Low birth weight, as a proxy for fetal growth, is associated with various chronic diseases later in life(Reference Barker and Clark1, Reference Roseboom, de Rooij and Painter2). Fetal growth depends on multiple genetic factors and environmental exposures, derived from the mother. In this respect, maternal nutrition during pregnancy has been shown to play a critical role(Reference Barker and Clark1, Reference Wu, Bazer and Cudd3, Reference Steegers-Theunissen4).
Because developing organ systems directly respond with permanent adaptations to the availability of nutrients during critical periods of rapid development, timing of adequate maternal nutrition is important to determine the effects both in the fetus and child(Reference Roseboom, de Rooij and Painter2, Reference Steegers-Theunissen4–Reference Hoet and Hanson6). Moreover, first evidence has suggested that fetal growth is vulnerable to maternal nutrition, especially during the preconception period and first weeks of gestation, since it has the potential to affect epigenetic mechanisms in the placenta and fetus(Reference Wu, Bazer and Cudd3, Reference Steegers-Theunissen4, Reference Steegers-Theunissen and Steegers7–Reference Heijmans, Tobi and Stein9). Folate is for this reason of great interest. Together with vitamin B12 it plays a critical role in homocysteine metabolism. The folate-dependent homocysteine pathway is important for protein, lipid and DNA synthesis. In addition, folate provides methyl groups for the synthesis of methionine and its derivate S-adenosyl-methionine. The latter is the most important methyl donor in the human body for DNA methylation, and represents one of the best known epigenetic mechanisms(Reference Tamura and Picciano10, Reference Bailey and Gregory11).
During pregnancy folate demand increases because of placental and fetal growth and development(Reference Tamura and Picciano10). For this reason pregnant women with a folate deficiency are at an increased risk for various reproductive failures, including neural tube defects and congenital malformations(12–Reference Steegers-Theunissen, Boers and Blom16). As a consequence, most European countries, including the Netherlands, recommend that women planning pregnancy use a folic acid supplement during the periconception period in addition to a healthy diet(Reference de Bree, van Dusseldorp and Brouwer17, Reference Botto, Lisi and Robert-Gnansia18). Several studies have shown positive associations between maternal folic acid intake and fetal growth(Reference Baumslag, Edelstein and Metz19–Reference Fawzi, Msamanga and Urassa27). However, the majority of these studies were focused on the effects of folic acid on fetal growth in mid- and late pregnancy. Only few studies assessed the associations in early pregnancy and failed to be consistent(Reference Rolschau, Kristoffersen and Ulrich23, Reference Ronnenberg, Goldman and Chen28). Thus, despite the fact that early pregnancy is the most important period for placental development, embryogenesis and fetal programming, relatively little is known about the implications of folic acid supplementation during this particular pregnancy period on fetal growth(Reference Steegers-Theunissen and Steegers7, Reference Waterland and Jirtle8, Reference Sinclair, Allegrucci and Singh29, Reference Rosenquist and Finnell30).
The aim of the present study was to investigate the effects of low-dose periconception folic acid supplementation on fetal and placental growth in low-risk singleton pregnancies. Also, we studied the associations between periconception folic acid intake and low birth weight, small for gestational age (SGA) and preterm birth.
The study was embedded in the Generation R Study, Rotterdam, the Netherlands, a population-based prospective cohort study from early pregnancy onwards. The Generation R Study was designed to identify early environmental and genetic determinants of growth, development and health from fetal life until young adulthood. The Generation R Study is conducted in Rotterdam, the second largest city in the Netherlands, and eligible women were those who were resident in the study area and delivered between April 2002 and January 2006. The study aimed to enrol women in early pregnancy (gestational age < 18 weeks), but enrolment was possible until the birth of the child. All midwifery practices and three hospitals located in the study area participated during the prenatal phase. The overall response rate was about 61 %, and is based on the number of children born to eligible mothers during the inclusion period. The Medical Ethics Committee of the Erasmus Medical Centre has approved the study, and all participants gave their written informed consent(Reference Jaddoe, Bakker and van Duijn31, Reference Jaddoe, Mackenbach and Moll32).
