Howard D. Sesso, ScD, MPH, researcher in the landmark Physicians’ Health Study I and II and the Women’s Health Study, discussed the need to improve clinical trial designs at the Council for Responsible Nutrition’s Science in Session conference.
When a scientist involved in landmark nutrition studies says there’s a problem with clinical trials, you sit up and you listen.
The trouble has to do with how we factor in sex differences when designing clinical trials and interpreting their results, said Howard D. Sesso, ScD, MPH. Sesso speaks from experience. As the associate director of the Division of Preventive Medicine at Brigham and Women’s Hospital, and associate professor of medicine at Harvard Medical School, he’s been part of such pivotal nutrition studies as the Physicians’ Health Study I and II and the Women’s Health Study. Sesso shared his thoughts about clinical trial design on October 19 during the Council for Responsible Nutrition’s (CRN; Washington, DC) Science in Session conference.
“When we think about dietary supplements and a lot of the large-scale and smaller-scale clinical trials that have been done,” said Sesso, “the notion of how sex factors into those results, I would actually say, has been lost, unfortunately.”
First, there’s the problem of restricting studies to a single sex population, Sesso said. While studies like the Physicians’ Health Study and Women’s Health Study—which still stand as well-designed, well-conducted clinical trials—focused on one sex only, you lose the chance to study the same outcome in the other sex. Why does this matter? “By limiting the results to predominantly men, or just men, or just women, we lose the value of subgroup analysis,” Sesso explained.
Subgroup analysis matters because it gives us a window into how the intervention performed in other populations. And it extends beyond gender, Sesso said. The same goes for differences in age, race and ethnicity, nutrition status, diet, body weight, and many other factors that help us richly contextualize results. By not diversifying and building adequate subgroups into a study, “the problem is you wind up with results that are pertaining to one group but not the other, or [just] a certain age range,” he said.
There’s also the practical drawbacks of limiting a study—drawbacks involving wasted time and money that might get decision-makers’ attention. Say a study done in men shows promising results that researchers realize should also then be explored in females. Researchers have to initiate an entirely new study, which they could have avoided by including both sexes from the get-go. Sesso said this happened in the case of the Physicians’ Health Study, which was conducted in men but then ultimately extended to women. This was “not the most efficient way to do it,” he said. “It should have really been done all at once in some capacity.”
Then there’s the problem of not digging deeply enough into data to show how study results applied to different subgroups, such as how interventions might have impacted men and women differently. This is especially the case with meta-analyses, Sesso said. “Where is the separation by men versus women, by age, things like that? Those are things that are just not emphasized enough in these meta-analyses.”
He lamented: “We focus on the overall picture and not the specifics.” And these specifics matter. They matter in terms of how we understand and apply study results. They matter in how we make public health recommendations for the broad population.
They also matter when comparing results of different clinical trials in order to make broadscale conclusions—and, indeed, to move nutrition science forward. If more studies were designed more similarly and to include more subgroups, their results could be more easily stacked against each other. “Whether we’re looking at large-scale trials or small-scale trials, they need to complement each other in much more direct, functional ways,” Sesso said. “There’s just too much heterogeneity…across these trials over the course of time that have made it difficult to put together a really sound recommendation.”
“In the past, what we’ve done with our trials is we’ve been too limiting, frankly. We’ve focused on intervention and clinical outcomes and didn’t build in the mechanisms simultaneously so we could roll a number of different trials together,” he concluded.
The good news is that we’re learning to avoid these obstacles. As examples, Sesso pointed to VITAL (the VITamin D and OmegA-3 TriaL) and COSMOS (the Cocoa Supplement and Multivitamin Outcomes Study). Here, researchers purposefully embraced more specificity and more subgroups. “Our more recent trials that we’ve done have included both men and women so we can actually look directly at effect modification by sex, by age, and other important components,” he said, “so that when you do a trial, you can actually test the same intervention in a wider swath of people.”
Sesso’s hope is that we can “start to flip the script a bit” and factor in these considerations when designing studies. “We’ve been doing it backward,” he said. By designing studies more holistically, we may find promising results in some, but not all, of the subjects—and then we can follow up with those likely to see benefits. But if we’re not looking at those subjects in the first place? It’s likely we’re missing a valuable piece of the puzzle.