Similar Associations of Parental Prenatal Smoking Suggest Child Blood Pressure Is Not Influenced by Intrauterine Effects
Maternal smoking in pregnancy may be associated with higher offspring blood pressure; however, results of previous studies have been inconsistent and included varying confounder adjustments. We studied the association between maternal smoking in pregnancy and offspring blood pressure at 7 years in the Avon Longitudinal Study of Parents and Children, accounting for important social and environmental confounders and using partner smoking to investigate intrauterine effects. Analysis was carried out in 6509 children with maternal smoking data and 7149 children with partner smoking data. In models adjusting for child age and sex, modest differences in systolic blood pressure were observed between children of mothers who did and did not smoke during pregnancy (β=0.64 mm Hg; 95% CI: 0.09 to 1.20; P=0.02). Adjusting for all of the confounders attenuated this difference toward the null (β=0.05 mm Hg; 95% CI: −0.59 to 0.68; P=0.9), mostly because of adjustment for breastfeeding, maternal education, and family social class. Associations were similar between maternal and partner smoking with offspring systolic blood pressure (for partner smoking: β=0.62 mm Hg; 95% CI: 0.17 to 1.07; P=0.07 minimally adjusted and β=0.26 mm Hg; 95% CI: −0.36 to 0.87; P=0.4 fully adjusted), providing further evidence that differences in child blood pressure observed in minimally adjusted models are not because of a biological influence of maternal smoking on the intrauterine environment.
- blood pressure
- confounding factors (epidemiology), cohort
- maternal smoking
- prenatal exposure delayed effects
According to the theory of developmental origins of health and disease, fetal exposure to insults or undernutrition in utero alters the structure and function of organs and physiological systems in the fetus. This leads to permanent or long-lasting effects, which influence later susceptibility to disease in adult life.1 Maternal smoking during pregnancy has been associated with adverse offspring outcomes, such as low birth weight and preterm birth,2,3 reduced stature,4,5 high body mass index (BMI),6–8 and higher body fat.9 In addition, there is some evidence to suggest that exposure to maternal smoking in utero is associated with higher blood pressure in children.10–14 Determining influences of blood pressure in childhood is important in understanding the pathogenesis of heart disease, because blood pressure “tracks” during life, and children with high blood pressure are likely to become hypertensive adults.15
Findings from studies investigating the association between maternal smoking in pregnancy and offspring blood pressure in childhood have been inconsistent, with positive associations reported in some studies10–14 and null associations in others.16–18 Also, smoking behaviors tend to cluster with various social and environmental factors that relate to blood pressure and, thus, may confound the relationship between maternal smoking and offspring blood pressure in later life. Thus, detailed analyses that adjust for potential confounding factors are important to determine whether maternal smoking is associated with offspring blood pressure independent of other factors. The use of risk factors in family members as a marker of background confounding has been described previously with respect to the relationship between obesity and mortality.19 A similar approach can be adopted to analyze confounding in relation to associations with maternal smoking and offspring blood pressure by using partner smoking. We compared the association between maternal smoking and offspring blood pressure with that between partner smoking and offspring blood pressure as a test of confounding. If smoking exposure in utero results in higher offspring blood pressure, one would expect to see a much stronger association with maternal smoking than with partner smoking, because the dose of cigarette smoke to the fetus from the mother will be much larger with maternal smoking. If, however, maternal smoking is a marker of other familial behaviors, then one would anticipate that the associations would be similar for maternal and partner smoking. Therefore, the purpose of this study was to investigate the association between maternal and partner smoking during pregnancy and offspring blood pressure, taking into account a range of potential confounders using the Avon Longitudinal Study of Parents and Children (ALSPAC).
The ALSPAC is a prospective cohort study investigating the health and development of children. A full description of the methodology is available elsewhere20 and on the study Web site (www.alspac. bris.ac.uk). Briefly, pregnant women residing in 3 health districts in Bristol, England, with an expected date of delivery between April 1, 1991, and December 31, 1992, were invited to take part in the study. Of these women, 14 541 enrolled, and 13 678 had a singleton, liveborn child. Detailed information was obtained from mothers during pregnancy using self-report questionnaires, and relevant information at birth was abstracted from obstetric records and birth notifications. The entire cohort of children was invited to the ALSPAC clinic at 7 years of age, during which blood pressure measurements were taken. A total of 8297 attended the clinic. Blood pressures were available in 7638 singleton children. Of these children, information on maternal smoking during any trimester was available in 6509 children and information on partner smoking in 7149 children. Ethical approval of the study was obtained from the ALSPAC Law and Ethics Committee and 3 local research ethics committees.
