Infant growth during the first year of life following a pregnancy lifestyle intervention in routine care—Findings from the cluster‐randomised GeliS trial

Lifestyle interventions in pregnancy may influence postpartum development and obesity risk in offspring. The impact of lifestyle interventions as health system‐based approaches is unclear.

decades. 2,3 the worldwide prevalence of childhood obesity increased continuously, which is particularly alarming as childhood obesity was identified as a predictor for obesity during adolescence or later in life. 4 For these reasons, the World Health Organization declared childhood obesity as one of the major public health challenges of the 21 st century. 5 The aetiology of childhood obesity is multifactorial. Research suggests that several antenatal exposures augment individual obesity risk. 6,7 For example, a positive association between maternal pre-pregnancy BMI and the infant's obesity risk has consistently been observed. 6,8 Among all antenatal factors, maternal obesity is the strongest predictor for infant adiposity, increasing the risk by up to sixfold. 6,8 Moreover, total gestational weight gain (GWG) or GWG above a certain threshold (excessive GWG) 9 have been shown to influence weight development of offspring and their obesity risk. 6,8,10,11 Further, early postnatal factors such as breastfeeding pattern and formula feeding, introduction of solid foods, activity or inactivity level, and also sleeping behaviour and exposure to antibiotics are currently discussed to modify the infant's obesity risk in the short-and long-term. 7,12,13 Due to the potential to reverse health consequences of childhood obesity if weight normalisation is achieved by puberty, 14 identifying effective early-life strategies that tackle modifiable risk factors are urgently needed. Within the last decade, several randomised controlled trials have been initiated primarily with the aim to improve maternal health outcomes during pregnancy and postpartum. Metaanalyses confirmed the potential of these lifestyle interventions to modestly ameliorate GWG and some other health outcomes such as the odds of caesarean section or the extent of maternal postpartum weight retention 1 year after birth. 15,16 Although some studies also focused on long-term infant development as a secondary goal, a recent systematic review was unable to draw any conclusion on their influence on obesity outcomes in early childhood. 17 Reviewed studies had either small to moderate sample sizes or were conducted in community settings and academic facilities. To date, nothing is known about the effect of a large-scaled antenatal lifestyle intervention performed under real-life conditions on early life development of the offspring. Herein, we assessed the influence of the "Gesund leben in der Schwangerschaft" ("healthy living in pregnancy," GeliS) lifestyle intervention, that was conducted within the German routine care system, on infant growth during the first year of life.
The cluster-randomised GeliS trial primarily aimed to reduce the proportion of women with excessive GWG by offering comprehensive counselling alongside routine visits. While the intervention was not successful in limiting excessive GWG, it yielded some improvements in maternal antenatal dietary and physical activity behaviour, 18,19 as well as modest between-group differences in maternal breastfeeding behaviour. 20 Herein, our principle aim was to investigate the influence of the GeliS intervention on offspring weight development during the first year of life as well as on further developmental variables including obesity outcomes. We also examined the intervention effect on infant feeding practices. Additional cohort analyses sought to assess factors potentially influencing infant weight, BMI and obesity risk at 12 months of age.

| The GeliS study: Design and setting
The GeliS study is a prospective, cluster-randomised, controlled, open intervention trial, conducted in Bavaria (Germany). The primary outcome was to reduce the proportion of women with excessive GWG as defined by the Institute of Medicine (IOM). 9,21 Primary and some secondary outcomes have been published recently. [18][19][20][22][23][24] Details on the study design and setting, and clusterrandomisation, have been described in the study protocol. 21 In brief, the randomisation resulted in one control (CG) area and one interven-   The following criteria were considered, a priori, as reasons for exclusion: 21,22 miscarriage or late loss of pregnancy, terminations, pregnancy complications that interfered with the intervention, and maternal death during the course of the trial. If women were no longer reachable during the follow-up phase, did not provide contact details, or withdrew participation, they were defined as drop-outs in the follow-up.

