Conflicts of interest The authors have stated explicitly that there are no conflicts of interest in connection with this article.
Evidence regarding an effect of marine n-3 fatty acids on preterm birth: a systematic review and meta-analysis
Article first published online: 14 JUN 2011
© 2011 The Authors Acta Obstetricia et Gynecologica Scandinavica© 2011 Nordic Federation of Societies of Obstetrics and Gynecology
Acta Obstetricia et Gynecologica Scandinavica
Volume 90, Issue 8, pages 825–838, August 2011
How to Cite
SALVIG, J. D. and LAMONT, R. F. (2011), Evidence regarding an effect of marine n-3 fatty acids on preterm birth: a systematic review and meta-analysis. Acta Obstetricia et Gynecologica Scandinavica, 90: 825–838. doi: 10.1111/j.1600-0412.2011.01171.x
- Issue published online: 12 JUL 2011
- Article first published online: 14 JUN 2011
- Accepted manuscript online: 3 MAY 2011 03:30AM EST
- Received: 14 October 2010, Accepted: 20 April 2011
- fish oil;
- gestational age;
- marine n-3 fatty acids;
- preterm delivery;
- preterm birth;
- pregnancy duration;
- Top of page
- Conclusions and recommendations
Background. Preterm delivery remains a substantial healthcare problem, complicating 5–10% of pregnancies, and is the major cause of perinatal morbidity and mortality in the developed world. Few effective methods to prevent preterm delivery have been identified to date. Objective. To review systematically the evidence from randomized controlled trials with respect to the hypothesis that increased consumption of marine n-3 fatty acids in pregnancy can prevent preterm birth. Setting. Electronic searches of the following databases were performed: PubMed (1995–2009), SCOPUS including EMBASE (1995–2009), and Cochrane Library. A combination of key words and text words related to fish oil, marine n-3 fatty acids, fish consumption, preterm birth, preterm delivery, prematurity, pregnancy duration, gestational age, parturition, delivery and pregnancy were used. Methods. A systematic review of randomized controlled trials of relevance was conducted. Three trials were included, comprising 921 women for whom data on gestational age and 1 187 women for whom data on birthweight were available. Results. Overall, 46 (8.9%) of 516 women who received n-3 fatty acids gave birth before 37 completed weeks of gestation, compared with 66 (16.3%) of 405 in the control group [relative risk 0.61; 95% confidence interval (CI) 0.40–0.93; p<0.05]. Data on delivery before 34 completed weeks showed the same trend (relative risk 0.32; 95% CI 0.09–0.95). Overall, the mean birthweight was 71g higher in women who received n-3 fatty acids during pregnancy (95% CI 4.73–138.12; p<0.05). The rate of low birthweight was not statistically significantly different between the intervention and the control groups. The mean gestational age at delivery was significantly higher by 4.5days in the intervention group supplemented with n-3 fatty acids compared with placebo (95% CI 2.3–6.8; p<0.05). Conclusions. Marine n-3 fatty acids administered in pregnancy reduce the rate of preterm birth and increase birthweight.
intrauterine growth restriction
randomized controlled trial
spontaneous preterm labor
- Top of page
- Conclusions and recommendations
Preterm birth (PTB) complicates 5–10% of pregnancies, and accounts for the majority of neonatal deaths in the developed world. Preterm birth is associated with a high morbidity in infancy, and the sequelae of PTB include significant mental, neurological and physical disability. Survival of the very premature infant has improved with advances in neonatal intensive care. However, these infants often stay in intensive care units for long periods after delivery, causing high distress and anxiety among relatives and incurring high healthcare costs (1). Despite substantial research, there are few effective methods to prevent PTB, and the etiology of spontaneous preterm labor (SPTL) and PTB is multifactorial and poorly understood.
The long-chain n-3 fatty acids consist primarily of eicosapentaenoic acid (EPA, 20:5n3) and docosahexaenoic acid (DHA, 22:6n3). They are designated marine because they are abundant in fat from animals of marine origin (2) and can be acquired by ingesting fish-oil supplements or by eating seafood. Marine n-3 fatty acids are involved in key cellular functions, such as the production of eicosanoids and transmembrane passage. Docosahexaenoic acid is present in high concentrations in certain tissues, such as the brain and retina. A substantial body of evidence supports the fact that marine n-3 fatty acids may have beneficial therapeutic or prophylactic roles in relation to certain cardiovascular (3) and atopic diseases (4), and that DHA in particular may be essential to early cognitive development (5). The aim of this meta-analysis was to review systematically the evidence from randomized controlled trials (RCTs) to address the hypothesis that increased consumption of marine n-3 fatty acids in pregnancy can prevent SPTL and PTB.
