Overview

A total of 117 reports, describing 89 unique studies, investigated questions pertinent to this systematic review of the evidence concerning the effects of omega-3 fatty acids on child and maternal health. The questions regarding the influence of the intake of omega-3 fatty acids on pregnancy outcomes, such as duration of gestation, preeclampsia, eclampsia or gestational hypertension and infants SGA were address separately, since RCTs were identified that answered each of these questions separately.

The questions regarding the child's outcomes, such as growth patterns, neurological development, cognitive development and visual function are divided in a series of questions: one question is related to the maternal intake of omega-3 fatty acids for each outcome; another question is associated with the infant's intake of human milk; two questions have been lumped together regarding the infant's intake of formula, with or without breast milk; a separate question addressed the infant's intake of omega-3 fatty acids from other sources (diet, supplements); and, a final set of questions relate to biomarkers in maternal, fetal or infant's blood, and the association with the clinical outcomes.

For each group of outcomes, we present a synthesis of the key findings with respect to each question. This includes a critical appraisal of the group of trials from which results are drawn. The broader implications of these findings, including potential future research, are highlighted. We begin with the safety issues concerning all the included studies.

Evidence Synthesis and Appraisal

Adverse events, contraindications, and intolerance are often under-reported in human experimental studies. Many studies do not report any data on adverse events, and so it is frequently not clear whether or not an adverse event had actually occurred in these studies. Furthermore, even if a study reports an adverse event, the study authors do not always state explicitly if this adverse event was related to the study intervention or some other factor(s). An additional problem that aggravates the assessment of adverse event data, is that some authors do not clarify whether the number of adverse events reflects the total number of event occurrences across all patients (i.e., a single patient may experience more than one adverse event during the study period), or the number of patients who had experienced at least one adverse event. This information should be reported in order to distinguish between the two scenarios.

Overall, omega-3 fatty acids supplementation in pregnant women, breastfeeding mothers and preterm and term infants, was very well tolerated and did not generate any serious adverse events across the included RCTs. The safety data was reported in 21 RCTs.

In pregnant women, the adverse events related to the omega-3 fatty acids intake were mild and transient, with nausea and gastrointestinal discomfort being the most commonly reported.^{230, 233}

For both the term and preterm population, change in number of stools and flatulence were the most common adverse events related to the omega-3 supplemented formulas. However, most of the serious adverse events were related to the fact that the infants were premature with low birth weights, which increases the occurrence of necrotizing enterocolitis (NEC), bleeding problems, infections and respiratory failure, among others in the case of preterm infants.^{104, 182, 193, 201, 203, 205, 207, 212, 218, 227, 251, 257, 258, 261, 265, 266, 268, 273, 286, 287} In general, none of the withdrawals were due to the interventional formula.

Fifteen average poor quality (Jadad: 2.8/5) RCTs addressed the question of the influence of omega-3 fatty acids intake during pregnancy on the duration of gestation.^{31, 41, 288, 290, 291, 293–295 295, 296} Seven trials included otherwise healthy pregnant women,^{141, 196, 209, 231, 232, 234, 235} the remaining eight studies included a high-risk population of pregnant women, yet with different types of risk factors (i.e., IUGR, premature delivery, preeclampsia, etc). Ten studies did not find a significant difference between intervention groups in the duration of gestation measured as mean of gestational age at delivery.^{141, 196, 230–235} However, four average poor quality (Jadad score 2/5) studies observed that the omega-3 fatty acid group had a significantly greater duration of gestation after treatment compared with the unsupplemented group.^{209, 230}

Omega-3 fatty acids did not have a significant effect on the proportion of premature deliveries in ten studies.^{31, 209, 233, 234, 238} Only Smuts et al. observed a noticeable lower percentage of premature deliveries in mothers taking omega-3 fatty acid supplements, yet this study was underpowered (small sample) to measure the statistical significance of such observation.²³²

Other variables, such as length of the intervention and background diet, were different among the identified trials. Most studies began the treatment during the second trimester of pregnancy,^{141, 196, 230, 231, 233, 235, 238} while the remaining trials enrolled their subjects during the third trimester. Fish consumption in the background diet, one of the most important effect modifiers, was used as a covariate in only one trial.²⁰⁹ After adjusting for this effect modifier, the results did not change, and the fish oil group still had a longer duration of gestation than the olive oil group.²⁰⁹

Other covariates used to control the results were the compliance with the intervention,²⁰⁹ current smoking status,^{233, 234} as well as maternal BMI and number of prior pregnancies.²³⁴ The only variable that had an impact on the results was the smoking status in Smuts et al's study.²³⁴ The duration of gestation was significantly longer in the high-DHA group in the nonsmokers.²³⁴

Meta-analysis of the incidence of premature deliveries was performed pooling the data of eight RCTs that compared the use of capsules containing DHA+EPA,^{31, 41, 291} and two trials using high DHA eggs^{294, 296} with control group. Both meta-analysis failed to find a statistical difference between groups. The limitation of combining the studies using DHA+EPA versus control, is that the population of pregnant women included in seven trials was high risk for premature delivery in different ways (twin pregnancy,³¹ antecedent of premature delivery,³¹ antecedent of GHT and IUGR,^{31, 291} and threatening pre-eclampsia³¹). Only one study included healthy Danish women.⁴¹ Subgroup analysis was not possible given the lack of individual data for each of the six RCTs included in Olsen et al. 2000.³¹ Another limitation of this approach is the length of intervention. While five trials started in the second trimester of pregnancy,^{31, 291} three began the intervention during the third trimester (shorter period of time and likely not meaningful to see a significant effect).^{31, 41}

These findings suggest that there is inconsistent evidence of the use of omega-3 fatty acids supplements during the second or third trimester of pregnancy to reduce the incidence of premature pregnancies in high and low risk populations. Nevertheless, the overall effect does not show a significant difference between study arms.

The association between the maternal biomarkers during pregnancy and the duration of gestation was assessed in four studies.^{234, 239–241} The study by Smuts et al. was an RCT that compared the use of DHA-enriched eggs intake with ordinary eggs in healthy pregnant women.²³⁴ This study did not observe a significant correlation between the maternal RBC content of DHA and the duration of gestation, however, the study found a significantly positive correlation between the infant RBC DHA at birth and this pregnancy outcome.²³⁴

Three observational trials,^{239, 240} found a significantly positive association between the maternal plasma content of AA (at 34–35 weeks of GA) and the duration of gestation, whereas, Rump et al.'s cross-sectional study did not find any correlation between maternal biomarker content and duration of gestation.²⁴¹ The study by Elias and Innis was a single prospective cohort of pregnant women that reached a term delivery,²⁴⁰ and the study by Reece et al.²³⁹ was a case-control study that compared the maternal content of RBC omega-3 and omega-6 fatty acid biomarkers at 34 weeks of gestation and at delivery in preterm and term pregnancies. This study found that the preterm deliveries had a significantly higher content of AA (omega-6) and DPA (omega-6), reflecting a relative reduction in the omega-3 fatty acids. The omega-6/omega-3 ratio was higher in preterm deliveries or in 34-week pregnant women, compared with samples taken after term deliveries.²³⁹

These findings suggest that there is an uncertain association between the maternal biomarkers during pregnancy and the duration of gestation, independently of the maternal intake.

Eight RCTs addressing the question concerning the influence of maternal intake of omega-3 fatty acids during pregnancy in the incidence of gestational hypertension (GHT), preeclampsia or eclampsia were identified with a quality score approaching good internal validity (Jadad: 2.9/5).^{209, 230, 233, 236, 237} Six studies compared the use of fish oil supplements containing DHA and EPA with placebo (generally olive oil). The population characteristics of these studies were very diverse, since one of them included healthy Danish pregnant women,²⁰⁹ while the others included high-risk pregnant women (i.e., preeclamptic, twin pregnancies, IUGR or preeclampsia in previous pregnancies, etc).^{230, 233, 236, 237} The incidence of GHT in these populations, after the use of omega-3 fatty acids or placebo did not differ in six of seven studies.^{209, 230, 233, 237, 238} The study by D'Almeida et al. was the only poor quality trial conducted in South Africa that observed a reduction of the incidence of GHT in the magnesium oxide group, compared to the omega-3 FA supplemetation and the placebo groupsa (no significance assessed).²³⁶ Regarding the incidence of preeclampsia (triad of hypertension, edema and proteinuria), six studies showed that compared with placebo, supplementation with omega-3 fatty acids did not have a significant effect.^{230, 233, 234, 237, 238}

Only one study conducted in South Africa observed a statistically significant difference between groups, showing that the fish oil group had a lower incidence of preeclampsia compared with placebo and magnesium oxide.²³⁶

Meta-analysis was possible for the outcome related to the incidence of gestational hypertension. Two studies were included in the analysis,^{230, 233} which selected a population of women at high risk of developing GHT. The overall effect size was nonsignificant between groups.

It appears that there is some evidence to suggest that supplementation with omega-3 fatty acids during the second or third trimester of pregnancy does not reduce the incidence of gestational hypertension, preeclampsia or eclampsia in healthy or high-risk pregnant women. However, the results were not adjusted for the potential covariates or confounders, such as background diet, grade of risk for GHT or preeclampsia in the current pregnancy, smoking status, and age, among others.

No RCTs were identified to investigate the association between the omega-3 or omega-6/omega-3 ratio content of maternal biomarkers and the incidence of preeclampsia-eclampsia or gestational hypertension. We identified five observational trials that addressed this question, yet the incidence of preeclampsia could not be assessed given the study designs.^{179, 229, 242–244} Four studies selected preeclamptic women and normal pregnant women as controls.^{229, 242–244} Al et al. selected women with GHT and healthy pregnant women as controls,¹⁷⁹ and Craig-Smith et al. also included women with GHT and chronic hypertension.²⁴³ Wang et al. and Hofmann et al. found that the maternal plasma content of AA did not differ significantly between preeclamptic and normal pregnant women.^{229, 242} On the other hand, Craig-Smith et al. observed that the women with chronic hypertension had a significantly higher plasma content of AA compared with women with preeclampsia, GHT or normal pregnant women.²⁴³ Shouk et al. observed that the women with preeclampsia had a significantly higher AA content compared with normal women, although the plasma measurement was different from the other studies (mcg/L).²⁴⁴ Results regarding total PUFA content, total omega-3 fatty acids, total omega-6 fatty acids, DHA, EPA and other PUFAs did not follow a consistent pattern across the studies. The results are very inconsistent among the studies.

