Chrononutrition during Pregnancy and Its Association with Maternal and Offspring Outcomes: A Systematic Review and Meta-Analysis of Ramadan and Non-Ramadan Studies

Much evidence suggests that food intakes and eating patterns are major determinants of the phase of peripheral circadian clocks, and desynchronization between them is thought to contribute to the development of metabolic disorders. However, much remains to be understood about how different dimensions of chrononutrition during pregnancy affect pregnant women’s and their offspring’s health outcomes. Therefore, we systematically reviewed and integrated all emerging evidence on chrononutrition during pregnancy (including meal skipping, meal frequency, night eating, and (Ramadan) fasting) and their relationships with maternal and offspring outcomes. The results suggest that meal skipping and night eating during pregnancy were generally associated with adverse pregnancy and birth outcomes, whereas no strong conclusion could be reached for meal frequency. In our meta-analysis, Ramadan fasting did not seem to be related with birth weight or gestational age at birth, but evidence for other mother–offspring outcomes was inconsistent. To further elucidate the effect of chrononutrition factors on maternal and offspring health outcomes, larger and well-conducted prospective cohort and interventional studies are needed. In addition, information on covariates such as physical activity, sleep, diet quality and quantity, fasting days, fasting period per day, and trimester exposure should also be collected and considered during analysis.


Introduction
The nutritional status and dietary intake of women during pregnancy are not only related to their own health, but also the health of their offspring. Since nutritional requirements during pregnancy differ considerably from those of non-pregnant women, the daily reference values of many nutrients and energy intake are different during pregnancy to ensure healthy pregnancy weight gain, optimal fetal development, and favorable longterm outcomes of offspring [1]. Maternal malnutrition and excessive gestational weight gain (GWG) increase the risk of adverse offspring and maternal health outcomes [2]. For example, inadequate GWG was associated with increased risks of preterm delivery and delivering small-for-gestational age infants; excessive GWG was associated with pregnancyinduced hypertension, cesarean delivery, macrosomia [3], and childhood obesity [4]. Hence, optimal dietary intake and nutritional status during pregnancy are vital for maternal and child health.
In addition to dietary content (energy intake and nutrients), multiple lines of evidence now suggest that chrononutrition (interaction between nutrition and the circadian rhythm)

Search Strategy
We searched two online databases (PubMed and Embase) from database conception until 19 September 2022 for relevant papers that investigated maternal chrononutrition and mother-offspring outcomes. The search terms were developed according to the PEO (Participants AND Exposure AND Outcome) framework. A combination of the following keywords was used for the search: participants (pregnancy OR gravidity OR gestation OR antenatal OR postpartum OR mother etc.) and exposure (chrononutrition OR meal frequency OR meal skipping OR eating episode etc.). The exact search strategy for the databases is shown in Supplementary Table S1. Since the inclusion of outcome keywords resulted in limited improvement in search specificity and was less comprehensive with regard to a range of health outcomes corresponding to maternal chrononutrition, we left out the outcome keywords.

Study Selection
The identified citations from the online searches were de-duplicated before first-stage screening using titles and abstracts. Full texts of potentially relevant articles were then retrieved and subjected to second-stage screening.
We included original, peer-reviewed, experimental (randomized or non-randomized clinical trials), and observational (prospective/retrospective cohort, case-control, and crosssectional) studies reporting data on chrononutrition factors during the peri-pregnancy period (pre-pregnancy, pregnancy, and post-delivery) in relation to maternal or child health outcomes. We excluded (1) meeting abstracts, comments, editorials, study protocols, reviews, and meta-analyses; (2) animal studies and in vitro studies; (3) studies that did not study chrononutrition factors during the peri-pregnancy period in relation to maternal or child outcomes; (4) studies without a control group (e.g., case reports and series); and (5) articles not written in English.

