The Association Between Midpregnancy Food Intake Timing Patterns and Gestational Weight Gain

Abstract

Background:

This article aims to characterize midpregnancy food timing profiles and examine their association with gestational weight gain (GWG).

Materials and Methods:

This secondary data analysis of a randomized controlled trial of two gestational diabetes screening approaches included 641 individuals with primary exposures and outcomes data. Food timing indicators (i.e., first and last eating episode time, caloric midpoint time, and the total eating window) were assessed using two 24-hour dietary recalls conducted in midpregnancy. Latent profile analysis was used to identify distinct food timing profiles based on these indicators. Regression analyses explored the associations between individual food intake timing indicators, food timing profiles, and GWG.

Results:

We identified four food timing profiles: extended window eating (n = 133; earliest first eating episode and the longest eating window), restricted window eating (n = 120; latest first eating episode and shortest eating window), early eating (n = 188; earliest caloric midpoint), and typical eating (n = 200; food intake aligning with the sample median). Participants with a restricted window eating profile (vs. typical eating profile) had an increased risk of insufficient GWG (unadjusted relative risk [RR] = 1.85, 95% confidence interval [CI] 1.12, 3.05). Each hour increase in the timing of the last eating episode was associated with 0.39 kg (0.03, 0.75) higher GWG. Both associations were attenuated in adjusted models and no longer statistically significant in adjusted models.

Conclusions:

We identified four distinct midpregnancy food timing profiles, but these profiles were not independently associated with GWG. These findings suggest that midpregnancy food timing may not play a major role in GWG.

Introduction

Excessive or insufficient gestational weight gain (GWG), based on Institute of Medicine (IOM) guidelines,¹ is associated with an increased risk for a range of adverse perinatal and child outcomes (e.g., gestational diabetes, preeclampsia, low birth weight, and childhood obesity).^1

–6 According to the latest CDC data, ∼50% of individuals had excessive GWG, 20% had insufficient GWG, and only one-third had adequate GWG.⁷ Developing and implementing strategies to promote healthy GWG should be prioritized.

Interventions designed to promote healthy GWG include diet counseling, which typically consists of making healthier food choices (e.g., consuming less saturated fat, eating more fruits and vegetables, etc.) to improve diet quality and to match energy intake with energy expenditure.⁸ While overall diet quality is considered a key contributor to healthy GWG, recent attention has been directed toward the role of food timing patterns.^9,10 Food timing indicators such as eating windows (i.e., the timeframe that people consume their meals and snacks) and the timing of the first and last eating episodes are associated with obesity in general adult populations.¹¹ For example, late-night eating has been linked to poor metabolic profiles, including increased insulin resistance and higher body mass index (BMI).^12
–14

Late eating may contribute to excess GWG and impair maternal glucose regulation by disrupting circadian rhythms, which affect macronutrient metabolism and glucose tolerance. In nonpregnant individuals, late eating has been shown to reduce fat oxidation and increase fat accumulation over time.^15
–17 During pregnancy, most GWG is due to fat-free mass and volume expansion, with fat mass accumulation varying by prepregnancy weight.^18
–20 For individuals with overweight or obesity, additional fat storage is unnecessary, potentially making late eating a contributor to excess GWG. Additionally, late eating may raise the risk of gestational diabetes by both increasing weight gain and reducing glucose tolerance at night, a concern during pregnancy when insulin resistance naturally rises to support fetal growth.^21,22

Research on food intake timing patterns during pregnancy and GWG is limited by small sample sizes and homogeneous populations, restricting generalizability.^23
–25 For example, studies by Balieiro²³ and Gontijo²⁴ in Brazil and Xiong et al.²⁵ in China examined chronotype, food timing, and circadian energy distribution, but none included more than 234 participants. This gap highlights the need for further research to explore how food timing impacts GWG and identify potential intervention targets.

This study addresses existing research gaps by characterizing midpregnancy food timing patterns in pregnant individuals, using latent profile analysis to identify distinct food timing profiles based on the combination of individual food timing indicators (i.e., the timing of the first and last eating episode, caloric midpoint time, and the total eating window). Additionally, we examine the association between individual food timing indicators, food timing profiles, and GWG. We hypothesize that having a shorter eating window, consuming the first and last meals earlier in the day, and consuming most calories earlier in the day will be associated with a higher likelihood of insufficient and excessive GWG.

Materials and Methods

Study design

This is a secondary data analysis of the Comparison of Two Screening Strategies for Gestational Diabetes (GDM2) trial, a single-center, parallel-arm, comparative effectiveness trial that aimed to compare the rates of perinatal outcomes among women randomized to receive either the International Associations of Diabetes and Pregnancy Study Groups (IADPSG) or Carpenter−Coustan gestational diabetes (GDM) screening approach.^26,27 Between June 2015 and February 2019, participants were recruited from 10 obstetrics clinics affiliated with University of Pittsburgh Medical Center Magee-Womens Hospital. Recruitment methods included physician referrals, flyers, brochures, electronic newsletters, and social media advertisements. This diverse approach aimed to minimize selection bias by reaching a broad spectrum of women receiving prenatal care in general obstetrics and Maternal-Fetal Medicine clinics. The University of Pittsburgh’s Institutional Review Board approved the study design and procedures. Participants provided written informed consent before enrollment. An independent data safety monitoring board within the University of Pittsburgh Clinical and Translational Science Institute provided oversight. GDM2 trial was registered at clinicaltrials.gov (NCT02309138).

Participants were recruited before routine screening for GDM between 18 to 28 weeks of gestation. Women were eligible if they were between 18 and 45 years old and had a singleton pregnancy at screening. Exclusion criteria included: (1) preexisting type 1 or 2 diabetes mellitus, (2) diabetes diagnosed <24 weeks gestation, (3) hypertension requiring medication, (5) corticosteroid use in the past 30 days, (6) anticipated preterm delivery due to maternal or fetal indications before 34 weeks gestation, (7) advanced HIV, (8) severe liver disease, or (9) gastric bypass surgery or other illness that precluded the participant from drinking the glucola solution (Supplementary Table S1). A total of 921 participants were enrolled in the GDM2 trial. For this secondary analysis, we excluded participants with missing food recall (n = 109) or gestational weight data (n = 78).