These discrepancies across the studies can be explained given the differences in the study designs, case ascertainment, severity of preeclampsia, appropriate technique of lipid extraction and manipulation, measurements of FA in plasma (% weight of total FA, mcg/L or mol/L) background diet, age, gestational age, and other variables like alcohol intake, tobacco use and supplements that were not assessed.

Regarding the influence of omega-3 fatty acids supplementation during pregnancy on the incidence of SGA infants, fourteen average poor quality scores (Jadad: 2.85/5) RCTs with addressed this question. The definition of SGA was diverse across the included studies, using the smaller percentile (PC) as the upper limit (i.e., PC < 3 or PC < 5 or PC < 10 for gestational age). Most of the studies evaluated the mean birth weight, instead of the incidence of SGA infants. In the majority of the studies, mean birth weight was not influenced by the intervention. Despite the fact that the selected populations in the trials were so different (e.g., high risk vs. healthy women), the results seem to be very consistent across the studies. None of the trials adjusted their results for the maternal background diet, which can be an important effect modifier.

Meta-analysis was performed for two different variables. The birth weight (mean value) was combined in two studies that were comparable in terms of type of intervention and population. The overall size of the effect was nonsignificantly different between groups (supplemented vs. unsupplemented).^{230, 233} The other outcome was the incidence of infants with IUGR in three studies,^{230, 233, 238} with a nonsignificant overall effect of supplementation during pregnancy. These findings are consistent with the results of the remaining included studies.

Six studies addressed the question regarding the association between the omega-3 or omega-6/omega-3 ratio content of maternal biomarkers and the incidence of SGA infants.^{196, 240, 241, 245–247} de Groot et al.'s RCT found a significantly positive correlation between the maternal plasma and RBC DHA content and birth weight, however, this relationship was nonsignificant when measured at delivery.¹⁹⁶ Among the observational studies, three investigators compared the maternal biomarker content in women at risk of IUGR with healthy controls.^245–247 Two of them found that the women with IUGR fetuses had a significantly lower content of LA (omega-6) in the plasma.^{246, 247} The content of DHA, EPA, AA, total omega-3 and omega-6 fatty acids, however, did not show a constant pattern across the studies. Two observational studies did not observe a correlation between maternal plasma biomarkers and birth weight,^{241, 247} consistent with the result in the RCT.¹⁹⁶ Elias and Innis did not define the birth weight for GA, so their results are difficult to interpret in the context of correlation of maternal PUFA with SGA infants.³⁴⁶

These discrepancies in the study results may be due to many variables that play a relevant role in the lipid profile, such as population characteristics (healthy pregnant women, high risk of IUGR, women with IUGR), background diet, lipid extraction and manipulation, lipid fraction (TGL, PL, CE), and timing of drawing the blood samples.

No studies were identified to address the question of the influence of the omega-3 fatty acids from sources other than formula or human milk, and any of the child's clinical outcomes (e.g., growth patterns, neurological and cognitive development, and visual function).

One good quality RCT addressed the question of the influence of maternal omega-3 fatty acids intake during pregnancy on the growth patterns outcomes.¹⁴¹ There was no statistical differences between infants from mothers that were taking the supplementation with omega-3 and omega-6, or omega-6 fatty acids predominantly, on the weight, length and head circumference (HC) from birth to 12 months of age.¹⁴¹ The infants were also breastfed exclusively during the first three months of life, and their mothers were still taking the interventional oils. Thus, these results also apply to the question of the maternal breast milk content of omega-3 fatty acids and growth patterns.

Helland et al. included a large sample (n=590) of healthy pregnant women from Norway, yet this study only used the completers in the analysis (n=341) given the large number of dropouts.¹⁴¹ The fact that only 57% of the included women were included in the analysis, makes the results more difficult to interpret. The intake of marine omega-3 fatty acids is relatively high in Norway compared with other countries.^{347, 348} The pregnant and lactating women have high concentration of DHA in plasma phospholipids and breast milk, and a great majority of Norwegian mothers also breastfeed their infants up to at least 3 months after giving birth, thus providing their infants with preformed DHA.¹⁴¹

One good quality RCT evaluating omega-3 supplementation in Norwegian mothers,¹⁴¹ one poor quality RCT,²⁴⁸ and two observational studies were identified to answer the question related to the influence of omega-3 fatty acid content of maternal breast milk on the growth patterns in term infants.^{249, 302} No studies were identified to answer this question for the preterm population.

The two RCTs showed no apparent effects of breast milk, with maternal intake of omega-3 (DHA) or omega-6 fatty acids (AA), on the growth patterns at any time point.^{141, 248} The single prospective cohort of a small sample of Swedish mother/term infant pairs, where the infants were receiving almost exclusively breast milk for 3 months, showed a positive correlation between the maternal mother's breast milk content of AA/DHA and the infant's rate of increase of HC at 1 and 3 months of age.²⁴⁹ No associations were found between the HC and LA or ALA, or between HC and AA or DHA in breast milk.²⁴⁹

On the other hand, a cross-sectional study that included two different cohorts of term infants from Africa (two different cities with different intakes of PUFAs) was identified.³⁰² Despite the limitations of including a study with this type of research design, the differences in weight-for-age and weight-for-height z-scores and weight gain (g) were significantly lower in infants from Ouagadougou (low omega-3 fatty acids intake) compared with infants from Brazzaville (high omega-3 intake).³⁰² There are several problems with the interpretation of these results, such as the fact that the included cohorts corresponded to a completely different population (location, maternal education, home characteristics, feeding practices, maternal diet, etc.). Thus, the differences in the growth patterns could be due to all these baseline discrepancies rather than a real statistical difference. The conflicting findings across the studies demonstrate the need for further appropriate research on this association.

Twenty RCTs, with an overall mean quality score of 2.64/5 (i.e., poor quality), addressed the question of the influence of omega-3 fatty acid supplement of infant formula on the growth patterns in preterm infants.^{185, 191, 193, 198, 201, 207, 212, 218, 225, 250–259, 273} Eighteen studies failed to find an effect of the omega-3 and omega-6 fatty acids supplementation in preterm formulas on the growth parameters at several time points.^{185, 193, 198, 201, 207, 212, 218, 225, 250–259} The growth outcomes measured were the mean (SD) weight, length and head circumference, the normalized z-score of weight, length and HC and the weight, length and HC gain.

Two studies found that the omega-3 fatty acids supplemented group had a significantly lower weight at 6, 9 and 18 months of CA.^{191, 273} Both studies included healthy preterm infants and provided formulas containing DHA+EPA, as well as a control formula for comparison. The duration of the supplementation was different across the 19 trials (range from 3 weeks to 12 months CA). Interestingly enough, two studies by the same author (Fewtrell et al.) showed opposite effects in the growth pattern outcomes.^{321, 322} The results were different probably due to the different length of intervention (33 days vs. 9 months), dose of DHA and EPA (DHA 0.17 g/100 ml vs. 0.5 g/100 ml) and source of PUFAs (egg-TGL vs. fish oil).

Meta-analysis was performed for two different growth outcomes—weight and length at 4 months of CA. The results of the meta-analysis performed on the mean weight and length measured at 4 months, in the studies that compared the use of formula supplemented with DHA+AA with control formula,^{201, 207} showed that the overall effect was nonstatistically significant. No other combinations were possible, given the differences in the intervention length, measuring points and type of growth parameter (mean, z-score, mean change). Overall, there is some evidence from 20 RCTs that the omega-3 fatty acids supplementation may not have an impact on the growth parameters. This findings are consistent with the meta-analysis done by Simmer and Patole in 2003.³⁴⁹

Eighteen average good quality (Jadad: 3.2/5) RCTs addressed the question of the influence of omega-3 fatty acid supplement of infant formula on the growth patterns in term infants.^{104, 182, 203, 205, 223, 227, 260–270}

The effects across these studies on the growth outcomes, such as weight, length and head circumference, were nonstatistically different between study arms. Yet, some inconsistent differences were found across five trials at certain timepoints and subgroup of patients.^{120, 325, 328, 329, 332} The supplementation with omega-3 and/or omega-6 fatty acids has not demonstrated any benefit regarding the growth of term infants across these trials.

The studies were rather diverse in terms of intervention characteristics (type of formula, content of PUFA, duration of intervention, cointerventions), as well as the timing of the outcome measures (e.g., 2, 4, 6, 9, 12 months of age).

Meta-analysis was only possible for two studies that had the same intervention as well as the timing of the outcomes.^{104, 205} We decided to measure only two time points that corresponded to the background diet as a potential confounder. Consequently, 4 and 12 months of age were the time points selected. Four months of age is when the infants were exclusively fed with the formula, after which they began solid foods that were not controlled in any of the trials. The overall effect of formulas containing DHA+AA or DHA compared with control formula was nonstatistically significant at 4 or 12 months of age for any of the growth parameters (weight, length or HC in mean (SD)). This is consistent with the rest of the included studies and with a meta-analysis prepared by Simmer in 2003.³⁵⁰

Only four trials adjusted the results for potential confounders, such as gender, maternal education, parental socioeconomic status and center, failing to find any change in the results.^{203, 205, 263, 266}

Regarding the association between the growth patterns in preterm and term infants and the omega-3 or omega-6/omega-3 fatty acid content of maternal or fetal biomarkers, no studies were identified to answer these questions.

