Effect of n-3 Long-chain Polyunsaturated Fatty Acid Intake during Pregnancy on Maternal, Infant, and Child Health Outcomes: A Systematic Review

Authors

  • Beth Imhoff-Kunsch,

    Corresponding author
    1. Hubert Department of Global Health, Rollins School of Public Health
      Beth Imhoff-Kunsch, MPH, PhD, Hubert Department of Global Health, Rollins School of Public Health, Emory University, 1599 Clifton Road, N.E. – GHI Rm 6.431, Atlanta, GA 30322, USA. E-mail: beth.imhoffkunsch@gmail.com
    Search for more papers by this author
  • Virginia Briggs,

    1. Orthopedics and Rehabilitation Department, University of Massachusetts Medical School, Worcester, MA, USA
    Search for more papers by this author
  • Tamar Goldenberg,

    1. Hubert Department of Global Health, Rollins School of Public Health
    Search for more papers by this author
  • Usha Ramakrishnan

    1. Hubert Department of Global Health, Rollins School of Public Health
    2. Doctoral Program in Nutrition and Health Sciences, Graduate Division of Biological and Biomedical Sciences, Emory University, Atlanta, GA
    Search for more papers by this author

Beth Imhoff-Kunsch, MPH, PhD, Hubert Department of Global Health, Rollins School of Public Health, Emory University, 1599 Clifton Road, N.E. – GHI Rm 6.431, Atlanta, GA 30322, USA. E-mail: beth.imhoffkunsch@gmail.com

Abstract

Evidence from observational studies and randomised trials has suggested a potential association between intake of n-3 long-chain polyunsaturated fatty acids (LCPUFA) during pregnancy and certain pregnancy and birth outcomes. Marine foods (e.g. fatty sea fish, algae) and select freshwater fish contain pre-formed n-3 LCPUFA, which serve as precursors for bioactive molecules (e.g. prostaglandins) that influence a variety of biological processes. The main objective of this analysis was to summarise evidence of the effect of n-3 LCPUFA intake during pregnancy on select maternal and child health outcomes. Searches were performed in PubMed, EMBASE, and other electronic databases to identify trials where n-3 LCPUFA were provided to pregnant women for at least one trimester of pregnancy. Data were extracted into a standardised abstraction table and pooled analyses were conducted using RevMan software. Fifteen randomised controlled trials were eligible for inclusion in the meta-analysis, and 14 observational studies were included in the general review. n-3 LCPUFA supplementation during pregnancy resulted in a modest increase in birthweight (mean difference = 42.2 g; [95% CI 14.8, 69.7]) and no significant differences in birth length or head circumference. Women receiving n-3 LCPUFA had a 26% lower risk of early preterm delivery (<34 weeks) (RR = 0.74; [95% CI 0.58, 0.94]) and there was a suggestion of decreased risk of preterm delivery (RR = 0.91; [95% CI 0.82, 1.01]) and low birthweight (RR = 0.92; [95% CI 0.83, 1.02]). n-3 LCPUFA in pregnancy did not influence the occurrence of pre-eclampsia, high blood pressure, infant death, or stillbirth. Our review of observational studies revealed mixed findings, with several large studies reporting positive associations between fish intake and birthweight and several reporting no associations. In conclusion, n-3 LCPUFA supplementation during pregnancy resulted in a decreased risk of early preterm delivery and a modest increase in birthweight. More studies in low- and middle-income countries are needed to determine any effect of n-3 LCPUFA supplementation in resource-poor settings, where n-3 PUFA intake is likely low.

Evidence from observational studies has suggested a potential association between intake of marine foods during pregnancy and certain pregnancy and birth outcomes. Researchers observed in the 1980s that women living in the Faroe Islands, where seafood is commonly consumed, had longer gestational duration and heavier babies than women living in Denmark.1,2 Additionally, researchers found that Greenland Inuits had lower rates of pre-eclampsia than Danish women. These observations instigated future research founded on the hypothesis that intake of marine foods in pregnancy might influence maternal and neonatal health outcomes.

Marine foods are rich sources of the n-3 long-chain polyunsaturated fatty acids (LCPUFA) docosahexaenoic acid (DHA, 22:6n-3) and eicosapentaenoic acid (EPA, 20:5n-3), bioactive LCPUFA that serve as structural and functional constituents of cell membranes. These LCPUFA act as signalling molecules and thereby influence the regulation of numerous biological processes such as eicosanoid synthesis and gene expression. Intake of n-3 LCPUFA could affect the synthesis of bioactive molecules such as prostaglandins, which in turn influence an array of biological activities including vasodilation, placental blood flow, cervical ripening, and onset of labour.

Long-chain polyunsaturated fatty acids, particularly DHA and arachidonic acid (AA, 20:4n-6), are integral for fetal neural and retinal development, accrete extensively in these tissues in the last trimester of pregnancy, and are preferentially transported across the placenta to the developing fetus.3–6 The fetus has limited ability to synthesise DHA and, therefore, its supply is almost entirely dependent upon maternal transfer.7 Studies have demonstrated that supplementing pregnant women with fish oil, DHA, or DHA-enriched foods improves both maternal and neonatal n-3 PUFA status.8–14

Numerous epidemiological studies and randomised trials have examined the influence n-3 LCPUFA intake during pregnancy on a variety of maternal, neonatal, and child health (MNCH) outcomes and several meta-analyses have been published.15–17 This systematic review not only updates a previous meta-analysis that included trials published through December 2005, but also reviews evidence from observational studies of associations between n-3 LCPUFA consumption during pregnancy and MNCH outcomes.

Methods

Literature search strategy

We initially searched electronic databases including PubMed, EMBASE, PopLine, the Cochrane Library, and Web of Science in January 2011 to identify trials for this review. The following Medical Subject Heading (MeSH) terms were included in the search: fish oil or DHA or EPA or marine oil or n-3 LCPUFA AND pregnancy or pregnant or maternal AND high blood pressure or hypertension or pre-eclampsia or eclampsia or preterm birth or birthweight or low birthweight (LBW) or birth length or head circumference or intrauterine growth restriction or morbidity or mortality or stillbirth or growth. In addition, bibliographies in key papers were hand-searched for relevant studies. We included only papers with human subjects published in English and imposed no restrictions on year of publication.

