SEARCH

SEARCH BY CITATION

Keywords:

  • Smoking during pregnancy;
  • children;
  • fat mass;
  • lean mass;
  • partner smoking

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Objective: Maternal smoking during pregnancy has been shown to be associated with obesity in the offspring, but findings have been based mainly on BMI, which is derived from height and weight. This study examined the association between maternal and partner smoking during pregnancy and offspring total fat, truncal fat, and lean mass in childhood.

Research Methods and Procedures: Analysis was based on 5689 white singletons born in 1991–1992 and enrolled in the Avon Longitudinal Study of Parents and Children, with maternal smoking data recorded for at least one trimester in pregnancy and their own body composition assessed by DXA at mean age 9.9 years.

Results: Smoking at any time during pregnancy was associated with higher offspring BMI [0.18 (95% confidence interval, 0.12 to 0.25) standard deviation units] and total fat mass [0.17 (95% confidence interval, 0.12 to 0.23) standard deviation units], after adjustment for age, sex, height, and height squared for total fat mass. These associations were not attenuated by adjustment for the confounding factors that were measured. Maternal smoking was also associated with lean mass and, to a lesser extent, truncal fat mass. Associations with partner's smoking were in the same direction but weaker than those of the mother's for all outcomes.

Discussion: Maternal smoking at any time during pregnancy is associated with higher offspring total fat mass at mean age 9.9 years. However, as the associations with partner smoking were only a little weaker than those with maternal smoking, confounding by social factors rather than a direct effect of maternal smoking is a possible explanation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Childhood obesity is a major health problem in developed nations, and in the United Kingdom, rates have more than doubled during a recent 10-year period (1). Dietz (2) suggested that there are three critical time periods for the development of obesity: the prenatal period, the period of adiposity rebound, and adolescence. However, Dietz and Gortmaker (3) concluded more recently that the relevance of these critical periods to the prevalence of adult obesity is still uncertain. For example, they are less convinced that birthweight per se is important, as both high (4) and low birthweight (5) have been shown to be associated with increased obesity. Furthermore, subsequent analyses (for example, Reference (6) have highlighted the strong associations between early measures of growth and later obesity, suggesting that infancy and, in particular, rapid growth within infancy may also be a critical determinant of obesity risk.

Previous studies have identified a number of genetic, intrauterine, environmental, and lifestyle factors that may be associated with childhood obesity. For example, increased prevalences of overweight and obesity in the offspring if the mother smoked during pregnancy have been demonstrated (6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16). However, these studies were all based on BMI, which is known not to be a good measure of adiposity in children, as it cannot distinguish between fat and lean mass (17); it has been shown that, although a high BMI is a good indicator of excess fat mass, BMI differences among thinner children may be largely attributable to fat-free mass (18). In addition, because all except one study used dichotomized outcomes resulting in loss of information and reduced power, the associations may actually be stronger than those reported (7). Some studies also included skinfold measurements in the offspring (7, 13, 14), which are a more direct measure of fatness but are prone to measurement error. The study reported by Vik et al. (7) was the only study to use continuous measurements. To our knowledge, no studies have investigated associations between maternal smoking and fat distribution or lean mass in the offspring. In addition, none have examined associations with paternal smoking to check whether maternal smoking is of specific importance.

We, therefore, used direct measures of total fat, truncal fat, and total lean mass in the offspring, obtained with DXA at approximately age 9 years, to better describe associations across the continuum with maternal and partner smoking during pregnancy, in a large contemporary cohort of children.

Research Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Study Population

The Avon Longitudinal Study of Parents and Children (ALSPAC)1 is a population-based study investigating environmental and other factors that affect the health and development of children. The study methods are described in detail elsewhere (19) and on the study website (www.alspac.bris.ac.uk).

In brief, pregnant women living in three health districts centered in Bristol, England, who had an expected date of delivery between the start of April 1991 and end of December 1992 were eligible. There were 14,541 women enrolled in the study, and of these, 11,211 had a white singleton liveborn child.