A total of 12 studies addressed the question of the association between growth patterns in preterm and term infants and the omega-3 or omega-6/omega-3 fatty acid content of child biomarkers. Five RCTs included a preterm population of infants,^{185, 191, 201, 207, 212} five RCTs^{143, 203, 205, 262, 263} and a prospective single cohort²⁷¹ included a term population of infants and Woltil et al., which was consciously described only in the preterm section of this question, selected a group of VLBW preterm and term infants.²²⁵

All the RCTs that included a preterm population, assessed the correlation between the infant's plasma and RBC content of AA and the growth outcomes, such as weight (mean, gain), length and HC.^{185, 191, 201, 207, 212} Carlson et al. found a significantly positive correlation between the weight and length z-scores from 2 to 12 months of CA and the plasma and RBC AA.¹⁸⁵ However, Uauy et al. observed a negative correlation between the RBC AA content and the length z-score at 57 weeks (PCA).²¹² Two studies found a positive correlation between the RBC AA and the weight and length at 1 month CA²⁰⁷ and at 2 months CA.²⁰¹ These two studies also found a significantly positive correlation between the same biomarker and weight gain.^{201, 207} Only Carlson et al. detected a positive correlation between the plasma and RBC AA and the HC at 2 and 4 months.¹⁸⁵

Carlson et al., in another study, found a negative correlation between the weight-for-length z-score and the RBC DHA at 5 months of age.¹⁹¹ Woltil et al. found a positive correlation between the weight, length and HC gains, and the plasma and RBC DHA content in preterm and term infants.²²⁵

Five RCTs measured the correlation between the plasma or RBC PUFAs and growth outcomes in term infants.^{143, 203, 205, 262, 263} Two studies did not find a significant correlation between the omega-3 fatty acids in plasma or RBC and weight.^{203, 262} However, Jensen et al. observed a significant positive correlation between the weight at 4 months and the plasma AA content at the same time point.²⁰³ Innis et al., on the contrary, did not find a significant correlation between growth patterns and the plasma and RBC AA content in term infants.²⁶³

Makrides et al. found a significantly negative correlation between plasma DHA at 16 weeks and weight at 12 and 24 months of age.²⁰⁵ Consistent with the findings in Innis et al.'s cohort of term infants, with a negative correlation of RBC and plasma DHA and infant's weight at 6 months of age, yet not at 12 months.²⁷¹ Guesnet et al. also found a negative correlation between the plasma and RBC EPA at birth and the length gain over 6 weeks.¹⁴³

It appears to be a negative correlation between weight and the plasma or RBC content of DHA, and a positive correlation between weight and the content of AA in plasma or RBC. However, not all of the studies found this association. The content of omega-6 fatty acids (AA) as a biomarker may be related to weight gain in infants. The content of DHA seems to be inversely related to weight gain, yet no significant clinical outcomes were detected.

There was one good quality RCT that addressed the question of the influence of omega-3 fatty acids intake during pregnancy and the neurological development outcomes.¹⁴¹ Helland et al. randomized a sample of pregnant women to receive either cod liver oil (DHA + EPA) or corn oil (LA +ALA) until 3 month post-delivery. This study failed to find a significant difference between groups in maturity as evaluated from the EEGs, neither at day 1 of life nor at 3 months of age.¹⁴¹

Two studies, one RCT¹³⁸ and one single prospective cohort design,²⁸⁴ addressed the question of the omega-3 fatty acid content of maternal breast milk, with or without known maternal intake of omega-3 fatty acids, influence on the neurological development in term or preterm human infants.^{138, 284} Gibson et al. randomized healthy mothers of term infants who intended to breastfed with increasing doses of DHA-rich algal oil. The infants were exclusively breastfed for 3 months. There was no difference between groups in the Bayley's Developmental Index (PDI score) at 12 and 24 months of age, however, none of the groups were acting as a control group (no omega-3 fatty acids).¹³⁸ Another issue with the interpretation of these results is that the infants were only exclusively breastfed for the first 3 months of life, which introduces potential confounding factors, such as the background diet of the infants after this age. Other potential confounders were controlled in a post-hoc analysis, which found that there were no associations with any sociodemographic variables at 1 year. The only association at 2 years of age was between PDI and the level of education of the partner.¹³⁸

Agostoni et al. evaluated the neurodevelopmental indices at 1 year of age in a single prospective cohort of term infants who were exclusively breastfed for at least 3 months in Italy.²⁸⁴ After correcting for potential confounders such us parity and mother's characteristics (i.e., age, education, smoking habits), breastfeeding for 6 months or longer was not significantly correlated to the mean PDI result compared with subjects breastfed for 3 to 6 months (n=15).²⁸⁴ There was no correlation between PDI and the milk fat content at any time point.

The results of these two different design studies showed that maternal breast milk might not have an influence on the neurological outcome, measured with the PDI scale of the Bayley's Index.

Six average good quality (Jadad: 4.2/5) RCTs were identified to assess the neurological development of preterm infants (< 37 weeks of GA) supplemented with omega-3 fatty acids in infant formula with or without breast milk intake.^{193, 207, 254, 258, 272, 273} The outcomes assessed were the PDI scale of the Bayley's Developmental Index, the Knobloch, Passamanick and Sherrards' Developmental Screening Inventory (five subscales), the neurological impairment evaluated by a pediatrician, BAEP, and NCV studies.

The results showed that, for the PDI scale, two of five studies did not observe a significant difference between the supplemented and the control formula.^{258, 273} Two studies found that the supplemented formula groups had a significantly higher score (better) than the control group.^{193, 207} However, O'Connor et al. only observed this difference in the group of infants that consumed > 80% infant formula and whose weight at birth were <1,250 g.²⁰⁷ On the other hand, van Wezel-Meijler et al. found a significantly better PDI score in the control group compared with the supplemented group at 3, 6 and 24 months, yet this difference did not reach statistical significance when adjusted for birth weight and number of SGA infants.²⁷² Only Fewtrell et al. found that there was no difference between groups in the neurological impairment assessment at 9 and 18 months CA, and in the Knobloch, Passamanick and Sherrards' Developmental Screening Inventory score.²⁵⁸

For the studies that measured the Bayley's PDI score, we could not combine them for meta-analysis given the lack of information at certain time points (i.e, 4 or 12 months of age). Two studies included patients who were also breastfed,^{207, 258} which could have introduced bias given the content of PUFAs in human milk. In some cases, the duration of supplementation was different than the time to outcome measure, or endpoint (e.g., intervention lasted 6 months and PDI was measured at 24 months). Infants that tolerated enteral feeding began their solid food at around 4 months of age. This background diet added to the formulas was not controlled in the trials, which can modify the effect of the intervention. Other factors, such as maternal diet, second hand smoking, and socioeconomic status are potential confounders, as well as parental stimulation at home.

Four studies used a non-randomized reference standard group of mothers who decided to breastfeed exclusively.^{193, 207, 254, 273}

Overall, there is not consistent evidence to suggest that the omega-3 fatty acids supplementation of infant formula, with or without breast milk, influences the neurological development in preterm infants. These findings also corresponds with the meta-analysis done by Simmer and Patole.³⁴⁹

Eight average good quality (Jadad: 4.25/5) RCTs addressed the question regarding the influence of omega-3 fatty acids supplement in infant formula, with or without human milk, on the neurological development of term infants.^{104, 176, 182, 203, 205, 227, 227, 265} The main outcome measured was the Bayley's Developmental Score system, the PDI. None of the seven studies that assessed this outcome found a statistically significant difference between diet groups at different follow-ups.^{104, 182, 203, 205, 227, 265} The endpoints were measured at 6, 12, 18 and 24 months of age.

There were other type of outcomes measured, like the Brunet-Lézine test in an Italian trial,¹⁷⁶ which showed a significantly better result in the LCPUFA supplemented group compared with the control group at 4 months of age (after exclusive formula intake). However, this result was not significant at 24 months of age, possibly due to the potential covariates and confounders after 20 months of lack of intake.¹⁷⁶

All the studies included healthy term infants, although the sources and type of omega-3 fatty acids supplementation, as well as the duration of the intervention, were different across the studies. Other potential confounders that were not assessed in the analysis were the lack of information regarding the background diet from 4 months of age until the time of assessment, and the absolute and relative amount of omega-3 and omega-6 fatty acids intake that was associated with the infant formulas. This last piece of information was not provided in any of the included trials. Jensen et al. was the only trial that compared the use of LCPUFA precursors such as LA (omega-6) and ALA (omega-3) in different ratios.²⁰³ The remaining studies used DHA and AA as type of LCPUFA, yet from completely different sources (egg lipids, vegetable oils, fish oil).