We searched articles for the following specific outcomes of interest:

  • 1Maternal morbidity (hypertension, pre-eclampsia);
  • 2Maternal mortality;
  • 3Preterm birth (<37 weeks), early preterm birth (<34 weeks);
  • 4LBW (<2500 g);
  • 5Intrauterine growth restriction or small for gestational age;
  • 6Stillbirth;
  • 7Infant growth, morbidity and mortality; and
  • 8Child growth, morbidity and mortality.

Inclusion criteria for the meta-analysis

  • • Randomised controlled trials (RCTs).
  • • The participants were pregnant women, and the intervention was provided during pregnancy.
  • • Women in the intervention group received n-3 LCPUFA (i.e. DHA and/or EPA found in fish or algal oils). Trials that provided only n-3 PUFA such as α-linolenic acid were excluded.
  • • In the case of co-interventions, both groups received the same co-intervention (e.g. cereal bar + DHA vs. cereal bar).
  • • One or more of the aforementioned MNCH outcomes was assessed.
  • • Both high and low risk pregnancies were eligible for inclusion.

Inclusion and exclusion criteria for observational studies included in the general review

  • • Individual dietary intake of foods containing n-3 LCPUFA (i.e. fish and marine foods) during pregnancy was assessed.
  • • Ecological studies, case series, case reports, and cross-sectional studies were excluded.
  • • All included studies reported information about one or more of the aforementioned outcomes of interest.

Data extraction, statistical methods, and assessment of quality

All relevant trial data (e.g. study context, design and limitations, intervention specifics, and effects estimates) were entered into an Excel-based data extraction form designed specifically for this study (available upon request). At least 75% of data were double-entered to ensure accuracy. When necessary, the author calculated quantitative values, for example standard deviations, based on numbers found in the text or tables. We contacted corresponding authors from two of the trials to obtain numbers needed for the meta-analysis (SDs from the Ramakrishnan et al.18 trial and disaggregated data on stillbirths and infant deaths from the Makrides et al.19 trial). Rates of infant death (0–12 months) were calculated based on the number of live births (i.e. we subtracted stillbirths from the denominator where possible).

We conducted pooled analyses using Review Manager software (RevMan version 5; Copenhagen, Denmark) where data were available from more than one trial. Statistical heterogeneity was assessed for each outcome with sufficient data using I2, where I2 = 100% × (Q − df)/Q (Q = Cochran's heterogeneity statistic, df = degrees of freedom).20 Fixed effect models were used in cases of low to moderate heterogeneity (I2 < 50%), and random effect models were used in cases of moderate to high heterogeneity (I2 ≥ 50%). We generated risk ratios (RR) with 95% confidence intervals (CI) for dichotomous outcomes, and mean differences (MD) with 95% CI for continuous outcomes.

We assessed the quality of evidence for outcomes using an approach used by the Child Health and Epidemiology Reference Group (CHERG), based on Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology.21–23 This approach grades the overall strength of the evidence for outcomes based on individual study design and quality, consistency and volume of evidence across studies, and the effect size and strength of association across studies. Factors such as sequence generation, allocation concealment, blinding, loss to follow-up, relevance to outcome of interest, and selective reporting were considered when grading individual trials. Grades for the individual included RCTs can be found on Webtable 1.

Results

Results from the meta-analysis

A flow diagram outlining details of the literature search is shown in Figure 1. We identified 15 RCTs that examined the effect of n-3 LCPUFA in pregnancy on select MNCH outcomes.18,19,24–39 All control groups were provided a placebo, either in the form of capsules, milk-based supplements, eggs, or cereal bars. We present a summary of individual trial characteristics in Table 1. In 12 of the 15 included trials, n-3 LCPUFA supplementation began in the second trimester of pregnancy. The Olsen et al.26 and Sanjurjo et al.37 trials provided supplements early in the third trimester of pregnancy, and some women in the Bulstra-Ramakers et al.29 trial began supplementation in gestation week 12 or 13. Twelve of the trials were conducted in higher income countries and three were conducted in low–middle-income countries (Mexico,18 Angola25 and Bangladesh38). In 11 of the trials, the supplements were DHA + EPA, often as fish oil. Two trials provided DHA alone,18,34 and two provided ‘low EPA’ fish oil.28,36 Doses of DHA ranged from 80 mg/day to 2.2 g/day. Three of the 15 trials provided DHA or DHA + EPA in food or beverage vehicles; one trial provided a milk-based supplement,28 one provided DHA-enriched eggs,34 and one provided cereal bars.36 Reasons for exclusion of data and trials are detailed in Appendix 1. Of note, we could not include the NUHEAL multi-centre European trial because SDs for our birth outcomes of interest were not reported.14,40

Figure 1.

Flow chart identifying randomised controlled trials evaluating the effect of n-3 LCPUFA in pregnancy on maternal, neonatal, and child health outcomes. LCPUFA, long-chain polyunsaturated fatty acids.

Table 1.  Description of trials included in the meta-analysis of the effect of n-3 LCPUFA supplementation in pregnancy on maternal and child health outcomes
Country, first author, publication yearDesign, nPopulationInterventionControlDurationOutcomes
  1. BL, birth length; BP, blood pressure; BW, birthweight; DB RCT, double blind randomised controlled trial; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; EPB, early preterm birth; FO, fish oil; FOS, fructo-oligosaccharide; gest, gestation; GLA, gamma linolenic acid; IUGR, intrauterine growth restriction; LA, linoleic acid; LBW, low birthweight; LCPUFA, long-chain polyunsaturated fatty acids; LNA, α-linolenic acid; NICU, neonatal intensive care unit; PB, preterm birth; PIH, pregnancy induced hypertension; preg, pregnant; RCT, randomised controlled trial; SGA, small for gestational age; supp, supplement.

