Current national and international guidelines

Search strategy

To identify guidelines on fish consumption in pregnancy, three search strategies were employed: (i) keywords were used for Internet searches (‘Pregnancy’ OR ‘Pregnant’, ‘Mercury’ ‘Fish’, ‘Guideline’ OR ‘Advice’); (ii) the FAO’s website on dietary guidelines^{( 11 )} was explored country by country to check for relevant guidance on fish consumption; and (iii) the European Food Safety Authority’s (EFSA) report on the health benefits of seafood^{( 12 )} includes a list of countries that have dietary guidelines – these guidelines were accessed and searched for specific advice relating to pregnancy. This yielded international guidelines from three organisations (Table 1) and national guidelines from nineteen countries (two in North America, two in Australasia, twelve in Europe and three in Asia; Table 2). All were direct advice from national or government agencies, except the guidance from Japan, which was a translation available on the Internet, and the Korean guidelines, which were reported in a newspaper article. A web-based translation service was used where necessary. Several other countries provide guidance for adults but not for pregnant women in particular, such as Greece (5–6 portions/week)^{( 13 )}, Brazil (include fish as part of a varied diet)^{( 14 )} and India (include fish as part of a varied diet)^{( 15 )}.

Table 1 Recommendations from multinational organisations on fish consumption for pregnant women related to mercury

Table 2 Recommendations on fish consumption related to mercury in pregnancy in individual countries

ND, not dated.

* Includes an infographic illustrating the categories and portion sizes. Endorsed by the American College of Obstetricians and Gynaecologists with the additional advice that pregnant women should avoid raw and undercooked seafood^{( 76 )}.

† Other guidelines and advisories for North American populations are shown in Oken et al. (2012)⁽ Reference Oken, Choi and Karagas ^{28 )}.

‡ Based on the NHS Choices website ‘Should pregnant and breastfeeding women avoid some types of fish?’^{( 21 )}.

§ Online translation from original.

║ Newspaper article reporting advice from the Korea Health Promotion Foundation.

¶ Raw shellfish is not advised in pregnancy as it can be a microbiological hazard, but cooked shellfish can be eaten freely^{( 22 )}.

** High levels of Cd, Pb, polychlorinated biphenyls and/or dioxins.

†† A serving is defined as 4 oz for an adult (about 110 g). The guidance also applies to ‘young children’, who are advised to eat 1–2 servings of fish/week starting at age 2 years (child’s serving defined as 2 oz (about 55 g)).

‡‡ Under the guidance, canned tuna does not count as oily fish so is not included in the maximum of two portions of oily fish per week. However, because of the higher Hg level in tuna, if eating canned tuna, the advice is not to pick fresh tuna as one of the tally of oily fish.

§§ The NHS Choices website ‘Your pregnancy and baby guide. Foods to avoid in pregnancy’ omits these points^{( 20 )}.

║║ A type of whelk.

