Associations Between Prenatal Loss of Control Eating and Cardiovascular Health During Pregnancy

Abstract

Objective:

Loss of control (LOC) eating (feeling unable to control food type/amount eaten) during pregnancy is common and linked to risk for poor cardiovascular health (CVH), but it is unclear whether prenatal LOC eating directly relates to CVH during pregnancy. The current study tested associations between prenatal LOC eating and CVH during pregnancy in a sample with prepregnancy body mass index (BMI) ≥ 25.

Method:

At 12–20 weeks’ gestation, participants (N = 124) self-reported: prenatal LOC eating, diet, physical activity, nicotine use, sleep; height/weight were measured. Data were collected during 2015–2017. We dichotomized LOC eating (0 = absent; 1 = present) and scored CVH metrics using Life’s Essential 8 to create a composite CVH score (range = 0–100; higher = better). Linear and binary logistic regression models tested if LOC eating is related to composite CVH score and odds of scoring low (0)/moderate–high (1) on each CVH metric, respectively. All models employed propensity score adjustment, since those with/without LOC eating may differ in ways affecting CVH, and covaried for: age, gestational age, prepregnancy BMI, ethnicity, race, education, and income.

Results:

Compared to those without, participants with LOC eating had significantly poorer composite CVH scores (b = −9.27, t(111) = −2.70, p < .01) and lower odds of scoring moderate–high on nicotine use (OR = 0.20, 95% CI [0.04, 0.85], p = .03) and sleep duration (OR = 0.19, 95% CI [0.04, 0.83], p = .03) CVH metrics.

Conclusions:

Prenatal LOC eating was associated with poorer CVH during pregnancy in this sample with prepregnancy BMI ≥ 25, even after controlling for propensity of experiencing LOC eating and known risk factors for poor CVH. Thus, prenatal LOC may represent a modifiable factor related to prenatal health risk.

In 2010, the American Heart Association created Life’s Simple 7, a composite measure of cardiovascular health (CVH) that aggregates seven of an individual’s health behaviors (i.e., diet, physical activity, nicotine exposure) and physiological factors (i.e., body mass index [BMI], blood lipids, blood glucose, blood pressure) known to influence risk for morbidity and mortality (Lloyd-Jones, Allen, et al., 2022). In 2022, Life’s Simple 7 was renamed Life’s Essential 8 with the inclusion of sleep as an additional health behavior, due to the preponderance of evidence that sleep independently contributes to CVH and health outcomes (Lloyd-Jones, Allen, et al., 2022). Since the inception of this composite CVH measure, thousands of studies have examined the prevalence and correlates of CVH in diverse populations across the lifecourse and have consistently demonstrated robust associations between poor composite CVH scores and increased cardiovascular disease risk and incidence (Lloyd-Jones, Allen, et al., 2022). Despite the widespread use of both Life’s Simple 7 and Life’s Essential 8 in nonpregnant populations, research employing a composite measure of CVH in pregnant populations is lacking. The dearth of research in this area is of concern because the limited existing evidence indicates that pregnant individuals may be particularly at risk for poor CVH and that poor composite CVH scores during pregnancy are associated with adverse health outcomes.

Results from the few studies that have used Life’s Simple 7 in pregnancy have documented that pregnant individuals have significantly poorer CVH than nonpregnant individuals (Perak et al., 2020) and that poorer composite CVH scores during pregnancy are linked to increased risks for preeclampsia and unplanned primary cesarean delivery, greater subclinical atherosclerosis postpartum, and higher relative risk for poor CVH among adolescent offspring (Benschop et al., 2019; Perak, Lancki, Kuang, Labarthe, Allen, Shah, Lowe, Grobman, Lawrence, et al., 2021; Perak, Lancki, Kuang, Labarthe, Allen, Shah, Lowe, Grobman, Scholtens, et al., 2021). These findings suggest that poor CVH is prevalent during pregnancy and associated with short- and long-term negative health consequences (Benschop et al., 2019; Perak et al., 2020; Perak, Lancki, Kuang, Labarthe, Allen, Shah, Lowe, Grobman, Lawrence, et al., 2021). In line with these findings, other studies have found that indices for many of the metrics comprising Life’s Simple 7, including diet (Moran et al., 2013), physical activity (Borodulin et al., 2008), and BMI (Addo, 2011), are generally poor during pregnancy. Given that indices of sleep have also been found to be poor during pregnancy (Meers & Nowakowski, 2022), the use of Life’s Essential 8, which includes sleep as an additional CVH metric, offers the opportunity to capture more of the pregnancy-related variance in CVH. To identify targets for interventions designed to improve CVH during pregnancy, it is important to examine potentially modifiable factors related to health risk during this critical period.