We did not include the comparisons made with the reference standard group, breastfed infants, given that those infants were not randomized and belonged to a different population. Only one study included human milk as a cointervention of the infant formulas.²²⁷ This study did not find differences between groups in any of the neurological outcomes (i.e., Bayley's PDI, and BRS at 6 and 12 months).²²⁷

Meta-analysis of the outcome measured with the Bayley's PDI was conducted in three RCTs that compared the use of formula supplemented with DHA+AA with control formula.^{104, 205, 227} The overall effect size at 12 months was nonstatistically significant between groups. No other time points could be combined. These conclusions are consistent with the meta-analysis done by Simmer in 2003.³⁵⁰

One cross-sectional study conducted in the United States assessed the association of maternal LCPUFA content (DHA) in plasma and RBC at delivery and the neurological status of their newborns.²⁷⁴ Maternal DHA was negatively associated with active sleep (AS), AS:QS (quiet sleep) and sleep-wake transition, and positively associated with wakefulness (postpartum day 2).²⁷⁴ The ratio of n-6:n-3 in maternal plasma was positively associated with AS, AS:QS and sleep-wake transition, and negatively associated with wakefulness (day 2). On day 1, the ratio of n-6:n-3 in maternal plasma was negatively associated with QS and positively associated with arousals in QS.²⁷⁴ These results mean that lower amounts of AS and the greater amounts of QS observed in the infants exposed prenatally to higher DHA concentrations suggest greater CNS maturity. Furthermore, the lower AS:QS observed in the infants in the high-DHA group shows that their sleep organization soon after birth was approaching that of normal, older infants.³³⁸

When the cohort was analyzed by maternal DHA plasma concentration, the high DHA group (>3.0% by wt of total fatty acids) did not significantly differ from the low DHA group (≤3.0% by wt of total fatty acids) in terms of maternal age, race, parity, duration of gestation, maternal education, infant birth weight and length, infant HC and Apgar score at 1 and 5 minutes.²⁷⁴ However, infants from mothers with high plasma DHA concentrations had significantly less AS and had a lower AS:QS compared with infants of mothers with low plasma DHA concentrations. Furthermore, infants in the high DHA group had significantly less sleep-wake transition and more wakefulness than did infants in the low DHA group on postpartum day 2.²⁷⁴

The difficulty with the interpretation of these results lies in the research design. Cross-sectional studies are appropriate to measure prevalence, yet not appropriate for measuring the etiological association between two variables, such as maternal biomarkers at delivery and neurological development in the infant. The outcomes assessed in this study are related to sleep patterns rather than other neurological functions such as motor, sensation and brain development, which can be associated with the CNS maturity of the infant at birth.

No studies were identified to answer the question about the association with fetal biomarkers. Four RCTs^{176, 182, 203, 205} and one observational study²⁷¹ addressed the question regarding the association of the child content of omega-3 and/or omega-6 and the neurological outcomes.

Three RCTs^{182, 203, 205} and a prospective cohort study²⁷¹ evaluated the association between the infant's plasma and RBC DHA content and the Bayley's PDI score in term infants. All these studies assessed this association in healthy term infants. Two RCTs found a significant positive correlation between the plasma DHA and the PDI score.^{203, 205} However, the timing of assessment was different for both studies. Makrides et al. measured both the blood content of biomarkers and the PDI at 12 months of age,²⁰⁵ while Jensen et al. measured the plasma and RBC content of PUFA at 120 days of age and the PDI at 12 months.²⁰³ The formula intake was also different in both trials. Two other studies (including the observational study), did not find a significant correlation between the PDI and the infant content of PUFA in plasma or RBC.^{182, 271}

Innis et al. did not find a statistically significant relation between the infant RBC DHA or AA status at 2 months of age and the Bayley's PDI score at 6 and 12 months of age.²⁷¹ But given the research design of this study, the interpretation of the results is very limited. Bias could have been introduced due to several potential effect modifiers that could underestimate results, such as maternal diet of the breastfed infants, child's background diet after 3 months of age, as well as other environmental factors that can influence the content of LCPUFAs and the neurological development in infants. The results, across the studies, are not consistent enough to draw any conclusions.

Two studies addressed the question of the influence of omega-3 fatty acids intake during pregnancy and the visual function in term infants.^{235, 275} There were no studies identified that included a preterm population.

The first study was a double-blinded RCT that assessed the retinal function of term infants of mothers that were or were not taking DHA during pregnancy.²³⁵ This trial failed to find a significant effect of DHA supplementation during pregnancy on the retinal sensitivity (ERG) measured at birth in term infants. The cross-sectional study was conducted in Cuba and measured the visual function of a cohort of term infants from mothers who had a high intake of high-fat fish during pregnancy and breastfeeding.²⁷⁵ This study failed to find a statistically significant difference in mean visual function values between the exclusively breastfed group and the infants who were also receiving formula.²⁷⁵ However, the purpose of this study was to evaluate the correlation between the visual function at 2 month of age and their blood LCPUFA biomarkers; and, no correlations were found.²⁷⁵ The interpretation of such research design on the clinical outcomes is very difficult given the lack of an appropriate comparator, randomization, blinding and other variables necessary to produce more accurate results.

These findings suggest that maternal intake of omega-3 fatty acids supplements may not effect visual function outcomes in term infants. Yet, better-conducted studies are required to support this conclusion.

Five studies addressed the question regarding the influence of human milk content of omega-3 or omega-6/omega-3 fatty acids on the visual function of term infants.^{138, 140, 248, 275, 276} Two were RCTs,^{138, 248} one was a prospective cohort study,²⁷⁶ and two were cross-sectional studies.^{140, 275} The RCTs did not detect a statistical difference in the VEP acuity among infants of mother who were or were not receiving DHA at any age (from 12 weeks to 8 months of age).^{138, 248} No studies were identified in the preterm population.

Two observational studies found a significant association between the DHA content of breast milk and visual function in term infants at 4 months of age,¹⁴⁰ and at 3.5 years old.²⁷⁶ The Cuban cross-sectional study, on the other hand, did not observe this correlation at 2 months of age.²⁷⁵

The correlation between the DHA content in breast milk and visual function was not consistent with the clinical outcomes measured in breastfed term infants of mothers who were or were not taking supplements containing high DHA.

The influence of omega-3 fatty acids supplementation of infant formula, with or without maternal breast milk, on the visual function in preterm infants was evaluated in nine RCTs with an average quality score approaching good internal validity (Jadad: 2.9/5).^{185, 191, 198, 201, 207, 212, 251, 254, 272} Five studies used the VEP as the main outcome measure,^{198, 207, 212, 254, 272} while six trials measured the visual acuity with the Teller's Acuity Card Procedure for binocular vision.^{185, 191, 201, 207, 251, 272} Only two trials measured the ERG to evaluate the retinal function of the infants, and did not detect a significant effect with LCPUFA supplementation compared with control formula.^{198, 212}

Of the five studies that measured VEP, two did not find a statistical difference between feeding groups at any time point (1, 3, 4, 12 months of CA).^{254, 272} Three studies found that compared with the unsupplemented group, infants fed with LCPUFA-supplemented formula had a better or faster maturation of visual function, in terms of significantly shorter waves in the VEP.^{198, 207, 212} O'Connor et al., however, only detected this positive effect at 6 months, but not at 4 months of CA.²⁰⁷ Uauy et al. included VLBW preterm infants (60% Black),²¹² whereas Faldella et al.¹⁹⁸ and O'Connor et al.²⁰⁷ included healthy preterm infants with an appropriate weight for GA.

Among the studies that evaluated the visual acuity using the Teller's Acuity Card test, only two studies found a significant difference between groups.^{185, 191} Carlson et al. observed a higher acuity in the LCPUFA group compared with the control group at 2 months of CA, but not at 4 and 12 months.¹⁹¹ The same significant difference favoring the supplemented group was seen in the other Carlson et al. study at 2 and 4 months of CA, but not from 6.5 to 12 months of CA.¹⁸⁵

A meta-analysis of the relevant visual outcomes was performed, comparing the studies by the type of omega-3 fatty acids used in the supplemented formula (DHA or DHA+AA) and control formula, and by the type of outcome (VEP and Teller's test of visual acuity). For the VEP visual acuity outcomes, only two studies were combined.^{207, 212} O'Connor et al. found that the use of formulas with DHA+AA resulted in a better VEP measurements compared with control formula, but only at 6 months of age. At 4 months of CA, none of the interventions showed a significant difference.^{207, 212}

Regarding the behavioral visual acuity measured with the Teller's Card test, compared with controls, there was no significant effect of DHA-supplementation at 2,4,6 or 9 months of CA,^{185, 201} or DHA+AA supplementation at 2, 3, 4 or 6 months of CA.^{191, 201, 207, 212, 272}

Only O'Connor et al. allowed their infants to receive breast milk besides the formula.²⁰⁷ The results were controlled for the amount of formula taken (>80%) in contrast with the breast milk, and the differences were still not significant for both outcomes (VEP and Teller).²⁰⁷

The differences across the trials were mostly related to the intervention characteristics (amount of formula, type of supplementation, duration of intervention) and some population characteristics, such as birth weight (VLBW, AGA), race/ethnicity distribution, and socioeconomic status, among others. These differences could explain the discrepancies in the results. These findings are consistent with the meta-analysis done by Simmer and Patole.³⁴⁹ However, the conclusions of another meta-analysis conducted by SanGiovanni et al.³⁵¹ were somewhat different. Their meta-analysis of four studies showed that at 2 and 4 months of age there was a statistically significant difference between the DHA and control groups in the visual resolution acuity (behavioral test).³⁵¹ They did not observe a significant overall effect after 4 months of age. In SanGiovanni et al., the comparisons used in the meta-analysis were taken from the same trial that included more than two dietary groups (corn oil vs. soy/marine oil, soy vs. soy/marine oil, human milk vs. corn oil and human milk vs. soy oil). We did not use this approach given that we considered more appropriate to combine the dietary groups without omega-3 FA as control group and the intervention groups discriminated by content of DHA+AA or DHA alone. Therefore, their approach to do meta-analysis is different from ours and that could be the result of the discrepancies between them.

