Cynthia Graber and Nicola Twilley: Personalized Science Reporting With a Bite

The defrosted muffins we ate each morning of the two-week nutrition study were hard to choke down — leaden, overly sweet, reminiscent of mattress-stuffing. By the third day, we’d stuck a needle in each of our fingertips for blood draws, and the resulting bruises made it difficult to type.

It took so much time to weigh every ingredient for every non-muffin dish we consumed — and then to photograph the resulting lunch or dinner — that our meals often were cold by the time we ate them.

One thought kept us going: This will make a great Gastropod episode. How better to report on a personalized nutrition study for a podcast about the science and history of food than by taking part in it?

We had reached out to British genetic epidemiologist Tim Spector in late 2018. His dozens of studies on more than 13,000 twins in the United Kingdom, whom he’d been investigating for more than two decades, had led him to suspect that variations in how we metabolize food might not just be genetic.

After all, in sets of identical twins, with identical genomes, one twin might be 30 or 40 pounds heavier than the other. As founder of the British Gut Project, Spector already had begun to explore the connections between weight and the gut microbiome — the trillions of microbes that inhabit our intestines.

At Gastropod, we love nothing more than guts, microbes, and emerging science, so we emailed Spector to see whether he had any upcoming research we might want to cover. Spector told us that he knew the gut microbiome was only one of a host of factors that shape our individual reaction to different foods. Other potentially significant influences include sleep, stress, and exercise.

To try to tease this all out, Spector had joined with a group of internet entrepreneurs and a global team of researchers to launch the world’s largest and most ambitious study of personalized nutrition to date. Called PREDICT, the study was to include 700 identical twins, 300 individual British volunteers, and 100 subjects from the United States.

The range of individual variation in responses to food was the focus of the first phase of the study. It also sought to pin down how much of that difference is genetic, how much is due to gut microbes, and how much is due to any of the dozens of other factors that could affect our metabolic processes. In the end, Spector told us, he hopes that understanding those differences might help individuals reduce their risk of weight gain or developing diabetes and other diet-related diseases.

While most of the participants were recruited from Spector’s original cohort of British twins, he told us there was a small validation arm of the study, the 100 aforementioned American subjects, based at Massachusetts General Hospital in Boston. Cynthia lives in Boston, and Nicky was going to be traveling there as part of a Gastropod reporting trip. Did we want to participate, he asked?

Of course. We were in.

We’d both drawn on our personal experiences in past science reporting. Nicky had been given a third degree burn in an MRI machine as part of a story on the neuroscience of pain, published in The New Yorker. Cynthia had written for Wired about her family’s medical history, and her personal decision to refuse to be tested for the BRCA mutation (linked to a higher risk of breast and ovarian cancer).

While neither of us had participated as subjects in a scientific study before, PREDICT offered the perfect opportunity to illustrate the challenges inherent in nutrition research — something we frequently flag for Gastropod listeners when discussing what we do and don’t know about the health impacts of food. What better way to bring to life the difficulties of standardizing intake or the huge gaps in food recall that researchers deal with than by struggling through the process ourselves. Plus, perhaps we’d learn a little something about our individual metabolic responses while doing so.

And so, in February 2019, we spent a very long day at Massachusetts General Hospital. We had continuous blood glucose monitors stapled onto our upper arms, we gave samples of spit, urine, and feces, and we learned to use the food logging app developed specially for the study. We ate bright blue muffins for breakfast (and spent the next couple days anxiously looking for their telltale color in the toilet pan).

From Boston, we traveled to Washington, D.C., for the 2019 AAAS Annual Meeting, each with a bag of defrosted standardized muffins and blood prick kits. After two days watching friends and colleagues enjoying complimentary cocktails and passed hors d’oeuvres at the various AAAS receptions while we weighed out almonds, carrots, and chunks of canned tuna, we began to question exactly how much we were willing to suffer for our art.

All the while, for Gastropod, we recorded the sounds of the hospital day, our reactions to the muffins, and the reactions of others to our bizarre weighing and photographing rituals. The more difficult it was for us, the funnier we knew the resulting episode would be. We’ve found that humor is frequently an effective way to draw listeners into difficult, complicated scientific topics.

But while we thought our personal experience added something valuable to the science reporting — making a research study concrete, compelling, and understandable for our listeners — not everyone agreed. Editors at The New York Times, for which we wrote an accompanying article published the same day as our Gastropod episode, were concerned that we might be unable to report objectively on a study in which we were participating. In the end, we wrote the Times article as a conventional news story without the personal details we brought to our Gastropod episode.

Objectivity is a hard thing to be objective about, of course, but we genuinely felt we could report dispassionately about a study in which we were involved, and that our participation enabled us to probe the science more effectively. We met with Spector in person only after we’d completed the two-week study, and we were able to have a more comprehensive conversation about the methodology, assumptions — and limitations — underlying the trial precisely because we’d been through the process ourselves.

For example, we were asked regular questions via the online app about our mood and stress levels. Such information might help the team learn more about the complicated relationship between food and mood, Spector explained. But, as participants, we were aware of external factors the app wasn’t collecting. Cynthia’s aunt died in the middle of the study; Nicky is an eternal optimist. How could the team sort out the influence of food as opposed to other factors, when the app didn’t collect such meta-level information?

The study’s potential impact on public health — a far more nuanced series of trade-offs than Spector and his colleagues seem to foresee — became clearer to us because of our participation. For example, given how much time we spent weighing and logging food, it quickly became obvious that this kind of personalized nutrition advice will likely remain the preserve of people who already have the time and interest to optimize their health — in other words, those who need it the least. In addition, receiving our own preliminary results from the nutrition study gave us unusual insight into some of the challenges scientists face in converting their research findings into health prescriptions.

The Massachusetts General Hospital team emailed us each some early results in May, analyzing our individual metabolic responses to fat, carbohydrates, and protein, based largely on our blood levels of glucose, insulin, and triglycerides after consuming meals of standardized muffins and drinks. The two of us did have some marked differences, particularly in how we responded to standard muffins that were high in fat.

We discussed these results with Spector and asked what changes he might recommend in our diets based on the findings. Our reactions, over the next few days, were surprisingly complicated and, at times, emotional. (You can listen to the full Gastropod episode here.) In short, our experience gave us the insight to explore whether this type of personalized data might help — or harm — a participant’s personal relationship to food.

While discovering our own metabolic strengths and weaknesses was unexpectedly challenging, we’ve found the study’s early findings, presented by Spector at the American Society for Nutrition’s annual conference in June, extremely interesting. As we reported in the Times, the study found that genes play a limited role in how we process fats and carbohydrates. Among identical twins, only about half of the amount and duration of an individual’s post-meal blood glucose level could be attributed to genetic influence — and less than 30 percent for insulin and triglyceride response. The more important factors influencing how our bodies metabolize food seem to be environmental factors such as sleep, stress, exercise, and the diversity and population of our individual gut microbiome.

Meanwhile, the PREDICT study continues. Spector has recruited some of the “super-loggers” from the first phase to take part in a subsequent iteration, with a stronger focus on meal timing, and is currently recruiting participants for an even larger second phase. Would we advise signing up?

Well, Cynthia would certainly never do it again. Two weeks of food logging was more than enough. Nicky is typically more of a glutton for punishment, but the thought of having those muffins again is chilling. Still, in general, would we both consider being part of a study again to provide greater clarity and depth to our science reporting? Absolutely.

Nicola Twilley won a 2018 AAAS Kavli Science Journalism Award in the magazine category. Cynthia Graber won the award in 2004 in the radio category.

Associated homepage image: Dave Szalay for Gastropod.