Evidence for beneficial effects of fish on child health and development

The guidance in the UK was developed from the recommendations of the Scientific Advisory Committee on Nutrition published in 2004. These were based on calculation of the Hg content of fish that would result in exposure at a provisional tolerable weekly intake of 1·6 µg/kg body weight for pregnant women as being sufficient to protect against adverse effects on neurodevelopment in the fetus (3·3 µg/kg body weight for breast-feeding women)^{( 36 )}. Similar approaches have been adopted by other countries, but they fail to take into account the potential risk-payoff from fish consumption⁽ Reference Hibbeln, Davis and Steer ^{37 –} Reference Maycock and Benford ^{39 )}: fish is a rich source of protein, as well as other nutrients required for fetal neurodevelopment including iodine, Se, choline, vitamin D and long-chain n-3 fatty acids. Iodine levels particularly have been shown to be low in pregnant women in the UK⁽ Reference Bath, Walter and Taylor ^{40 )} and this has been associated with adverse effects on offspring IQ (intelligence quotient)⁽ Reference Bath, Steer and Golding ^{41 )}. Indeed, studies from a UK birth cohort have shown evidence of a positively beneficial effect of eating fish in pregnancy on a range of developmental outcomes in the child⁽ Reference Golding, Hibbeln and Gregory ^{42 –} Reference Golding, Gregory and Iles-Caven ^{44 )}. Evidence for risk is also sometimes based on studies from the Faroe Islands, where Hg exposure is derived from consumption of pilot whales rather than fish⁽ Reference Grandjean, Weihe and White ^{45 )}. The evidence for beneficial effects of fish consumption on many aspects of maternal health and child development has increased in recent years. In the UK, the Avon Longitudinal Study of Parents and Children (ALSPAC; observational birth cohort) includes data on prenatal measures of Hg exposure, together with maternal fish consumption and a range of childhood outcome indicators. In this cohort, consumption of two to three portions of fish per week is associated with beneficial effects on child development, suggesting that limiting fish intake might actually be detrimental⁽ Reference Hibbeln, Davis and Steer ^{37 )}. Fish consumption made only a small contribution to the variation in blood levels of Hg during pregnancy⁽ Reference Golding, Steer and Hibbeln ^{46 )}. There was no effect on the likelihood of the baby being born with a low birth weight or preterm; indeed, birth weight was lower in the babies of mothers who did not eat fish during pregnancy, suggesting that fish consumption has a beneficial effect on birth weight⁽ Reference Taylor, Golding and Emond ^{47 )}. Other measures of child development, such as child behaviour, social, motor and communication skills, and IQ, similarly showed no association with prenatal Hg exposure⁽ Reference Golding, Hibbeln and Gregory ^{42 –} Reference Gregory, Iles-Caven and Hibbeln ^{48 )}. These findings have been substantiated by similar evidence from outside the UK, for example from the Seychelles Development Study, where there is daily fish consumption and Hg exposure levels are about ten times higher than typical exposures in the USA: a variety of neurodevelopmental tests have been applied at ten age points in 24 years of follow-up without any evidence of associations with prenatal exposure to Hg⁽ Reference Myers, Davidson and Cox ^{49 –} Reference van Wijngaarden, Thurston and Myers ^{51 )}. Similarly, no associations of prenatal fish intake or Hg exposure with cognitive outcomes were found in children aged about 7 years in Project Viva in the USA, despite adjustment for long-chain n-3 fatty acids (DPA+EPA) and Se⁽ Reference Oken, Rifas-Shiman and Amarasiriwardena ^{52 )}. In the Norwegian Mother and Child Cohort Study, seafood intake was positively associated with birth weight, whereas Hg exposure was negatively associated, suggesting that the balance of the risks and benefits of seafood might need further quantification⁽ Reference Vejrup, Brantsaeter and Knutsen ^{53 )}. However, unlike the other studies described, prenatal Hg exposure was calculated from dietary intakes (FFQ) and may not represent Hg exposure in the same way. Other studies have shown similar positive associations of child neurodevelopment with prenatal fish intake⁽ Reference Starling, Charlton and McMahon ^{54 )}.

How much fish do pregnant women eat?

It would therefore seem to be disadvantageous if the guidelines had the unintended consequence of reducing fish consumption in pregnant women. To understand the impact of these guidelines on fish consumption, it is necessary first to have accurate nationally representative data specifically from pregnant women. Fish consumption in women of childbearing age and pregnant women has consistently been shown to be below recommended levels. In a compilation of data on pregnant women from nineteen European birth cohort studies with recruitment from 1996 to the date of publication (2014)⁽ Reference Leventakou, Roumeliotaki and Martinez ^{55 )}, the median fish intake ranged from 0·4 times/week in the Netherlands (the Generation R study) to 4·5 times/week in Spain (Childhood and Environment Project (INMA)). The median oily fish intakes in Italy (Genetic and Environment: Prospective Study on Infancy in Italy (GASPII)), Portugal (Generation XXI), Spain (INMA) and Poland (Polish Mother and Child Cohort Study (REPRO-PL)) were more than twice the overall median intake of 0·5 times/week. Portion sizes of different fish types varied from 100 to 150 g across cohorts that included this information. Thus, in fourteen of the nineteen studies, the median intake was less than 2–3 times/week, and no study reported an intake of oily fish of more than 1 time/week (six studies had no data on oily fish intake). A more recent (2017) compilation of seventeen cohorts in eleven European countries plus one cohort in the USA, which included some of the same studies as the previous compilation, found an overall consumption of 1·5 times/week (oily fish 0·6 times/week): women in all but three of the cohorts (Spain, INMA; Portugal, Generation XXI; Italy, NINFEA) ate fish less than 2 times/week, and none of the cohorts reported oily fish intake more than 1 time/week⁽ Reference Stratakis, Roumeliotaki and Oken ^{56 )}.

