This is the third paper in what I would call a “curiously well done” study looking at the rates of death and cardiovascular disease based on diet. The first two papers established clear connections between more fat in the diet being healthier (Just Read: Validation of Low(er) Carb High(er) Fat Diets: The PURE Study and CVD/Overall Death ) and with better intermediate biomarkers ( Just Read: More Validation of Low(er) Carb High(er) Fat Diets: The PURE Study and Lipids ) which gives credence to the validity of food frequency questionnaires used.
Interspersed in the papers and commentary are references to older thinking about nutrition that seem out of place – it seems that even when the data says something new, people are challenged to undo something that’s been said before.
Like the others, the cohort of people studied is huge (135,355 people in 18 countries) and spans a large swath of humanity in terms of income, education, and diet, so their exclusion gives a much better sense of the effect of eating a healthier version of these foods.
Noting: Potatoes and tubers were excluded as vegetables. Fruit and vegetable juices were excluded as well. These foods represent very large carbohydrate loads if included in the diet.
Paying attention to the adjustments:
The primary model adjusted for age, sex, energy intake, current smoking status, urban or rural location, physical activity, baseline diabetes, education, and other dietary variables (white meat, red meat, bread, and cereal intake), and study centre (as a random effect). For the analyses of fruit intake, we adjusted for vegetable intake, and conversely, analyses of vegetable intake were adjusted for fruit intake. We adjusted for variations in socioeconomic status using education level and household income or wealth index; these produced similar results.
This is key because people with different health behaviors are likely to eat differently, which will confound things quite a bit.
The semi-curious results: As summarized in the attached commentary:
However, when Miller and colleagues adjusted for other potential dietary and non-dietary confounders, the associations were substantially attenuated. In the multivariable adjusted model, the risk of total and non-cardiovascular mortality was in the order of 22â€“23% lower for participants consuming at least three servings per day of fruits, vegetables, and legumes, compared with participants who consumed less than one serving per day of these foods. The associations remained significant for fruits and legumes but not for vegetables (emhpasis added by me).
How to put this together with the entire study:
- As I’ve mentioned previously here, it’s now impossible to interpret any study on nutrition without knowing who the specific people are doing the study, what their interests are and which side they are on things.
- What is possible now is to see the social media reaction, and from my review of blog posts and tweets about this study, I see mostly proclamations of the value of fruits and vegetables in the diet with some leaning toward promotion of the plant based diet.
- I think the latter assessment is not validated either way, since the results were adjusted for meat intake – in other words, this is not “vegetables good, meat bad.”
- I am surprised by the fact that after adjustment vegetable intake does not correlate with reduced cardiovascular disease or death, even with tubers removed
- Remembering the other papers coming from the same study, fat is still important in the diet and removing it results in more death.
I listened to the podcast interview of the authors and found it very helpful. Their assessment (I am paraphrasing, listen for yourself) is that the carbohydrate component of fruits, vegetables, legumes (minus tubers) is low enough that increased intake is still consistent with the other findings in the study.
The curiosity in this paper is this statement in one of the “whys” fruits and vegetables may contribute to less disease:
Fruits and vegetables are good sources of dietary fibre, which has been shown to reduce the insulin response to carbohydrates, and decrease total and LDL cholesterol.
Except in the Lipids paper, they said this
Focusing on a single lipid marker such as LDL cholesterol alone does not capture the net clinical effect of nutrients on cardiovascular risk. The current recommendation to reduce total fat and saturated fatty acids, which leads to a de facto increase in carbohydrate intake, is not supported by our data.
Regardless, I am impressed at how much information has come out of this study. I see they are continuing to enroll subjects and are nearing 200,000 people. I am sure this data will be discussed for a very, very, very long time.