Medical Examiner

The Scales Have Tipped on Fish and Pregnancy

The benefits of eating fish during pregnancy vastly outweigh the risks. So why do the recommendations still put women off from seafood?



When I was pregnant two years ago, I spent a lot of time on food—both eating it and Googling it. There was so much to consider: nutrition and cravings on one hand, foodborne illnesses, potential toxins, and the risk of heartburn on the other. I once spent 45 minutes deciding which hoagie to order for lunch, and that was before I had even picked my condiments.

Take, for example the question of fish. Fish is a good thing to eat, generally, but what about the risk methyl mercury poses to a fetus? I read the Environmental Protection Agency/Food and Drug Administration guidelines, which cautioned me to limit my intake to 12 ounces per week. It sounded pretty frightening, so I cut back. I ate more veggies and more chicken. Unfortunately, by skimping on fish, I probably wasn’t reducing my risk of harm, and I may have been skimping on important brain-boosting nutrients that aren’t found elsewhere.

I’m not worried—my daughter is an avid Lego architect who does a killer rendition of her favorite song, “Snuggle Puppy.” I mean, it’s pretty obvious the girl is a genius! But my confusion about how much fish to include in my diet is not unique. A 2003 study by Emily Oken at Harvard showed that pregnant women who saw the same guidelines I saw tended to reduce their fish intake, too.

The problem is, they shouldn’t. Philip Spiller, the former director of the FDA’s Office of Seafood, and his colleague Michael Bolger, the former director of chemical assessment at the FDA, spent years completing a large quantitative risk assessment to find out whether the benefits of eating fish outweighed the possible harms for pregnant women. Their conclusion? Pregnant women should eat more fish, not less. Clark Carrington, the main scientist on the project, even used his free time to create a calculator where pregnant women could enter their favorite fish to see the benefits and the risks in numerical form. Unless you are consuming outlandish quantities of seafood (more on that later), the benefits almost always win out. (Sushi, it should be noted, presents a different issue: The risk it poses is because it might carry harmful microbes, which don’t have a chance to be killed during cooking.)

The FDA was aware that its guidelines were discouraging fish consumption. So on Jan. 18, two days before the inauguration, it released new guidelines. Do these guidelines make use of Carrington’s handy dandy tool to show women how safe and beneficial fish are? They do not. Instead, the new recommendations simply state that pregnant women should eat 8–12 ounces of fish per week, rather than stating that pregnant women should limit their fish intake to 12 ounces a week. It still warns women that “methyl mercury can affect the central nervous system, particularly the developing brain of a fetus.” Spiller, Bolger, and Carrington’s research is not specifically mentioned, thanks to a dispute with the Environmental Protection Agency. The new guidelines are still based on an EPA reference dose from the early 2000s.

In other words, the new guidelines don’t solve the problem.

On Thursday, a dozen former FDA officials and other scientists submitted a petition to the FDA asking them to withdraw the guidelines. They request that it be made more clear that the benefits of eating fish outweigh the risks, even for neurodevelopment. In particular, they request language that follows the Food and Agricultural Organization of the United Nations and World Health Organization advice to “emphasize the net neurodevelopmental benefits to offspring of women who consume fish … and the neurodevelopmental risks to offspring of women who do not consume fish.”

“It’s important for pregnant women to understand that 12 ounces is not the edge of danger, i.e. that 12 ounces is not a dividing line between safe and unsafe or anywhere close to it,” Spiller says.

* * *

Spiller started working on the fish guidelines almost 20 years ago. At the time, he says, the research on the risks of methyl mercury was very contradictory. It was known that seafood with extremely high levels of mercury, such as fish from the highly polluted waters of Minamata Bay in Japan, could cause developmental delays among the children of women who ate it. But it was not known if these same consequences would apply to women who ate fish with low doses of mercury, like the salmon and cod sold in the supermarkets of the United States. Along with being a pollutant, mercury is a naturally occurring element (often occurring in nature naturally bonded to methyl, hence methyl mercury). All fish have at least some methyl mercury, but the question—as always with toxicity—is how much is harmful?

