Background

Approximately two-thirds of women of reproductive age are overweight or obese [1], and across the range of body mass index (BMI), gestational weight gain (GWG) in excess of guidelines is more common than GWG within or below guidelines, contributing to maternal obesity risk, pregnancy complications, and unsuccessful breastfeeding [1]. Adverse infant outcomes associated with maternal obesity and excessive GWG include birth defects, macrosomia, shoulder dystocia, perinatal mortality, hyperinsulinemia, developmental delays, childhood obesity and cardiovascular disease [2–15]. Maternal diet is increasingly recognized as an important factor in the developmental origins of health and disease. Antecedents of obesity may develop in utero [16, 17], and data suggest that altering maternal prenatal diet impacts offspring body composition [18–23] as well as a range of adverse child outcomes including birth defects [24], cancer [25–28], type 1 diabetes [29], and asthma symptoms [30]. Research is needed to identify dietary determinants of excessive GWG and postpartum weight retention, and inform best practices for weight management during pregnancy and the postpartum period [31].

The problem of obesity and weight gain in pregnancy is linked to the larger epidemic of obesity in the US which, along with poor diet quality, contributes to numerous adverse health outcomes, including reduced fecundity and fertility and chronic diseases such as cardiovascular disease, sleep disorders, and many cancers [32]. The poor diet quality of the US population, characterized by excessive intake of total energy, added sugar, fat and sodium, and inadequate intake of fruit, vegetables and whole grains, is well-documented [33–37]. Mirroring findings regarding weight management in pregnancy, weight management interventions in the general population have been only marginally successful, with less than optimal initial weight loss and poor long-term maintenance [38–41].

The most proximal cause for excess body weight is eating beyond energetic needs. An emerging hypothesis for widespread excess energy intake leading to weight gain and increased adiposity is excess “hedonically motivated food intake” [42], which posits that the neural reward response to highly palatable food cues acts to override the homeostatic processes that historically balanced energy intake with energy expenditure, e.g., [42, 43]. Human neuroimaging research has revealed differential mesolimbic/mesocortical reward circuitry response to both consuming and viewing images of highly palatable foods versus control stimuli (e.g., non-food objects and tasteless solution) as a function of weight status [44]. Differences in reward-related brain activity in the nucleus accumbens in response to food images predicted weight change across 6 months among college students [45]. These neuroimaging results dovetail with research using behavioral assessments of food reinforcement value – the degree to which individuals are willing to work for food rewards – which also implicates the reinforcing value of food as an important determinant of dietary intake and weight status [46].

Evidence suggests that the neural reward response to food stimuli is dependent on a number of food attributes, indicating that highly palatable foods may be more likely than others to contribute to hedonically-driven overeating. Neural reward response is positively associated with perceived energy content of food in images [47–49], and as such, foods high in reinforcement value are generally high-fat, high-sugar, high-sodium selections [50–55]. It is likely, then, that foods processed to contain added fat, sugar, and salt including desserts/pastries, candy, sweetened cereal, snack chips, cheese, fried foods, and processed meat – all highly prevalent in the US diet [56, 57] – are high in reinforcement value. In comparison, nutrient-dense foods containing minimal added sugar, fat and sodium such as whole grains, vegetables, fruit and legumes [58, 59] may be less likely to result in hedonic overeating, and are referred to herein as “normo-palatable” foods. Relative to their highly palatable counterparts, images of these normo-palatable foods elicit a lower reward and attentional response [48, 60–62]. Consequently, the pervasive availability, accessibility and low cost of highly palatable foods may displace normo-palatable foods in the diet [63], leading to excessive intake of energy, sugar, fat, and sodium, inadequate intake of nutrient-dense foods, and excess weight gain.

