In this large prospective national birth cohort study, we did not find any statistically significant association between the amount of folate intake from the diet or supplements and spontaneous PTD in uncomplicated pregnancies. Folic acid supplementation starting more than 8 weeks before conception was associated with an increased HR for spontaneous PTD.
When interpreting the results, the selection of the study population has to be kept in mind: all known risk-pregnancies due to maternal disease, pregnancy complications or fetal malformation have been excluded from the analysis. There might be an association between the amount of folate intake from diet or supplements and spontaneous PTD in those pregnancies excluded.
Our results, demonstrating no significant protective effect of the maternal folate intake or folic acid supplementation on the spontaneous PTD risk, support a number of earlier observational studies [36–41] and RCTs [6, 42, 43]. A reanalysis of the most recent Cochrane review, based on data from 21 studies and one of the largest RCTs as well as a recent meta-analysis of all RCTs published to date, could not confirm the effect of the maternal folate status on the gestational length or risk of PTD [23, 24]. Shaw et al. found a comparable association of increased PTD risk and pre-conceptional folic acid supplementation when analyzing data from the US National Birth Defects Study . In addition, extensive supplementation with multivitamins with a major folic acid component was associated with an increased risk of PTD in a study by Alwan et al. .
However, some recent observational studies have found that folic acid supplementation reduces the risk of PTD [20–22]. In some cases, this association was linked to pre-conceptional folic acid supplementation for 1 year or longer  or third-trimester folic acid supplementation , raising questions about extended supplementation schemes compared to the NTD prevention scheme. A protective effect of folic acid supplementation was supported by a modest reduction in the PTD rate after the introduction of folate fortification of foods .
One of the most obvious explanations for these conflicting results could be the dosage of folic acid. While most of the studies finding an association with gestational length or PTD were based on comparably high doses of folic acid (≥5000 μg/d [20, 22, 44], ≥2500 μg/d [22, 45] and ≥500 μg/d [46–48]), very few women in our study population consumed as much as 5000 μg/d of supplemental folic acid, while only 9% consumed >500 μg/d and 15% consumed >400 μg/d. However, the Hungarian RCT, one of the biggest performed so far, did not find any effect of a high dosage of 8000 μg/d of folic acid supplementation on PTD . Unfortunately, the folic acid dosage was not indicated in all of the studies [21, 35].
The assessment of folate intake from supplementation alone or when studying populations with different dietary folate intakes are additional factors compromising comparability and generalizability between studies. While recent US studies are performed against the background of mandatory folate fortification of food [21, 41], other studies have examined supplementation effects in folate-deficient populations . Few studies have assessed the effects of both dietary folate and folic acid supplementation separately  or combined [20, 47, 49], and adjustments for bioavailability are rare. In this Norwegian study population, only 0.4% of the participants reached the Nordic Nutrition Recommendation of 500 μg/d with their dietary folate intake (adjusted with a factor of 0.6 for bioavailability; 2.8% without adjustment), and 31% of the participants achieved the recommended level with their total folate intake (39.6% without adjustment). After stratification for dietary folate intake, the early initiation of folic acid supplementation was significantly associated with spontaneous PTD in the subgroup of women with low but not high dietary folate intakes. There were no significant associations between high total folate intakes and PTD risk in the MoBa study population.
Confounding is always an issue when assessing the effect of a single environmental factor on a complex outcome like PTD. For example, it is well established that women with high levels of education, privileged socioeconomic status and healthier overall diets are more likely to use supplements during pregnancy [50, 51] and less likely to experience PTD than women without these characteristics. Some observational studies failed to adjust for these confounders, and the effect attributed to folic acid supplementation might in fact be confounded by overall health and lifestyle behaviors. While the strength of the significance was moderate, the association with the early onset of folic acid supplementation in the current study remained significant even after extensive adjustment for maternal characteristics such as socioeconomic and life-style parameters as well as obstetric anamnesis.