Thirteen RCTs, of average good quality (Jadad: 3.61/5), addressed the question of the influence of the omega-3 fatty acids supplementation of infant formula, with or without breast milk intake, on the visual function outcomes in term infants.^{104, 182, 203, 205, 227, 263, 264, 266, 269, 270, 277, 352}

The outcomes assessed were the VEP in nine trials,^{104, 182, 203, 205, 264, 266, 269, 270, 352} visual acuity (binocular vision) using the Teller's Card test (behavioral visual function) in five studies,^{104, 227, 263, 277} retinal function using the ERG in one study,¹⁸² and stereoacuity using the FPL in three studies.^{182, 269, 270}

Five of nine studies did not find a significant difference between groups in the VEP at any age.^{104, 203, 205, 264, 266} Whereas, the other four trials did find a significantly better VEP in the LCPUFA-supplemented group compared with the control group at a number of time points, from 1.5 to 13 months of age.^{182, 262, 269, 270} The meta-analysis performed on this particular outcome, by LCPUFA content of DHA alone (or with the addition of AA), versus control, showed that the studies that compared DHA supplemented formula with control formula did not have an overall significant effect at any age.^{104, 182, 205} Conversely, in seven studies that compared the use of DHA+AA formula with placebo, there was no difference between groups at any age,^{104, 182, 205, 262, 264, 269, 270} with the exception of four studies that found a significant difference at 12 months of age.^{104, 182, 269, 270}

One of five studies that evaluated behavioral visual acuity with the Teller's test,²⁷⁷ found a significantly better acuity in the LCPUFA formula group compared with the control group at 2 months of age, yet not at 4, 6, 9 or 12 months. The remaining four studies did not observe a significant difference between groups in this outcome, at any time point.^{104, 227, 263} The meta-analysis performed on this outcome showed that, in studies comparing the use of DHA+AA with a control intervention, acuity was only significantly better in the DHA+AA group at 2 months of age,^{104, 182, 277} but not at 4, 6, 9 or 12 months of age.

These findings suggest that there are conflicting results across the trials regarding the efficacy of the omega-3 fatty acids supplementation of infant formula on the visual function outcomes. These conclusions are consistent with the meta-analysis done by Simmer in 2003.³⁵⁰ Another meta-analysis performed by SanGiovanni et al., also showed that there was a significantly better visual acuity (Teller's Card test) in the DHA supplemented group compared with the control group at 2 months of age, yet this effect was not seen at any other age. This result is also consistent with our findings.³⁵³

One study measured the association between the maternal content of biomarkers at 2 months postpartum and the visual acuity (Teller's Card Test) in term infants at 2 months of age. This study failed to find a significant correlation.²⁷⁵ No studies were identified to assess the association of the omega-3 fatty acids content in fetal biomarkers and the visual function outcomes. However, 21 studies assessed the question of the association between child's omega-3 or omega6/omega-3 fatty acids biomarkers and the visual function outcomes. Five studies included a preterm population,^{185, 198, 212, 278, 279} while 16 included term infants. Of the five studies in the preterm group, three were RCTs,^{185, 198, 212} and two were cross-sectional studies.^{278, 279} Of the 16 term infant studies, nine were RCTs,^{138, 182, 203, 248, 262–264, 269, 270} and seven were observational studies.^{140, 271, 275, 278, 280–282}

In all the preterm RCTs, the results were conflicting. In the study by Birch et al, the LCPUFA content of RBC DHA/DPA ratio correlated with both FPL and VEP at 57 weeks PCA.²¹² Based on ANOVA, there was a statistically significant correlation between RBC DHA at 2 months and visual acuity at 2 and 4 months, in the Carlson et al. study.¹⁸⁵ Faldella et al. found a negative correlation between the RBC DHA and the N4 and P4 wave latency of the VEP at 52 weeks PCA.¹⁹⁸

In two preterm cross-sectional studies, the results also were divergent.^{278, 279} Birch et al. found that the LogMAR (VEP) acuity was significantly associated with the end-product ratio [DHA n-3/DPA n-6] in total RBC lipids. For FPL acuity, the results were the same for both the breastfed and formula-fed groups.²⁷⁸ Whereas, Leaf et al. observed a positive correlation between scotopic b wave (ERG) implicit time and percentage composition of DHA in both plasma and RBC PL. A similar relationship was seen with total omega-3 LCPUFA in both plasma and RBC PL. There was a positive correlation between both RBC AA and total omega-6 LCPUFA and scotopic a-b amplitude. No significant relationships were seen between photopic ERGs and either plasma or RBC LCPUFAs.²⁷⁹

Given the different designs and interventions (human milk or formula), it is very challenging to draw a conclusion in the preterm population.

In the term population, of the seven RCTs that had an infant intake, four^{182, 264, 269, 270} reported associations between milk or blood biomarkers (plasma/RBC DHA and/or AA content) and the sweep VEP acuity measures. Of these trials, three^{182, 269, 270} found statistically significant negative linear regression coefficients indicating that higher RBC DHA content was associated with a better sweep VEP acuity in infants at different age time points. The remaining study²⁶⁴ suggested that the RBC DHA content was not associated with the measured sweep VEP acuity at 4 months of age. The results of both trials^{182, 264} that looked at the RBC EPA and AA content in relation to the measure of sweep VEP acuity, indicated that neither RBC AA nor EPA content was associated with the sweep VEP acuity during the first year of the infants' life. One study,²⁶⁹ that investigated the relationship between infant's plasma DHA and AA content, found that higher plasma contents of both DHA and AA were associated with better sweep VEP acuity at 4 and 13 months of age.

The relationship between the infants' blood biomarkers and the measures of infant amplitude of VEP acuity were reported in two trials.^{203, 262} Both trials suggested that RBC DHA correlated negatively with the amplitude of VEP acuity (in log MAR), measured at 4^{203, 262} and 7.5²⁶² months of age (i.e., infants at 4 and 7.5 months of age who on average had a higher RBC DHA content, tended to have a lower log MAR or better VEP acuity). The former trial²⁰³ also showed that there was no correlation between either plasma or RBC DHA content at 4 months of age, and the latency measure of VEP acuity obtained at either 4 or 8 months of age. The same trial,²⁰³ however, found a statistically significant negative correlation between plasma-DHA content and the amplitude of VEP acuity both measured at 4 months of age.

One study reported the association(s) of the plasma DHA or RBC DHA content in relation to the measure of Teller's visual acuity.²⁶³ The plasma or RBC DHA content did not correlate with the Teller's acuity, measured at 3 months of age.

Only two trials reported the associations between the infants' RBC DHA content and their stereoacuity (in log seconds) measured at 4²⁶⁹ and 12²⁷⁰ months of age. Both trials found that there was no association between the two factors.

The correlation between plasma- and RBC DHA and ERG parameters in infants was reported in one trial.¹⁸² None of the Naka-Rushton parameters except for log k (in scotopic troland seconds) was significantly correlated with plasma or RBC DHA content at either 1.5 or 4 months of age. There was a statistically significant negative correlation between the RBC DHA content and log k in the infants at 1.5 months of age.

None of the RCTs that measured the association of the infant's biomarkers after exclusive breast milk intake and the visual acuity outcomes found any significant correlation.^{138, 248}

The seven observational studies were very heterogeneous in term of exposure characteristics and population, as well as outcomes. Most of them used breast milk as the main exposure, as well as formula. However, the overall association was that in four cross-sectional studies there was a nonsignificant correlation between infant's biomarkers and the visual acuity at any age.^{140, 275, 281, 282} In three studies, there was significant correlation between the biomarkers and the visual acuity.^{271, 278, 280} Yet, the biomarkers and the outcomes were different in each. Birch et al. found that there was a positive correlation between the infant's RBC DHA/DPA ratio and the stereoacuity,²⁷⁸ whereas, Makrides et al. observed a positive correlation between the RBC DHA and LA and the VEP (logMAR).²⁸⁰ Finally, Innis et al. also detected a positive association between the RBC DHA at 2 months and the visual acuity (Teller's test) at 2 and 12 months of age, but not at 4 and 6 months of age.²⁷¹

Overall, there was a lack of pattern of correlation between the infant's biomarkers in blood and the visual function outcomes across 21 studies that addressed this issue.

One RCT addressed the question regarding the influence of maternal intake of omega-3 fatty acids during pregnancy on the cognitive development in infants.¹⁴¹ This study measured the cognitive development using the Fagan Test of Infant Intelligence at 6 and 9 months of age in the infants of mother who had taken either cod liver oil (DHA+EPA) or corn oil (LA+ALA) during pregnancy and lactation. There was no differences between groups in the novelty preference at both time points.¹⁴¹ There was a follow-up study at 4 years of age that measured the Kaufman Assessment Battery for Children (K-ABC), which is a measurement of intelligence and achievement designed for children between 2.5 years and 12.5 years old.³⁵⁴ The supplemented group (DHA+EPA) had significantly higher scores than children in the corn oil group (mothers) on the Mental Processing Composite of the K-ABC at 4 years old. However, the scores in the Sequential Processing Scale, the Simultaneous Processing Scale and the Nonverbal Scale among children who were born to mothers who were given cod liver oil were non statistically different from the control group.¹⁴¹

The latter relationship may be relevant, although the clinical importance of this result has yet to be determined. The potential confounders such as infant's diet after the exclusive breastfeeding, medications, supplements and other variables that could affect the results, were not measured at the time of the outcome (4 years of age).

Three studies were identified to respond to the question of the influence of maternal content of omega-3 fatty acids in breast milk influences the cognitive development in infants.^{138, 141, 284} Two were RCTs^{138, 141} and one was a prospective cohort.²⁸⁴ The study by Helland et al. was an RCT described above,¹⁴¹ and the study by Gibson et al. was a double-blind RCT that included mother of term infants who intended to breastfeed.¹³⁸ They were randomized to receive five increasing doses of DHA (algal oil) during the first 3 months postpartum. The mean Bayley's MDI score did not differ between groups at 1 or 2 years of age.¹³⁸ The environmental factors that were associated with the Bayley's MDI at 1 year of age were the home stimulation test, partner smoking status, length of breastfeeding and the 3-month DHA status of breast milk and infant blood. The only one that was still correlated to the Bayley's MDI at 2 years was the home stimulation test.¹³⁸

This study was underpowered to detect a significant difference between groups in the MDI scores, which makes it very difficult to draw a conclusion. There was no comparator without omega-3 fatty acids. The infants were fed solid foods before the measurements (Bayley's score), which can be a potential effect modifier.

