Unfavourable birth weight, as characterised by either low birth weight (<2500 g) or macrosomia (>4000 g), accounts for approximately 20 % of births worldwide(Reference Englund-Ögge, Brantsæter and Juodakis1). Low birth weight is more prevalent in low- and middle-income countries and especially prevalent among vulnerable populations (lower socio-economic status)(2). The prevalence of macrosomia has increased markedly in the past few decades especially among high-income countries and women with increased maternal weight and gestational weight gain(Reference Surkan, Hsieh and Johansson3). Both low birth weight and macrosomia are associated with infant morbidity and mortality, as well as increased risk of poor health status later in life(Reference Risnes, Vatten and Baker4). For example, low birth weight is linked to increased risks of non-communicable diseases such as cardiovascular disease and diabetes, while macrosomia is associated with increased risks of obesity and cancers in adulthood(2,Reference Risnes, Vatten and Baker4) . Mothers’ health behaviours during pregnancy, such as their dietary intake and activity levels, may influence fetal development and be an important modifiable determinant of healthy birth weight(Reference Murphy, Stettler and Smith5,Reference Fazzi, Saunders and Linton6) .
Pregnancy is a critical time for caution about diets. Maternal dietary habits can have a direct effect on the growing fetus(Reference Blumfield, Hure and MacDonald-Wicks7) and subsequently the outcomes of pregnancy(Reference Kind, Moore and Davies8). However, findings from previous research on the relationship between maternal dietary intake and birth weight remain inconclusive. A Norwegian study of 65 904 mother–child dyads linked poor diet to increased risks of both low birth weight and macrosomia(Reference Englund-Ögge, Brantsæter and Juodakis1). Two studies, one from Japan (803 infant–mother dyads) and the other from Denmark (44 612 infant–mother dyads), found that a low-quality diet was associated with an increased risk of low birth weight(Reference Okubo, Miyake and Sasaki9,Reference Knudsen, Orozova-Bekkevold and Mikkelsen10) . These findings contrast with a US study of forty-one infant–mother dyads, which suggested that poorer diet quality may be related to higher birth weight(Reference Grandy, Snowden and Boone-Heinonen11). Another study, from the USA, found no association between maternal dietary intake and birth weight(Reference Poon, Yeung and Boghossian12).
Another health behaviour during pregnancy that may influence birth weight is sedentary behaviour, which is mainly attributed to prolonged screen time. With an escalation in the use of digital technologies over the last two decades, prolonged and excessive screen time has become a public health concern. The increased sedentary time during pregnancy is associated with higher levels of LDL-cholesterol in mothers and a larger newborn abdominal circumference. However, its association with child birth weight remains inconclusive(Reference Fazzi, Saunders and Linton6). A Japanese study showed that the prevalence of excessive mobile phone use was significantly higher among pregnant women compared with non-pregnant women(Reference Lu, Oda and Ohba13). A study conducted in Australia found over one-third of study participants (408 first-time mothers) spent more than 3 h/d on screens(Reference Wen, Simpson and Baur14). Similarly, a Chinese study, which included 2345 pregnant women, found more than a quarter of participants had prolonged television and computer viewing time (2 h or more/d), and more than three-fifths of participants reported prolonged mobile phone viewing time (1 h or more/d)(Reference Xu, Liu and Rao15). The association between screen time during pregnancy and birth weight has, to our knowledge, only been addressed in a Japanese study, which suggested that excessive mobile phone use may be linked to an increased risk for low birth weight(Reference Lu, Oda and Ohba13).
A better understanding of the association between maternal dietary and sedentary behaviours during pregnancy and birth weight will inform future health promotion interventions. Research on the association between mothers’ dietary and sedentary behaviours and birth weight is scarce and results are inconclusive. To address this knowledge deficit, this study investigated whether mothers’ dietary and sedentary behaviours during pregnancy are associated with their children’s birth weight outcomes.