Angola, D’Almeida, 199225RCT, 1 maternity centre, n = 150Preg women ≤4 months gestEvening primrose + FO capsules, 296 mg GLA + 144 mg EPA + 80 mg DHA/dayOlive oil (without vitamin E)4 months gest until deliveryHigh BP, pre-eclampsia, eclampsia, LBW
Australia, Dunstan, 200439DB RCT, 1 hospital, n = 83Preg women <20 weeks gest diagnosed with allergic diseaseFO capsules, 1.1 g EPA + 2.2 g DHA/dayOlive oil capsules, 2.7 g n-9 oleic acid/day20 weeks gest until deliveryBW, BL, head circum, gest duration
Australia, Makrides, 201019DB RCT, 5 maternity hospitals, n = 2399Preg women with singleton pregnancies <21 weeks gestCapsules containing 800 mg DHA + 100 mg EPA/dayVegetable oil capsules without DHAEnrolment until deliveryPB, EPB, BW, LBW, NICU admissions, stillbirth, infant death
Bangladesh, Tofail, 200638DB RCT, house-to-house identification of eligible women, n = 400Preg women at 25 weeks gest4 daily capsules containing 1 g FO each, 1.2 g of DHA and 1.8 g of EPA total daily doseSoy oil capsules containing 2.25 g LA and 0.27 g LNAEnrolment until deliveryStillbirth, infant death
Denmark, Olsen, 1992,26 Salvig, 199624RCT, 1 midwife clinic, n = 533Preg women approximately 30 weeks gest visiting study clinic and were scheduled for a routine midwife assessmentFO capsules containing 2.7 g n-3 fatty acids/day1. 1 g olive oil capsules 72% oleic acid, 12% LA 2. no suppEnrolment until deliveryHigh BP, pre-eclampsia, PB, gest duration, BW, BL, stillbirth
Europe, Olsen, 200027RCT, 19 hospitals, n = 1647Preg women, >16 weeks gest, 5 prophylactic trials with uncomplicated preg who in an earlier preg had PB, IUGR, PIH or who currently had a twin preg, 2 therapeutic trials with current preg with threatening pre-eclampsia or suspected IUGR1. prophylactic trials: FO capsules, 1.3 g EPA + 0.9 g DHA/day; 2.thereapeutic trials: FO capsules, 2.9 g EPA + 2.1 g DHA/dayOlive oilProphylactic trials: 20 weeks gest until delivery; therapeutic trials: 33 weeks gest until deliveryHigh BP, pre-eclampsia, PB, EPB, IUGR, BW, LBW, SGA, gest duration, stillbirth, infant death, NICU admission
Germany, Bergmann, 200828DB RCT, n = 144Healthy preg Caucasian women ≥18 years, 21 weeks gest, willing to breast feed for ≥3 monthsMilk-based supp with vitamins and minerals, 4.5 g FOS + 200 mg DHA prepared from low EPA FO/day1. milk-based supp with vitamins and minerals/day; 2. milk-based supp + 4.5 g FOS/dayEnrolment until 37 weeks gest, optional 2 weeks–3 months postpartumGest duration, BW, BL, head circum
Mexico, Ramakrishnan, 201018DB RCT, 1 large hospital and 3 small clinics, n = 1040Preg women, 18–35 years, 18–22 weeks gest, planning to deliver at the study hospital, exclusively or predominately breast feed for ≥3 months, and live in the area for ≥2 years after deliveryCapsules containing 400 mg DHA/dayCapsules with corn/soy oil blendEnrolment until deliveryGest duration, PB, BW, BL, head circum, LBW, IUGR, stillbirth, infant death
Netherlands, Bulstra-Ramakers, 199429DB RCT, university hospital + 7 regional hospitals, n = 63Preg women with history of IUGRCapsules containing DHA + EPA (3 g EPA/day)Coconut oil capsules12–14 weeks gest until deliveryHigh BP, LBW, IUGR, PB, EPB, gest duration, stillbirth, neonatal death
Norway, Helland, 2001, 2003, 200830–32DB RCT, 2 hospitals, n = 590Preg women, 19–35 years, 17–19 weeks gest, healthy, nulli- or primiparous10 mL cod liver oil/day10 mL corn oil/dayEnrolment until 3 months postpartumPB
Spain, Sanjurjo, 200437Blinded RCTPreg women recruited during routine outpatient examsDHA + EPA (200 mg DHA/day)Placebo (unspecified)26–27 weeks gest until deliveryGest duration, BW
UK, Onwude, 199533DB RCT, 1 hospital, n = 232Preg women, 19–26 weeks gest who: 1. had a previous birth with ≥1 small babies, PIH, or an unexplained stillbirth; 2. were primiparous with abnormal uterine arcuate artery Doppler blood flow at 24 weeks gest2.7 g MaxEpa containing 1.62 g EPA + 1.08 g DHA/dayMatching air-filled capsulesEnrolment until 38 weeks gestHigh BP, pre-eclampsia, PB, EPB, LBW, IUGR, neonatal death, stillbirth
US, Smuts, 200334DB RCT, 1 hospital, n = 291Preg women, 16–36 years, 24–28 weeks gest with singleton gest12 DHA enriched eggs/week (mean of 133 mg DHA/egg)12 ordinary egg/week (mean of 33 mg DHA/egg)Enrolment until deliveryGest duration, PB, BW, LBW, BL, head circum, NICU admission
US, Judge, 200736DB RCT, n = 30Preg women, 18–35 years, <20 weeks gestCereal based bars, 300 mg DHA as low EPA FO; 3, 5, or 7 times/weekCereal based bars containing corn oil; 3, 5, or 7 times/week24 weeks gest until deliveryGest duration, BW, head circum, infant length
US, Harper, 201035DB RCT,13 centres, n = 852Preg women, 16–21 6/7 weeks gest, with a history of prior spontaneous singleton PB and a current singleton gestCapsules, 1.2 g EPA + 0.8 g DHA/day + intramuscular 17 α-hydroxy-progesterone caproate injectionsCapsules containing a minute amount of inert mineral oil + intramuscular 17 α-hydroxy-progesterone caproate injectionsEnrolment until 36 6/7 weeks gest (or delivery if occurred first)PB, EPB, LBW, SGA, NICU admission

