ASSOCIATIONS OF EXPOSURE TO PHYSICAL ACTIVITY AND SMOKING DURING PREGNANCY AND OFFSPRING BODY COMPOSITION:
THE HEALTHY START STUDY by
CURTIS STUART HARROD
B.A., University of Colorado at Colorado Springs, 2007 M.P.H., University of Colorado, Anschutz Medical Campus, 2010
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment
of the requirements for the degree of Doctor of Philosophy
Epidemiology Program 2014
ii This thesis for the Doctor of Philosophy degree by
Curtis Stuart Harrod has been approved for the
Epidemiology Program by
Tasha E Fingerlin, Chair Dana Dabelea, Advisor
Lisa Chasan-Taber Regina M Reynolds
Deborah H Glueck
iii Harrod, Curtis Stuart (Ph.D., Epidemiology)
Associations of Exposure to Physical Activity and Smoking during Pregnancy and Offspring Body Composition: The Healthy Start Study
Thesis directed by Professor Dana Dabelea
ABSTRACT
Our objective was to examine the associations of pregnancy physical activity and prenatal smoking with offspring mass [i.e. body mass, fat mass (FM), fat-free mass (FFM)]. We analyzed mother-offspring pairs participating in the longitudinal Healthy Start study who delivered before November 1st, 2013. The Pregnancy Physical Activity Questionnaire was used to assess total energy expenditure and meeting American Congress of Obstetricians and Gynecologists (ACOG) guidelines for physical activity during early-, mid- and late-pregnancy. Data on the quantity and duration of prenatal smoking during early-, mid- and late-pregnancy were collected through self-report. We found a significant inverse linear trend between total energy expenditure during late-pregnancy and neonatal FM (Ptrend = 0.04). Total energy expenditure during early- and mid-pregnancy was not significantly associated with neonatal mass, nor was meeting ACOG guidelines during pregnancy. We also observed dose-dependent and time-specific relationships of prenatal smoking significantly reducing neonatal body mass, FM, FFM and the ratio, FM to FFM. Exposure to prenatal smoking before late-pregnancy was not associated with reductions in neonatal mass compared to unexposed offspring. Despite systematic growth restriction observed at birth, at 5 months, exposed and unexposed offspring had comparable FM (P = 0.61) and FFM (P = 0.41), and following further adjustment for birth weight, exposed offspring had significantly greater FFM (P = 0.04).
iv The change in FFM from birth to postnatal follow-up was also significantly greater in exposed relative to unexposed offspring, even after adjustment for FFM at birth (P = 0.04). In a large cohort, we observed that increasing levels of late-pregnancy total energy expenditure were significantly associated with reduced neonatal adiposity, suggesting that behavioral intervention during pregnancy may help reduce the risk of obesity in offspring. Additionally, exposure to prenatal smoking was associated with systematic growth restriction and that late-pregnancy smoking appears to be the primary period for growth restriction. During early-life, offspring exposed to prenatal smoking demonstrated significant compensatory growth, independent of characteristics, including birth weight. This is suggestive of a programmed mechanism in the offspring as a result of exposure to prenatal smoking, and may contribute to an increased risk of obesity later in life.
The form and content of this abstract are approved. I recommend its publication. Approved: Dana Dabelea
v TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION ... 1
Background ... 1
Pregnancy Physical Activity ... 1
Prenatal Smoking ... 4
Significance... 8
Maternal Physical Activity and Neonatal Body Composition ... 8
Prenatal Smoking and Neonatal Body Composition ... 8
Prenatal Smoking and Postnatal Growth ... 9
Specific Aims and Hypotheses ... 9
Specific AIM 1 ... 9
Specific AIM 2 ... 10
Specific AIM 3 ... 11
II. REVIEW OF THE LITERATURE ... 12
Introduction ... 12
Pregnancy Physical Activity and Neonatal Body Composition ... 12
Prenatal Smoking and Neonatal Body Composition ... 12
Prenatal Smoking and Early-life Body Composition... 13
Search Strategy ... 14
Studies: Pregnancy Physical Activity and Neonatal Body Composition... 15
Studies: Prenatal Smoking and Neonatal Body Composition... 18
Studies: Prenatal Smoking and Postnatal Growth ... 21
Offspring Exposed to Prenatal Smoking and Postnatal Diet ... 24
vi
Neonates Exposed to Pregnancy Physical Activity ... 25
Neonates Exposed to Prenatal Smoking ... 27
Exposure to Prenatal Smoking and Body Composition Later in Life ... 28
Conclusions ... 29
III. RESEARCH DESIGN AND METHODS ... 31
Overview of the Study Design ... 31
Data Collection Methods and Variable Development ... 34
Offspring Body Composition ... 34
Development of FM, FFM and F:FFM Variables ... 35
Development of SGA Variable ... 35
Total Energy Expenditure ... 36
Development of Total Energy Expenditure Variable ... 36
Prenatal Smoking Status ... 36
Development of Smoking Exposure Variable ... 37
Infant Diet ... 38
Data Analysis ... 39
Statistical Power... 40
Statistical Power for AIM 1 ... 41
Statistical Power for AIM 2 ... 41
Statistical Power for AIM 3 ... 43
IV. PREGNANCY PHYSICAL ACTIVITY AND NEONATAL BODY COMPOSITION: THE HEALTHY START STUDY ... 45
Introduction ... 45
Materials and Methods ... 46
Healthy Start Study ... 46
vii Outcomes – Neonatal Body Composition, Birth Weight and
Small-for-Gestational Age ... 48
Covariates ... 48
Data Analysis ... 49
Results ... 50
Meeting ACOG Guidelines during Early-, Mid-, and Late-pregnancy .... 56
Total Energy Expenditure and Small-for-Gestational Age ... 56
Discussion ... 56
V. QUANTITY AND TIMING OF PRENATAL SMOKING ON NEONATAL BODY COMPOSITION: THE HEALTHY START STUDY ... 61
Introduction ... 61
Methods... 62
Healthy Start Study ... 62
Exposure - Prenatal Smoking... 63
Outcomes – Neonatal Body Mass and Composition ... 64
Covariates ... 64
Data Analysis ... 65
Results ... 66
Dose-dependent Effects ... 69
Smoking throughout Pregnancy Compared to Non-smokers ... 70
Smoking before Late-pregnancy Compared to Non-smokers ... 71
Smoking throughout Pregnancy Compared to before Late-pregnancy ... 71
Sensitivity Analyses - Secondhand Smoke during Late-pregnancy ... 71
Discussion ... 72
VI. EXPOSURE TO PRENATAL SMOKING AND EARLY-LIFE BODY COMPOSITION: THE HEALTHY START STUDY ... 76
viii
Introduction ... 76
Methods... 77
Healthy Start ... 77
Exposure – Prenatal Smoking ... 78
Outcomes – Offspring Body Composition and Anthropometrics ... 78
Covariates ... 79
Data Analysis ... 80
Results ... 80
Exposure to Prenatal Smoking and Offspring Outcomes at Follow-up .... 83
Changes in Offspring Outcomes between Delivery and Follow-up ... 84
Discussion ... 86
VII. DISCUSSION AND FUTURE DIRECTIONS ... 91
REFERENCES ... 97
APPENDIX A ELECTRONIC SEARCH STRATEGY ... 109
ix LIST OF TABLES
TABLE
I. Summary of Data Collected on Healthy Start Participants by Research Visit .. 32 II. Summary of estimating duration of prenatal smoking ... 38 III. Summary of estimating quantity of cigarettes smoked during pregnancy ... 38 IV. Summary of Collected Data by AIM and Hypothesis ... 40