Methods
Study design
A secondary data analysis was conducted using the baseline data from a three-arm randomised controlled trial, Communicating Healthy Beginnings Advice by Telephone (CHAT)(Reference Wen, Rissel and Baur16), a health promotion intervention aimed at reducing early risk factors for childhood obesity. The CHAT study delivered an intervention targeting body mass index (BMI), eating and screen time behaviours of children in the first years of life. Participants were assigned into telephone, short message service or control arms. Staged interventions were delivered for mothers from the antenatal stage (third trimester of pregnancy) until children were aged 2 years. The CHAT randomised controlled trial was conducted across four local health districts in New South Wales (NSW), Australia: Sydney, South Eastern Sydney, South Western Sydney and Southern NSW. The trial was approved by the Sydney Local Health District Ethics Review Committee (Protocol No. X16–0360 and LNR/16/RPAH/495). Further details of the CHAT randomised controlled trial’s protocol have been published elsewhere(Reference Wen, Rissel and Baur16).
Participants
Pregnant women in their third trimester (28–34 weeks) were recruited from eight hospital sites within the above four districts between February and July 2017. Potential participants were approached by research assistants with participant information sheet at the antenatal clinics and invited to participate in the study. Women were eligible to participate in the study if they were 16 years or above, between 28 and 34 weeks pregnant, capable to communicate in English, owned a mobile phone device and resided in the recruitment areas. Women were excluded if they had a severe medical condition, were expecting multiple births or with babies with known major fetal anomalies. Written consent was obtained from each participant. Participants were then required to fill in a registration form to provide their contact details for baseline data collection. The recruitment process was shown in Fig. 1. Details of the CHAT study recruitment and participants have been published elsewhere(Reference Ekambareshwar, Mihrshahi and Wen17).
Recruitment process
Measures and data collection
Data collection was conducted by a market survey company using the computer-assisted telephone interview. All data were collected at baseline when mothers were 28–34 weeks pregnant except the birth weight and child sex data. Data on child birth weight and sex were collected when mothers completed the telephone survey at 6 months post-partum.
Child birth weight
Mothers were asked to provide the birth weight and child sex noted in the My Personal Health Record (the ‘Blue Book’), which was issued to all parents in NSW when their child was born. The newborn birth weight was measured and recorded together with other birth data in the Blue Book(18). The birth weight was grouped into three categories based on the common birth weight cut-off points: low birth weight (<2500 g), normal birth weight (2500 to <4000 g) and macrosomia (≥4000 g)(2,Reference Dennedy and Dunne19) .
Mothers’ dietary and sedentary behaviours
Dietary behaviour was assessed using questions derived from the NSW Adult Population Health Survey, which was an annual cross-sectional telephone-based survey of residents aged 16 years and older living in NSW(20,21) . The validity and reliability of the survey questions have been demonstrated in adult population in NSW(Reference Riley, Rutishauser and Webb22) and have previously been used to examine dietary behaviours among pregnant women(Reference Wen, Simpson and Baur14,Reference Wen, Simpson and Rissel23) . Mothers were asked about their frequency of processed meat, fast food and potato crisps intake (times per day or week or month). Responses were recorded as times per day/per week/per month, rarely/never and don’t know/refused. Responses were further dichotomised into ‘Yes’ and ‘No’, where ‘No’ referred to rarely/never based on the diet recommendation for pregnant women (avoid proceed meat, fast food and potato crisps)(24). The amounts of vegetables and fruit (serves per day or week) and soft drinks intake (cups per day or week or month) were also measured. Vegetable, fruit and soft drink consumptions were further categorised into two groups (Yes/No) based on whether the mother met the recommended daily consumption of vegetables (≥5 serves/d), fruit (≥2 serves/d) and soft drink (no) during pregnancy(24). A continuous variable (ranging from 0 to 6) called ‘Dietary behaviour’ was generated based on the number of dietary recommendations the mother met. The higher the score, the better the dietary behaviour. Another continuous variable (ranging from 0 to 4) called ‘Junk Food’ was generated based on the number of ‘Junk Food’ items (processed meat, fast food, chips and soft drink) consumed by the mother. The higher the score, the worse the dietary behaviours (online Supplemental File).
