A recent study attempted to quantify the association of ten dietary factors with deaths from cardiovascular disease and diabetes. Headlines about the study were misleading.
How likely is bacon to kill you?
I’ve been seeing differing headlines about a recent study:
- MDLinx said, “Bacon, soda and too few nuts tied to big portion of US deaths.”
- USNews said, “Poor Diet Tied to Half of U.S. Deaths from Heart Disease, Diabetes.”
- Reuters said, “Poor diet tied to nearly half of U.S. deaths from heart disease, stroke, diabetes.”
Since they said different things, I had to wonder which headline was the most accurate, and the whole thing just didn’t sound quite right to me. It fits too neatly with the alternative medicine myth that if you just eat right, you won’t get sick. I could just see people reading those headlines and concluding “If I just stop the bacon and soda and eat more nuts, I’ll be half as likely to die.” I don’t question that diet plays an important role in health, but that role is frequently exaggerated by people with a “natural medicine” bias. The CDC has estimated that nearly half of US deaths could be prevented with lifestyle changes; but that includes not only healthy diet but also exercise, avoiding overweight, and avoiding tobacco and alcohol.
I went to the source and read the study itself: “Association Between Dietary Factors and Mortality from Heart Disease, Stroke, and Type 2 Diabetes in the United States,” by Micha et al., published in JAMA on March 7, 2017. As it turned out, the Reuters headline was the most accurate.
Objective: To estimate associations of intake of 10 specific dietary factors with mortality due to heart disease, stroke, and type 2 diabetes (cardiometabolic mortality).
Methods: They used a comparative risk assessment model to evaluate previously published data. Data from the National Health and Nutrition Examination Survey (NHANES) was based on 24-hour dietary recalls. Data on disease–specific mortality was taken from the National Center for Health Statistics, which covers the entire US population. They developed models that tried to compensate for potential confounders.
They studied fruits, vegetables, nuts/seeds, whole grains, unprocessed red meats, processed meats, sugar-sweetened beverages, polyunsaturated fats, seafood omega-3 fats, and sodium. They chose not to study several other dietary factors (monounsaturated fats, vitamin D, magnesium, calcium, antioxidant vitamins, dairy products, cocoa, coffee, and tea) because of insufficient evidence for a causal relationship. They looked only at cardiometabolic conditions, not at other conditions like cancer, osteoporosis, gallstones, inflammatory diseases, depression, cognitive function, or micronutrient deficiencies.
- High sodium intake (> 2000 mg/d) was linked to 9.5% of cardiometabolic deaths
- Low nuts/seeds (<20.2 g/d) to 8.5%
- High processed meats (>0 g/d) to 8.2%
- Low seafood omega-3 fats (<250 mg/d) to 7.8%
- Low vegetables (<400 g/d) to 7.6%
- Low fruits (<300 g/d) to 7.5%
- High sugar-sweetened beverages (>0 g/d) to 7.4%
- Low whole grains (<125 g/d) to around 6% (not specified in text; I estimated from graph)
- Low polyunsaturated fats (<11% energy/d) to 2.3%
- High unprocessed red meats (>14.3 g/d) to 0.4%
- Taken together, non-optimal intakes of these 10 dietary factors were associated with 45.4% of all cardiometabolic deaths.
They did some subgroup analyses for individual diseases, age, sex, education, and race/ethnicity; and they looked for trends between 2002 and 2012. They found a larger diet-related proportional mortality for men than for women, for younger than older ages, for blacks and Hispanics than whites, and for those with low and medium education than for those with high education. Between 2002 and 2012 they found several improvements but these were not uniform.
In the Discussion section, they pointed out some of the strengths and weaknesses of their study and stressed that their model does not prove that changes in these dietary habits reduce disease risk and that model estimates do not make assumptions about the feasibility of society-wide interventions to improve diet.
In an accompanying editorial, Noel T. Mueller, PhD, MPH, and Lawrence J. Appel, MD, MPH pointed out that in attributing death to diet, precision counts. They say:
It is generally accepted that a sub- optimal diet is causally related to CMD, but scientists debate which factors are responsible and the relative importance of each factor given the challenges of isolating and estimating the potential effects of individual nutrients and foods, especially in observational studies.
They have three caveats:
- Evidence from observational studies is subject to pitfalls and confounders.
- We don’t know whether the 10 selected factors comprise the right set.
- Dietary factors are interrelated and modified by each other.
They say: “Despite these limitations and concerns, the study by Micha et al is quite relevant to public health nutrition policy, a critically important issue both in the United States and around the world.” Whatever the actual numbers, they undeniably showed that better diet quality (with reference to the 10 items studied) is associated with fewer cardiometabolic deaths. What they did not show was that you could take people with poor diet quality and reduce their death rate by improving the quality of their diet. It might seem intuitively obvious that that would work, but even the most compelling assumptions can be wrong. The history of science has taught us not to assume but to test. Hypotheses about interventions remain to be tested, and will require further studies.
Conclusion: nothing really new
The study’s findings are interesting and contribute to our understanding of relative risks, but there are really no big surprises here. We already knew a healthy diet was beneficial, not just for cardiometabolic disease but for cancer and other conditions; and we have a pretty good idea what constitutes a healthy diet. Evidence from various sources is converging on the consensus that fruits, vegetables, whole grains, nuts/seeds, and omega-3s from seafood are good for us and that sugar-sweetened beverages, processed meats, and salt are not.
Opinion: everything in moderation
I found it interesting that the weakest association was with unprocessed red meats. That confirms my opinion that small amounts of red meat (for instance, the occasional steak or hamburger) can be part of a healthy diet. For that matter, if you eat a lot of fruits, vegetables, nuts, whole grains and salmon, you can probably feel free to indulge in bacon on occasion without worrying that it will kill you. We have enough to worry about without denying ourselves the pleasures of the foods we love. The goal of living longer should be put into context with the goal of enjoying life.
This article was originally published in the Science-Based Medicine Blog.