V. Statistical Power for the Association between Maternal Physical Activity and FM (AIM 1), Respective of Sample Size and Detectable Difference ... 41 VI. Statistical Power for the Association between Maternal Physical Activity
and FFM (AIM 1), Respective of Sample Size and Detectable Difference... 41 VII. Statistical Power for the Association between Prenatal Smoking and FM
(AIM 2), Respective of Sample Size and Detectable Difference ... 42 VIII. Statistical Power for the Association between Prenatal Smoking and FFM
(AIM 2), Respective of Sample Size and Detectable Difference ... 42 IX. Statistical Power for the Association between Prenatal Smoking and F:FFM
(AIM 2), Respective of Sample Size and Detectable Difference ... 42 X. Statistical Power for the Association between Late-pregnancy Smoking and
FM (AIM 2), Respective of Sample Size and Detectable Difference ... 43 XI. Statistical Power for the Association between Late-pregnancy Smoking and
FFM (AIM 2), Respective of Sample Size and Detectable Difference ... 43 XII. Statistical Power for the Association between Prenatal Smoking and
Postnatal Growth (AIM 3), Respective of Sample Size and Detectable
Difference ... 44 XIII. Characteristics of Healthy Start mother-neonate pairs delivered between
March 2010 and November 2013 (N = 826) ... 52 XIV. Adjusted means and standard errors (SE) for birth weight, neonatal fat mass
and fat-free mass, by quartiles of total energy expenditure and meeting ACOG guidelines during early-, mid- and late-pregnancy ... 54 XV. Adjusted odds ratios for the associations between total energy expenditure
and meeting ACOG guidelines during early-, mid- and late-pregnancy and small-for-gestational age ... 55
x XVI. Characteristics of Healthy Start mother-neonate pairs used in complete case
analyses by prenatal smoking status (N = 916) ... 68 XVII. Summary effects of time specific exposure to prenatal smoking on neonatal
body mass and composition ... 72 XVIII. Characteristics of Healthy Start mother-offspring pairs by prenatal smoking
status (N = 670) ... 82 XIX. Adjusted means and standard errors (SE) by offspring prenatal smoking
status for head circumference, sum of skinfolds, weight-for-length, fat mass and fat-free mass at delivery, postnatal follow-up and change between
xi LIST OF FIGURES
FIGURE
1. Relationships between Total Packs Smoked during Pregnancy and Neonatal Body Mass, Fat-free Mass and Fat Mass ... 70
1 CHAPTER I
INTRODUCTION
The goals of this dissertation are to explore associations between physical activity during pregnancy and neonatal body composition [i.e. fat mass (FM) and fat-free mass (FFM)] and birth weight, and exposure to prenatal smoking and neonatal and early-life body composition and anthropometric measures.
Background
Pregnancy Physical Activity
The current American Congress of Obstetricians and Gynecologists (ACOG) guidelines for physical activity during pregnancy recommend 30 minutes of moderate activity on most days of the week.1 Randomized controlled trials (RCT) have shown that physical activity during pregnancy can greatly affect birth weight.2,3,4 However, time-specific effects or critical periods were suggested as pregnant women who engaged in physical activity only during early stages of gestation had increased birth weight infants, whereas women who exercised only late in gestation had lower birth weight infants.2 The latter finding suggests a biologic mechanism based on energy expenditure and fetal growth related to the timing of FM accretion, which occurs following the 28th week of gestation to term.2,5 This potential mechanism was supported by a cohort study that analyzed women who were recreational runners and/or aerobic dancers during late-pregnancy and increased their pre-late-pregnancy activity levels by 50%.3 The results suggested that compared to matched controls, birth weight was reduced by 310 g and two-site skinfolds thickness assessment showed a 1.5 mm reduction in size. Although absolute FM levels were not directly measured, the authors estimated that, on average,
2 FM was decreased by 220 g,3 and FFM was not significantly different between groups. This provides evidence for a primary effect of late-pregnancy total energy expenditure on FM, while FFM appears to be preserved.
Early studies on the association between physical activity during pregnancy and infant weight related outcomes produced heterogeneous results,3,4,6-9 perhaps due to maternal nutrition, specifically the consumption of carbohydrates.2,10 Evidence has shown that diets of high relative to low-glycaemic food sources contributes to increased
postprandial glucose and insulin responses among pregnant women, leading to increased gestational weight gain (GWG) and exacerbated feto-placental growth.10 Glucose and insulin are obesogenic growth hormones and maternal physical activity has been shown to reduce maternal substrates, such as free fatty acids and glucose from being delivered to the fetus.2 Moreover, exercise during pregnancy increases maternal circulation, which results in better blood flow and oxygenation to the fetus.2 Insulin-like growth factors (IGF)-1 and -2 have been shown to be important hormones in fetal growth.11-13 Another potential mechanism that may be contributing to effects on fetal growth, as a result of maternal physical activity, are reduced concentrations of IGF-1 and -2 among offspring of exercisers compared to controls.14
The positive effects of maternal physical activity during pregnancy are believed to be related to reducing the risk of large-for-gestational age (LGA) (i.e. above the 90 percentile for birth weight, given gestational age and sex)15,16 without increasing the risk of small-for-gestational age (SGA) (i.e. under the 10 percentile for birth weight, given gestational age and sex);14,15 thus, enhancing the probability of having a leaner, normal birth weight offspring or average-for-gestational age (AGA) (i.e. 10th to 90th percentile
3 for birth weight, given gestational age and sex) birth.17
Although the risk of LGA may be decreased, a primary concern of clinicians, researchers and interventionists is increasing the risk of SGA among offspring exposed to high amounts of physical activity during pregnancy. Not only is there evidence to suggest that the risk of SGA is not increased, but there is evidence that exercise during pregnancy results in a protective effect from SGA.15 Nevertheless, an older prospective study among vigorously active, lean women who continued to exercise during pregnancy found an increased risk of SGA relative to women who discontinued exercising prior to the 28th week of gestation.18 There is additional evidence that an inverse relationship may be true based on the risk of macrosomia, as maternal physical activity during pregnancy has been shown to reduce the risk of LGA.15,16
Besides body composition, researchers have demonstrated no significant differences in crown-heel length or head circumference among neonates exposed to maternal physical activity relative to controls.3 This is important to highlight, as these are two commonly reported adverse effects of exposures that are harmful to the fetus, such as prenatal smoking.