Effect of n-3 LCPUFA supplementation in pregnancy on select MNCH outcomes

A summary of results from the meta-analyses and the grading of overall evidence are presented in Table 2. n-3 LCPUFA supplementation in pregnancy did not influence the occurrence of high maternal blood pressure or pre-eclampsia (Figures 2,3). A pooled estimate from eight trials indicated that there was no significant difference in gestational duration between the n-3 LCPUFA and placebo groups (Figure 4). The risk of preterm birth and early preterm birth was generally lower in the n-3 LCPUFA groups; however, only the risk of early preterm birth was statistically significantly lower in the n-3 LCPUFA group (RR = 0.74; [95% CI 0.58, 0.94]) (Figures 5,6). Birthweight was modestly higher among infants born to women in the n-3 LCPUFA group (RR = 42.2; [95% CI 14.8, 69.7]); however, risk of LBW was not significantly different (RR = 0.92; [95% CI 0.83, 1.02]) (Figures 7,8). Birth length, head circumference and risk of small for gestational age (SGA) or intrauterine growth restriction (IUGR) were similar between treatment groups (Figures 9–11). Risk of admission to the neonatal intensive care unit, stillbirth and infant death was also similar between treatment groups (Figures 12–14).

Table 2.  Summary of findings and overall assessment of quality of evidence of the effect of n-3 LCPUFA in pregnancy on select MNCH outcomes
No. of trialsQuality assessmentSummary of findings
Quantitative heterogeneityConsistency of resultsGeneralisable to resource-poor populationsGeneralisable to intervention of interestNo. of participantsPooled estimate
  • Table format adapted from Walker, Fischer-Walker, Bryce et al., International Journal of Epidemiology, 2010;39:i21–i31 and Cochrane Review Manager.

  • a

    Risk ratio (95% CI).

  • b

    Mean difference (95% CI).

  • IUGR, intrauterine growth restriction; LBW, low birthweight; LCPUFA, long-chain polyunsaturated fatty acids; SGA, small for gestational age.

Risk of high maternal blood pressure: Overall quality of evidence grade = moderate/low  
5I2 = 0%; P = 0.64; unclear allocation concealment in several trialsAll trials reported null findings4 of the 5 trials were conducted in high income countries4 of the 5 trials provided n-3 LCPUFA capsules of varying doses; 1 trial provided fish oil + evening primrose oil18311.09 [0.90, 1.33]a (fixed effects model)
Risk of pre-eclampsia: Overall quality of evidence grade = moderate/low
4I2 = 29%; P = 0.24; unclear allocation concealment in 1 trialAll trials reported null findings3 of the 4 trials were conducted in high income countries3 of the 4 trials provided n-3 LCPUFA capsules of varying doses; 1 trial provided fish oil + evening primrose oil16830.86 [0.59, 1.27]a (fixed effects model)
Mean difference in gestational duration (days): Overall quality of evidence grade = moderate
8I2 = 40%; P = 0.08; unclear allocation concealment in several trials; 1 trial excluded 4 babies born <36 weeks7 of the 8 trials reported null findings; 1 reported longer gestational duration in the fish oil group7 of the 8 trials were conducted in high income countries3 of the 8 trials provided n-3 LCPUFA in a food or drink vehicles; 5 provided n-3 LCPUFA capsules of varying doses28020.87 [−0.11, 1.84]b (fixed effects model)
Risk of preterm birth (<37 weeks): Overall quality of evidence grade = moderate
9I2 = 0%; P = 0.66; unclear allocation concealment in several trialsAll trials reported null findings. Direction of effect favoured n-3 LCPUFA in 7 of the 9 trials8 of the 9 trials were conducted in high income countries8 of the 9 trials provided n-3 LCPUFA capsules of varying doses65050.91 [0.82, 1.01]a (fixed effects model)
Risk of early preterm birth (<34 weeks): Overall quality of evidence grade = moderate
5I2 = 0%; P = 0.42; unclear allocation concealment in 1 trial4 of the 5 trials showed a tendency toward lower risk in the n-3 LCPUFA group; 1 was statistically significantAll trials were conducted in high income countriesAll trials provided n-3 LCPUFA capsules of varying doses43430.74 [0.58, 0.94]a (fixed effects model)
Mean difference in birthweight (g): Overall quality of evidence grade = moderate
9I2 = 27%; P = 0.21; unclear allocation concealment in several trialsDirection of effect in 7 trials favoured n-3 LCPUFA; 2 trials reported statistically significant findings8 of the 9 trials were conducted in high income countries6 of the 9 trials provided n-3 LCPUFA capsules of varying doses; 3 trials provided n-3 LCPUFA in a food or drink vehicles602042.22 [14.76, 69.68]b (fixed effects model)
Risk of low birthweight: Overall quality of evidence grade = moderate
8I2 = 11%; P = 0.34; unclear allocation concealment in several trials1 trial significantly favoured n-3 LCPUFA; 7 others reported null findings6 of the 8 trials were conducted in high income countries7 of the 8 trials provided n-3 LCPUFA capsules of varying doses65110.92 [0.83, 1.02]a (fixed effects model)
Mean difference in birth length (cm): Overall quality of evidence grade = moderate/low
6I2 = 54%; P = 0.06; unclear allocation concealment in several trialsAll trials reported null findings5 of the 6 trials were conducted in high income countries3 of the 6 trials provided n-3 LCPUFA capsules of varying doses19890.27 [−0.13, 0.67]b (random effects model)
Mean difference in head circumference (cm): moderate/low
5I2 = 87%; P < 0.001; unclear allocation concealment in several trials4 trials reported null findings4 of the 5 trials were conducted in high income countries2 of the 5 trials provided n-3 LCPUFA capsules of varying doses; 3 trials provided n-3 LCPUFA in a food or drink vehicles1456−0.21 [−0.84, 0.42]b (random effects model)
Risk of SGA or IUGR: moderate
5I2 = 0%; P = 0.69; unclear allocation concealment in 1 trialAll trials reported null findings4 of the 5 trials were conducted in high income countriesall trials provided n-3 LCPUFA capsules of varying doses34611.06 [0.92, 1.21]a (fixed effects model)
Risk of admission to neonatal intensive care unit: low
4I2 = 90%; P < 0.001; unclear allocation concealment in 1 trial1trial significantly favoured n-3 LCPUFA; 3 reported null findingsAll trials were conducted in high income countries3 of the 4 trials provided n-3 LCPUFA capsules of varying doses56650.85 [0.61, 1.20]a (random effects model)
Risk of stillbirth: moderate/low
8I2 = 0%; P = 0.51; unclear allocation concealment in several trials; relatively few events (n = 30 total)All trials reported null findings6 of the 8 trials were conducted in high income countriesAll trials provided n-3 LCPUFA capsules of varying doses70380.80 [0.50, 1.26]a (fixed effects model)
Risk of infant death: moderate/low
6I2 = 0%; P = 0.89; unclear allocation concealment in several trials; relatively few events (n = 18 total)5 trials showed a tendency toward favouring n-3 LCPUFA; one toward favouring placebo4 of the 6 trials were conducted in high income countriesAll trials provided n-3 LCPUFA capsules of varying doses62350.69 [0.38, 1.23]a (fixed effects model)
Figure 2.

