Flip-flopping headlines — Part One

The latest study

A recent study published in the British medical journal Lancet has elicited headlining stories claiming so many different things that consumers and healthcare professionals may be left more confused than ever. The problem is, not a single story gave an accurate presentation of what the study and body of science has actually shown. Sadly, consumers don’t realize that they’ve been caught in the middle of, not a medical debate, but a political situation with upwards of $162 billion at stake. [That’s with a “b.”]

Saddest of all, politics has affected the clarity, soundness and helpfulness of health advise for pregnant women and new mothers, and needlessly worried them.

The headlines have gone from confusing, to claiming extraordinary benefits from eating more fish to dire warnings for expectant mothers who don’t.

The Washington Post teaser said: Flip-Flopping Fish Advice

...Do flip-flopping scientific results make you feel like a fish out of water?

The AP Medical Writer’s version made its way around the world: Children of women who eat fish are smarter, study says

LONDON — Women who eat seafood while pregnant may be boosting their children’s IQ in the process, according to new research published Friday in The Lancet. The results of the study were surprising, say the authors, and contradict American and British recommendations that pregnant women should limit seafood and fish consumption to avoid potentially high levels of mercury....

The study, led by Dr. Joseph Hibbeln of the United States’ National Institutes of Health, tracked the eating habits of 11,875 pregnant women in Bristol, Britain....Hibbeln and his colleagues concluded that women who ate more than 340 grams per week of fish or sefood — the equivalent of two or three servings a week — had smarter children with better developmental skills. Children whose mothers ate no seafood were 48 percent more likely to have a low verbal IQ score, compared to children whose mothers ate high amounts of seafood.

WebMD Medical News said: Research Shows FDA Advice Does More Harm Than Good to Babies

Pregnant women who limit their fish consumption to recommended government levels may be doing their unborn babies more harm than good, according to a study from the National Institutes of Health.

Researchers found that women who ate less than 12 ounces of fish or other seafood a week while pregnant were more likely to have children with verbal and other developmental delays than women who ate more than 12 ounces each week...."Regrettably, these data indicate that the [FDA-EPA] advisory apparently causes the harm that it was intended to prevent, especially with regard to verbal development," Hibbeln says.

And American Council of Science and Health’s Health Facts and Fears wrote: Eat More Fish: Your Baby and Your Heart Will Thank You

....The sad truth is that, over the course of the past few years, many infants have been born with lower brain function than they might have had, due to the unscientific warnings issued by activists groups and our own government agencies. Hopefully, as study after study shows how misguided these advisories are, officials at the FDA or even the EPA will come to their senses and tell women the truth: eat more fish for your baby's health.

The actual science never flip-flopped, the results of this study were not surprising, the association it found did not demonstrate causation, nor did the study offer any evidence that babies have been born with mental deficits since the latest 2004 FDA-EPA fish advisories.

Before we examine the body of evidence and what’s behind the spins we’ve been hearing, let’s look at this new study with the same critical eye we would apply to any other research, such as fears about red meat and cancer, or obesity and prostate cancer, or dietary fat and infertility.

In this study, the researchers culled through a database known as the Avon Longitudinal Study of Parents and Children (ALSPAC) looking for associations between the number of servings of seafood reported being eaten by the expectant mothers at 32 weeks gestation and their children’s later development. The ongoing ALSPAC project began in 1991 recruiting 14,541 pregnant women in Bristol England, with a due date between April 1, 1991 and December 31, 1992. The database is maintained at the University of Bristol Institute of Child Health and thus far, more than 160 papers using the data have been published. In fact, in July 2004, the ALSPAC Study Team published a study in Epidemiology that was nearly identical to this one and found the same results....

Expectant mothers were mailed questionnaires four times during their pregnancy, asking them questions about how often they ate seafood and about their education, social situations and other demographic information, and 8,946 women returned the questionnaires. Of those, 8,801 mothers also completed at least one questionnaire asking them to score their children’s development and behavior in four areas (gross motor, fine motor, communication and social skills) at ages 6, 18, 30, 42 and 81 months.

Already, you’re probably wondering about the accuracy and how tightly we can interpret and compare such self-reported questionnaires. The authors compared the self-completed results with tests done under controlled conditions by trained psychologists for 1,045 of the children at 18 months and found a correlation coefficient [explained here] of only 0.54. In other words, there wasn’t a linear relationship.

At 8 years of age, 5,449 children were tested at the ALSPAC research clinic for IQ using the Weschler Intelligence Scale for Children. There’s your Trojan Number.

Since the researchers didn’t know portion sizes, they estimated actual consumptions using common consumptions in the UK. They put all of this information into their computer and, using computer modeling, looked for correlations among the women who’d eaten no fish, 1-340 grams (1-3 servings) per week and more than 340 grams (> 3 servings) per week. They reported the correlations as odds ratios, explained here.

They were unable to create a single odds ratio tenable enough to be beyond what could have mathematically popped up by random chance or statistic error. The correlations were so inconsequential that in any other study would be recognized as clinically meaningless.

In other words, they were unable to show that the amount of fish that expectant mothers ate mattered at all: eating lots of fish was not associated with smarter and more developmentally advanced children, eating no fish was not associated with children of lower cognitive abilities or development delays.

But let’s look more closely at the conclusions being made from the minor risk ratios between the kids of mothers who’d eaten no fish to those eating the most fish. Among the 13 different cognitive, behavioral and developmental measures examined in this study, the biggest number they could create after adjusting for confounders, was a 48% higher odds ratio associated with suboptimum verbal IQ among 5,407 child questionnaires at 8 years of age, although their performance IQ risks were 2% lower and full scale IQ risk ratios were only 29% higher. Clearly, these ratios are considerably below 200% that would be necessary before a relationship for this type of study would be considered tenable. The other odds ratios for various developmental skill measures were even more minor, but with no consistency from age to age. And there was virtually no difference between the mothers eating 1 to 3 servings and those eating more than 3 servings. So there was no dose response; more wasn’t better.

But making any credible conclusions about findings between the mothers who’d eaten no fish and those eating 1 or more servings a week is impossible to attribute to just the fish they ate, despite any undescribed statistical adjustments for confounding factors the researchers might have made in their computer model. Why? There were significant socio-economic and demographic differences between those mothers which would point to other potential causes for any differences, notably developmental and social opportunities for the children, and the carefulness or thoroughness of the returned questionnaires. For example, those mothers eating no fish as compared to the most fish were twice as likely to be living in crowded homes of more than 1 person per room, to be smokers and of low education; and five times more likely to be an unmarried and teenage mother.

Like so many studies claiming to find a certain food intake or lifestyle factor is associated with better health, they commonly turn out to be more indicative of better socioeconomic status than something magical in the food or behavior itself, lending a cautionary note to interpreting such correlations.

In Part Two, we’ll look at what the body of evidence shows because, as we know, a single study is just one piece in the scientific process.

© 2007 Sandy Szwarc. All rights reserved.