Overall, these studies indicate that maternal physical activity affects birth weight and neonatal compartments. They also suggest differential effects of maternal physical activity on neonatal FM and FFM. Specifically, that FM is reduced while FFM appears to not be significantly altered. Therefore, physical activity during pregnancy is suggestive of a beneficial reduction in FM and sustained levels of FFM, indicating that activity does not result in systematic fetal growth restriction. This is paramount as growth restriction in
4 shown to be associated with negative short- and long-term health outcomes of the
offspring, and macrosomia and LGA, which may be prevented by energy expenditure during pregnancy, are also associated with adverse outcomes in the offspring.19-21
Prenatal Smoking
Prenatal smoking is one of the most common, yet preventable teratogens. Research has shown that 1 out of 4 women will stop smoking when they become aware that they are pregnant, and approximately another 25% will quit at some point during pregnancy.22 However, women who are heavy smokers, multiparous, or have an unqualified job or a partner that smokes continue to smoke at greater levels during pregnancy.22 Studies have indicated that the prevalence of prenatal smoking varies between 10.4 and 17%.23-26 Recent surveillance from the Pregnancy Risk Assessment Monitoring System (PRAMS) indicate that from 2000 to 2010, the prevalence of prenatal smoking in the United States has only decreased from 13.3% to 12.3%.27 Acute effects of prenatal smoking include medically indicated and spontaneous pre-term birth (e.g. <37 weeks gestation)28,29 and spontaneous abortion.30
Prenatal smoking has been shown to be associated with low birth weight (LBW) and SGA or IUGR.31-35 Primary determinants of birth size are placental function and length of gestation.36 Studies have suggested that prenatal smoking may decrease birth size as a result of reduced intra-abdominal organ size, bone mineral content and density, birth length, and arm and head circumferences.5,37-42 Recently, a study found that prenatal smoking lowers birth weight more than illicit drug use.43
Intrauterine tobacco exposure has been shown to be associated with reduced neonatal abdominal and head circumference. The former is associated with a decrease in
5 intra-abdominal organ size and the latter is suggestive of neuro-developmental
impairment.39-41 A reduction in intra-abdominal organ size may lead to short and long-term developmental and functional consequences, which may predispose or “program” the offspring to adverse outcomes such as cardiovascular or metabolic diseases.44 Studies have suggested that neurodevelopmental impairment may also have short-45 and long-term46 consequences for the offspring. Head circumference was shown to be a predictor of postnatal neurobehavioral outcomes measured by Mental Developmental Index (MDI) and Psychomotor Developmental Index (PDI).45 A study found that adolescents who were exposed to tobacco in utero as opposed to those who were not, but their mother’s reported smoking before and/or after pregnancy, had poorer academic performance in English, Mathematics and Science.46 Because this analysis was based on women who smoked during pregnancy compared to women who smoked, but not during pregnancy, the results are supportive of programming mechanisms as a result of intrauterine tobacco exposure. One could speculate that the negative impact that prenatal smoking has on fetal neurodevelopment may in part, be the reason for poorer postnatal neurobehavioral
outcomes and adolescent academic performance. Additional evidence was provided by a recent study that showed that exposure to tobacco in utero was associated with a
decreased age at menarche.47
Time-specific effects or critical periods have been suggested with respect to fetal growth. Women who stopped smoking early in gestation had neonates with similar birth weight and other anthropometric measures relative to non-smoking mothers.17,39,48-50 Further, evidence supports that the critical period of growth restriction may be the third trimester of pregnancy, as the risk of SGA was not significantly different between
late-6 pregnancy smoking and smoking throughout gestation.51 Additionally, there are data that display a strong dose-dependent effect of prenatal smoking on fetal growth during the third trimester based on 1 to 8 cigarettes/day; however, this effect leveled off after 8 cigarettes/day,52 indicating a potential threshold of effect on fetal growth.
Multiple mechanisms explaining the association between prenatal smoking and IUGR have been proposed. Tobacco contains carbon monoxide and if exposed within the maternal body, carboxyhemoglobin is formed inhibiting oxygen transfer to the placenta.53 Further, fetal exposure to nicotine results in fetal hypoxia, causing a narrowing of blood vessels (i.e. vasoconstriction),54 which results in increased umbilical artery blood flow restriction.39 An enzyme that may mediate the association between prenatal smoking and LBW is endothelial nitric oxide synthase (eNOS). This enzyme is responsible for
regulating blood vessel dilation, and was found to be significantly decreased in maternal smokers compared to non-smokers.31 Smoking affected both the activity and
concentration of the enzyme and as a result, adequate blood flow was not delivered to the fetus, resulting in a phenotypically undernourished neonate. Studies suggested that eNOS explains close to 25% of the association between prenatal smoking and birth weight.31 Additionally, similar to maternal physical activity, but possibly through different pathways, prenatal smoking was shown to decrease circulating concentrations of IGF, specifically IGF-1 and IGFBP-3.39
In the U.S., one out of every five children is classified as obese.55,56 Several epidemiologic studies have shown long-term consequences of intrauterine tobacco exposure on childhood health, including an increased risk for childhood
7 involved. Von Kries and colleagues retrospectively analyzed prenatal smoking status among current overweight [body mass index (BMI) > 90th percentile] and obese (BMI > 97th percentile) youth. The results suggest that the odds of being exposed to tobacco in
utero was 1.5-times greater among overweight children and 2-times greater among obese
children.57 In a population-based cohort study,58 the odds of prenatal smoking were 30% and 40% greater among overweight and obese participants, respectively. A systematic review on the association between offspring exposure to prenatal smoking and childhood overweight showed that prenatal smoking increased the risk of childhood
overweight/obesity by 50%.59 Despite the evidence, there remains uncertainty if the association of prenatal smoking with childhood obesity is confounded by social factors.62 Nevertheless, using the Northern Finland Birth Cohort, Morandi and colleagues63
recently showed that prenatal smoking was a stronger predictor of both childhood and adolescent obesity than GWG and maternal and paternal BMI. Interestingly, the authors also generated a clinical predictive probability calculator for offspring adiposity, in which prenatal smoking was a primary factor.
The above studies demonstrate associated detrimental effects of exposure to prenatal smoking on the growth and development of the offspring. However, it is not known whether and how tobacco exposure in utero may influence direct measures of neonatal body composition, including the ratio of F:FFM. Knowledge of these effects may help us understand how programming of childhood adiposity by intrauterine tobacco exposure operates.
8 Significance
Maternal Physical Activity and Neonatal Body Composition
With our proposed study, we will add knowledge to the existing body of literature on the association between total energy expenditure during pregnancy and neonatal body composition. To our knowledge, this will be the first study assessing this association by use of air displacement plethysmography (ADP) [PEA POD; Cosmed, USA, Concord, CA] to measure neonatal body composition. Using these highly accurate measures will enable us to tease out the effects of energy expenditure during pregnancy on specific neonatal body compartments, such as FM and FFM. Further, evidence based on
pregnancy physical activity and direct measures of neonatal body composition have been implemented in highly controlled, clinical trials, which may have reduced external validity. Macrosomia and LGA are highly important to prevent as this may program the offspring to acute or chronic morbidities. Understanding the complexity of these
associations in an observational setting is of high significance.
Exploring the influence that maternal physical activity has on neonatal body composition will also elucidate potential intrauterine effects that could alleviate short- and long-term outcomes, such as childhood obesity. Certain behavioral factors as physical activity may be easier to modify and potentially lessen the adverse effects of increased FM among offspring. In turn, this may provide knowledge that is beneficial to researchers and obstetricians.
Prenatal Smoking and Neonatal Body Composition
A critical barrier to preventing the obesity epidemic is a lack of understanding of the etiology of the disease. This study might not only be extrapolated to offspring
9 exposed to tobacco in utero, but potentially all IUGR births. By better understanding the developmental process of the fetus, as a result of in utero tobacco exposure, this may facilitate a greater understanding of programming mechanisms that may be mitigated by preventive efforts to alleviate adverse childhood and later-life outcomes.
Prenatal Smoking and Postnatal Growth
Infants that demonstrate rapid growth within the first several months after birth have an increased risk for later chronic diseases.64,65 Several studies have shown that fetal growth-restriction leads to compensatory acceleration in growth and rate of development as an infant, also known as catch-up growth.66 It is of interest to assess whether and how intrauterine tobacco exposure may influence growth in FM and FFM in the first months of life. Understanding how offspring compensate for growth restriction, as a result of intrauterine tobacco exposure, will provide further evidence towards understanding the early stages of development, and potential mechanisms responsible for the long-term programming of childhood adiposity. Postnatal growth is of particular significance given that childhood BMI and other adiposity measures, even among 2-5 year old children, have been associated with measures in adulthood.67 Studies have indicated that catch-up growth in offspring exposed to prenatal smoking occurs within the first year of life.66,68,69 Specific Aims and Hypotheses
Specific AIM 1
To explore the association of pregnancy physical activity, including time-specific relationships with neonatal body composition, birth weight and SGA. We hypothesize that:
10 during late-pregnancy will be inversely associated with neonatal FM. More specifically, as physical activity levels increase, neonatal FM will decrease. These associations will be independent of maternal age, prenatal smoking status, race/ethnicity, GWG,
pre-pregnancy BMI, educational attainment, household income, gravidity, gestational age at delivery, chronological age at PEA POD and offspring sex. Effect modification of the above associations by pre-pregnancy BMI and maternal race/ethnicity will be explored.