Data Extraction
The data extracted from the included articles included the first author's name, publication year, study design, city and country of study, population, sample size, study period, duration of fasting per day (for Ramadan studies), personal characteristics of participants (age, BMI), exposure factors, outcomes, gestational age of exposure assessment, and covariates. If the literature did not mention fasting hours during Ramadan (n = 20), we estimated the fasting time as the duration from the Islamic Fajr to Maghrib prayers (from dawn to sunset) according to the reported study city and calendar/reported dates of Ramadan fasting.

Risk of Bias Assessment
To assess risk of bias in the included studies, Mixed Method Appraisal Tool Version 2018 (MMAT) was used. The (A) quantitative non-randomized studies criteria used in the MMAT include the following domains: (1) whether the sample is representative of the population; (2) appropriateness of measurement; (3) completeness of results; (4) adjustment of confounders; and (5) compliance with interventions (or in observational studies, whether exposure occurred as intended). The (B) quantitative randomized controlled trials criteria include the following domains: (1) is randomization appropriately performed? (2) Are the groups comparable at baseline? (3) Are there complete outcome data? (4) Are outcome assessors blinded to the intervention provided? (5) Did the participants adhere to the assigned intervention? Within each domain, each study was categorized based on whether they had met the criteria (marked as "Yes") or not (marked as "No") (Supplementary Table S2). For the purpose of subgroup analysis in the meta-analysis only, we categorized the studies as having low, medium, or high risk of bias if they did not meet ≤2, 3, or ≥4 (out of 5) of the domain criteria, respectively.

Data Analysis
Due to the mostly heterogenous results from the included studies with various definitions of exposure and outcomes, especially for non-Ramadan studies involving various aspects of maternal chrononutrition factors, most of the results and discussions were narrative (i.e., without meta-analysis). However, since there was a sizeable number of studies involving Ramadan fasting and birth weight/low birth weight and gestational age at birth/preterm birth, we conducted meta-analyses to generate summary estimates. A smaller number of studies also looked at the influence of Ramadan fasting on common maternal outcomes (blood glucose and weight gain), and we pooled these estimates as well.
For meta-analyses, pooled mean differences with 95% confidence intervals were derived for continuous outcomes, whereas pooled odds ratios with 95% confidence intervals were calculated for binary outcomes. A random effects meta-analysis was used in anticipation of heterogeneity due to differences in study design. The Cochran Q test and I 2 statistic were used to evaluate statistical heterogeneity among studies, and I 2 values of 25%, 50%, and 75% correspond to low, moderate, and high degrees of heterogeneity, respectively [18]. For outcomes with a sufficient number of studies, we conducted stratified analyses to identify potential sources of heterogeneity by different study-level characteristics (number of participants, age, duration of fasting per day, region, BMI, and MMAT). If studies presented data grouped by fasting days or duration of fasting per day, we selected the group with the longest fasting duration (i.e., most extreme contrast compared to the non-fasting group). If a study provided effect estimates by trimester, we derived a within-study pooled mean and standard deviation first before including these data in the meta-analysis. However, for

Study Characteristics
The included literature investigated the relationships of chron pregnancy with maternal, fetal, neonatal, and childhood outcomes. Th factors assessed can be broadly categorized into meal skipping (n = 7), n meal frequency (n = 11), night fasting duration (n = 2), and Ramadan f mentioned above, due to substantially heterogenous exposure and o Ramadan fasting literature on meal frequency, meal skipping, night eati ing duration were reviewed qualitatively. For the Ramadan fasting lit portant birth outcomes [gestational age at birth (n = 11); risk of preterm weight (n = 19); and risk of low birth weight (n = 4)] and maternal outco