- Top of page
- Conclusions and recommendations
Electronic searches of the following databases were performed: PubMed (1995–2010), SCOPUS including Embase (1995–2010), and Cochrane Library. A combination of key words and text words related to fish oil, marine n-3 fatty acids, fish consumption, preterm birth, preterm delivery, prematurity, pregnancy duration, gestational age, parturition, delivery and pregnancy were used. The search strategy included the use of a validated filter for identifying RCTs and was combined with a topic-specific use of MeSH terms in PubMed. For studies identified which resulted in multiple publications, the data from the publication with the report of the primary end-points and the largest sample size were used.
Randomized controlled trials comparing long-chain n-3 fatty acid supplementation with placebo or no supplementation, in singleton pregnant women were identified. All studies deemed appropriate were retrieved and reviewed independently by two screening authors (JDS and RFL) to determine inclusion. Disagreements were resolved through consensus discussions.
The primary outcome of interest was PTB of less than 37 completed weeks of gestation. Secondary outcomes were very PTB before week 34 of gestational age, length of gestation, birthweight and low birthweight.
Study quality assessment
We developed a quality assessment tool for studies of the effectiveness of fish oil using recommended guidance for undertaking systematic reviews of health technology assessment (6). The format for the quality assessment tool has been previously used in a systematic review of the efficacy of nifedipine as a tocolytic (7,8). Study methodological quality was assessed using a tailored quality checklist, which structured items in two broad categories: topic-specific or method-specific items. Furthermore, these items were divided into three general subcategories: selection bias, performance bias and measurement bias. All quality items were scored as follows: 1, adequate; 2, inadequate; or 3, not stated. Authors of the original papers were contacted if clarification was needed, or if the original papers implied certain assumptions about the methodology. Consensus among the authors was reached through discussion and re-evaluation.
In the assessment of selection bias (randomization and concealment), computer-generated, random-sequenced numbers were considered to be adequate, and the use of alternation, case record numbers, birth dates or days of the week was considered to be inadequate. With respect to the assessment of concealment of allocation from the providers and patients, this was considered to be adequate if randomization was through: (1) centralized, real-time or pharmacy-controlled randomization; (2) serially numbered identical containers; or (3) robust methods to prevent foreknowledge of the allocation sequence to clinicians and patients. The use of alternation, case record numbers, birth dates or days of the week, open random number lists or serially numbered envelopes (regardless of opacity), which might be subject to manipulation, was considered inadequate. Assessment of performance bias was considered to be adequate if both the care provider and the study patient were blinded to the administered agent and inadequate if they were not blinded. Assessment of attrition bias was considered adequate if all those cases who dropped out or who were lost to follow up in the analysis were included or if the numbers and reasons for withdrawal were reported for each group and if the description of methods permitted analysis using the intention-to-treat principle. The assessment of attrition bias was considered to be inadequate if only numbers and not the reasons were provided for each study group, or if the description did not permit analysis following the intention-to-treat principle.
The checklist for topic-specific issues involved the use of items related to populations, interventions and outcomes pertinent to the use of fish oils for the prevention of preterm birth. In the assessment of selection bias, the similarity of the groups at baseline was considered to be adequate or inadequate depending on whether or not the distributions in each study group were equal for the following variables: age, parity, socioeconomic status, race, smoking, substance abuse, and history of prior late spontaneous abortion (<24 weeks) and preterm birth (<37 weeks). With respect to performance bias, adequate stipulation of the preparation, dose, type and administration for each group was assessed. The assessment of the use of fish oils was deemed to be adequate if the dosage, route and frequency of administration were reported in the methodology. The control was recorded as being either a placebo or no intervention. The measurement bias was assessed by the method used to calculate gestational age. The gestational age was considered adequate if calculated by accurate menstrual dates and/or ultrasonography, but inadequate if the last menstrual period was used with no mention of regularity of cycle or certainty of dates. Long-term follow up was adequate if reported and inadequate if not reported.
Two reviewers (JDS and RFL) scanned the abstracts and titles. Potentially relevant articles were aquired and evaluated independently by the two reviewers. No blinding of authorship was performed. All outcome data were extracted in duplicate from all reports independently by the two reviewers. Information was extracted on study group methodology, study group demographics, intervention details (dose, type, placebo or no treatment) and the outcomes, i.e. PTB prior to week 37, gestational age and birthweight.