At baseline (visit 1), between 24 and 28 weeks of gestation, all participants received a nonfasting 50 g glucola solution to drink in 5 − 10 minutes, followed by a blood draw one hour after completing the glucola solution. Participants with a blood glucose value of 200 mg/dL or higher (11.1 mmol/L or higher) after the initial 50 g glucola solution were presumed to have gestational diabetes and excluded from the study. Participants with a blood glucose value <200 mg/dL proceeded with visit 2, during which they were randomized at a 1:1 ratio to receive either a 2-hour 75 g (IADPSG) or 3-hour 100 g (Carpenter−Coustan) oral glucose tolerance test between 25 and 32 weeks of gestation.²⁶

Diet assessment and food timing indicators

Dietary intake was assessed using two 24-hour dietary recalls on random nonconsecutive weekdays. Diet data were processed using the Automated Self-Administered 24-hour (ASA24®) Dietary Assessment Tool, which generates detailed reports on nutrient intake and food group consumption.²⁸ We used an average of the two assessments to calculate direct timing variables: first eating episode, last eating episode, eating window (difference between first and last eating episode), and caloric midpoint time (clock time at which 50% of the total daily calories were consumed). An eating episode was defined as consuming a calorie-containing beverage or meal.

We used the Healthy Eating Index 2015 (HEI-2015) to determine the participants’ overall diet quality. The HEI score is based on 13 different aspects of someone’s diet, including total vegetables, greens/beans, total fruit, whole fruit, whole grains, dairy, total protein, seafood and plant protein, fatty acid ratio, saturated fats, sodium, refined grains, and added sugar. Scores range from 0 to 100, with higher scores indicating better diet quality.²⁹

Gestational weight gain

GWG was defined as the difference in measured weight at delivery and self-reported prepregnancy weight. We used IOM guidelines to define insufficient, excessive, and adequate GWG.¹ According to these guidelines, adequate GWG is defined as a weight gain of between 11.34 and 15.88 kg for people with a normal weight (i.e., BMI 18.5–24.9 kg/m²), 6.80 and 11.34 kg for people with prepregnancy overweight (i.e., BMI 25–29.9 kg/m²) and between 4.99 and 9.07 kg for people with prepregnancy obesity. (i.e., BMI ≥30.0 kg/m²). Insufficient and excessive GWG was defined as weight gain below or above these guidelines, respectively. Adjusted maternal weight gain was defined as total GWG minus infant birth weight to account for the contribution of fetal weight to weight gain.

Covariates

Baseline covariates included demographic, behavioral, psychosocial, and clinical risk factors for GWG. Demographic characteristics included self-reported age and race (i.e., American Indian/Alaska Native, Asian, Native Hawaiian or other Pacific Islander, Black, White, or more than one race). Participants were also asked if they were Hispanic or Latinx. We grouped participants according to the United States census as Asian, White, Black, and “all other racial identities” because of small sample sizes. Participants also self-reported their employment status (i.e., part-time, full-time, or not working), education, marital status (i.e., married or/living with a partner or single/divorced, widowed), income, the number of adults in the household, current smoking status (i.e., cigarettes, e-cigarettes, or cigars per day), alcohol consumption, and current health status. Physical activity was assessed using the Godin Leisure-Time Exercise Questionnaire, a self-report tool that measures weekly frequencies of mild, moderate, and strenuous exercise during leisure time.³⁰ A Leisure Score Index is calculated by summing the frequency of mild, moderate, and strenuous activities, multiplying each type of activity by its respective metabolic equivalent task value. To assess insomnia symptom frequency, participants were asked how often they experienced trouble falling asleep, trouble staying asleep, waking several times per night, and waking after their usual amount of sleep feeling tired or worn out in the month. Scores for each question ranged from 0 (not at all) to 5 (22 to 31 days). The total insomnia symptom frequency was defined as the sum of each question and grouped into tertiles (i.e., 0–6, 7–13, and 14–20). Higher scores indicate more frequent insomnia symptoms.

Clinical characteristics included self-reported prepregnancy BMI and current health status. Participants self-reported their current health status as poor, fair, good, or excellent. We categorized their responses into two groups: poor/fair and good/excellent. Psychosocial factors included depressive symptoms and perceived stress. The Edinburgh Postnatal Depression Scale (EPDS) was used to measure depression symptoms.³¹ This 10-item self-reported scale identifies symptoms of depression such as feeling sad or miserable, the inability to cope, and sleep disruption in the past week. The scores range from 0 to 30, with higher scores indicating more depressive symptoms. An EPDS score of 13 or higher prompted us to provide supportive resources for depression. The scale has acceptable sensitivity (86%) and specificity (78%) for predicting depression. The Perceived Stress Scale (32) is a 14-item scale used to assess how often participants experienced stress in the past month. Scores range from 0 to 56, with higher scores indicating more perceived stress. In a general population, the test-retest correlations ranged between 0.55 and 0.85.³²