Study Characteristics
The included literature investigated the relationships of chrononutrition during pregnancy with maternal, fetal, neonatal, and childhood outcomes. The chrononutrition factors assessed can be broadly categorized into meal skipping (n = 7), night eating (n = 8), meal frequency (n = 11), night fasting duration (n = 2), and Ramadan fasting (n = 41). As mentioned above, due to substantially heterogenous exposure and outcomes, the non-Ramadan fasting literature on meal frequency, meal skipping, night eating, and night fasting duration were reviewed qualitatively. For the Ramadan fasting literature, some important birth outcomes [gestational age at birth (n = 11); risk of preterm birth (n = 7); birth weight (n = 19); and risk of low birth weight (n = 4)] and maternal outcomes [weight gain (n = 8) and blood sugar (n = 5)] were analyzed quantitatively, while other outcomes were less frequently reported and thus reviewed qualitatively. Studies investigating meal skipping during pregnancy and maternal/offspring outcomes are summarized in Figure 2 and Table 1.
In two self-matched case-crossover studies (used to assess health event triggers) from the United States, compared with a week [20] or three days [21] before labor (control period), the odds of imminent spontaneous labor were 5 [21] times as high within 24 h (hazard period) of skipping one or more meals [20,21]. However, these results may be affected by reverse causality since irregular meal intake can be a sign rather than a cause of the approaching labor due to a lack of appetite [20].
In summary, these results indicate that skipping meals during pregnancy may be related to maternal malnutrition [22][23][24], GDM [25], and weight status [26]. However, since most studies are cross-sectional in nature, it is unclear whether these associations are causal.

Night Eating
Studies investigating night eating during pregnancy and its associations with maternal/offspring outcomes are summarized in Figure 3 and Table 2.

Night Eating
Studies investigating night eating during pregnancy and its associations with maternal/offspring outcomes are summarized in Figure 3 and Table 2.
Two studies explored the associations of maternal night eating with GWG and postpartum weight retention (PPWR). A prospective cohort study from Brazil collected three 24 h dietary recalls in each trimester and pregnant women were classified as having a "lower" or "higher" night eating pattern if caloric consumption during 1900-0559 h was below or above the median of the population, respectively, for at least two trimesters. The results showed that women in the higher night eating group had lower energy intakes (P = 0.009) but gained more weight than recommended in the third trimester compared to the lower night eating group (P < 0.05) [30]. Another study from Singapore defined night eating as consuming >50% of total daily energy intake during 1900-0659 h in pregnant women, which was assessed using a 24 h dietary recall at 26-28 weeks gestation. After Two studies explored the associations of maternal night eating with GWG and postpartum weight retention (PPWR). A prospective cohort study from Brazil collected three 24 h dietary recalls in each trimester and pregnant women were classified as having a "lower" or "higher" night eating pattern if caloric consumption during 1900-0559 h was below or above the median of the population, respectively, for at least two trimesters. The results showed that women in the higher night eating group had lower energy intakes (p = 0.009) but gained more weight than recommended in the third trimester compared to the lower night eating group (p < 0.05) [30]. Another study from Singapore defined night eating as consuming >50% of total daily energy intake during 1900-0659 h in pregnant women, which was assessed using a 24 h dietary recall at 26-28 weeks gestation. After confounder adjustment, night eating at 26-28 weeks gestation was associated with higher odds of substantial PPWR ≥5 kg at 18 months (OR = 2.04, 95% CI: 1.06, 3.94) [31].
There were two studies that investigated metabolic consequences of night eating. The first prospective cohort study from Singapore divided pregnant women into overweight (≥23 kg/m 2 ) and lean (<23 kg/m 2 ) groups and then classified them as being predominantly daytime (pDT) or predominantly nighttime (pNT) eaters based on whether they consumed a greater percentage of calories from 0700 to 1859 h or from 1900 to 0659 h. The results showed that pNT eaters had lower energy intakes at 26-28 weeks gestation in the lean group (p = 0.019) but not in the overweight group. However, after adjusting for confounders, pNT feeding was associated with a higher fasting glucose level in the lean group (β = 0.16 mmol/L, 95%CI: 0.05, 0.26); no association was noted for postprandial plasma glucose outcome [32]. Another cross-sectional study from Turkey reported that a higher total score from the Night Eating Questionnaire was positively associated with hemoglobin A1C (HbA1c), insulin resistance, insulin, and high density lipoprotein (HDL) cholesterol levels at 28-38 weeks gestation (p < 0.05) [33]. These observational findings were supported by a recent randomized controlled trial involving 103 Israeli women with GDM who were assigned to either the chrononutritional and sleep hygiene intervention group (n = 33) or the control group (n = 70) at 25-29 weeks gestation [34]. The chrononutritional recommendations for the intervention group from 25-29 weeks gestation to delivery were as follows: (1) reduce carbohydrate intake by 10 to 15% during the evening interval (1800-0600); (2) eat breakfast up to 30 min from waking up and eat dinner within and up to 12 h from waking up; and (3) avoid eating 1.5 h before going to sleep. The trial showed a significant reduction of suboptimal glycemic control (defined by <80% of the plasma glucose values at target) in the intervention group compared to the control group (RR = 0.28, 95% CI: 0.18, 0.81). However, the intervention had no effect on birth weight and gestational age at birth [34].
Overall, eating at night during pregnancy seems to be associated with higher preterm birth risk, impaired glucose metabolism, poor sleep quality, higher GWG, and higher risk of PPWR. These adverse complications may be related to disrupted circadian rhythms by night eating [35].