| Participants and lifestyle intervention
While participants in the CG attended standard antenatal care and were provided with a flyer, which outlined general information on a healthy antenatal lifestyle and the importance of breastfeeding, the IG received a comprehensive lifestyle intervention programme.
Details on the lifestyle intervention programme have been described elsewhere. 21 In brief, the IG received three antenatal (12 th -16 th , 16 th -20 th and 30 th -34 th week of gestation) and one postpartum (6 th -8 th week postpartum) face-to-face counselling sessions alongside routine care visits. These sessions lasted between 30 and 45 minutes and were given by previously trained midwives, medical personnel or gynaecologists in their practices. Participants were counselled on adequate GWG according to the IOM recommendations, 9 and the importance of a healthy antenatal lifestyle in relation to optimal offspring development during childhood. In accordance with national and international recommendations, [25][26][27] women received lifestyle advice for the antenatal as well as the postpartum period including information on a healthy dietary and physical activity behaviour. 21 Breastfeeding advice according to German recommendations were outlined, with additional emphasis and information on the importance of breastfeeding for both mothers and their offspring. 21,25 Additionally, women were provided with information on the introduction of complementary food, infant hunger and satiety signals and infant feeding practices according to the recommendations. 21,25 A follow-up observation programme, which was identical in both groups, started subsequent to the intervention phase. Infant anthropometrics in the postpartum period, measured in paediatric practices and documented in routinely used health records, were enquired within a phone interview in the 12 th month after birth. Moreover, data on infant feeding practices were collected via a set of questionnaires completed by women 12 months after birth. More information on the follow-up programme is provided elsewhere. 21

| Data collection and outcomes
Anthropometric and sociodemographic characteristics of participating women were collected by means of a screening questionnaire at the time of recruitment. The women's weight was measured in participating gynaecological or midwifery practices during the course of pregnancy and 6-8 weeks postpartum and was retrieved from maternity records. Maternal pre-pregnancy BMI was calculated based on self-reported pre-pregnancy weight. GWG was defined as the difference between the last measured weight before delivery and the first measured weight at the time of recruitment. Preterm delivery was defined as giving birth before the 37 th week of gestation. 22 Measurements of infant anthropometrics at birth (time point U1) were performed in hospitals and documented in maternity and birth records. Offspring with birth weight above the 90 th percentile for gestational age were defined as being "large for gestational age" and below the 10 th percentile for gestational age as being "small for gestational age." Infant development during the first 12 months was assessed at paediatric practices within the routine health check-up programme for infants. Weight, length and head circumference of the infants were measured at five time points in these health examinations (U2: 3 rd -10 th day postpartum; U3: 4 th -6 th week postpartum; U4: 3 rd -4 th month postpartum; U5: 6 th -7 th month postpartum; U6: 10 th -12 th month postpartum) and documented in the routinely used infant health records. Weight, length and BMI were converted into sex-specific percentiles and z-scores for age using a German reference group. 28 According to the German recommendations for children aged between 0 and 18 years, infants with a BMI-for-age-percentile below 10.0 were classified as being underweight, while a percentile above 90.0 and 97.0 was defined as being overweight and obese, respectively. 28 Data on infant feeding practices were collected within the set of questionnaires via questions adapted from the "German Health Interview and Examination Survey for Children and Adolescents" (KiGGS). 29 "Any breastfeeding" was defined as breastfeeding at any time and "exclusively breastfeeding" as breastfeeding without the addition of any formula or complementary food. Moreover, selfreported paternal weight and height were enquired within the set of postpartum questionnaires.

| Statistical analysis
Power calculation was conducted based on the primary study outcome (excessive GWG defined by the IOM 9 ) and was described elsewhere. 21 Power calculations were not performed for secondary outcomes. All analyses presented herein were performed using SPSS software (IBM SPSS Statistics for Windows, version 24.0, IBM Corp, Armonk, New York).
Baseline characteristics are depicted for all mother-infant pairs that completed the active phase and thus entered the follow-up period.
Analyses on infant anthropometrics included all mother-infant pairs that provided data for the corresponding time point (U1-U6) except those who were lost to follow up. For outcomes measured more than once, likelihood-based mixed models for repeated measures according to Bell et al. 30 were fit using data from each visit (U1-U6). Through the inclusion of visit number (as a factor) and group assignment, and their interaction, these models provide point estimates and 95% confidence intervals (CIs) for the mean differences between groups at each visit. In adjusted analyses, maternal age, pre-pregnancy BMI category, parity, infant sex, infant age in days at the corresponding visit and study region were considered as further independent variables. For outcomes only assessed at U6, and infant feeding practices, between-group differences were investigated using linear, binary logistic or proportional odds ordinal logistic regression models fit with generalised estimating equations (GEEs) according to Donner et al. 31 These results are presented as estimated mean differences or odds ratios (OR) along with the 95% CI. In adjusted models, the same covariates as for likelihood-based mixed models for repeated measures were considered with one exception for feeding pattern; instead of infant age, the time interval between questionnaire completion date and birth date was used.
To identify predictors of infant weight and the risk for overweight or obesity at the 12 th month postpartum, mother-infant pairs in the IG and CG were pooled to form one cohort. For all cohort analyses, unadjusted linear and binary logistic regression models were applied, as well as models adjusted for the aforementioned covariates and group assignment. As defined a priori, 22 analyses related to GWG excluded subjects with preterm delivery.
For all analyses, P values below .05 were considered as statistically significant. No adjustment was made for multiple comparisons.  Table 1 shows characteristics of mother-infant pairs entering the follow-up including weight characteristics of the father. Mean selfreported weight of mothers and fathers were comparable between groups. In total, 64.9%, 23.1% and 12.0% of women had pre-pregnancy normal weight, overweight and obesity, respectively. As previously noted, 22 Table 1).
The proportions of mother-infant pairs that were lost to followup were comparable between groups (IG: 10.1%; CG: 11.5%). As shown in Table S1, characteristics of women that were lost to followup differed slightly from those remaining in the study in terms of educational level, history of GDM, parity, country of birth and smoking status. Infant birth outcomes were comparable between those that were lost to follow-up and those remaining in the study.

