Active prenatal tobacco smoke exposure, classified using self-report or serum cotinine concentrations, was imprecisely associated with higher BMI and odds of overweight BMI at 2 or 3 years of age compared to unexposed children. Our results using serum cotinine concentrations were more suggestive of increased BMI among children with prenatal SHS exposure than the results using self-reported exposures. Self-reported secondhand tobacco smoke exposures were misclassified to a greater extent than self-reported active smoking, and accordingly, the point estimates were attenuated when compared to estimates based on serum cotinine concentrations.
Prior studies of prenatal tobacco smoke exposure and BMI have relied on self-reported tobacco smoke exposures and many studies have compared actively smoking women to non-smokers, without considering the impact SHS exposure. This misclassification may have contributed to an underestimate of the impact of active prenatal tobacco smoke exposure on childhood BMI. Our results show that self-reported tobacco smoke measures fail to accurately quantify secondhand tobacco smoke exposures and may result in biased estimates of association.
Several investigators have reported increased BMI and risk of overweight BMI among 2–4 year old children born to active smokers.7, 8, 10, 11, 37
ORs between prenatal active smoking and overweight BMI in prior studies range from 1.2 to 2.2.7, 8, 10, 37
Our estimated association between active smoking and BMI were less precise than prior estimates, similar to Oken et al. and Adams et al. (OR: 2.2),7, 8
and larger than Chen et al. and Whitaker et al. (OR: 1.2–1.5).10, 37
Consistent with our findings, Oken and colleagues did not report increased BMI among 3 year old children born to women with self-reported prenatal secondhand tobacco smoke exposure.8
However, Leary et al. reported that 10 year old children born to women whose partners smoked regularly during pregnancy had BMI 0.1 standard deviations higher than unexposed children.9
The discrepancy in results could be due to differences in the tobacco smoke exposure questionnaire, temporal or geographic variations in secondhand exposure, or the age of follow-up.
Increased weight at 2 or 3 years of age was responsible for increases in BMI among children born to women with serum cotinine concentrations consistent with secondhand and active tobacco smoke exposures. Children with prenatal tobacco smoke exposures were born lighter and shorter and grew heavier over the first 3 years of life compared to unexposed children. Attained height at 1, 2, or 3 years of age was similar between categories of prenatal tobacco smoke exposure.
Prenatal tobacco smoke exposure may influence childhood BMI by restricting fetal growth as a result of vascocontriction and hypoxemia.14–16, 38
Restricted fetal growth may lead to rapid weight gain, resulting in increased BMI and adiposity during early childhood.16, 39
In addition, tobacco smoke constituents may act on various hormonal systems that change metabolic programming.40, 41
There has also been some suggestion that prenatal tobacco smoke exposure may alter infant appetite and leptin concentrations.42–44
The results of this study should be considered in light of several limitations. Confounding due to dietary, lifestyle, or socioeconomic factors may eliminate the observed elevation in BMI among children born to women with serum cotinine concentrations indicative of SHS exposure. Women who smoke or are exposed to higher levels of SHS during pregnancy are less likely to provide optimal diets and exercise for their children.45, 46
Residual confounding may eliminate the positive association between prenatal secondhand tobacco smoke exposure and BMI at 2 and 3 years of age. We attempted to indirectly control for confounding due to dietary or lifestyle factors by including breast feeding duration and maternal BMI in our models. In addition, we controlled for other markers of socioeconomic status, including household income and maternal education which are associated with diet and exercise.47, 48
We also used HOME Inventory scores as a proxy for socioeconomic, dietary, and exercise factors not captured by other variables. Our results were not substantially different when we included HOME Inventory scores in our models.
While serum cotinine is considered an ideal biomarker for tobacco smoke exposure, it only reflects exposure over the last 2 to 3 days, which introduces the potential for exposure misclassification.49, 50
Women could have been exposed intermittently (e.g., only on weekends). However, we do not believe this is a substantial source of bias since maternal serial serum cotinine concentrations were highly correlated and we used 2 or more prenatal serum samples from each woman to quantify prenatal exposure.
Increased BMI can result from increases in fat mass or fat-free mass. Other measures of body composition, including densitometry, bioelectric impedance, dual energy x-ray absorptiometry (DXA), and isotope dilution can derive these separate components of body composition and provide more accurate quantification of excess adiposity.51
In addition, BMI is differentially correlated with direct measures of adiposity, with higher correlation among children ≥ 85th
While there are limitations to using BMI as a marker of early childhood adiposity, BMI is endorsed by the World Health Organization to measure and track body composition across the life-course.27
In addition, prior studies report that BMI measurements before 2 years of age are correlated with other direct measures of adiposity.53
Future research should examine the association between prenatal tobacco smoke exposures using direct measures of fat and fat-free mass and study children in later childhood when BMI measurements become more reflective of adult adiposity.4
A substantial proportion of women and their children dropped out of the study by 3 years of age. Attrition in our study was associated with socioeconomic factors, which are related to health and caregiving behaviors that influence childhood adiposity and tobacco smoke exposures. The association between prenatal serum cotinine concentrations and BMI was slightly strengthened when we restricted to children who completed all four follow-up visits.
Women in this study were required to be living in housing built before 1978. We do not believe this limits the generalizability of our study, since almost 70% of US housing units were built prior to 1978.54
However, we analyzed a relatively homogenous sample of mothers who come from higher socioeconomic backgrounds which may reduce the generalizability of these results to samples drawn from populations with lower SES.
These results suggest that prenatal serum cotinine concentrations may be associated with increased BMI and odds of overweight BMI at 2 and 3 years of age. Given the high prevalence of tobacco smoke exposure and detrimental health consequences of increased BMI, additional research should be conducted examining the relationship between prenatal SHS exposures and childhood BMI and adiposity. Specifically, future studies should use validated biomarkers of tobacco smoke exposure to reduce misclassification of SHS exposure and examine this association at later ages. The presented findings add to a growing body of literature suggesting that prenatal environmental insults, including secondhand tobacco smoke exposure, may impact the public’s health by influencing the risk of later life health outcomes through excess childhood BMI.55, 56