Detailed information was obtained from the mother (about herself and her child) and her partner by means of questionnaires. A randomly selected 10% subsample of children (the Children in Focus subgroup), ranging in age from 4 months to 5 years, were invited to attend regular research clinics, where detailed physical examinations were undertaken. From age 7 years onward, the whole cohort of children was invited to attend regular research clinics. After the study's restriction to white singleton live births, maternal smoking data for at least one trimester in pregnancy were recorded for 10,282 children, and the examination at the 9-year clinic was attended by 6470, of whom 6160 had DXA data recorded; this allowed 5689 children with data on both maternal smoking and 9-year DXA variables to be used for the analysis. Ethical approval of the study was obtained from the ALSPAC Law and Ethics Committee and Local Research Ethics Committees.

Smoking during Pregnancy

In the 18-week antenatal questionnaire, the mother was asked whether she smoked tobacco 1) in the first 3 months of pregnancy and 2) in the past 2 weeks. Positive responses (cigarettes, cigars, pipes, or “other”) were grouped together to create dichotomous variables to represent smoking in the first and second trimesters, respectively. In the 32-week antenatal questionnaire, the mother was asked how many cigarettes she was currently smoking per day, and this was categorized into a dichotomous variable to represent smoking in the third trimester. Responses from the three trimesters were combined to create a variable for any smoking during pregnancy.

The number of times the mother smoked per day was recorded for the first 3 months of pregnancy and also the past 2 weeks in eight categories (0, 1 to 4, 5 to 9, etc., up to 30+ times), in the 18-week antenatal questionnaire. This information, along with the current number of cigarettes smoked per day from the 32-week antenatal questionnaire, was used to derive the number smoked per day in each of the first, second, and third trimesters (grouped as none, 1 to 9, 10 to 19, and 20+ times).

In the 18-week antenatal questionnaire sent to the partner, he was asked whether he had smoked regularly in the past 9 months. In the 18-week antenatal questionnaire sent to the mother, she was also asked whether her partner smoked. Partner's smoking was, therefore, taken as his own response if available (95% agreement with maternal response where both sets of data were available); otherwise, the mother's response was used.

Offspring Body Composition

Height was measured with shoes and socks removed using a Harpenden stadiometer (Holtain, Ltd., Crymych, Pembs., United Kingdom), and weight was measured using a Tanita TBF 305 body fat analyzer and weighing scales (Tanita United Kingdom, Ltd., Yewsley, Middlesex, United Kingdom). BMI was calculated as weight (in kilos) divided by height squared (in meters). Total fat, central fat, and lean mass were measured using a Lunar Prodigy DXA scanner (GE Medical Systems Lunar, Madison, WI). The scans were visually inspected and realigned where necessary. Truncal fat mass was estimated using the automatic region of interest that included chest, abdomen, and pelvis.

Potential Confounders

Potential confounders were those that were shown to be predictive of obesity at age 7 years in this cohort (6). Explanations of maternal height, prepregnancy BMI, age parity, education, and maternal/partner social class are provided elsewhere (20). At enrollment, the mother's partner was also asked to record his height and weight, which were used to calculate BMI. The date of the last menstrual period, as reported by the mother at enrollment, and the actual date of delivery were used to estimate gestation; if there was a discrepancy of more than 2 weeks between the menstrual-based estimate and one from an early ultrasound scan, the latter was used instead.

From the 6-month postnatal questionnaire, a variable was derived for exclusive breastfeeding, coded as exclusive breastfeeding beyond 2 months of age, partial breastfeeding (breastfeeding had been stopped or was non-exclusive by 2 months), and never breastfed. Exclusive breastfeeding was defined as no solids, milk formulas, or other drinks, except vitamins, minerals, medicines, and/or water (note that including infants who had ingested water is not consistent with the World Health Organization's definition of exclusive breastfeeding). The mother was also asked to record the age in months her child was introduced to solids, which was grouped into ≤2, 3 to 4, and ≥5 months of age. Infant sex and birthweight were recorded in the delivery room and abstracted from obstetric records and/or birth notifications. In the 30-month questionnaire, the mother was asked how much time her child spent asleep at night (grouped into <10.5 or ≥10.5 hours), and in the 38-month questionnaire, she was asked how much time they spent watching television per week (grouped into ≤8 hours or >8 hours).