Smoking During Pregnancy
Information on maternal smoking was collected from antenatal questionnaires sent to mothers at 18 and 32 weeks gestation. In the 18-week questionnaire, information was gathered on the smoking of tobacco (cigarettes, cigars, pipes, or “other”) in the first 3 months of pregnancy and in the last 2 weeks. Positive responses were grouped together to create dichotomous variables representing smoking in the first and second trimesters, respectively. In the 32-week questionnaire, mothers were asked to indicate the number of cigarettes that they were currently smoking per day. This was collapsed into a dichotomous variable to represent smoking in the third trimester. Responses from the 3 trimesters were combined to create variables for smoking at any stage during pregnancy, smoking in the first trimester only, and smoking in all 3 of the trimesters.
Information on partner smoking was collected from questionnaires sent to partners at 18 weeks gestation. Partners were asked if they had smoked regularly in the last 9 months. Mothers were also asked, in their 18-week questionnaire, if their partners smoked. Thus, partners’ smoking was taken as their own response, if available; otherwise, the mother’s response was used. Responses for partner smoking available in both mothers and partners (N=5216) indicate 94.6% agreement between mothers’ and partners’ responses in relation to partner smoking.
Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured using a Dinamap 9301 vital signs monitor. A child size cuff was used for children with an upper arm circumference of <18 cm and a small adult cuff for children with an upper arm circumference of ≥18 cm. Initial inflation was set to 130 mm Hg. Two readings of SBP and DBP were recorded, and the mean for each measure was calculated. Child’s demeanor during the session (silent, talking, or fidgeting), the time of blood pressure measurement (grouped into morning and afternoon), room temperature, and observer for the measurement were also noted. Current age of child at measurement was calculated from the date of clinic attendance and child’s date of birth.
At initial enrollment with ALSPAC, mothers were asked to provide their date of birth, height, and prepregnancy weight. From these data, mothers’ age at childbirth and maternal BMI (weight/height2, with weight in kilograms and height in meters) were calculated. The date of last menstrual period was also reported at enrollment and, together with the actual date of delivery, was used to estimate gestational age. However, if this differed by >2 weeks from the estimate calculated from an early ultrasound assessment, the latter was assumed to be the more accurate. Infant sex and birth weight were obtained from obstetric records and/or birth notifications. Details of all previous pregnancies resulting in either a live birth or stillbirth, which enabled parity to be derived, were gathered from the 18-week questionnaire. In the 32-week questionnaire, mothers were asked to record their highest education level, which was collapsed into none/Certificate of Secondary Education (national school exams at 16 years), vocational, O level (national school exams at 16 years, higher than certificate of secondary education), A level (national school exams at 18 years), or degree. Mothers also recorded their occupation, as well as their partners’, which were used to allocate them to social categories (I, II, III nonmanual, III manual, IV, and V, where I [professional] is the highest category and V [unskilled manual worker] the lowest) according to the 1991 Office and Population Censuses and Surveys standard.21 A single variable (head of household social class) was derived from the lowest social class of both partners. We used lowest social class because it has been used in previous analyses on this cohort. This measure was chosen in preference to the highest because it distributes the children more evenly across socioeconomic groups (results were similar whether lowest social class or highest social class was used; data not shown). Breastfeeding information was collected from questionnaires sent to mothers at 6 months. This information was used to derive 3 categories of breast-feeding: (1) exclusively breastfeeding beyond 2 months of age, (2) partial breastfeeding for breastfeeding that has been stopped or was nonexclusive by 2 months; and (3) never breastfed.