Prior work suggests that loss of control eating (LOC) may represent one potentially modifiable factor associated with health risk during pregnancy. LOC, defined as the subjective experience of feeling unable to control the type or amount of food consumed during an eating episode, is a commonly reported symptom of disordered eating in both clinical and community samples (Goldschmidt et al., 2015; Hudson et al., 2007; F. Solmi et al., 2014; Sonneville et al., 2013) and is the core psychological feature of binge eating disorders (Goldschmidt, 2017), which have been shown to be effectively treated among nonpregnant individuals with interpersonal psychotherapy and cognitive behavioral therapy (Wilson et al., 2010). Notably, while other disordered eating symptoms typically remit during pregnancy (Crow et al., 2008), LOC remains prevalent and may even present for the first time during pregnancy (Donofry et al., 2021; Kolko et al., 2017; Micali et al., 2018). For example, a large, population-based study of pregnant individuals found that 36% of the sample (N = 11,132) reported prenatal LOC (Micali et al., 2018); another study of pregnant individuals with BMI ≥25 revealed that 37% of the sample (N = 257) reported at least one prenatal LOC episode, with rates of LOC during pregnancy more than double those in the three months prior to pregnancy or at six months postpartum (Donofry et al., 2021). Independent of prepregnancy BMI, prenatal LOC is also associated with adverse health outcomes linked to higher cardiovascular disease risk (Elizabeth et al., 2020; Ferketich & Binkley, 2005; McDowell et al., 2019), including more depressive symptoms and stress, higher total energy intake, and higher intake of processed foods and sugary snacks (Kolko et al., 2017; Micali et al., 2018). Moreover, compared to individuals who do not endorse LOC, those who report prenatal LOC have been shown to experience two-fold greater odds of exceeding gestational weight gain guidelines (Levine et al., 2023). While these findings indirectly suggest an association between prenatal LOC and risk for poor CVH during pregnancy, it remains unclear whether prenatal LOC and CVH during pregnancy are directly related.

Taken together, the prevalence and consequences of poor CVH during pregnancy (Benschop et al., 2019; Perak et al., 2020; Perak, Lancki, Kuang, Labarthe, Allen, Shah, Lowe, Grobman, Lawrence, et al., 2021), and the links between prenatal LOC and factors associated with poor CVH (Kolko et al., 2017; Levine et al., 2023; Micali et al., 2018), suggest that prenatal LOC may directly relate to CVH during pregnancy. Documenting associations between prenatal LOC and CVH, in addition to those previously documented between LOC and eating behavior, weight, and psychological distress/impairment (Colles et al., 2008; Kolko et al., 2017), would provide further indication of whether prenatal LOC is associated with cardiovascular disease risk. Associations between prenatal LOC and CVH during pregnancy could also inform treatment and prevention efforts, as screening for prenatal LOC in medical settings, which can be accomplished via a brief self-report survey, may prove useful in early detection of individuals who may benefit from intervention to address LOC-related morbidity or additional monitoring. Given the increased prevalence of both LOC and poor CVH among individuals with BMI ≥25 (Meany et al., 2014; Opio et al., 2020; Powell-Wiley et al., 2021; F. Solmi et al., 2014; Sonneville et al., 2013), and rising rates of prepregnancy BMI ≥25 (Driscoll & Gregory, 2020), it is especially pertinent to investigate the relationship between prenatal LOC and CVH during pregnancy in this high-risk subgroup. Accordingly, within a community sample of pregnant individuals with BMI ≥25 between 12 and 20 weeks of gestation, the current study first aimed to test the association between the presence of prenatal LOC and a composite CVH score based on the American Heart Association’s Life’s Essential 8. To determine whether prenatal LOC impacts individual CVH metrics, the current study also aimed to test associations between the presence of prenatal LOC and likelihood of scoring low or moderate–high on metrics comprising the composite CVH score. We hypothesized that the presence of prenatal LOC would be associated with a significantly poorer composite CVH score during pregnancy and significantly lower likelihood of scoring moderate–high on individual metrics comprising the composite CVH score.

Method

Participants and Study Procedure

The current study is a secondary analysis of baseline data from a longitudinal study of perinatal eating and weight patterns among individuals with prepregnancy BMI ≥25 (Levine et al., 2023). The current study/present analyses were not preregistered. The data, code used to analyze the data, and nonproprietary materials from the current study are available from the corresponding author upon reasonable request. All study procedures were approved by the University of Pittsburgh Institutional Review Board (PRO11070083). Pregnant individuals were recruited from obstetrics clinics within the University of Pittsburgh Medical Center. The current study includes participants (N = 125) who completed the measures needed to calculate a composite CVH score, including a sleep measure that was added midway through the larger study (described in further detail below). Inclusion criteria for the larger study were prepregnancy BMI ≥25 (i.e., self-reported prepregnancy weight [kg] divided by self-reported height [m²]), singleton pregnancy, and age ≥14 at enrollment. Exclusion criteria were type I diabetes, medications or diagnosed conditions affecting weight, participating in weight management programs, and/or acute psychiatric symptoms warranting immediate intervention (e.g., suicidality). Participants age ≥18 provided written informed consent. Participants age <18 provided verbal assent, and a parent/legal guardian provided written informed consent.

A baseline assessment was administered at 12–20 weeks of gestation, during which all participants in the larger study completed a self-report demographic survey, self-report measures of psychological distress and eating behavior, a clinical interview about experiences of LOC in the first 3 months of pregnancy, self-report measures of diet, physical activity, and nicotine use, and measurements of height and weight. Data collection for the baseline assessment occurred between September 2012 and January 2017. A measure of sleep was added to the baseline assessment battery in February 2015. Accordingly, the current study includes baseline data collected from participants between February 2015 and January 2017.