Specific AIM 2
To examine the dose-dependent and time-specific relationships of prenatal smoking with neonatal body mass, FM and FFM, using data from the Healthy Start cohort. We hypothesize that:
Increased levels of prenatal smoking (i.e. total packs) will be associated with lower neonatal body mass, FM and FFM, but similar F:FFM due to comparable proportionate reductions in neonatal FM and FFM. These associations will be
independent of maternal age, race/ethnicity, physical activity, GWG, pre-pregnancy BMI, educational attainment, household income, gravidity, gestational age at delivery,
chronological age at PEA POD and offspring sex. Effect modification of the above associations by pre-pregnancy BMI and race/ethnicity will be explored.
Compared to never smokers, prenatal smoking before late-pregnancy will not be significantly associated with neonatal body mass, FM or FFM.
Compared to never smokers, prenatal smoking throughout pregnancy will be associated with reduced neonatal body mass, FM or FFM.
Compared to mothers who smoked before late-pregnancy, mothers who smoked throughout pregnancy will have reduced neonatal body mass, FM or FFM.
11 Specific AIM 3
To investigate the associations of prenatal smoking and postnatal growth from birth to 5 months of age. We hypothesize that:
Independent of gestational age at delivery, chronological age at PEA POD, offspring sex, maternal race/ethnicity, educational status, household income, gravidity, GWG, pre-pregnancy BMI, exclusivity of breast feeding and mean total energy
expenditure during pregnancy, at postnatal follow-up, offspring exposed to prenatal smoking will have demonstrated compensatory growth with higher, or at least
comparable FM and FFM compared to unexposed offspring. Further, changes in FM and FFM between delivery and postnatal follow-up will be significantly greater among exposed relative to unexposed offspring, independent of neonatal FM and FFM, respectively.
12 CHAPTER II
REVIEW OF THE LITERATURE Introduction
In this literature review, we used a comprehensive search strategy to identify eligible studies that were reviewed based on associations between maternal behaviors during pregnancy (i.e. physical activity and prenatal smoking) and directly measured offspring body composition.
Pregnancy Physical Activity and Neonatal Body Composition
Over the past few decades, we have begun to understand the effects of physical activity during pregnancy. Early apprehension to recommend moderate- or high-intensity physical activity during pregnancy was based on the potential for delivering a phenotypically undernourished infant or causing harm to the fetus. Several studies have been conducted to assess pregnancy physical activity and the risk of IUGR or SGA, as well as the probability of LGA. As a result, we are beginning to understand that maternal physical activity positively impacts offspring during intrauterine life and postnatally. Further, in general, activity during pregnancy has been shown to be safe for both the mother and fetus.2
Prenatal Smoking and Neonatal Body Composition
In the 1960s, around 30-40% of women smoked during pregnancy. Overtime, the prevalence has decreased, but form 2000 to 2010, the prevalence of prenatal smoking only decreased from 13.3% to 12.3%.27 Because of the existing, stagnant levels of prenatal smoking, further understanding of growth and development among exposed offspring is imperative to mitigate adverse effects associated with exposure.
13 Early studies exploring prenatal smoking and neonatal body composition were conducted in the 1980’s.5,70 Researchers used surrogate measures of FM and FFM to estimate body composition. In 1981, D’Souza and colleagues70 analyzed 452 mother-neonate pairs and found that mother-neonates exposed to tobacco in utero had significantly lower birth weight and head circumference, but skinfold thicknesses were not different. Subsequent to this study, Harrison and colleagues5 analyzed 285 full-term, Caucasian neonates, 109 who had been exposed to tobacco in utero and 176 who had not. They assessed neonatal body composition through ponderal index and skinfold measurements. Results suggested that birth weight, length and arm circumference were negatively affected, but subcutaneous fat was not. The researchers concluded that the reduction in birthweight was primarily through decreases in FFM.5
Subsequent to these studies, researchers have further explored similar indirect measures of body composition and shown similar results.42,71-73 These studies primarily relied on anthropometric measures such as circumferences and skinfolds. Nevertheless, there appears to be a gap in knowledge related to direct measures of neonatal body composition and how they are affected by prenatal smoking.
Prenatal Smoking and Early-life Body Composition
The relationship between prenatal smoking and early-life body composition, using direct measures, is not fully understood. The “fetal origins hypothesis” of disease as a result of intrauterine tobacco exposure has been speculated upon with regard to the increased risk of later development of obesity. This hypothesis postulates that
susceptibility to later chronic diseases is programmed during intrauterine life. We will explore this hypothesis within the published literature.
14 For this review, we have searched the literature for existing knowledge on how physical activity during pregnancy alters neonatal FM and FFM, and how prenatal smoking influences neonatal and early-life FM and FFM. To do this, we implemented the following comprehensive search strategy to identify relevant studies.
Search Strategy
We generated a comprehensive electronic search strategy that was implemented within Web of Science (WOS) [Science Citation Index Expanded (SCI-Expanded) (1974-present) and Social Sciences Citation Index (SSCI) (1974 to present)] (Appendix A). An initial search was implemented on November 5th, 2012 and yielded 182 citations, an updated search was conducted on March 7th, 2014 and yielded 442 overall citations. Reference list, retrospective and prospective author, and lateral searches (i.e. cited by or references by) of studies that met inclusion criteria for this literature review were
conducted. Additional papers of interest were also identified by committee members and colleagues. If papers were identified as eligible through secondary screening, the search strategy was revised to increase search sensitivity. Citations were screened by titles and abstracts. If they were deemed of interest, full-text was acquired and reviewed based on the following inclusion criteria:
1. The study sample included offspring exposed to pregnancy physical activity and prenatal smoking.
2. Pregnancy physical activity was a principal component of the study and total energy expenditure was estimated indirectly (e.g. questionnaire) or directly (e.g. actigraphs).
15 3. Prenatal smoking was a principal component of the study (i.e. treated as a
primary explanatory variable; focus of the study was on prenatal smoking) and was collected by subjective (e.g. self-report) or objective measures (e.g. cotinine levels).
4. Neonatal (pregnancy physical activity or prenatal smoking) or early-life (prenatal smoking only) FM or FFM was measured directly, that is by ADP, dual-energy x-ray absorptiometry (DXA), isotope dilution, hydrostatic weighing, MRI, multifrequency bioimpedance or total body electrical conductivity.
Studies that attempted to estimate body composition through birth weight, skinfold instruments or other surrogates, but did not include direct measures were not fully reviewed here, nor were studies where prenatal smoking status or pregnancy physical activity was collected as a potential confounder and only adjusted for in analyses. Further, studies that conducted ultrasonographic examinations were not
comprehensively reviewed, but were used for editorial purposes within the review. Studies: Pregnancy Physical Activity and Neonatal Body Composition
Based on our eligibility criteria, we identified four studies that analyzed maternal physical activity and directly measured neonatal body composition. Within these studies, there was heterogeneity with regard to maternal physical activity regimens and timing of gestation.