From the regular measurements made on the Children in Focus subgroup, the following variables, based on the findings of Reilly et al. (6), were derived: weight gain during infancy, calculated as the 12-month weight minus birthweight; weight standard deviation (SD) scores at 8 and 18 months (quartiles); rapid growth, defined as a weight gain of at least 0.67 SD unit in the first 2 years [derived variable grouped into catch-down, no change, and catch-up; see Reilly et al. (6) for further explanation of this variable]; and early adiposity rebound, based on the change in BMI up to 60 months (grouped as: by 43 months, by 61 months, and after 61 months).

A puberty questionnaire was filled in by the mother or other caregiver when the child was approximately 9 years old; it included questions on developmental stage (21). Pubertal stage for boys was based on pubic hair development and for girls was based on the most advanced stage for pubic hair and breast development. Data were used only if the puberty questionnaire was administered within 16 weeks of the DXA scan; 74% of the children had puberty data, which reduced to 64% after this restriction was imposed.

Statistical Analysis

Means and SDs were calculated for continuous variables, and proportions were calculated for categorical variables. Further analysis was based on internally derived SD scores for BMI, total fat, truncal fat, and total lean mass to allow comparison of the regression coefficients across outcome measures. These were calculated by subtracting the mean from the individual's value, then dividing by the SD; the mean and SD were based on the whole sample. As BMI, total fat, and truncal fat had skewed distributions, logged variables were used for calculation of the SD scores.

The associations between potential confounding factors and the offspring outcomes were assessed using linear regression, as were relationships between potential confounders and the maternal smoking variables Associations between each maternal smoking variable and each offspring outcome were examined after adjustment for: sex and age of the child at the time of the DXA scan (Model 1), plus maternal factors (age, parity, height, BMI), partner factors (height, BMI), social factors (social class, maternal education), and infant feeding factors (breastfeeding and age at introduction of solids) (Model 2), plus birthweight and gestation (Model 3). Additionally, adjustment was made for the early life risk factors for childhood obesity identified by Reilly et al. (6) that were not already included (nighttime sleep duration at 30 months, television viewing at 38 months, weight gain during infancy, weight SD score at 8 and 18 months, catch-up growth, and early adiposity rebound), although this substantially reduced the numbers on which the models were based. All measures except BMI were adjusted for height and height squared to take into account differences in stature (there was evidence of quadratic relationships with height). Models for truncal fat were also presented with and without adjustment for total fat mass to compare whether any observed associations were similar and to see whether the association observed between maternal smoking and truncal fat mass was independent of total fat mass. Associations in male and female subjects separately were compared by including interaction terms for sex and smoking variables in the models. Analyses were repeated after restriction to all boys and pubertal Stage 1 and 2 girls. To compare the effect sizes for maternal and partner smoking, Models 1, 2, and 3 were fitted for any maternal smoking during pregnancy, restricted to those with partner data available, and for partner smoking instead of maternal smoking. Model 1 was also fitted to include both maternal and partner smoking variables. All analyses were performed using Stata software, version 8 (StataCorp, College Station, TX).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Geometric mean BMI was 17.5 [interquartile range (IQR), 15.7 to 19.0] kg/m2, total fat mass was 7.3 (IQR, 4.7 to 11.0) kg, and truncal fat mass was 2.7 (IQR, 1.7 to 4.4) kg. Mean lean mass was 24.6 (IQR, 22.3 to 26.5) kg. The percentage of mothers who had smoked in at least one trimester of pregnancy was 19.8% (18.1% in the first, 14.0% in the second, and 14.3% in the third trimester).

All of the potential confounders are summarized in Table 1. Children who attended the examination did not differ from those who did not with respect to their mother's and her partner's BMI and also their own birthweight and gestation. However, their mothers were slightly taller and older and were more likely to have smoked, their mothers’ partners were taller, and the children were more likely to have come from more affluent and better-educated families, to have been breastfed, to have been introduced to solids at 3 to 4 months of age as opposed to earlier or later, to be female, and to have no older siblings.

Table 1.  Summary of potential confounders for 5689 white singleton children enrolled in ALSPAC with information on maternal smoking in pregnancy and DXA data at mean age 9.9 years
 No.Mean (standard deviation) or %
  1. ALSPAC, Avon Longitudinal Study of Parents and Children; CSE, certificate of secondary education. Education levels and social classes are ranked from lowest to highest. Education level O refers to school exams taken at age 16 years. Education level A refers to school exams taken at age 18 years.