Data from each independent study involving a binary outcome were organized in a two-by-two contingency table containing the number of patients classified according to the ‘disease status’ (such as diseased vs. healthy) and the fish-oil treatment applied (treated or nontreated). The data from each individual study were analyzed first using Fisher's exact test with computation of odds ratio and confidence interval (CI), as well as the p-value. Several meta-analyses were conducted by combining data from several of the four studies based on several criteria, e.g. including all studies contrasting fish-oil-treated with placebo-treated women.
In each meta-analysis involving a binary outcome, the contingency tables of the individual studies included were expanded into a global table with three indicator variables for each patient in each study: outcome, treatment and study identity (ID). The effect of the treatment on the outcome was estimated using a logistic regression model in which different studies were treated as fixed effects, as recommended elsewhere (http://www-users.york.ac.uk/~mb55/intro/introcon.htm). The p-value, odds ratio and its confidence interval were derived from the logistic regression model.
The data available from each individual study involving a continuous outcome (such as birthweight) were the means and standard deviations of the outcome in each group (treated and untreated). Data in each study were analyzed using a two-tailed two-sample equal variance t-test. As the raw data from the individual studies were not available, meta-analysis for continuous outcomes was performed using the following steps: (1) we drew a random sample of normally distributed values with the size, mean and standard deviation corresponding to those reported by the authors of each study in each group (treated or untreated); (2) a linear model was used to estimate the effect of treatment on the outcome while treating each study ID as a fixed effect; the effect size of the treatment thus obtained, as well as confidence intervals and p-values were extracted from the linear model; and (3) steps 1 and 2 were repeated 1000 times and the mean of the statistics described at step 2 was computed, except for the p-values, for which the median was computed instead. These summary statistics were assigned as the meta-analysis result.
For both types of meta-analyses involving continuous and binary outcomes, a weight coefficient was determined for each study in each meta-analysis as the ratio between the sample size of the individual study and the one of all studies considered in a given meta-analysis. These weights were provided only as a rough estimate of how much each study weights into the overall result of a given meta-analysis. However, the overall effect size and significance in meta-analyses were derived from the logistic and linear regression models as described above, which basically perform a within strata (study) estimation of the effect of treatment and summarize these effects in a global effect. All computation and plots were performed using the R statistical environment (http://www.r-project.org).
- Top of page
- Conclusions and recommendations
The flow of the electronic search is shown in Figure 1. Of the 80 potentially relevant citations identified, 67 were excluded based on the title or on review of the abstract. Based on abstract review, hard copies of 13 studies were obtained. After a detailed review of these, three studies (four trials) fulfilled the inclusion criteria and were included in the meta-analysis (9–11). One study (9) had two independant arms (listed as earl-PD and earl-IUGR) to signify previous pregnancy with PTB or intrauterine growth restriction (IUGR), respectively, which quoted our prescribed outcomes and was considered as two independent trials in this review (Table 1). Of the 10 studies excluded (Table 2), three were based on the same population as reported in another study (12–14), four reported no useful data according to the outcome (15–18) and three used a combination of n-3 fatty acids and another intervention that could not be considered separately (19–21). The four included trials comprised 921 women with respect to data on gestational age and 1 187 women with respect to data on birthweight.