Statistical analyses

For our exposures of interest, we used continuous food intake indicators, including (1) the timing of the first eating episode, (2) the timing of the last eating episode, (3) caloric midpoint time, and (4) the total duration of the eating window. These individual indicators assess different aspects of eating timing and caloric distribution throughout the day, but they may also have interactive or synergistic effects. Therefore, we used latent profile analysis (LPA) to identify unmeasured classes (i.e., food timing profiles) from our measured continuous food timing indicators. Models with 2−10 classes were fitted to the data. Model fit was assessed using a combination of criteria to ensure a robust and interpretable solution. The Bayesian Information Criterion (BIC) balances model fit and complexity, with lower values indicating better model fit. In addition, we used the likelihood ratio test (LRT) to compare nested models and assess whether the inclusion of additional classes significantly improved model fit. We also considered the smallest class size, setting a threshold that any class should contain at least 5% of the total sample. This criterion helps to avoid overfitting and ensures that each identified class represents a sufficiently large subgroup within the population. Finally, we assessed the interpretability of the class solutions by examining the patterns of food timing indicators within each class. The final decision on the number of classes was based on triangulating these criteria: we selected the model that provided a balance of good fit indices (BIC, LRT), acceptable class sizes, and clear, interpretable class distinctions that aligned with theoretical expectations and prior research.³³

Descriptive statistics were calculated to compare participant demographics, diet, and clinical and psychosocial characteristics between the food timing profiles. For the primary aim, we used multivariable linear regression to examine the relationship between individual food timing indicators, food timing profiles identified by LPA, and total GWG adjusted for infant birth weight. We fit unadjusted and adjusted models that included covariates that could affect either the exposure or outcome variables. The adjusted models included education, marital status, income, employment status, perceived stress, insomnia symptoms, prepregnancy BMI, physical activity, gestational diabetes status, and gestational age at delivery. When food time profiles were the primary exposure, we used the ‘best’ food timing profile as the reference groups in all models. In selecting the ‘best’ food timing profile as our reference group, we selected the profile with the highest prevalence of meeting IOM’s adequate GWG recommendations. We performed a sensitivity analysis using total GWG unadjusted for infant birth weight. The results of the primary and sensitivity analyses were nearly identical. Thus, this article only presents the primary analysis (total GWG adjusted for infant birth weight).

We used Poisson regression to calculate the relative risk for insufficient and excessive GWG separately by the individual food timing indicators and food timing profiles using the same modeling approach described above, except we did not include prepregnancy BMI in our models. We excluded BMI from the Poisson regression models because prepregnancy BMI standardizes the outcome. Adequate GWG was the reference group in each analysis.

All analyses were performed in R (R version 4.0.2, RStudio Version 1.3.1073). We used the gtsummary package (version 1.5.2) to generate descriptive and regression tables and the depmixS4 package (Version 1.5.) for LPA.

Results

Of the 921 GDM2 study participants, 641 participants were included in the final analytic sample (Supplementary Fig. S1). Individuals included (vs. excluded) in the analysis had a higher prevalence of gestational diabetes (11% vs. 5.4%) and were older (median age 29 vs. 28 years). There were also significant differences in income levels, with a higher proportion of included individuals earning more than $61,000 (38% vs. 27%). Education level also varied significantly, with those included in the analysis more likely to have education beyond college (26% vs. 13%). Employment status (employed 70% vs. 61%) and depressive symptoms (median score 5.0 vs. 6.0) also differed. Finally, a lower proportion of those included in the analysis smoked cigarettes currently (12% vs. 19%) (Supplementary Table S2).

We identified four unique food timing profiles. Supplementary Figure S2 contains the model fit indices for selecting the final class solution. The BIC values indicate a good fit at four classes, with minimal gains in lowering BIC from adding more classes. The ‘extended window eating’ (n = 133) profile had the earliest first eating episode, at approximately 6:30 AM, and continued until 9:15 PM, resulting in the longest eating window of approximately 14.5 hours. This profile suggests a flexible approach to meal timing, with a caloric midpoint at around 2:30 PM. In contrast, the ‘restricted window eating’ (n = 120) profile started eating at 11:00 AM and finished by 7:20 PM, adhering to a more confined eating window of about 8.75 hours, with a caloric midpoint at 3:17 PM. The ‘early eating’ (n = 188) profile typically begins their food intake at 7:45 AM and concludes at 7:30 PM, maintaining a traditional eating window of approximately 11.64 hours, with a caloric midpoint at 1:30 PM. Lastly, the ‘typical eating’ (n = 200) profile had food intake timing patterns closest to the sample median.³⁴ Specifically, they started eating at 9:00 AM and ended at 9:00 PM, over an 11.75-hour window, with a caloric intake midpoint at 3:30 PM. Table 1 provides a more detailed summary of each food timing profile.

Table 1.

Food Timing Characteristics Overall and by Food Timing Profile

	Food timing profiles
Characteristic	Overall, n = 641^a	Extended window eating, n = 133^a	Restricted window eating, n = 120^a	Early eating, n = 188^a	Typical eating, n = 200^a
First eating episode	8:30 AM (7:22 AM, 9:45 AM)	6:30 AM (5:15 AM, 7:15 AM)	11:00 AM (10:00 AM, 12:19 PM)	7:45 AM (7:15 AM, 8:30 AM)	9:00 AM (9:30 AM, 10:00 AM)
Last eating episode	8:22 PM (7:30 PM, 9:25 PM)	9:15 PM (8:37 PM, 10:00 PM)	7:20 PM (6:00 PM, 8:30 PM)	7:30 PM (7:00 PM, 8:00 PM)	9:00 PM (8:30 PM, 10:00 PM)
Eating window, hours	11.75 (10.50, 13.00)	14.50 (13.87, 16.25)	8.75 (6.72, 9.25)	11.64 (10.87, 12.49)	11.75 (11.00, 12.50)
Caloric midpoint time	2:45 PM (1:00 PM, 4:30 PM)	2:30 PM (12:37 PM, 4:15 PM)	3:17 PM (1:00 PM, 5:15 PM	1:30 PM (12:30 PM, 3:00 PM	3:30 PM (2:15 PM, 5:03 PM

Food intake profiles were identified using latent profile analysis.

Median (interquartile range [IQR]).