The sedentary time question was derived from the NSW Adult Population Health Survey in 2015(25). Participants were asked about their daily time of sitting. The mean and median screen time were both 5 h/d. Therefore, sedentary time was dichotomised into ‘≤5 h/d’ and ‘>5 h/d’. Dietary and sedentary time questions used in this survey are listed in the online Supplemental File.
Mothers’ demographics
Questions from the NSW Adult Population Health Survey(20) were used to collect information on mothers’ age, country of birth, language spoken at home, education level, household income, marital and employment status, whether they were first-time mothers, pre-pregnancy weight and height, smoking status and gestational diabetes status (including pre-pregnancy diabetes). Mothers’ pre-pregnancy weight and height data were converted into pre-pregnancy BMI (kg/m2). Mothers were grouped into four categories based on their pre-pregnancy BMI: <18·5 kg/m2 (underweight), healthy weight (18·5 to <25 kg/m2), overweight (25·0 to <30 kg/m2) or obesity (≥30 kg/m2)(Reference Li, Zeng and Wang26). Gestational age was estimated based on mothers’ self-reported weeks pregnant on the day they were recruited, and the child’s date of birth collected at the 6 months telephone survey.
Statistical analysis
Statistical analysis was performed using Stata software, version 16 (StataCorp LP 2016). Statistical significance was set at a P value <0·05 for all statistical tests, and testing was two-sided. Mothers’ demographics and their dietary and sedentary behaviours were presented in numbers and percentages by three birth weight groups (low, normal and macrosomia). They were compared across the birth weight status by using Chi-squared tests. Multinomial logistic regression models were built to investigate whether children’s birth weight status was associated with mothers’ dietary and sedentary behaviours during pregnancy. Chi-squared and Pearson’s correlation tests were conducted to assess the multicollinearity between the variables before adding them into the regression model. Since the CHAT study was a randomised controlled trial and the intervention started at late pregnancy, in order to take intervention effect into account, all multinomial logistic regression models were adjusted for intervention allocation. Based on previous literature(Reference Wen, Simpson and Rissel23,Reference Coelho, Cunha and Esteves27) , maternal age, country of birth, language spoken at home, pre-pregnancy BMI, first-time mother, gestational diabetes during pregnancy, smoking status during pregnancy, infant sex and gestational age were considered as potential confounding factors. Backward elimination approach was used to identify confounding factors. All potential confounding factors were included in a multinominal logistic regression model. The least significant variables were progressively dropped until only those with P < 0·05 remained. Adjusted risk ratios and the 95 % CI were reported. To test the robustness of the association between maternal dietary and sedentary behaviour during pregnancy and children’s birth weight status, sensitive analysis were conducted by comparing the relative risk ratios for low birth weight or macrosomia between bivariate and multinomial logistic regression models. Chi-squared test for trend was used to examine whether the risk of low birth weight or macrosomia decreased or increased with increasing the number of dietary recommendations met by mothers and ‘Junk Food’ items consumed by mothers.
Results
Baseline data were available for 1155 women. A total of 1132 birth weight data were obtained and included in this current study. Maternal characteristics are shown in Table 1. Sixty-three percentage of mothers were born overseas, predominantly in Asia, approximately 54 % of mothers spoke English at home, 61 % had normal pre-pregnancy BMI, 97 % were non-smokers and 68 % did not have diabetes (including gestational diabetes). The mean birth weight was 3·32 kg (sd 0·51, range: 1·15–4·90 kg). The majority of infants were of normal birth weight (87 %), with only 5 % of low birth weight and 8 % having macrosomia. There were significant differences in mothers’ country of birth, language spoken at home, pre-pregnancy BMI, diabetes and being a first-time mother among children’s birth weight status. Mean and median gestational age were both 39 weeks.