|Study||Population||Intervention||End-point data||Description of withdrawals|
|Smuts et al. (2003) (11)||n=350, enrolled week 24–28, 291 completed the study||Docosahexaenoic acid egg 33mg vs. 133mg||83%||Yes|
|Olsen et al. (2000) (9)||n=232 with previous preterm delivery enrolled week 18–21||Supplementation with 2.7g n-3 fatty acid capsules vs. capsules of olive oil||98%||Yes|
|Olsen et al. (2000) (9)||n=280 with previous intrauterine growth restriction, enrolled week 18–21||Supplementation with 2.7g n-3 fatty acid capsules vs. capsules of olive oil||94%||Yes|
|Olsen et al. (1992) (10)||n=533, low risk, enrolled week 30||Supplementation with 2.7g n-3 fatty acid capsules vs. capsules of olive oil or no supplementation||100%||Yes|
|Study||Reasons for exclusion|
|Olsen et al. (1994) (14)||Reported the same population as in another study|
|Onwude et al. (1995) (18)||Lack of useful data according to the outcome – spontaneous preterm delivery|
|Bulstra-Ramakers et al. (1995) (15)||Lack of useful data according to the outcome – spontaneous preterm delivery|
|Borod et al. (1999) (12)||Abstract of subsequently published randomized trial, included in the review (11)|
|de Groot RH et al. (2004) (16)||Lack of useful data according to the outcome – spontaneous preterm delivery. Intervention with α-linolenic acid, which is a precursor of docosahexaenoic acid.|
|Khoury et al. (2005) (20)||The intervention was a cholesterol-lowering diet without specification of the content of n-3 fatty acids|
|Knudsen et al. (2006) (17)||Lack of useful data according to the outcome – spontaneous preterm delivery. No placebo|
|Olsen et al. (2007) (13)||Reported the same population as in another study, included in the review (FOTIP trial, (9))|
|Mardones et al. (2008) (21)||The intervention was micronutrients in addition to n-3 fatty acids. The selected n-3 fatty acid was α-linolenic acid, which is a precursor of docosahexaenoic acid, and the dose was not specified|
|Harper et al. (2010) (19)||The intervention was progesterone in addition to n-3 fatty acids|
Study quality assessment
The results of the evaluation of the studies’ adherence to the criteria within the two domains of quality (method- and topic-specific items) are presented in Table 3 and Figures 2 and 3. For all the method-specific items of quality, the majority of the studies were adequate (Figure 2). For 11 of the 14 topic-specific items of quality, the majority of the studies were considered adequate. For all of the 14 topic-specific items of quality, none of the studies was considered inadequate (Figure 3).
|Olsen et al. (1992) (10)||Olsen et al. (2000) (9)||Smuts et al. (2003) (11)|
|Method-specific items of quality|
|A priori sample size calculation||Unstated||Adequate||Adequate|
|Topic-specific items of quality|
|Previous preterm birth||Unstated||Adequate||Adequate|
|Previous late miscarriage||Unstated||Adequate||Unstated|
|Route of administration||Adequate||Adequate||Adequate|
|Long-term follow up||Unstated||Unstated||Unstated|
Characteristics of included studies
The characteristics of the included studies are shown in Table 1. Two studies included pregnant women considered to be at low risk (10,11), and one study (9) with two independent trials included only women with a previous PTB or previous IUGR, and therefore a high risk of recurrence. One study was performed in Denmark (10), one study in the USA (11) and one as a multicenter study in Europe (9). The estimated baseline intake of n-3 fatty acids was expected to be different among the study populations, with the highest intake in Denmark. In two studies, the supplementation of n-3 fatty acids was equal (9,10), but the duration of the supplementation differed. In the third study, the dose of n-3 fatty acid was lower (11). In all studies, the intervention was DHA, and in the Danish and European studies an additional amount of eicosapentaenoic acid (EPA) was part of the intervention. In two studies, the main outcome was gestational age, while data on PTB and birthweight were presented as secondary outcomes (10,11). In the European study (9), with two included trials, the main outcomes were recurrence of PTB and IUGR. Data on birthweight and gestational age were reported as secondary outcomes.
Overall, 46 (8.9%) of 516 who who received n-3 fatty acids delivered before 37 completed weeks of gestation, compared with 66 (16.3%) of 405 in the control group (relative risk 0.61; 95% CI 0.40–0.93; p<0.05; Figure 4). Data on delivery before 34 completed weeks showed the same trend (relative risk 0.32; 95% CI 0.09–0.95).
Overall, the mean birthweight in women who received n-3 fatty acids during pregnancy was higher by 71g than that of women who did not receive the treatment (95% CI 4.73–138.12; p<0.05; Figure 5). The rate of low birthweight was not significantly different between the intervention and the control groups (Figure 6).
The mean gestational age at delivery was significantly higher by 4.5days in the intervention group supplemented with n-3 fatty acids compared with placebo (95% CI 2.3–6.8; p<0.05; Figure 7).
- Top of page
- Conclusions and recommendations
This systematic review and meta-analysis, in which only randomized placebo-controlled studies with data on SPTL and PTB were included, demonstrated that n-3 fatty acid supplementation during pregnancy reduces the risk of PTB and increases birthweight. The included studies represent different populations with respect to the risk profile concerning PTB. Although the results from RCTs (9,11,22–24) are promising and supported by our meta-analysis, there is a paucity of RCTs which are large enough to demonstrate whether an increased intake of marine n-3 fatty acids is associated with a reduced risk of PTB in the general population of pregnant women as opposed to only those women with a previous PTB. Equally, the sample size is not sufficient to demonstrate a reduced risk of the complications seen in association with PTB, such as neonatal infections.