Folic acid use
Pregnant women were asked by questionnaire at enrolment (median 15·4 weeks of gestation, 95 % range 10·2–24·8) whether they had used a folic acid supplement periconceptionally (folic acid dosage of 0·4–0·5 mg/d, according to the advice of the Health Council of the Netherlands), and when supplementation was started(33, 34). Self-reported folic acid use was categorized into three groups: (1) preconception start: defined as preconception start of folic acid supplement intake at any moment prior to conception; (2) start before 8 weeks: defined as start of folic acid supplementation from the moment that pregnancy was recognized but before the eighth week of gestation; and (3) no use: defined as no use of folic acid supplementation at all. Because our interest was mainly focused on start, and use, of folic acid supplementation during the period just before and after conception, i.e. the periconception period, we did not include women who only started folic acid supplementation after the eighth week of gestation in our main analysis (n 46). However, because this group may present an interesting comparison group, we did include them in a sensitivity analysis. About 15 % of the women reported to have used folic acid as part of a multivitamin supplement regimen. The doses of folic acid in these multivitamins were approximately 0·4–0·5 mg/d.
Fetal ultrasound examinations were carried out in early pregnancy (median 13·5 week of gestation, 95 % range 10·6–17·5), mid-pregnancy (median 20·6 weeks of gestation, 95 % range 18·6–23·4) and late pregnancy (median 30·4 weeks of gestation, 95 % range 28·4–33·0). The ultrasound examinations were used for establishing both gestational age and fetal growth characteristics(Reference Verburg, Steegers and De Ridder35). Fetal biometry, including head circumference, biparietal diameter, abdominal circumference and femur length, was measured transabdominally during each ultrasound examination. Crown–rump length was measured in early pregnancy. Dating of pregnancy was performed using the first ultrasound measurement of either crown–rump length (gestational age until 12 weeks and 5 d of gestation, and crown–rump length measurement smaller than 65 mm), or biparietal diameter (gestational age from 12 weeks and 5 d of gestation onwards, and biparietal diameter larger than 23 mm)(Reference Verburg, Steegers and De Ridder35). Estimated fetal weight was calculated for mid- and late pregnancy using the formula of Hadlock et al. (Reference Hadlock, Harrist and Sharman36). Longitudinal growth curves and gestational age-adjusted standard deviation scores were constructed for all fetal growth measurements. These gestational age-adjusted standard deviation scores were based on reference growth curves from the whole study population, and represent the equivalent of z-scores(Reference Verburg, Steegers and De Ridder35).
Medical records completed by community midwives and obstetricians were used to obtain information about gestational age at birth (weeks), birth weight (g), gender and placental weight (g), and to calculate placental index (placental weight/birth weight)(Reference Bleker, Buimer and van der Post37, Reference Kloosterman38). Prematurity was defined as the birth of an infant before 37·0 weeks of gestation, and low birth weight was defined as a birth weight < 2500 g. Being SGA was defined as a standard deviation score ≤ − 2·3 ( < 5th percentile) and based on standard deviation curves derived from this cohort(Reference Verburg, Steegers and De Ridder35).
Information about maternal age, educational level, ethnicity, smoking and parity was obtained from the questionnaire at enrolment in the study. Educational level was assessed by the highest completed educational level of the mother and classified into two categories: (1) primary school; and (2) secondary school, university or college(Reference Jaddoe, Mackenbach and Moll32, 39). Ethnic background was defined from information from the country of birth of the woman herself and her parents and categorized as follows: (1) western, including Dutch, European, North American, and Oceanean; (2) Moroccan; (3) Turkish; (4) Antillean and Surinamese; and (5) other non-Western, including African, Asian, South American and Central American(Reference Jaddoe, Mackenbach and Moll32, 40). At the first ultrasound examination, maternal height (m) and weight (kg) were measured and BMI (in kg/m2) was calculated. Information on fertility treatment and pregnancy complications, including pre-eclampsia and gestational diabetes, was obtained from midwives and obstetricians. Women who became pregnant after fertility treatment, including in vitro fertilization and intracytoplasmatic sperm-injection, were excluded from analyses.
