Maternal intake of seafood and supplementary long chain n-3 poly-unsaturated fatty acids and preterm delivery

Preterm delivery, which is defined as spontaneous or iatrogenic delivery before gestational week 37, is the major cause of perinatal mortality and morbidity and is an important risk factor of long-term physical and mental disabilities [1‐4]. In Scandinavia and some other European countries, the rate is around 5–7% of all deliveries, while in the United States it is as high as 12% [5].

Preterm delivery accounts for a high financial burden on healthcare and is a considerable trauma for those involved [6‐8]. Several factors have been shown to be associated with preterm delivery, including maternal demographic characteristics, reproductive history, infection, and biological and genetic markers [1, 9‐11]. However, the aetiologies for preterm delivery are largely unknown and currently there is no effective treatment to reduce the rate of preterm delivery. Hence, it is important to identify potential modifiable factors in order to prevent the complications and cost associated with preterm delivery.

Seafood is a rich source of essential nutrients including protein, selenium, iodine, vitamin D, and the marine long chain n-3 polyunsaturated fatty acids (LCn-3PUFA) which have important structural and physiological roles in the body, including neurological, immune, and cardiovascular systems [12‐14].

In prospective observational studies, high levels of maternal fish consumption during pregnancy have been associated with longer gestation [15‐17] and lower prevalence of preterm delivery [16, 18], but the results are not found in all studies [19‐21]. The beneficial effects of fish consumption have primarily been attributed to the LCn-3PUFA eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA). Randomized controlled trials have shown lower risk of preterm delivery, and particularly early preterm delivery (<34 weeks) in women supplemented with EPA and DHA during pregnancy [22‐25].

The current dietary advice to pregnant women in Norway and other countries is to include lean and fatty fish as part of a balanced diet and to limit or avoid consumption of contaminated species [26‐28]. In Norway there is a long tradition not only for eating seafood, but also for use of cod liver oil. More studies of maternal seafood and LCn-3PUFA intake in relation to preterm delivery are needed to disentangle, if possible, the role of different types of fish and supplementary LCn-3PUFA [29, 30]. In the Norwegian Mother and Child Cohort Study, women reported in detail their intakes of food and dietary supplements during pregnancy, making it possible to quantify their total seafood consumption and subcategories of lean and fatty fish, as well as LCn-3PUFA contributed by use of dietary supplements and LCn-3PUFA contributed by food (i.e., fish) [31‐34].

The aim of the present study was to examine associations of maternal seafood and LCn-3PUFA supplement intakes with the risk of preterm delivery. We hypothesized that higher intake of seafood is associated with lower risk of preterm delivery and that associations vary by seafood categories. We investigated the associations with all preterm deliveries and with the outcome stratified as spontaneous and iatrogenic preterm delivery and as early, moderate and late preterm delivery.

Consumption of any seafood was reported by 97.7% of all women in this study, 5.7% reported no lean fish, 13% reported no fatty fish, 10.7% reported no shellfish and 91% reported no consumption of fish liver. The median intakes of total seafood, lean fish and fatty fish were 33.4, 18.5 and 8.3 g/day, respectively. Lean and fatty fish constituted on average 56 and 34% of the total seafood intake (Table 1). The median total seafood intake corresponds to 1–2 servings per week. The median amount of LCn-3PUFA from total seafood was 0.25 g/d (mean 0.37 g/d), with median 0.039 g/d from lean fish (mean 0.044 g/d)) and 0.188 g/d from fatty fish (mean 0.298 g/d). On average, fatty fish contributed 75% and lean fish 11% to the LCn-3PUFA intake from food (Table 1). Use of LCn-3PUFA supplements was reported by 67% of the women. The supplement users were divided into two groups based on the median intake of supplemental LCn-3PUFA (<0.30 g/d and ≥0.30 g/d). In the low group, LCn-3PUFA intakes ranged from 0.01 g/d-0.29 g/d (median 0.16 g/d) and in the high group from 0.30 g/d-8.8 g/d (median 0.80 g/d).

Maternal characteristics differed across increasing categories of seafood intake and between LCn-3PUFA non-supplement and supplement users (Table 2). Maternal age and energy intake increased with increasing seafood intake, while BMI decreased. Women in the lowest consumption categories included more nulliparous women and more women with low education and income than to those in the higher consumption categories. Increasing seafood intake was associated with not only increasing intakes of fish and shellfish, but also by increasing intakes of LCn-3PUFA from food and from supplements. LCn-3PUFA supplement users had lower BMI, included fewer first time mothers and fewer smokers, and included more women with high education and income than non-supplement users (Table 2).