Measures

Baseline Characteristics

Participants self-reported age, ethnic and racial identities, educational background, annual household income, and date of their last menstrual period (to calculate gestational age at the time they completed their baseline assessment; Wegienka & Baird, 2005). Neither sex assigned at birth nor gender identity were assessed in this study. Prepregnancy BMI was calculated by dividing participants’ self-reported prepregnancy weight (kg) by their height (m²) as measured by research staff at the baseline assessment.

Prenatal LOC

LOC was assessed by trained, master’s-level clinicians using the Eating Disorder Examination-Pregnancy Version, a structured clinical interview for evaluating disordered eating attitudes/behaviors adapted to the context of pregnancy (Emery et al., 2017). To probe for prenatal LOC, the following question(s) were asked to participants who reported times when they felt that they had, or might have, eaten too much at one time during the first three months of pregnancy: “Did you have a sense of loss of control at the time?” “Did you feel you could have stopped eating once you had started?” “Did you feel you could have prevented the episode from occurring?” Prenatal LOC episodes during which participants consumed either objectively (i.e., unambiguously) or subjectively (i.e., based on the respondent’s perception) large amounts of food were included because of data linking LOC, regardless of episode size, to distress/impairment, disturbed eating behavior, and weight-related factors (Goldschmidt, 2017). Given data suggesting that the prevalence of any LOC during pregnancy relates to greater gestational weight gain and higher likelihood of exceeding gestational weight gain guidelines (Levine et al., 2023), prenatal LOC was dichotomized (absent = 0; present = 1) in the current study. Eating Disorder Examination-Pregnancy Version interviews were audio-recorded for reliability; trained, independent raters coded a random sample (n = 20) of the interviews to assess interrater reliability (Cohen’s κ), which was excellent (κ = .99).

CVH During Pregnancy

CVH during pregnancy was operationalized using the American Heart Association’s Life’s Essential 8, a standardized tool to measure and monitor CVH in individuals and populations (Lloyd-Jones, Allen, et al., 2022). Life’s Essential 8 is comprised of the following eight metrics that contribute to health promotion and preservation across the life course: “diet” (i.e., diet quality), “physical activity” (i.e., minutes of moderate and/or vigorous-intensity physical activity per week), “nicotine exposure” (i.e., use of combustible tobacco/other nicotine delivery systems [NDS], secondhand smoke exposure), “sleep health” (i.e., sleep duration), “BMI” (i.e., current weight [kg] divided by current height [m²]), “blood lipids” (i.e., nonhigh-density lipoprotein cholesterol [mg/dl]), “blood glucose” (i.e., fasting blood glucose [mg/dl] or hemoglobin A1C [%]), and “blood pressure” (i.e., systolic and diastolic blood pressures). Scoring systems consistent with the newest clinical guidelines are used to calculate scores for each CVH metric, ranging in increments from 0 to 100, with higher score categories indicating better adherence to recommendations. CVH metric scores are averaged to calculate a continuous composite CVH score, ranging from 0 to 100, with a higher score indicating better CVH. In the current study, the following five CVH metrics were assessed: diet quality, physical activity, nicotine use, sleep duration, and current BMI. Using the scoring system for Life’s Essential 8, we calculated scores for each of these available metrics and averaged them to create a composite CVH score.

Diet Quality.

Diet quality scores were calculated using Healthy Eating Index (HEI)-2015 total scores, which were determined using information collected during 24-hr dietary recalls administered by master’s-level clinicians who received certifications after completing on-site training (University of Minnesota, 2023). The 24-hr dietary recall is a structured interview used to gather detailed information about all foods and drinks consumed in the past 24 hr (24-Hour Dietary Recall (24HR) at a Glance, n.d.), and the HEI-2015 is a measure of diet quality that quantifies the extent to which a set of foods align with key recommendations of the 2015–2020 Dietary Guidelines for Americans (Healthy Eating Index, 2022). Participants completed two 24-hr dietary recalls, one on a weekday and one on a weekend day, which were averaged to calculate total average dietary intake. Total average dietary intake was then used to calculate HEI-2015 component scores, which each reflect one of 13 food groups (e.g., whole fruits, greens and beans, added sugars) and range from either 0 to 5 or 0 to 10. A higher component score indicates better adherence to recommendations for that component. Component scores are summed to create an HEI-2015 total score, ranging from 1 to 100, with a higher score indicating better diet quality. A categorical diet quality score (i.e., 0, 25, 50, 80, 100) was determined for each participant based on their HEI-2015 total score (Lloyd-Jones, Allen, et al., 2022; see Table 1).

Table 1.