Effect of n-3 LCPUFA supplementation during pregnancy on risk of high maternal blood pressure. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 3.

Effect of n-3 LCPUFA supplementation during pregnancy on risk of pre-eclampsia. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 4.

Effect of n-3 LCPUFA supplementation during pregnancy on gestational duration. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 5.

Effect of n-3 LCPUFA supplementation during pregnancy on risk of pre-term birth (<37 weeks). LCPUFA, long-chain polyunsaturated fatty acids.

Figure 6.

Effect of n-3 LCPUFA supplementation during pregnancy on risk of early pre-term birth. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 7.

Effect of n-3 LCPUFA supplementation during pregnancy on birthweight. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 8.

Effect of n-3 LCPUFA supplementation during pregnancy on risk of low birthweight. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 9.

Effect of n-3 LCPUFA supplementation during pregnancy on birth length. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 10.

Effect of n-3 LCPUFA supplementation during pregnancy on head circumference. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 11.

Effect of n-3 LCPUFA supplementation during pregnancy on risk of SGA or IUGR. IUGR, intrauterine growth restriction; LCPUFA, long-chain polyunsaturated fatty acids; SGA, small for gestational age.

Figure 12.

Effect of n-3 LCPUFA supplementation during pregnancy on risk of admission to the neonatal intensive care unit. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 13.

Effect of n-3 LCPUFA supplementation during pregnancy on risk of stillbirth. LCPUFA, long-chain polyunsaturated fatty acids.

Figure 14.

Effect of n-3 LCPUFA supplementation during pregnancy on infant death. LCPUFA, long-chain polyunsaturated fatty acids.

We conducted subgroup analyses including only high risk pregnancy trials (n = 4)27,29,33,35 and found no between-group differences in the risk of pre-eclampsia (2 trials; RR = 0.80, [95% CI 0.50, 1.29]), preterm birth (4 trials; RR = 0.88, [95% CI 0.76, 1.02]), LBW (4 trials; RR = 0.90, [95% CI 0.76, 1.08]), SGA (2 trials; RR = 0.98, [95% CI 0.74, 1.30]), stillbirth (2 trials; RR = 0.68, [95% CI 0.11, 4.08]), and neonatal death (3 trials; RR = 1.01, [95% CI 0.32, 3.24]) (data not shown). Additionally, when high risk trials were excluded from the main meta-analyses, the direction and strength of the associations did not change significantly (data not shown).

Results from the general review

Overview of included observational studies

Fourteen observational studies examining fish and/or seafood intake during pregnancy were identified41–55 (Table 3). Eleven of the studies were prospective cohorts, and three were retrospective in design. The studies included a total of 114 006 women and all used questionnaires, most commonly validated Food Frequency Questionnaires (FFQs), to measure fish and/or seafood intake. The majority of studies took place in high income countries (US, Denmark, France, Canada, Iceland, Norway, England). Most studies included cohorts recruited from more than one centre, ranging from two to nine centres.

Table 3.  Description of studies included in the general review of the effect of n-3 LCPUFA in pregnancy on maternal and child health outcomes
First author, publication yearStudy designNumber of subjectsParticipant attributesTiming of dietary assessment questionnaireaMean intakeOutcomes measuredConclusionsAssociation (Yes/No)
  • a

    Validated Food Frequency Questionnaires were used in many, but not all studies.

  • b

    Estimates were adjusted for potential confounding factors.

  • aOR, adjusted odds ratio; BL, birth length; BP, blood pressure; BW, birthweight; circum, circumference; FFQ, Food Frequency Questionnaire; gest, gestation; HC, head circumference; IUGR, intrauterine growth restriction; LBW, low birthweight; LCPUFA, long-chain polyunsaturated fatty acids; PB, preterm birth; SGA, small for gestational age.