First, independent Student's t test and Mann–Whitney U test were used to test differences in characteristics between women who used periconception folic acid and women who did not use a folic acid supplement. Next, associations of folic acid supplementation with fetal growth characteristics (head circumference, abdominal circumference, femur length, estimated fetal weight) and birth characteristics (birth weight, placental weight) in low-risk singleton pregnancies were assessed using univariate linear regression models (model A). The linear regression analyses that were based on fetal growth characteristics (head circumference, abdominal circumference, femur length and estimated fetal weight) were restricted to those mothers who enrolled and had their pregnancies dated in early pregnancy (78 %). The consideration of confounding variables in our analyses was primarily determined a priori and based on earlier literature. These included time of enrolment in study, gestational age, maternal age, height, weight, parity, ethnicity, fetal gender, educational level, smoking, alcohol use, primary or secondary antenatal care, gestational diabetes and pre-eclampsia(Reference Kramer41, Reference Timmermans, Jaddoe and Mackenbach42). Potential confounders were selected as a result of exploratory analyses and were included in the analyses if the effect estimates of the fetal growth parameters changed more than 10 %. By using this approach, type of antenatal care, alcohol consumption, pre-eclampsia and gestational diabetes were not included in the final multivariate regression model (model B). To analyse the associations of folic acid supplementation with the risks of low birth weight, SGA and preterm birth, we used univariate and multivariate logistic regression models, with a similar approach to select confounders. Effect modification was tested by multiplying folic acid supplement use with the covariates educational level, ethnicity, smoking habits, parity and BMI. Subsequently, and under the condition of a P value < 0·1, multivariate linear regression analyses were performed in strata of that specific determinant. Last, to test any possible associations of folic acid supplementation started after the eighth week of gestation with fetal growth, we also performed similar analyses with the cohort categorized into four groups, namely (1) preconception start; (2) start before 8 weeks; (3) start from 8 weeks onwards; and (4) no folic acid use. The size of the effect estimates (regression coefficients and OR) are presented with their 95 % CI. Statistical analyses were performed using the Statistical Package of Social Sciences version 15.0 for Windows (SPSS Inc., Chicago, IL, USA).
Between 2002 and 2006, 8880 women enrolled during pregnancy in the Generation R Study(Reference Jaddoe, Mackenbach and Moll32). Forty-six women (0·5 %) started using a folic acid supplement after the eighth week of gestation and information on folic acid supplementation was missing in 25 % of the women (n 2274). Of the remaining 6560 women, forty-seven in vitro fertilization or intracytoplasmatic sperm-injection pregnancies, sixty-three twin pregnancies and seventy-seven fetal deaths were identified, and twenty women were lost to follow up. Data on 6353 low-risk singleton pregnancies (71·6 %) were available.
Characteristics of the women per folic acid category are presented in Table 1. Of all women 39·2 % (n 2493) started folic acid supplementation preconceptionally. Approximately 31 % of all women (n 1983) started folic acid supplementation after pregnancy recognition, and 29·5 % of the women (n 1877) did not use folic acid supplementation at all. The age of the cohort ranged from 15·3 to 46·3 years with a median of 29·8 years, and the lowest median age was found in women who did not use folic acid (27·8 years). The percentage of women with a higher educational level was also lowest in this group (26·9 %). In the whole cohort the largest ethnic groups were the Dutch and other western (63 %). The percentage of women who did not use folic acid was highest among non-western women, including Moroccan, Turkish, Surinam and Antillean women (total of 71·6 %).
* Median with 95 % range.
The associations between folic acid use and fetal growth characteristics in mid- and late pregnancy are presented in Table 2. Periconception folic acid supplementation was associated with trends towards significantly larger head circumference and abdominal circumference, in mid- and late pregnancy, compared to women who did not use folic acid. In addition, similar trends towards larger femur length in mid- and late pregnancy were observed for women who periconceptionally used a folic acid supplement, compared to women who did not use folic acid. However, these effect estimates were not significant.
RC, regression coefficient; Ref., reference.
* For details of subjects and procedures, see Methods and Table 1. Results are from linear regression analysis. Values are regression coefficients (in mm) and reflect the difference in growth for each characteristic measured in mid-pregnancy (median 20·5 weeks, 95 % range 18·8–23·1) or late pregnancy (median 30·4 weeks, 95 % range 28·5–32·7), compared to no folic acid use.
† Adjusted for gestational age at time of ultrasound measurement.
‡ Adjusted for gestational age at time of ultrasound measurement, maternal age, height, weight, parity, ethnicity, fetal gender, educational level and smoking.
