Worldwide, more than 41 million children under 5 years of age are affected by childhood obesity and being overweight, which poses one of the major challenges of the 21st century [1
]. Research suggests that babies born with specific anthropometric characteristics are more susceptible to childhood obesity and to an increased risk of obesity later in life [2
]. For example, neonatal outcomes such as high birth weight (>4000 g) and being born large for gestational age (LGA; >90th percentile for gestational age) were shown to be early markers for an increased obesity risk in infancy [3
A healthy lifestyle during pregnancy is discussed to enhance the maternal health status, positively impact fetal development, and improve neonatal as well as obstetric outcomes [5
]. Within the last few years, approaches focusing on the prenatal lifestyle of the mother-to-be have been initiated to avoid an early obesogenic environment in utero. Lifestyle interventions frequently target a modification of the maternal diet and/or physical activity (PA) behavior [7
]. However, the role of PA in the prevention of adverse neonatal and obstetric complications is controversially discussed. While some reported PA to lower the risk for LGA and high birth weight [8
], others suggested positive associations between PA and birth weight dependent on the PA intensity [10
]. Moreover, the effect of prenatal PA on the risk for preterm and caesarean delivery is highly debated, as some analyses showed beneficial associations between prenatal PA and the risk of preterm birth and cesarean section, while others did not demonstrate any effects of prenatal PA on gestational age at birth or mode of delivery [11
]. Thus, evidence is still inconclusive. The activity mode, time point during pregnancy, and intensity seem to be particularly important in this heterogeneous picture on the effect of PA on neonatal and obstetric parameters. Nevertheless, no negative effects of prenatal PA on obstetric or neonatal outcomes have been found so far.
In order to encourage expectant mothers to engage in adequate levels and intensity of PA, the American College of Obstetricians and Gynecologists (ACOG) established recommendations in the report “Physical Activity and Exercise During Pregnancy and the Postpartum Period” [16
]. These recommendations have been adapted for Germany [17
]. The key message of these recommendations is that pregnant women without contraindications should not stop exercising during the course of pregnancy. In the absence of medical or obstetric complications, pregnant women are recommended to engage in moderate-intensity PA for at least 20–30 min per day on most or all days of the week [16
In Bavaria, Germany, the large-scale “Gesund leben in der Schwangerschaft”/ “Healthy living in pregnancy” (GeliS) trial was initiated offering a lifestyle intervention program that is linked to routine prenatal care [18
]. The GeliS trial sought to improve maternal and neonatal health outcomes as well as obstetric parameters in order to positively influence the long-term health development of both mothers and infants. The primary objective of the GeliS trial was to reduce the proportion of women with excessive gestational weight gain. The GeliS intervention was not successful in reducing the incidence of excessive gestational weight gain [19
]. However, it led to some improvements in maternal antenatal dietary [20
], and PA behavior [21
]. Moreover, some minor but likely clinically irrelevant benefits in maternal postpartum weight development and breastfeeding behavior were observed [22
The present work aims to investigate associations between early and late prenatal PA behavior and neonatal and obstetric parameters in the entire GeliS cohort. Moreover, we sought to describe differences in neonatal and obstetric outcomes between women meeting or not meeting the PA recommendations given by the ACOG.
In this secondary analysis of the GeliS cohort, we were able to show associations between early and late prenatal PA and infant anthropometrics as well as several neonatal and obstetric outcomes. Moreover, we could comprehensively investigate the influence of differences in the PA behavior between women who met and did not meet the prenatal PA recommendations on these outcomes. Lastly, we could associate different PA intensities with infant anthropometrics and the risk of adverse neonatal and obstetric outcomes.
Infants of women meeting the PA recommendations in late pregnancy were born significantly heavier and tended to be larger than offspring of inactive women. However, the estimated differences between groups were small. We observed that late sedentary PA was inversely associated with birth weight and positively associated with the risk of low birth weight but not for SGA. Engaging in light-intensity activities in early pregnancy was associated with a decreasing risk of the offspring to be born SGA. Our findings on infant birth weight correspond to the results of Koushkie Jahromi et al. [31
] and Badon et al. [32
] who, likewise, reported that babies born to women who exercised during pregnancy were heavier than those born to non-exercising women [31
] with no impact of early sedentary behavior on birth weight [32
]. However, our results contrast the observations of Bisson et al. [33
] who also used the PPAQ to explore effects of PA on infant birth weight in a Canadian cohort. Although study samples seemed to be comparable in terms of BMI, educational level, and parity, these authors observed a reduction in infant birth weight by 2.5 g with each increase of 1 MET-h/week in the level of sports and exercise [33
]. A possible explanation for the discrepancy with our results might be that the authors included more covariates in their model such as maternal education and smoking status, paternal weight, and infant sex, among others.
In general, the opinion about the effect of prenatal PA on infant birth weight is conflicting. Some studies report no significant association [14
], while others report a positive [31
] or a negative association [38
]. The considerable heterogeneity in the assessment of PA data complicates the comparison of study results. Furthermore, inconsistency of results may arise from differences in type, frequency, timing, and duration of PA [40
]. There is some evidence suggesting an inverted U-shaped relationship between PA intensity and birth weight [10
]. Although this hypothesis has been criticized [9
], it provides a possible explanation for our observations and seems to explain why we found late sedentary-intensity to be inversely associated with birth weight and positively associated with the risk for a low birth weight, while moderate-intensity PA in early and late pregnancy appeared to be linked to a slight increase in the odds of high birth weight. Bisson et al. [10
] explain this relationship by alterations in glucose availability and uteroplacental blood flow in response to exercise [41
], which seems to be reduced by exercise intensity [42
] and compensated afterward, resulting in enhanced fetal growth. A comprehensive estimation of the influence of PA on neonatal body composition might elucidate the debate on its effect on birth weight [35
In our cohort, infants of active women at T0 and T1 tended to be at higher risk for being born LGA. This observation was shown to be significant in women meeting the recommendations in early as well as in late pregnancy. However, it is difficult to explain this finding since we could not identify any PA intensity during the course of pregnancy, which was associated with an increasing risk of LGA. Current research has either found no influence of PA on the incidence of LGA [12
], or observed PA to be protective against LGA [8
], which calls into question whether our observations were chance findings.