Drouillet, 200945Prospective cohort/Retrospective questionnaires1805French women <24 weeks gest, singleton pregnancy1) Asked at recruitment about diet in the year before pregnancy;Fish consumed 8.2 times/month (last 3 months of pregnancy)BW, BL, HCNo association between seafood intake and newborn anthropometric measures. Positive associations between higher seafood intake before pregnancy and BW, BL and HC reported in overweight womenbNo (BW, BL, HC)
2) After delivery
Guldner 200753Prospective cohort2398French women living in Brittany1st trimesterFish (20.4 g/day) Shellfish (19.7 g/day)BW, gest duration, LBW, PB, SGAHigher fish intake associated with an increase in gest duration of 1.03 days [95% CI 0.10–1.97]; high shellfish (but not fish) intake associated with increased risk of SGAbNo (BW, LBW, PB, SGA);
Yes (gest duration)
Yes (SGA- select group)
Halldorsson, 200752Prospective cohort44 824Danish women, fluency in Danish language25 weeks gestFish fat = 7.6 g/day (in highest fish intake group)BW, BL, HC, gest duration, SGAHighest intake of fatty fish positively associated with risk of SGA (aOR = 1.18, [95% CI 1.03, 1.35]) and inversely associated with BW, BL and HC; no significant associations found for lean fishbYes – fatty fish (BW, BL, HC, SGA)
No (gest duration)
Haugen, 200855Prospective cohort26 125Norwegian women, 21–38 years, BMI = 19–32, singleton pregnancy18–22 weeks gest66% consumed fishPBFish intake ≥2 times/week was associated with a decrease in risk of PB: (aOR: 0.84; [95% CI 0.74, 0.95])bYes (PB)
≥2 times/week
Mendez, 201042Prospective cohort657Spanish women with high seafood intakes1) 1st ultrasound visit (mean 14 weeks);88 g seafood/weekBW, gest duration, SGAHigher intake of crustaceans and canned tuna was associated with increased risk of SGA; higher intakes of other fish not associated with SGAbYes (SGA) No (BW, gest duration)
2) Trimester 3
Muthayya, 200946Prospective cohort676Indian women, age 17–40 years, <20 weeks gest, singleton pregnancy1) Trimester 1;EPA intake ≈ 2–3 mg/dayLBWWomen who did not eat fish during the 3rd trimester had a significantly higher risk of delivering a LBW baby (OR: 2.49, P = 0.019)bYes (LBW)
2) Trimester 2;
3) Trimester 3DHA intake ≈ 10–11 mg/day
Oken, 200447Prospective cohort2109Women in US, <22 weeks gest, singleton pregnancy1) Trimester 1;EPA + DHA = BW, gest duration, LBW, PB, SGAHigher seafood intake in the 1st trimester was inversely associated with BW (94 g lower BW in the highest seafood intake group)bYes (BW); No (gest duration, LBW, PB, SGA)
2) Trimester 2;0.16 g/day
3) Trimester 3
Olsen, 199041Prospective cohort11 980Danish women36 weeks gestation9% of women consumed 5 + fish meals/month (reference)BW, BL, gest durationIn non-smokers, BW and HC were positively associated with maternal fish consumptionbYes (BW, HC); No (gest duration, BL)
Olsen, 199349Retrospective cohort1362Danish womenAfter delivery3.3 seafood meals per weekBW, BL, gest durationIntake of 3 + seafood meals/week was associated with increased BW and BL (P = 0.001), but not gest durationbYes (BW and BL); No (gest duration)
Olsen, 199543Prospective cohort965Danish Women30 weeks gestation25 g/day fishBW, BL, gest durationHigher intake of n-3 LCPUFA not associated with BW (P = 0.4), BL (P = 0.5) or gest duration (P = 0.5)bNo (BW, BL, gest duration)
Olsen, 2002, 200644,48Prospective Cohort8729Danish women, ≤16 weeks gest, singleton pregnancy1) 16 weeks gestation;15.8 g fish and 0.182 g n-3 LCPUFA per day.Gest duration, PB, LBW, IUGRZero fish meals at 16 weeks was associated with decreased gest duration (3.91 days shorter; [95% CI 2.24, 5.58]) and higher risk of PB and LBWbYes (gest duration, PB, LBW);
2) 30 weeks gestation
No (IUGR)
Popeski, 199150Retrospective cohort300Inuit womenAfter delivery (a subset of 27 women reported seafood intake)Not reportedHigh BP, BW, gest durationLower diastolic BP in higher seafood group (78.2 mmHg) vs. lower group (81.5 mmHg), P < 0.005Yes (high BP);
No (BW and gest duration)
Rogers, 200454Prospective cohort11 585Women from southwest England32 weeks gestation32.8 g/day fishBW, PB, LBW, gest duration IUGRHigher risk of IUGR in those eating no fish (aOR: 1.37, [95% CI 1.02, 1.84])bNo (BW, PB, gest duration);
Yes (IUGR)
Thorsdottir, 200451Retrospective cohort491Icelandic women, age 20–40 years, 38–43 weeks gest, singleton pregnancyAfter delivery47 g/day (total fish)BW, BL, HCLowest quartile of fish consumption was associated with lower BW, BL and HC. Higher levels of fish liver oil consumption were associated with lower BL and HCbYes (BW, BL, HC)

Food Frequency Questionnaires were administered once in nine of the studies, and at least twice in five studies. In 11 studies, FFQs were administered during pregnancy, and three of the studies collected the dietary information following delivery. Mean fish intake was reported as g per day or number of fish meals consumed. Consumption ranged from 3.4 g to 47 g fish per day, or eight to 12 fish meals per month. One study calculated only EPA and DHA consumption, and one study did not report intake. Reported outcomes included birthweight, birth length, head circumference, gestational duration, preterm birth, LBW, IUGR or SGA, and maternal hypertension. Nearly all included observational studies reported estimates adjusted for potential confounding factors.

Findings from observational studies

All studies but one55 reported outcome information on birthweight or LBW, and six studies found either a positive association between fish consumption and birthweight,41,49,51 or an inverse association between fish consumption and LBW or IUGR.44,46,48,54 In the largest study (n = 44 824), Halldorsson et al.52 reported an increased risk of SGA in babies of women who consumed >60 g/day of fatty fish compared with ≤5 g/day (odds ratio = 1.24; [95% CI 1.03, 1.49]). There was no relationship between lean fish consumption and occurrence SGA. Three other studies reported increased risk of SGA or lower birthweight among groups consuming the highest amounts of seafood; however, this relationship was often specific to certain types of seafood (e.g. high intake of crustaceans/shellfish, but not fish).42,47,53 Birth length and head circumference were assessed in six and three studies, respectively, and findings varied. Eight of ten studies that examined gestational duration did not find an association with fish consumption; however, two studies reported associations.44,53 Olsen et al. found an increase of 4 days in gestational duration among infants of women who consumed fish in the first two trimesters of pregnancy (n = 8729).44,48 Guldner et al. reported a decrease of gestational age by one day in women who consumed fish less than once a month compared with women who consumed fish twice a week (n = 2398).53 Two of the larger prospective cohort studies found a higher occurrence of preterm birth in groups that consumed less fish.55,56

Comments

Findings from this meta-analysis indicate that babies born to mothers supplemented with n-3 LCPUFA during pregnancy had modestly higher mean birthweight. Although babies in the n-3 LCPUFA group were heavier, this did not translate into a clear significant reduction in LBW (RR = 0.92; [95% CI 0.83, 1.02]). The risk of early preterm birth was significantly lower in the n-3 LCPUFA supplemented group, and the risk of preterm birth was borderline non-significant (RR = 0.91; [95% CI 0.82, 1.01]). n-3 LCPUFA supplementation in pregnancy afforded no benefits on other MNCH outcomes including high maternal blood pressure, pre-eclampsia, gestational duration, birth length, head circumference, IUGR or SGA, stillbirth and infant death.

