Influence of physical activity before and during pregnancy on infant’s sleep and neurodevelopment at 1-year-old

The aim of this study was to investigate the association between maternal physical activity (PA) before and during pregnancy and sleep and developmental problems in 1-year-old infants. We used data from a nationwide cohort study in Japan that registered 103,062 pregnancies between 2011 and 2014. Participants were asked about their PA before and during pregnancy, and the sleep and development of their children at the age of 1 year. Maternal PA was estimated using the International Physical Activity Questionnaire and was expressed in METs per week. We defined scores below the cut-off points of the Ages and Stages Questionnaire (ASQ) as abnormal for infant development. Based on the levels of PA before or during pregnancy, the participants were divided into five groups. In mothers with higher PA levels, the risk ratio for bedtime after 22:00 or abnormal ASQ scores in their 1-years-old infants were lower. These associations were observed for PA before and during pregnancy. Higher levels of maternal PA, both before and during pregnancy, may reduce sleep and developmental problems in infants.

Association of maternal PA before and during pregnancy with sleep problems in 1-year-old infants. Low levels of maternal PA before and during pregnancy were associated with an increased risk ratio for bedtime after 22:00 but were not associated with other sleep outcomes (Tables 2 and 3). The risk ratio of bedtime after 22:00 in the group with the highest levels of PA (Quartile 4) both before and during pregnancy was lower than the reference group (PA before pregnancy, RR = 0.91, 95% CI = 0.87-0.95; PA during pregnancy, RR = 0.88, 95% CI = 0.84-0.92).
Association of maternal PA before and during pregnancy with development in 1-year-old infants. Low levels of maternal PA both before and during pregnancy were associated with an increased risk of abnormal ASQ scores. Compared to the reference PA group before pregnancy, the group with the lowest levels of PA (PA = 0) had higher risk ratios of abnormal scores in the following domains of the ASQ: fine motor skills (RR = 1.15, 95% CI = 1.05-1.26), problem-solving (RR = 1.20, 95% CI = 1.09-1.33), and personal-social skills (RR = 1.29, 95% CI = 1.04-1.60) ( Table 2).
In other domains of the ASQ, including communication and gross motor skills, maternal PA both before and during pregnancy was not associated with significant risk ratios of abnormal scores.
The association between the risk ratios of abnormal ASQ scores and maternal PA levels, both before and during pregnancy, did not change in the subgroup analysis that excluded women with HDP and GDM (Table S2).