| Infant anthropometrics
Mean infant weight and further anthropometric outcomes are shown in Table 2. Figure S1 illustrates infant weight development during the first year of life for IG and CG, depicted as interquartile ranges. The relationship between infant weight in the CG and IG groups changed over the course of 12 months. From birth until the 3 rd -4 th month F I G U R E 1 Participant flow postpartum, mean infant weight was lower in the IG compared to the CG. At the 6 th -7 th month postpartum, mean weight was similar in both groups, and at 12 months IG infants were estimated to be heavier (Table 2, Figure S1). Statistical significance of between-group differences was only observable at U2 (3 rd -10 th day postpartum) and U3 (4 th -6 th week postpartum) ( Table 2).
The mean length of IG infants was significantly lower than in the CG until the 10 th day after birth (Table 2). Likewise, IG infants had a smaller head circumference compared to the CG from the 4 th week until the 12 th month postpartum with significant findings at the U3 (4 th -6 th week postpartum), U4 (3 rd -4 th month postpartum) and U6 (10 th -12 th month postpartum). At 12 months of age, neither infant BMI nor any other age-and sex-specific outcomes differed significantly between the two groups. The proportions of infants in the different weight categories at 12 months postpartum were comparable between groups ( Table 2). The incidence of overweight (> 90 th BMI percentile) did not differ significantly between groups (IG: 12.2%; CG: 11.8%; adjusted OR: 1.02, 95% CI 0.73 to 1.43; P = .893). The mean weight and BMI in infants of mothers with overweight was significantly higher in the IG compared to the CG (Table S2). Moreover, T A B L E 1 Characteristics of mother-infant pairs entering the follow-up mean weight in infants of mothers with obesity was lower in the IG, however lacking statistical significance (Table S2, P = .081). There was significant evidence of an interaction effect between group allocation and maternal pre-pregnancy BMI (Table S2).

| Infant feeding practices
As previously reported, 20 women in the IG were more likely to breastfeed their infants exclusively. Further characteristics of infant From linear regression models fit using GEEs controlled for maternal pre-pregnancy age, maternal pre-pregnancy BMI, parity, sex, infant age (days) except age-and sex-specific percentiles and z-scores (not controlled for infant age and sex). d All z-scores and percentiles were calculated according to Kromeyer-Hauschild et al. 28 e From proportional odds ordinal logistic regression models fit using GEEs controlled for maternal pre-pregnancy age, maternal pre-pregnancy BMI, parity.
feeding practices are shown in Table 3. There was no significant evidence of any differences in formula-related outcomes. A higher proportion of infants in the IG received whole-grain products (95.6% vs. 90.8%; P = .003). The proportion of infants receiving other specific solid foods did not differ between groups ( Table 3). The timing of introducing some solid foods such as vegetables, fruits, fruit juice and puree prepared with milk to infants differed significantly by 3-12 days per food component between groups with women in the IG introducing these foods later to their infants (data not shown).

| Determinants of infant weight and risk for overweight at the 10 th -12 th month of life
In the GeliS cohort, infant weight and BMI in the 10 th -12 th month postpartum were positively associated with maternal pre-pregnancy BMI and GWG (Table S3, Table S4). There was a negative association between infant weight but not infant BMI and maternal pre-pregnancy age (Table S3, Table S4). An increase in paternal BMI was linked to an increase in infant BMI but not infant weight in the 10 th -12 th month postpartum (Table S3, Table S4). Infant weight and BMI at 1 year of life were positively related to large for gestational age and high birth weight and inversely associated with small for gestational age (Table S3, Table S4). There was a negative association between preterm birth, any breastfeeding or exclusive breastfeeding and infant weight but not infant BMI (Table S4). Moreover, higher GWG or paternal BMI as well as a high birth weight or large for gestational age significantly increased the odds of being overweight or obese at the time point 10-12 months postpartum (Table 4).

