Background
Consistent evidence of an influence of maternal dietary intake during pregnancy on infant body size and composition in human populations is lacking, despite robust evidence in animal models.
Objective
We sought to evaluate the influence of maternal macronutrient intake and balance during pregnancy on neonatal body size and composition, including fat mass and fat-free mass.
Study Design
The analysis was conducted among 1040 mother-offspring pairs enrolled in the prospective prebirth observational cohort: the Healthy Start Study. Diet during pregnancy was collected using repeated 24-hour dietary recalls (up to 8). Direct measures of body composition were obtained using air displacement plethysmography. The National Cancer Institute measurement error model was used to estimate usual dietary intake during pregnancy. Multivariable partition (nonisocaloric) and nutrient density (isocaloric) linear regression models were used to test the associations between maternal dietary intake and neonatal body composition.
Results
The median macronutrient composition during pregnancy was 32.2% from fat, 15.0% from protein, and 47.8% from carbohydrates. In the partition multivariate regression model, individual macronutrient intake values were not associated with birthweight or fat-free mass, but were associated with fat mass. Respectively, 418 kJ increases in total fat, saturated fat, unsaturated fat, and total carbohydrates were associated with 4.2-g ( P = .03), 11.1-g ( P = .003), 5.9-g ( P = .04), and 2.9-g ( P = .02) increases in neonatal fat mass, independent of prepregnancy body mass index. In the nutrient density multivariate regression model, macronutrient balance was not associated with fat mass, fat-free mass, or birthweight after adjustment for prepregnancy body mass index.
Conclusion
Neonatal adiposity, but not birthweight, is independently associated with increased maternal intake of total fat, saturated fat, unsaturated fat, and total carbohydrates, but not protein, suggesting that most forms of increased caloric intake contribute to fetal fat accretion.
Introduction
There is significant interest in the role of maternal dietary intake during pregnancy on offspring body size and composition. Epidemiologic studies have demonstrated that human fetal growth can be influenced by variations within the normal range of dietary intake in high-income counties. Animal studies have consistently demonstrated a relationship between prenatal intake of a high-fat diet and offspring adiposity and metabolic dysfunction. In rodent models, overfeeding of a high-fat or high-cholesterol diet during pregnancy and lactation results in an obese phenotype in the offspring that closely resembles human metabolic syndrome, including increased adiposity. Human studies are less conclusive, and prone to substantial methodological limitations. Most studies of pregnant women have relied on food frequency questionnaires (FFQ), prone to measurement error, which can result in a loss of statistical power to detect a true effect. Inconsistencies in the findings from previous studies of the role of maternal dietary intake on offspring body size at birth or composition may be explained, in part, by inadequate adjustment for measurement error associated with FFQ when applying energy-adjustment models. Observational studies of pregnancy diet relying on FFQ suffer from residual confounding stemming from underreporting of intake, and systematic reporting bias related to the tendency for individuals with higher body mass index (BMI) and higher habitual caloric intake to underreport intake. In addition, inconsistent results across studies may arise from the categorization of continuous dietary intake data into quartiles, expressing results as “the top vs bottom intake of the group” rather than “per unit intake” making comparisons across studies difficult. To address some of these limitations, we sought to characterize the relationship between maternal dietary intake during pregnancy using repeated 24-hour recalls collected via an automated multipass method and direct measures of neonatal body composition in the Healthy Start Study prebirth cohort. To expand comparability with other human studies, the relationship with neonatal outcomes was explored using 2 approaches: a partition model to assess the role of energy intake from specific macronutrients, and a nutrient density model to assess the role of macronutrient composition holding energy constant.
Materials and Methods
The Healthy Start Study recruited and enrolled 1410 pregnant girls and women at ≤24 weeks’ gestation from prenatal obstetric clinics at the University of Colorado Hospital in Aurora, CO. A detailed description of the Healthy Start Study methods has been published elsewhere. Briefly, pregnant girls and women ages ≥16 years who enrolled in the study were invited to participate in 2 research visits during their pregnancy, the first between 8-24 weeks’ gestation (median gestational age = 17 weeks) and the second in mid/late pregnancy between 24-32 weeks’ gestation (median gestational age = 27 weeks). A third research visit was conducted in the hospital within 72 hours after delivery (median postnatal age = 1.0 day) when offspring fat mass (FM) and fat-free mass (FFM) were assessed using air displacement plethysmography (PEA POD, COSMED, Rome, Italy) and anthropometric measures were obtained. All research measurements were obtained by trained research assistants and study nurses. The Healthy Start Study protocol was approved by the Colorado Multiple Institutional Review Board and registered as an observational study at clinicaltrials.gov ( NCT02273297 ).