Discussion
In the present study, lower maternal PA levels before and during pregnancy increased the risk ratios of abnormal scores on the infants' ASQ at 1 year of age. Higher maternal PA levels were associated with lower risk ratios of abnormal ASQ scores. Similarly, maternal PA levels before and during pregnancy were inversely associated with infant late bedtime at or after 22:00, as shown by the risk ratio. However, other sleep outcomes were not associated with maternal PA levels. This is the first study to show that maternal PA levels before pregnancy influence measures of development and concur with previous findings associated with PA during pregnancy. Higher PA levels during preconception may decrease the risk of developmental delay.
Regarding infant sleep problems, lower maternal PA levels before and during pregnancy were associated exclusively with late bedtime in 1-year-old infants. In this study, the proportion of infants who fell asleep after 22:00 was approximately 20%, which is a larger percentage than other sleep outcomes. Thus, a small but significant difference in late bedtime was detected. A high level of maternal activity during pregnancy has been reported to improve maternal sleep 15,16 . The sleep cycle develops from the fetal period 17 , and the association between maternal sleep during pregnancy and infants' sleep has also been reported 18,19 . Therefore, maternal PA during pregnancy may affect infants' sleep through maternal sleep. No other studies have addressed the direct association between maternal PA and infant sleep patterns. Further investigations are required to evaluate these associations. This is the first study to show that maternal PA before pregnancy may influence infant developmental outcomes. Regarding PA during pregnancy, there have been many previous studies that reported an association between PA during pregnancy and language development [20][21][22] . The association between PA during pregnancy, motor function, and social skills remains inconclusive 5,6 . A recent RCT study reported a positive association between maternal exercise and infant neuromotor outcomes at 1-month-old 23  www.nature.com/scientificreports/ association between maternal PA and child development, it was also reported that the association became insignificant as the child matured 21,24 . Future studies should investigate the association between maternal PA before and during pregnancy and development in older children.
There are several hypotheses on how maternal PA before and during pregnancy affects infant neurodevelopment. The first hypothesis involves maternal inflammation, which affects fetal neurodevelopment in utero and may cause developmental disorders 25 . One study reported that exercise intervention in pregnant women reduced inflammatory cytokines 26 . Therefore, high maternal PA levels may protect fetal neurodevelopment from inflammation.
The second hypothesis is that maternal PA directly affects neurodevelopment in infants. In an experiment with mice and rats, exercise during pregnancy improved neurogenesis in the hippocampus, memory, and learning outcomes [27][28][29][30][31] . From this study, it can be inferred that for humans, PA during pregnancy may have a beneficial influence on fetal neurodevelopment.
The third hypothesis is that maternal activity may stimulate fetal sensory systems, such as vestibular function. A study of preterm infants reported that auditory, tactile, visual, and vestibular interventions increased nipple feeding and decreased the length of infant hospitalization 32 . As fetal vestibular function develops from early pregnancy 33,34 , maternal PA may stimulate the fetal vestibular system to positively affect neurodevelopment.
Perinatal complications, such as GDM, HDP, and perinatal depression, have negative effects on child neurodevelopment 1,2,35,36 . These complications are known risk factors of developmental disorders 11 . However, in our subgroup analysis, which excluded cases of HDP and GDM, an association between abnormalities in ASQ scores and maternal PA before and during pregnancy was found. This finding implies that HDP and GDM may not be the only complications associated with maternal PA before and during pregnancy.
This study has several limitations. First, this was an observational study, so there could be unmeasured confounding factors, such as parental life rhythm or sleep cycle. Second, maternal PA and infant outcomes (infant sleep problems and ASQ scores) were evaluated using a self-reported questionnaire, so there could be some bias. In particular, maternal PA before pregnancy was reported at recruitment in the first trimester of pregnancy. On the other hand, the strength of the present study was that it was based on national data. Additionally, this was the first study to focus on the association between maternal PA before pregnancy and infant development.
In conclusion, lower maternal PA before or during pregnancy was associated with negative effects on infant development and increased risk of late bedtimes in 1-year-old infants. In contrast, higher maternal PA before or during pregnancy may have positive effects on infant development and decrease the risk of late bedtimes in 1-year-old infants. www.nature.com/scientificreports/ Study participants. The data used in this study were obtained from the JECS, an ongoing large-scale cohort study. The JECS was designed to follow children from the prenatal period to the age of 13 years. The detailed protocol of the study and the baseline profile of participants in the JECS have been previously reported previously 37,38 . The participants answered a questionnaire about lifestyle and behavior twice during pregnancy. The questionnaire completed at recruitment was referred to as M-T1, and the questionnaire completed later during mid-and late pregnancy was M-T2. The mean gestational weeks (SD) at the time of responding to M-T1 and M-T2 were 16.4 (8.0) and 27.9 (6.5) weeks, respectively. Participants also answered a questionnaire about their offspring one year after delivery (C-1y). Between 2011 and 2014, 103,062 pregnant women were recruited from 15 regions throughout Japan (Fig. 1). Of these, we excluded 26,694 pregnancies due to the following reasons: previous participation in the study (n = 5647), multiple fetuses (n = 949), miscarriage or stillbirth (n = 3,676), congenital anomaly or disease at 1 month of age (n = 3553), missing information on maternal age at delivery (n = 7), delivery before 37 weeks or after 42 weeks of gestation (n = 4184), lack of information about maternal PA in the M-T1 and M-T2 (n = 1109), and no response to questions about children's sleep and development at C-1y (n = 7569). The remaining 76,368 participants (74,971 with M-T1 data and 72,700 with M-T2 data) were included in the analysis.

Methods
Exposure: maternal PA. We used the Japanese short version of the International Physical Activity Questionnaire (IPAQ) to evaluate maternal PA, for which test-retest reliability and criterion validity were reported elsewhere 39,40 . Participants reported their mean PA per week before pregnancy on the M-T1 questionnaire based on recall and their mean PA per week during pregnancy on the M-T2 questionnaire. We calculated PA in terms of metabolic equivalent of a task (MET), measured as the number of minutes per week (METs-min/week) 39 . PA, as defined in the IPAQ, includes all activities of daily life, such as work, housework, and leisure activities.
We divided the participants into five groups based on their level of PA before pregnancy. We also divided the participants into five groups based on their levels of PA during pregnancy. In each of the five groups, the "PA = 0" group consisted of participants whose PA was 0. The other participants were divided into four groups using PA quartile points. The groups were labeled Quartiles 1-4 in ascending order of PA. Quartile 1 referred to the group with the lowest PA levels among the four groups. Quartile 4 referred to the group with the highest PA levels. To visualize the effects when the amount of activity was very low, we defined the Quartile 1 groups as the reference groups for the purpose of statistical analysis instead of the PA = 0 groups.