In 2000, the National Academy of Sciences was tasked with developing a reference dose: the amount of mercury that the EPA would say was acceptable to eat each day. At the time, the scientific community was perhaps reasonably much more focused on the harms of methyl mercury than on the benefits of fish. The NAS based its analysis on one large study in the Faroe Islands, an island between Iceland and Norway. The Faroe Islands study had identified pregnant women, tracked their mercury intake, and followed their children as they grew up. At age 7, study authors found that children of women with high mercury concentrations performed worse on several neurological tests such as the Finger-Tapping Test and the Boston Naming Test. By the time the kids had reached 22, they found that a tenfold increase in prenatal mercury exposure was associated with a decrease of 2.2 IQ points.

Sounds scary, right? But, there were questions about the Faroe Islands study and its applicability to the U.S. On Faroe Islands, people don’t only eat fish—they also eat pilot whale. Thanks to its higher place on the food chain, pilot whale can have 100 times the concentration of methyl mercury compared to fish, as well as high concentrations of PCBs and other chemicals. Were the mercury intakes of these women applicable to American women?

Pal Weihe, one of the authors of the Faroe Islands study, agreed this was an open question. In his role as chairman of the public health board in the Faroe Islands, he said devising a public message was pretty easy. “My recommendation has been very clear to people,” Weihe said. “Stop eating pilot whale meat and blubber, because they are not beneficial.” I asked if he thought the Faroe Islands study results could have been caused by the consumption of pilot whale. Weihe said he couldn’t answer that question—yet. His team is currently following a cohort of children that were born after the whale advisory was released, and they should have the results next year.

But, in 2000, the NAS—with its focus on risks—based the reference dose largely on the Faroe Islands study. It did not include data from another study in the Seychelles, an island off the coast of East Africa, that had found no detrimental effects. In addition, the NAS built a tenfold “safety factor” into the reference dose. In other words, it calculated an acceptable level of intake for methyl mercury, based primarily on the Faroe Islands study, and then divided that number by ten, just to be safe. Other agencies around the world have used a safety factor of four or five. That makes the U.S. reference dose the most cautious in the world, at .1 microgram per kilogram of bodyweight per day. (In Canada, for instance, where they used a safety factor of 5, their reference dose is 0.2 micrograms per kilogram of body weight per day.)

When the NAS report was released, it noted that as many as “60,000 children are born each year at risk for neurodevelopmental effects” in the U.S. What it really meant, as it later told the FDA, was that 60,000 pregnant women might be exposed to mercury above the reference dose, though the NAS wasn’t sure if this exposure actually meant the women’s babies were at risk of neurodevelopmental delays. But this claim “basically caused the media to jump out of their skin,” Spiller said. There was widespread concern among the public.

The FDA needed to issue a recommendation to pregnant women about how much fish they ought to eat, but the low reference dose and the lack of certainty about causation offered by NAS put the FDA in a gray area, scientifically speaking. “We didn’t know if people were actually being harmed above the reference dose, or what the nature of the harm was,” Spiller said. “Anything and everything we would do at that point would have to be done as a matter of precaution. We had no analytical basis for it.”

Their deliberations were circuitous, Spiller said. There was a feeling within the agency, for instance, that it didn’t want to undermine the American Heart Association’s recommendations to eat two servings of fish a week, which was good for heart health. “Of course, that has nothing to do with fetal neurodevelopment, it’s a different issue altogether,” Spiller said. “But if the AHA was saying eat at least two servings, we couldn’t say eat less than two servings. So we decided to say don’t eat over two servings.”

In 2001, the FDA recommended that pregnant women and women who might become pregnant limit their fish intake to 12 ounces a week, and avoid several high-mercury fish, including shark, swordfish, king mackerel, and tilefish. In 2004, it issued a similar guideline in partnership with the EPA, who had jurisdiction over recreational fishing.

