Insights into Nutritional Epidemiology: An Interview with Walter Willett

Interview by Iris Yan

HHPR Editor Iris Yan interviewed Walter Willett, M.D., Dr. P.H., a professor of Epidemiology and Nutrition at the Harvard School of Public Health and professor of Medicine at Harvard Medical School. He is a widely cited author in nutritional epidemiology, having conducted some of the most comprehensive studies on diet and disease, published 1,700+ original research papers and reviews, and authored multiple books including the textbook "Nutritional Epidemiology" and the bestseller “Eat, Drink, and Be Healthy: The Harvard Medical School Guide to Healthy Eating.”

Iris Yan (IY): Thank you so much for taking the time to talk with me today! First of all, what inspired you to get into the nutritional epidemiology research that you are doing now?

Dr. Walter Willett (WW): I’ve always been interested in food. I grew up in an agricultural family with many generations of dairy farmers. I studied food science when I was an undergrad at Michigan State and then went on to study medicine and then did a doctoral degree at Harvard in epidemiology. So, it was somewhat natural for me to put these together under the umbrella of nutritional epidemiology, which tries to answer the basic questions of how what we eat affects our long-term health and well-being.

IY: That’s really interesting! Also, I noticed you do both qualitative survey-type research as well as more quantitative scientific findings. How do you feel that those two components play into your overall research?

WW: Well, I don’t really like to separate them. An essential part of our work has really been to develop methods to assess dietary intake of individuals, because if you can’t measure your exposure, then you can’t study long-term effects. That had been the core of why nutritional epidemiology had not been developed when I started looking at this in the 1970s. At the core of our research has been dietary assessment using mostly self-reported information, but carefully looking at the sources of error and validity also. Along with it, we have been using biomarkers of diet and metabolism, and recently genetics and other omics technology.

We really get the best understanding when we use multiple tools. At the core of it, still our best information does come from our self-reported dietary assessment measurements. We knew from the beginning that they couldn’t be perfect, but people are able to report their own dietary information somewhat surprisingly well. Contrary to what a lot of people believed when we got started, not everyone eats the same. There are very large differences in individual food choices that are influenced by many different factors. Because there are these differences in what people eat, we can learn about the consequences of food choices.

IY: Can you talk a little bit more about what you mean by controlling for sources of error when people are self-reporting their diets?

WW: First of all, at the beginning of this work, most epidemiologic studies in nutrition were case-control studies. We suspected early on, and this was supported by data, that people do report their diets differently because they have a disease. Instead, we use prospective studies where we collect dietary disease while people are healthy and then we follow them to see who develops a disease. Prospective design has taken away much of the bias that we had in retrospective studies. We also know that nothing can be reported with perfect accuracy, so much of what we’ve done has been developing statistical methods to correct for measurement errors. Also, even if you could measure diet perfectly at one point in time, there’s enough change in food supply and individual choices over time that we really need to have repeated measurements when we are doing studies that last for decades. Our studies have been unique in having many repeated measurements over a period of decades.

IY: Going off of that, you mentioned that what people eat changes a lot over time. What are some of the biggest factors that you think - in addition to food supply - such as advertising or food trends, that have really big influences on what people are eating?

WW: I would like to believe that people change their diet because we have good scientific evidence, but of course life is much more complicated than that. But still, the basis of that needs to be solid scientific information and education. Sometimes it takes decades for scientific knowledge to be translated into actual changes in diet. For trans-fat intake, there was a huge pushback even within the scientific community at first about its importance, but as our findings were replicated, it became accepted knowledge that high trans-fat intake increases risk of heart disease and many other conditions. This was quickly conveyed by different types of media, and for part of the population, that was enough. But then we got trans-fat labelling applied to foods, and some people understood that that was something to be concerned about & reduced their consumption. More importantly, that was also an incentive for manufacturers to reformulate products to be trans free or low trans-fat. Then the FDA officially banned it. It took a whole series of events over about 30 years. That’s a good story.

The other story is that vast amounts of money, many billions of dollars a year, are spent promoting unhealth foods and beverages. Coca Cola and Pepsi spend vast amounts of money getting people to drink their beverages. There are some big counterforces that do influence people in the wrong direction, and in some ways, they have the advantage in terms of resources because their profits are so huge. The cost of ingredients in a bottle is about 2-3 cents, and the rest of that is profit. There is thus a huge incentive to keep people drinking their beverages, and they have lots of resources for promoting them.