Mothers’ baseline characteristics by infant birth weight status
SMS, short message service.
* HSC, higher school certificate (Year 12).
† TAFE, technical and further education.
P, P-value of Chi-squared test.
Maternal dietary and sedentary behaviours by child’s birth weight status are presented in Table 2. Most mothers met dietary recommendations for fruit (≥2 serves/d, 71 %) and soft drink (no soft drink, 63 %). More than half of mothers (60 %) had sedentary time less than the average (5 h/d). Less than half of mothers met dietary recommendations for fast food (38 %), processed meat (48 %) and chips (25 %). Only 9 % of mothers met recommendation for vegetable (5 serves per day). While there were no significant differences in mothers’ fruit, vegetable, soft drink, fast food and sedentary time among children’s birth weight status, there were significant differences in mothers’ consumption of processed meat and chips among children’s birth weight status. Higher proportion of mothers who ate processed meat and chips were observed in the macrosomia group. Although, overall, the differences in mothers’ fast food consumption among children’s birth weight status were not significant, higher proportion of mothers who ate fast food was observed in the macrosomia group.
Mothers’ dietary and sedentary behaviours during third trimester by infant birth weight status
P, P-value of Chi-squared test.
Results from multinomial logistic regression analyses are presented in Table 3. Mothers who ate fast food (risk ratio (RR) 1·64, 95 % CI (1·03, 2·63)), processed meat (RR 1·91, 95 % CI (1·22, 2·99)) and chips (RR 2·16, 95 % CI (1·18, 3·95)) were more likely to have macrosomia than children with normal birth weight. However, after adjusting for confounders, macrosomia was only associated with mothers’ processed meat consumption. The macrosomia RR was attenuated by 6 %. Compared with mothers who did not eat processed meat, mothers who ate processed meat were more likely to have infants with macrosomia than normal birth weight with adjusted risk ratio 1·80 (95 % CI (1·12, 2·89)). Nevertheless, macrosomia was associated with mothers’ ‘Junk Food’ consumption (expressed as the number of four ‘Junk Food’ items – soft drink, fast food, processed meat, and chips the mother consumed) after adjusting for confounders. For every increase in the items of ‘Junk Food’ consumed by a mother, a 31 % increase in the relative risk of macrosomia would occur (adjusted risk ratio 1·31, 95 % CI (1·07, 1·60)). The Chi-squared test for trend also showed a strong evidence that the risk of macrosomia increased with increasing the items of ‘Junk Food’ consumed by a mother (P < 0·0001). When examining mothers’ whole diet, macrosomia was associated with mothers’ dietary behaviour (expressed as the number of dietary recommendations met by the mother) after adjusting for confounders. For every increase in the number of dietary recommendations met by a mother, a 16 % reduction in the relative risk of macrosomia would occur (adjusted risk ratio 0·84, 95 % CI (0·71, 0·99)). The Chi-squared test for trend also showed that the risk of macrosomia decreased with an increasing number of dietary recommendations met by a mother (P = 0·002). For mothers’ ‘Junk Food’ consumption and dietary behaviour, the macrosomia RR was attenuated by 7 % and 5 % after controlling for confounders. There were no significant associations between children’s birth weight status and mothers’ sedentary time. And children’s low birth weight was not associated with mothers’ dietary and sedentary behaviours during pregnancy.
Risk of low birth weight and macrosomia by mothers’ dietary and sedentary behaviours during third trimester
RR, risk ratio; ARR, adjusted risk ratio. Adjusted for mothers’ language spoken at home, pre-pregnancy BMI, diabetes (including gestational diabetes), gestational age, infant sex and intervention allocations.
Junk food, processed meat, fast food, chips and soft drink, ranging from 0 to 4.
Dietary behaviour, number of dietary recommendations met by mothers, ranging from 0 to 6.