Basic, behavioral, and neuroimaging research has identified a number of important factors associated with inter-individual differences in the neural reward response to food stimuli and its plasticity. Repeated overeating of highly palatable foods is hypothesized to be a function of individuals’ amplified reward response to highly palatable foods and/or associated food cues [64–67]. Studies have demonstrated the reward response to cues for energy dense food is more pronounced in obese versus normal weight subjects [62], and predicts subsequent weight gain over 6–12 months [61]. Experiments that randomly assigned young adults to consume high-fat/high-sugar foods daily over 14–22 days showed increased food reinforcement for their assigned food relative to controls [68, 69], echoing findings with rodents [70]. Such findings provide compelling behavioral evidence that “wanting” for highly rewarding foods increases with repeated intake. Also in line with the neuroplasticity hypotheses, sensitivity to specific highly palatable foods has been shown to decrease over time in subjects frequently consuming these items [71]. Subjects assigned to consume high-fat/high-sugar foods daily over 14–22 days [72–74] or even 3-month periods [75] reported reduced “liking” of the foods relative to baseline and control foods that were not consumed daily. Taken together, laboratory and experimental evidence from non-pregnant samples implicate the importance of both food-specific attributes and between-individual differences in the role of the neural reward response to food stimuli on dietary intake and weight change. However, no study to date has evaluated these constructs in the context of maternal weight change during pregnancy and postpartum.

In addition to the contributions of reward responsivity and food reinforcement value to eating behavior, evidence also indicates that decreased self-regulation may be a risk factor for overeating and excess weight gain. Both self-report and laboratory measures of impulsivity correlate positively with caloric intake [76, 77] and body mass index [78–80]. Two studies demonstrated that the influence of food reward sensitivity on intake is modified by behavioral control. Findings from both studies indicate that food reward sensitivity is positively associated with food intake during an “eating in the absence of hunger” research paradigm (EAH) particularly in the presence of high versus low impulsivity [81, 82]. Dietary restraint, defined as actively attempting to control intake in order to produce weight loss or prevent weight gain, has also been positively associated with food reward sensitivity [64], and has been found to moderate the effect of food reward on adult energy intake and body weight [83]. Restraint was inversely related to snack food intake in one cross-sectional observational study of college students with normal weight status, but was positively related to snack food intake in overweight subjects [84]. This suggests a distinction between “successful” and “unsuccessful” restrainers, potentially due to differences in self-regulation skills as well as food reward sensitivity [85]. Women of normal, overweight, or obese weight status who demonstrated restrained eating behaviors prior to pregnancy were more likely to experience excessive weight gain [86], possibly suggesting that efforts to control eating were relaxed during pregnancy. Recent evidence indicates that low impulsivity is associated with reduced functional connectivity between the reward and affect regions of the brain, a process involved in cost/benefit evaluation of primary rewards [87]. Further, neural response associated with greater generalized and food-specific impulsivity is related to elevated BMI and predicts future weight gain [48, 88–90]. Despite this growing body of evidence in non-pregnant populations, no study to date has examined the role of food reward and its interaction with behavioral control in the context of GWG and postpartum weight change.

Previous interventions to improve dietary intake and promote healthy body weight based on current health behavior theories have yielded minimal success, suggesting the need for a more comprehensive understanding of the determinants of eating behavior. The reward-driven motivation to eat, culminating in widespread susceptibility to the influence of high food reinforcement value (“hyper-palatability”) on overeating, may be a central factor that has been absent from interventions to influence eating behaviors and weight change. Progress in understanding determinants of GWG and postpartum weight change may be advanced by extending recent findings from lab-based research on the neural reward response to highly palatable food cues to large-scale population-based research. Unique aspects of pregnancy further support the scientific utility of investigating the role of these constructs in diet and weight change. The period of pregnancy and postpartum is a bounded time of expected weight change, offering a window during which influences on weight change may be investigated. An increase in dietary intake is socially normative (e.g., beliefs regarding “eating for two”), and consequently, efforts to restrain intake may be relaxed. Additionally, because many women do not return to their pre-pregnancy weight, excess GWG and postpartum weight retention represent important risk factors for long-term excess weight. The proposed research investigates the influence of food reward sensitivity, home food environment, and behavioral control on maternal dietary intake and weight outcomes during pregnancy and postpartum. The study will further explore these constructs in the context of other aspects of eating behavior such as dietary restraint and motivation for healthful eating, as well as weight-related biomedical, psychosocial and behavioral factors including genetics, physical activity, stress, sleep and depression.

Conceptual model

The hypothesized core conceptual model (Fig. 1) underlying this research considers the primary constructs to be examined in this study with regard to their hypothesized influence on eating behavior and weight change.