Figure 1 shows the differences in standard deviation scores (z-scores) of fetal weight from mid-pregnancy until birth between the three folic acid categories. No significant differences in standard deviation scores between the folic acid supplementation groups were observed for estimated fetal weight in mid-pregnancy. This effect changed over time with significant higher estimated fetal weight in late pregnancy for women who preconceptionally started folic acid supplementation (sd 0·10, 95 % CI 0·02, 0·19), and significant higher birth weight for both women who preconceptionally started folic acid supplementation and women who started after pregnancy recognition, compared to no folic acid supplementation at all (sd 0·16, 95 % CI 0·09, 0·23 and sd 0·13, 95 % CI 0·06, 0·20, respectively). After adjustment for potential confounders birth weight was 68 g higher in women who preconceptionally started folic acid supplementation (95 % CI 37·2, 99·0) and 53 g higher in women who started supplementation after pregnancy recognition (95 % CI 23·60, 83·18), compared to birth weight of newborns of women who did not use folic acid supplementation.
Table 3 shows the associations between folic acid use and placental weight. Placental weight was approximately 10 g more in women who preconceptionally started using a folic acid supplement (95 % CI 0·32, 20·49) compared to placental weight of women who did not use folic acid supplementation. After adjusting for potential confounders this effect estimate did not change. For placental index, a trend was observed towards a smaller index in both women who used preconception folic acid (β − 0·004, 95 % CI − 0·007, − 0·002), as well as in women who started after pregnancy recognition but before the eighth week of gestation (β − 0·003, 95 % CI − 0·006, − 0·001), compared to women who did not use folic acid. This trend, however, was not significant anymore after adjusting for potential confounders.
RC, regression coefficient; Ref., reference.
* For details of subjects and procedures, see Methods and Table 1. Results are from linear regression analysis. Values are regression coefficients and reflect the difference for each characteristic (placental weight in g) compared to no folic acid use.
† Adjusted for gestational age at birth (placental weight).
‡ Adjusted for gestational age at birth (placental weight), maternal age, height, weight, parity, ethnicity, fetal gender, educational level and smoking.
Table 4 shows the associations between folic acid use and birth outcomes. Preconception start of folic acid was associated with a decreased risk of low birth weight (OR 0·47, 95 % CI 0·33, 0·68) as well as a decreased risk of SGA (OR 0·38, 95 % CI 0·23, 0·63), compared to women who did not use folic acid supplementation. After adjustment for potential confounders these effect estimates did not further change. In addition, after adjustment for potential confounders, start of folic acid supplementation after pregnancy recognition was also associated with a decrease of having a child with low birth weight (OR 0·61, 95 % CI 0·40, 0·94), compared to women who did not use a folic acid supplement. Folic acid supplementation (either preconception start, or start after pregnancy confirmation) was not significantly associated with the risk of preterm birth after controlling for confounders.
Ref. reference; SGA, small for gestational age.
* For details of subjects and procedures, see Methods and Table 1. Results are from logistic regression analysis.
† Adjusted for gestational age at birth (not prematurity).
‡ Adjusted for gestational age at birth (not prematurity), maternal age, height, weight, parity, ethnicity, fetal gender, educational level and smoking.
Parity significantly modified the effect of periconceptional folic acid use on birth weight (P < 0·001) (Fig. 2). After adjustment for potential confounders, preconception start of folic acid among multiparous women was associated with about 240 g higher birth weight (95 % CI 195·13, 282·92), compared to nulliparous women who did not use a folic acid supplement. In contrast, preconception start of folic acid among nulliparous women was about 55 g higher (95 % CI 15·44, 94·68), compared to nulliparous women who did not use a folic acid supplement. We did not observe further significant effect modification on the multiplicative scale by educational level, ethnicity, smoking or BMI.
Last, the analysis performed to test possible associations of folic acid supplementation started after the eighth week of gestation with fetal growth, did not reveal any differences between (1) non-users, preconception users and women who started before eight weeks, and (2) folic acid supplementation from 8 weeks onwards, and similarly, no association of this latter group with fetal growth (likely due to small numbers in this group; n 46).
In this prospective population-based cohort study we demonstrate that low-dose periconception folic acid supplementation is associated with increased fetal growth compared to non-users, and that parity significantly modifies the effect. Low-dose periconception folic acid supplementation is also associated with reduced risks of having a child with low birth weight or being SGA at birth. No significant association was observed between periconception folic acid supplementation and preterm birth.