Six average good quality (Jadad: 4.4/5) RCTs addressed the question of the influence of formula intake, with or without breast milk, on the cognitive development of preterm infants.^{185, 193, 207, 258, 272, 273} The main outcome measured was the Bayley's MDI score, at different time points in the five RCTs.^{193, 207, 258, 273, 355} Overall, four of the five trials did not find that the supplementation of infant formula with omega-3 fatty acids had an effect on this particular outcome at 3, 6, 12, 18 and 24 months of age. This remained true even after controlling for potential effect modifiers such as site, gender, birth weight, maternal education, gestational age, and human milk intake, among others.²⁰⁷ Except for one trial which found that sex was an important covariate, males in the supplemented formula group had a significantly higher score than those in the control group at 18 months.²⁵⁸ Only one study, which included preterm and term infants, found that the supplemented groups had greater scores than the control group at 118 weeks PMA, and the term infants had higher scores than the preterm infants.¹⁹³

Regarding the Fagan test of Infant Intelligence outcome, two studies found a significant difference between the omega-3 fatty acids group and the control group.^{185, 207} Carlson et al. observed that the DHA group had significantly more discrete looks in the novelty test,¹⁸⁵ however, at 12 months the DHA-supplemented group had a significantly lower novelty preference compared with the control group. Whereas, O'Connor found that the DHA+AA (egg-TGL/fish) group had a significantly greater mean novelty preference look compared with the DHA+AA (fish/fungal) formula and the control group at 6 months.²⁰⁷

O'Connor et al. also found that there was no significant differences between groups in the Infant version of the MacArthur Communicative Development Inventories (a standardized parent-report instrument) at 9 months CA and 14 months CA.²⁰⁷

Meta-analysis was not possible given the heterogeneity across the studies for each of the different outcomes. This heterogeneity was observed in the intervention characteristics (meaning dose, source of omega-3 fatty acids, duration of intervention), cointerventions, and timing of the outcomes measures. Other potential confounding factors can be associated with the discrepancies in the study results such as background diet, breast milk intake, and environmental factors (parental education, stimulus at home, smoking status at home, etc), as well as the use of different assessment tools. It is thought that global measures of cognitive development (Griffith, Bayley, Brunet-Lezine scales) may not be sensitive enough to detect differences in normal infants supplemented with or without DHA. It is likely that specific functional tests (Fagan's test, Means-end problem solving test) would be more sensitive and specific in detecting these differences in assessing the adequacy of DHA intake on optimizing neurocognitive development. The more specific tests used during infancy have been shown to have a better correlation with testing later in childhood than the global infant tests.^{356, 357}

Overall, most of the studies did not find a significant effect of the omega-3 fatty acids supplementation in preterm infants on the cognitive developmental outcomes using the Bayley's MDI scale. Nonetheless, a question remains as to which would be the best instrument to measure this particular outcome. These conclusions are consistent with the meta-analysis done by Simmer and Patole.³⁴⁹

Eight good quality RCTs were identified to address the question of the influence of omega-3 fatty acids supplementation of infant formula, with or without breast milk intake, on the cognitive development in term infants.^{104, 182, 203, 205, 223, 227, 265} The mean outcome that was measured across seven RCTs was the Bayley's MDI score at different time points.^{104, 182, 203, 205, 227, 265} All but one of the studies did not find a significant difference between groups (supplemented vs. control) in this outcome at 6, 12 and 18 months of age. Only Birch et al. observed that the DHA+AA group had a significantly higher score compared with the control group at 18 months of age.¹⁸²

There were five other different cognitive outcomes measured across the trials. The Knobloch, Passamanik, and Sherrards Development Screening Inventory test, performed at 9 months of age in the study by Lucas et al., and the Fagan Test of Infant Intelligence, performed at 6 and 9 months of age in two other trials by Auestad et al., did not reveal an effect with omega-3 fatty acids supplementation.^{227, 265} The IQ (Stanford-Binet), Receptive Vocabulary (PPVT-R), Expressive Vocabulary, and Visual-Motor Index scores, as well as the Problem-Solving scores, did not differ between groups in two studies.^{104, 223}

Regarding the Infant version of the MacArthur Communicative Development Inventories, Auestad et al. found that the DHA group had a significantly lower vocabulary production score compared with the control group at 14 months of age.¹⁰⁴ Yet, the other Auestad et al. study found that at 14 months, the DHA+AA (fish/fungal) group had a significantly higher vocabulary expression score than those fed with DHA+AA (egg-TG) supplemented formula.²²⁷ Both Auestad et al. studies did not reveal a between-group significant difference at 9 months.²²⁷

A meta-analysis of the main outcome used across the trials, the Bayley's MDI score at 12 months of age, was performed. Three studies were identified to be appropriately comparable in terms of type of supplementation (DHA+AA) and population characteristics (healthy term infants).^{104, 205, 227} The overall size of the effect was nonstatistically different between study groups.

Overall, it appears that the supplementation with omega-3 fatty acids does not have an effect on the cognitive development outcomes. These conclusions are consistent with the meta-analysis done by Simmer in 2004.³⁵⁰ Although the design of the studies is very appropriate, they have some limitations. The studies did not measure the total dose of omega-3 or omega-6 fatty acids contained in the formulas, since they failed to account for the total amount of formula intake per day. They also were unsuccessful in controlling for background diet, in the infants (from 4 months of age) and the mothers (breastfed infants). There were also discrepancies in the intervention length and the outcome measures (e.g., formula given until 4 months of age and Bayley's MDI measured at 12 months of age) within each trial and across all the included studies.

An attempt to control for potential confounders was appropriately done in almost all the studies. However, none of them use the omega-3 fatty acids dose as a covariate. Instead, they used the plasma or RBC DHA content, or the type of diet.

Only one study allowed the infants to be breastfed as a cointervention.²²⁷ Nevertheless, the use of both supplemented formula and breast milk, did not show an effect on the cognitive development when compared with breast milk alone (control formula).

No studies were identified to answer the questions of the association of omega-3 or omega-6/omega-3 fatty acids content of maternal or fetal biomarkers and the cognitive development in term or preterm infants.

Six studies addressed the question of the association of omega-3 or omega-6/omega-3 fatty acids content of child biomarkers and the cognitive development in term infants.^{138, 182, 203, 205, 271, 285} Four of them were good quality RCTs,^{138, 182, 203, 205} and two were single prospective cohort studies.^{271, 285} There were no studies identified to address the same question in preterm children.

Gibson et al found that the infants were exclusively breastfed for 3 months. There was a significant correlation between the Bayley's MDI score at 1 year old and DHA indices in plasma and RBC at 12 weeks of age, yet this correlation was not seen at 2 years of age.¹³⁸

Birch et al. found that the MDI score at 18 months was positively correlated with plasma and RBC DHA at 4 months of age. None of the other plasma biomarkers (LA, AA, ALA, EPA) were correlated with the MDI at 18 months, however the RBC-LA and RBC ALA were negatively correlated with the MDI at 18 months of age.¹⁸² None of the biomarkers measured at 12 months of age were correlated with the MDI at 18 months of age.¹⁸²

Jensen et al. and Makrides et al. did not observe a significant correlation between the PUFA content in infant's plasma and RBC, and the Bayley's MDI at 1 and 2 years of age.^{203, 205} However, these studies used a different type of intervention—Jensen et al. used increasing ratios of LA/ALA in four groups,²⁰³ whereas, Makrides et al. used three formulas with LCPUFAs (DHA+AA vs. DHA alone vs. control).²⁰⁵

Finally, both observational studies failed to find a significant correlation between the biomarkers and the cognitive outcomes.^{271, 285} Innis et al. did not find a statistically significant relation between the infant RBC DHA or AA status at 2 months of age and the Bayley's MDI score at 6 and 12 months of age, as well as the Novelty Preference at 6 and 9 months.²⁷¹ Ghys et al. did not observe a correlation between the DHA and AA concentration in infant's plasma or RBC and the cognitive development at 4 years of age. Small but significant associations occurred with maternal IQ, birth weight, duration of breast-feeding, maternal smoking during pregnancy, and paternal educational attainment.

Meta-analysis of these associations was not possible given the differences in the intervention characteristics, as well as in the timing of the blood samples and the cognitive outcomes measures. In general, there are discrepancies in the results related to the association between the child's biomarkers and the cognitive developmental outcomes.

Clinical Implications

The intake of omega-3 fatty acids in the present review's collection of interventional studies by maternal and child populations did not appear to be associated with moderate or severe adverse events. Supplementation studies enrolling pregnant women typically utilized controlled, capsule delivery of relatively simple interventions (e.g., fish oil, containing DHA); and, supplementation appeared to be well tolerated, with some mild, mostly gastrointestinal events occurring occasionally. A similar pattern was observed in supplementation studies with child populations. However, a few factors make it very difficult to identify the specific or collective safety profiles of the individual omega-3 fatty acids in studies investigating their influence on child outcomes.

First, there was a wide variety of types of omega-3 fatty acid employed in these studies. Second, more than just a single omega-3 fatty acid was typically employed in these pediatric trials. The latter observation likely has strong implications for what can be understood as the meaningfulness of possible differences or similarities in the adverse event profiles associated with the respective study groups (i.e., “intervention,” “control”), even in RCTs considered well-controlled in other ways (e.g., allocation concealment; blinding).

In a study comparing the effects of DHA and an olive oil placebo (i.e., “no-DHA”), for example, typically added to the active and placebo formulations are the exact same constituents (e.g., other omega-3 fatty acids; omega-6 fatty acids; iron; anti-oxidants). However, the possibility that individually or collectively these cointerventional or background elements could “interact”—metabolically speaking—differently with DHA and olive oil to potentially produce different “synergistic” influences on clinical outcomes suggests that, in these studies: a) what is meant by the “intervention” and “control” is more complicated than a simple distinction between “DHA present” and “DHA absent;” and b) the exact absolute and relative influences of DHA on clinical outcomes in this example cannot be readily isolated. Especially problematic for interpretation are those interventional studies whose specific cointerventional or background constituents included various other omega-3, omega-6 or omega-9 fatty acids, which constitute various metabolites along the metabolic pathway (from the parent EFAs, LA and ALA). In short, the dynamic interplay among these fatty acid contents (e.g., competition for enzymes), and how this interplay may influence outcomes, may differ in important ways depending on whether DHA or olive oil is added to this combination of cointerventional or background constituents.