Fig. 1
figure 1

Pregnancy Eating Attributes Study (PEAS) conceptual model

Full size image

Food reward sensitivity is a within-person characteristic posited to influence dietary intake and body weight outcomes. All else held constant, individuals with higher food reward sensitivity are hypothesized to consume more food, leading to excess energy intake and increased risk of excessive weight gain and overweight/obesity.

Behavioral control (self-regulation) is the other key personal factor influencing dietary intake and weight outcomes [81–83]. Individuals with high food reward sensitivity but sufficiently high behavioral control of intake may not consume excessively, whereas excess intake would be more likely for individuals with high food reward sensitivity and low self-regulation.

The primary environmental factor included in this model is the reinforcement value of foods in the home. The degree to which food environments are predominated by foods sufficiently high in reinforcement value so as to induce hedonic overeating is hypothesized to influence the relationship between food reward sensitivity, behavioral control, dietary intake and weight outcomes [83]. Individuals with high food reward sensitivity and low behavioral control who are exposed to mainly highly palatable food in their environment are hypothesized to be at increased risk for excess energy intake and elevated weight status. Conversely, even individuals with high food reward sensitivity and low behavioral control may be less like to experience excessive energy intake and weight gain if the home food environment contains few or no highly palatable foods.

More distal influences assessed in this study but not included in the conceptual model include personal factors such as stress, sleep and depression, which are hypothesized to influence food reward sensitivity, behavioral control and dietary intake directly. Consequently, these factors also impact dietary intake and weight outcomes through indirect pathways. Physical activity is hypothesized to influence energy expenditure and weight outcomes directly. Genes are hypothesized to have direct effects on all interpersonal model components.

Study aims

The overarching aim of this research is to examine the role of food reward sensitivity, food reinforcement value, and behavioral control in maternal weight change and dietary intake during pregnancy and postpartum. Primary research questions include the association of food reward sensitivity with maternal dietary intake and weight outcomes; the moderating role of behavioral control and the availability of high-reinforcement value foods in the home environment; and differences in food reinforcement value of fruits and vegetables in their natural form versus highly processed sweet and savory snacks. Secondary research questions include examination of the interplay of food reward sensitivity, self-control, and the home food environment with other eating-related behaviors; the role of maternal food reward sensitivity and dietary intake on infant feeding behavior and body composition, and the roles of maternal sleep, stress and depression as potential moderators of the effect of food reward sensitivity on dietary intake, and change in body weight and body composition.

Design and methods

Study design

PEAS is a prospective observational study of 450 women without evidence of psychiatric or eating disorders, recruited in early pregnancy (≤12 weeks gestation), including those of normal weight (BMI 18.5–24.9 kg/m2), overweight (BMI 25–29.9 kg/m2) and obese (BMI ≥30 kg/m2). Women are followed through pregnancy and until 1-year postpartum, along with their infants from birth to 1 year, with collection of anthropometrics, blood, stool, and urine specimens, previous and current medical and demographic information, and dietary intake and eating and physical activity behaviors. We anticipated an 11 % attrition rate from enrollment to delivery and another 12 % attrition rate from delivery to 1 year postpartum for a total sample size of approximately 350 women-child dyads.

Participants

Participants are recruited from women obtaining prenatal care at the obstetrics clinics at the University of North Carolina (UNC) at Chapel Hill Healthcare System with two locations, one in the hospital and the other off campus – Timberlyne. Inclusion criteria include the following: confirmed pregnant ≤12 weeks gestation at enrollment; uncomplicated singleton pregnancy anticipated; age ≥18 and <45 at screening; willingness to undergo study procedures and provide informed consent for her participation and assent for the baby’s participation; BMI ≥18.5 kg/m2; able to complete self-report assessments in English; access to Internet with email; plan to deliver at the UNC Women’s Hospital; and plan to remain in the geographical vicinity of the clinical site for 1 year following delivery. Exclusion criteria include the following: pre-existing diabetes (type 1 or type 2); multiple pregnancy; participant-reported eating disorder; any fetal anomaly requiring surgery with hospital admission following delivery (e.g., neural tube defects, gastroschisis, cardiac defects, Trisomy 21); any medical condition contraindicating participation in the study such as chronic illnesses or use of medication that could affect diet or weight (e.g., cancer, HIV, active renal disease, myocardial infarction in the last 6 months, chronic steroid use, thyroid disease requiring medication, or autoimmune disease such as rheumatoid arthritis, lupus, antiphospholipid antibody syndrome, scleroderma); psychosocial condition contraindicating participation in the study (e.g., bipolar disorder, schizophrenia, major affective disorder, and substance abuse). Recruitment of participants was initiated November 2014 and expected to be completed by December 2016.

