We’ve all seen the back and forth in the news about diet studies, even within the last few months.

No need to cut down on red and processed meat for health reasons, headlines screamed at us back in September. Not so fast! others roared back, saying red meat is OK to eat is bad science. Soon, an uproar erupted over the red meat study, with experts claiming “egregious abuse of evidence” and reporters pointing to the researcher’s past ties to the meat and food industry.

After a week or two, the hubbub died down, only to be replaced by a slew of fresh headlines touting a new study that’s just shown — surprise! — red meat and highly processed foods are linked to increased cancer risk.

It’s enough to give a person whiplash.

What is with all the dithering in diet studies? Are scientists just fickle? Is nutrition research that inexact? The pendulum swings are frequent and frustrating and they create uncertainty with the public, not just about whether bacon or butter or beer is good or bad for your health but whether nutritional research itself is even reliable.

All the back and forth is frustrating for scientists, too. Ross Prentice, PhD, of Fred Hutchinson Cancer Research Center just penned an editorial about the challenges of nutritional research for the Annals of Internal Medicine.

“It’s not just within the context of red meat,” Prentice said. “The same thing happens over and over. We get dietary recommendations put together by expert committees and the data are reviewed. But when subsequent, so-called systematic reviews of specific recommendations take place, the data don’t meet reliability standards.”

Why is it so hard to pinpoint what’s good, bad and ugly when it comes to food and our health? We sat down with the longtime public health researcher to discuss the challenges and the changes he believes are necessary to add a bit more rigor, and a bit less roller coaster, to dietary research.

Why is there so much back and forth with regard to whether eggs or fat or red meat or processed meat are good or bad for us?

Diet and nutrition are tough areas to study and I have substantial respect for scientists working in this arena. The human diet over a person’s life span involves a complex mixture of nutrients, foods, dietary practices and preparations that influence the milieu surrounding the cells throughout the body. The problem is we don’t have objective ways of assessing what people eat, even in the short term.

The standard approach for five decades has been ask people what they eat — usually in the form of a food-frequency questionnaire with a list of about 100 food items. Study participants check how often they eat the foods and the portion sizes for each of these foods, then those are added up and compared to a nutrient database that breaks down particular foods or dishes according to the nutrient content.

Those estimates of either nutrient intake or specific food intake from self-reported data are then associated with disease incidence in study cohorts. Almost all the literature is based on observational studies of this type — not many are randomized controlled trials [the gold standard of clinical research] — because RCTs of diet and chronic disease outcomes are expensive, long-term and logistically complicated.

So a convincing body of evidence is difficult to achieve?

Yes, available information is mostly based on studies of association rather than causation, using methods that fall short of proving chronic disease effects, especially in view of the crucial dietary measurement issues. The whole gestalt produces reports that seem very uncertain in terms of the standards that are applied elsewhere in the scientific community for reliable evidence.

The majority of the back and forth relates to different opinions about the reliability of the self-reported dietary information. Critics doing systematic reviews attach very light weight to those data sources. [Read more on the red meat study’s research approach here.] But many of the people who develop dietary guidelines defend the approach — and continue to do more of the same types of studies.

We have quite a history of using these nutritional epidemiology methods and not getting clear results, or sometimes getting contradictory results. Experts are brought together and they end up saying we don’t know enough to make causation claims. There aren’t that many diet and chronic disease effects that can be viewed as well established. It’s kind of a mess, actually.

Why is self-reported data so unreliable? Do people just have bad memories or lie about how much they eat?

We can’t tell whether or not people are deliberately misreporting. We are a body image-conscious society and people might tend to not report as much intake of high energy, fatty foods and desserts, without even being consciously aware of it.

But it’s not a matter of finding who to blame. The research issue concerns the whole set of tools that are used for dietary assessment. We can’t just rely on what’s been done; there are well-documented problems with current approaches. It’s time to try additional approaches, along with what’s already being done.

What would you do differently?

We sometimes have the ability to assess short-term dietary intake reliably through the use of objective measures in urine or blood or other body fluids. Established biological measures, or biomarkers, can assess the intake of total proteins, mostly expressed as nitrogen in the urine. Similarly, sodium and potassium can be recovered in the urine, although it might take a few days of urine collection for precise estimation. There is also an excellent, but rather expensive, biomarker of the short-term consumption of total energy (calories).