Our results showed that women who were active in late pregnancy were less likely to give birth prematurely, which was similarly observed by others [45
]. Likewise, we identified sedentary PA in early and late pregnancy to be, by trend, associated with an increase in the odds of preterm delivery, even though statistical evidence was lacking. These observations are in line with some investigations showing either no effect of PA on the risk for preterm birth [12
], or a protective influence [47
]. However, we observed that vigorous-intensity PA in late pregnancy appeared to be linked to an increasing risk for preterm birth. This is in contrast with observations from a prospective study [48
], a systematic review [49
], and a meta-analysis [11
], which showed that higher leisure-time and vigorous PA did not change the incidence of, or reduce the risk for preterm deliveries. Nevertheless, the impact of prenatal PA on pregnancy duration was studied in women with different weight classifications, which might explain inconsistencies with our results since not all trials included women with normal and overweight outcomes simultaneously.
In the present analyses, we found no difference in the incidence of caesarean delivery between active and inactive women during the course of pregnancy. This corresponds to other current investigations [13
] even though some studies observed a lower caesarean section rate in exercising women [8
]. However, we observed that TALIA and light-intensity PA in early pregnancy were associated with a marginal increase in the odds of caesarean delivery. Poyatos-Léon et al. [15
], likewise, assessed the association between PA intensities and the risk for caesarean section deliveries and provided a possible extension of our findings. They concluded that, in general, exercise during pregnancy appears to decrease the risk of caesarean delivery. In particular, women who engaged in exercise during the second and third trimester seemed to increase their likelihood of a normal delivery [15
This secondary analysis of the GeliS trial has some limitations. As noted by others [40
], we did not include dietary intake as a covariate. Previously, we reported small effects of the maternal diet on neonatal weight-related parameters [51
], and are aware that maternal dietary behavior might have biased the present results. Further potential confounders such as the maternal employment, living circumstances, or other lifestyle factors (smoking, drinking) might have influenced the maternal PA level, but were not considered in the present analyses. Moreover, we used a self-administered PPAQ, which, although a being validated and easily applicable tool, was filled out by participants at both time points without supervision or accompanying interview. Self-reports rely on the subjective estimation of participants who had to remember their PA level and type of performed sports for the past four weeks. As reported by others [52
], the presented self-reported data might have been susceptible to over-reporting and under-reporting. Despite our efforts to exclude over-reporting, this mode of PPAQ self-administration might have biased our results. While done by others [30
] and recommended by the questionnaire developers (personal communication), assessing the adherence to the PA recommendations by means of the PPAQ might present a methodological shortcoming and might explain some inaccuracies with observed effects of PA intensities. We acknowledge that objectively measuring prenatal PA using accelerometers might provide a more precise assessment and might provide a more accurate estimation of changes and variations in the PA level during the course of pregnancy. Since the GeliS trial was performed in a large cohort within the real-life setting of routine prenatal care, using other methods of PA data collection in place of or in addition to the PPAQ was not feasible. A minor limitation is that our cohort differed slightly from the general German population of women of child-bearing age in terms of educational level and BMI categories [53
], which needs to be considered when generalizing our findings. Moreover, neonatal and obstetric outcomes were collected from different hospitals, and data collection was, thus, not completely standardized. In addition, there was no possibility to collect offspring´s body composition measurements, which may have expanded some of our findings beyond associations with crude measurements of body weight and BMI. Lastly, we are aware that some pregnancy-induced complications such as premature contractions that potentially led to a reduction of maternal PA, could have resulted in obstetric complications (e.g., preterm birth) and could have biased our results.
There are several strengths of this analysis that merit particular attention. Current research has either investigated women with normal weight or studied women with overweight and obesity. The GeliS cohort comprised women of all BMI categories, which allowed us to report the impact of prenatal PA for women in all weight ranges. Irrespective of shortcomings, the PPAQ is a valid and easily applicable tool that enables an extensive description of PA behavior during pregnancy. Using the PPAQ, we could not only estimate the impact of different PA intensities, but could also sub-group participants according to their activity level into meeting or not meeting the ACOG prenatal PA recommendations. This gave us the opportunity to report differences in obstetric and neonatal outcomes between sub-groups. Data for this study were collected within a public health approach under real-life conditions without requiring further measurements or tools, which allowed for methodological advantages. First, we were able to follow participants and collect data longitudinally over the course of pregnancy, which allowed us to observe the impact of prenatal PA in early as well as in late pregnancy. Second, we were able to collect data on a relatively large study sample. This provided a comprehensive and valuable assessment of the influence of antenatal PA on infant health outcomes as well as obstetric parameters.
To the best of our knowledge, there is no other trial that has provided a comprehensive description of the impact of early as well as late prenatal PA on obstetric and neonatal outcomes in such a large sample and that was additionally able to demonstrate the influence of PA intensities on these parameters.