Maternal measurements during pregnancy
At the 2 research visits during pregnancy, maternal height was measured with a stadiometer (Accustat, Ross Laboratories, Bardonia, NY) and weight measured with a calibrated scale (Tanita Corp, Tokyo, Japan). Questionnaires were administered to assess demographic information and physical activity levels. Maternal prepregnancy BMI was calculated using maternal height measured at the first research visit and prepregnancy weight was obtained from medical records (83.7%) or self-reported at the first research visit (16.2%). Physical activity during pregnancy was measured using the Pregnancy Physical Activity Questionnaire and metabolic equivalent values were estimated as described in detail in previous publications. Prenatal smoking was ascertained through interview-administered questionnaires at each research visit.
Maternal diet during pregnancy
Diet during pregnancy was assessed with repeated 24-hour dietary recalls using the National Cancer Institute (NCI) Automated Self-Administered 24-Hour Recall (ASA24) utilizing the multipass method. Participants completed up to 8 recalls (approximately 1/mo) beginning at their first study with 82% having at least 2 diet recalls. Data from the ASA24 were collected and processed by the Nutrition Obesity Research Center, University of North Carolina at Chapel Hill. My Pyramid Food Equivalents and macronutrient and micronutrient components for each dietary recall were derived through the ASA24 system using nutrient values provided by the US Department of Agriculture My Pyramid Food Equivalents database (Versions 1.0 and 2.0) and the Food and Nutrient Database for Dietary Studies (Versions 1.0 and 4.1).
Dietary data
The NCI measurement error model was used to estimate usual dietary intake during pregnancy from the repeated ASA24. The NCI method is a 2-part nonlinear mixed effects model from which individual estimates of usual macronutrient intake can be generated using a combination of single and multiple dietary recalls. The model separates usual intake of nutrients into 2 parts: the probability of consumption on a particular day, and given that the food was consumed, the amount eaten on the consumption day. For dietary components consumed on an almost daily basis by nearly everyone, such as the macronutrients in the present analysis, the probability of consumption is close to or equal to 1, thus the “amount-only” model was sufficient to predict the usual amount consumed. To increase precision of predicted estimates of usual dietary intake of macronutrients and micronutrients, covariates known to impact dietary intake were incorporated. Covariates included in the NCI model were chosen a priori and included smoking at any time during pregnancy (yes/no), prepregnancy BMI (normal weight, overweight, obese), gravidity, and observed mean total daily energy (kJ/d). Nonepisodically consumed nutrients including total fat; saturated fatty acid (SFA); monounsaturated fatty acid; polyunsaturated fatty acid; total carbohydrates; carbohydrates from sugar including fruit, added sugars, and beverages; and protein were estimated using only the first part of the NCI model (amount-only model). SAS macro-code from the NCI World Wide Web site was used to implement all nutrition models.
Neonatal measures
Offspring’s birth length, weight, head circumference, and skin-fold thickness were measured by trained nurses. Neonatal body composition, FM, and FFM were calculated from total mass and density using air displacement plethysmography (PEA POD). Body composition was recorded twice for each neonate, with a third measurement taken if the first 2 percent body fat values were >2 percentage points apart. Values used in this report are the average of the 2 closest measures. Gestational age at birth was abstracted from medical records. Neonatal chronological age at PEA POD was calculated as the difference in days between the date of birth and the PEA POD research visit.
Statistical analysis
Characteristics of participating mothers and offspring at birth are presented as mean ± SD or number and percent in Table 1 . Maternal usual dietary intake of total energy (kJ), macronutrients, and micronutrients during pregnancy are presented as median grams per day or percentage of total daily caloric intake along with the 25th and 75th percentile.