Discussion
In this study, we found that mothers’ consumption of processed meat during late pregnancy was associated with macrosomia. Although no statistically significant associations were found between children’s birth weight status and mothers’ fruit, vegetable, soft drink, fast food and chips consumptions, the risk of macrosomia decreased with increasing the number of dietary recommendations met by a mother. These findings suggest that encouraging mothers to meet more dietary recommendations, especially avoiding processed meat during pregnancy, may reduce the chance of having high birth weight.
The study also showed the risk of having macrosomia increased with an increase in the items of ‘Junk Food’ consumed by the mother. This finding is aligned with another Australian study on 368 first-time mothers showing maternal ‘junk food’ diet increased the risk of having a macrosomic infant(Reference Wen, Simpson and Rissel23). Similarly, a Brazilian longitudinal study analysing 1298 pregnant women also found a positive association between salty snacks (i.e. food with high concentration of simple carbohydrates, fats and low amounts of protein and micronutrients) and high birth weight(Reference Coelho, Cunha and Esteves27).
The current study did not find associations of maternal fruit and vegetable intakes with their child’s birth weight. These findings are aligned with a systematic review, which found inconclusive evidence of a specific protective effect of fruit and vegetable consumption on infant birth weight(Reference Murphy, Stettler and Smith5). In contrast, a large-scale prospective cohort study among pregnant Norwegian women found protective effects of diets comprising vegetables, fruit and water on birth outcomes(Reference Englund-Ögge, Brantsæter and Haugen28). Similarly, a study among Danish pregnant women found a small but statistically significant positive association between fruit and vegetable consumption and birth weight(Reference Mikkelsen, Orozova-Bekkevold and Knudsen29). It should be noted that only 9 % of our study participants met the recommended vegetable consumption, and hence the association between vegetable consumption and child birth weight could be attenuated since the majority of the participants did not meet the dietary recommendations.
The current study found that more than half of the mothers did not meet the dietary recommendations for vegetable (≥5 serves/day), fast food (no), processed meat (no) and chips (no) during pregnancy. Although the majority (91 %) did not meet the recommendation for vegetables, more than half reported having 2–4 serves of vegetables per day. The vegetable consumption of our study participants was comparable with another study in Australia that found only 10 % of pregnant women met the recommendation, with a median of 2 serves per day(Reference Malek, Umberger and Makrides30). The proportion of mothers who ate processed meat was also much less in our study (52 %) than the previous study (89 %)(Reference Malek, Umberger and Makrides30). In the current study, 32 % of women had diabetes (including gestational diabetes). This was higher than the Australia wide data (16·1 % in 2017–2018)(31), which may be attributed to the high percentage of women from Asian backgrounds(Reference Menard, Sotunde and Weiler32). Regarding the birth weight status, the rate of macrosomia (8 %) and low birth weight (5 %) in the current study was comparable to the population data in NSW in 2017 at 9·5 % and 6·7 %, respectively(33).
In this study, we did not find associations between mothers’ sedentary time and children’s birth weight status. In contrast to the data on maternal physical activity and pregnancy outcomes, research on association between sedentary behaviours and birth weight is scarce and with conflicting findings. Three studies found no association between sedentary behaviour during pregnancy and birth weight(Reference Fan, Huang and Cui34–Reference Ruifrok, Althuizen and Oostdam36). One UK study found that sedentary behaviours (assessed by asking if participants mostly sitting during pregnancy) were inversely associated with infant size at birth(Reference Both, Overvest and Wildhagen37). A Japanese study also found that excessive mobile phone use during pregnancy was a significant predictor of lower birth weight(Reference Lu, Oda and Ohba13). In contrast, a cross-sectional study on 112 pregnant women in UK found increased sedentary time in third trimester was linked to macrosomia(Reference Reid, McNeill and Alderdice38). To our knowledge, there are no specific recommendations developed for sedentary time during pregnancy. Recommendations for pregnant women emphasise reducing sitting time and breaking up long periods of sitting or standing still(39). More research on the association between sedentary time during pregnancy and birth outcomes is needed to inform guidelines for pregnant women.