Origin of hypothesis
Population comparisons and clinical case studies produced the hypothesis that dietary marine n-3 fatty acids could delay the timing of spontaneous labor and delivery (25). Birthweights in the Faroe Islands, where fishing is the predominant way of living, were found to be higher on average by approximately 200g compared with birthweights on the mainland of Denmark (26). A case report suggested that Faroese women had longer pregnancy durations and higher birthweights (27), and this stimulated a discussion about a possible nutritional explanation related to fish intake. More detailed comparisons of birth statistics showed that gestations were longer in Faroese women compared with women on the Danish mainland, by approximately three to four days on average. This could account for approximately half of the observed difference in the mean birthweight between the two populations (25). Records kept in historical archives showed that the mean birthweight in the Faroese had been 170g higher before the Second World War compared with the present time, and this matched observed changes in the traditional Faroese lifestyle and how these tended to increase the birthweight (28).
Several different mechanisms have been suggested. Originally, it was proposed that the long-chain n-3 fatty acids might reduce the activity of eicosanoid promoters of the parturition process, particularly prostaglandins (PGs) F and E, and increase the activity of eicosanoids with myometrial relaxant properties, such as prostacyclins (29,25). Biochemical studies confirmed that EPA and DHA influence the formation of PGs in fetal membranes (30,31). In the sheep model, the infusion of an emulsion of long-chain n-3 fatty acids stopped SPTL induced by betamethasone (32). Reduced synthesis of myometrial PGH synthase-2 mRNA, which is the main rate-limiting enzyme for producing PGs in the myometrium, was observed in the fish-oil group compared with the control group, and it was suggested that this might be part of the mechanism underlying the effect on the parturition process (33). Experiments in the rat also demonstrated longer gestations in those rats that received a diet high in fish oil compared with arachis oil, but the underlying mechanisms were not studied (34). A third mechanism is based on the theory that marine n-3 fatty acids have an effect on the electrical activity of the heart, and these anti-arrhythmic properties (35) may be why fish oils are protective against sudden death following myocardial infarction (3). Since the myometrium, like the myocardium, has constant electrical and contractile activity (33), even during the prelabor phase, one possibility might be that marine n-3 fatty acids have a similar ‘anti-arrhythmic’ effect on the myometrium, which might explain their effect on delaying the initiation of labor, as well as their proposed tocolytic properties (36).
Three observational studies have demonstrated positive associations between biomarkers for the intake of marine n-3 fatty acids and the length of gestation (37–39). In a much larger, prospective observational study conducted among 8 729 Danish women, low fish consumption recorded in the first trimester was shown to be a strong risk factor for PTB (40), with the strongest association among women with an intake below 0.15g n-3 fatty acids per day. It was also possible to define a subgroup of 764 women with large exposure contrasts (41). In this subgroup, women with zero fish intake during both the first and the second trimester of pregnancy had a risk of PTB that was more than 10-fold, and they also had an 8.6day longer gestational age compared with those who reported a high intake of fish. Several other observational studies (42–48) did not show an association between the estimated intake of marine n-3 fatty acids and duration of pregnancy, although some suggested a trend (42,45).
The first RCT conducted with the specific aim of examining the hypothesis (10) showed a longer mean duration of pregnancy of 4.0days and a corresponding increase in mean birthweight of 100g in women who received fish oil. Compared with the 20% of women who reported the lowest regular intake of fish at baseline, the mean gestation length was 7.4days longer in the fish-oil group. A European series which co-ordinated six different RCTs was conducted among high-risk women (9). In the trial which included women who had had a previous PTB (n=228), the recurrence rate was reduced from 33 to 21% (odds ratio 0.54; 95% CI 0.30–0.98). The risk of delivery before 34 weeks gestation was also reduced from 13.3 to 4.6% (odds ratio 0.32; 95% CI 0.11–0.89), and the mean gestation and mean birthweight increased by 8.5days and 209g, respectively. Data from these studies are included in our meta-analyses.
The effect of fish oil on pregnancy duration depends on the background intake of fish. If a woman has a relatively high fish intake, fish oil supplementation appears to provide no benefit, suggesting the existence of a threshold or ‘saturation’ level of intake (13). In contrast, relatively low doses of marine n-3 fatty acids of 0.2g/day or less may have substantial benefits for women with a prior low intake of marine n-3 fatty acids (11,23,40). Furthermore, the Aarhus trial (10), the FOTIP trial (9) and the Aarhus observational study (40) suggest that in women with an intake above a certain level, no additional benefits were observed by further increasing the intake. The Aarhus observational study estimated the critical level to be as low as 0.15g marine n-3 fatty acids per day (40).