IY: You mentioned the FDA and product labelling – Are there other policies you feel like should be put in place to address other findings from your research or promote health eating in general? Either from the scientific community or from the government to encourage people along that way?

WW: There are many actions & policies at different levels that can be taken to promote health eating. You can have policies even within a family – they don’t have to be just national or international. Combinations of taxes or incentives can influence food choices very extensively. That can happen within schools, worksites, or cities also. The powerful parts of the food industry are not so powerful at city and local levels, so there is a lot of possible action there even while we’re gridlocked at the national level. For instance, under an Obama administration program, school food quality went from 58 to 81 (out of 100) in only 4 years. It shows that you can have major changes in a short period of time. Institutions also procure large amounts of foods, and if food services have policies that don’t include unhealthy foods and make healthy foods affordable, that can have important consequences. You can almost look at any policies where there are possibilities to make healthy food choices more available and affordable, and that does change behavior.

IY: We were talking earlier about how soft drink manufacturers have low production costs, and I think we might say that’s true for a lot of low quality and unhealthy foods, whereas for organic foods or produce, that is a lot more expensive. Do you think putting these sorts of tax incentives might help even that balance?

WW: Absolutely – putting incentives or subsidies on healthier choices can make a very big difference. We did a little study on our own food service at Harvard, where we subsidized the salad bar, and that did have an important impact, increasing salad bar sales within our own community. That works both ways, either with subsidies or taxes, where we can just shift the prices within a food service. It can be revenue neutral overall, but we can make the playing field, which is now tilted towards unhealthy choices, at least a level playing field and ideally tilted towards those healthier foods.

IY: Also, just for a quick follow up – how do you determine which diseases to look at? For example, how do you know we’ll look at adolescent consumption of some nutrient to study the occurrence of dementia later on?

WW: We choose our hypotheses and research focus based on several considerations. In our long-term studies, our priorities are chosen for us because we look at the conditions that accrue in greatest frequency, which means that they’re important questions. For example, in the Nurses’ Health Study, our original large study of diet and health in women, there’s no escaping that breast cancer is the number one issue in middle age women. It has become even more challenging than anticipated, since we thought that we’d have answers within 5 or 10 years, but as we’ve learned from other lines of epidemiology, many of the determining factors actually happen early in life. We enrolled another 116,000 women starting off earlier, collecting data from their mothers and even their mother’s pregnancy experiences, and we’re just starting to have enough follow up to look at some of those issues. So again, our priorities to a large extent are set by the frequency at which these health issues are occurring. Now, these participants are moving into later life, and neurodegenerative consequences are becoming more and more common, so we now have enough data to begin to study those. These data are reflective of what’s going on with the general population, so neurodegenerative diseases are becoming a higher research priority for us.

IY: For sure. In the long term, how do you see the field of food science and nutritional epidemiology changing in the coming years, and how has it been changing so far?

WW: It is changing at multiple levels. For example, the NIH is promoting personal nutrition, and I think there is a lot to be learned from integrating genetics, genomics, and metabolomics into nutrition research, but personalized nutrition for a few people is really just making tweaks around the margin. Really big impacts come from overall changes in food supply and cultural change. If we put all the emphasis on personalized nutrition, that can actually be dangerous because it undermines the most powerful tools we have in public health, which are cultural and policy changes that have a more profound impact on our health overall. Personalized research can be most revealing in terms of understanding mechanisms, but we also need to include bigger picture implementation that deals with education about nutrition and well-being and the influences of economic and policy factors the influence what we eat. Closely related to this are the huge and growing gaps in diet quality across socioeconomic groups that translate directly into rates of obesity, diabetes, cardiovascular disease and premature death: food justice must be addressed.

We have also done more research at the ends of the life cycle, for obvious reasons. Our research is focused on adults because that is where the disease is occurring, but we do see that the choices of foods during childhood and adolescence are for some cancers, such as breast cancer, are more important than what people are eating in midlife. However, it is more complex and sometimes takes a longer time when you are studying food choices during adolescence. Also, neurodegenerative conditions such as dementia and Parkinson’s are becoming more important, and our work looks at how we can prevent or reduce risk of those conditions via food choices.

Integrating human health effects with environmental effects is also very important. As a human society we are facing an existential threat of climate change at an accelerating rate, and we need to look at everything we do through an environmental lens. It turns out that our food choices have a big impact on greenhouse gas production, and we must take that into account in our research and our policy decisions.

IY: Definitely. Thank you so much for your time & for speaking with me! I’ve learned a lot.