Meal Frequency
Studies investigating meal frequency during pregnancy and its associations with maternal/offspring outcomes are summarized in Figure 4 and Table 3.
Three cohort studies investigated the relationship between maternal meal frequency and neonatal outcome. Salunkhe et al. assessed the meal frequency of 380 Indian pregnant women at 14-26 weeks gestation using a 24 h dietary recall and found that participants who consumed four or five meals a day had higher energy intakes than those having ≤3 meals per day (p < 0.001). In addition, participants consuming two and three meals a day had a higher risk of delivering low birth weight (RR = 4.1, 95% CI: 3.39, 4.65 for two meals and RR = 11.9, 95% CI: 6.8, 20.9 for three meals) and preterm birth (RR = 4.9, 95% CI: 2.79, 8.49 for two meals and RR = 3.1, CI: 1.8, 5.4 for three meals) infants as compared to participants having ≥4 meals per day [36]. Using principal component factor analysis, Englund-Ögge et al. derived several "meal frequency patterns", namely "snack pattern", "main meal pattern", and "dinner pattern" among 65,487 Swedish pregnant women; greater adherence to the "main meal pattern" (implying regular intakes of breakfast, lunch, and dinner) was associated with a reduced risk of preterm birth [hazard ratio (95% CI): 0.89 (0.80, 0.98) and 0.90 (0.81, 0.99) for the third and fourth quartiles of "main meal pattern" score, respectively] [37]. Ainscough et al. recruited only pregnant women with overweight and obesity from Ireland and identified three mutually exclusive meal frequency pattern categories ["main meal pattern" (3 main meals + 0-3 snacks), "large meal pattern" (≤2 main meals + <2 snacks), and "snack pattern" (3 main meals + >3 snacks or ≤2 main + ≥2 snacks)] based on preferences for consuming main meals or snacks. There were no differences in nutrient and energy intakes across the meal patterns at 16 and 28 weeks gestation. However, pregnant women with a large meal pattern at 28 weeks gestation were more likely to deliver macrosomic babies compared to those with main meal and snack patterns (p = 0.008 based on a chi-squared test) [38].
Two cohort studies explored the association of meals pattern during pregnancy with GWG and PPWR. Ainscough et al. showed that term GWG was lower in women with the aforementioned large meal pattern than those with the main meal or snack patterns at 16 weeks gestation [38]. Another cohort study from Singapore assessed the relationship between eating episodes at 26-28 weeks gestation and PPWR. No associations were observed between eating episodes and PPWR (weight at 18 months postpartum first antennal visit weight ≥5 kg) [31]. LGA birth A significant reduction in suboptimal glycemic control in the intervention compared to control group. The intervention had no effect on birth weight or gestational age at birth.