More specifically, in the UK (where the recommended intake is for at least 2 portions fish/week with at least 1 portion oily fish/week; Table 2), women aged 19–64 years participating in the National Diet and Nutrition Survey (NDNS) from 2008 to 2012 ate a mean of 22 g fish/d (about 1 portion/week) including just 8 g oily fish/d (about 0·3 portion/week)^{( 57 )}; mean consumption in pregnant women enrolled in ALSPAC was 235 g/week (about 1·5 portions/week) but 12 % ate no fish at all⁽ Reference Hibbeln, Davis and Steer ^{37 )}; in the Southampton Women’s Study, total fish consumption was 1·8 times/week and oily fish consumption was about 0·5 times/week⁽ Reference Leventakou, Roumeliotaki and Martinez ^{55 ,} Reference Fisk, Calder and Inskip ^{58 )}. Findings in the USA are similar: women of childbearing age in the National Health and Nutrition Examination Surveys (NHANES) ate a median of 81 g/week (about 0·5 portion/week) and 23 % reported not eating any fish^{( 59 ,} Reference Cusack, Smit and Kile ^{60 )}; mean fish intake in pregnant women was 1·5 portions/week and 14 % never ate any fish. In Australia, mean intakes are a little higher at about 28 g/d⁽ Reference Taylor, Collins and Patterson ^{61 )}, but still fall far short of national recommended intakes (Table 2).

The methodology used in the surveys conducted is critical to the interpretation of studies on fish consumption: dietary recalls or dietary records are not ideal to capture an item that might be infrequently consumed and have the potential to underestimate fish/seafood intake. For example, in data compiled by EFSA, all country-level surveys were conducted with 24 h dietary recalls or dietary records and this was noted as being likely to ‘have the potential for overestimating the high ends of the distribution of fish/seafood consumption’^{( 16 )}. EFSA also noted that conversion of values from daily to weekly to enable comparisons can magnify inaccuracies, as well as there being considerable between-country variation in mean portion size. Methods based on FFQ are likely to be more accurate for this type of low frequency food, but still present difficulties over the length and depth of detail in the questionnaire⁽ Reference Oken, Guthrie and Bloomingdale ^{62 )}.

What evidence is there for the effect of guidelines?