| Maternal characteristics, N = 1040 | Mean ± SD or N (%) |
|---|---|
| Age, y, mean | 27.87 ± 6.11 |
| Race/ethnicity | |
| Non-Hispanic white | 570 (54.81) |
| Hispanic | 256 (24.62) |
| Non-Hispanic black | 153 (14.71) |
| Other | 61 (5.87) |
| Prepregnancy BMI, kg/m 2 | 25.53 ± 6.07 |
| Prepregnancy BMI status | |
| Underweight | 31 (2.98) |
| Normal weight | 551 (52.98) |
| Overweight | 258 (24.81) |
| Obese | 200 (19.23) |
| Maternal level of education less than high school | 140 (13.46) |
| Primiparous | 667 (64.13) |
| Maternal smoking during pregnancy–any | 90 (8.65) |
| Maternal GDM; by design these girls and women were excluded | – |
| Offspring characteristics | |
| Gestational age, wk | 39.60 ± 1.10 |
| Female sex | 511 (49.13) |
| Birthweight, g | 3283.18 ± 432.54 |
| Birthweight z score | –0.41 ± 0.88 |
| Postnatal age at PEA POD, d | 1.64 ± 2.34 |
| Fat mass, g | 296.48 ± 152.94 |
| Fat mass, % | 9.13 ± 3.96 |
| Fat-free mass, g | 2852.98 ± 336.47 |
| Waist circumference, cm | 29.57 ± 2.43 |
| Head circumference, cm | 34.26 ± 2.09 |
| Sum of skinfolds, cm | 15.21 ± 3.68 |
Two modeling approaches were used to examine the associations between maternal prenatal dietary intake and neonatal body composition outcomes: FM, FFM, and birthweight. The first model was a partition model, which estimates the change in neonatal outcomes associated with a 418 kJ increase in intake of the macronutrient of interest, adjusted for all other macronutrient intake (ie, total energy intake is not held constant). The second model was a nutrient density model that estimated the change in neonatal outcomes associated with a 1% isocaloric increase in a specific macronutrient, offset by a concomitant drop in other nutrients (total energy is held constant). For both modeling approaches a base model included adjustment for offspring sex; gestational age at birth; postnatal age at the PEA POD measurement; and maternal age, gravidity, race/ethnicity (non-Hispanic, Hispanic, non-Hispanic black, and other), smoking at any time during the pregnancy (yes/no), and quartiles of metabolic equivalent values of physical activity levels during pregnancy. Effect modification of the relationship between each macronutrient intake and neonatal outcomes by prepregnancy BMI was evaluated and found to be nonsignificant at the alpha = 0.10 level for all macronutrients examined. Thus a second model (model 2) controlled for prepregnancy BMI in addition to each of the variables in the base model. Finally, gestational weight gain was not controlled for in the multivariate model under the hypothesis that it is on the causal pathway between dietary intake and infant fat accretion. Estimated coefficients are presented in tabular form with 95% confidence intervals. P values <.05 were considered statistically significant, and all analyses were done using software (SAS, Version 9.3; SAS Institute, Cary, NC).
Materials and Methods
The Healthy Start Study recruited and enrolled 1410 pregnant girls and women at ≤24 weeks’ gestation from prenatal obstetric clinics at the University of Colorado Hospital in Aurora, CO. A detailed description of the Healthy Start Study methods has been published elsewhere. Briefly, pregnant girls and women ages ≥16 years who enrolled in the study were invited to participate in 2 research visits during their pregnancy, the first between 8-24 weeks’ gestation (median gestational age = 17 weeks) and the second in mid/late pregnancy between 24-32 weeks’ gestation (median gestational age = 27 weeks). A third research visit was conducted in the hospital within 72 hours after delivery (median postnatal age = 1.0 day) when offspring fat mass (FM) and fat-free mass (FFM) were assessed using air displacement plethysmography (PEA POD, COSMED, Rome, Italy) and anthropometric measures were obtained. All research measurements were obtained by trained research assistants and study nurses. The Healthy Start Study protocol was approved by the Colorado Multiple Institutional Review Board and registered as an observational study at clinicaltrials.gov ( NCT02273297 ).
Maternal measurements during pregnancy
At the 2 research visits during pregnancy, maternal height was measured with a stadiometer (Accustat, Ross Laboratories, Bardonia, NY) and weight measured with a calibrated scale (Tanita Corp, Tokyo, Japan). Questionnaires were administered to assess demographic information and physical activity levels. Maternal prepregnancy BMI was calculated using maternal height measured at the first research visit and prepregnancy weight was obtained from medical records (83.7%) or self-reported at the first research visit (16.2%). Physical activity during pregnancy was measured using the Pregnancy Physical Activity Questionnaire and metabolic equivalent values were estimated as described in detail in previous publications. Prenatal smoking was ascertained through interview-administered questionnaires at each research visit.