Clapp and colleagues4 implemented an RCT enrolling 46 women who were not regular exercisers at enrollment. These mothers were randomly assigned at 8-weeks of gestation to an exercise group featuring weight-bearing components that were to be followed 20 minutes a day, 3-5 times a week until delivery (n = 22) or no exercise (n =
16 24). Each mother allocated to the exercise group performed one of the three weight-bearing exercises: treadmill, step aerobics or stair stepper. The researchers used total body electrical conductivity to measure neonatal body composition and found that neonates of exercisers relative to controls had increased absolute levels of FFM (3.23 kg vs. 3.04 kg; p <0.05), but similar levels of FM (0.43 kg vs. 0.40 kg; p >0.05). Other results were indicative of increased growth. Comparing neonates of exercises to neonates of controls, birth weight (3.75 kg vs. 3.49 kg; p <0.05) and crown-heel length (51.8 cm vs. 50.6 cm; p <0.05) were significantly increased. The findings are suggestive of maternal physical activity enhancing fetal growth.4
Following their initial study, Clapp and colleagues6 again implemented an RCT, but this time they enrolled 75 women who regularly exercised before gestation. At 8-weeks of gestation, mothers were also randomly assigned to one of the three weight-bearing exercises (e.g. treadmill, step aerobics or stair-stepper)6. The participants were further allocated to one of three physical activity regimens from baseline until delivery: 1.) 20 minutes a day, 5 days a week until 20 weeks of gestation, then gradually increase to 60 minutes a day, 5 days a week by 24 weeks of gestation and maintain until delivery (Lo-Hi); 2.) 40 minutes a day, 5 days a week from baseline until delivery (Mod-Mod); 3.) 60 minutes a day, 5 days a week until 20 weeks of gestation, then gradually decrease to 20 minutes a day, 5 days a week by 24 weeks of gestation and maintain until delivery (Hi-Lo).6 Of the 75 subjects, 26 were randomized to Lo-Hi, 24 to Mod-Mod and 25 to Hi-Lo. The results suggested that neonates of mothers who were allocated to the Hi-Lo compared to Mod-Mod and Lo-Hi groups, had significantly greater absolute amounts of FFM (3.42 kg vs. 3.16 kg and 3.06 kg; p <0.001, respectively) and FM (0.48 kg vs. 0.27
17 kg and 0.29 kg; p <0.0001, respectively). Further, differences were found with regard to increased birth weight, crown-heel length, Ponderal index, head/abdomen ratio and body fat percentage. The results demonstrated that maternal physical activity during mid- and late-gestation decreased fetoplacental growth; however, early-gestational physical activity, followed by a decrease in intensity, exacerbated fetal growth.6
Around a decade later, Hopkins and colleagues14 implemented a community based RCT enrolling 84 healthy nulliparous women and randomly assigned them to home-based stationary cycling (n = 47) or no exercise (n = 37). The intervention group was instructed to not exceed 5 sessions of 40-minutes of cycling per week, and
encouraged to maintain this regimen until delivery. The researchers found that neonates of exercisers had significantly reduced body weight (3,683 g vs. 3,936 g; p <0.05) and FFM (3,341 g vs. 3,564 g; p <0.05) compared to controls. On average, they also had less FM (292 g vs. 323 g) and FM% (7.86% vs. 8.17%). The authors further found that mean crown-heel length (50.8 cm vs. 51.0 cm) and head circumference (35.0 cm vs. 35.2 cm) were comparable between the two groups.14 These results support the notion that, through energy expenditure during pregnancy, specific compartments of the developing fetus are targeted, as opposed to a systematic reduction in size.
Pomeroy et al.74 were the first researchers to our knowledge to conduct an observational study exploring pregnancy physical activity and neonatal body
composition. In this small cohort study, 30 pregnant women and their offspring were followed during pregnancy and up to 4 months postpartum. The researchers assessed pregnancy physical activity during 28-32 weeks of gestation using a combined heart rate-movement sensor over 10 days. Total accelerometry counts adjusted for wear time
18 were used to estimate total physical activity during late-pregnancy.74 Offspring body composition was measured by PEA POD. Spearman correlation coefficients were used to show correlation between total pregnancy physical activity and offspring body composition.
Correlation between pregnancy physical activity and offspring FM and FM% were not statistically significant (ρ = 0.31, 95% Confidence Interval [CI]: -0.07 to 0.61 and ρ = 0.19 [-0.23 to 0.52], respectively). However, maternal physical activity was significantly positively correlated with neonatal FFM (ρ = 0.52 [0.17 to 0.74]).74 Using objective measures of pregnancy physical activity and offspring body composition, researchers found that physical activity during late-pregnancy was positively correlated with offspring FFM, indicating that maternal physical activity likely enhance
fetoplacental growth.
Studies: Prenatal Smoking and Neonatal Body Composition
Our search revealed that two studies directly analyzed the association between prenatal smoking and neonatal body composition.34,75 Additionally, three other studies analyzed the association between prenatal smoking and fetal growth as measured by ultrasound.38,39,76 Although these studies did not assess neonatal body composition we will still review the knowledge that has been gleaned from them.
Lindsay and colleagues34 implemented a prospective cohort study, in which data on prenatal smoking were collected during pregnancy, and if at any time participants reported regular smoking, they were classified as a smoker. All infant anthropometric and body composition measures were taken within 24-hours of birth. Total body
19 electrical conductivity was used to measure fat-free mass (FFM) and infants birth weight was subtracted from FFM to obtain FM.
Total body electrical conductivity produces an oscillating current creating a magnetic field. Once an object is placed in the field, electrolytes, which are dissolved in water found in FFM, are conductive compared to FM, which is essentially anhydrous.34 Electrical conductivity is an accurate measure of FFM relative to chemical analysis of carcasses.77
After adjusting for maternal age, education level and payment status, Lindsay and colleagues34 suggested that neonatal FM was not significantly reduced among exposed offspring, relative to unexposed offspring (343±164 g vs. 387±216 g; p = 0.32). On average, FM was reduced among exposed compared to unexposed neonates, and although the difference may not be statistically significant, there might be clinical relevance of an 11.4% reduction in neonatal FM.34 The authors further demonstrated, following adjustment for the aforementioned characteristics, that neonatal FFM was significantly reduced among exposed offspring, relative to unexposed offspring (2,799±292 g vs. 2,965±359 g; p = 0.02).34
Recently, using a large, cross-sectional study, Au et al.75 explored the effects of maternal and fetal factors on neonatal FM, using PEA POD. The sample consisted of 599 mother-neonate pairs. All neonates were term births and measured by PEA POD within 48-hours of birth. Prenatal smoking status was abstracted through medical records and categorized dichotomously as smoked during pregnancy or did not smoke during pregnancy. PEA POD was used to estimate neonatal FM and FFM in both proportionate and absolute terms.
20 Out of the 599 mothers analyzed, only 25 (4.2%) were classified as prenatal smokers. Following adjustment for offspring sex and gestational age, maternal ethnicity, GWG, pre-pregnancy BMI, parity and GDM status, the researchers found that birth weight was statistically significantly reduced by 301 g (95% confidence interval: -492.2 to -109.9; p <0.01). No results were presented on FM or FFM in either proportionate or absolute terms, but the authors did state in the text that prenatal smoking did not have a significant effect on FM%.75
Although our inclusion criteria limited review of studies to those that had measures of neonatal body composition instead of fetal biometrics, we will briefly highlight three studies that analyzed prenatal smoking and fetal growth.