The food timing profiles showed statistically significant differences in demographic, clinical, and behavioral characteristics. The early eating profile had a slightly higher median age (31 years) than the other classes. There were qualitative differences in income and education between the four profiles, with the early eating profile having the highest proportion of individuals with income over $61,000 (57%) and a formal education beyond college (39%). The early eating profile also had a higher proportion of married (72%) and employed (84%) individuals than the other classes. Health status appeared to be generally better in the early eating profile, with 93% reporting their health as ‘Excellent/Good.’ The extended window and restricted eating profiles had a lower frequency of insomnia symptoms than the early eating and daytime eating profiles. However, the restricted window eating profile had a higher percentage of participants who currently smoked (23%) and the highest median prepregnancy BMI. The four classes had similar distributions of gestational diabetes and randomization assignment (Table 2).

Table 2.

Demographic, Clinical, Psychosocial, and Behavioral Characteristics by Food Timing Profiles

Characteristic	Overall, n = 641^a	Daytime eating, n = 200^a	Extended window eating, n = 133^a	Restricted window eating, n = 120^a	Early eating, n = 188^a	p-value^b
Gestational diabetes						>0.90
No	571 (89%)	178 (89%)	117 (88%)	108 (90%)	168 (89%)
Yes	70 (11%)	22 (11%)	16 (12%)	12 (10%)	20 (11%)
Randomization arm						0.80
Carpenter Coustan	313 (49%)	94 (47%)	66 (50%)	56 (47%)	97 (52%)
IADPSG	328 (51%)	106 (53%)	67 (50%)	64 (53%)	91 (48%)
Age, years	29.0 (26.0, 32.0)	28.0 (25.0, 32.0)	29.0 (26.0, 32.0)	27.0 (23.0, 31.0)	31.0 (28.0, 33.0)	<0.01
Income level						<0.01
≤$30,000	285 (44%)	95 (48%)	56 (42%)	88 (73%)	46 (24%)
$31,000–60,000	110 (17%)	37 (19%)	20 (15%)	18 (15%)	35 (19%)
≥$61,000	246 (38%)	68 (34%)	57 (43%)	14 (12%)	107 (57%)
Education						<0.01
High school/GED or less	167 (26%)	49 (25%)	31 (23%)	62 (52%)	25 (13%)
Some college/vocational school	142 (22%)	52 (26%)	24 (18%)	37 (31%)	29 (15%)
College degree	165 (26%)	51 (26%)	42 (32%)	12 (10%)	60 (32%)
Beyond college	167 (26%)	48 (24%)	36 (27%)	9 (7.5%)	74 (39%)
Marital status						<0.01
No	315 (49%)	105 (53%)	63 (47%)	95 (79%)	52 (28%)
Yes	326 (51%)	95 (48%)	70 (53%)	25 (21%)	136 (72%)
Race						<0.01
Black or African American	185 (29%)	65 (33%)	34 (26%)	67 (56%)	19 (10%)
White	377 (59%)	106 (53%)	81 (61%)	39 (33%)	151 (80%)
All other races	79 (12%)	29 (15%)	18 (14%)	14 (12%)	18 (9.6%)
Currently employed						<0.01
No	193 (30%)	66 (33%)	35 (26%)	62 (52%)	30 (16%)
Yes	448 (70%)	134 (67%)	98 (74%)	58 (48%)	158 (84%)
Health status						0.04
Fair/Poor	74 (12%)	28 (14%)	12 (9.0%)	20 (17%)	14 (7.4%)
Excellent/Good	567 (88%)	172 (86%)	121 (91%)	100 (83%)	174 (93%)
Perceived stress	4.0 (2.0, 6.0)	5.0 (2.0, 7.0)	5.0 (3.0, 6.0)	5.0 (3.0, 7.0)	4.0 (2.0, 6.0)	0.03
Depressive symptoms	5.0 (2.0, 9.0)	6.0 (2.0, 9.0)	6.0 (3.0, 8.0)	6.0 (3.0, 11.0)	4.0 (2.0, 8.0)	0.02
Insomnia symptom frequency						0.50
0–6	311 (49%)	93 (47%)	70 (53%)	64 (53%)	84 (45%)
7–13	239 (37%)	74 (37%)	44 (33%)	42 (35%)	79 (42%)
14–20	91 (14%)	33 (17%)	19 (14%)	14 (12%)	25 (13%)
Godin leisure physical activity score						0.70
<24	300 (47%)	100 (50%)	61 (46%)	54 (45%)	85 (45%)
≥24	341 (53%)	100 (50%)	72 (54%)	66 (55%)	103 (55%)
Currently smokes						<0.01
No	563 (88%)	170 (85%)	120 (90%)	93 (78%)	180 (96%)
Yes	78 (12%)	30 (15%)	13 (9.8%)	27 (23%)	8 (4.3%)
Currently drinks						0.40
No	620 (97%)	193 (97%)	128 (96%)	114 (95%)	185 (98%)
Yes	21 (3.3%)	7 (3.5%)	5 (3.8%)	6 (5.0%)	3 (1.6%)
Prepregnancy BMI, kg/m²	25.2 (21.9, 30.4)	25.1 (22.0, 30.7)	24.5 (21.3, 28.7)	26.6 (21.9, 32.5)	25.8 (22.0, 30.3)	0.20
Total kcals/day	1,994.5 (1,473.4, 2,500.7)	2,090.1 (1,636.9, 2,622.7)	2,190.1 (1,715.3, 2,613.7)	1,541.7 (996.9, 2,090.0)	1,935.2 (1,479.2, 2,396.2)	<0.01
Carbohydrate, %	60.6 (55.3, 65.3)	60.5 (55.8, 65.4)	61.1 (55.8, 67.2)	59.1 (53.0, 63.9)	61.1 (55.4, 65.2)	0.04
Protein, %	19.4 (16.3, 23.3)	19.2 (15.9, 22.9)	18.7 (16.3, 23.3)	20.6 (16.3, 23.9)	19.6 (16.6, 23.4)	0.4
Fat, %	19.6 (16.6, 22.7)	19.4 (16.6, 22.7)	19.5 (15.8, 22.1)	20.8 (18.3, 24.0)	19.3 (16.4, 22.1)	0.001
Health Eating Index	54.5 (48.3, 59.9)	54.6 (48.4, 59.2)	55.8 (49.9, 61.8)	47.7 (43.7, 53.7)	56.7 (50.3, 61.8)	<0.01
Adjusted GWG, kg	11.4 (7.6, 15.7)	11.3 (7.5, 15.0)	11.3 (8.2, 16.3)	10.4 (6.2, 16.9)	11.8 (8.1, 15.7)	0.60
GWG recommendations						0.02
Excessive	365 (57%)	104 (52%)	80 (60%)	68 (57%)	113 (60%)
Recommended	167 (26%)	65 (33%)	33 (25%)	21 (18%)	48 (26%)
Inadequate	109 (17%)	31 (16%)	20 (15%)	31 (26%)	27 (14%)

n (%); Median (IQR).