Means and SDs were calculated for continuous variables, with geometric means and interquartile ranges presented for variables with skewed distributions. Proportions are presented for categorical variables. Associations between confounding factors and offspring blood pressure were analyzed using linear regression. Associations between maternal or partner smoking and offspring blood pressure were assessed, adjusting for confounders using 5 models: (1) child’s sex and age at blood pressure measurement; (2) sex, age, and maternal factors (age, parity, height, and BMI); (3) sex, age, and social factors (social class and maternal education); (4) sex, age, and breastfeeding; and (5) all potential confounders. If associations remained after adjustment for all of the confounders, we planned to include further models with birth weight, gestation, and child’s current BMI as potential mediators. Simultaneous models were also built by including both maternal and partner smoking together in each of the 5 statistical models described (thus providing associations of maternal and partner smoking on blood pressure that are independent of each other, at each of the 5 levels of confounder adjustment). Results presented relate specifically to the first (minimally adjusted) and fifth (fully adjusted) statistical models. Analyses were carried out using Stata version 9.
Characteristics of the 6509 children (boys: 3281; girls: 3228) in whom data on blood pressure and maternal smoking were available are summarized in Table 1. Compared with mothers excluded because of missing data on smoking, child’s blood pressure, or confounders, mothers in our analysis were more likely to have had <2 previous pregnancies, to have a university degree, to be older at child birth, and to be taller. Compared with children excluded because of missing data, children in our analysis were more likely to have come from a family of nonmanual social class, to have been breastfed in infancy, and to have had higher birth weight and gestational age. A comparison of results in children with smoking and blood pressure data (N=6509) with those in children with parental smoking, blood pressure, and complete confounder data (N=3739) indicated that the association of maternal smoking and offspring blood pressure was the same in both analyses. This provides some reassurance that it is unlikely that missing data are biasing our results.
The number of mothers who smoked at any stage during pregnancy was 1310 (20.1%), 842 mothers (13.9%) smoked in all 3 trimesters, and 224 (4.1%) quit smoking after the first trimester. In the first trimester, 598 mothers (9.2% of the sample) smoked 1 to 9 cigarettes, 449 (6.9%) smoked 10 to 19 cigarettes, and 155 (2.3%) smoked ≥20 cigarettes. Similar proportions were observed for mothers who smoked in the second and third trimesters. Of the 7149 partners included in this analysis, 2402 (33.6%) smoked during the pregnancy.
Table 2 shows the associations of potential confounders with child blood pressure at 7 years. Child SBP at 7 years was positively associated with child’s age at blood pressure measurement, child’s current BMI, maternal prepregnancy BMI, and maternal height and inversely associated with birthweight, gestation, maternal age at child birth, maternal parity, family social class, highest level of maternal education attained, and breastfeeding in infancy. Associations were similar, although much weaker, with child DBP at 7 years.
Maternal smoking during any trimester of pregnancy was associated with shorter stature, younger age at child birth, offspring with lower birth weight, a manual family social class, nondegree education, and lower prevalence of breastfeeding but not associated with maternal prepregnancy BMI, offspring gestational age at birth, and number of previous pregnancies (data not shown). Maternal smoking in any trimester of pregnancy was associated with a modest increase in child SBP at 7 years when adjusting only for child age at blood pressure measurement and sex (Table 3; model 1). This attenuated toward the null after adjusting for all of the confounders (model 5), in particular after adjustment for maternal education and social class (model 3), and breastfeeding (model 4). In models adjusting only for child’s age and sex (model 1), further adjustment for breastfeeding resulted in the largest attenuation with a 54.7% reduction in the association. Adjusting model 1 for maternal education resulted in a reduction of 48.4% and social class a reduction of 42.2%. To determine whether breastfeeding modifies the association between maternal smoking and offspring blood pressure, formal tests for interaction were carried out. There was no evidence of an interaction between maternal smoking and breastfeeding in infancy on child blood pressure at 7 years (P>0.1, minimally and fully adjusted models). In fully adjusted models, associations remained the same when further adjustments were made for factors influencing blood pressure measurements: child’s demeanor, hour of the day, room temperature, and observer (data not shown). Because no differences in blood pressure remained after adjustment for all of the confounders, further analyses of birthweight, gestation, and child BMI as potential mediators of the association between maternal smoking and child blood pressure were not carried out. No association was observed between maternal smoking in pregnancy and child DBP at 7 years (Table 3).