Outcome 1: infant sleeping problems.
One year after delivery, information on infant sleep habits was collected via parent-reported questionnaires (C-1y). The participants answered questions regarding their infant's sleep time in the previous 24 h, in 30-min increments. They were also asked whether their children cried at night, and if so, the crying frequency ("rarely, " "1-3 times in a month, " "1-2 times in a week, " "3-4 times in a week, " "5 times in a week or more") was reported. In this analysis, we focused on five points. First, we determined the number of nocturnal awakenings from maternal responses to infants' sleeping periods. We defined ≥ 3 awakenings as too many because a previous study reported that the upper limit of the number of awakenings during the night was 2.5 for 1-year-old infants 41 . Second, we determined whether the infants awoke more than once and whether they stayed awake for more than 1 h during the night. If so, these were defined as unusual. Third, we analyzed the duration of nighttime sleep (20:00-07:59). We regarded less than 8 h of sleep as too short because past research reported that the mean duration of sleep for this age group was 8.3 h 41 . Fourth, we determined the www.nature.com/scientificreports/ infants' bedtime. In this study, about 65% of 1-year-old infants slept later than 21:00, and about 20% slept later than 22:00. Therefore, we defined bedtime after 22:00 as too late. Fifth, we obtained information about crying at night in the past month. If the mother answered that her infant cried during the night, and the frequency of crying at night was more than five times per week, we defined the case as "crying at night".
Outcome 2: infant development. We used the Japanese version of the Ages and Stages Questionnaire (ASQ), third edition, to evaluate infant development. The C-1y questionnaire included ASQ. ASQ captures developmental delay in five domains: communication, gross motor skills, fine motor skills, problem-solving, and personal-social characteristics. The answer to each question is one of the following: "yes, " "sometimes, " or "not yet. " The scores were 10, 5, and 0 points, respectively. Each ASQ domain was composed of six questions, and the total score ranged from 0 to 60. The cut-off point for each domain in the Japanese version was 2SD below the mean, and all the cut-off points were determined by age groups in a previous study 42   www.nature.com/scientificreports/ Covariates. Information on maternal age at delivery, pre-pregnancy body mass index (BMI), parity, gestational age at birth, infertility treatment, type of delivery, current history of hypertensive disorders of pregnancy and diabetes or gestational diabetes, infant birth weight, and infant sex were collected from medical records. Information about smoking habits, alcohol consumption, educational background, household income, maternal psychological distress at 1 year after delivery, doctor diagnosis of asthma and atopic dermatitis in children up to 1 year of age, and feeding status were collected via self-administered questionnaires. Maternal depression has been reported to affect infant development 43 . In the present study, we did not know whether the participants had a mental illness after delivery. Thus, maternal psychological distress was assessed using the Kessler 6 44,45 questionnaire at C-1y. In concordance with previous studies, participants with a score of five or more were categorized as having distress 46 .
Statistical analyses. We used a log-binominal regression model to explore the association of maternal PA with each outcome and to estimate the risk ratio (RR) of each outcome and the 95% confidence intervals (CIs). We initially adjusted for maternal age at delivery and then further adjusted for smoking habits (never smokers, ex-smokers who quit before pregnancy, smokers during early pregnancy), alcohol consumption (never drinkers, ex-drinkers who quit before pregnancy, drinkers during early pregnancy), pre-pregnancy BMI (< 18.5, 18.5-24.9, ≥ 25.0 kg/m 2 ), parity (0, ≥ 1), infertility treatment (no ovulation stimulation/artificial insemination by sperm from husband, assisted reproductive technology), type of delivery (vaginal or cesarean section), gestational age at birth (37-38, 39-41 weeks), infant sex (boys, girls), psychological distress at 1 year after delivery (yes, no), doctor diagnosis of asthma and atopic dermatitis at 1 year of age, and feeding (breast milk, formula, both). The covariates to be added to the multivariate model were determined by referring to the previous literature as potential risk factors for developmental disorders 11,47 . Due to the large sample size in this study, we used the risk factors contained in the dataset in the multivariate model as covariates whenever possible. However, since there were many missing data on household income and educational background, we excluded them from the covariates of the multivariate model after confirming that the results did not change significantly even if they were included in the model. We did not complete the missing data. Thus, the multivariate analysis was limited to those participants that had all the covariate data. We also performed a subgroup analysis excluding women with HDP and GDM to investigate the influence of these factors on infant development.
In this study, we used a fixed dataset "jecs-an-20180131, " which was released in March 2018. Stata version 15 (StataCorp LP, College Station, TX, USA) was used for all statistical analyses.
The statistical analyses of this study were conducted in a manner that was similar to that of our previous study 48 .