Procedures

Potential participants are identified through the electronic clinical appointments and medical records database. These women are approached regarding the study by research staff and provided with information regarding study participation, including referral to information on the study website. All participants provide signed informed consent to participate in the study. After delivery, participants additionally provide signed informed consent for their child’s participation. Follow-up comprises designated data collection at scheduled clinic visits for the mother during pregnancy and postpartum, and for the baby after delivery, as well as self-report assessments that the participant completes online by secure connection at the study website. Certain specified medical history, medication use, laboratory data and pregnancy complications are extracted from the electronic medical record system post-delivery. Study procedures were approved by the UNC Institutional Review Board.

Measures

Study assessments are conducted prenatally at each trimester of pregnancy, and postpartum at approximately 2 months, 6 months, 9 months and 12 months. A summary of the data collection schedule is provided in Table 1.

Table 1 Pregnancy Eating Attributes Study (PEAS) data collection and schedule
Full size table

Outcome measures include maternal anthropometrics (height and weight; waist, hip and mid upper arm circumferences; and triceps, iliac crest, and thigh skin folds), maternal dietary intake assessed using 24-hour dietary recalls [91, 92], infant eating behavior measures [93–96], and infant anthropometry (birthweight, weight, length, head/abdominal/mid-arm circumferences, skinfolds). Research staff are trained and certified on all these measures prior to formal data collection.

Individual’s food reward sensitivity is measured using several validated questionnaires [85, 97–99]. Reinforcement values of a variety of foods are assessed using adaptations to existing measures [100, 101]. The psychometrics of the measures will be evaluated. Related eating constructs will be assessed including restrained, external and emotional eating [102], motivation for healthy eating [103], eating competence [104], cravings and aversions (developed by the investigators for this study), and food preferences [105].

Self-regulation is measured using two validated questionnaires [106, 107]. The presence of foods with high reinforcement value in the home environment will be assessed using the Home Food Inventory [108], which yields an obesogenic home food availability score. Additional assessments include maternal weight history [109], physical activity [110], perceived stress [111, 112], sleep quality [113], nausea/vomiting, provider advice regarding GWG, and postpartum depression [114]. Data on participant obstetric history, health status, medication use, genetic screening, pregnancy progression (including lab and ultrasound data), and pregnancy complications are extracted from the electronic medical record. Biospecimens collected include maternal blood, urine, rectal swab, cord blood and child rectal swab. Participant demographic information including education level, family income, household composition, marital status, and race/ethnicity are obtained through maternal self-report.

Power analyses

Power analyses are based on examination of the association of food reward sensitivity with GWG, energy intake, and diet quality. For analysis of GWG, the null hypothesis to be tested is that a subject’s weight gain during pregnancy is independent of the subject’s food reward sensitivity, using a regression model with weight gain as the dependent variable and food reward sensitivity score as the independent variable. The null hypothesis is then the regression coefficient beta in the model is zero. For the power calculation, we assume a mean food reward sensitivity score of 2.28 with standard deviation of 0.76, an overall mean GWG of 29.7 pounds with standard deviation of 11.7, both measured at three time points during pregnancy, and utilizing the average of the three measurements. Further assuming a correlation between two food reward sensitivity measurements of 0.7, with a standard deviation of the average score of 0.68, taking retention into account, and an effective sample size of 400 women, with a two-sided significance level of 0.05, the power will be at least 90 % to detect a regression coefficient of 2.76, interpreted as the estimated gain in gestational weight for every unit increase in the food reward sensitivity score.

For analysis of energy intake, the null hypothesis to be tested is that a subject’s energy intake is independent of the subject’s food reward sensitivity, using a regression model with energy intake as the dependent variable and food reward sensitivity as the independent variable. The null hypothesis is then the regression coefficient beta in the model is zero. Assuming energy intake has a mean of 2296 kcal and standard deviation of 453 kcal, and an effective sample size of 400, with a two-sided significance level of 0.05, the power will be at least 90 % to detect a regression coefficient of 106.83 kcal, interpreted as the estimated increase of dietary intake for every unit increase in the food reward sensitivity score.