An RCT of DHA supplementation during pregnancy, also included in the meta-analyses, was conducted in a predominantly black USA population with low baseline consumption of marine n-3 fatty acids (11). The mean gestational age was 274.1days in the DHA-supplemented group compared with 271.6days in the control group. Another RCT of DHA supplementation during pregnancy was conducted in a population who had a baseline DHA intake far below that which was recommended. The purpose of the study was to evaluate the possible benefit on problem solving during infancy (49). The gestational age at birth in the DHA-supplemented group turned out to be significantly greater, at 39.9days compared with 39.0days in the control group (p=0.019).
In contrast, several well-conducted trials show no benefit of n-3 fatty acids on pregnancy duration and PTB (15,17,18,50). One trial was conducted among women with an increased risk of IUGR or hypertension in pregnancy owing to similar problems in a previous pregnancy (15). Treatment was started between 12 and 14 weeks postmenstrual age, and the fish oil provided 3g EPA, though no information was given on the content of DHA. No estimate of mean gestational length was provided, but women delivering at gestational ages <28, 28–32, 32–34, 34–37 and >37 weeks occurred with frequencies of 1, 1, 1, 5 in the fish-oil group and 24 and 1, 2, 3, 4 and 21 in the coconut-oil group, respectively. All three SPTLs and PTBs occurred in the coconut-oil group. Although these trends were compatible with an effect, the trial lacked power to support the hypothesis. In another trial (18), women at increased risk of either pre-eclampsia or IUGR, either because of a past obstetric history or because of abnormal umbilical artery Doppler measurements in the present pregnancy, were randomized to receive nine capsules per day filled with either fish oil or air. The fish-oil capsules provided 2.7g of n-3 fatty acids per day. The mean length of gestation (38.1 weeks) was very similar in the two groups. Women who delivered at gestational ages 32, 32–37 and 37+weeks occurred with frequencies of 3, 19 and 91% in the fish-oil group and 2, 17 and 100% in the air group, respectively. However, only 49% (55 of 113) and 57% (68 of 119) of deliveries, respectively, were spontaneous. The high proportion of elective delivery in this high-risk population may have made the detection of any true effect on the timing of spontaneous delivery difficult, but the trial provided no support to the hypothesis.
In another RCT, 590 Norwegian women were randomized to receive 10ml daily of either fish oil, which provided around 2g daily of marine n-3 fatty acids, or corn oil (5g per day), which is the main constituent of linoleic acid, an n-6 fatty acid (50). Women were recruited in weeks 17–19 of pregnancy, and because the intention was also to examine the effect of supplementation on the neurodevelopment of offspring, they were asked to take the supplements until three months after delivery. The two groups were similar with respect to gestational age at birth, which was 279.6days in the fish-oil group and 279.2days in the corn-oil group. Significant numbers of women (116 of 291 and 106 of 272, respectively) were excluded from the two groups owing to withdrawals from the study, postrandomization. As the length of gestation for these women was not established, it was not possible to perform an intention-to-treat analysis for the end-point. Fatty acids were quantified in umbilical plasma phospholipids. As expected, in the fish-oil group, supplementation led to substantial increases in EPA and DHA, and in the corn-oil group, of arachidonic acid. When gestational length was compared with DHA concentrations in umbilical plasma phospholipids, mean gestational length was greater by 7.1days in the highest compared with the lowest quartile of DHA (282.5 vs. 275.4days; p=0.001). Elective cesarean sections occurred significantly more often in the fish-oil group (6.3%) than the corn-oil group (1.2%). When elective caesarean sections and induced deliveries were added together, 17.7% (31 of 175) occurred in the fish-oil group and 7.8% (13 of 166) in the corn-oil group. The baseline intake of long-chain n-3 fatty acids was estimated to be relatively high, at 0.5g/day, and PTB occurred at a low rate of 1%. Although this trial did not support the hypothesis, the substantial attrition in the study population, combined with a high baseline intake of marine n-3 fatty acids in this population, may have reduced the opportunity to observe any true effect of fish oil. The Danish National Birth Cohort recruited 101 046 women in their pregnancy for long-term follow up (51,52). An RCT was conducted within this cohort (17). Three thousand and ninety-eight women reporting low dietary fish intake in an interview carried out at 12 weeks gestation were randomized in a ratio of 1:1:1:1:1:1:2 into six treatment groups and a control group. The treatment groups were offered to receive, by mail, capsules containing flax oil or fish oil in five different doses. The control group was not offered any oil capsules. No differences were seen across the groups with respect to gestational length, but a possible explanation for this is that compliance may have been very low.