Pearson’s Chi-squared test; Kruskal-Wallis rank sum test; Fisher’s exact test.

IADPSG, International Associations of Diabetes and Pregnancy Study Groups; GWG, gestational weight gain; GED, General Education Development.

There was no difference in GWG among the four food timing profiles (Table 3). However, we identified unadjusted associations between some individual food timing indicators and GWG. Specifically, our unadjusted models showed that each hour delay in the timing of the last eating episode was associated with 0.39 kg (0.03, 0.75) higher GWG. There was no association in the adjusted models or between any other food timing indicator and GWG.

Table 3.

The Association Between Midpregnancy Food Timing Patterns and Total Gestational Weight Gain

		Unadjusted	Adjusted*
Characteristic	Median (IQR)	Beta (95% CI)	Beta (95% CI)
First eating episode	8:30 AM (7:22 AM, 9:45 AM)	−0.081 (−0.34, 0.179)	0.04 (−0.22, 0.3)
Last eating episode	8:22 PM (7:30 PM, 9:25 PM)	0.39 (0.03, 0.75)	0.29 (−0.06, 0.63)
Eating window, hours	11.75 (10.50, 13.00)	0.20 (−0.02, 0.41)	0.08 (−0.13, 0.3)
Caloric midpoint time	2:45 PM (1:00 PM, 4:30 PM)	−0.02 (−0.25, 0.21)	0.00 (−0.22, 0.22)

Food timing profiles	n (%)	Marginal Means, kg (95% CI)	Marginal Means, kg (95% CI)
Daytime Eating	200 (31%)	11.24 (9.94, 12.55)	11.22 (9.96, 12.49)
Extended Window Eating	133 (21%)	11.35 (9.84, 12.86)	11.31 (9.86, 12.77)
Restricted Window Eating	120 (19%)	10.77 (9.18, 12.37)	10.77 (9.20, 12.34)
Early Eating	188 (29%)	11.36 (9.89, 12.82)	11.30 (9.94, 12.67)

*Adjusted models include age, education, married/living with a partner, income, employment status, perceived stress, insomnia symptom frequency, prepregnancy BMI, physical activity, gestational diabetes status, gestational age at delivery, and race.

IQR, interquartile range; CI, confidence interval; BMI, body mass index.

There was no relationship between individual food timing indicators or food timing profiles and the risk of excessive GWG (Table 4). In the unadjusted analysis, participants with a restricted window eating profile (vs. typical eating profile) had an 85% increased risk of insufficient GWG. However, this association was attenuated in adjusted models. There were no other differences in insufficient GWG risk between the food timing profiles. Participants with insufficient gestational weight had their first eating episode nearly 30 minutes later and their last eating episode nearly 20 minutes earlier than participants who had recommended GWG. However, these differences were not statistically significant (Table 5).

Table 4.

The Associations between Midpregnancy Food Timing Patterns and Excessive Gestational Weight Gain

Characteristic	Adequate GWG, n = 167^a	Excessive GWG, n = 365^a	Unadjusted	Adjusted
Characteristic	Adequate GWG, n = 167^a	Excessive GWG, n = 365^a	RR	RR
First eating episode	8:31 AM (8:12 AM, 8:51 AM)	8:31 AM (8:18 AM, 8:44 AM)	1.00 (0.95, 1.05)	0.99 (0.94, 1.04)
Last eating episode	8:25 PM (8:10 PM, 8:39 PM)	8:19 PM (8:09 PM, 8:28 PM)	0.99 (0.92, 1.06)	0.98 (0.92, 1.05)
Eating window, hours	12.00 (10.75, 13.20)	11.75 (10.50, 13.23)	1.00 (0.96, 1.04)	1.00 (0.96, 1.04)
Caloric midpoint time	2:59 PM (2:36 PM, 3:31 PM)	2:49 PM (2:33 PM, 3:04 PM)	0.99 (0.95, 1.03)	0.99 (0.94, 1.03)
Food timing profiles
Typical Eating	65 (38%)	104 (62%)	—	—
Extended Window Eating	33 (29%)	80 (71%)	1.15 (0.86, 1.54)	1.17 (0.87, 1.57)
Restricted Window Eating	21 (24%)	68 (76%)	1.24 (0.91, 1.68)	1.18 (0.86, 1.61)
Early Eating	48 (30%)	113 (70%)	1.14 (0.87, 1.49)	1.17 (0.89, 1.55)

Adequate and excessive gestational weight gain is based in Institute of Medicine recommendations.

We used Poisson regression to calculate the relative risk for excessive gestational weight gain by the individual food timing indicators and food timing profiles. Adjusted models include gestational diabetes status, income, education, married/living with a partner, insomnia symptom frequency, physical activity, perceived stress, employment status, gestational age at delivery.

Median (IQR); n (row %).

GWG, gestational weight gain; RR, relative risk.

Table 5.