Bernstein and colleagues76 conducted ultrasonographic examinations between 27 and 37 weeks of gestation with all subjects receiving at least two examinations separated by 4 weeks. The authors found that the growth rate of head circumference was not affected, but abdominal circumference, fetal weight and muscle area were all
significantly reduced. The growth rate of fetal thigh fat deposition was reduced among mothers who smoked; however, absolute differences in mass measured between 33 and 37 weeks of gestation were not significantly different.76
Subsequently, Hindmarsh and colleagues38 and Pringle et al39 used the
University College of London Fetal Growth study to analyze 1,650 low-risk, singleton Caucasian pregnancies. Ultrasonographic examinations were conducted at 20 and 30 weeks of gestation. At 20 weeks of gestation, head and abdominal circumferences, and femur length were not significantly different between fetuses exposed to smoking relative to unexposed fetuses. At 30 weeks, there were significant reductions in femur
21 length and abdominal circumference, but no differences in head circumference.38,39 Pringle et al.39 found that at birth, weight, length and head circumference were all reduced among exposed offspring.
It should be noted that recently the Women And Their Children’s Health study (WATCH)78 published a protocol with the plan of following 180 mother-neonate pairs from the second trimester of gestation to 4-years of age. During pregnancy, they propose to estimate fetal body composition through ultrasonographic examination. However, during infancy and early-life, they do not plan on using direct body composition measures postnatally. Instead they will be using estimates such as skinfolds and BMI.
An additional study directly measured neonatal FM and FFM by DXA within two weeks of birth; however, prenatal smoking was not a principal component of the study. Their findings did suggest that neonates exposed to prenatal smoking had significantly decreased total FM and FFM, and marginally decreased FM% and increased FFM%.79 The varying percentages of FM and FFM may be explained by a greater proportionate reduction of absolute FM relative to FFM.
Studies: Prenatal Smoking and Postnatal Growth
Several studies have analyzed weight and BMI as indicators of body composition among offspring exposed to tobacco in utero. Rather than relying on inferior measures of body composition, which have been shown to have high specificity, but low
sensitivity for identifying obesity,80 body composition should be directly measured by accurate techniques such as ADP, DXA or hydrostatic weighing. Nevertheless, we did not identify any studies analyzing prenatal smoking and direct measures of early-life body composition. We identified and reviewed three studies that met our eligibility
22 criteria, but analyzed offspring exposed to prenatal smoking and childhood and
adolescent body composition.
Using 5,689 white singleton births enrolled in the AVON Longitudinal Study of Parents and Children (ALSPAC), Leary and colleagues62 assessed offspring (mean age = 9.9 years) body composition by DXA. The authors were the first to examine prenatal smoking and directly measured offspring body composition. Prenatal smoking was analyzed dichotomously with regard to any trimester during pregnancy (i.e. yes/no during any trimester) and by trimester (i.e. yes/no during specific trimesters). In order to compare regression coefficients across outcome measures, mean values were subtracted from individual values and divided by the SD. The mean and SD were calculated based on the entire cohort. Further, BMI, total FM and truncal fat had skewed distributions and as a result were log transformed. These SD scores by outcomes of interest were then analyzed using multiple linear regression. After adjusting for sex, age (at DXA scan), maternal, paternal and infant feeding factors, birth weight and gestational age, offspring exposed to prenatal smoking during any trimester compared to unexposed offspring had higher childhood BMI (β = 0.24; p <0.001), total FM (β = 0.19; p <0.001), truncal fat (β = 0.20; p = 0.02) and total FFM (β = 0.10; p <0.001). Truncal fat, adjusted by total FM, became non-significant on additional adjustment for offspring and parental factors. Similar findings were observed when prenatal smoking was analyzed by trimester.81
Subsequently, Syme and colleagues82 analyzed body composition of 505 (237 exposed and 268 unexposed), 12-18 year old participants of varying stages of puberty. Subjects were categorized as early (stages 1-3) or late (stages 4-5) pubertal stage. The authors measured subcutaneous abdominal fat (SAF) and intra-abdominal fat (IAF) by
23 MRI and total body FM by multifrequency bioimpedance. Among early stage subjects, SAF, IAF and FM were not different between exposed and unexposed offspring. However, among exposed late stage subjects, increases in SAF and IFA were
observed.82 These associations remained significant after controlling for characteristics including birth weight and breastfeeding. Despite previous evidence showing increases in body compartments of exposed children 10 years of age,62 this study found no differences in early stage pubertal subjects. However, Syme and colleagues found that the association between prenatal smoking and offspring body composition is modified by stage of puberty.82 These findings suggest that due to exposure to prenatal smoking, an occult mechanism mediates exacerbated offspring growth over and above normal growth patterns observed during puberty, emphasized by adipocyte proliferation.
The Saguenay Youth Study conducted by Haghighi and colleagues83 analyzed offspring body adiposity among 13-19 year old sexually mature (i.e. Tanner stage 4 or 5) adolescents exposed (n = 180) and unexposed (n = 198) to prenatal smoking. They implemented a cross-sectional study design, treated prenatal smoking as a dichotomous variable (i.e. offspring of mothers who smoked 1 cigarette per day during
mid-pregnancy were classified as exposed; offspring of mothers who did not smoke at least 1-year before or during pregnancy were classified as unexposed) and measured body adiposity through multifrequency bioimpedance. Following statistical adjustment for sex, age and height, the results suggest that adolescents exposed to intrauterine tobacco exposure had increased levels of FM (β = 1.7 kg; p = 0.009). Further adjustment for birth weight, breastfeeding duration, gestational age and family income slightly attenuated the estimate, but the effect size remained statistically significant. Body
24 weight and BMI were also both significantly greater among exposed compared to
unexposed offspring.83
Offspring Exposed to Prenatal Smoking and Postnatal Diet
Haghighi and colleagues83 also analyzed fat intake based on prenatal smoking status. They measured this macronutrient by using a 24-hour food recall questionnaire complemented by six questions pertaining to eating habits and fruit and vegetable intake.83 Using these data, the researchers estimated the percentage of fat intake in the adolescent’s diet. They found that those who were exposed to prenatal smoking had an increased propensity for fat in their diet (β = 3.4% increase; p = <0.001).83 Although this is the first study to report this association based on offspring exposed to prenatal
smoking, increased fat in the diet had been previously observed with regard to offspring exposed to maternal undernutrition.84 Exposure to both prenatal smoking and maternal undernutrition during pregnancy are known risk factors for fetal growth restriction, which is a predictor of later-life obesity.85
This is important to explore and further understand, as this may be a mechanism that is programmed during pregnancy, as a result of exposure to prenatal smoking. The authors provide further evidence that growth and development may have been
permanently altered or programmed by exposure to tobacco in utero as exposed offspring had significantly reduced amygdala volume (β = -95 mm3; p <0.001), as measured by MRI.83 Additional statistical adjustment attenuated the effect size, but it remained statistically significant. Although the amygdala is known for regulation of aggression and fear, there is evidence that it plays a role in stimulus-reward
25 show similar brain reward system processes as the pleasure inducing effects of illicit drug use.89
Overall, underlying mechanisms of the association between prenatal smoking and increased offspring adiposity are unknown; however, these findings support an increased proclivity of fat in the diet of exposed offspring. Further, slight structural variations of the amygdala may mediate this process.83
Discussion
Neonates Exposed to Pregnancy Physical Activity
An initial RCT by Clapp and colleagues4 analyzing maternal physical activity and directly measured neonatal body composition found that neonates of exercising mothers had increased FFM relative to controls. One observational study also found that total pregnancy physical activity was positively correlated with FFM.74 These findings appear to contradict two RCTs51,56 in different populations, as FFM was significantly decreased in both trials. However, the dose of physical activity (i.e. 20 minutes a day, 3-5 days a week) administered in the initial study was essentially the same as their
subsequent study (i.e. the Hi-Lo group), following the 20th week of gestation.6 The mothers allocated to the Hi-Lo group had neonates with the largest amount of FM and FFM relative to the other exercise groups. Thus, this evidence may be suggestive of a specific amount of energy expenditure, especially late in pregnancy, to decrease neonatal FFM, but more importantly FM.