The overall proportion of preterm delivery in the study population was 5.4% (n = 3630), comprising 3.1% (n = 2051) spontaneous and 2.2% (n = 1419) iatrogenic preterm deliveries, while information about delivery initiation was missing for 88 preterm cases (0.1%). Of all deliveries, 2659 (4.0%) were late preterm deliveries, 491 (0.7%) were moderately preterm and 480 (0.7%) were early preterm deliveries (Table 3). In crude analyses using continuous seafood intake variables, women with preterm deliveries had lower intakes of total seafood, lean fish and fatty fish than women with term deliveries. This was also found for LCn-3PUFA from food, while there was no difference in the amount of LCn-3PUFA from supplements between the groups. The differences in seafood intakes were seen particularly for the subcategories spontaneous and iatrogenic preterm delivery and for late preterm delivery (Table 3).

Examining seafood intake by categories with the never/rarely as the reference category, showed lower prevalence of preterm delivery for all other intake categories (Table 4). For total seafood intake, the lowest risk estimates were observed in the two highest intake categories with adjusted HR: 0.72 (95%CI: 0.62, 0.83) for 2- ≤ 3 servings per week and HR: 0.72 (95%CI: 0.61, 0.85) for ≥3 servings per week, p-trend <0.001 (Table 3). When lean and fatty fish were examined as separate variables, significantly lower risk for preterm delivery was observed for lean fish intake in all intake categories except the highest category (p-trend: 0.005). For fatty fish, lower risk was seen only for the intake category corresponding to 1- ≤ 2 servings per week (p-trend 0.411). LCn-3PUFA from supplements was not associated with overall preterm delivery (Fig. 2 and Table 4).

We further examined whether the associations between LCn-3PUFA from supplements and preterm delivery differed between groups of women with low and high seafood intake. Low seafood intake was defined as no seafood intake and intakes below the 5th percentile. Total seafood intake was also ranked into tertiles, quintiles and deciles, and the association between LCn-3PUFA from supplements and preterm delivery examined in each strata separately. No significant association between LCn-3PUFA from supplements and preterm delivery was observed at any level of seafood intake (results not shown). Furthermore, we examined LCn-3PUFA from supplements as a continuous variable and ranked into quintiles and deciles. No significant associations with preterm delivery were indicated (results not shown).

We repeated the analysis using the calculated LCn-3PUFA intake from food as the exposure and the lowest quintile as reference category. Lower risk estimates were seen in all the other quintiles (Additional file 1: Table S1), with the lowest HR observed in quintile 4 (HR 0.83, CI: 0.75, 0.92), p-trend 0.002. However, when we summed the intake of LCn-3PUFA from food and supplements, the association with preterm delivery was significant only in the third quintile (Additional file 1: Table S1). When LCn-3PUFA from food was included in a model with seafood as the exposure, the hazard ratios for seafood and preterm delivery remained significant, but with wider confidence intervals (i.e., for 2–3 servings/week HR: 0.77 (95%CI: 0.64, 0.92) and for ≥3 servings/week HR: 0.75 (95%CI: 0.62, 0.92).

We further analysed associations between seafood intake and subcategories of preterm delivery. Any intake of seafood beyond no/rare intake was associated with lower prevalence of both spontaneous and iatrogenic preterm deliveries (Table 5), with HRs similar to those seen for any preterm (20–30% risk reduction). In these analyses, lean fish had a stronger influence on the association for iatrogenic than for spontaneous preterm delivery (p-trend: 0.012 for iatrogenic vs 0.219 for spontaneous), while fatty fish and LCn-3PUFA from supplements did not show a significant trend with either spontaneous or iatrogenic preterm delivery.

When preterm delivery was examined separately according to late, moderate or early onset, any increase in seafood beyond no/rare intake was associated with lower risk in the late preterm group, with HRs resulting in 20–30% risk reduction (p-trend <0.001) (Table 6). Although not statistically significant, the risk estimates for moderately and early preterm delivery were comparable to those for late preterm delivery. In these analyses LCn-3PUFA from supplements was not associated with late or moderately preterm delivery, while a non-significant trend was indicated for early preterm delivery, HR 0.84 (CI: 0.67, 1.05) and HR 0.81 (CI: 0.65, 1.01) for the two supplement intake groups versus no LCn-3PUFA from supplements (p-trend 0.072) (Table 6).

Statistical significance of the adjusted associations was similar whether the seafood and LCn-3PUFA variables were modelled as continuous or categorical variables.

Obesity is a risk factor of preterm delivery, and we examined the associations between seafood intake and preterm delivery separately in women with pre-pregnant BMI <25 (n = 45,119) and those with BMI ≥ 25 kg/m² (n = 20,165). The prevalence of preterm delivery was 5.0% in the group with BMI <25 and 6.3% in the groups with BMI ≥25 kg/m². Lower risk of preterm delivery with seafood intake above 1 serving a week (20-30% risk reduction) was seen in both strata, although slightly stronger HRs were observed in women with BMI <25 than in those with BMI ≥25 kg/m² (Additional file 2: Table S2).