In general, there is substantial heterogeneity between trials with respect to population, baseline n-3 levels, and the amount and timing of n-3 fatty acid supplementation. Nevertheless, two recent meta-analyses of trials of fish-oil supplementation in low-risk pregnancy concluded that fish oil has a prolonging effect on the duration of pregnancy. Makrides et al. (53) concluded that fish-oil supplementation was associated with a 2.6days (95% CI 1.0–4.1) longer mean gestation, whereas Szajewska et al. (54) concluded that it was associated with 1.6days longer mean gestation. Both reviews questioned the clinical significance of this effect, and so did a recent narrative review (55). However, a recent systematic review of the literature of preventive agents against preterm birth concluded that fish oil is one of the promising agents (56). In contrast, a meta-analysis of trials of fish-oil supplementation in high-risk pregnancy did not confirm the effect on duration of pregnancy, although n-3 supplementation was associated with a lower rate of early PTB before 34 weeks of gestation (57). The high incidence of induced deliveries in two of the included trials in this meta-analysis could have interfered with the possibility of showing an effect on SPTL.
Evidence for a rapid effect on timing of birth
There is evidence to suggest that the effect of an increased intake of marine n-3 fatty acids on duration of gestation may be rapid (36). Randomized controlled trials have demonstrated that fish-oil supplementation can significantly prolong gestation and could delay timing of delivery substantially when administered from week 30 (10), and even from week 33 of gestation (9). These studies supported the hypothesis of a relatively rapid effect, as did an observational study that used fatty acids measured in erythrocyte phospholipids as a biomarker for intake (38). The association seen with length of gestation was stronger when fatty acids were measured in a phospholipid fraction with a fast turnover of fatty acids (phosphatidylcholine). This probably reflects intake during a relatively narrow time scale prior to blood sampling than would be evident if the fatty acids were measured in other phospholipid fractions, such as phosphatidyl ethanolamine (38), which have a slower turnover (58).
Linoleic acid, an n-6 fatty acid, is the parent fatty acid of PGF2α and PGE2, which are essential to the parturition process. Pregnant rats fed on a diet that was low in linoleic acid but high in linolenic acid (the parent n-3 fatty acid) had impaired parturition (59). While this was not unexpected, when the investigators exchanged linolenic acid with linoleic acid, from day 19 of pregnancy (which is only two days prior to the expected date of delivery in the rat), parturition was normalized. In the rat at least, there seems to be a rapid effect of dietary fatty acids on the process of parturition. In another animal study, six sheep had an intravenous infusion of n-3 fatty acids after labor had been induced with betamethasone (32), whereas six control sheep had an infusion with neutral lipids. In the fish-oil group, the onset of both labor and time to delivery was delayed. In two animals that received n-3 fatty acids, premature labor stopped completely. These findings suggest that, in the sheep, long-chain n-3 fatty acids may have tocolytic properties (36).
Possible long-term effects of increased intake of marine n-3 fatty acids in pregnancy
Increased intake of marine n-3 fatty acids in pregnancy may have beneficial effects with respect to atopic diseases and cognitive development in offspring. A follow-up study over 16years of children from the Aarhus trial showed that children whose mothers took fish oil in pregnancy had a lower risk of developing asthma compared with those children whose mothers had taken olive oil (60). Based on data from the Danish National Birth Cohort, developmental attainment in the first 18months was greater in children of mothers who had a relatively high intake of fish during their pregnancy. The relation appeared to be graded across the fish-intake distribution of this population of Danish women (61).
Potentially modifying factors
Several factors may modify the relation between the intake of marine n-3 fatty acids and the duration of pregnancy, such as genetic factors and the intake of other nutrients. α-Linolenic acid (ALA, 18:3n.3) the parent fatty acid of the long-chain n-3 fatty acids, may be important. Conversion of ALA to EPA and DHA is generally thought to occur slowly in humans (62,63). However, recent studies have indicated that supplementation with ALA elevates erythrocyte levels of EPA and DHA, and the dose of ALA can be easily achieved by dietary modification (64). Sources of ALA are widely available in leafy vegetables, vegetable oils derived from flaxseed, canola and soybeans, and nuts such as walnuts and almonds. There may also be a genetic–environmental interaction in the relation between maternal n-3 fatty acids and the duration of pregnancy. A number of candidate genes have been proposed, such as TNF-α, interleukin (IL)-4, IL-6 and IL-10, TNFR1, TNFR2 and IL1R2, MMP1, MMP9, SERPINH, VEGF and factor VII (65,66) Genes involved in fatty acid metabolism and cyclo-oxygenase 2 production, such as FADS1 and FADS2, would also be of great interest (67–70).