Scoring and Distributions of CVH Metrics

CVH metric (LE8 score units)	Low score (0)		Moderate-high score (1)
	LE8 score= points	n	LE8 score = points	n
1. Diet quality (percentiles)	1st-24th = 0	1	50th-74th = 50	54
	25th-49th = 25	67	75th-94th = 80	2
			≥95th = 100	0
		n = 68		n = 56
2. Physical activity (min/week of at least moderate intensity)	0 = 0	62	60–89 = 60	9
	1–29 = 20	11	90–119 = 80	5
	30–59 = 40	10	120–149 = 90	9
			≥150 = 100	18
		n = 83		n = 41
3. Nicotine use (combustible tobacco/NDS use)	Current smoker = 0	14	Quit 1-<5 years = 50	9
	Quit <1 year/current NDS use = 25		Quit ≥5 years = 75	11
		16	Never smoker = 100	74
		n = 30		n = 94
4. Sleep duration (average hours of sleep/night)	<4=0	2	6-<7 = 70	25
	4-<5 = 20	6	9-<10 = 90	12
	5-<6 or ≥10 = 40	11	7-<9 = 100	68
		n = 19		n = 105
5. Current BMI (kg/m2)	≥40.0 = 0	22	25.0–29.9 = 70	46
	35.0–39.9 = 15	26	<25.0 = 100	1
	30.0–34.9 = 30	29
		n = 77		n = 47

Note. CVH = cardiovascular health; LE8 = Life’s Essential 8; NDS = nicotine delivery systems; BMI = body mass index.

Physical Activity.

Physical activity scores were calculated using the Paffenbarger Physical Activity Questionnaire. The Paffenbarger Physical Activity Questionnaire is an eight-item, self-report survey that assesses engagement in various physical activities over the past week, including number of flights of stairs climbed, number of city blocks walked, and duration of sports and recreational activities (Paffenbarger et al., 1978). Trained research staff administered the survey as an interview to seek clarification (e.g., about activity type and duration) when needed and enhance accuracy of reporting. Activities reported by participants were assigned codes from the 2011 Compendium of Physical Activities to derive metabolic equivalent (MET) values (Ainsworth et al., 2011). Each activity was categorized as light (i.e., <3.0 METs), moderate (i.e., 3.0–5.9 METs), or vigorous intensity (i.e., ≥6.0 METs; Pate et al., 1995). A categorical physical activity score (i.e., 0, 20, 40, 60, 80, 90, 100) was determined for each participant based on their minutes of moderate- and/or vigorous-intensity physical activity per week (Lloyd-Jones, Allen, et al., 2022; see Table 1).

Nicotine Use.

Nicotine use scores were calculated using responses to nicotine use-related questions on a health survey. Originally, the health survey asked participants to self-report cigarette smoking history, current cigarette smoking status, and cigarette quit date (if applicable); the health survey was revised after data collection had begun to additionally ask participants to self-report use of other NDS (e.g., e-cigarettes, vaporizers, hookah), including NDS use history, current NDS use status, and NDS quit date (if applicable). Thus, all participants in the current study had information available on combustible tobacco use, but a subset of participants (n = 15) who completed the health survey prior to its revision were missing information on NDS use. Given the substantially lower prevalence of NDS use compared to combustible tobacco use during the period in which data collection for the larger study was ongoing (Phillips et al., 2017), we retained the subset of participants missing NDS use data in the current study’s final sample and utilized their combustible tobacco use data to calculate their nicotine use scores. A categorical nicotine use score (i.e., 0, 25, 50, 75, 100) was determined for each participant based on their combustible tobacco and/or NDS use (Lloyd-Jones, Allen, et al., 2022; see Table 1). As neither version of the health survey assessed whether participants were living with an active indoor smoker in the home, this information was not reflected in their nicotine use scores.

Sleep Duration.

Sleep duration scores were calculated using the Pittsburgh Sleep Quality Index. The Pittsburgh Sleep Quality Index is a 19-item, self-report questionnaire that assesses global sleep quality, and seven individual components of sleep quality (e.g., sleep latency, sleep duration, habitual sleep efficiency), over the past month (Buysse et al., 1989). A categorical sleep duration score (i.e., 0, 20, 40, 70, 90, 100) was determined for each participant based on their average hours of sleep per night (Lloyd-Jones, Allen, et al., 2022; see Table 1).

Current BMI.

BMI scores in early pregnancy were calculated by dividing participants’ current weight (kg) by their current height (m²), which were both measured by research staff during the baseline assessment at 12–20 weeks of gestation. A categorical current BMI score (i.e., 0, 15, 30, 70, 100) was determined for each participant based on their BMI as measured at this point during their pregnancy (Lloyd-Jones, Allen, et al., 2022; see Table 1).