Some studies use self-reported diet with an objective measure that correlates with the dietary factor under evaluation but does not actually assess intake. These measures can be useful, but they usually do not provide a reliable assessment of actual intake.

Intake biomarkers would be a more accurate way to determine what a person really ate so you could more accurately measure dietary impact on their health?

Biomarkers, if they’re actually reflecting intake, could be used for reliable disease association studies by themselves, if you have stored specimens for deriving these measures.

Johanna Lampe, PhD, RD, Marian Neuhouser, PhD, RD, other Hutch colleagues and I published a paper last year that did just that for some vitamin A and vitamin E-related micronutrients. We conducted a feeding study that used intake biomarkers, derived from blood micronutrient concentrations and study participant characteristics obtained routinely in a subset of more than 5,000 women in the Women’s Health Initiative study to accurately estimate short-term micronutrient consumption.

We then examined these biological assessments of intake to subsequent chronic disease risk.

The micronutrients we studied in the research, published in the American Journal of Clinical Nutrition, included alpha-carotene [found in yellow-orange and dark-green veggies like carrots, sweet potatoes, squash, broccoli, spinach, green beans and collards]; beta-carotene [found in yellow-orange fruits like cantaloupe, mangoes, papayas and dark-green veggies like kale and spinach]; carotenoids such as lutein plus zeaxanthin, or L+Z, [plant chemicals found in bright red, yellow and orange fruits and veggies like squash, carrots, grapefruit, oranges and apricots] and alpha-tocopherol or vitamin E [commonly found in seeds, nuts, leafy green veggies and vegetable oils].

What did you find?

We found that somewhat lower risks of specific cardiovascular outcomes, breast cancer and diabetes were associated with a higher intake of alpha- and beta-carotene; that a lower risk of diabetes was associated with higher L+Z intake; and that elevated risks of certain cardiovascular outcomes were associated with a higher intake of alpha-tocopherol, or vitamin E. Additionally, we showed that suitable biomarkers can be calculated from blood specimens obtained in large cohorts and applied directly in disease-association analyses. 

That’s pretty cool. Are you doing more of this biomarker-driven nutritional research?

Our research group is also studying small molecules that circulate in body fluids, seeking novel dietary markers. We’re working in collaboration with Dan Raftery, PhD, a metabolomics expert from the Hutch and the University of Washington Medical School.

And we’re strongly advocating for a much larger research enterprise to identify intake biomarkers for other components of diet. We still have a very short list, but this approach could be a big part of the answer.

So it’s not that our nutritional guidelines are wrong. We absolutely know that eating lots of processed foods and sugar is unhealthy. It’s just the data we have leaves too much “wiggle room.” Do we need to do more? Ban junk food from advertising on TV like they did with cigarettes in the ’60s?

There’s a lot of improvement needed in the American diet. Poor quality diets are likely a big part of the reason we have such high risks for various chronic diseases. Yes, there might be a need to ban bad food ads.

Even though we may need to continue to depend substantially on observational studies to answer the many diet and chronic disease questions of importance, we can learn from research on cigarette smoking and its impact on the risk of lung cancer and various other chronic diseases, which also relied on observational data.

But cigarette smoking epidemiology is much simpler since there are many nonsmokers, never-smokers and smokers. With diet, you don’t have never-eaters. Everybody has to consume calories. Furthermore, the human diet is such a complex mixture of nutrients, foods and practices that it becomes a difficult challenge to identify what might be bad and what might be beneficial. A lot of the notions about what’s good and bad in the dietary guidelines are reasonable, but it just can’t be said to be based on a convincing body of evidence.

In summary, there’s a great need for the development of additional intake biomarkers, perhaps using metabolomics [the study of small molecules, metabolites, and how they interact with cells and tissues], microbiomics [the study of microorganisms found in, say, the human gut or mouth] or other high-dimensional platforms. Doing so may be able to help us avoid another 50 years of uncertainty concerning the impact of diet on the risk of various cancers and other chronic diseases.

This article was originally published on January 28, 2020, by Hutch News. It is republished with permission.