Maternal diet during pregnancy
Diet during pregnancy was assessed with repeated 24-hour dietary recalls using the National Cancer Institute (NCI) Automated Self-Administered 24-Hour Recall (ASA24) utilizing the multipass method. Participants completed up to 8 recalls (approximately 1/mo) beginning at their first study with 82% having at least 2 diet recalls. Data from the ASA24 were collected and processed by the Nutrition Obesity Research Center, University of North Carolina at Chapel Hill. My Pyramid Food Equivalents and macronutrient and micronutrient components for each dietary recall were derived through the ASA24 system using nutrient values provided by the US Department of Agriculture My Pyramid Food Equivalents database (Versions 1.0 and 2.0) and the Food and Nutrient Database for Dietary Studies (Versions 1.0 and 4.1).
Dietary data
The NCI measurement error model was used to estimate usual dietary intake during pregnancy from the repeated ASA24. The NCI method is a 2-part nonlinear mixed effects model from which individual estimates of usual macronutrient intake can be generated using a combination of single and multiple dietary recalls. The model separates usual intake of nutrients into 2 parts: the probability of consumption on a particular day, and given that the food was consumed, the amount eaten on the consumption day. For dietary components consumed on an almost daily basis by nearly everyone, such as the macronutrients in the present analysis, the probability of consumption is close to or equal to 1, thus the “amount-only” model was sufficient to predict the usual amount consumed. To increase precision of predicted estimates of usual dietary intake of macronutrients and micronutrients, covariates known to impact dietary intake were incorporated. Covariates included in the NCI model were chosen a priori and included smoking at any time during pregnancy (yes/no), prepregnancy BMI (normal weight, overweight, obese), gravidity, and observed mean total daily energy (kJ/d). Nonepisodically consumed nutrients including total fat; saturated fatty acid (SFA); monounsaturated fatty acid; polyunsaturated fatty acid; total carbohydrates; carbohydrates from sugar including fruit, added sugars, and beverages; and protein were estimated using only the first part of the NCI model (amount-only model). SAS macro-code from the NCI World Wide Web site was used to implement all nutrition models.
Neonatal measures
Offspring’s birth length, weight, head circumference, and skin-fold thickness were measured by trained nurses. Neonatal body composition, FM, and FFM were calculated from total mass and density using air displacement plethysmography (PEA POD). Body composition was recorded twice for each neonate, with a third measurement taken if the first 2 percent body fat values were >2 percentage points apart. Values used in this report are the average of the 2 closest measures. Gestational age at birth was abstracted from medical records. Neonatal chronological age at PEA POD was calculated as the difference in days between the date of birth and the PEA POD research visit.
Statistical analysis
Characteristics of participating mothers and offspring at birth are presented as mean ± SD or number and percent in Table 1 . Maternal usual dietary intake of total energy (kJ), macronutrients, and micronutrients during pregnancy are presented as median grams per day or percentage of total daily caloric intake along with the 25th and 75th percentile.
| Maternal characteristics, N = 1040 | Mean ± SD or N (%) |
|---|---|
| Age, y, mean | 27.87 ± 6.11 |
| Race/ethnicity | |
| Non-Hispanic white | 570 (54.81) |
| Hispanic | 256 (24.62) |
| Non-Hispanic black | 153 (14.71) |
| Other | 61 (5.87) |
| Prepregnancy BMI, kg/m 2 | 25.53 ± 6.07 |
| Prepregnancy BMI status | |
| Underweight | 31 (2.98) |
| Normal weight | 551 (52.98) |
| Overweight | 258 (24.81) |
| Obese | 200 (19.23) |
| Maternal level of education less than high school | 140 (13.46) |
| Primiparous | 667 (64.13) |
| Maternal smoking during pregnancy–any | 90 (8.65) |
| Maternal GDM; by design these girls and women were excluded | – |
| Offspring characteristics | |
| Gestational age, wk | 39.60 ± 1.10 |
| Female sex | 511 (49.13) |
| Birthweight, g | 3283.18 ± 432.54 |
| Birthweight z score | –0.41 ± 0.88 |
| Postnatal age at PEA POD, d | 1.64 ± 2.34 |
| Fat mass, g | 296.48 ± 152.94 |
| Fat mass, % | 9.13 ± 3.96 |
| Fat-free mass, g | 2852.98 ± 336.47 |
| Waist circumference, cm | 29.57 ± 2.43 |
| Head circumference, cm | 34.26 ± 2.09 |
| Sum of skinfolds, cm | 15.21 ± 3.68 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