Child age (years)56899.9 (0.3)
Child height (cm)5689139.7 (6.3)
Maternal age (years)568929.2 (4.5)
Maternal height (cm)5509164.2 (6.6)
Maternal BMI (kg/m2)526222.9 (3.7)
Partner height (cm)4258176.4 (6.8)
Partner BMI (kg/m2)422025.2 (3.3)
Birthweight (kg)56243.4 (0.5)
Gestation (weeks)568939.6 (1.7)
Sex (%)  
 Male280049.2
 Female288950.8
Parity (%)  
 Primapara254645.5
 Multipara305154.5
Social class (%)  
 V1853.3
 IV73713.2
 III Manual150227.0
 III Non-manual146726.3
 II147826.5
 I2013.6
Maternal education (%)  
 None/CSE70012.3
 Vocational4758.4
 O levels202835.7
 A levels153927.1
 Degree93716.5
Breastfed (%)  
 Exclusive183234.4
 Partial263049.4
 Never86416.2
Age at introduction to solids (%)  
 ≤2 months75013.8
 3 to 4 months445682.2
 ≥5 months2133.9

All of the potential confounders were associated with at least one offspring outcome (Table 2). Mothers who smoked at any time during pregnancy were more likely to be younger, shorter, less educated, and from lower social classes, not to have breastfed, to have introduced their child to solids earlier, to have had shorter partners, and to have given birth to lighter babies (p ≤ 0.01 for all). There were no differences in maternal and partner BMI, parity, gestation, and sex of the child according to smoking status (p ≥ 0.3 for all).

Table 2.  Univariate regressions of offspring BMI, total fat, truncal fat, and total lean mass at mean age 9.9 years on potential confounders
  BMI (SD score) Total fat (SD score) Truncal fat (SD score) Total lean (SD score)
  1. SD, standard deviation; CI, confidence interval; CSE, certificate of secondary education. p values for trend given if more than two categories. Education level O refers to school exams taken at age 16 years. Education level A refers to school exams taken at age 18 years.