Also to note, proportionally, the reduction in neonatal FM was larger than FFM in both RCTs that found significant decreases in neonatal body composition.6,14 In part,
26 this may be the reason that evidence is suggestive of no significant increased risk of SGA among women who exercise during pregnancy.
Using the same study population of 84 healthy, nulliparous women, Hopkins and colleagues90 further analyzed the association between maternal physical activity and neonatal body composition by assessing hormonal changes in exercisers compared to controls. They found that exercise training had no impact on IGF axis, but it increased leptin levels and marginally decreased FFAs during late-pregnancy. The authors concluded that increased maternal leptin levels along with decreased availability of FFAs among treatment participants, may have occurred as a result of exercise induced placenta alterations, which may explain the reduced neonate size (FM and FFM) relative to controls.90
The evidence displayed in the literature highlights time-specific effects of maternal physical activity on neonatal body composition. This is not surprising as neonatal body composition is related to the timing of gestation, such as FM and FFM being primarily accreted in mid- to late-gestation.
None of the previously described studies analyzed resistance training as a form of maternal physical activity; instead they all analyzed aerobic exercise4,6,14 or total physical activity.74 It may be of interest to analyze varying or combined methods of physical activity and its effect on offspring body composition. Resistance training promotes maternal muscle growth (i.e. lean mass), which in turn, increases the need for glucose and as a result may decrease maternal substrate availability to the fetus. A recent meta-analysis in a non-pregnant population showed beneficial effects of reducing fasting
27 glucose and insulin levels based on a combination of aerobic and resistance training compared to a diet only intervention.91
Neonates Exposed to Prenatal Smoking
Even though Lindsay and colleagues showed that FFM was significantly reduced among neonates exposed compared to unexposed to prenatal smoking, and that FM was not significantly reduced, the proportionate reduction in mass was greater in FM relative to FFM, 11.4% and 5.6%, respectively. The clinical implications of losing greater than one-tenth of total FM may be of importance and overlooked based on a non-statistically significant finding.
It is important to highlight that FM is primarily accreted after 28 weeks of gestation until term. Thus, the ultrasonographic examinations may have been limited with the timing of exams. Further, if Bernstein et al76 had a larger sample to analyze, the absolute mean difference between FM among neonates exposed and unexposed to prenatal smoking may have been statistically significant.
Outside of the dearth of evidence on the associations between prenatal smoking and direct measures of neonatal FM and FFM, the study by Lindsay et al.34 analyzed only 129 term infants with just 30 exposed to tobacco in utero.34 Additionally, Au et al.75 analyzed 599 term neonates, but only 25 (4.2%) were exposed to prenatal smoking. Although Au et al.75 adjusted for several important confounders in their analyses, Lindsay and colleagues34 estimates were not adjusted for by critical factors such as GWG, and pre-pregnancy BMI was not considered in the analyses. The authors only adjusted estimates by covariates that were significantly different between smokers and non-smokers. Thus, because differences in pre-pregnant weight, GWG, gestational
28 diabetes mellitus (GDM) and neonatal sex were not different, they were not explored in the analyses.
Other studies have shown that pre-pregnancy BMI is an effect modifier of the association between prenatal smoking and birth weight33, and that GDM exacerbates fetal growth.92 Lastly, evidence has also suggested that prenatal smoking has a more complex association with neonatal body composition,52 which may be obscured with a dichotomous assessment of smoking status.
Exposure to Prenatal Smoking and Body Composition Later in Life
Exposure to late-pregnancy smoking has been shown to have a strong dose-effect on birth weight and similar decreases in birth weight compared to exposure to smoking throughout pregnancy.51,52 Yet, using the ALSPAC cohort, Leary and colleagues62 found equivalent increases in offspring FM and FFM at approximately 10 years of age, based on prenatal smoking during pregnancy and by trimester. This suggests that exposure to prenatal smoking during pregnancy, regardless of the timing, may result in programming effects that increase the risk of childhood obesity later in life. However, further study is needed.
In a subsequent study, Leary and colleagues81 also analyzed prenatal smoking and offspring stature. Following adjustment for offspring and parental characteristics, prenatal smoking was associated with reduced leg length and leg-to-trunk ratio. Reduced offspring height was present in models adjusted only for offspring characteristics, but was attenuated by adjustment of additional parental covariates.81
To highlight the importance of using direct measures of body composition, a recent study estimated subcutaneous fat using skinfold measurements among offspring
29 exposed to prenatal smoking compared to unexposed offspring and found no significant associations. The authors stated that their findings may contradict those found in
previous studies because of their use of inferior measurement techniques.93 To date, Leary and colleagues62 used the largest cohort study to evaluate the association between prenatal smoking and directly measured childhood body composition and found
significant increases in adiposity. These findings have been reproduced by other studies reviewed here82,83 with one study82 implicating puberty as an effect modifier.
Conclusions
To our knowledge only one, small cohort study74 has analyzed pregnancy physical activity and neonatal body composition, which was measured by PEA POD. Further, only one cross-sectional study75 has been conducted based on the association between prenatal smoking and neonatal body composition, as measured by PEA POD. No studies were identified that analyzed exposure to prenatal smoking and early-life body composition.
Hopkins and colleagues demonstrated that maternal physical activity during pregnancy appears to lead to leaner, normal birth weight neonates. This was exhibited by similar average crown-heel length and head circumference, along with no increased risk of SGA.14 These associations should be further explored using larger samples and with an observational study design, as opposed to the RCTs to determine if such changes are observed in a non-clinical and more generalizable setting.
Although the data are limited, based on the evidence, it appears that growth restriction in neonates exposed to in utero tobacco is primarily driven by decreases in neonatal FFM. More studies using larger samples, particularly of those exposed to
30 tobacco in utero are needed to assess absolute differences in body composition of
exposed and unexposed offspring. Additionally, future studies should analyze prenatal smoking as a continuous (dose-effect) rather than a dichotomous variable. Nevertheless, quantity and duration are both necessary for such a variable throughout pregnancy.
By use of indirect measures of body composition, the evidence is mixed with regard to the association between exposure to prenatal smoking and increased FM during childhood.66,93-101 Yet, when using direct measures of body composition the evidence suggests that intrauterine tobacco exposure accelerates post-natal growth, particularly FM, relative to unexposed offspring.62,82,83 Rigorously designed studies are needed along with consideration of additional confounding factors, such as offspring diet and physical activity.
31 CHAPTER III
RESEARCH DESIGN AND METHODS Overview of the Study Design
The Healthy Start Study is an ongoing, longitudinal cohort study that enrolls ethnically diverse women early in pregnancy. Pregnant women are primarily recruited at the prenatal obstetrics clinics located at the University of Colorado Hospital (UCH) Outpatient Pavilion within the Anschutz Medical Campus of the University of Colorado - Denver. This provides access to a population of ethnically-diverse pregnant women. Participants enrolled in Healthy Start were invited to participate in three prenatal research visits and a postnatal follow-up.
Table I provides an overview of the data that will be used in this dissertation by timing of pregnancy, all of which are collected through the Healthy Start study.
Maternal physical activity levels are ascertained through a validated102 pregnancy
physical activity questionnaire (PPAQ) during each research visit. The PPAQ consists of 33 questions, two of which are open-ended and 31 that are based on activities that vary in duration (i.e. hours/day or hours/week), which are reported in ranges, intensity (i.e. sedentary, light, moderate and vigorous) and type (i.e. household/care giving,
occupational, and sports/exercise). The goal of using the PPAQ is to obtain an estimate of the participant’s daily total energy expenditure within a general 24-hour period within early-, mid- and late-pregnancy.