The Associations Between Midpregnancy Food Timing Patterns and Insufficient Gestational Weight Gain

Characteristic	Adequate GWG, n = 167^a	Inadequate GWG, n = 109^a	Unadjusted	Adjusted
Characteristic	Adequate GWG, n = 167^a	Inadequate GWG, n = 109^a	RR	RR
First eating episode	8:31 AM (8:12 AM, 8:51 AM)	9:00 AM (8:36 AM, 9:24 AM)	1.08 (0.98, 1.18)	1.02 (0.93, 1.12)
Last eating episode	8:25 PM (8:10 PM, 8:39 PM)	8:06 PM (7:48 PM, 8:23 PM)	0.94 (0.84, 1.05)	0.97 (0.87, 1.08)
Eating window, hours	12.00 (10.75, 13.20)	11.50 (9.50, 12.92)	0.93 (0.86, 1.00)	0.97 (0.90, 1.05)
Caloric midpoint time	2:59 PM (2:36 PM, 3:31 PM)	2:49 PM (2:21 PM, 3:17 PM)	0.99 (0.92, 1.06)	0.98 (0.91, 1.06)
Food timing profiles
Typical eating	65 (68%)	31 (32%)	—	—
Extended window eating	33 (62%)	20 (38%)	1.17 (0.66, 2.03)	1.32 (0.74, 2.35)
Restricted window eating	21 (40%)	31 (60%)	1.85 (1.12, 3.05)	1.55 (0.90, 2.68)
Early eating	48 (64%)	27 (36%)	1.11 (0.66, 1.87)	1.25 (0.72, 2.16)

Adequate and excessive GWG is based in Institute of Medicine recommendations.

We used Poisson regression to calculate the relative risk for insufficient GWG by the individual food timing indicators and food timing profiles. Adjusted models include gestational diabetes status, income, education, married or living with a partner, insomnia symptom frequency, physical activity, perceived stress, employment status, gestational age at delivery.

Median (IQR); n (row %).

GWG, gestational weight gain; RR, relative risk.

Discussion

Our study aimed to explore the association between midpregnancy food timing patterns and GWG in a well-characterized sample of over 600 pregnant patients in a comparative effectiveness trial of two GDM screening approaches. We identified four distinct food timing profiles: typical eating, extended window eating, restricted window eating, and early eating. Despite these differing eating behaviors, there were no significant differences in total adjusted GWG between the four food timing profiles. There were also no statistically significant differences in the association among the individual food timing indicators or food timing profiles and the risk of having excessive GWG. However, participants with a restricted eating window profile had the highest risk of insufficient GWG in unadjusted models, which was no longer significant after adjusting for covariates.

The results of the current study partially align with prior studies that examined the relationship between food timing and GWG. Balieiro et al. examined the effects of chronotype on eating patterns, energy, and macronutrient intake and distribution, and GWG in a cohort of 100 Brazilian pregnant people.²³ They found that pregnant people with an ‘eveningness’ chronotype ate breakfast later, had higher energy and carbohydrate intake at dinner, and gained more excess weight in the third trimester than people with a ‘morning’ chronotype (2.24 ± 0.25 vs. 1.22 ± 0.14, p < 0.001). In the same study population, Gontijo et al. found that those who consumed their first and last meals earlier in the day had better diet quality but no difference in GWG.²⁴ Xiong et al. used latent profile analysis to examine the effects of a circadian distribution of energy and macronutrient intake and GWG in a cohort of 234 Chinese pregnant people.²⁵ They identified four eating profiles characterized as having high energy and macronutrient intake at (1) night, (2) afternoon to early evening, (3) late morning to early afternoon, and (4) in the morning. People with a high energy and macronutrient intake in the afternoon to early evening profile had high GWG rates. In contrast, we did not identify associations between the timing of the first meal and GWG, but people who ate their last meal later in the day did have higher GWG in unadjusted models. The differences in the study results may be due to differences in sample characteristics, exposure and outcome assessments, and statistical analysis, which should be explored further in future research.

In nonpregnant adults, consuming more calories later in the day is associated with adverse metabolic consequences, including overweight and obesity, even after adjusting for physical activity or sleep.^35

–39 Mechanistically, eating later in the day may impose a misalignment between metabolism needs and endogenous circadian rhythms, creating an energy imbalance that leads to weight gain. As humans our metabolic processes are optimized to be active during the day and restful at night. For instance, indirect calorimetry data have shown that in healthy adults, carbohydrate oxidation was highest in the biological morning, and fat oxidation was highest in the biological evening in preparation for fasting during overnight sleep.⁴⁰ Late eating thus may challenge our metabolism, which has been optimized for handling food during the daytime. In support of this, interventional studies in healthy adults have demonstrated that eating a late dinner or snack reduces both dietary fat oxidation and total body fat oxidation, which potentially could lead to the development of obesity in the future.^15,41

While we did not find strong evidence linking food timing to GWG in our adjusted models, these findings contribute to the limited research on food timing patterns during pregnancy. Studies on US adults have typically excluded pregnant persons from dietary analyses,^42,43 leaving gaps in understanding how meal timing may affect gestational outcomes. Our results could inform traditional nutritional guidance, though further research is needed to explore whether specific timing behaviors could optimize gestational weight outcomes. Additionally, lifestyle modifications during pregnancy are important, given the limited use of pharmacotherapy for cardiometabolic conditions during this time and the long-term health implications of excessive weight gain in pregnancy, including an increased risk of type 2 diabetes in those with gestational diabetes.⁴⁴