The Healthy Eating Index-2010 (HEI-2010) is the primary indicator of diet quality. The HEI2010 score measures conformance to the 2010 Dietary Guidelines for Americans, and is comprised of 12 component scores corresponding to dietary guidelines for intake of total fruit, whole fruit, total vegetables, greens and beans, whole grains, dairy, total protein foods, seafood and plant proteins, fatty acids, refined grains, sodium, and empty calories [115]. The null hypothesis to be tested is that a subject’s HEI-2010 score is independent of the subject’s food reward sensitivity, using a regression model with HEI-20101 as the dependent variable and food reward sensitivity as the independent variable. The null hypothesis tests whether the regression coefficient beta in the model is zero. Assuming a mean HEI-2010 score of 52.7 with a standard deviation of 43.2, and a two-sided significance level of 0.05, the power will be at least 90 % to detect a regression coefficient of 10.21, interpreted as the estimated increase of HEI-2010 score with every unit increase in food reward sensitivity score.

Substudies

Three substudies are embedded within the PEAS observational cohort study in order to examine food reward sensitivity and related constructs in greater depth than could be conducted in the main study. Substudies include focus groups, an experimental study using an eating in the absence of hunger paradigm, and functional magnetic resonance imaging (fMRI) of neural response to food stimuli.

Focus groups

A series of focus groups is conducted to provide in-depth exploration of participants’ perceptions relevant to food reward value and other influences on eating during pregnancy. Participants are recruited from the main cohort, including participants of normal weight, overweight, and obese weight categories, for a total of approximately 80 women. Inclusion/exclusion criteria are the same as those used for the main study.

Eating in the absence of hunger

This substudy involves a behavioral experiment that investigates hedonic eating using an eating in the absence of hunger (EAH) paradigm. Aims of the substudy include examining the relationship of food reward sensitivity and BMI with EAH and the potential modifying role of self-control. Participants are recruited from the main cohort, to include participants from the normal weight, overweight, and obese weight categories, for a total of approximately 50 women. Women participate in the behavioral experiment at any point during the window of their second trimester. Inclusion/exclusion criteria are the same as those used for the main study. Additionally, eligible women must have no allergies or aversions to foods served in the substudy protocol.

Functional neuroimaging

This substudy will evaluate brain response to multiple types of food stimuli, varied on hedonic value, as well as an examination of possible weight related differences in resting state functional connectivity and brain network organization using functional magnetic resonance imaging (fMRI). Specifically, this substudy will examine the relationship of brain response to food stimuli (e.g., cue-elicited anticipation and during receipt) with current weight status and GWG, test for the ability of brain response to food stimuli to predict postpartum weight retention (prospectively), evaluate between-food differences in brain response to food stimuli, as well as test for associations of brain response to food stimuli with survey and behavioral measures of food reward sensitivity. Approximately 75 participants, including normal weight, overweight, and obese women, are recruited from the PEAS cohort to participate at their six-month postpartum study visit and undergo behavioral and survey assessments of food reward, executive functioning and one fMRI session.

Discussion

The problem of maternal overweight, obesity, and excessive GWG is a critical public health concern. Findings from basic research in animals and humans indicate the reward-driven motivation to eat culminating in widespread susceptibility to the influence of highly palatable foods on overeating may be a central factor that is absent from current predominant theoretical frameworks explaining eating behavior. Research on the neurobiology of eating behavior has raised important unanswered questions that must be addressed in order to enable the translation to population-based research. The degree to which food reward sensitivity is associated with dietary intake and weight change during pregnancy is not well-understood, nor is the interplay of food reward sensitivity with other relevant constructs including self-control, the food environment, and other eating and health-related behaviors. This observational study addresses important knowledge gaps by examining the implications of food reward sensitivity for maternal diet and weight change. PEAS aims to advance understanding of the determinants of eating behavior and weight change, informing future interventions for improving maternal diet and obesity risk, and leading to improved maternal and child health trajectories.

Abbreviations

BMI, body mass index; EAH, eating in the absence of hunger; fMRI, functional magnetic resonance imaging; GWG, gestational weight gain; HEI-2010, healthy eating index-2010; PEAS, pregnancy eating attributes study; UNC, University of North Carolina