As we have recently discussed diet and its effects on health, it seems reasonable to ask whether there is a diet that is demonstrably healthy. A recent investigation attempted to answer this question.
This study was aimed at developing a healthy diet score that is associated with health outcomes and is globally applicable. It used data from the Prospective Urban Rural Epidemiology (PURE) study and tried to replicate it in five independent studies on a total of 245 000 people from 80 countries.
A healthy diet score was developed on the basis of the data from 147 642 people from the general population, from 21 countries in the PURE study. The consistency of the associations of the score with events was examined in five large independent studies from 70 countries.
The healthy diet score was developed based on six foods each of which has been associated with a significantly lower risk of mortality [i.e. fruit, vegetables, nuts, legumes, fish, and dairy (mainly whole-fat); range of scores, 0–6]. The main outcome measures were all-cause mortality and major cardiovascular events [cardiovascular disease (CVD)].
During a median follow-up of 9.3 years in PURE, compared with a diet score of ≤1 point, a diet score of ≥5 points was associated with a lower risk of:
- mortality [hazard ratio (HR) 0.70; 95% confidence interval (CI) 0.63–0.77)],
- CVD (HR 0.82; 0.75–0.91),
- myocardial infarction (HR 0.86; 0.75–0.99),
- stroke (HR 0.81; 0.71–0.93).
In three independent studies with vascular patients, similar results were found, with a higher diet score being associated with lower mortality (HR 0.73; 0.66–0.81), CVD (HR 0.79; 0.72–0.87), myocardial infarction (HR 0.85; 0.71–0.99), and a non-statistically significant lower risk of stroke (HR 0.87; 0.73–1.03). Additionally, in two case-control studies, a higher diet score was associated with lower first myocardial infarction [odds ratio (OR) 0.72; 0.65–0.80] and stroke (OR 0.57; 0.50–0.65). A higher diet score was associated with a significantly lower risk of death or CVD in regions with lower than with higher gross national incomes (P for heterogeneity <0.0001). The PURE score showed slightly stronger associations with death or CVD than several other common diet scores (P < 0.001 for each comparison).
The authors concluded that consumption of a diet comprised of higher amounts of fruits, vegetables, nuts, legumes, and a moderate amount of fish and whole-fat dairy is associated with a lower risk of CVD and mortality in all world regions, but especially in countries with lower income where consumption of these natural foods is low. Similar associations were found with the inclusion of meat or whole grain consumption in the diet score (in the ranges common in the six studies that we included). Our findings indicate that the risks of deaths and vascular events in adults globally are higher with inadequate intake of protective foods.
The authors rightly stress that their analyses have a number of limitations:
First, diet (as in most large epidemiologic studies) was self-reported and variations in reporting might lead to random errors that could dilute real associations between diet scores and clinical outcomes. Therefore, the beneficial effects of a healthier diet may be larger than estimated.
Second, the researchers did not examine the role of individual types of fruits and vegetables as components in the diet score, since the power to detect associations of the different types of fruits and vegetables vs. CVD or mortality is low (i.e. given that the number of events per type of fruit and vegetable was relatively low). Recent evidence suggests that bioactive compounds and, in particular, polyphenols which are found in certain fruit or vegetables (e.g. berries, spinach, and beans) may be especially protective against CVD.
Third, in observational studies, the possibility of residual confounding from unquantified or imprecise measurement of covariates cannot be ruled out—especially given that the differences in risk of clinical events are modest (∼10%–20% relative differences). Ideally, large randomized trials would be needed to clarify the clinical impact on events of a policy of proposing a dietary pattern in populations.
Fourth, the use of the median intake of each food component as a cut-off in the scoring scheme for each diet may not reflect the full range of consumption or provide a meaningful indicator of consumption associated with the disease. However, the use of quintiles instead of medians within each study or within each region yielded the same results indicating the robustness of our findings.
Fifth, the level of intake to meet the cut-off threshold for each food group in the diet score may differ between countries. However, in sensitivity analyses where region-specific median cut-offs were used to classify participants on each component of the diet score, the results were similar to using the overall cohort median of each food component. Further, with unprocessed red meat and whole grains included or excluded from the diet score in these sensitivity analyses, the results were again similar.
Sixth, misclassification of exposures cannot be ruled out as repeat measures of diet were not available in all studies. However, the ORIGIN study, in which repeat diet assessments at 2 years were conducted, showed similar results based on the first vs. second diet assessments. This indicates that misclassification of dietary intake during follow-up was not undermining the findings.
Seventh, one unique aspect of the study is the focus on only protective foods, i.e. a dietary pattern score that highlights what is missing from the food supply, especially in poorer world regions, but this does not negate the importance of limiting the consumption of harmful foods such as highly processed foods. While the PURE diet score had significantly stronger associations with events than other diet scores, the HRs were only slightly larger for PURE than for most other diet scores. However, the Planetary score was the least predictive of events. The analyses provide empirical evidence that all diet scores (other than the Planetary diet score) are of value to predicting death or CVD globally and in all regions of the world.
So, what should we, according to these findings, be looking for and how much of it should we consume? Here is the table that should answer these questions:
|Fruits and vegetables||4 to 5 servings daily||1 medium apple, banana, pear; 1 cup leafy vegs; 1/2 cup other vegs|
|Legumes||3 to 4 servings weekly||1/2 cup beans or lentils|
|Nuts||7 servings weekly||1 oz., tree nuts or peanuts|
|Fish||2 to 3 servings weekly||3 oz. cooked (pack of cards size)|
|Dairy||14 servings weekly||1 cup milk or yogurt; 1 ½ oz cheese|
|Whole grainsc||Moderate amounts (e.g. 1 serving daily) can be part of a healthy diet||1 slice (40 g) bread; ½ medium (40 g) flatbread; ½ cup (75–120 g) cooked rice, barley, buckwheat, semolina, polenta, bulgur, or quinoa|
|Unprocessed meatsc||Moderate amounts (e.g. 1 serving daily) can be part of a healthy diet||3 oz. cooked red meat or poultry|