Folate is a B vitamin that serves as a substrate in many underlying pathways of cellular processes, including cell multiplication, apoptosis, intracellular signalling and programming, which are all implicated in fetal and placental growth and development(Reference Tamura and Picciano10, Reference Bailey and Gregory11). Therefore, additional supply of folic acid during pregnancy is likely to influence fetal growth. Furthermore, the placental function is of main importance for the growth and development of the fetus. Thus, when folic acid directly affects placental growth and development, it also indirectly affects fetal growth(Reference Steegers-Theunissen and Steegers7, Reference Iyengar and Rajalakshmi21, Reference Bleker, Buimer and van der Post37, Reference Kloosterman38). Previous studies support the present findings and show positive associations between increased folic acid intake, higher birth weight, and reduced risks of low birth weight and SGA(Reference Baumslag, Edelstein and Metz19–Reference Fawzi, Msamanga and Urassa27). However, the majority of these studies were focused on increased folic acid intake in mid- and late pregnancy. Since fetal growth is greatest in the second half of pregnancy, and a larger supply of folic acid during this particular period may directly serve as a substrate for increased cellular synthesis, their results were not really unanticipated(Reference Cunningham, Gant, Leveno, Cunningham, Gant and Leveno43). Only a few studies assessed the associations between folate in early pregnancy and fetal growth(Reference Rolschau, Kristoffersen and Ulrich23, Reference Ronnenberg, Goldman and Chen28). Rolschau et al. (Reference Rolschau, Kristoffersen and Ulrich23) found an increased birth weight in infants of mothers who preconceptionally started folic acid supplementation. Interestingly, they reported this effect only to be present in week 43 of gestation. They also reported reduced prevalences of low birth weight and SGA in newborns of women who preconceptionally started supplementation and, in contrast with the present study, a reduced risk for preterm birth. However, like most other studies that investigated associations between folic acid supplementation and fetal growth, they used very high doses of folic acid (up to 2·5 mg), compared with the low doses of 0·4–0·5 mg/d in the present study(Reference Goldenberg, Tamura and Cliver20, Reference Rolschau, Date and Kristoffersen22, Reference Rolschau, Kristoffersen and Ulrich23, Reference Tamura, Goldenberg and Freeberg25, Reference Neggers, Goldenberg and Tamura26, Reference Palma, Perez-Iglesias and Prieto44).
The association between periconception folic acid supplementation and fetal growth could also be explained by optimalization of the folate-dependent homocysteine pathway. Reduced folate status is associated with elevated homocysteine and higher serum levels of homocysteine have been associated with decreased fetal growth and placental vasculopathy(Reference van der Molen, Verbruggen and Novakova45). In most cases hyperhomocysteinaemia can be treated by low-dose folic acid supplementation. Additionally, it has been shown that folate has the potential to improve endothelial function independent from homocysteine status(Reference Nafee, Farrell and Carroll46).
Previous studies showed that folic acid supplementation of 5 mg in mid- and late pregnancy increased placental size, cell number (DNA) and protein content(Reference Iyengar and Rajalakshmi21, Reference Rolschau, Date and Kristoffersen22). In the present study women who preconceptionally started taking a folic acid supplement did not only have significantly larger newborns but larger placentas as well, compared to women who did not use folic acid. However, this association was not found for women who started folic acid supplementation after pregnancy recognition. The present findings could implicate that periconception low-dose folic acid influenced placental growth via other pathways rather than through improved placental vasculogenesis or its role in placental nucleic acid synthesis.
The results as shown in the present study on periconception folic acid supplementation might suggest that fetal and placental programming are affected by additional folic acid supply at different time frames throughout the periconception period. Folate is essential for DNA methylation, an important epigenetic mechanism that plays a regulatory role in genome programming and imprinting during pregnancy(Reference Tamura and Picciano10, Reference Bailey and Gregory11). Differences in quantitative methylation may affect genes implicated in embryogenesis and fetal growth(Reference Waterland and Jirtle8, Reference Nafee, Farrell and Carroll46, Reference Waterland and Michels47). It has been shown that variations in preconception exposure to folic acid can lead to epigenetic modifications of the genome in the offspring associated with adiposity, insulin resistance and high blood pressure in adulthood(Reference Waterland and Jirtle8, Reference Sinclair, Allegrucci and Singh29). This might also apply to the present results, suggesting that periconception folic acid supplementation may cause epigenetic modifications in the preimplantation embryo which may result in increased placental and fetal growth patterns(Reference Waterland and Jirtle8, Reference Pennisi48). However, at this moment these speculations need to be studied in further detail.