Potential confounderFactor testedβ95% CIpβ95% CIpβ95% CIpβ95% CIp
Age (years) 0.290.21, 0.37<0.0010.320.24, 0.41<0.0010.310.23, 0.39<0.0010.630.55, 0.71<0.001
Sex (vs. male)Female0.120.07, 0.18<0.0010.560.51, 0.61<0.0010.570.52, 0.62<0.001−0.62−0.67, −0.57<0.001
Height (cm) 0.0500.046, 0.054<0.0010.0690.066, 0.073<0.0010.070.06, 0.07<0.0010.130.12, 0.13<0.001
Maternal age (years) −0.01−0.01, −0.0030.002−0.01−0.02, −0.0040.001−0.01−0.02, −0.005<0.0010.01−0.0002, 0.010.06
Parity (vs. primapara)Multipara0.01−0.04, 0.060.7−0.06−0.11,−0.0050.03−0.04−0.10, 0.010.09−0.02−0.07, 0.040.6
Maternal height (cm) 0.0001−0.004, 0.0040.960.01−0.01, 0.01<0.0010.0090.005, 0.013<0.0010.040.03, 0.04<0.001
Maternal BMI (kg/m2) 0.080.08, 0.09<0.0010.070.06, 0.08<0.0010.070.06, 0.08<0.0010.040.04, 0.05<0.001
Partner height (cm) −0.002−0.01, 0.0030.40.0100.005, 0.014<0.0010.010.004, 0.01<0.0010.0310.026, 0.035<0.001
Partner BMI (kg/m2) 0.080.07, 0.09<0.0010.070.06, 0.08<0.0010.070.06, 0.08<0.0010.060.05, 0.06<0.001
Social class (vs. V)IV0.190.03, 0.35 0.190.03, 0.35 0.180.02, 0.34 0.220.06, 0.38 
 III Manual0.12−0.03, 0.27 0.11−0.04, 0.26 0.09−0.06, 0.24 0.180.02, 0.33 
 III Non-manual0.07−0.09, 0.22 0.07−0.08, 0.22 0.04−0.12, 0.19 0.220.07, 0.37 
 II0.03−0.12, 0.18 0.03−0.12, 0.18 −0.004−0.16, 0.15 0.210.06, 0.36 
 I−0.09−0.28, 0.11<0.001−0.05−0.25, 0.150.001−0.08−0.28, 0.11<0.0010.19−0.003, 0.390.1
Maternal education (vs. CSE/none)Vocational0.10−0.11, 0.13 −0.02−0.13, 0.10 −0.02−0.14, 0.10 0.07−0.04, 0.19 
 O levels−0.09−0.18, −0.01 −0.06−0.15, 0.02 −0.08−0.16, 0.01 0.01−0.08, 0.09 
 A levels−0.12−0.21, −0.04 −0.07−0.16, 0.02 −0.09−0.18, −0.0004 0.05−0.04, 0.14 
 Degree−0.27−0.36, −0.17<0.001−0.21−0.31, −0.11<0.001−0.24−0.34, −0.15<0.0010.08−0.02, 0.180.1
Breastfed (vs. exclusive)Partial0.160.10, 0.22 0.110.05, 0.17 0.110.05, 0.17 0.070.01, 0.13 
 Never0.230.15, 0.31<0.0010.170.09, 0.25<0.0010.180.10, 0.26<0.001−0.01−0.09, 0.070.6
Age introduced to solids (vs. ≤2 months)3 to 4 months−0.13−0.20, −0.05 −0.08−0.16, −0.001 −0.08−0.16, −0.003 −0.20−0.28, −0.13 
 ≥5 months−0.25−0.40, −0.10<0.001−0.15−0.31, −0.0010.02−0.16−0.31, −0.010.02−0.32−0.47, −0.17<0.001
Birthweight (kg) 0.260.21, 0.31<0.0010.190.14, 0.24<0.0010.150.09, 0.20<0.0010.520.48, 0.58<0.001
Gestation (weeks) 0.020.001, 0.030.040.01−0.0003, 0.030.060.01−0.003, 0.030.10.01−0.01, 0.020.2

Table 3 shows the associations between smoking during pregnancy and each of the offspring outcomes. For interpretation of the regression coefficients, SDs were 0.15 kg/m2, 0.57 kg, and 0.68 kg on the logarithmic scale for BMI, total fat, and truncal fat mass, respectively, and 3.19 kg for lean mass. After minimal adjustment [age and sex (Model 1) plus height and height squared], smoking at any time during pregnancy was associated with an increase in both offspring BMI and total fat mass of similar magnitude. There was also a clear association with increased lean mass, although the effect size was less than one-half the effect sizes of BMI and total fat mass. There was an association between maternal smoking and truncal fat mass, but this association was much weaker in models that adjusted for total fat mass. None of the associations except those with truncal fat mass (in the model adjusted for total fat mass) were attenuated by adjustment for maternal, partner, social, and infant feeding factors (Model 2) or, additionally, birthweight and gestation (Model 3); if anything, some were slightly strengthened. In the subgroup of 358 children (in the Children in Focus 10% subsample of the main cohort) with data available for all of the early life risk factors identified by Reilly et al. (6), regression coefficients were further reduced by 21% for BMI and fat mass and 11% for lean mass if full adjustment was made. Similar results were seen if the smoking data were analyzed for each trimester separately (Table 3).

Table 3.  Regressions of offspring BMI, total fat, truncal fat, and total lean mass at mean age 9.9 years on maternal smoking variables (any smoking)
 BMI (SD score) Total fat (SD score) Truncal fat (SD score) Truncal fat (SD score)*Total lean (SD score)
  • SD, standard deviation; CI, confidence interval. Model 1 was adjusted for sex and child's age at DXA scan; Model 2 was additionally adjusted for maternal, partner, social, and infant feeding factors. Model 3 was additionally adjusted for birthweight and gestation. Total fat, truncal fat, and total lean were adjusted for height and height squared.

  • *

    Truncal fat adjusted for total fat in all models.

