Diet and Health: The Need for New and Reliable Approaches

Andrew Mente; Mahshid Dehghan; Salim Yusuf


Eur Heart J. 2020;41(28):2641-2644. 

In This Article

Why are Causal Connections in Nutrition Difficult to Establish?

Making causal connections between diet and CVD is enormously challenging (Figure 1). Diet is one of the most complex exposures and essential to health (unlike external exposures such as tobacco, occupational hazards, etc.). Given its complexities (and the likelihood that the effects of even large differences in diets may only be modest), it is hardly surprising that the results of different studies are inconsistent or inconclusive.

Figure 1.

Schematic diagram summarizing the use of diverse methods to provide robust answers on diet and health. The use of such complementary methods can be achieved through extensive collaborations and funding efforts from both academia and industry, with checks to minimize bias.

Because large and prolonged randomized trials of diet and clinical outcomes are extremely challenging to conduct, the field has mainly relied on epidemiological studies or short-term trials on the effects of diet on intermediate endpoints such as blood pressure or lipids. However, the effects on a limited number of physiological outcomes may not necessarily predict the net impact on clinical outcomes in populations.[9,20,21] In addition, in almost all large observational studies, diet is self-reported, which may contribute to both systematic and random measurement errors and can attenuate associations between diets and outcomes and lead to inconsistent findings between studies. Other problems include biases in analyses (e.g. the degree and approach to covariate adjustments may change the results of the same study), that apparently 'protective foods' are more expensive and so consumed more often by wealthier or more educated people, and publication biases. Further, during the review process of manuscripts, reviewers with strongly held views sometimes 'block' publication of results that they are not comfortable with, even if the methods are relatively sound.[22]

These factors mean that it may not be wise to depend entirely on results from observational studies or small randomized trials of intermediate markers to make policies related to diet and health.