Thus, the ability to reliably associate the presence, or absence, of specific adverse effects with specific omega-3 fatty acids may be impeded by the inclusion of background constituents within studies of formula supplementation. At best, inferences may be drawn with respect to often very complex combinations of constituents. This research strategy adds considerable “noise” to studies, which precludes the identification of clear “signals” regarding the adverse effects associated with specific omega-3 fatty acids. Moreover, definitions of interventions in the different studies were often diverged, even though they appeared to share the same key active ingredient, such as “DHA.” This clinical heterogeneity complicated attempts to compare studies.

The evidence pertaining to the possible impact of supplementation with omega-3 fatty acids on predefined pregnancy outcomes showed either evidence of no effect, or the results were inconclusive. Results suggested the absence of effects with respect to the impact of supplementation on the incidence of GHT, preeclampsia or eclampsia, as well as on infants being born SGA (measured via birth weight and incidence of IUGR). However, regarding evaluations of the duration of gestation, some discrepancies were observed, although most of the studies failed to detect a statistically significant effect.

Regarding the questions of the biomarker content during pregnancy, and its possible association with pregnancy outcomes, nothing conclusive can be asserted. There was considerable heterogeneity in the research designs (i.e., experimental versus observational), the types of biomarker that were evaluated, the timing of these measurements, and the types of intervention given to study participants (i.e., source of omega-3 fatty acids; omega-6 fatty acids; omega-6/omega-3 ratio intake).

Overall, results concerning the impact of the intake of omega-3 fatty acids on the development of infants are primarily, although not uniformly, inconclusive. The inconsistencies in study results may be due to differences in the: a) definitions of the type and source of omega-3 fatty acids; b) omega-6/omega-3 fatty acid intake ratio in the intervention, the background diet, or both; c) absolute and/or daily amounts of formula supplementation received by the children; or, d) duration of the intervention. Most of the studies did not control for the absolute or daily amounts of formula ingested by the child populations, which lessens our ability to draw unequivocal inferences about the value of this supplementation. Moreover, making clear sense of the absolute or relative effects of individual omega-3 fatty acids, or even omega-3 fatty acid combinations, on child outcomes is complicated by the same problem of “noise” described with respect to the safety evidence in child supplementation studies. It is very difficult to reliably ascribe definite benefits, or the absence thereof, to specific omega-3 fatty acids.

Looking at specific categories of child outcome, growth patterns were not affected by the intake of omega-3 fatty acids via human milk or formula supplementation in either term or preterm infants. With biomarker data obtained exclusively from infant population sources, results across the different studies concerning the association between child biomarker content and growth outcomes were inconsistent, and thus inconclusive.

The neurological development outcomes were influenced somewhat by the omega-3 fatty acids supplementation of infant formula in preterm infants,^{193, 207} although not all of the studies found evidence for a benefit. Overall, however, the results must be considered inconclusive for preterm offspring. On the other hand, term infants did not receive any benefit from the intervention in the short- or longterm. A reliable association between infant biomarker content and neurological outcomes for both term and preterm infants was not supported, because of the lack of consistency in the results across the studies.

Visual function outcomes provided the most inconsistent data in both the preterm and term infant populations. This suggests an inconclusive response to the question of the value of omega-3 fatty acid intake for visual development. This same observation characterizes the results concerning the association between biomarker content and visual outcomes.

In the preterm population, the only type of clinical outcome that showed a significant favorable effect related to the intake of omega-3 fatty acids was the Fagan test of Infant Intelligence (i.e., “novelty preference looks”) at 6 and 12 months of age.^{185, 207} It assesses cognitive function. However, the scores on the Bayley's Developmental Index (MDI) were not influenced by infant supplementation at any age.^{185, 207, 258, 273} In most of these studies, the intervention was stopped months before the final cognitive assessment was performed (i.e., 12 or 18 months). This observation suggests a likely problem in interpreting the results. Between the end of the intervention period and the final cognitive evaluation, dietary intake was not measured and controlled for analytically. This factor may have contributed to what was observed at the final outcome evaluation. Other factors that could have influenced the outcomes included child illnesses, perceptual-cognitive stimulation, smoking, and parental education.

In the term population, while there was some disagreement in results across the trials, most of them reported a lack of effect using the Bayley's MDI. The association between biomarker content and cognitive outcomes has yet to be determined.

In summary, definitions of the maternal population in studies of pregnancy outcomes varied considerably, yet no conclusive evidence for benefit was identified. Results based on both term and preterm study populations were also inconclusive, although these studies typically entailed interventions of the complex nature discussed earlier. Thus, when it came to the set of child developmental/health questions investigated in our review, it must be asked whether or not the included studies could have been expected to provide unequivocal evidence regarding the value of all, or individual, omega-3 fatty acids in influencing child health? Could these studies have been expected to permit the isolation of the impact of the omega-3 fatty acids in these populations? That said, had the results been conclusive one way or the other, much of the included research studies lacked strong applicability to the North American population.

What, then, are the research implications?

Research Implications and Directions

Questions for which no evidence was identified clearly require empirical studies. The studies enrolling child populations typically exhibited sound quality, defined in terms of Jadad total scores. However, these investigators typically failed to design studies where the specific effects of omega-3 fatty acids could be isolated. While this outcome may have been necessary, given the expectation that all of the constituents were likely important for child health, the results were difficult to interpret. Biomarkers measure the content of specific fatty acids of different lipid fractions in plasma (individual fatty acids or content in triglycerides, cholesterol esters or phospholipids), cell membranes (red blood cells, platelets) or tissues (such as adipose, umbilical cord). These biomarkers are used to reflect dietary intake or as a surrogate measurement of the fatty acid content of various tisses that are not readily available for measurement. The essential n-3 and n-6 fatty acids content of these biomarkers reflect the exogenous intake of these fatty acids within hours to years. The inherent difficulty with using membrane and accessible tissue biomarkers as surrogate measurements of the fatty acid content of for example, the brain or retina, is the difference in preferential deposition of these fatty acids in different membranes and tissues and the rate of turnover. For example, DHA is preferentially accumulated in the brain and retina but not in the red cell membrane. As well, once DHA is deposited in the brain and retina, the amount is relatively resistant to turnover even with subsequent dietary n-3 fatty acid or DHA deficiency, whereas RBC membrane levels would decrease.

During different stages in life there are changes fatty acid metabolism, storage and turnover that affect the fatty acid profile of the various biomarkers. The choice of biomarker is dependent on the intervention and outcome of interest. For example, during pregnancy there are significant changes in lipid metabolism with increased fat storage in the early stages and mobilization in the later stages. If the outcome of interest is the effect of maternal intake of n-3 fatty acids on pregnancy outcomes, then markers that reflect shorter term dietary intake should be used (plasma lipid fractions, RBC membranes). During periods of growth and development in infancy, there is rapid accumulation of n-3 fatty acids that are preferentially deposited in neurologic tissues, which may not be reflected in the available biomarkers. Again, it is likely that RBC membrane fatty acid content more closely reflects the content in neurologic tissue than from plasma or adipose tissue. It is clear that further research is required to establish the predictive value of available biomarkers or the development of new biomarkers of n-3 fatty acid status on clinical outcomes.

One key implication is that the most likely question that the included child outcome studies might have been able to address is whether formula supplementation “cocktails,” which included at least one type of omega-3 fatty acid content, could provide a benefit to child health. The overarching question concerning the role of omega-3 fatty acids in child health that we aimed to address with this review might have been too narrow especially in light of: a) expectations that the omega-6 fatty acids alone (e.g., AA), or possibly in combination with the omega-3 fatty acids, might substantially influence child health; and b) knowledge that the available, relevant studies invariably employed interventions including elements other than the omega-3 fatty acids. Thus, one key contribution of our review may be that we have now raised an additional question: can questions concerning the possible impact of any of the EFAs on child health be conceived without concurrently considering the (e.g., interactive) roles of both the omega-3 and omega-6 fatty acids?

That said, one possible strategy for research entails defining interventions according to specific omega-6/omega-3 fatty acid intake ratios, which would be achieved via the co-modification of the intake of omega-3 and omega-6 fatty acids. While the ideal design with which to test questions of efficacy is the RCT, pilot work using less complex designs would need to be done first. These would help establish intake ratios with some potential to benefit child outcomes. It might then be observed that different intake ratios positively influence different developmental outcomes, or yield different safety profiles.

Decisions as to the “appropriate” or “reasonable” intake ratios for use as interventions in RCTs could then be made based on what is considered an acceptable benefit/safety profile and/or what are the most important outcomes—and the timing of their assessment—requiring modification. It may turn out that in a preliminary cohort study, exposure to EFAs is most beneficial for early neurodevelopment.

Evidence concerning the metabolic interplay of the fatty acid contents in biomarkers might also help shape the “appropriate” or “reasonable” intake ratio. This preliminary work could demonstrate that certain combinations of fatty acids actually produce antagonistic, rather than synergistic, effects, metabolically speaking. In this way the optimal combinations of EFA (e.g., DHA+AA), and sources thereof (e.g., marine, plant) could be identified, including circumstances where it is an antagonistic metabolic dynamic that is desired, since it appears to produce important clinical effects. Work with biomarker data could thus be helpful in designing studies and not just as a means to predict clinical outcomes, or to make sense of relationships between patterns of EFA intake and clinical outcomes. Nevertheless, to produce readily interpretable results, at least two additional strategies would be helpful.

