Better stats. Better science. Better outcomes.

Better stats. Better science. Better outcomes.

Through statistics, Duke's Amy Herring is changing what we know about chemical exposure, uncovering adverse health outcomes for pregnant women and their babies.

All of us are exposed to hundreds of synthetic chemicals every day with unknown effects on our health. That's worrisome enough, but some chemicals could threaten healthy pregnancies by interfering with the rapid physiological changes that take place during growth and development.

In a sea of chemical exposure, how can researchers pinpoint which ones might be critical for pregnant women to avoid? It's difficult question to answer, says Amy Herring, the Sara and Charles Ayres Distinguished Professor of Statistics at Duke. The chemicals are so ubiquitous that virtually everyone has some exposure. Add to that, it's unethical to purposely expose anyone to potentially harmful chemicals during a research trial, making it impossible to do a randomized controlled study.

That means researchers have to conduct observational experiments, in which they measure the levels of one or more chemicals in a pregnant woman's blood or urine to assess her exposure. Then they follow the baby's developmental progress over months or years to see if they can discern any effect.

With today's technology, it's now possible to measure the levels of hundreds of chemicals in bodily fluids, including plasticizers, pesticides, flame retardants, and ingredients in personal care products. With so many, it is hard to tease out the effects of any one of them.

Amy Herring writing formula on whiteboard in front of students

“You might be looking at a chemical in nail polish,” Herring says, “but if I get a pedicure, I may be more likely to use color or chemical treatments on my hair, or to enjoy perfumed bath and body products, so that it's hard to point a finger at a single compound.

Herring helps researchers discover more about the effects of chemical exposure by using her superpower—statistics. She uses that superpower to find relationships between exposures and effects within huge datasets. She particularly enjoys the complexities that come along with studies involving variables that are difficult to quantify—like chemical exposure.

Herring has a lot of tools in her statistical toolbox, but sometimes there isn't an existing technique to ferret out particularly complex relationships in a dataset. In that case, she and her graduate students develop new statistical methods. “We're always improving on the state of the art,” she says.

One of her collaborators, Duke environmental epidemiologist Kate Hoffman, says, “That statistical magic — I have no idea how it happens. That's Amy's cool factor.”

Herring's work is paying off. She and her colleagues have already discovered that exposure to certain flame retardants during pregnancy is associated with increased activity and impulsivity in early childhood. Along with Hoffman and Heather Stapleton, Ph.D., the Ronie-Richele Garcia-Johnson Distinguished Professor of Environmental Sciences and Policy, she is working to identify other chemical exposures that might be unsafe for pregnant women and children, such as chemicals used in certain personal care products.

Their work has the potential to transform future outcomes by allowing women to avoid chemicals that might harm their unborn babies. That's particularly important to Herring, whose own son was born with a very rare birth defect (“He's fine now, except for an impressive chest scar,” she says).

“I was drawn to biostatistics because of the chance to have an impact on people's health and well-being,” she says. “And having the ability to collaborate with exceptional research teams to provide the best possible information for pregnant women to use in making health and lifestyle choices is particularly meaningful to me.”