The associations of cola intake with adverse birth outcomes among pregnant women after assisted reproductive technology treatment and women naturally conceived: a birth cohort study

Abstract The influence of cola intake on birth outcomes is unclear. This study sought to describe and compare the associations between cola intake and adverse birth outcomes among women following assisted reproductive technology (ART) and women spontaneously conceived (SC). Participants (736 ART women and 1,270 SC women) were from the Chinese National Birth Cohort collected in Anhui province. Cola intake was assessed by self-reported questionnaires at each trimester. Outcome measures including preterm birth (PTB) and low birth weight (LBW) were extracted from medical records. The association between cola intake during pregnancy and PTB was found using multivariable log-binomial regression in combined ART and SC women. Separately, for ART women, cola intake during pregnancy increased the risk of PTB (risk ratios were 2.10, 1.65, and 1.81 for all three trimesters, respectively, all p < 0.05), and cola intake in the 1st trimester increased the risk of LWB (risk ratio 2.58, 95% confidence interval 1.29 to 5.16). Cola intake during pregnancy was not associated with PTB or LBW for SC women. Our findings indicate a detrimental effect of cola intake during pregnancy on birth outcomes for ART women. Thus, avoidance of cola intake should be counselled by medical doctors in women prescribed with ART treatment.


Introduction
Adverse birth outcomes mainly include preterm birth (PTB), and low birth weight (LBW), which are the leading causes of perinatal mortality and long-term morbidity in children (World Health Organization, 2018).Infertile couples have been shown to be at higher risks of PTB and LBW (Luke, 2017).Approximately half of infertile couples in developed or developing countries seek assisted reproductive technology (ART) treatment, and this is an opportunity for preconception optimisation (Boivin et al., 2007).Modifiable lifestyle risk factors such as drinking habit can be regulated to improve birth outcomes.
The consumption of cola is one of the largest sales worldwide (G� ornicka et al., 2014;Mirmiran et al., 2007), but the potential detrimental effect of cola on birth outcomes tends to be neglected compared with alcohol and coffee.Cola contains many chemical compounds such as sugar (sucrose or high-fructose corn syrup), phosphoric acid and caffeine, and these individual compounds have been shown to be associated with infertility and adverse birth outcomes (Goran et al., 2019;Imai et al., 2010;Klonoff-Cohen et al., 2002).However, the overall effect of cola intake on adverse birth outcomes, after controlling for intakes of other caffeine-containing beverages and functional beverages, has not been investigated.Therefore, this study aimed to investigate and compare the associations of cola intake with PTB and LBW among ART pregnant women and spontaneous conceived (SC) pregnant women.

Study design and population
Participants were from the Chinese National Birth Cohort in Anhui province, which is a prospective and longitudinal study including two parallel cohorts (i.e.ART and SC cohorts).The birth cohort targets the effect of environmental and genetic factors on birth outcomes and the differences between ART and SC populations.This study used the data from the ART and SC cohorts that were performed in Anhui province.The ART cohort in Anhui province recruited 2,210 infertile couples with indications for in vitro fertilization or intracytoplasmic sperm injection treatment at the Reproductive Center of the First Affiliated Hospital of Anhui Medical University between May 2017 and May 2021.Donation of oocyte was excluded for ART cohort.The SC cohort in Anhui province recruited 1,508 SC pregnant women who were in the 1st trimester immediately after the confirmation of pregnancy by ultrasound assessment at the Maternal and Child Health Care Center in Ma'anshan city of Anhui province between May 2017 and September 2018.The study was approved by the ethics committee of Anhui Medical University (20160270).All participants provided written informed consent.All research procedures were carried out in accordance with the Declaration of Helsinki.

Study procedure
Based on the time of recruitment, the baseline for ART cohort was pre-pregnancy, while the baseline for SC cohort was in the 1st trimester.The maternal age, height, weight and multiparity were recorded using self-reported questionnaires at baseline.Dietary, beverage intake and exercise were assessed using selfreported questionnaires at baseline and follow-ups in each trimester (i.e. at gestational weeks 4, 24, 32 respectively).All questionnaires were processed in a separated quiet room with the guidance of a trained research nurse.The reasons of infertility and gender of neonates were extracted from medical record.