Smoking variable and modelβ95% CIpβ95% CIpβ95% CIpβ95% CIpβ95% CIp
Smoking any trimester               
 Model 1 (N = 5689)0.180.12, 0.25<0.0010.170.12, 0.23<0.0010.190.14, 0.25<0.0010.020.003, 0.050.030.070.04, 0.11<0.001
 Model 2 (N = 3664)0.200.12, 0.28<0.0010.170.10, 0.24<0.0010.190.12, 0.26<0.0010.02−0.01, 0.050.10.080.03, 0.130.001
 Model 3 (N = 3621)0.240.16, 0.32<0.0010.190.12, 0.26<0.0010.200.12, 0.27<0.0010.02−0.01, 0.050.20.100.05, 0.15<0.001
Smoking 1st trimester               
 Model 1 (N = 5671)0.190.12, 0.25<0.0010.170.11, 0.23<0.0010.190.13, 0.25<0.0010.02−0.003, 0.040.10.070.03, 0.11<0.001
 Model 2 (N = 3664)0.210.13, 0.29<0.0010.180.11, 0.25<0.0010.200.12, 0.27<0.0010.02−0.01, 0.050.10.080.03, 0.130.001
 Model 3 (N = 3621)0.250.16, 0.33<0.0010.190.12, 0.27<0.0010.200.13, 0.28<0.0010.02−0.01, 0.050.20.110.06, 0.16<0.001
Smoking 2nd trimester               
 Model 1 (N = 5671)0.190.11, 0.26<0.0010.190.12, 0.25<0.0010.210.15, 0.28<0.0010.030.01, 0.060.020.070.03, 0.110.001
 Model 2 (N = 3664)0.200.11, 0.29<0.0010.170.09, 0.26<0.0010.190.11, 0.27<0.0010.03−0.0004, 0.070.050.080.02, 0.130.01
 Model 3 (N = 3621)0.240.15, 0.34<0.0010.190.11, 0.27<0.0010.200.11, 0.28<0.0010.03−0.003, 0.070.070.100.05, 0.16<0.001
Smoking 3rd trimester               
 Model 1 (N = 5636)0.190.12, 0.26<0.0010.190.12, 0.25<0.0010.210.15, 0.27<0.0010.02−0.003, 0.050.080.060.02, 0.100.003
 Model 2 (N = 3651)0.220.13, 0.32<0.0010.180.10, 0.26<0.0010.200.12, 0.28<0.0010.01−0.02, 0.050.40.080.03, 0.130.004
 Model 3 (N = 3608)0.280.18, 0.37<0.0010.200.11, 0.28<0.0010.210.12, 0.29<0.0010.01−0.02, 0.040.60.110.06, 0.17<0.001

Findings were similar to the above for smoking at any time during pregnancy if restriction was made to those with complete confounder information (data not shown) and if only those in early puberty (153 Stage 3 girls excluded) were used (data not shown). When the sexes were analyzed separately, stronger associations were seen in girls for all outcomes except truncal fat mass when adjusted for total fat mass. However, the difference reached conventional significance for BMI only after adjusting for age, whereby the increase was 0.26 (95% confidence interval, 0.16 to 0.35) SD units in girls compared with 0.10 (95% confidence interval, 0.02 to 0.20) SD units in boys if the mother smoked, and the statistical evidence for an interaction was not strong (p = 0.02), especially considering the large number of tests that had been performed.

Five hundred and nine women smoked 1 to 9 times, 381 smoked 10 to 19 times, and 117 smoked at least 20 times a day. For BMI and, particularly, total fat mass and truncal fat mass, there were suggestions of quadratic relationships, with the greatest increase in outcome when the mother was smoking 10 to 19 times per day (data not shown). For lean mass, the greatest increase in outcome was associated with the mother smoking >20 times per day (data not shown). These associations were stronger after adjustment for potential confounders. Findings for the first and second trimesters were similar.