Findings from the present meta-analysis are generally in agreement with those from the 2006 Cochrane review conducted by Makrides et al. (n = 6 included trials with a total of 2783 women).15 In the 2006 Cochrane review, infants born to n-3 LCPUFA-supplemented mothers were 31% less likely to be born very early (<34 weeks), and had a higher mean birthweight than controls (mean difference = 47 g). Although we included nine additional trials in the present meta-analysis and had different inclusion criteria than the Cochrane review, we also found a decreased risk of early preterm birth and a higher mean birthweight in the n-3 LCPUFA group. Makrides et al. found that women supplemented with n-3 LCPUFA experienced longer gestational duration than controls (2.6 days longer). We did not find this effect on gestational duration after including five additional trials in the meta-analysis. Similar to our analysis, these authors found no significant effect of n-3 LCPUFA on other MNCH outcomes such as pre-eclampsia, stillbirth and infant death.

Two additional meta-analyses of prenatal n-3 LCPUFA supplementation were conducted and published within a year of the Cochrane review. A meta-analysis by Szajewska et al. focused on low-risk pregnancies, while one by Horvath et al. focused on high-risk pregnancies.16,17 Although the analysis by Szajewska et al. had different inclusion and exclusion criteria than the Cochrane review, authors arrived at similar conclusions about gestational duration, preterm birth (<37 weeks) and LBW. The Szajewska et al. review included six trials with a total of 1278 women and showed that n-3 LCPUFA supplementation during pregnancy prolonged gestation by 1.57 days [95% CI 0.35, 2.78 days], but had no effect on preterm birth, LBW or other birth outcomes. The Horvath et al. review focused on high-risk pregnancies and included four studies with 1264 women. The analysis of gestational duration included two studies with 295 women and showed no difference in duration of pregnancy >37 weeks between groups (RR: 0.99, [95% CI 0.9, 1.1]). Likewise, the analysis of preterm delivery <37 weeks including three trials and 523 women showed no difference between groups (RR: 0.82, [95% CI 0.6, 1.1]). Similar to findings of the Cochrane review and of our current meta-analysis, this analysis indicated that supplementation with n-3 LCPUFA decreased the risk of early preterm delivery (<34 weeks) (RR: 0.39, [95% CI 0.18, 0.84]). Other maternal and neonatal outcomes such as rates of pre-eclampsia, LBW, IUGR and birthweight did not differ by intervention group.

Overall, the three previous meta-analyses showed that supplementation with n-3 LCPUFA is safe in pregnancy and is generally well tolerated. Rates of serious adverse events were similar between treatment groups and occurrence of side effects (e.g. vomiting, nausea and diarrhoea) was generally similar except for belching and bad taste, which occurred more frequently in the n-3 LCPUFA-supplemented groups. The Makrides and Szajewska analyses showed a prolongation of gestational duration with n-3 LCPUFA supplementation, and the Makrides et al., Szajewska et al., Makrides et al., and Horvath et al. reviews showed a decreased risk of early preterm delivery (<34 weeks) with n-3 LCPUFA supplementation.

Several studies have examined relationship between biomarkers of n-3 LCPUFA status and fetal growth and gestational duration. Among women in a fishing community in the Faroe Islands, Grandjean et al. found that higher cord serum concentrations of DHA were associated with increased gestational duration.57 In a small study by Olsen et al., higher ratios of n-3 LCPUFA to arachidonic acid in erythrocytes were associated with an increase in gestational duration by 5.7 days in 37 Danish women ([95% CI 1.4–10.1 days]; P = 0.02).58 Lucas et al. observed increased gestational duration in 454 Nunavik newborns with higher concentrations of n-3 LCPUFA in cord plasma.59 A study by van Eijsden et al. examined the relationship between fetal growth and varying concentrations of plasma phospholipids.60 Lower concentrations of n-3 LCPUFA were associated with lower birthweight (125 g lighter) and two times higher risk of SGA births compared with women with higher concentrations. Results from our meta-analysis are not in accord with these findings of longer gestational duration with higher n-3 LCPUFA status; however, the observations of a positive association between birthweight and n-3 LCPUFA status support our findings.

Limitations

Like many meta-analyses, the present analyses are limited by heterogeneity in the trial designs. The n-3 LCPUFA supplement varied by trial and variations included (1) the type of n-3 LCPUFA provided, (2) the doses of DHA and EPA provided, and (3) the vehicle in which the n-3 LCPUFA was provided. Doses of DHA varied by trial and ranged from 80 mg/day to 2.2 g/day. Many of the trials provided fish oil capsules, while some provided foods such as eggs, a milk-based supplement, and cereal bars with added n-3 LCPUFAs. The duration of n-3 LCPUFA supplementation was different among the trials, and the timing of supplementation in pregnancy differed. Finally, our meta-analysis included supplementation in both low and high risk pregnancies. A strength of our analysis is that many of the outcomes had quite large samples (e.g. >6000 babies were included in the analysis of birthweight), and statistical heterogeneity was low for many outcomes. Although the quality of included trials was generally moderate, several trials were potentially biased because of issues such as inadequate or unknown allocation concealment. Results describing more rare events such as infant death and stillbirth should be interpreted with care because of the relatively few reported events.