Statistical Analysis

All analyses were performed in RStudio, Version 4.3.1 (R Core Team, 2023). Prior to conducting analyses, descriptive statistics of the sample were calculated. To test whether the presence of prenatal LOC was associated with composite CVH score during pregnancy, a linear regression model was specified with prenatal LOC as a dichotomous independent variable (absent = 0; present = 1) and composite CVH score as a continuous dependent variable. To test whether the presence of prenatal LOC was associated with likelihood of scoring low or moderate–high on metrics comprising the composite CVH score, five binary logistic regression models were specified with prenatal LOC as a dichotomous independent variable (absent = 0; present = 1) and CVH metric scores (0 = low; 1 = moderate–high) as dichotomous dependent variables. Binary logistic regression models were chosen over ordinal logistic regression models for easy interpretation and because, as shown in Table 1, distributions across score categories were nonnormal in this sample, and some categories had insufficient numbers of participants. Previous studies have used cutoffs of 0–49 (low), 50–79 (moderate), and 80–100 (high) to reduce the number of CVH metric score categories (Sun et al., 2023; Zhang et al., 2023), but issues with nonnormal distributions and insufficient numbers of participants per score category would remain if these were used in the current study. Thus, we used a cutoff of 50 (i.e., the midpoint) to create “low” and “moderate–high” CVH metric score categories, with CVH metric scores below 50 points categorized as “low” and those of 50 points or above as “moderate–high” (see Table 1). Retaining the “low” category and combining the “moderate” and “high” categories used in previous research produced more even distributions and allowed for conservative estimates of how prenatal LOC affects metrics comprising CVH (i.e., we collapsed the “moderate” CVH score category into a “moderate–high,” rather than “low–moderate,” CVH metric score category).

Because the current study was cross-sectional, and participants with and without prenatal LOC may differ in ways that affect CVH, we employed propensity score adjustment to help reduce overt bias (Austin, 2011; Elze et al., 2017). Propensity scores are defined as predicted probabilities of an exposure given measured confounders (i.e., variables affecting both the exposure and outcome) and can be estimated by logistic regression of the exposure on potential confounders (Austin, 2011; Coffman et al., 2016; Rosenbaum & Rubin, 1983). When used to adjust statistical models, propensity scores balance the distributions of measured confounders so that differences in the dependent variable (e.g., CVH) between “exposed” (e.g., prenatal LOC present) and “unexposed” (e.g., prenatal LOC absent) individuals with similar propensity scores can be used to estimate the causal effect of the exposure (e.g., prenatal LOC). Propensity scores for prenatal LOC, representing risk for prenatal LOC relative to others in the sample, were calculated for each participant using potential confounders measured at baseline (e.g., psychological distress, eating behavior; see Table S1 in the online supplemental materials) and included as a covariate in all models. While it is not necessary to include confounders used to calculate propensity scores as additional covariates in final models since their confounding is already adjusted for through the propensity scores, doing so is considered a doubly robust method and recommended because it can compensate for insufficient covariate balance (Elze et al., 2017; Kang & Schafer, 2007). Therefore, age, gestational age, prepregnancy BMI, ethnic and racial identities, education, and income were included as additional covariates in all models. Age, gestational age, and prepregnancy BMI were meancentered to aid in the interpretation of results. Statistical significance was set at p < .05.

Of note, the American Heart Association uses different CVH scoring systems for nonpregnant “adults” (i.e., age ≥20) versus “children” (i.e., age ≤19), and six participants in the current study were age ≤19 (range: 17.04–19.89 years). Due to empirical evidence indicating minimal differences in diet quality, physical activity, nicotine use, and BMI between individuals age ≥20 and ≤19 during pregnancy (Chang et al., 2017; Giddens et al., 2000; Steinl et al., 2019; Tobacco Use among Children and Teens, 2023), participants age ≤19 were retained and scored as adults. However, a sensitivity analysis was performed excluding participants age ≤19 to test whether their inclusion significantly influenced results.

Results

Sample Characteristics

Of the 257 participants from the larger study, 130 had sleep data. Five of these 130 participants were excluded from the present analyses due to missing diet (n = 3) and nicotine use (n = 2) data necessary to calculate the composite CVH score, and one participant was excluded due to missing racial identity data necessary to calculate a prenatal LOC propensity score. The final sample size for the current study was 124. Characteristics of the sample, including the composite CVH score and CVH metrics, are displayed in Table 2. Table 1 provides additional detail about CVH metric scores.

Table 2.

Sample Characteristics (N = 124)

Characteristic	M (SD)	Range	n (%)
Age (years)	29.05 (5.39)	17.04–40.80
Gestational age (weeks)	17.10 (2.31)	12.14–20.00
Prepregnancy BMI (kg/m2)	32.50 (6.48)	23.45–55.13
Ethnic identity
Hispanic			4 (3.23)
Non-Hispanic			120 (96.77)
Racial identitya
Asian			1 (0.81)
American Indian/Alaskan Native			3 (2.42)
Black/African American			59 (47.58)
White			69 (55.65)
Education
High school graduate/GED or less			32 (25.81)
Some college/technical school			47 (37.90)
Four-year college degree			22 (17.74)
Postgraduate degree			23 (18.55)
Annual household income
<$30,000			73 (58.87)
≥$30,000			51 (41.13)
LOC present			24 (19.35)
LOC episodesb	2.04 (0.95)	1.00–4.00
Composite CVH score	52.50 (15.04)
CVH metric measure	Median	Range	n (%)
Diet quality (HEI-2015 total score)	48.36	22.61–80.32
Physical activity (min/weekc)	2.50	0.00–780.00
Nicotine use (current combustible tobacco and/or NDS use)			20 (16.13)
Sleep duration (hr/night)	7.00	3.00–12.00
Current BMI (kg/m2)	31.60d	23.20–65.60

Note. BMI = body mass index; GED = high school equivalency degree; CVH = cardiovascular health; HEI = Healthy Eating Index; NDS = nicotine delivery systems; LOC = loss of control eating.