First, the nutrients obtained via the background diet would also need to be factored into the definition of the intake ratios. Second, to control for the possibility that it is the volume of intake of supplementation that positively influences child outcomes, daily or weekly amounts of intake should be measured, and the corresponding data are entered into covariate analysis. For ethical reasons, this approach would likely be preferable to one whereby a minimum or maximum volume of intake is established.

These strategies would complement the other, typically necessary research-design elements, and maximize the meaningful interpretability of even RCT results (e.g., control for caloric/energy intake across study groups). Data regarding the maternal preconceptional and perinatal diets should be retrieved before a study begins.³⁵³ Data concerning the maternal diet during pregnancy or breastfeeding may help explain (the lack of) beneficial effects with respect to child outcomes. Likewise, data regarding the dietary intake of children following the termination of the intervention period (e.g., at 4 months), yet preceding a longer term followup (e.g., at 12 months), need to be collected to help explain (the lack of) beneficial effects on child outcomes.

Many of these variables were not assessed in the studies focusing on child outcomes in our review. Failure to control for these or other variables, either experimentally or analytically, complicate or preclude the meaningful interpretation of results. Also very important is the need to take into account the possible influences of key confounders, such as mother's smoking or alcohol consumption. If it is assumed that EFA content in mother's biomarkers may be associated with child outcomes, then these and other factors with the ability to negatively influence the fatty acid content of biomarkers need to be evaluated. These factors are likewise important when trying to make sense of maternal outcome data.

Future child outcome trials will always be faced with the problem of selection bias inherent in appropriately giving women the choice of whether or not to breastfeed, and then excluding those who decide to breastfeed from being randomized to study groups varying in terms of the constituents defining formula supplementation. As we did in this review, data from children of mothers who breastfed can be used as a reference point from which to understand results produced by supplementation. That said, it must also be appreciated that the choice not to breastfeed could also influence child outcomes in ways that are as yet unclear.

The relevance of the instruments chosen by the investigators to measure the neurological development, cognitive development and visual function are perhaps open to debate. Future research might benefit from the work of a panel to establish the most important outcome constructs as well as the most reliable and valid instruments. Candidate outcomes and instruments should include, yet without being restricted to, those instruments utilized in studies included in our review (e.g., Bayley's Developmental Index, Fagan test of Infant Intelligence, EEG).

Regarding pregnancy outcomes, the issue of the length of the omega-3 fatty acid intervention may be an important one. Most of the studies initiated the intervention during the second or third trimester of pregnancy. Almost none provided it before, or at the beginning of, the pregnancy. One empirical question is whether or not ingesting omega-3 fatty acids for a longer period of time might increase their contents in maternal stores, which in turn could have a beneficial impact on maternal or child outcomes.

Most of the interventions given to maternal populations identified in our review were relatively simple, in that they did not contain the myriad constituents such as those received in formula supplementation studies. However, while the problem of “noise” discussed above with respect to child outcome studies typically did not characterize the maternal outcome investigations, studies relating to pregnancy outcomes might consider concurrently modifying the intake of both omega-3 and omega-6 fatty acid contents for the purposes of evaluating the possible beneficial impact of specific omega-6/omega-3 fatty acid intake ratios.

In preparing such intake ratio interventions, the exact source, type and doses of fatty acids will require definition in pilot work. As with interventions given to child populations, the efficacy and safety of those provided to maternal populations needs to be balanced. Moreover, the possible interactions of fatty acid contents and other types of supplementation routinely taken during pregnancy (e.g., vitamins, iron) should likely be fully understood to assure that positive clinical outcomes are afforded.

In the studies of pregnancy outcomes per se, a number of factors need to be controlled either experimentally (e.g., stratification) or analytically, which will permit meaningful inferences to be drawn from results. These variables include the maternal background diet, smoking, alcohol consumption, obstetric history, other supplementation, medication, and socioeconomic status. Most of the included studies did not control for maternal background diet, for example.

Future studies hoping to investigate the possible role played by biomarker data—obtained from the mother, fetus or child—in understanding the relationship between the intake of specific nutrients and clinical-developmental outcomes should likely be undertaken as an integral part of RCTs evaluating this relationship. Observational studies lack the types of controls required to best minimize bias from known and unknown confounders. The timing of the measurement of biomarker data is also very important. If an argument can be made to conduct followup assessments of clinical-developmental outcomes at specific time points, or according to specific milestones, then it might be reasonable to evaluate the fatty acid content of biomarkers at these same times. If there is no concurrence in the measurement of these two classes of outcomes, then it may be difficult to detect the most meaningful parallels in the respective patterns of results.

Finally, in order to maximize the applicability of the evidence to the reference standard established in our review—the North American population—it would be helpful to conduct more research in North America. Furthermore, evidence concerning otherwise healthy populations should likely be obtained, before attempts are made to understand the interrelationships among intake, biomarkers, and clinical-developmental outcomes in populations with specific disorders or problems (e.g., celiac disease; malnutrition).

Limitations of the Review

One of the main limitations was that we did not investigate studies assessing the possible impact of the intake of omega-3 fatty acids on the fatty acid content of biomarkers. While it might be assumed that omega-3 fatty acids, when ingested, eventually find their way into pertinent biomarkers, it may be the case that it is actually the failure to become incorporated in pertinent biomarkers that prevents (some or all of) the fatty acid contents from positively influencing clinical-developmental outcomes. Thus, problems complicating or preventing their accretion should likely be understood before interpretations can be accepted that omega-3 fatty acids have no effect on clinical-developmental outcomes in various populations.

Another limitation is the difficulties that we were faced to identify studies that addressed some of the questions, specially the association between fetal biomarkers and clinical outcomes and the influence of other sources of omega-3 fatty acids on the child's clinical outcomes.

Safety data obtained from RCTs are typically under-reported. Thus, the exclusive focus on RCT evidence for certain questions in our review may have allowed us to miss key adverse effects data contained in reports of studies employing less inherently rigorous types of study design.

The quality assessment of observational studies was conducted using items we modified from existing instruments. A design-specific, total quality score was then generated for each study, from which a single summary value was derived. (i.e., A, B, C). This simplification permitted the entry of these values into summary matrices. However, the design-specific cutpoints used to assign these values were established without any validational basis, and so their value is likely extremely limited. The modified instruments themselves were also never subjected to a validational exercise. The applicability indices, while continuing the work we did when we systematically reviewed the evidence for the health effects of omega-3 fatty acids on asthma,¹⁶³ likewise did not receive validational support.

We recognize that the issue of investigating the possible impact of the background diet's omega-6/omega-3 fatty acid intake ratio within studies evaluating the health effects of omega-3 fatty acids is a very complex one. There are many ways to produce the same ratio, for example. Ratios of 30:2 and 15:1 are equivalent, yet the absolute amounts may also need to be taken into consideration when appreciating the possible benefits of omega-3 fatty acid supplementation. Moreover, there are multiple definitions of each of these classes of fatty acid (i.e., omega-3 vs omega-6 fatty acids), and the types of dynamic metabolic interaction between fatty acids appears to depend greatly on which fatty acids are involved. One likely needs to distinguish the absolute and relative amounts of the short- versus long-chain fatty acids, for example. EPA and DHA (i.e., long-chain omega-3 fatty acids) have markedly different metabolic properties than ALA (i.e., short-chain omega-3 fatty acid). The same may be said about LA (i.e., short-chain omega-6 fatty acid) when compared with AA (i.e., long-chain omega-6 fatty acid). The interaction of EPA and AA is different from the interaction of DHA and AA. Moreover, AA and DHA do not compete for positions in cell membrane phospholipids: AA may be found in PI, while DHA is contained in PS and PE. That said, future research might end up concluding that especially in the North American diet—where much more omega-6 fatty acid content is consumed when compared with omega-3 fatty acid content—the best way to alter the omega-6/omega-3 fatty acid intake ratio is to focus exclusively on increasing the intake of (especially long-chain) omega-3 fatty acids.

Finally, time constraints made it impossible to perform additional meta-analysis of other time points relating to the neurological and cognitive outcomes.

Conclusion

Studies investigating the influence of omega-3 fatty acids on child and maternal health revealed the absence of a notable safety profile (i.e., moderate-to-severe AEs). Pregnancy outcomes were either unaffected by omega-3 fatty acid supplementation, or the results were inconclusive. Results suggested the absence of effects with respect to the impact of supplementation on the incidence of GHT, preeclampsia or eclampsia, as well as on infants being born small for gestational age. However, regarding evaluations of the duration of gestation, some discrepancies were observed, although most of the studies failed to detect a statistically significant effect. Biomarker data failed to clarify patterns in pregnancy outcome data.

Results concerning the impact of the intake of omega-3 fatty acids on the development of infants are primarily, although not uniformly, inconclusive. The inconsistencies in study results may be attributable to numerous factors.

In addition, making clear sense of the absolute or relative effects of individual omega-3 fatty acids, or even omega-3 fatty acid combinations, on child outcomes is complicated or precluded by the following problem. Studies typically employed interventions that involved various cointerventional or background constituents (e.g., omega-6 fatty acids), yet whose metabolic interactions with the omega-3 fatty acid(s) were not taken into account in interpreting the results. The dynamic interplay among these fatty acid contents (e.g., competition for enzymes), and how this interplay may influence outcomes, may differ in important ways depending on whether DHA or olive oil is added to this combination of cointerventional or background constituents, particularly in the maternal population. This strategy prevented the isolation of the exact effects relating to the omega-3 fatty acid content. It is thus very difficult to reliably ascribe definite child outcome-related benefits, or the absence thereof, to specific omega-3 fatty acids. Biomarker data failed to clarify patterns in child outcome data.

Future research should likely consider investigating the impact of specific omega-6/omega-3 fatty acid intake ratios, in no small part to control for the possible metabolic interactions involving these types of fatty acid. To produce results that are applicable to the North American population, populations consuming high omega-6/omega-3 fatty acid intake ratios should likely be randomized into trials also exhibiting better control of confounding variables than was observed, especially in the present collection of studies of child outcomes.