Women exposed to tobacco smoke in utero were more likely as adults to be obese and have hypertension. For T2DM we observed slightly increased odds, however, the results were not statistically significant. The odds of developing GDM in pregnancy were also increased in women exposed to tobacco smoke in utero; to our knowledge this association has not been examined previously. Our results were similar after adjusting for birth weight and concurrent BMI.
In animals, fetal nicotine exposure results in increased adiposity and body weight, altered perivascular adipose tissue composition and function, raised blood pressure, elevated fasting serum insulin concentrations, and enhanced insulin response to a glucose challenge (Bruin et al. 2010
). These changes may have long-lasting adverse effects in the exposed offspring. The doses of nicotine used in animal experiments results in serum cotinine levels (136 ng/ml) (Bruin et al. 2010
) that are comparable to those found among moderate smokers (Eskenazi and Bergmann 1995
). In humans, other less studied constituents of tobacco smoke (Bruin et al. 2010
) might also contribute to the adverse outcomes observed.
An association of in utero
exposure to tobacco smoke with subsequent childhood obesity has been consistently observed (Oken et al. 2008
). Results from a birth cohort showed that this association became stronger with age (up to 33 years), and seemed to be independent of early-life and other adult confounding factors (Power et al. 2010
). Results from other studies, however, raise the possibility that the association may be confounded by social factors (Iliadou et al. 2010
; Leary et al. 2006
A previous meta-analysis showed a small increase in BP (adjusted β = 0.62 mmHg; 95% CI: 0.19, 1.05) among children and adolescents exposed to tobacco smoke in utero
(Brion et al. 2008
). However, among adults (~ 45 years of age) this finding tended to be null after accounting for life-time confounding factors (Power et al. 2010
In studies of in utero
tobacco smoke and T2DM or elevated percentage of glycosylated hemoglobin, the association was gone after accounting for adult adiposity and other life-time confounding factors (Power et al. 2010
; Thomas et al. 2007
). Similarly, our results suggested an association between in utero
tobacco smoke and T2DM; we also observed an attenuation of the OR after adjusting for adult BMI.
tobacco smoke was not associated with cardiovascular risk factors other than obesity in a recent study (Power et al. 2010
). However, in another study parental smoking during childhood (a surrogate for in utero
exposure), was associated with a higher prevalence of metabolic syndrome (Hunt et al. 2006
). [Metabolic syndrome is diagnosed when at least three of the following conditions are present: central obesity, elevated triglycerides, reduced high-density lipoprotein cholesterol, high blood pressure, and elevated fasting glucose (Alberti et al. 2009
).] We did not evaluate metabolic syndrome as an outcome, although women exposed to tobacco smoke in utero
were more likely to have at least two of the prepregnancy outcomes studied. Obesity, hypertension, and T2DM are chronic conditions that become more prevalent at older ages (Deshpande et al. 2008
; Ong et al. 2007
; Pi-Sunyer 2002
); as this cohort ages, the associations we observed may be estimated more precisely, perhaps becoming easier to detect as more women develop the outcomes of interest. The association of in utero
exposure to tobacco smoke with GDM may also have implications for future evaluations of the association between in utero
exposure to tobacco smoke and T2DM (GDM is a risk factor for T2DM).
In the present study and previous studies, many factors (e.g., social, demographic, and lifestyle) were associated with in utero
tobacco smoke; thus, confounding by unmeasured factors may explain some or all of the associations we identified (Donovan and Susser 2011
). And many examples exist where once confounding by lifetime factors is accounted for, the relationship weakens. However, adjustment for risk factors that are affected by the exposure (i.e., birth weight and BMI) is controversial because standard methods used to assess the direct effect of an exposure on a given outcome after controlling for an intermediate variable may result in bias (Cole and Hernan 2002
). In some studies, father’s smoking is as strongly associated with total fat mass and blood pressure in children as is mother’s smoking, which has been interpreted as evidence of confounding (Brion et al. 2007
; Leary et al. 2006
). However, among pregnant women exposed to secondhand tobacco smoke, levels of cotinine (the primary metabolite of nicotine) in fetal fluids can be higher than those found in maternal serum (Jauniaux et al. 1999
). Thus, in utero
exposure to tobacco smoke from the partner may truly have an impact on the fetus.
