ABOVE: ©ISTOCK.COM, fcafotodigital

Dietary and nutritional advice is notoriously fraught with misinformation and unreliable findings. The Institute for Health Metrics and Evaluation (IHME) attempts to offer clarity with a new tool, described in a set of papers published today (October 10) in Nature Medicine. The approach employs a 5-star rating system to categorize how much evidence exists for a given risk factor-health outcome correlation. The higher the star rating, the more scientific evidence exists that there is a correlation between a risk factor (smoking, for example) and a health outcome (heart disease).

The set of papers includes a methodology paper and four “proof-of-concept” studies in which scientists tested the tool’s validity on the effects of smoking, blood pressure, consuming unprocessed meat, and consuming vegetables. In the four studies, IHME scientists found strong correlations between consumption of non-starchy vegetables and better health: The vegetables correlated to decreased risk of ischemic stroke (3 stars) and to a lesser extent decreased risk of heart disease (2 stars). They also found 5-star evidence that smoking can increase the risk of diseases such as lung and laryngeal cancer, and that high blood pressure increases risk of ischemic heart disease. However, the correlation between red meat consumption and health issues such as colon and rectum cancer, heart disease, and diabetes were all 2 stars or weaker. “I was very surprised at how many of the diet-risk relationships were much weaker” than expected, study author and IHME founder Christopher Murray tells STAT.

Some experts express doubt regarding the model’s utility. “Although these analyses are interesting, they perhaps only confirm that it is hard to accurately measure diet,” Aston Medical School dietician Duane Mellor says in comments provided to the Science Media Centre, an independent, nonprofit science press office based in the UK. The IHME model does not incorporate individual study biases and variations in the method of data collection and cannot account for the complex interactions between multiple foods in a person’s diet, Mellor continues, citing the example that while smoking is more of a binary (you either smoke or you don’t), a person might swap one food for a very similar food, changing their health outcome as compared to another individual who omits a group of food entirely. Reducing these subtleties down to a single number might not make for entirely accurate results, he says.

That doesn’t mean the tool won’t be useful, Kevin McConway, a statistician at The Open University in the UK, tells the Science Media Centre. “If one does want to reduce a set of studies of association between a potentially risky exposure and a health outcome to one number, or a few numbers, this new method seems to be a plausible and generally appealing way to do that,” he says.