In our study sample, preterm delivery was more prevalent in nulliparous (6.3%) than in parous women (4.3%), and we also examined the associations between seafood intake and preterm delivery separately in nulliparous (n = 34,731) and parous women (n = 32,276). Lower HRs for preterm delivery were seen in both strata, but the effect size was smaller and the confidence intervals wider in parous than in nulliparous women (Additional file 2: Table S2). LCn-3PUFA from dietary supplements was not associated with preterm delivery in any of these analyses (Additional file 2: Table S2 and Additional file 3: Table S3).

Finally we tested whether the associations between seafood intake and preterm delivery were consistent in subcategories of maternal age, education and smoking. The risk estimates were comparable in all sub-strata, but in small groups e.g., smokers, the confidence intervals were wider (results not shown).

The main finding in the present study was the significant association between maternal total seafood intake and lower likelihood of preterm delivery. The association was mainly explained by intake of lean fish. The results for the calculated intake of LCn-3PUFA from food reflected the results for seafood intake, while LCn-3PUFA from supplements was not associated with the outcome, with the exception of a borderline significant trend for early preterm delivery. Lower prevalence of preterm delivery with increasing seafood intake was observed for both the subcategories spontaneous and iatrogenic preterm delivery. Significantly lower HR was observed for moderately increased intakes relative to no/low intake, while we observed no additional risk reduction for the highest intake categories.

A beneficial relationship between seafood consumption and preterm delivery has been reported in several studies [15‐18, 45]. In a previous study in nulliparous women in MoBa focusing on the Mediterranean diet, we found that eating fish at least twice weekly during pregnancy was associated with lower risk of preterm delivery [46]. We defined Mediterranean diet according to five a-priori defined criteria, and associations with preterm delivery were examined for adherence to each criterion and for all criteria combined. Only the criterion ‘eating fish at least twice weekly’ was significantly associated with the outcome (adjusted OR: 0.84; 95% CI: 0.74, 0.95) [46]. A more recent study from MoBa examining maternal dietary patterns showed a reduced risk of preterm delivery associated with high adherence to a traditional dietary pattern characterised by i.e., potatoes, fish dishes and lean fish [47]. The present study complements the previous studies by including more detailed quantification of seafood intakes, by separating between lean and fatty fish intakes, and by examining LCn-3PUFA from food and supplements as independent exposure variables.

Associations between fish consumption and LCn-3PUFA with increased length of gestation observed in previous studies have been explained by the anti-inflammatory properties of these fatty acids modulating the inflammatory pathways leading to cervical ripening and initiation of labour and delivery [48‐51]. Klebanoff et al. (2011) studied fish consumption, erythrocyte fatty acids, and preterm delivery in a high-risk population of women with a prior preterm delivery. The women participated in a randomized controlled trial of n-3 supplementation [52]. Women were randomized to receive either an LCn-3PUFA supplement or placebo starting in mid-gestation (weeks 16–21). The results showed that women who reported the lowest fish consumption at gestational weeks 16–21 had higher risk of recurrent preterm birth than those who ate fish more frequently. The lowest occurrence of preterm birth was seen among women who ate fish approximately 2–3 times/week, while more frequent fish consumption was not associated with further risk reduction [52]. Interestingly, the lowest occurrence of preterm birth was observed among women in the second quartile of erythrocyte LCn-3PUFA concentrations, and no benefit of LCn-3PUFA supplementation was found, regardless of baseline fish consumption and erythrocyte n-3 concentration [52]. A multicentre randomized control trial in seven European countries comprising women with previous pregnancy complications found that fish oil supplementation (2.7 g LCn-3PUFA daily) delayed the onset of delivery for women with low (mean 16 g/d) and medium fish intake (mean 23 g/day), but not for women with high fish intake (mean 36 g/day) at baseline [53], suggesting that the effect of fish oil supplementation on timing of delivery depends on a woman’s fish intake. In the current study, however, no association between LCn-3PUFA from supplements and overall preterm delivery was evident at any level of seafood intake. However, our finding of borderline significantly lower risk of early preterm delivery (<32 weeks) in women using LCn-3PUFA is in line with results from experimental studies in the US and Australia which found that supplementation with 0.6–0.8 g/d DHA reduced the prevalence of early preterm births in low risk pregnancies [22, 23]. Contrary to our findings, a prospective study in pregnant women in Massachusetts found no associations between intake of fish or LCn-3PUFA from food and preterm delivery [20]. Furthermore, another recent study from the US reported an increased risk of preterm in women with high intakes of lean fish [21].