Beverage intake
Participants were asked 'How often did you drink cola in the past year' at baseline.The responses were 'never or rarely', 'only in special occasion (e.g.festivals, being a guest or entertainment)', 'every month, but less than once a week', 'every week regardless of seasons, but less than once a day' and 'everyday'.Because less than 1% of pregnant women in ART cohort drank cola more often than special occasion in this study, a response of 'never or rarely' was defined as non-cola intake, and other answers including 'only in special occasion' were defined as cola intake.The similar question was also asked regarding the intake of alcohol, tea, coffee and functional beverage (e.g.Mizone and Red Bull), and the answers were transformed to binary variables using the same strategy.All the above questions were repeated in follow-ups, except that the timeframe was specified as 'since the last visit'.
Any cola intake during pregnancy was defined as cola intake in at least one of the three trimesters, and no cola intake was defined as no cola intake in all three trimesters.The consistent cola intake during pregnancy was defined as the total times of defined cola intake in all three trimesters, and the range was from zero trimester to three trimesters.The accumulation of cola intake was done for participants who presented all three visits.

Outcomes
Gestational week at delivery and infant birth weight were extracted from medical record.PTB was defined as birth before 37 weeks of pregnancy.LBW was defined as birth weight < 2500 kg.The analyses for PTB and LBW were limited in singleton birth and non-smoker because multiple birth and smoker had considerably poorer birth outcomes than singletons and non-smoker (Ashoor et al., 2022;van Heesch et al., 2014).
Participants were asked the average consumption frequency of each food item and the intake weight of each consumption.The responses for consumption frequency were 'everyday', 'four to six times a week', 'one to three times a week', 'one to three times a month', 'few times a week' and 'rarely'.The responses for intake weight of each consumption were 'less than 25 g', '25 g to 50 g', '50 g to 100 g', '100 g to 150 g', '150 g to 200 g', '200 g to 300 g' and 'greater than 300 g'.Total energy of the 25 kinds of food intake were calculated based on the China Food Composition Database (Y.-X.Yang, 2005;Y. Yang et al., 2002).Maternal BMI was calculated as height(m)/ weight(kg) 2 .The frequency of exercise was recorded as rarely, <3 times/week or �3 times/week.The reason of infertility was classified into four groups: female infertility, male infertility, both, and other unclear reason.

Statistical analysis
Continuous data were presented as mean (standard deviation) and categorical data as percentages.Normality tests of continuous data were performed by visual inspection of the histogram and distribution curve.Independent-sample t-tests and chi-squared tests were used to compare the differences of participant characteristics in the 1st trimester between the ART and SC cohort.Uni-and multivariable log binomial regression analyses were performed to examine the associations between cola intake and adverse birth outcomes.Conventional risk factors that were physiologically or statistically correlated with both exposures and outcomes were included as confounders in the adjusted models.Age, BMI at baseline, multiparity, gender of neonate, alcohol, tea, coffee, functional beverage, total food energy and exercise frequency were adjusted in multivariable model 1.The interaction between cola intake and cohort was determined by including an interaction term of cola intake with a cohort variable (SC or ART) in multivariable model 1 to indicate whether the association between cola intake and adverse birth outcomes differs between cohorts.The reason of infertility was additionally adjusted for ART cohort in multivariable model 1.The association between cola take during pregnancy and adverse birth outcomes was further adjusted for cola intake pre-pregnancy for ART women in multivariable model 2 in order to determine whether the association between cola intake during pregnancy and adverse birth outcomes is independent of cola intake before pregnancy.All statistical tests were 2-sided with the significance level set at p � 0.05.All analyses were performed using Stata 16.1 (StataCorp LP, TX, USA).

Participant characteristics
Figure 1 presents the flowchart of this study.In ART cohort, 932 (42.2%) women have successfully given birth after treatment, of which 162 (17.4%) gave multiple birth and 34 (3.6%) smokers were excluded from this study.In SC cohort, 238 women were excluded from this study for the following reasons: 8 (0.5%) refused to conduct the questionnaires, 206 (13.7%) gave multiple birth, and 24 (1.6%) were smokers.Therefore, 726 ART women and 1,270 SC women fulfilled the inclusion criteria and were included in the analysis.
The differences of participant characteristics between ART and SC population in the 1st trimester are presented in Table 1 and those in 2nd and 3rd trimester are in supplementary table.ART women were older, had a higher BMI, exposed to less cola, alcohol, tea, coffee and functional beverage, had higher food energy intake, and did less exercise than SC women.ART women were more likely to have preterm delivery and deliver LBW neonates compared with SC women.

The association between cola intake and adverse birth outcomes in combined ART and SC women
Table 2 summarises the association of cola intake with PTB and LBW in both cohorts.
Cola intake in the 2nd and 3rd trimester was associated with PTB.The difference in the magnitude of the association between two cohorts was not statistically significant, indicated by p values for the interaction being greater than 0.05, but the difference is substantial as demonstrated by 55%, 26% and 44% higher risk in the 1st, 2nd and 3rd trimester among ART women compared with SC women for PTB, and 116% and 52% higher risk in the 1st and 3rd trimester among ART women for LBW.