Meal Frequency
Studies investigating meal frequency during pregnancy and its associations with maternal/offspring outcomes are summarized in Figure 4 and Table 3.
Three cohort studies investigated the relationship between maternal meal frequency and neonatal outcome. Salunkhe et al. assessed the meal frequency of 380 Indian pregnant women at 14-26 weeks gestation using a 24 h dietary recall and found that participants who consumed four or five meals a day had higher energy intakes than those having ≤3 meals per day (P < 0.001). In addition, participants consuming two and three meals a day had a higher risk of delivering low birth weight (RR = 4.1, 95% CI: 3.39, 4.65 for two meals and RR = 11.9, 95% CI: 6.8, 20.9 for three meals) and preterm birth (RR = 4.9, 95% CI: 2.79, 8.49 for two meals and RR = 3.1, CI: 1.8, 5.4 for three meals) infants as compared to participants having ≥4 meals per day [36]. Using principal component factor analysis, Englund-Ögge et al. derived several "meal frequency patterns", namely "snack pattern", "main meal pattern", and "dinner pattern" among 65,487 Swedish pregnant women; greater adherence to the "main meal pattern" (implying regular intakes of breakfast, lunch, and dinner) was associated with a reduced risk of preterm birth [hazard ratio (95% CI): 0.89 (0.80, 0.98) and 0.90 (0.81, 0.99) for the third and fourth quartiles of "main meal pattern" score, respectively] [37]. Ainscough et al. recruited only pregnant women with overweight and obesity from Ireland and identified three mutually exclusive meal frequency pattern categories ["main meal pattern" (3 main meals + 0-3 snacks), "large meal pattern" (≤2 main meals + <2 snacks), and "snack pattern" (3 main meals + >3 snacks or ≤2 main + ≥2 snacks)] based on preferences for consuming main meals or snacks. There were no differences in nutrient and energy intakes across the meal patterns at 16 and 28 weeks gestation. However, pregnant women with a large meal pattern at 28 weeks gestation were more likely to deliver macrosomic babies compared to those with main meal and snack patterns (P = 0.008 based on a chi-squared test) [38].
Two cohort studies explored the association of meals pattern during pregnancy with GWG and PPWR. Ainscough et al. showed that term GWG was lower in women with the Two studies investigated the association between eating frequency and blood glucose regulation in pregnant women. One cross-sectional study included 1061 Singapore pregnant women and used 24 h dietary recalls to determine eating episodes. The results showed that women with more frequent eating episodes had greater total energy intakes (p < 0.001), and each additional daily eating episode was associated with 0.15 mmol/L higher 2 h glucose (95% CI: 0.03, 0.28) but not with fasting glucose [39]. Another randomized crossover trial with isocaloric meals provided to 10 women with GDM showed that increasing meal frequency to six meals a day for one day, as compared with three meals a day, could significantly affect glycemic excursions measured by continuous glucose monitoring. Specifically, higher meal frequency reduced the peak glucose level, standard deviation, coefficient variation, mean amplitude, and the largest amplitude of glycemic excursions (all p < 0.05), but no differences were observed for the mean and lowest glucose levels [40].
Five cross-sectional studies, all conducted in Ethiopia, found that eating fewer than three meals a day increased the risk of anemia in pregnant women by 1.9-3.9 times as compared with eating more than three meals a day [41][42][43][44][45].
In summary, studies on the associations of meal frequency during pregnancy with maternal and child outcomes provided mixed findings. Higher meal frequency may be associated with lower risk of low birth weight, preterm birth [36], anemia [41][42][43][44][45], and less glycemic variability [40], but it may have the potential to increase postprandial gestational glycemia [39]. More studies are needed to determine the optimal meal frequency for pregnant women.

Non-Ramadan (Night) Fasting Duration
In a cross-sectional analysis within a Singapore prospective cohort, women with longer night-fasting intervals had lower total energy intakes (p < 0.001), and each hourly increase in night fasting interval during 26-28 weeks gestation was associated with a 0.03 mmol/L lower fasting glucose (95% CI: −0.06, −0.01 mmol/L). However, no association was noted for 2 h postprandial blood glucose [39]. In the same cohort, there was no consistent association between higher night fasting duration during pregnancy and substantial PPWR (≥5 kg) at 18 months [31]. Given the dearth of research investigating health impacts of habitual fasting duration during pregnancy, more studies in this area are warranted.

