Dietary intake and thereby the risk of chronic diseases and excessive weight gain can be altered by many lifestyle behaviors. Evidence is accumulating that sleep is one of these factors [1
]. Shorter sleep, in particular, may affect dietary intake through a number of possible mechanisms [3
]; shorter sleep gives more time and opportunities for eating, it may change the timing of eating, and may induce hedonic feeding. Also, it has been suggested that short sleep alters the release of the appetite-related hormones. Studies have observed lower leptin and higher ghrelin concentrations, which are associated with increased dietary intake, in short sleepers [4
Most research on sleep and dietary intake has focused on macronutrients or single nutrient intakes [4
], whereas examining the diet holistically has been suggested to better reflect the complexity of dietary intake [5
]. The two most common methods to study overall diet are diet quality indices that are based on pre-existing knowledge of diet-disease associations (e.g., dietary recommendations) or data-reductions methods (e.g., principal components analyses) [6
]. To date, evidence from trials in adults and children suggests that sleep restriction increases food and total energy intakes [1
]. Moreover, short sleep duration (<6–7 h) as compared to adequate sleep (7–8 h) was associated with poorer diet quality as assessed with diet quality scores in Canadian children [7
], European adolescents [8
], and Iranian adult women [9
]. According to another review, mainly including cross-sectional studies, diets that were high in fat were associated with more sleep disorders, whereas Mediterranean-style diets and diets high in carbohydrates were associated with fewer insomnia symptoms [10
Beyond dietary intake itself, sleep might also alter eating behaviors possibly leading to concurrent changes in dietary intake and the risk of chronic disease. It has been reported that sleep deprivation or short sleep duration (<6 h) increases portion size, the likelihood of main meal skipping, and was associated with greater energy intake from snacks and longer eating periods [11
Sleep quality and sleep duration are often compromised in pregnant women and deteriorate over the course of pregnancy [15
], mainly due to physical discomfort and pain [17
]. Sleep behavior may, therefore, be an important risk factor of unhealthy dietary practices during pregnancy and creates a window for development of public health education programs. We are aware of only one study that investigated sleep quality and dietary intake among pregnant women. Chang et al. showed positive significant associations between night time sleep disturbance and dietary fat intake and between shorter time to fall asleep, and higher fruit and vegetables intake in overweight and obese American women during pregnancy [18
]. However, diet quality or dietary patterns were not studied. Thus, it is yet unknown whether sleep is associated with dietary patterns and eating behaviors during pregnancy. Hence, the aim of this study is to understand the association between dietary patterns and eating behaviors in apparently healthy pregnant women using the Growing Up in Singapore Towards Healthy Outcomes (GUSTO) cohort.
Our results showed that good sleep quality was associated with better diet quality, as denoted by greater HEI-SGP scores and greater adherence to the VFR pattern, although these associations attenuated when taking anxiety scores into account. After additional adjustment for anxiety, good quality sleepers showed to be associated with lesser adherence to the SFN pattern in this cohort of pregnant women. No association was observed between sleep duration and dietary intake of dietary behaviors.
Only a few other studies examined the relationship of sleep quality with diet. Similarly to our positive findings of sleep quality with diet quality and the VFR pattern, fewer insomnia symptoms have been reported with greater adherence to the Mediterranean diet [32
]. The Mediterranean diet is characterized by high intakes of vegetables, legumes, fruits and cereals, fish, a moderate intake of alcohol, and low intakes of dairy, and saturated fatty acids, and is considered a healthy diet [33
]. Moreover, a study investigating single food groups found that individuals with good sleep quality, defined as PSQI-Japan score ≤3, had significantly higher total vegetable and rice intakes in middle-aged Japanese females [34
]. Another study in Japanese men and women further confirmed that a high rice intake was significantly associated with a 46% lower risk of poor sleep quality [35
]. Suggested explanations were the high glycemic load of rice and the high content of melatonin of rice, which may favor good sleep [35
]. In contrast to our findings, Chang et al. reported no associations between sleep quality using the PSQI and fat, and fruit and vegetable intakes among American overweight and obese pregnant women [18
]. However, this study only examined women who were already overweight or obese and did not study overall diet or eating behaviors. Also, Cheng et al. [36
] and Stern et al. [37
] observed no associations between diet quality assessed by the ‘Alternate-HEI’ score and probable insomnia among American males or the women health initiative insomnia rating scale among postmenopausal women, respectively.