Strengths and limitations
The strengths of the current study are the large sample size and prospective design of the study. The prospective design, whereby dietary and sedentary behaviours were assessed before the outcome (birth weight), has minimised recall bias. To our knowledge, this is the first study that has examined the associations of maternal dietary and sedentary behaviours with birth weight in Australia. This current study contributes to an under-researched area.
However, this investigation has several limitations. First, gestational age data were estimated based on mothers’ self-reported weeks pregnant, which is subjected to estimation error. Birth weight is determined by both fetal growth and the duration of gestation(Reference Klebanoff and Yip40). And rapid gestational weight gain was also associated with macrosomia(Reference Goldstein, Abell and Ranasinha41). However, our study did not have access to the participants’ medical records to collect those data. Second, both the maternal dietary and sedentary behaviours were based on questionnaires reported by the pregnant women. Self-reported data, especially dietary habits, have often been criticised for their subjective nature and estimation error. Previous studies that investigated dietary habits have also showed that female participants tend to under-report their unhealthy food consumption in a way to portray themselves more favourably(Reference Macdiarmid and Blundell42). Although the questions used to measure dietary and sedentary behaviours were both derived from questionnaires that have previously been validated and widely used in adult populations(Reference Riley, Rutishauser and Webb22,25) , and been used to assess health behaviours during pregnancy(Reference Wen, Simpson and Baur14,Reference Wen, Simpson and Rissel23) , the questionnaire has not been specifically validated for use during pregnancy. Third, human diet is complex and the dietary questions may not capture the whole picture of dietary intake. For instance, alcohol intake during pregnancy that may influence birth weight(45) was not assessed in the survey; the whole energy intake (the amount of food) was not assessed in the survey either. Additionally, the dietary questions predominantly assessed the Anglo-Celtic type of food and beverages. Cultural food preferences were not taken into consideration. Around 63 % of our study population were born outside of Australia. The survey may not have captured the overall dietary intake of some participants. Fourth, the survey only asked about time spent sitting on a weekday but not on a weekend day. Sedentary time may vary between weekend and weekdays. Therefore, mothers’ sedentary time we assessed may not reflect their sedentary time correctly. In our study, 40 % (n 443) mothers spent more than 5 h/d on sitting. Fazzi et al. assessed twenty-six studies on sedentary behaviours during pregnancy, with time spent in sedentary behaviours ranging from 7 to 18 h/d(Reference Fazzi, Saunders and Linton6). Lastly, both dietary and sedentary behaviours measures in the current study only captured a snapshot of the third trimester pregnancy; they may not reflect the dietary and sedentary behaviours during the entire pregnancy.
Conclusion
Maternal dietary behaviours are independently associated with macrosomia. Our findings provide valuable insights for future health promotion interventions aiming to improve pregnancy outcomes by reducing mothers’ unhealthy dietary behaviours.
Acknowledgements
The authors would like to acknowledge all the participating families in the study, the telephone survey company, research assistants, Child and Family Health nurses, and the project partners from the Sydney, South Eastern Sydney, South Western Sydney and Southern New South Wales Local Health Districts, Australia.
Financial support
The CHAT RCT was funded by the NSW Health Translational Research Grant Scheme 2016 (ID number: TRGS 200).
Conflicts of interest
There are no conflicts of interest.
Authorship
Study concept and design: L.M.W. and L.B. Data analysis and interpretation: H.X., Y.W. and L.M.W. Drafting and approval of the manuscript: H.X. and L.B. drafted the manuscript with literature review. Y.W., P.P., L.A.B. and L.M.W. provided critical reviews and comments. All authors reviewed and approved the final version of the manuscript.
Ethics of human subject participation
This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Ethics Review Committee of Sydney Local Health District (Protocol No. X16–0360 and LNR/16/RPAH/495) on 17 October 2016. Written informed consent was obtained from all participants.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980023002161
Open access