Associations of partner smoking and child SBP at 7 years were similar to those observed for maternal smoking (Table 3). As with maternal smoking, partner smoking was also associated with modest increases in child SBP at 7 years in the minimally adjusted model (model 1), which attenuated considerably after adjustment for all of the confounders (model 5) and which was mostly because of adjusting for family social class and partner education (model 3). Similar results were observed using simultaneous models with both maternal and partner smoking included together in each of the different models (Table 3), that is, that modest associations of maternal and partner smoking with offspring blood pressure (minimally adjusted) subsequently attenuated toward the null after adjusting for all of the confounders and that the associations of maternal smoking and offspring blood pressure were similar to associations of partner smoking and offspring blood pressure.
The association of parental smoking on offspring blood pressure was also analyzed for the effect of both parents smoking during pregnancy (compared with both nonsmokers). As with separate maternal or partner models, modest differences in blood pressure in minimally adjusted models were observed with attenuation toward the null after full adjustment, although with slightly larger differences in blood pressure observed (β=0.88; 95% CI: 0.19 to 1.57; P=0.01 minimally adjusted and β=0.48; 95% CI: −0.53 to 1.49; P=0.3 fully adjusted).
Compared with maternal smoking in any trimester, the association with child SBP at 7 years was slightly stronger for maternal smoking in all trimesters (Table 3). However, the association between maternal smoking in all trimesters and child SBP was attenuated after adjustment for all of the confounders (model 5), again, mostly because of social factors (model 3) and breastfeeding (model 4). No differences in child SBP were observed in any model between nonsmoking mothers and mothers who smoked in the first trimester but quit thereafter (data not shown). When a dose–response was analyzed with number of cigarettes smoked in pregnancy in the first trimester, there was some evidence of a trend in the minimally adjusted model, but this attenuated toward the null after adjustment for full confounders. This was also observed for number of cigarettes smoked in the second and third trimesters (data not shown).
In this cohort of children whose mothers were pregnant during the early 1990s, there was little evidence to suggest that maternal smoking during pregnancy was independently associated with offspring blood pressure at 7 years. Although in age- and sex-adjusted models modest increases in SBP were observed in children of mothers who smoked during pregnancy, adjusting for all of the confounders attenuated this association. This attenuation was mostly because of adjustments made for breastfeeding, education, and social class (influences that have been found to relate to higher blood pressure in later life22,23), with the largest attenuation because of adjusting for breastfeeding.
Of the previous studies reporting increased blood pressure in children of mothers who smoked during pregnancy,10–14 none have adjusted for all 3 of the factors (breastfeeding, education, and social class) found in the present analysis to attenuate the observed differences in child blood pressure associated with maternal smoking. However, 2 studies did adjust for income and education,11,12 and 2 adjusted for socioeconomic position (SEP).10,13 Previous studies have also reported that maternal smoking in pregnancy is associated with elevated neonatal blood pressure in the first 3 to 4 days of life.24,25 This association was found to persist on re-examination at 4, 9, and 12 months of age but was no longer evident at 24 months.25 Thus, it is possible that maternal smoking affects offspring blood pressure in the early phase of infancy before 2 years of age but not thereafter.
To determine whether maternal smoking influences child blood pressure via biological effects on the intrauterine environment, associations with partner smoking were analyzed for comparison. Stronger associations with offspring blood pressure would be expected for maternal compared with partner smoking if maternal smoking does indeed have a biological influence on the intrauterine environment and subsequent offspring blood pressure. This has been illustrated in a previous study whereby components of offspring stature were more strongly associated with maternal smoking than with partner smoking.4 However, in our study, associations with offspring blood pressure were found to be similar for maternal and partner smoking. This provides further evidence that modest differences in child blood pressure associated with maternal smoking in minimally adjusted analyses are not likely to be a result of biological effects on the intrauterine environment and, rather, appear to be a marker for other familial factors.
Weak evidence for a modest difference in offspring blood pressure was observed where both parents smoked during pregnancy compared with both nonsmokers. Although it is possible that this may reflect a small dose–response effect of exposure to smoke in utero from both parents smoking during pregnancy, it is likely that this difference reflects residual confounding from unmeasured confounders not included in the present analysis or because of measured variables being unable to fully capture the nature of the confounding variable.