In the present study, LCn-3PUFA from fish paralleled the results obtained for seafood, while LCn-3PUFA from supplements was not associated with overall preterm delivery. The association between seafood intake and preterm delivery remained significant also when adjusted for LCn-3PUFA from food, indicating that the observed associations between seafood consumption and lower prevalence of preterm delivery cannot be explained only by LCn-3PUFA. Alternative explanations could be other components in fish (e.g., proteins, iodine, and selenium) or other aspects of fish consumption (e.g., foods typically eaten with fish or foods displaced by fish) might modulate the inflammatory pathways leading to delivery. The effects of fatty fish, lean fish and LCn-3PUFA intake on inflammation have been studied in some intervention studies [54, 55]. In one study, participants were randomly allocated to receive dietary advice plus either 300 g of fatty fish (salmon) or 300 g of lean fish (cod) per week for 6 months, or only dietary advice. Interestingly, the effect estimates did not differ between the lean and fatty fish groups, and a significantly lower concentration of the systemic inflammation marker C-reactive protein was found in both fish groups compared to the control group [54]. Another study investigated the effects of weight loss and seafood consumption in three intervention groups (salmon, cod, or LCn-3PUFA supplementation) and controls on inflammation parameters during energy restriction. The largest decrease in inflammation parameters was observed for salmon consumption, and no decrease was seen for LCn-3PUFA supplementation [55]. It could also be speculated that fish consumption may influence the gut microbiota which is crucial for optimal functioning of the digestive and immune systems [56]. Two experimental studies explored whether fish consumption influenced gut microbiota composition and local markers of gut inflammation without convincing results [54, 57]. However, in one of these studies a significant effect on the systemic inflammatory markers was observed for both lean and fatty fish [54].

An issue of particular concern when it comes to seafood consumption in women of childbearing age, is the concomitant exposure to environmental pollutants, e.g., methylmercury, polychlorinated biphenyls and perfluorinated alkylated substances, which may counteract the beneficial effects of fish consumption [28, 58‐61]. However, a systematic review of environmental contaminant exposures and preterm birth found no consistent evidence for chemical exposure and increased risk of preterm delivery [62]. Likewise, studies in MoBa examining associations between maternal exposure to environmental pollutants primarily contributed by seafood and pregnancy outcomes showed no increased risk of preterm delivery [63, 64]. The result of the present study corroborates the current advice to pregnant women to include fish and seafood as part of a balanced diet and restricting the intake of species and items with known high concentrations of environmental pollutants [26, 65].

The strengths of this study include the prospective design and detailed information about maternal diet, demography, socioeconomic factors, and pregnancy outcomes. Thanks to the large sample size we were able to examine the associations between seafood intake and preterm delivery in subcategories. Participants in MoBa were recruited from urban and rural, coastal and inland regions and represented different socioeconomic groups. Dietary intake was assessed using an FFQ that was specifically developed and validated for use in this cohort [31, 32]. The FFQ included questions with special emphasis on various seafood items, and the validation study showed that the reported intakes of lean and fatty fish and LCn-3PUFA supplements were reflected by blood and urine samples [33]. The food frequency method challenges respondents with rather complex cognitive tasks and are more suitable for ranking participants according to high and low intakes than for precise intake calculations. Therefore, differential misclassification of seafood and n-3 supplement intakes cannot be excluded, but is as likely to occur in women with term delivery as in women with preterm delivery.

The somewhat low response rate in MoBa is a concern and participants are not representative of all pregnant women in Norway. The potential bias due to self-selection has been evaluated by comparison of eight exposure-outcome associations reported both in MoBa and in the MBRN. Significant differences in prevalence estimates were found for most variables, e.g., MoBa participants included less smokers than the general pregnant population. In spite of this, there were no statistically relative differences in association measures, e.g., prenatal smoking and low birthweight [66]. It is a strength that the women who do participate remain in the study, illustrated by a response rate of more than 90% for the three extensive questionnaires answered during pregnancy [39]. However, the study is observational and we cannot rule out the possibility that residual or unmeasured confounding may still exist.

This study showed that maternal seafood consumption was associated with reduced risk of preterm delivery. The association was mainly explained by intake of lean fish. The association was seen for subcategories spontaneous, iatrogenic and late preterm delivery. Intake of LCn-3PUFA from seafood was associated with preterm delivery, while intake of LCn-3PUFA from supplements was associated only with early preterm delivery. The association between supplementary LCn-3PUFA and preterm delivery did not differ between women with low or high seafood intake. Furthermore, the association between seafood intake and reduced risk of preterm delivery remained significant when adjusted for LCn-3PUFA from food, indicating that the observed associations cannot be explained only by LCn-3PUFA and that other properties related to fish and seafood consumption may also be of importance. The findings corroborate the current advice to include fish and seafood as part of a balanced diet during pregnancy. Preterm delivery is a serious public health problem resulting in high societal costs, and no effective preventive strategies for the general population exist. Dietary changes have low cost and low risk compared with medical interventions and even moderate changes may be of public health importance.