The association between cola intake and adverse birth outcomes in ART women
In ART women, cola intake during pregnancy, but not prepregnancy, was significantly associated with the higher risk of PTB in both the uni-and multi-variable models (Table 3).Moreover, ART women exposed to cola in any of the three trimesters associated with a higher risk of PTB compared with those without cola intake throughout the pregnancy, and this is independent of cola intake pre-pregnancy.Cola intake in the 1st trimester increased the risk of LBW compared with non-consumers after adjusting for the cola intake pre-pregnancy.
Table 5 shows the association between consistency of cola intake in the three trimesters and the risk of PTB and LBW.ART women who had cola intake in two of the three trimesters showed an increased risk of PTB compared with those without cola intake throughout the pregnancy (Table 5).Only 2.4% of pregnant women (n ¼ 12) drank cola in all three trimesters, and all of them had a normal weight neonate.Overall, the risk of PTB increased with the consistency of cola intake, indicated by the p for trend.

The association between cola intake and adverse birth outcomes in SC women
In SC women, cola intake in each or any of the three trimesters was not significantly associated with either PTB or LBW (Table 4).Exposure to cola in all three trimesters significantly associated with higher risk of PTB, and consistency of cola intake showed a trend to increase the risk of adverse birth outcomes (Table 5).

Discussion
This study is the first to investigate the influence of cola intake on adverse birth outcomes in both ART and SC women.This study suggested that pregnant women in the 2nd and 3rd trimester exposed to cola had higher risk of PTB; More specifically, ART women who had cola intake during pregnancy had a higher risk of PTB after adjusting for cola intake pre-pregnancy and other beverages.The consistency of cola exposure during pregnancy showed a relationship with the higher risk of PTB.Cola intake in the 1st trimester was associated with the increased risk of LBW in ART women; and unlike ART women, cola intake was not associated with either PTB or LBW in SC women.Many maternal characteristics have been shown to be associated with adverse birth outcomes, including maternal demographic characteristics, obstetric history and diabetes (Ashoor et al., 2022).In the current study, the proportion of cola exposure in SC women was 43.5% in the 1st trimesters, while the proportion among ART women was substantially decreased with 7.7%, which was sharply reduced from pre-pregnancy with 59.5%.However, the cola intake in the 2nd and 3rd trimesters among ART women tended to rise.This may be because ART women pay more attention to their drinking behaviours in the early stage of pregnancy.Although the proportion of alcohol, tea, coffee and functional beverages intake among ART women in the 1st trimester were substantially less than that among SC women, the proportion of cola intake was the highest than other beverages (i.e.tea, functional beverage, coffee and alcohol) in both cohorts, reminding that the potential detrimental role of cola on birth outcomes has been ignored.
This study revealed the universal and detrimental effect of cola intake on the PTB in both ATR and SC cohorts, but the associations of cola intake with PTB and LBW were clinically different between the two cohorts.The equivocal influence of caffeine-containing beverages (e.g.coffee, tea, functional beverage, and cola) on ART outcomes has been investigated in few studies (Al-Saleh et al., 2010;Choi et al., 2011;Klonoff-Cohen et al., 2002;Laizure et al., 2017;M� ınguez-Alarc� on et al., 2018), but the independent effect of cola intake on ART outcomes has never been investigated.Our findings showed that the adverse influence of cola intake during pregnancy on PTB was independent of other caffeine-contained beverages and total food energy among ART women.Moreover, we found that the effect of cola intake during pregnancy on PTB was independent of cola intake before pregnancy, and that there was a positive relation between consistency of cola exposure during pregnancy and the risk of PTB.The lack of substantial association between exposure of cola intake in all three trimesters and PTB could be because of the small sample (n ¼ 12).It is expected that all 12 ART women had normal weight neonates based on that the incidence rate of LBW infants among ART cohort was 7.3% as shown  At least one trimester in all three trimesters refers to any cola intake was defined as at least one cola intake in at least one of the three trimesters, and no cola intake was defined as no cola intake in all three trimesters for participants who presented all three visits.RR, the ratio of probability of an outcome in cola exposure group to non-cola exposure group.CI, confidence interval.� , p value for the RR < 0.05.Multi-model 1 was adjusted for age, BMI at baseline, multiparity, gender of neonate, alcohol, tea, coffee, functional beverage, total intake energy, exercise frequency and the reasons of infertility.Multi-model 2 was further adjusted for cola intake pre-pregnancy above the multi-model 1.
in Table 1.Association between cola intake and LBW was not observed in combined population but appeared in ART cohort.Although the underlying cause of the association between cola intake during pregnancy and PTB, such as the contained ingredients including caffeine, artificial sweeteners and the aluminium in cans, could not be speculated in this study.
In SC cohort, we found that cola intake was not significantly associated with either PTB or LBW, but the strength of association between ART and SC cohorts was not significantly different from each other.The lack of significant association between cola intake and birth outcomes in SC cohort may be due to the much lower incidence of PTB and LBW in this population as compared with ART women (PTB: 3.1% vs 7.3%; LBW: 4.1% vs 13.7%), which is likely a reflection of power issue.Advising SC women to reduce, if not avoid, the intake of cola during pregnancy is warranted unless evidence from future large cohort studies and clinical trials suggests the contrary.
The strengths of this study include the two comprehensive parallel cohorts (i.e.SC and ART cohorts), which enables a direct comparison of the effects of cola intake on birth outcomes in both populations.There are several limitations in this study.Firstly, SC women were recruited from the 1st trimester and did not have data on cola intake pre-pregnancy, which prevented us from evaluating the influence of cola intake pre-pregnancy.Secondly, there was a considerable rate of loss to follow-up (6.2%) in the ART cohort because ART women tended to have antenatal appointment at community hospital health centre instead of large general hospital (e.g. the First Affiliated Hospital of Anhui Medical University, where the ART cohort was performed), but there was still a total of 736 women included in this study.Thirdly, the frequency of cola intake was evaluated by a simple question.Omission of the type, brand and nutritional content of cola (e.g., diet, decaffeinated, sugar free, regular, coca cola or pepsi cola), duration of cola At least one trimester in all three trimesters refers to any cola intake was defined as at least one cola intake in at least one of the three trimesters, and no cola intake was defined as no cola intake in all three trimesters for participants who presented all three visits.RR, the ratio of probability of an outcome in cola exposure group to non-cola exposure group.CI, confidence interval.� , p value for the RR < 0.05.Multi-model 1 was adjusted for age, BMI at baseline, multiparity, gender of neonate, alcohol, tea, coffee, functional beverage, total intake energy, exercise frequency and the reasons of infertility.Multi-model 2 was further adjusted for cola intake pre-pregnancy above the multi-model 1. intake and amount of consumption in the selfreported questionnaire impeded more stratification analysis.Fourth, questionnaire did not assess the intake of other sugar-contained beverages (e.g.other carbonate beverages, juice or soda water), and these may have biased the association between cola intake and birth outcomes.Lastly, although this study included accessibly maximum number, the sample size was modest and formal sample size calculation was not conducted.Therefore, this study may be underpowered to observe a statistically significant association of cola intake and PTB and LBW.