It is also possible that parents who smoke during pregnancy are likely to continue to smoke beyond the prenatal period and that exposure to cigarette smoke in infancy or childhood may have modest effects on child blood pressure. However, because no data on parental smoking are currently available in this cohort beyond the prenatal period, we were unable to explore this possibility in the present analysis. It is also possible that, whereas no differences in offspring blood pressure were detected at the age of 7 years, differences could emerge at a later age. Thus, it may be of interest for future studies to analyze blood pressure in older offspring.
In comparing associations between maternal and partner smoking, this study uses a novel way of determining whether observed associations reflect a specific effect of maternal smoking on the intrauterine environment or whether the association may be confounded by other environmental or social factors. An additional strength of this study is the availability of detailed information on potentially important social and environmental factors, in particular, breastfeeding and partner smoking, which few studies have and no other studies of maternal smoking and offspring blood pressure have adjusted for. Because breastfeeding occurs temporally after maternal smoking, it is possible that breastfeeding is on the causal pathway between maternal smoking and offspring blood pressure, for example, if mothers who smoked in pregnancy produced less breast milk and were, thus, less able to breastfeed (with this lack of breastfeeding being, in turn, related to higher blood pressure in children). If this were the case, it has been argued that the inclusion of intermediate variables in epidemiological analyses may introduce bias26; however, we are not aware of any evidence to support such a relationship, and we think it is unlikely. Breastfeeding could also modify the association between maternal smoking and offspring blood pressure. Formal tests for interactions in our analyses suggest that this is not the case in this cohort.
A strong argument for the inclusion of breastfeeding is that it is so strongly socially patterned it can be considered a proxy for unmeasured confounders related to SEP, because SEP may not be fully captured by social class and maternal education. This study demonstrates that, in addition to maternal education and social class, which studies usually adjust for, adjusting for breastfeeding is also important in determining an unconfounded association between maternal smoking and offspring blood pressure. Adjusting for breastfeeding in populations where rates of breastfeeding differ from that of the ALSPAC cohort or where the nature of the associations between breastfeeding and unmeasured socioeconomic confounders differ may not produce the same effect observed in this study. However, results from this cohort suggest that consideration of breastfeeding as a possible confounder should be included in planning future analyses of maternal smoking and offspring blood pressure.
Methodologic weaknesses regarding the reporting of maternal smoking during pregnancy in ALSPAC have been discussed previously4 but are mentioned briefly here. First, the smoking data were self-reported by mothers and partners. However, a meta-analysis of comparisons between biochemical measures and self-reported smoking found the self-reported data to be accurate.27 Second, the questions used to collect data on maternal smoking during pregnancy differed across trimesters. Third, only cigarette smoking was reported in the third trimester. However, this would have resulted in only a small number of smokers being excluded (22 mothers smoked other forms of tobacco in first trimester).
Furthermore, blood pressure was measured using an automated oscillometric device, which is ideal for large population studies but produces slightly different values compared with a sphygmomanometer. Results from the 1994 Office for Population Censuses and Surveys Dinamap Calibration Study and other previous studies indicate that, compared with sphygmomanometers, blood pressure values from the Dinamap oscillometric instrument are ≈6- to 8-mm Hg greater for SBP and within 1-mm Hg difference for DBP.28 However, the Office for Population Censuses and Surveys calibration study also indicated that the oscillometric device was highly reliable with repeat measures yielding correlation coefficients of 0.88 for SBP and 0.83 for DBP.
In conclusion, in this large British cohort of children born in the early 1990s, modest differences in blood pressure at 7 years associated with maternal smoking in pregnancy do not persist after adjustment for variables relating to SEP, namely, social class, maternal education, and breastfeeding. The similarities of associations between maternal and partner smoking with offspring blood pressure provide further evidence that modest associations observed between maternal smoking in pregnancy and offspring blood pressure do not reflect a specific effect of maternal smoking on the intrauterine environment.