^aParticipants could indicate >1 racial identity.

^bPresent in the first 3 months of pregnancy among participants who endorsed prenatal LOC.

^c≥ Moderate intensity.

^dM(SD) = 33.98(6.98).

Prenatal LOC and Composite CVH Score During Pregnancy

As shown in Table 3, the presence of prenatal LOC was associated with significantly poorer composite CVH scores during pregnancy. This relationship remained significant after adjusting for prenatal LOC propensity scores and other relevant covariates. Participants who reported prenatal LOC had composite CVH scores during pregnancy that were, on average, approximately 9 points lower than participants who did not report prenatal LOC. Additionally, results revealed that older age, higher prepregnancy BMI, and a high school-level education or less were associated with having a significantly poorer composite CVH score.

Table 3.

Association Between the Presence of Prenatal LOC and Composite CVH Score

Coefficient	Estimate (SE)	P
(Intercept)	50.58 (3.15)	<.001
Prenatal LOC present	−9.27 (3.43)	.008
Prenatal LOC propensity score	6.44 (6.11)	.295
Age (M-centered)	−0.71 (0.25)	.006
Gestational age (M-centered)	−0.31 (0.50)	.531
Prepregnancy BMI (M-centered)	−0.67 (0.18)	<.001
Hispanic ethnic identity	−0.53 (6.83)	.939
Black/African American racial identity	−0.88 (2.91)	.764
Other or >1 racial identity	−0.99 (4.80)	.836
High school graduate/GED or less	−7.03 (2.96)	.019
Four-year college degree	3.03 (3.86)	.433
Postgraduate degree	7.51 (4.36)	.088
Annual household income >$30,000	6.87 (3.58)	.058

Note. Reference groups for categorical variables = prenatal LOC absent; non-Hispanic ethnic identity; White racial identity; annual household income < $30,000; some college/technical school. n = 52 identified as only Black/African American, n = 61 identified as only White, and n = 11 identified as other or >1 racial identity (e.g., Black and White). LOC = loss of control eating; CVH = cardiovascular health; BMI = body mass index; GED = high school equivalency degree. Bolded p values indicate statistical significance (p < .05).

Prenatal LOC and CVH Metric Scores During Pregnancy

Compared to their counterparts without prenatal LOC, participants who reported prenatal LOC were significantly less likely to fall into a moderate–high score category for nicotine use (OR = 0.20, 95% confidence interval [CI] [0.04, 0.85], p = .03) and sleep duration (OR = 0.19, 95% CI [0.04, 0.83], p = .03) CVH metrics. However, the odds of falling into low or moderate–high score categories for diet quality (OR = 0.79, 95% CI [0.22, 2.74], p = .71), physical activity (OR = 0.76, 95% CI [0.19, 2.78], p = .68), and current BMI (OR = 1.60, 95% CI [0.24, 11.32], p = .63) CVH metrics did not significantly differ between participants with and without prenatal LOC (see Figure S1 in the online supplemental materials).

Sensitivity Analysis Excluding Participants Age ≤19

When the six participants age ≤19 were excluded from analyses, results remained largely the same. The presence of prenatal LOC among participants age ≥20 was still associated with significantly poorer composite CVH scores during pregnancy, even after adjusting for prenatal LOC propensity scores and other relevant covariates, b = −9.52, t(105) = −2.61, p = .01. Older age, b = −0.70, t(105) = −2.56, p = .01, higher prepregnancy BMI, b = −0.69, t(105) = −3.64, p < .001, and a high school-level education or less, b = −7.28, t(105) = −2.31, p = .02, were still associated with having significantly poorer composite CVH scores. Participants age ≥20 who reported prenatal LOC, compared to those who did not, were still significantly less likely to fall into a moderate–high score category for the nicotine use CVH metric (OR = 0.13, 95% CI [0.04, 0.85], p = .02) but not diet quality (OR = 0.70, 95% CI [0.22, 2.74], p = .61), physical activity (OR = 0.87, 95% CI [0.19, 2.78], p = .84), or current BMI (OR = 1.88, 95% CI [0.24, 11.32], p = .52) CVH metrics. Participants age ≥20 who reported prenatal LOC, compared to those who did not, were also still less likely to fall into a moderate–high score category for the sleep duration CVH metric (OR = 0.19, 95% CI [0.04, 0.83], p = .049), although the statistical significance of this effect was attenuated.