While previous research was limited by small sample sizes and homogeneous populations, our study included a relatively large and diverse sample of over 600 pregnant individuals, enhancing the generalizability of our findings. Strengths of our study include a well-characterized clinical sample, identification of food intake timing profiles, and the inclusion of multiple covariates. However, several limitations exist. First, the use of 24-hour recalls only in the second trimester may not capture typical or changing eating patterns throughout pregnancy. Future research should assess food timing across pregnancy and examine its association with GWG. Second, we lacked data on sleep and wake times, limiting our ability to explore the interaction between meal and sleep timing. Third, LPA, while useful, may produce patterns specific to this sample that may not generalize to other populations. Additionally, although our sample size was larger than prior studies, it may have been insufficient to detect smaller eating profiles or subgroup differences. For example, insufficient GWG risk was highest in participants with a restricted eating window profile in unadjusted but not adjusted models, possibly due to the small size of this subgroup. Another limitation is the absence of data on morning sickness and hyperemesis gravidarum, conditions that can persist beyond the first trimester and potentially influence dietary intake and weight gain. We also acknowledge that seasonal changes may influence behaviors dietary intake, physical activity, and nocturnal artificial light exposure. However, the effect of light on weight regulation is complex.⁴⁵ An accurate evaluation of the effect of artificial light exposure on GWG would necessitate detailed assessment of light exposure, which was not collected as part of the current study. Future studies should examine other factors, such as artificial light exposure and the interaction of food timing with sleep and physical activity, to provide a more comprehensive understanding of their combined impact on GWG.

In conclusion, our study indicated that food timing patterns were not independently associated with GWG. The lack of significant results may be attributed to the complex interplay of sociodemographic, behavioral, and clinical factors. Future research should measure food timing throughout pregnancy and consider potential interactions with sleep patterns, physical activity, and circadian rhythms. Additionally, examining the impact of food timing on other pregnancy-related outcomes, such as metabolic health or birth outcomes, could provide valuable insights for clinical practice and dietary recommendations.

Data Share Statement

Data described in the article, code book, and analytic code will be made available upon request pending application and approval.

Footnotes

Authors’ Contributions

M.S.H.: Conceived the article, analyzed data, wrote the article, and had primary responsibility for the final content. E.M.D.: Cesigned the study. All the authors read the article, provided feedback, and approved the final version.

Author Disclosure Statement

Over the past 3 years, D.B. has served as a paid or unpaid consultant to Sleep Number, Idorsia, and Eisai. All consulting agreements have been for a total of less than $5,000 per year from any single entity. D.B. is an author of the Pittsburgh Sleep Quality Index, Pittsburgh Sleep Quality Index Addendum for PTSD (PSQI-A), Brief Pittsburgh Sleep Quality Index (B-PSQI), Daytime Insomnia Symptoms Scale, Pittsburgh Sleep Diary, Insomnia Symptom Questionnaire, and RU_SATED (copyrights held by the University of Pittsburgh). These instruments have been licensed to commercial entities for fees. He is also coauthor of the Consensus Sleep Diary (copyright held by Ryerson University), which is licensed to commercial entities for a fee. He has received grant support from National Institute of Health, Patient-Centered Outcomes Research Institute, Agency for Healthcare Research and Quality, and the Veterans Affairs E.M.D. is a United States Preventive Services Task Force (USPSTF) member. This article does not necessarily represent the views and policies of the USPSTF.

Funding Information

This work was supported by funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development [R01HD079647 (PI: Davis)], the University of Pittsburgh Clinical and Translational Science Institute [UL1TR001857].

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2

Supplementary Table S1

Supplementary Table S2

References

American College of O, Gynecologists. ACOG Committee opinion no. 548: Weight gain during pregnancy. Obstetrics and Gynecology, 2013; 121(1):210–212.

Voerman

, Santos

, Inskip

, et al.; LifeCycle Project-Maternal O, Childhood Outcomes Study G. Association of gestational weight gain with adverse maternal and infant outcomes. JAMA, 2019; 321(17):1702–1715.

Catalano

, Presley

, Minium

, et al. Fetuses of obese mothers develop insulin resistance in utero. Diabetes Care, 2009; 32(6):1076–1080.

Howell

, Powell

. Effects of maternal obesity on placental function and fetal development. Reproduction, 2017; 153(3):R97–R108.

Dude

, Grobman

, Haas

, et al. Gestational weight gain and pregnancy outcomes among nulliparous women. Am J Perinatol, 2021; 38(2):182–190.

Catalano

, Shankar

. Obesity and pregnancy: Mechanisms of short term and long term adverse consequences for mother and child. BMJ, 2017; 356:j1.

Anonymous. QuickStats: Gestational Weight Gain Among Women with Full-Term, Singleton Births, Compared with Recommendations—48 States and the District of Columbia, 2015. MMWR Morb Mortal Wkly Rep, 2016; doi: 10.1558/5/mmwr.mm6540a10

Shieh

, Cullen

, Pike

, et al. Intervention strategies for preventing excessive gestational weight gain: Systematic review and meta-analysis. Obes Rev, 2018; 19(8):1093–1109.

Davis

, Rogers

, Coates

, et al. The impact of meal timing on risk of weight gain and development of obesity: A review of the current evidence and opportunities for dietary intervention. Curr Diab Rep, 2022; 22(4):147–155.

10.

Ahluwalia

. Chrononutrition-when we eat is of the essence in tackling obesity. Nutrients, 2022; 14(23).

11.

Schuppelius

, Peters

, Ottawa

, et al. Time restricted eating: A dietary strategy to prevent and treat metabolic disturbances. Front Endocrinol (Lausanne), 2021; 12:683140.

12.

Thomas

, Zaman

, Cornier

M-A

, et al. Later meal and sleep timing predicts higher percent body fat. Nutrients, 2020; 13(1).

13.

Takahashi

, Tahara

. Timing of food/nutrient intake and its health benefits. J Nutr Sci Vitaminol (Tokyo), 2022; 68(Supplement):S2–S4.

14.

Dashti

, Gómez-Abellán

, Qian

, et al. Late eating is associated with cardiometabolic risk traits, obesogenic behaviors, and impaired weight loss. Am J Clin Nutr, 2021; 113(1):154–161.

15.

Kelly

, McGuinness

, Buchowski

, et al. Eating breakfast and avoiding late-evening snacking sustains lipid oxidation. PLoS Biol, 2020; 18(2):e3000622.

16.

