Don't blame nutrition: blame epidemiology

Normally, my blogs are built around a recently published scientific paper of relevance to public health nutrition. This week’s blog is a rant. It is a rant against the tsunami of second-rate science that dominates the media and that confuses the consumer allowing all sorts of drivel to be peddled as nutritional science. The truth is that the great majority of these stories are based on nutritional epidemiology, an academic area that gathers vast amounts of data on people and uses statistical models to sort the wheat from the chaff. Their statistical models are bound by the “known knowns” but they necessarily exclude the “known unknowns” and the “unknown unknowns” and thank you, Donald Rumsfeld for this addition to our lexicon. For example, a journalist writing in the Irish times cited a paper, which was built on such epidemiological models relating sugar to cardiovascular disease and ended up by quoting the publication asserting that as the % calories from sugar in the diet rose, the risk of cardiovascular disease rose by some huge figure. The journalist argued that the higher sugar intakes “led” to a rise in cardiovascular disease. It didn’t because there was no intervention that would “lead” anyone anywhere. The statistical model implied so and that implication was limited by the constraints of the “known knowns” of the statistical wizardry. But there was no experiment in which subjects had their diet manipulated and then followed to look for changes in cardiovascular risk factors such as might be measured in blood or detected by imaging or simple monitoring blood pressure. No, this study, like all epidemiological studies does not provecause and effect. They are merely interestingobservations of association awaiting confirmation by the science of experimental nutrition.

No matter what the nutritional challenge is, there are normally, with some few exceptions, the means to experimentally test the observed associations of epidemiology. As the Nobel Laureate in Immunology, Sir Peter Medawar wrote: “If politics is the art of the possible, then science is the art of the soluble.”

This rant is written in the heart of the Chianti region of Tuscany where village shops abound in beautiful fresh fruit, vegetables and legumes, fresh fish, lean meat, nuts, breads and of course beautiful red wine. This is the Mediterranean diet and unless you’ve been a hermit here in Europe for the last 30 years, you will know that the Mediterranean diet is the most healthful on the planet. Or so the epidemiologists told us. But it took a consortium of Spanish scientists and their funders to put in place the critical test of the hypothesis that the Mediterranean diet was all it was cooked up to be. The Predimed[1] study, as it is called, had the following design as outlined in the abstract: “In a multicenter trial in Spain, we randomly assigned participants who were at high cardiovascular risk, but with no cardiovascular disease at enrollment, to one of three diets: a Mediterranean diet supplemented with extra-virgin olive oil, a Mediterranean diet supplemented with mixed nuts, or a control diet (advice to reduce dietary fat). Participants received quarterly individual and group educational sessions and, depending on group assignment, free provision of extra-virgin olive oil, mixed nuts, or small nonfood gifts. The primary end point was the rate of major cardiovascular events (myocardial infarction, stroke, or death from cardiovascular causes). On the basis of the results of an interim analysis, the trial was stopped after a median follow-up of 4.8 years”.

The outcome was clear. Both the intervention diets reduced cardiovascular disease by about 30%. Now we have experimental data to search, analyse and interpret, knowing that the data is the outcome of a dietary intervention study. Some may not like the outcome and some may criticise the design or query its policy implications. But in as far as is humanly possible, this consortium of Spanish scientists has adhered to the correct scientific rigour and has not stopped at epidemiological observations.

The media simply do not get this vitally important difference of an association between two factors and a proven effect of the causative factor (the Mediterranean diet) on the effectunder study (cardiovascular disease). Consider for example the major feature article, which recently appeared in Time magazine by Brian Walsh. The work of the epidemiologist Ancel Keys, who first presented evidence of a link between saturated fats and heart disease, is rubbished in this article. But Walsh never once refers to the follow up to that epidemiological observation. Both Keys and his colleague Pepe Grande subsequently conducted multiple dietary intervention studies in psychiatric hospitals in which the major part of protein, carbohydrate and fat were provided as milks with the subjects allowed a small selection of vegetables. The results were conclusive. As the % calories from saturated fats rose, blood cholesterol levels also rose while when polyunsaturated fats were increased, blood cholesterol levels fell. In fact the effects was so consistent that a set of predictive equations were established which can still operate today. Decades later, Martijn Katan at Wageningen University in the Netherlands, also concluded several dietary intervention studies which showed a role of olive oil type fats (monounsaturated) and an undesirable effect of trans fats. Once again, a rich stream of experimental data could inform policy. Indeed, Katan still operates a web-based tool for calculating how blood lipids change in response to changes in the composition of dietary fats (http://www.katancalculator.nl).

None of these dietary intervention studies appears in Walsh’s article. He relies solely on epidemiological studies. He argues that nutritionists were wrong to advocate a low fat diet. We didn’t. Going back as far as the first set of dietary guidelines issued in the mid 1970s by Senator George McGovern, nutritionists asked for a modest reduction in fat intake (from 40% to 35 % of energy) but with major reductions in saturated fat intakes (based on experimental evidence). To this day, that advice still persists within national and international bodies that are charged with issuing dietary advice. It was the media, the food industry and the “pop nutritionists” which sold an ever-decreasing fat intake as the holy grail of diet. We in nutrition knew back then that a low fat diet would raise blood triglycerides and re-shape that structure of the so-called bad cholesterol LDL, into a higher quantity of very small and dense LDL particles. We in nutrition also knew that studies on fasting subjects, which epidemiologists insisted on using, bore no relation to the manner in which the body metabolised fats after a meal, an area, which is today’s new “hot topic” in nutrition.

Public health nutrition has benefited greatly from the theories that nutritional epidemiology has generated. But public health nutrition that eschews experimental evidence where it exists, will ruin the public’s confidence in nutrition. So to those readers who influence policy in whatever way, might I ask that henceforth, any story you come across, which claims to link some nutritional pattern with some health outcome, should ask the following question: “Is the study based on a dietary intervention that allows cause and effect to be established?” If it isn’t, tread carefully until the evidence emerges from the true scientific tradition of experimentation