An important finding from the present analysis was the modifying effect of parity. The positive effect of parity on birth weight has been well established(Reference Bleker, Buimer and van der Post37, Reference Kloosterman38, Reference Kloosterman49, Reference Kloosterman and Huidekoper50). However, to our knowledge, the effect modification by parity on the association between periconception folic acid use and fetal growth has not been reported before. In the past, Kloosterman suggested that multiparous women offer, through remodelling of the maternal vascular structure in former pregnancies, a more favourable environment for placental development and function in subsequent pregnancies(Reference Bleker, Buimer and van der Post37, Reference Kloosterman38, Reference Kloosterman49). From this respect it can be hypothesized that periconception folic acid supplementation interacts with these vascular remodelling processes in multiparous women, thereby affecting placental and subsequent fetal growth. However, this needs to be studied further by other investigators.
The present study was embedded in a large prospective cohort study with a significant number of measurements performed in the mothers, which increases the accuracy of our effect estimates. However, some limitations should be addressed. First, because the response rate of the Generation R Study was approximately 61 %, selective participation of pregnant women may have influenced the observed associations(Reference Timmermans, Jaddoe and Mackenbach42, Reference van Delft and Ketting51, Reference Hoggatt52). In addition, complete information on folic acid supplementation was missing in approximately 25 % of the women, and this non-response could lead to selection bias if the association of periconception folic acid supplementation with fetal growth would differ between those with and without complete data. Even though this seems unlikely, it cannot be fully excluded.
The use of questionnaires to assess folic acid supplementation encompasses several limitations, including information bias and giving desirable answers. Yet, studies show that self-reported intake of folic acid correlates to folate serum levels(Reference Brantsaeter, Haugen and Hagve53). Moreover, we aimed to assess folic acid supplementation in early pregnancy to minimize possible recall bias. However, even though we used a clear definition of folic acid supplementation in the questionnaire misclassification, especially between those who started using a folic acid supplement preconceptionally and those who started after pregnancy recognition, should always be considered. Last, folic acid supplement use is related to socioeconomic status, lifestyle factors (dietary habits) and adverse health behaviours (smoking). Even though we were able to control for a large number of potential confounders available from our questionnaires, residual confounding is always of particular concern in vitamin supplement studies, and should for this reason be taken into account(Reference Bendich, Bendich and Butterworch54). For these reasons, the present findings would be much stronger if we had had availability of maternal, placental or neonatal folate biomarkers such as erythrocyte folate levels, or global methylation status.
In conclusion, periconception folic acid supplementation is significantly associated with increased fetal growth resulting in higher placental and birth weight, and decreased risks of having a child with low birth weight or being SGA. The effects are most pronounced in women who preconceptionally start using a folic acid supplement, and are modified by parity. To investigate the underlying pathways in more detail and possible consequences for postnatal growth and development, future studies are necessary.
The Generation R Study is conducted by the Erasmus Medical Centre in close collaboration with the School of Law and Faculty of Social Sciences of the Erasmus University Rotterdam, the Municipal Health Service Rotterdam area, Rotterdam, the Rotterdam Homecare Foundation, Rotterdam and the Stichting Trombosedienst & Artsenlaboratorium Rijnmond (STAR), Rotterdam. We gratefully acknowledge the contribution of general practitioners, hospitals, midwives and pharmacies in Rotterdam and, in particular, Frank Sleutels (PhD), of the Erasmus Medical Center, Department of Cell Biology, for his interpretation of the data analysis. The Erasmus Medical Centre Rotterdam, the Erasmus University Rotterdam and the Netherlands Organization for Health Research and Development (ZonMw) financially supported the first phase of the Generation R Study. There are no personal or financial conflicts of interest. The contributions of each author were as follows: study concept and design: S. T., V. W. V. J., A. H., R. P. M. S.-T., E. A. P. S.; acquisition of data: S. T., V. W. V. J., A. H., R. P. M. S.-T., E. A. P. S.; analysis and interpretation of data: S. T., V. W. V. J., R. P. M. S.-T., E. A. P. S.; drafting of the manuscript: S. T., V. W. V. J., R. P. M. S.-T.; critical revision of the manuscript for important intellectual content: S. T., V. W. V. J., A. H., R. P. M. S.-T., E. A. P. S.; statistical analysis: S. T., V. W. V. J., R. P. M. S.-T., E. A. P. S.; obtained funding: V. W. V. J., A. H., E. A. P. S.; administrative, technical or material support: A. H., E. A. P. S.; supervision: V. W. V. J., A. H., R. P. M. S.-T., E. A. P. S.