We observed an inverse association between the SFN pattern and sleep quality. Previously, we showed that participants who adhered highly to the SFN pattern had higher intakes of energy, protein, and fat, and lower intakes of carbohydrates, and dietary fiber [24
]; nutrients that have been suggested to be inversely associated with sleep [4
]. Additionally, a study in Japanese participants found that the food group noodles was, independently from other carbohydrate sources as rice and bread, associated with poorer sleep quality [35
While differences in diet quality scores between good and poor sleepers may appear moderate, there is evidence to demonstrate that the associations between diet quality and all-cause and cause-specific mortality are dose-response relationships [38
], suggesting that any improvement in diet quality will result in some health improvements. Furthermore, even moderate changes in diet quality have been shown to decrease the risk of death meaningfully [39
]. This evidence from other diet quality indices suggests that our rather small mean differences in diet between good/poor sleepers may be of clinical relevance.
We showed no association between sleep duration and dietary intake, which corroborates with two studies among non-pregnant women and men, [13
], but not with other studies [8
]. This discrepancy might be explained by the differences in methodology of the various studies, for example in definitions of short sleep duration (varying between <5 and <8 h), of diet quality indices (e.g., Alternate-HEI, HEI, Diet Quality Index for Adolescents with Meal index), or a different set of confounders included in the statistical models. Another explanation might be the difference in the cause of sleep problems in our population when compared to non-pregnant study samples [17
Evidence for the relationship between sleep and eating behaviors is very limited, although it is important as it can provide important context for future interventions. We observed no association between sleep quality or sleep duration with any of the dietary behaviors. In contrast, Kant et al. found that calories consumed from all snacks decreased with increasing sleep duration in an adult American population sample [12
], however, sleep quality was not studied. Similar to our findings, they did not find an association between the number of eating episodes and sleep duration. More studies are needed to further elucidate the relationship between sleep and eating behaviors, also looking into sleep quality.
Mental well-being is linked to poor sleep and has been associated with dietary intake [43
], and might thus modify or confound the relation between sleep and diet. For this reason, we excluded the women who had probable major depression (EPDS scores ≥ 15; n
= 45) and we additionally adjusted for STAI-state scores in model 2. Hence, our results showed that even for women without major mental health issues, there is a relation between sleep quality and dietary intake.
Both sleep and diet are paramount for overall health and mental well-being [4
], and this might be even more important in pregnant women. Sleep and diet during pregnancy can affect gestational weight gain, risk of delivery complications, and has an impact on the offspring’s health in later life [44
]. On top of this, it is well-known that sleep quality and sleep duration are often compromised in pregnant women and this deteriorates over the course of pregnancy [15
]. It is therefore crucial to make pregnant women aware of these risks and to improve their sleep practices. Health practitioners who are involved in antenatal care should be alert for sleep problems during pregnancy and educate women on healthy sleep hygiene practices and possible behavioral changes or therapies [47
The strengths of this study are the inclusion of a relatively large number of apparently healthy pregnant women, and the use of dietary patterns that enabled the examination of synergistic effects of nutrients. Furthermore, we assessed sleep quality that includes both quantitative aspects of sleep, such as sleep duration and qualitative subjective aspects of sleep, whereas most studies only investigate sleep duration. The PSQI questionnaire, moreover, enabled us to differentiate between good and poor sleepers.
However, we also need to acknowledge the limitations of our study. Sleep and dietary intake were assessed at a similar point in time, through which no conclusions can be drawn on the causality of the association. Trials have shown associations between sleep and diet in both directions (sleep affecting dietary intake and vice versa) [1
], which suggest a bi-directional relationship. Secondly, sleep was assessed by self-report on a subjective scale, which may induce misreporting and thereby attenuation of our results. Objective measurements, such as polysomnography or multiple actigraphs, could have strengthened our associations. Nevertheless, the PSQI was previously validated against objective sleep measures [27
]. Furthermore, we excluded quite a large number of participants who did not complete the sleep questionnaire (n
= 451) that could have led to selection bias. In general, the participants who completed the sleep questionnaire had higher household incomes, education levels, were more often employed, married, and adopted healthier lifestyles as compared to those who had incomplete sleep questionnaire (results not shown). Thirdly, dietary intake was assessed with only one 24-h recall and is thus not representing usual intake. Two or more 24-h recalls are recommended to capture some of the variability in dietary intake depending on the dietary habits of the population under study [48
]. Though, we think that our 24-h recall adequately represented the typical diet of our participants, as showed by the moderate correlations between the emerged dietary patterns from 24-h recall and those based on three-day food diaries that were administered in a subsample of the GUSTO cohort (n
= 255; Pearson’s correlation rVFR
= 0.48, rSFN