Allison

, Hopkins

, Ruggieri

, et al. Prolonged, controlled daytime versus delayed eating impacts weight and metabolism. Curr Biol, 2021; 31(4):908.

17.

Christian

, Carroll

, Teti

, et al. Maternal sleep in pregnancy and postpartum Part I: Mental, physical, and interpersonal consequences. Curr Psychiatry Rep, 2019; 21(3):20.

18.

Most

, Dervis

, Haman

, et al. Energy intake requirements in pregnancy. Nutrients, 2019; 11(8).

19.

Most

, Amant

, Hsia

, et al. Evidence-based recommendations for energy intake in pregnant women with obesity. J Clin Invest, 2019; 129(11):4682–4690.

20.

Dalfra

, Burlina

, Lapolla

. Weight gain during pregnancy: A narrative review on the recent evidences. Diabetes Res Clin Pract, 2022; 188:109913.

21.

Mason

, Qian

, Adler

, et al. Impact of circadian disruption on glucose metabolism: Implications for type 2 diabetes. Diabetologia, 2020; 63(3):462–472.

22.

Lain

, Catalano

. Factors that affect maternal insulin resistance and modify fetal growth and body composition. Metab Syndr Relat Disord, 2006; 4(2):91–100.

23.

Balieiro

LCT

, Gontijo

, Marot

, et al. Is chronotype associated with dietary intake and weight gain during pregnancy? A prospective and longitudinal study. Nutrition, 2022; 94:111530.

24.

Gontijo

, Balieiro

LCT

, Teixeira

, et al. Effects of timing of food intake on eating patterns, diet quality and weight gain during pregnancy. Br J Nutr, 2020; 123(8):922–933.

25.

Xiong

, Cui

, Dong

, et al. Effect of circadian distribution of energy and macronutrients on gestational weight gain in chinese pregnant women. Nutrients, 2023; 15(9).

26.

Abebe

, Scifres

, Simhan

, et al. Comparison of two screening strategies for Gestational Diabetes (GDM(2)) Trial: Design and rationale. Contemp Clin Trials, 2017; 62:43–49.

27.

Davis

, Abebe

, Simhan

, et al. Perinatal outcomes of two screening strategies for gestational diabetes mellitus: A randomized controlled trial. Obstet Gynecol, 2021; 138(1):6–15.

28.

Subar

, Kirkpatrick

, Mittl

, et al. The Automated Self-Administered 24-hour dietary recall (ASA24): A resource for researchers, clinicians, and educators from the National Cancer Institute. J Acad Nutr Diet, 2012; 112(8):1134–1137.

29.

Kirkpatrick

, Reedy

, Krebs-Smith

, et al. Applications of the healthy eating index for surveillance, epidemiology, and intervention research: Considerations and caveats. J Acad Nutr Diet, 2018; 118(9):1603–1621.

30.

Amireault

, Godin

. The Godin-Shephard leisure-time physical activity questionnaire: Validity evidence supporting its use for classifying healthy adults into active and insufficiently active categories. Percept Mot Skills, 2015; 120(2):604–622.

31.

Cox

, Holden

, Sagovsky

. Detection of postnatal depression. Development of the 10-item Edinburgh postnatal depression scale. Br J Psychiatry, 1987; 150:782–786.

32.

Cohen

, Kamarck

, Mermelstein

. A global measure of perceived stress. J Health Soc Behav, 1983; 24(4):385–396.

33.

Weller

, Bowen

, Faubert

. Latent class analysis: a guide to best practice. J Black Psychology, 2020; 46(4):287–311.

34.

Farsijani

, Mao

, Cauley

, et al. Comprehensive assessment of chrononutrition behaviors among nationally representative adults: Insights from National Health and Nutrition Examination Survey (NHANES) data. Clin Nutr, 2023; 42(10):1910–1921.

35.

Berteus Forslund

, Lindroos

, Sjostrom

, et al. Meal patterns and obesity in Swedish women-a simple instrument describing usual meal types, frequency and temporal distribution. Eur J Clin Nutr, 2002; 56(8):740–747.

36.

Wang

, Patterson

, Ang

, et al. Timing of energy intake during the day is associated with the risk of obesity in adults. J Hum Nutr Diet, 2014; 27(Suppl 2):255–262.

37.

Berg

, Lappas

, Wolk

, et al. Eating patterns and portion size associated with obesity in a Swedish population. Appetite, 2009; 52(1):21–26.

38.

Baron

, Reid

, Kern

, et al. Role of sleep timing in caloric intake and BMI. Obesity (Silver Spring), 2011; 19(7):1374–1381.

39.

, Musso

, Beccuti

, et al. Consuming more of daily caloric intake at dinner predisposes to obesity. A 6-year population-based prospective cohort study. PLoS One, 2014; 9(9):e108467.

40.

Rynders

, Morton

, Bessesen

, et al. Circadian rhythm of substrate oxidation and hormonal regulators of energy balance. Obesity (Silver Spring), 2020; 28(Suppl 1):S104–S113.

41.

, Brereton

, Schweitzer

, et al. Metabolic effects of late dinner in healthy volunteers-a randomized crossover clinical trial. J Clin Endocrinol Metab, 2020; 105(8):2789–2802.

42.

Kant

. Eating patterns of US adults: Meals, snacks, and time of eating. Physiol Behav, 2018; 193(Pt B):270–278.

43.

O’Connor

, Reedy

, Graubard

, et al. Circadian timing of eating and BMI among adults in the American Time Use Survey. Int J Obes, 2022; 46(2):287–296.

44.

Diaz-Santana

, O’Brien

, Park

, et al. Persistence of risk for Type 2 Diabetes after gestational diabetes mellitus. Diabetes Care, 2022; 45(4):864–870.

45.

Ishihara

, Courville

, Chen

. The complex effects of light on metabolism in humans. Nutrients, 2023; 15(6).

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB

0.03 MB

0.84 MB