Conclusions
Cola intake during pregnancy was associated with PTB in combined ART and SC women.Moreover, the association of cola intake with PTB and LBW among ART pregnancies was independent of cola intake pre-pregnancy and of other beverages intake, but the associations were not found in SC women.Our findings indicate that avoidance of cola intake during pregnancy may be beneficial for reducing adverse birth outcomes, especially for ART women, however, prospective studies with more information in terms of brand, quantity and type of cola are needed to confirm our findings.

Animal studies
Not applicable

Authors' contributions
SC and RH contributed to data collection, analysis and interpretation of data and drafted manuscript; XZ and GC contributed to analysis and interpretation of data and critical revision of manuscript; CL, HG, XX and HX contributed to data collection and quality control; FT and YC: project conception and study design; XP: project conception and study design and interpretation of data and critical revision of manuscript.ART, assisted reproductive technology.SC, spontaneously conceived.The consistency of cola intake in three trimesters refers how many trimesters of exposure cola in all three trimesters.RR, the ratio of probability of an outcome in cola exposure group to non-cola exposure group.CI, confidence interval.p value is for the RR.Multi-model 1 was adjusted for age, BMI at baseline, multi-parity, gender of neonate, alcohol, tea, coffee, functional beverage, total intake energy, exercise frequency and the reasons of infertility.� , only 12 ART women were exposed to cola in all three trimesters, and the birth weight of their neonates were heavier than 2500 g.

Table 1 .
Characteristics of study population in the 1st trimester.

Table 2 .
The association of cola intake with preterm birth and low birth weight in the spontaneously conceived women.

Table 3 .
The association of cola intake with preterm birth and low birth weight in the assisted reproductive technology women.

Table 4 .
The association of cola intake with preterm birth and low birth weight in the spontaneous conceived women.
RR, the ratio of probability of an outcome in cola exposure group to non-cola exposure group.CI, confidence interval.p value is for the RR.Multi-model 1 was adjusted for age, BMI at baseline, multi-parity, gender of neonate, alcohol, tea, coffee, functional beverage, total intake energy, exercise frequency and the reasons of infertility.

Table 5 .
The association between consistency of cola intake during pregnancy and preterm birth/low birth weight in the ART and SC cohort.