The estimation of a controlled direct effect of in utero
exposure to tobacco smoke on the outcomes (i.e., obesity, hypertension, T2DM, and GDM) would have required the use of causal methods and assumptions that do not necessary hold in the present setting (Cole and Hernan 2002
); therefore, we used a traditional approach. In our analyses adjusted for potential intermediate variables (e.g., BMI), we are implicitly assuming that there are no unmeasured common causes of the intermediate variable and the outcomes (i.e., hypertension, T2DM, and GDM), which is also a questionable assumption.
Our stratified analysis showed somewhat stronger associations of in utero
exposure to tobacco smoke with hypertension, T2DM, and GDM among smokers than among nonsmokers, but a stronger association with obesity among nonsmokers than among smokers. A potential explanation for the weaker association with obesity among smokers is that women who smoke tend to gain less weight as they age than do nonsmokers (CDC 2002
) or perhaps because the “effect” of in utero
exposure to tobacco smoke is additive (i.e., smokers may have other adverse exposures or behaviors related to their smoking and positively associated with obesity, thus in utero
smoking has little “effect” on their outcome).
Our results were derived from a cross-sectional analysis of self-reported early-life exposure to tobacco smoke and adult adverse outcomes. Women were unaware of the study hypothesis, so differential reporting of outcomes by exposure status seems unlikely.
Women reported their own in utero
exposure to tobacco smoke 14–47 years later. Random errors in the classification of the exposure might have occurred, causing an underestimation of the ORs. The reported exposure to maternal tobacco smoke in utero
by the adult offspring has been shown to be valid and reproducible (Cupul-Uicab et al. 2011a
; Simard et al. 2008
). In MoBa, the validity of the exposure was assessed indirectly using the birth weights of a subset of participants. In general, maternal smoking during pregnancy has been associated with an average reduction of 149 g in birth weight (Kramer 1987
); among MoBa participants the observed reduction was 181 g (Cupul-Uicab et al. 2011b
). The reproducibility of self-reported in utero
exposure to tobacco smoke in MoBa was high (weighted κ = 0.80) (Cupul-Uicab et al. 2011b
). The intensity of in utero
exposure to tobacco smoke was not ascertained in MoBa; therefore we could not evaluate a dose–response relationship. If women who experienced very intense exposure to maternal tobacco smoke in utero
were underrepresented in our study, we may have underestimated the ORs. Information on in utero
exposure tobacco smoke by trimester as well as the number of mothers who stopped smoking during pregnancy was not ascertained, limiting our ability to assess critical windows of exposure. We were unable to discriminate between in utero
tobacco smoke and childhood exposure; however, it is likely that women whose mothers smoked while pregnant with them (i.e., exposed in utero
) also experienced continuous exposure during childhood (Weinberg et al. 1989
). Information on childhood exposure to tobacco smoke from both parents as well as paternal smoking during the mother’s pregnancy was not ascertained in MoBa. Thus, confounding by childhood exposure to tobacco smoke from parents cannot be ruled out.
The outcomes studied were questionnaire-based or ascertained from the MBRN; however, we were able to assess the validity of the GDM within the MoBa cohort. The positive predictive value (88%) of the MBR against medical records in the present study was similar to what was reported previously for the MBRN in 1998 (89%, validated against medical records) before changes were introduced to the MBRN form (Stene et al. 2007
). Self-reported height and weight give accurate estimates of BMI among adults (Brunner Huber 2007
); and agreement of medical records with self-reported hypertension and diabetes in other populations has been moderate to high (Okura et al. 2004
). In addition, known predictors of the outcomes were confirmed in the present study. For example, BMI was negatively associated with education (); hypertension and diabetes (T2DM and GDM) were positively associated with prepregnancy BMI (data not shown). In utero
exposure to smoking was not associated with type 1 diabetes mellitus (data not shown), further supporting the specificity and validity of the diabetes outcomes. Nonetheless, random errors in the classification of the outcomes may have occurred, leading to an underestimation of the associations.
Women who participated in MoBa are not a representative sample of pregnant women from Norway; the prevalence of exposures and outcomes were biased. For example, the prevalence of smoking at the end of the pregnancy in Norway was 10.8% (MBRN; between 2000 and 2006) and the corresponding prevalence in MoBa was 6.1% (Nilsen et al. 2009
). We expected an underestimation of the prevalence of in utero
exposure to tobacco smoke in our sample because women who smoke as adults were more likely to report in utero
exposure (data not shown). However, previous analyses suggest that bias in the estimated parameters due to self-selection is negligible when evaluating exposure–outcome associations in the cohort (Nilsen et al. 2009
); this tendency may hold when assessing early life exposures in relation to adult outcomes.