Using the 5615 mother-partner pairs for which both had smoking data recorded, 33.0% of the partners smoked, and of these, 25.0% smoked when the mother did not. Table 4 shows the associations between maternal smoking and partner smoking for each of the offspring outcomes. Effects sizes for partner smoking were slightly smaller than those seen for maternal smoking (Table 3), and they were attenuated more after adjustment for potential confounders. In models in which both partner and maternal smoking were fitted, associations with outcomes were attenuated slightly for both maternal and paternal smoking

Table 4.  Regressions of offspring BMI, total fat, truncal fat, and total lean mass at mean age 9.9 years on maternal and partner smoking
 BMI (SD score)Total fat (SD score)Truncal fat (SD score)Truncal fat (SD score)*Total lean (SD score)
  • SD, standard deviation; CI, confidence interval. Model 1 was adjusted for sex and child's age at DXA scan. Model 2 was additionally adjusted for maternal, partner, social, and infant feeding factors. Model 3 was additionally adjusted for birthweight and gestation. Total fat, truncal fat, and total lean adjusted for height and height squared.

  • *

    Truncal fat adjusted for total fat in all models.

Smoking variable and modelβ95% CIpβ95% CIpβ95% CIpβ95% CIpβ95% CIp
Maternal smoking (any trimester) in subgroup with partner smoking recorded               
 Model 1 (N = 5615)0.170.11, 0.24<0.0010.170.11, 0.22<0.0010.190.13, 0.24<0.0010.030.00, 0.050.030.070.03, 0.10<0.001
 Model 2 (N = 3649)0.200.12, 0.28<0.0010.180.11, 0.25<0.0010.190.12, 0.26<0.0010.02−0.01, 0.050.10.080.03, 0.120.002
 Model 3 (N = 3606)0.240.16, 0.32<0.0010.190.12, 0.26<0.0010.200.12, 0.27<0.0010.02−0.01, 0.050.20.100.05, 0.15<0.001
Partner smoking               
 Model 1 (N = 5615)0.150.09, 0.20<0.0010.120.07, 0.17<0.0010.130.09, 0.18<0.0010.01−0.01, 0.030.40.040.01, 0.080.004
 Model 2 (N = 3649)0.110.05, 0.180.0010.080.02, 0.140.0060.080.03, 0.140.005−0.002−0.03, 0.020.90.040.01, 0.080.03
 Model 3 (N = 3606)0.110.05, 0.180.0010.080.02, 0.140.010.080.03, 0.140.005−0.004−0.03, 0.020.70.050.01, 0.080.02
Maternal and partner smoking (Model 1: simultaneous)               
 Mother0.130.06, 0.20<0.0010.130.07, 0.20<0.0010.150.09, 0.21<0.0010.030.001, 0.050.040.060.02, 0.100.004
 Partner0.110.05, 0.17<0.0010.080.03, 0.130.0010.090.04, 0.14<0.001−0.0001−0.02, 0.020.990.03−0.004, 0.060.09

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

This study, based on a large, contemporary cohort, is the first, to our knowledge, to examine associations between smoking in pregnancy and directly measured total fat, truncal fat, and total lean mass. We have demonstrated increases in offspring fat mass and, to a lesser extent lean mass, if the mother smoked during pregnancy.

Our findings for both BMI and total fat mass as continuous variables confirm what has already been shown in the literature for BMI and, in a few studies, skinfolds, both of which are generally used as dichotomous variables (6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16). Although our findings are consistent with the hypothesis that events occurring during the prenatal period seem to program the risk of later obesity (3), other explanations, discussed below, are possible.

Associations between maternal smoking and obesity may seem paradoxical, as it has been well established that smoking in pregnancy is associated with reduced offspring birthweight (see, for example, Reference (22). Possible explanations for these associations include: 1) mothers who smoke may increase feeding in infancy to help their child overcome their initial birthweight deficit; 2) nicotine acts as an appetite suppressant, so an infant exposed to nicotine in utero may demand more feeding when no longer exposed to nicotine postnatally; this “programming” of regulation of appetite has already been demonstrated in a primate study (23); 3) children exposed to prenatal smoking are more likely to be exposed to postnatal passive smoking; 4) the diets of smokers differ from those of non-smokers, so it is likely that the diets of children of smokers differ from those of non-smokers’ children (24, 25); 5) physical activity levels may be lower in the children of smokers (26).

Our study was the first to assess relationships between smoking in pregnancy and offspring lean mass, and we found increases in lean mass if the mother smoked. It is likely that maternal smoking is associated with both fat and lean mass in the offspring. However, it is possible that associations with lean mass are simply a reflection of associations with fat mass, as larger children will have more fat and lean mass; the correlation between these two components in our data was 0.39 (p < 0.001).