Discussion

Within a community sample of pregnant individuals with BMI ≥25, the current study examined the association between prenatal LOC and a composite CVH score based on Life’s Essential 8. Consistent with our first hypothesis, individuals with prenatal LOC had significantly poorer composite CVH scores during pregnancy than those without prenatal LOC. This novel finding of a direct relationship between prenatal LOC and CVH during pregnancy underscores the health significance of prenatal LOC and suggests it is a unique and modifiable predictor of poor CVH during pregnancy—which is especially important since estimated prevelances of prenatal LOC and poor CVH during pregnancy are so high (Donofry et al., 2021; Micali et al., 2018; Perak et al., 2020). Approximately 20% of the current sample endorsed prenatal LOC, which is lower than the 36%–37% found among pregnant individuals in prior studies (Donofry et al., 2021; Micali et al., 2018). This may in part be due to the fact that these prior studies examined rates of LOC across all of pregnancy, while the current study examined rates of LOC in only the first three months of pregnancy. Even so, the rate of LOC in the current sample is substantially higher than the 5%–10% typically observed among nonpregnant adults in the community (Goldschmidt et al., 2015; Hudson et al., 2007; F. Solmi et al., 2014), lending support to the notion that pregnancy represents a high-risk period for LOC. However, previous work has demonstrated that treatment-seeking status and method of assessment (e.g., questionnaire versus interview) are associated with inconsistent prevalence rates of LOC across studies with nonpregnant samples (He et al., 2017), and future work is needed to determine whether similar factors may relate to discrepant prevalence rates of prenatal LOC. Moreover, the average composite CVH score was approximately 53/100 in the current sample, which is lower than the average of 65/100 estimated among nonpregnant individuals in a large, nationally representative study using Life’s Essential 8 (Lloyd-Jones, Ning, et al., 2022). While prior research examining CVH during pregnancy used Life’s Simple 7, similarly poor CVH scores were observed among pregnant individuals (i.e., 5% had “high” composite CVH scores) compared to nonpregnant individuals (i.e., 13% had “high” composite CVH scores; Perak et al., 2020). Future research should test whether the association between LOC and CVH replicates in larger samples, across wider ranges of weight status and gestational age, and should include both pregnant and nonpregnant individuals to see if this relationship holds outside the context of pregnancy.

In addition to examining the overall association between prenatal LOC and CVH during pregnancy, we evaluated associations between prenatal LOC and scores for each of the individual metrics comprising CVH. Results from these analyses partially supported our second hypothesis that prenatal LOC would be associated with lower likelihood of scoring moderate–high on CVH metrics. Compared to individuals without, those with prenatal LOC had significantly lower odds of falling into moderate–high score categories for nicotine use and sleep duration CVH metrics. However, there were no significant relationships between prenatal LOC and likelihood of having low or moderate–high diet quality, physical activity, or current BMI CVH metric scores. Existing literature supports our finding that prenatal LOC relates to lower scores on sleep duration and nicotine use CVH metrics; prior research has found an increased prevalence of combustible tobacco and NDS use among individuals with diagnosed eating disorders, including binge eating disorder and bulimia nervosa (Morean & L’Insalata, 2018; M. Solmi et al., 2016), and some evidence suggests an association between LOC and poorer sleep (Allison et al., 2016; Manasse et al., 2022; Parker et al., 2022). Although previous research has shown associations between LOC and lower diet quality and higher BMI (Goldschmidt, 2017), we did not find associations between prenatal LOC and scores on these CVH metrics during pregnancy. It is possible that our data did not contain sufficient variability to capture these associations, given that diet quality among individuals with BMI ≥25, and physical activity among individuals across weight statuses, are generally lower during pregnancy (Laraia et al., 2007; Perak et al., 2020; Shin et al., 2016); further, our sample was comprised solely of pregnant individuals with BMI ≥25. Notably, other studies have found that several of the indices used to calculate CVH metrics included in Life’s Essential 8, including diet, physical activity, and sleep, significantly worsen throughout pregnancy, with declines that are maintained postpartum (Addo, 2011; Borodulin et al., 2008; Meers & Nowakowski, 2022; Moran et al., 2013), indicating the potential for changes in CVH risk across the perinatal period. Given the cross-sectional nature of the current study, in which participants were, on average, in their second trimester, future longitudinal research, with samples not selected for BMI, is needed to examine whether change trajectories in CVH throughout pregnancy and postpartum differ between those with and without prenatal LOC.

Findings from the current study should be interpreted within the context of its strengths and limitations. A major strength is the use of metrics from Life’s Essential 8, including the newly added sleep metric, to evaluate CVH in pregnancy, since research measuring CVH during pregnancy has been scarce. By allowing both a holistic view of a pregnant individual’s health status and a closer look at which individual health behaviors and factors are influencing it, measurement of CVH during pregnancy represents a potential method of predicting and intervening upon adverse prenatal events and health outcomes (Benschop et al., 2019; Perak, Lancki, Kuang, Labarthe, Allen, Shah, Lowe, Grobman, Lawrence, et al., 2021; Perak, Lancki, Kuang, Labarthe, Allen, Shah, Lowe, Grobman, Scholtens, et al., 2021). Another strength is the measurement of prenatal LOC with a specialized clinical interview administered by trained, master’s-level clinicians, which is generally preferred compared to self-report measures when assessing such complex, nuanced symptoms of disordered eating (Fairburn & Beglin, 1994). The current study also included pregnant individuals with prepregnancy BMI ≥25, who are at increased risk for both LOC and poor CVH (Meany et al., 2014; Opio et al., 2020; Powell-Wiley et al., 2021; F. Solmi et al., 2014; Sonneville et al., 2013). Furthermore, a majority of the current sample self-identified as Black (48%) and lower income (59%), and inclusion of individuals who identify as Black and/or lower income in research that aims to better understand factors contributing to poor perinatal outcomes is important because of the substantial inequities in perinatal health. Indeed, individuals who identify as Black and/or lower income are at significantly greater risk of adverse pregnancy and postpartum health outcomes than are those who identify as White and/or higher income due to structural and systemic injustices (Sheikh et al., 2022; Weck et al., 2008).