| DISCUSSION
To the best of our knowledge, the GeliS study is the first large-scaled trial that assessed the effect of an antenatal lifestyle intervention for pregnant women conducted under real-life conditions on infant development during the first year of life. We could not identify a substantial effect on infant outcomes at 10-12 months of age. Some significant between-group differences during the first year of life were observed, such as a slightly lower mean infant weight (3 rd -10 th day and 4 th -6 th week assessment) and length (birth and 3 rd -10 th day assessment) in the IG. Importantly, the estimated mean differences were small and might be explained by a number of outliers in the IG.
Moreover, there were no remarkable differences in BMI, age and sexadjusted BMI z-scores at these time points (data not shown). Given that the minor difference in weight and length disappeared by the 12 th month postpartum, we question the clinical relevance of observed significant differences. These concerns also extend to the findings on head circumference measurements during the first year.
Difficulties in measuring infant head circumference in a standardised way were frequently reported and may explain observed differences.
Apart from aforementioned findings, we found no significant between-group differences in other sex-and age-adjusted outcomes at the 12 th month postpartum including infant weight categories or the odds of being overweight. These findings suggest that the GeliS intervention did not have a sustained effect on infant anthropometrics up to the 12 th month of life. Thus, we were unable to confirm that our lifestyle intervention impacted infant obesity risk.
Our results are consistent with findings from our pilot trial where we did not find any significant difference in infant weight between groups in the 10 th to 12 th month postpartum. 32  As reported elsewhere, 20 we found small differences in the proportion of women exclusively breastfeeding, with a slightly longer duration of exclusive breastfeeding in the IG. This might partly explain group differences in infant growth in the first weeks, as weight gain T A B L E 4 Factors influencing the odds of being overweight or obese in the 10 th -12 th month after birth defined as BMI-for-age-percentile > 90.0 Overweight/obesity in the 10 th -12 th month after birth per month was shown to be higher in formula-fed infants. 46 We also observed that more infants in the IG were fed whole-grain products.
This is consistent with the eating pattern of women in the IG, who also consumed more whole-grain products after the intervention compared to the CG. 19 However, we found no evidence of betweengroup differences in the exposure to any other type of solid food. The absence of an intervention effect corresponds to observations in some other lifestyle intervention studies, 34 analyses on the cohort level might be valuable to assess whether differences in infant feeding patterns relate to the incidence of childhood diseases such as allergy, asthma and immune disorders.
Although the GeliS intervention was not able to prevent excessive GWG, 22 it yielded some improvements in maternal dietary and physical activity behaviour, 18,19 and was shown to be successfully The sex-and age-adjusted outcomes are not completely transferable to other populations as we used a German reference group. 28 Furthermore, we classified overweight and obesity in children under 5 years of age according to German recommendations standards, 28 which differs from World Health Organization standards. 50 In addition, the power calculation was based on the primary study outcome (excessive GWG) 21 and secondary outcomes such as infant growth or feeding patterns were not taken into account. Moreover, we observed differences in characteristics between mothers lost-to-follow-up and those who remained in the study, and cannot exclude a risk of bias, information not only on offspring development but also on infant feeding patterns during the first year of life. Although this intervention neither effectively influenced infant health outcomes nor feeding patterns, the GeliS cohort will continue to be followed throughout early childhood in order to investigate the potential long-term impact on obesity risk of both mother and child.
Evidence outlined above demonstrated a large heterogeneity of reported offspring variables which makes an overall estimation of the effectiveness of antenatal lifestyle interventions in improving infants health outcomes challenging. 17 Currently ongoing meta-analyses might close this gap. 44,45 In conclusion, this analysis provides unique data on the effect of a real-life lifestyle intervention for pregnant women across all BMI categories on their infant's development and nutrition during the first year of life. The GeliS intervention was not able to substantively modify infant growth or feeding patterns. However, it is worthwhile to follow mother-child pairs in order to investigate the potential effects later on and to assess adiposity outcomes in early childhood as well as the incidence of obesity. As we were not able to demonstrate the effectiveness of our intervention at scale, we encourage further research to address the question of how maternal antenatal lifestyle can be targeted at the public health level to significantly improve health and long-term obesity outcomes of the mother and her offspring. Leben.Bayern."), the AOK Bayern, the largest statutory health insurance in Bavaria, as well as the DEDIPAC consortium by the Joint Programming Initiative (JPI) "A Healthy Diet for a Healthy Life." The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.