Bolger, then the director of chemical assessment at the FDA, also remembered the issuing of this recommendation as somewhat haphazard. The FDA did not do any analysis or have any specific information to suggest these particular fish were causing harm to pregnant women. It simply knew those fish had the highest concentrations of mercury. “It was a precaution,” he said.

Still, coverage in the media focused on the reference dose and whether women exceeded it, rather than on whether exceeding the reference dose had any potential to cause harm. But still, the coverage was tough to ignore if you were pregnant—for me, it was a piece in Consumer Reports that turned me off from fish. People listened to the recommendations—when the Centers for Disease Control and Prevention looked into it, they found that only 6 percent of American women ate enough fish to exceed the reference dose. Were health officials worried about that 6 percent? Not really: The women were still eating well below the amount that might even begin to cause harm to a fetus.

It’s possible to see this as the system working. There was evidence of possible harm, and regulators took action to limit that harm. Except, remember, our understanding of harm had come primarily from one study on a population with an extremely high rate of exposure to the toxin.

Spiller and Bolger and others at the FDA felt that the reference dose was not necessarily the best way for the general population to think about methyl mercury. Consumers seemed to misunderstand it, and there was a widespread misperception (perhaps perpetuated by the NAS report) that the reference dose was a dividing line between safe and unsafe. And new information had started to shake up the scientific literature on methyl mercury. In 2007, for instance, a study with over 11,000 participants in the U.K. found that pregnant women who ate more than 12 ounces of fish a week had children with higher IQs.

The authors hypothesized that the benefits were due to the nutrients in fish, such as omega-3 fatty acids, which could be beneficial for brain development. (This had been suggested, for instance, by studies of genetic disorders that impacted the brain but were subsequently mitigated by DHA supplementation—DHA is an omega-3 fatty acid.) Similarly, in 2008, a study of over 25,000 women in Denmark found developmental benefits from eating fish. In 2016, a study of over 1,000 women in Spain found similar results. The scientific community started to realize that it was likely that both the nutrients in fish and the mercury in fish was acting on similar pathways in the brain, one in positive ways and the other in negative ways.

“These studies really caused us to take notice,” Spiller said. “The notion that the association could be beneficial had not occurred to us.”

Spiller, Bolger, Carrington, and their colleagues at the FDA realized that in order to understand the complexities and explain it to women, they had to create a model that would take into account both the nutrients in fish and the mercury in fish. They compiled all the epidemiological data they could get their hands on and used it to create a quantitative risk assessment.

The results of their calculations—called the Net Effects report—were a revelation. The FDA found that every fish, regardless of mercury content, was correlated with benefits. In the same way that mercury exposure was associated with lower IQ, omega-3 exposure was associated with higher IQ. “The peak benefit varies from fish to fish, but for the best fish, it’s a little over three IQ points,” Spiller said.

There was a point at which the risks of mercury would outweigh the benefits. But the thing was, for the vast majority of fish, this didn’t happen until you were eating a huge—even abnormal—amount of fish. “One of the big questions was, how much fish do you have to eat before it becomes adverse: That’s a big deal for fish-consumption advice,” Spiller said. “For very low-mercury fish, it gets crazy. It’s more than you could possibly eat. It gets to 50, 60, 70, or over 100 ounces of fish a week. It’s just impossible.”

Again, when it comes to toxicity, it’s all about the dosage. For example, according to the net effects report, you’d have to eat 1,537 ounces of shrimp a week before the risks outweighed the benefits. That’s about 96 pounds. According to the food blog Chowhound, you’d only need one-quarter of that amount to host a 75–80 person engagement party.

Granted, shrimp is a very low-mercury fish. But, for other low-mercury fish like salmon or cod, you’d have to eat 774 ounces and 173 ounces per week, respectively, for the risks to outweigh the benefits. A serving a fish is usually 4–6 ounces—and no one eats cod three times a day, seven days a week. Even for fish with medium amounts of mercury, like halibut, you’d have to eat roughly 72 ounces a week for the risks to outweigh the benefits.

The typical amount of fish eaten in America in 2015 was 4.77 ounces a week. Pregnant women are known to eat even less.