Non-Ramadan (Night) Fasting Duration
In a cross-sectional analysis within a Singapore prospective cohort, women with longer night-fasting intervals had lower total energy intakes (P < 0.001), and each hourly increase in night fasting interval during 26-28 weeks gestation was associated with a 0.03 mmol/L lower fasting glucose (95% CI: −0.06, −0.01 mmol/L). However, no association was noted for 2 h postprandial blood glucose [39]. In the same cohort, there was no consistent association between higher night fasting duration during pregnancy and substantial PPWR (≥5 kg) at 18 months [31]. Given the dearth of research investigating health impacts of habitual fasting duration during pregnancy, more studies in this area are warranted.

Ramadan Fasting
Studies investigating Ramadan fasting during pregnancy and its associations with maternal/offspring outcomes are summarized in Supplementary Table S3.

Weight Gain
Four studies examined the association between Ramadan fasting and GWG throughout pregnancy. Two of the studies showed lower GWG in the Ramadan fasting group compared with the non-fasting group [46,47], but the other two studies reported no difference between the two groups [48,49]. In our meta-analysis, GWG was not significantly different (pooled mean difference (MD): 0.5 kg; 95% CI: −1.27, 0.26; I 2 = 0.00%) in the Ramadan fasting group compared with the non-fasting group (Figure 5a). In addition, there were four studies that investigated the association between Ramadan fasting and GWG during Ramadan. Two of the studies showed that the Ramadan fasting group had a lower mean GWG during Ramadan than the non-fasting group [19,50], but the other two studies

Ramadan Fasting
Studies investigating Ramadan fasting during pregnancy and its associations with maternal/offspring outcomes are summarized in Supplementary Table S3.

Maternal Outcomes: Quantitative Analyses Weight Gain
Four studies examined the association between Ramadan fasting and GWG throughout pregnancy. Two of the studies showed lower GWG in the Ramadan fasting group compared with the non-fasting group [46,47], but the other two studies reported no difference between the two groups [48,49]. In our meta-analysis, GWG was not significantly different (pooled mean difference (MD): 0.5 kg; 95% CI: −1.27, 0.26; I 2 = 0.00%) in the Ramadan fasting group compared with the non-fasting group (Figure 5a). In addition, there were four studies that investigated the association between Ramadan fasting and GWG during Ramadan. Two of the studies showed that the Ramadan fasting group had a lower mean GWG during Ramadan than the non-fasting group [19,50], but the other two studies showed no differences between the groups [51,52]. In our meta-analysis, GWG during Ramadan was significantly lower in the fasting group compared with the non-fasting group (pooled MD: 0.65 kg; 95% CI: −1.02, −0.28; I 2 = 29.5%) (Figure 5b).

Other Outcomes for Ramadan Studies: Qualitative Analyses Maternal Health Outcomes
We categorized other maternal health outcomes reported as hematological parameters/lipid profiles, mode of delivery, oxidative stress markers, and pregnancy complications (Table 6). Overall, findings on the associations between Ramadan fasting and these outcomes were mixed, with studies showing direct, inverse, and no associations. For example, Ramadan fasting has been associated with both higher [62] and lower risk [47] of GDM, yet other studies reported no association [64,65]. Similarly, risk of caesarean section has been reported to be both higher [62] and lower [50] for fasting mothers, yet other studies reported no association between Ramadan fasting and mode of delivery [47,48,58,59,61,64,70]. Table 6. Associations of Ramadan fasting with maternal and child outcomes (compared to no Ramadan fasting).

Childhood Outcomes
A cohort study involved 191 children aged 4 to 13 years (Table 6). Of these, mothers of 98 children fasted throughout Ramadan during pregnancy, whereas mothers of 93 children did not fast. The results showed that, compared with the non-fasting group, Ramadan fasting during pregnancy was not associated with childhood intelligence quotient (IQ), height, or weight [49]. Since there was only one study investigating the association of maternal Ramadan fasting with childhood outcomes, a definitive conclusion could not be derived.