Our study was also the first to investigate associations between maternal smoking and offspring fat distribution. Although there was a suggestion of an increase in truncal fat mass if the mother smoked, associations were much weaker in models that adjusted for total fat mass. Hence, it seems that it is the total amount of fat, rather than the fat distribution, that is adversely influenced by maternal smoking. We found stronger associations between maternal smoking and both fat and lean mass in girls, compared with boys, although differences did not generally reach significance. Vik et al. (7) also found stronger associations in girls when considering the effect of smoking at the time of conception on offspring ponderal index and subscapular and triceps skinfolds at age 5 years, although no formal interaction tests were presented. However, Toschke et al. (10, 15) found no sex differences, and no other studies reported separate analyses for male and female subjects, so further investigation of this issue may be required.

It is likely that our results are due to confounding, as smoking is socially patterned. We have adjusted for a wide range of confounders, and adjustment has little effect on the regression estimates in general. However, we found that associations with partner's smoking were only a little weaker than those with maternal smoking; although adjustment for potential confounders reduced the associations, they still remained. These associations may have a biological basis, through passive smoking, but it is likely that residual confounding will, at the least, contribute to the association between maternal smoking and offspring body composition. It is likely that maternal smoking will be more strongly related to potential unmeasured confounders—such as additional aspects of diet and activity patterns—than to partner smoking, given the general tendency for infants and children to spend more time with the mother than with her partner. Thus, the somewhat greater magnitude of association with maternal smoking compared with partner smoking could reflect this stronger residual confounding in the case of maternal smoking. We have not been able to identify any other studies that have compared the associations of maternal and paternal smoking with offspring body composition, but it is important that our finding be confirmed.

It is possible that different results would have been obtained if all children whose mothers originally enrolled in the study were included in the analysis. However, some similarities between those who attended the physical examination and those who did not have been demonstrated. In addition, findings were similar if the minimally adjusted analysis was restricted to those with complete data on all confounders rather than including any with available data, providing some reassurance that attrition is unlikely to have biased results.

The smoking data relied on self-reports and were not validated in our study. However, a meta-analysis of studies that contained comparisons with biochemical measures found self-reported behavior to be accurate, as assessed by sensitivity and specificity (27). Furthermore, associations between maternal smoking in pregnancy and breastfeeding (28), size at birth and growth in infancy (29), wheeze in infancy (30), and preschool diet (25) have already been demonstrated in this cohort, thus supporting the validity of the smoking data.

Height and weight were measured and DXA scans were performed by trained fieldworkers, which should have minimized measurement error. For 122 children, we have DXA measurements that were repeated on the same day, and the repeatability coefficients (twice the SD of the difference between measurement occasions) (31) were 0.5, 0.6, and 0.7 kg for total fat, truncal fat, and total lean mass, respectively.

There are many reasons why women should not smoke during pregnancy, and our data are consistent with the current literature in providing further evidence that it may lead to increased fat mass in the offspring. However, the availability of partners’ smoking data has allowed our study to start to investigate in more detail the issue of confounding by social factors. In addition to the further work required to investigate possible sex differences in smoking-outcome associations and to compare the maternal and paternal smoking associations mentioned above, studies based in populations with different confounding structures would be valuable. Also, other studies are needed to confirm the association between maternal smoking and offspring lean mass and to investigate whether there are associations with offspring fat distribution (using DXA measures that distinguish between visceral and subcutaneous fat).

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We are extremely grateful to all of the women and children who took part in this study, the midwives for their help in recruiting the subjects, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses. ALSPAC is part of the World Health Organization-initiated European Longitudinal Study of Parents and Children. ALSPAC is supported by the Medical Research Council, the Wellcome Trust, the United Kingdom Department of Health, the Department of the Environment, the Department for Education and Employment, the NIH, and a variety of medical research charities and commercial companies. None of the authors had any financial or personal interest in any company or organization sponsoring this research.

Footnotes
  • 1

    Nonstandard abbreviations: ALSPAC, Avon Longitudinal Study of Parents and Children; SD, standard deviation; IQR, interquartile range.

  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References