) and to other populations with similar characteristics.
After confounder adjustment, we found a greater reduction in height at age 36 months in children from coal-using households with postnatal smoke exposure compared with those who were exposed to only indoor coal use or only cigarette smoke. However, the impact was not greater than expected from the sum of the 2 independent factors, which is plausible given similar constituents.41,42 Results are comparable to those from a study of 7 developing countries, which reported that maternal smoking and biofuel smoke exposure were each associated (separately) with a decrease in child's height.23 Nicotine from cigarette smoke may act directly on human growth plate chondrocytes, decreasing matrix synthesis and delaying skeletal growth.43
The literature relating breastfeeding to childhood height is inconsistent. Breastfeeding in the first few months of infancy is beneficial, but prolonged breastfeeding may have no advantage for childhood growth. Breastfeeding may even be disadvantageous when breast milk is given as a substitute for solid food in resource-deprived environments or owing to unawareness.44,45 However, other studies have observed a benefit for childhood height from both early breastfeeding46 and continued breastfeeding well into the second year.47,48 In this study, we found that children breastfed for any duration were taller than those who were never breastfed, irrespective of coal use.
Corroboration from other study populations would be desirable. Future research should examine critical time windows in gestation or early life during which children's growth is more susceptible to harmful exposures such as coal smoke. From a public health perspective, interventions such as replacing coal with cleaner fuels or, at a minimum, ensuring proper ventilation would be helpful.
In conclusion, we found that indoor coal use was associated with reduced height at age 36 months, adjusted for anthropometric and sociodemographic factors. Since half of the world population, mostly from developing countries, still relies on solid fuel for cooking14 with inadequate ventilation, this finding is of considerable public health concern.
Correspondence: Irva Hertz-Picciotto, PhD, Department of Public Health Sciences, Division of Epidemiology, MS1C, University of California, Davis, One Shields Avenue, Davis, CA 95616 (firstname.lastname@example.org).
Accepted for Publication: November 22, 2010.
Published Online: February 7, 2011. doi:10.1001/archpediatrics.2010.294
Author Contributions: Dr Hertz-Picciotto had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Sram and Hertz-Picciotto. Acquisition of data: Dostal, Sram, and Hertz-Picciotto. Analysis and interpretation of data: Ghosh, Amirian, and Hertz-Picciotto. Drafting of the manuscript: Ghosh and Hertz-Picciotto. Critical revision of themanuscript for important intellectual content: Ghosh, Amirian, Dostal, Sram, and Hertz-Picciotto. Statistical analysis: Ghosh, Amirian, and Hertz-Picciotto. Obtained funding: Dostal, Sram, and Hertz-Picciotto. Administrative, technical, and material support: Sram and Hertz-Picciotto. Study supervision: Sram and Hertz-Picciotto.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant SP/1b3/50/07 from the Czech Ministry of the Environment.
Previous Presentation: This paper was presented as a poster at the 43rd Annual Meeting of the Society for Epidemiologic Research; June 23-26, 2010; Seattle, Washington.
Ghosh R, Amirian E, Dostal M, Sram RJ, Hertz-Picciotto I. Indoor Coal Use and Early Childhood Growth. Arch Pediatr Adolesc Med. 2011;165(6):492–497. doi:10.1001/archpediatrics.2010.294