Despite evidence of fish consumption below recommended levels in pregnancy, there has been very little research on the impact of the guidelines on consumption levels or consideration of how consumption levels could be optimised. Most public messages usually struggle to have impact, but there is some evidence that messages that are more ‘alarming’ achieve greater change. As an illustration of this effect, the US Food and Drug Administration issued an advisory notice on avoidance of predatory fish and limitation of consumption of all other fish in 2001 (before this time there was no specific guidance for pregnant women as Hg from commercial fish was not thought to pose any significant health threats). The result of the advisory, which was widely promoted, was a reduction in total fish consumption in pregnancy by about 0·4 portions/week during the year following the notice, with diminished consumption of dark-meat fish, canned tuna and white-meat fish⁽ Reference Oken, Kleinman and Berland ^{63 )}. Shimshack and Ward⁽ Reference Shimshack and Ward ^{38 )} provided estimates of the observed effects of this advisory on Hg and n-3 fatty acid intakes and found that Hg intakes across the US population did fall by 17 %, but n-3 intakes also fell by 21 %, providing evidence of an unintended consequence. They attributed the fall in n-3 intake to ‘coarse information and broad behavioural guidance’ with lack of a detailed explanation of the recommendation that consumers select ‘a variety of other kinds of fish’⁽ Reference Shimshack and Ward ^{38 )}. The 2001 advisory was replaced in 2004 and again in 2017 (Table 3). For the 2004 guidance^{( 64 )}, there is some evidence that in the face of confusing and complex guidelines, and the lack of readily available advice, many women gave up eating fish: analysis of focus groups for twenty-two pregnant women who ate <2 portions fish/week in the USA in 2009/10 showed that many of them had received advice to limit their fish intake and knew that fish could contain Hg. Because of this advice and a lack of knowledge about which types of fish were safer to eat, many of the women reported that they would rather avoid fish altogether than risk harm to themselves or their baby. They felt that advice from a doctor on eating fish and readily available information on which fish are safe to eat would have encouraged them to eat more fish⁽ Reference Bloomingdale, Guthrie and Price ^{65 )}. There is direct evidence of lower intakes in pregnant women compared with non-pregnant women in Australia, where mean intakes were 28 g/d in pregnant women but significantly greater at 33 g/d in women who were not pregnant, trying to conceive or <1 year postpartum⁽ Reference Taylor, Collins and Patterson ^{61 )}.

Table 3 Changes in the US Food and Drug Administration (FDA)/Environmental Protection Agency guidelines on fish consumption for women planning to become pregnant and those who are pregnant or breast-feeding

In 2015 the American College of Obstetricians and Gynecologists issued guidance reflecting the 2014 FDA draft advice^{( 76 )}; this was superseded by a practice advisory in 2017 in line with the 2017 FDA advice^{( 77 )}.

* Cited in Shimshack and Ward⁽ Reference Shimshack and Ward ^{38 )}.

† First accessed 27 September 2015. Not available online when access attempted again on 5 October 2017 (originally available at http://www.fda.gov/Food/FoodborneIllnessContaminants/Metals/ucm393070.htm).

‡ About 330 g.

There is some evidence, however, that targeted education during pregnancy can result in increases in fish consumption. Fifty-five pregnant women in the USA who were low fish eaters (≤2 servings/month) were randomised to receive control messages, advice to eat low-Hg fish or advice to eat low-Hg fish plus coupons to buy fish: fish consumption increased in both intervention groups without an increase in blood Hg levels compared with baseline values⁽ Reference Oken, Guthrie and Bloomingdale ^{66 )}. Although this was a pilot study with small numbers of women, it does indicate that women are receptive and willing to increase their fish intake and can achieve this without increasing Hg exposure when given appropriate targeted advice.

Summary and conclusion

There is great variation in the content, complexity and presentation style of guidance for pregnant women on fish consumption between countries. This partly reflects local environmental conditions, species availability and consumption preferences, and to some extent local preferences for delivery of public health messages. The guidelines have largely been based on the Hg content of fish with far less consideration being given to the positive beneficial effects of nutrients provided by fish. There is evidence for low levels of several nutrients provided by fish – notably iodine – in pregnant women and it is essential that pregnant women are given balanced advice to make informed choices.

There is some evidence that pregnant women find the advice confusing and prefer to give up eating fish altogether rather than take the risk of harm. There is general agreement that pregnant women should eat at least two portions of fish per week, but this message is not always clear and prominent. Fish consumption falls below this level in many countries and this may have adverse effects on offspring health and development. Guidance needs to be clear, simple and memorable, and appropriately disseminated, to achieve impact⁽ Reference Taylor, Golding and Emond ^{67 )}. Guidance could include visual rather than narrative content. Use of technology, for example the development of apps, could enable women to record their fish consumption in real time and give feedback on compliance with guidance over a week or other time period.