Principal Findings
In this systematic review, the chrononutrition factors investigated in the original studies could be broadly classified into four categories: meal skipping, night eating, meal frequency, and (Ramadan) fasting. We found that meal skipping and night eating have a potential adverse influence on maternal and birth outcomes, including increased PPWR [26], GWG [30], risk of spontaneous preterm delivery [20,21], insomnia [28], impaired maternal nutritional status [22,23], and impaired glucose metabolism [25,32,33]. However, in the meal skipping literature, it is unclear whether these associations were causal, as most studies were cross-sectional in nature. Evidence on the associations of maternal meal frequency with maternal and offspring outcomes appeared mixed and inconsistent. In this meta-analysis, maternal fasting during Ramadan was associated with favorable shortterm maternal outcomes (lower GWG during Ramadan and fasting blood sugar) but not birth outcomes (birth weight, gestational age, and odds of low birth weight and preterm delivery). For other associations between Ramadan fasting and maternal and offspring outcomes that could not be pooled quantitatively, our qualitative review revealed no consistent associations.

Meal Skipping
According to the 2007-2016 National Health and Nutrition Examination Survey, people who skipped breakfast, lunch, or dinner had a lower diet quality and a higher energy intake in subsequent meals, yet the net effect of skipped meals on total daily energy intake was still negative [79]. However, Shiraishi et al. showed that skipping breakfast lowered protein intake but did not affect total energy intake [24]. Therefore, whether nutrient intake is a confounding factor for the effects of meal skipping on maternal and birth outcomes remains to be further explored.
Two studies found that meal skipping during pregnancy can cause the pituitary gland to release corticotropin-releasing hormone (CRH), which in turn stimulates the adrenal glands to secrete and release cortisol [80]. As maternal cortisol levels increase, placental CRH production also increases, which in turn aids labor by expanding the uterus and stimulating smooth muscle contractions. However, irregular eating patterns caused by loss of appetite may be a potential sign that labor is approaching rather than meal skipping leading to labor. Therefore, the causal relationship between meal skipping and spontaneous birth requires further investigation [20,21].

Night Eating
In human, endogenous melatonin levels start to increase approximately 2 h before natural sleep onset and peak approximately 5 h later [81]. Since eating at night can lead to a decrease in melatonin secretion [82], it can result in poor sleep quality [28,29]. During pregnancy, the peak of maternal melatonin levels gradually increase approximately four-fold from 24 weeks of gestation to term [83]. Melatonin protects fetal growth and development through its antioxidant [84] and homeostatic effects [85] on the placenta. In animal studies, melatonin has been shown to inhibit uterine contractions by inhibiting prostaglandin synthesis [86]. Therefore, reduced melatonin secretion due to disruption of circadian rhythms by night eating behavior [82] may increase the risk of preterm birth [27].
That night-eating is associated with higher GWG and an adverse metabolic profile during pregnancy is also plausible. In studies by Gontijo et al. and Loy et al., night eating resulted in lower total energy intake but higher GWG [30] and fasting glucose levels [32] compared to day eating. In an animal study, day-eating mice (which are nocturnal in contrast to human) gained more weight and adiposity compared with night-eating mice [87]. The day-eating mice also developed hyperinsulinemia, hypercorticosteronemia, and hepatic lipid accumulation with increased lipogenic gene expression. Thus, eating at a time that is not aligned with one's natural circadian rhythm may result in higher risk of obesity and metabolic disorders.

Meal Frequency
Globally, about 32.4 million pregnant women are anemic, and the rates are highest in low-income countries, especially those in Southeast Asia and Africa [88]. In our qualitative analysis, five studies from Ethiopia [41][42][43][44][45] and one study from India showed that lower meal frequency in pregnant women was associated with increased risk of anemia and preterm birth, respectively. However, since only one study showed that women with lower meal frequency had lower energy intake [36], it is unclear how lower meal frequency can cause anemia or preterm birth. It is also uncertain whether lower meal frequency is an indicator of meal skipping due to food insecurity, especially when the studies were conducted in developing countries.
A higher number of daily eating episodes was associated with higher 2 h glycemia in a cross-sectional analysis among pregnant women in a cohort study [39]. In the general population, a lower meal frequency (1-2 times compared with 3 times a day) in men was associated with higher risk of type 2 diabetes [89], but another cohort study reported no associations in women [90]. Thus, whether meal frequency is linked to blood glucose level and diabetes in the healthy population is still inconclusive. However, among patients with GDM, six meals a day led to better glycemic control than three meals a day in an isocaloric trial [40]. These data are similar to another study among type 1 diabetes patients, which observed that the number of meals was inversely associated with HbA1c and mean blood glucose measurements [91].