Associations with the early start of supplementation should be studied with particular caution. The early start of folic acid supplementation might partially identify a group of women with a history of adverse pregnancy outcomes who want to optimize conditions for their current pregnancy. As presented in Table 1, women who had previously experienced spontaneous abortions were more likely to initiate folic acid supplementation early in their subsequent pregnancies. However, the same association of the early initiation of folic acid supplementation and spontaneous PTD was found in primi-gravidae. Folic acid supplementation starting more than 8 weeks prior to conception might characterize women who planned a pregnancy but did not become pregnant during their first 2 cycles, thus constituting a subgroup of women with suboptimal fertility . The same association of the early start of supplementation was found in the group of women that became pregnant within the first month. Women who choose to start early with folic acid supplementation might be distinguished by some other characteristic that could be the causal link to spontaneous preterm delivery so that we cannot exclude confounding.
In addition to the amount of folic acid, the composition of supplements is another point of discussion. In some countries like Greece and Norway, commonly used supplements contain folic acid and/or iron only . In other countries, folic acid is mainly consumed in the form of multivitamins, making it difficult to differentiate the effects of multivitamin use and folic acid supplementation [21, 35, 41, 48]. Vitamins other than folic acid might explain the association between multivitamin use and PTD. Catov et al. found that in the Danish birth cohort, multivitamin use was associated with modestly decreased PTD rates, while there was no association with folic acid supplementation . As seen from Figure 1, vitamin A consumption (as a proxy for multivitamin supplementation) differed considerably from folic acid supplementation. However, the MoBa FFQ allowed us to calculate folic acid separately from other supplements, and adjusting for vitamin A consumption did not change the results for pre-conceptional folic acid supplementation.
Apart from the amount, timing and composition of folate exposure, differences in the definition of pregnancy outcomes hinder comparability. Most studies defined PTD as delivery at <37+0 weeks of gestation without indicating the range of gestational age. This information might be important, especially if the risk of early PTD is found to be associated with folate status, as suggested by this study and that of Bukowski et al. . Although PTD is a heterogeneous pregnancy outcome with distinct etiologies for different subgroups , not all studies analyzed clearly defined subgroups such as spontaneous PTD [20, 21, 37, 40, 47, 52].
Strengths and weaknesses
With a sample size of 65,668 pregnancies, this was a well-powered study for investigating the association of folate intake and pregnancy outcomes. Due to the large study sample, there were 1,628 cases defined as spontaneous PTD and 264 and 1,364 cases in the subgroups of early and late spontaneous PTD, respectively. The estimation of the gestational length by the second-trimester ultrasounds and the definition of a clear spontaneous PTD phenotype distinguishes this study .
The MoBa participation rate is 38.5%, and a demographic comparison with the MBRN in 2002 showed that single women and women <25 y of age are underrepresented in MoBa. Regarding PTD (7.2% in MoBa and 7.7% in MBRN), the differences are minor, and even the sub-group composition is similar to the distribution in the total population, with spontaneous PTD accounting for 42% of all PTD . Additionally, a recent study found no bias in 8 selected exposure-outcome associations .
The assessment of folate from both the diet and supplements is a clear strength of this study. Although all dietary assessment methods have limitations, the MoBa FFQ has been extensively validated in a sub-population of 119 MoBa participants using a 4-day weighed food diary and biological markers in the blood and urine as reference measures . The dietary supplement use was evaluated specifically. The total folate intake by the FFQ showed good agreement with the folate intake detailed by the food diary and was significantly reflected by the serum folate concentrations . In a subsample of an earlier MoBa version (2934 singleton pregnancies), Nilsen et al. did not find any significant associations of dietary folate intake, folic acid supplementation or plasma folate with PTD. This study also reported good agreement between the folate intake (dietary and supplements) by the MoBa FFQ and plasma folate concentration (r = 0.44, CI: 0.41-0.47) . As the relevant window of susceptibility for folate effects regarding pregnancy outcomes other than NTD is not yet known, the assessment of folate intake at different time points is a further strength of this study. The prospective design ensured that the women’s answers were not influenced by their knowledge of pregnancy outcomes.