Conclusions, recommendations, and research gaps

Results from this meta-analysis indicate that babies born to mothers supplemented with n-3 LCPUFA had a lower risk early preterm birth and were modestly heavier than babies born to placebo-supplemented women. The clinical significance of an increase in mean birthweight of 42 g is, however, unknown because the risk of LBW was not clearly lower (RR = 0.92, [95% CI 0.83, 1.02]) in the n-3 LCPUFA group. A low prevalence of LBW in many of the included trial populations might contribute to a lack of significant effect. Our review of observational studies revealed mixed findings; several studies revealed that higher seafood intake was associated with higher birthweight and lower risk of preterm birth, several studies showed no association between fish intake and MNCH outcomes, and a few of the studies showed an increased risk of SGA with high consumption of particular types of seafood (e.g. crustaceans).

Pregnant women in many low, middle and high income countries do not achieve the recommended intake of at least 200 mg of DHA per day, with the exception of coastal countries where fish and other marine foods are easily accessible, affordable and commonly consumed.7 For example, studies among pregnant women in India and Bangladesh report very low mean intakes of DHA (11 mg and 30 mg per day, respectively).61 Examination of LCPUFA content in breast milk reveals large variations in DHA concentrations in human milk worldwide, indicating differences in dietary intake of DHA.62 Furthermore, pregnant women in many low income countries do not consume the recommended amount of total energy as fat (20–35%).63 Studies among pregnant women in rural Bangladesh and Burkina Faso report mean intakes of 8% and 13% of total energy as fat, respectively.64,65 Many women living in under resourced settings not only consume low amounts of DHA and total fat, but also the essential fatty acids alpha-linolenic acid (ALA) and linoleic acid (LA), which are crucial for optimal growth and development.61 In non-coastal, under resourced settings, intake of n-3 LCPUFA supplements or increased consumption of marine foods in pregnancy is likely not economically feasible. Since 2008, the Chilean government has made DHA- + EPA-fortified powdered milk available to pregnant and lactating women.66 This cost-effective option might serve as a feasible solution to improving n-3 LCPUFA intake in resource-poor settings, though the effect of prenatal n-3 LCPUFA supplementation on MNCH outcomes in low income countries is largely unknown. However, because studies have shown that n-3 LCPUFA supplementation is safe in pregnancy, dietary counselling promoting consumption of marine foods, where feasible, could potentially benefit pregnant women by also increasing their intake of other important nutrients found in marine foods (e.g. protein, vitamin A and iron). Generally, consuming the recommended two fish meals per week does not pose a threat of adverse pregnancy outcomes due to environmental contaminants.67

Most randomised trials of prenatal n-3 LCPUFA supplementation have been conducted in high income countries and, therefore, our findings are not generalisable to populations in low- and middle-income settings. The recent trial in Mexico (1094 women) showed that prenatal DHA supplementation had no overall effect on maternal and newborn health outcomes; however, this study was conducted in a middle-class, fairly well educated urban population where the rate of LBW was low (<6%).18 The trial conducted in Angola was relatively small and was conducted in 1986 and is therefore not likely representative of all current low- and middle-income populations.25 The Bangladesh trial reported complete data on very few outcomes of interest for this meta-analysis.38 Further research is needed to determine any benefit of prenatal n-3 LCPUFA supplementation in populations in low income settings, especially where (1) intake of marine foods is low, and (2) the rate of adverse maternal and neonatal health outcomes, including LBW, is high. Several prenatal n-3 LCPUFA supplementation trials have suggested a benefit on subsequent infant and child outcomes such as cognitive development; however, results have been mixed and inconclusive. The Mexico 2005 trial revealed that infants in the n-3 LCPUFA group were generally healthier in the first 6 months of life compared with controls,68 but the outcome of infant morbidity has generally not been explored in other trials. Some studies have shown a potential benefit of prenatal fish oil supplementation on maternal mood, but results are also inconclusive.69

In conclusion, infants born to women supplemented with n-3 LCPUFA during pregnancy were modestly heavier and were less likely to be born very early (<34 weeks), compared with controls. More RCTs in resource-poor settings, where n-3 LCPUFA and essential fatty acid intakes are often low, would provided needed information about the potential benefits of supplementation in lower and middle-income countries.

Conflicts of interests

The authors declare that they have no competing interests.

Appendix

Appendix 1


Trials or data excluded from the meta-analysis

First author, yearReason for exclusion
  1. APGAR, Activity, Pulse, Grimace, Appearance, Respiration; LBW, low birthweight; MNCH, maternal, neonatal, and child health; PUFA, polyunsaturated fatty acids.

DeGroot, 200412The intervention was α-linolenic acid.
Helland, 2001, 200330,32The outcomes of birthweight, birth length, and head circumference were not included in our meta-analysis because 42% of randomised women (n = 249) were excluded because of either exclusion by the authors or attrition. The outcome of preterm birth was reported for the total sample (n = 590) and therefore included in our meta-analysis.
Innis, 200870The authors reported findings on a select group of infants. LBW and preterm infants were excluded from analyses.
Knudsen, 200671There was no in-person contact between participants and investigators (fish oil supplements were sent by mail). There was no discussion of blinding and allocation concealment.
Krauss-Etschmann, 2007, 200814,40Standard deviations were not published.
Malcolm, 2005, 200372Outcomes of interest were reported for only a select group of infants. Preterm babies and infants with low APGAR score were excluded from analyses.
Mardones, 200873n-3 + n-6 PUFA given as the intervention. The experimental group received higher doses of certain micronutrients than the control group.
Mihrshahi, 200374This intervention was in kids.
Olsen, 199075This large (n = 5022), controlled trial conducted in the UK in 1938 was not included in our analysis because the experimental group received multiple micronutrients, in addition to halibut liver oil. Therefore, any between-group differences in MNCH outcomes cannot solely be attributed to fish oil. The control group received no supplements.
Smuts, 200376This was a feasibility study.
Su, 200877None of our MNCH outcomes were included.
Tofail, 200638Birthweight, birth length, and head circumference were reported for only a subset of infants (those remaining in the trial through 10 months). Authors reported infants lost to follow-up at 10 months had smaller birth size, but did not report anthropometric measures for all 324 births.

Ancillary