Ramadan Fasting
Sick, pregnant, and menstruating women are exempted from Ramadan fasting, yet 70-90% of pregnant women prefer to fast [16], of which 40-55% fast for the entire Ramadan period [92,93]. In our meta-analysis, compared with the non-fasting group, Ramadan fasting was associated with favorable short-term maternal outcomes (lower GWG during Ramadan, lower fasting glycemia) but not birth weight and gestational age. These health implications may be short-lived and reversed after the end of the Ramadan month, due to festive celebration after the fasting month [94]. For example, Fernando et al. showed that mean weight loss with Ramadan fasting was 1.34 kg, but most of the weight was regained a few weeks post-Ramadan [95].
While previous research and our systematic review suggest that eating at night can disrupt circadian rhythms and be harmful to offspring [27], the analysis of Ramadan fasting (which involves eating during nighttime and a prolonged fasting interval among other things) did not find similar results. The consequences of Ramadan fasting may be affected by fasting days, fasting period per day, and dietary quality and quantity during Ramadan. Ramadan fasting is a special circumstance that lasts for only one month, and its influence may thus be lesser compared to habitual dietary behavior such as non-Ramadan night eating. Of note, some studies have shown that the overall dietary quality was improved during the Ramadan month [96]. Thus, it may also be possible that a prolonged daily fasting period and improved dietary quality attenuated any potential adverse influence of night eating. However, since objective measurement of fasting is difficult, it is unclear whether fasting behavior is consistent across populations and if Muslim pregnant women have different levels of fasting exposure or other factors associated with health outcomes. Many studies investigating Ramadan fasting did not adjust for potential confounding factors, further complicating evidence consolidation.

Strengths, Limitations, and Future Perspectives
In this systematic review, we identified potential studies from two databases using comprehensive search terms. We also included studies investigating different dimensions of chrononutrition during pregnancy, including meal skipping, meal frequency, night eating, and (Ramadan) fasting. Thus, to our knowledge, this is by far the most comprehensive review on the study of maternal chrononutrition and its associations with both maternal and offspring outcomes. However, we are hindered by many limitations present in the original studies. For example, most of the studies were cross-sectional, meaning temporal relationships cannot be easily established due to the possibility of reverse causality. Many studies also did not provide clear definitions of chrononutrition factors, such as the number of meals skipped, fasting days, or fasting period per day, which impedes the formation of strong conclusions. Moreover, the heterogenous study designs as well as the variable and scattered exposures and outcomes investigated also hindered quantitative meta-analysis for most associations of interest.
In order to further elucidate the effect of chrononutrition factors on maternal and offspring health outcomes, larger and well-conducted cohort and interventional studies are needed. Information on covariates such as diet quality and quantity, fasting days, fasting period per day, and trimester exposure should also be collected and considered during analysis. In addition to diet, exercise [97] and sleep [98] can regulate the circadian rhythm, so the interaction between diet, sleep, exercise, and circadian rhythm can be further explored in the future. Since chrononutrition factors are interrelated, more advanced statistical techniques, for example those that consider these factors holistically as "temporal dietary patterns" [99], may help to generate further insights.

Conclusions
In this systematic review, we showed that meal skipping and night eating during pregnancy was generally related to adverse maternal and birth outcomes, whereas strong conclusions cannot be reached for meal frequency and non-Ramadan daily fasting. Ramadan fasting did not seem to affect birth weight or gestational age at birth, but evidence for other maternal and offspring outcomes was inconsistent. Most of these results are not supported by strong evidence, indicating pressing needs for further research.