Many studies have assessed the effect of maternal smoking in pregnancy on offspring health in later life. However, associations with maternal smoking may be confounded with socioeconomic factors rather than having specific effects on the intrauterine environment. We have shown that adjusting for social class, parental education, and breastfeeding adequately captures the confounding of maternal smoking by SEP. Many studies currently include adjustments for some component(s) of SEP; however, most only include 1 or 2 variables, which may not be sufficient, and residual confounding could be interpreted as in utero effects. If a biological relationship exists between breastfeeding and later blood pressure, then breastfeeding could also be an intermediate variable in the association between maternal smoking and later blood pressure. If this were the case, adjusting for breastfeeding could introduce bias into the model. Thus, further research is required to ascertain the nature of the relationship of breastfeeding with both maternal smoking and offspring blood pressure. A novel approach has also been described in this article in relation to separating the effect of maternal smoking on the intrauterine environment from the association of maternal smoking with offspring health by virtue of it being a marker of familial socioeconomic environment. The comparison of maternal and partner exposures could be extended to future studies of maternal exposures and components of offspring health.
We are extremely grateful to all of the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses. We also thank Debbie Lawlor for her comments on an earlier version of this article. This publication is the work of the authors and M.-J.A.B. and A.R.N. will serve as guarantors for the contents of this article.
Sources of Funding
The United Kingdom Medical Research Council, the Wellcome Trust, and the University of Bristol provide core support for ALSPAC. M.-J.A.B. is jointly funded by the Overseas Research Students Awards Scheme and the University of Bristol. Contents of this article represent the views of the authors and not necessarily those of the funding bodies.
- Received November 29, 2006.
- Revision received December 16, 2006.
- Accepted March 14, 2007.
Barker DJP. Mothers, Babies, and Health in Later Life. Edinburgh, United Kingdom: Churchill Livingstone; 1998.
Al Mamun A, Lawlor DA, Alati R, O’Callaghan MJ, Williams GM, Najman JM. Does maternal smoking during pregnancy have a direct effect on future offspring obesity? Evidence from a prospective birth cohort study. Am J Epidemiol. 2006; 164: 317–325.
Reilly JJ, Armstrong J, Dorosty AR, Emmett PM, Ness A, Rogers I, Steer C, Sherriff A. Early life risk factors for obesity in childhood: cohort study. BMJ. 2005; 330: 1357.
Chen A, Pennell ML, Klebanoff MA, Rogan WJ, Longnecker MP. Maternal smoking during pregnancy in relation to child overweight: follow-up to age 8 years. Int J Epidemiol. 2006; 35: 121–130.
Williams S, Poulton R. Twins and maternal smoking: ordeals for the fetal origins hypothesis? A cohort study. BMJ. 1999; 318: 897–900.
Lawlor DA, Najman JM, Sterne J, Williams GM, Ebrahim S, Davey Smith G. Associations of parental, birth, and early life characteristics with systolic blood pressure at 5 years of age: findings from the Mater-University study of pregnancy and its outcomes. Circulation. 2004; 110: 2417–2423.
Morley R, Leeson PC, Lister G, Lucas A. Maternal smoking and blood pressure in 7.5 to 8 year old offspring. Arch Dis Child. 1995; 72: 120–124.
Whincup PH, Cook DG, Geleijnse JM. A lifecourse approach to blood pressure. In: Kuh D, Ben-Shlomo Y, eds. A Lifecourse Approach to Chronic Disease Epidemiology. Oxford, United Kingdom: Oxford University Press; 2004: 218–239.
Whincup PH, Cook DG, Papacosta O. Do maternal and intrauterine factors influence blood pressure in childhood? Arch Dis Child. 1992; 67: 1423–1429.
Law CM, Barker DJ, Bull AR, Osmond C. Maternal and fetal influences on blood pressure. Arch Dis Child. 1991; 66: 1291–1295.
Bergel E, Haelterman E, Belizan J, Villar J, Carroli G. Perinatal factors associated with blood pressure during childhood. Am J Epidemiol. 2000; 151: 594–601.
Office for Population Censuses and Surveys. Standard Occupational Classification. London, United Kingdom: HMSO; 1991.
Martin RM, Ness AR, Gunnell D, Emmett P, Davey Smith G. Does breast-feeding in infancy lower blood pressure in childhood? The Avon Longitudinal Study of Parents and Children (ALSPAC). Circulation. 2004; 109: 1259–1266.
Cole SR, Hernan MA. Fallibility in estimating direct effects. Int J Epidemiol. 2002; 31: 163–165.
Bolling K. The Dinamap 8100 Calibration Study; A Survey Carried Out by Social Survey Division of Office for Population Censuses and Surveys on Behalf of the Department of Health. London, United Kingdom: HMSO; 1994.