Limitations of the current study include the use of five of the eight CVH metrics included in Life’s Essential 8. Blood lipids, blood glucose, and blood pressure were not measured at the baseline assessment and thus not reflected in participants’ CVH scores. Although the health behaviors and BMI that were measured provide insight into participants’ CVH, future research should measure these biological markers to gain a more comprehensive view of how prenatal LOC relates to CVH during pregnancy. Furthermore, we did not adapt our CVH metrics for the context of pregnancy (e.g., dietary restrictions, gestational weight gain), and future research may want to consider whether doing so is necessary, as well as investigate the empirical basis for Life’s Essential 8’s current age-related scoring systems during pregnancy. While Life’s Essential 8 provides different scoring systems for nonpregnant individuals age ≥20 versus age ≤19, research suggests that low gynecological age (e.g., ≤2 years since menarche), in addition to a younger chronological age (i.e., age ≤16), may better explain differences in prenatal health outcomes and represent more meaningful cutoffs (Brosens et al., 2017; Gibbs et al., 2012). Our sensitivity analysis excluding participants age ≤19 revealed a similar pattern of results as to when they were retained in analyses and scored using age ≥20 guidelines, but future work is needed to determine the utility of stratifying by gynecological and/or chronological age. A second limitation to the current study is that nicotine use scores may be misestimated (i.e., participants may have scored more ideally) because a subset of participants did not complete the study’s revised health survey, which assessed for NDS use, and neither version of the health survey assessed whether participants were living with an active indoor smoker in the home. Had this information been available, it is possible that some participants without prenatal LOC would have scored lower on the nicotine use CVH metric, attenuating the association between prenatal LOC and lower odds of scoring moderate–high on the nicotine use CVH metric. Future research should assess all relevant information used to calculate the nicotine exposure metric score for Life’s Essential 8 and reexamine how they differ between individuals with and without prenatal LOC. Additionally, approximately half of the larger study’s sample (n = 132) were excluded from the present analyses because they were missing data necessary to calculate the composite CVH score. Notably, the majority (n = 127) were excluded due to structurally missing sleep data since the measure of sleep was introduced to the baseline assessment battery midway through data collection in February 2015. Given that sleep is now recognized as an important contributor to CVH and health outcomes, and, as a result, was added to the American Heart Association’s composite measure of CVH in 2022 (Lloyd-Jones, Allen, et al., 2022), future research investigating CVH should assess sleep among all participants in the sample. Finally, the cross-sectional nature of the data limited our ability to draw causal conclusions about relationships between prenatal LOC and CVH during pregnancy. It is possible that prenatal LOC occurs as a result of some CVH metrics (e.g., sleep) or that relationships are bidirectional. Notably, associations between the presence of prenatal LOC and composite CVH score and lower odds of having moderate–high nicotine use and sleep duration CVH metric scores were significant even after controlling for propensity of experiencing prenatal LOC and known risk factors for poor CVH, which helped to eliminate confounding and estimate the causal effect of prenatal LOC.

In conclusion, the current study suggests that the presence of prenatal LOC has negative implications for CVH during pregnancy among individuals with BMI ≥25. Findings from the current study further demonstrate the health significance of prenatal LOC and illustrate the potential utility of interventions that target prenatal LOC to mitigate health risk during pregnancy. The adaptation of evidence-based interventions for LOC in nonpregnant individuals (Wilson et al., 2010), which do not consider the unique eating, weight, and psychosocial factors that affect prenatal health (Baker et al., 1999; Faucher & Mirabito, 2020; Keely et al., 2017; Micali et al., 2018), to specifically address LOC during pregnancy could mitigate cardiovascular disease risk, in addition to risk factors for poor CVH previously shown to relate to prenatal LOC (e.g., suboptimal dietary intake, excess gestational weight gain, psychological distress; Kolko et al., 2017; Levine et al., 2023; Micali et al., 2018). However, further research is first needed to refine the measurement of CVH during pregnancy, examine the association between LOC and CVH in larger samples across the weight spectrum and throughout the perinatal period, as well as to include measures of pregnancy outcomes (e.g., preeclampsia, gestational diabetes), to better determine the generalizability, directionality, and consequences of the relationships observed in the current study.

Public Significance Statement.

Poor cardiovascular health during pregnancy is associated with short- and long-term adverse health consequences for birthing individuals and their children. The current study found that loss of control eating, the core psychological feature of binge eating disorders, during pregnancy was directly related to poorer cardiovascular health among a sample of pregnant individuals with body mass index ≥25. Thus, prenatal loss of control eating may represent a modifiable factor related to prenatal health risk.