Self-reported prenatal smoking status, duration and quantity are collected three times during study specific research visits. Data concerning self-reported exposure to second-hand smoke (SHS) or passive smoke are also collected during each visit.
32 Table I. Summary of Data Collected on Healthy Start Participants by Research Visit
Earlya Midb Latec Postnatald
Maternal Data Prenatal smoking† X X X Second-hand smoke X X X Prenatal weight X X Height X X Pre-pregnancy weight‡ Age X Race/ethnicity X Educational attainment X Gravidity X Physical activity X X X Diabetes status‡ X Offspring Data
Neonatal fat mass (g and %) X X
Neonatal fat-free mass (g and %) X X
Neonatal body mass X X
Crown-heel length X X Triceps skinfold X X Subscapular skinfold X X Mid-thigh skinfold X X Head circumference X X Abdominal circumference X X Mid-thigh circumference X X Sex X Offspring diet X
1- and 5-minute APGAR Scores‡
Gestational age‡ X X X
aEarly-pregnancy: 13-23 weeks of gestation bMid-pregnancy: 24-31 weeks of gestation cLate-pregnancy: Delivery hospitalization stay dPostnatal: 4-6 months postnatal
‡Measures are abstracted from medical records or collected from perinatal database
There are often concerns regarding the internal validity of studies using subject measures (e.g. self-reports), especially with regard to risky behaviors, such as smoking. However, our study design and data collection methodologies have been specifically implemented to assuage potential biases. A limitation of self-reported questionnaires is a vulnerability to bias, more specifically, to recall and reporting biases, which may lead to misclassification and apocryphal results. However, because Healthy Start is a prospective
33 longitudinal study, participants do not have knowledge of outcomes (e.g. neonatal FM and FFM) during ascertainment of exposure data. Therefore, if a bias such as under- or over-reporting of pregnancy physical activity or prenatal smoking occurs, it will likely be non-differential with respect to the outcomes of interest, and this type of misclassification is almost always biased towards the null.
Based on maternal physical activity, we will be using the PPAQ, which was specifically designed to ascertain total energy expenditure among pregnant women.
Based on prenatal smoking, validation studies have been conducted analyzing self-reported smoking status compared to exhaled carbon monoxide and cotinine levels, and the results suggest that mothers accurately report smoking status. Self-reporting smoking status compared to exhaled carbon monoxide levels produced an area under the ROC curve between 0.88 and 0.99 throughout gestation.103 Further, no significant differences were noticed in the accuracy of self-reported smoking status based on the quantity of smoking (e.g. light compared to heavy).103 Cotinine, which is a metabolite of nicotine that can be found in the blood, urine and saliva, is considered the gold standard towards delineating smokers and smokers. Out of 407 pregnant, self-reported non-smokers, only 6% were identified as smokers through plasma cotinine levels.104
Furthermore, a recent prospective cohort study found that the sensitivity and specificity of self-reporting smoking status compared to cotinine levels were 81% and 97%,
respectively. Based on the results from validation studies and how we will prospectively ascertain maternal smoking status, we are confident that our results will not be
substantially biased as a result of misclassification.
34 measures, including FM and FFM by PEA POD, are taken at birth and at postnatal
follow-up. The anthropometric measures consist of birth weight and crown-heel length, which enables calculation of ponderal index (PI), triceps, subscapular and mid-thigh skinfolds and head, abdominal and mid-thigh circumferences. The body composition measures, which are collected by PEA POD, include absolute levels (in grams) and percentage of FM and FFM. Additional information is collected, including demographic factors such as maternal age, race/ethnicity, educational attainment, gravidity; other behavioral factors, such as diet and alcohol consumption; as well as metabolic factors such as diabetes status (i.e. subjective and objective measures), pre-pregnancy BMI and GWG.
Because our population of mother-offspring pairs is sampled from Colorado and there is regional variability in altitude, we explored this potential bias of offspring that were gestated at varying levels of altitude (i.e. greater than 6,000 feet vs. less than or equal to 6,000 feet). If differences existed for any of the explored associations, we proposed to adjust for this in our statistical models.
Data Collection Methods and Variable Development
This section features a detailed description of the methodology pertaining to how primary and secondary outcomes, primary explanatory variables (PEV) and covariates are collected and developed into variables for analysis.
Offspring Body Composition
Offspring body composition is collected at delivery and postnatal visits. The PEA POD infant body composition system measures total body mass, FM and FFM. Each neonate is measured, if possible, 3 times at each visit. FM and FFM are calculated in both
35 percent and grams. How PEA POD, which is a densitometric technique, works can be found elsewhere.105 As compared to hydrostatic weighing (e.g. submerged underwater), the accuracy of PEA POD is equivalent, but is safer, easier to use and is less invasive. The PEA POD has been shown to be reliable and valid towards measuring neonatal body composition.105-107 Studies have shown that the mean percentage error in the volume measurements were < 0.05%.106 Furthermore, other commonly used techniques such as DXA have been shown to overestimate FM, even at 6-months of age.108
Development of FM, FFM and F:FFM Variables
FM and FFM are both measured by PEA POD. Because we repeatedly measured each offspring up to 3 times at delivery and postnatal visits, we will take the mean of the two closest measures, with respect to the visit, to assess both the absolute values (in grams) and the percentages of FFM and FM. By taking the average of the two closest measures, rather than all three, we will eliminate the possibility of including measures that are erroneous, which will likely reduce bias as a result of measurement error.
For AIMS 1 and 2, measures collected shortly after birth were analyzed by absolute (i.e. grams) estimates of neonatal FM and FFM. However, for AIM 3, we will also explore the change in early-life body composition. We took the difference in total body mass, FM and FFM from the delivery and postnatal PEA POD measures. Further, we adjusted for delivery measures in our models analyzing measures at follow-up.
Development of SGA Variable
We also calculated SGA using reference data. Birth weight was ascertained through medical record abstraction and maternal self-report. Using United States national reference data, 109 SGA was categorized as below or at or above the 10th percentile for
36 gestational age, given sex of the offspring.
Total Energy Expenditure
Using data from the three prenatal research visits, we estimated the participant’s daily total energy expenditure within a general 24-hour period on average during pregnancy, as well as total energy expenditure during early-, mid- and late-pregnancy.
Development of Total Energy Expenditure Variable
Because of concerns with over reporting, the lower end of each reported duration range in the PPAQ was taken and multiplied by a metabolic equivalent of task (MET) intensity value in accordance with the compendium of physical activities.110,111 However, because MET values of intensity are often generated from men and non-pregnant women, where possible, we applied pregnancy-specific MET values to calculate total energy expenditure.112 Mean daily energy estimates were then converted to MET-hrs/wk.
In order to reduce the potential for misclassification, we categorized total energy expenditure during early-, mid- and late-pregnancy by quartiles (i.e. 25th percentile or below [referent], indicates the lowest level of MET-hrs/wk and 75th percentile or above, indicates the highest level of MET-hrs/wk). Women were also categorized by meeting ACOG guidelines for physical activity if they did or did not have > 7.5 MET-hrs/wk in sports/exercise activities of moderate-intensity or greater (i.e. 30-minutes per day of activity at ≥ 3 METs multiplied by 5 days per week)1 during early-, mid- and late-pregnancy.
Prenatal Smoking Status
Smoking status is ascertained through an interview administered questionnaire, in which data based on the quantity and duration of early, mid- and late-pregnancy smoking