When the Net Effects report was completed in 2014, Spiller knew that his own agency’s public-facing guidelines sent the wrong message. Their assessment found that the optimum amount of fish to eat—the amount that allowed a pregnant woman to capture the most benefits from a fish—was between 8–10 ounces a week. That was within the “less than 12 ounces” guideline, but it didn’t change the fact that the entire premise of the fish guidelines, as conceived in 2001 and 2004, was faulty. “Advising how to get the most benefits for brain and nervous system development is a big change from advising how to stay safe from harm from methyl mercury,” Spiller said.

In addition to the FDA’s Net Effects assessment, the European Food Safety Authority, the Food and Agricultural Organization of the United Nations, and the WHO have all found that “when fish are eaten during pregnancy, improvements in fetal neurodevelopment are much more likely than deficits from methyl mercury.” If the guidelines weren’t shifted, Spiller was convinced women would continue to limit their intake based on fear rather than increase it based on benefits.

Spiller left the FDA in 2014. His boss, who was supportive of the Net Effects report, left around the same time. After that, he said, there was no one who felt they had ownership of the report. When the fish guidelines were released in January, they included no mention of the Net Effects report—and were instead based on the EPA’s reference dose from the early 2000s.

* * *

In their petition, Spiller and his former colleagues at the FDA say the new guidelines are misleading and very likely to be misunderstood. Instead of being worried about the hazards of methyl mercury, Spiller says, women should be trying to maximize the benefits that can be derived from fish. The new guidelines don’t come anywhere close to making that clear.

The 8–12 ounce guideline should be presented, they say, as an optimum range for enhancing neurodevelopment, not a maximum past which fetal harm is guaranteed. “It would be unfortunate if pregnant women were to misinterpret what is actually an optimum amount per week as being a dividing line between no risk and meaningful risk,” they write.

The petition’s authors say that women can maximize the benefits of eating fish by choosing fish that is high in nutrients like omega-3 fatty acids and low in methyl mercury. The current guidelines sort fish into categories based solely on their mercury content, without taking into account the omega-3 fatty acids or the other nutrients in fish. The petition proposes another method of categorizing fish—based on the Net Effects assessment—that would take both omega-3 fatty acids and methyl mercury into account, in order to allow women to chose fish more strategically.

Tom Brenna, professor of pediatrics and of chemistry at the University of Texas at Austin, and a leading expert on omega-3 fatty acids, signed the petition. He reminded me that “no amount of salmon, for instance, is known to cause problems.”

Weihe, one of the authors of the Faroe Islands study, had a message that was slightly more cautious but surprisingly similar. He does think mercury can be toxic at low levels, but says this has to be balanced against the nutrients in fish. “I recommended people to eat fish species low in mercury without any limitation,” Weihe said. “If you are eating low-mercury fish, I can’t imagine that you could eat so much that you would have any problem.”

He added that pregnant women should be aware of high-mercury fish and try to limit them. Spiller agrees, saying that if women want to exercise an abundance of caution, they should avoid the fish the FDA lists as high in mercury as long as they’re still eating plenty of other types.

* * *

There’s another issue at play here—the EPA’s resistance to adjustment. Both Bolger and Spiller suspect that the resistance was due to the agency’s own incentives. The reference dose doesn’t just inform fish recommendations, it’s also used to limit mercury emissions from coal-fired power plants, which is one of the ways too much mercury ends up in fish. Having a lower reference dose helps the EPA argue for emissions limits. (The EPA did not respond to a request for comment on this story.)

In the end, this speaks to the broader problem with the fear over fish. Being pregnant already makes you subject to plenty of fears, anxieties, and unwanted advice. If the government is going to add to that stress, it should do so for good reason. It seems like the initial recommendations were built on good intentions for public safety but now, decades later, our understanding of the risks and benefits has shifted. And the people making the choices about what to eat during their pregnancies deserve to know so they can do what’s best for them.

I know that when I was pregnant, the smell of cooked salmon made me sick. But if someone wants to eat pounds upon pounds of shrimp, she should know that it’s safe.