Potential unwanted effects of increasing the intake of marine n-3 fatty acids in pregnancy
Prolongation of pregnancy may not always be advantageous. In the European multicenter trial comparing 1 617 women (9), post-term delivery after 294days of gestation was substantially increased in the fish-oil arm, and in the Aarhus observational study (41,40), both the risk of post-term delivery and elective delivery was increased in women with high fish intake. Although generally regarded as a minor problem compared with preterm delivery, post-term delivery is a risk factor for perinatal complications such as stillbirth (71) and is generally undesirable. As a result, any treatment with fish-oil supplementation should be stopped whenever pregnancy has gone beyond term, or even the preterm period. Fish oil has been shown to increase the bleeding time, albeit that this is a laboratory measurement, and so the clinical relevance is uncertain. However, in two studies comprising 2 150 women, no significant increases in bleeding complications in the fish-oil group were detected (9,10).
Research in this area in future needs to address the following questions. If marine n-3 fatty acids delay the timing of SPTL and prevent PTB, what is the underlying mechanism? If they do prevent PTB, can they also prevent those complications such as infections associated with SPTL and PTB? If marine n-3 fatty acids delay timing of SPTL and PTB, what dose is necessary to obtain a clinically significant effect? Is there a dose above which no additional benefit can be obtained? Is it possible to establish a baseline intake below which the benefits of additional marine n-3 fatty acids on the prevention of SPTL or the delay in PTB are not evident? Could marine n-3 fatty acids be useful as a tocolytic in SPTL? Is it possible that other dietary factors, such as α-linolenic acid, genetic factors related to the parturition process, immune response or fatty acid metabolism interact? Does the intake of marine n-3 fatty acids in pregnancy have any long-term beneficial influences on child health and development of the offspring of women apart from a possible benefit on the timing of delivery? In particular, can marine-3 fatty acids affect neurodevelopment and the risk of atopic disease in infants?
Conclusions and recommendations
- Top of page
- Conclusions and recommendations
Evidence from a few large, well-conducted studies suggests that marine n-3 fatty acids may delay the timing of spontaneous delivery and be beneficial in relation to the prevention of PTB and its associated complications. These findings are supported by the present systematic review and meta-analysis. However, many studies have shown conflicting results, and more research is needed to substantiate, or reject, these theories. Some studies suggest that the dose–response relationship is only present at very low intakes. Others suggest that the effect of an increased intake of marine n-3 fatty acids on the timing of birth is a rapid one. Detecting such effects in observational prospective studies is unlikely, and future studies should be designed to address these issues. Future studies should clearly identify women who do not consume fish or who consume fish very rarely. Ideally, they should assess fish intake prospectively over prolonged periods of pregnancy, without affecting intake among study participants. Future randomized controlled trials of the effect of fish-oil supplementation on pregnancy duration should make every effort to optimize compliance to allocated treatment regimens, as well as excluding allocation and treatment bias, particularly in the control group, to prevent self supplementation with fish oil, in order to maintain low exposure in the control group. Further studies should also address cost vs. benefit and should be powered to demonstrate an effect on neonatal outcome.
- Top of page
- Conclusions and recommendations
No specific funding.
- Top of page
- Conclusions and recommendations
- 3Dietary supplementation with n-3 polyunsaturated fatty acids and vitamin E after myocardial infarction: results of the GISSI- Prevenzione trial. Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto miocardico. [See comments.] Lancet. 1999;354:447–55.
- 6Undertaking systematic reviews of research on effectiveness. Number 4 (2nd edn). 2001..
- 24People's League of Health. Nutrition of expectant and nursing mothers. BMJ. 1942;ii:77–8.
- 33Maternal intravenous administration of long chain n-3 polyunsaturates to the pregnant ewe in late gestation results in specific inhibition of prostaglandin h synthase (PGHS) 2, but not PGHS1 and oxytocin receptor mRNA in myometrium during betamethasone-induced labor. J Soc Gynecol Investig. 2000;7:233–7., , , .
- 38Pregnancy duration and the ratio of long-chain n-3 fatty acids to arachidonic acid in erythrocytes from Faroese women. J Intern Med Suppl. 1989;225:S185–9., , , .