Nutritional epidemiology ~ Faith or facts

In the course of a recent lecture on the subject of obesity, an issue arose in the discussion about an incremental approach to solutions for obesity. Thus individual action plans might seem very modest but a suite of such action plans could be successful in treating obesity. Allied to this viewpoint about incremental solutions came the old chestnut that sometimes, in epidemiology, we need to take action even if we don’t have definitive proof of the efficacy of that action. This chestnut, hankers back to the founder of modern day epidemiology, John Snow, who disconnected the tap from the water pump in Broad Street London, thereby ending a cholera outbreak. It is argued that Snow did not have definitive proof and this is the root of the view that sometimes, epidemiology needs to take a leap of faith. However, as Snow himself points out, of the 71 deaths he investigated, 61 were of people living in the vicinity of Broad Street and thus users of the pump. Three deaths occurred in children who lived nearer another pump but who went to school close to the Broad Street pump. Five others “preferred” the water at Broad street over their local pump. Only two deaths could not be linked to the guilty pump. All of this happened prior to our understanding of pathogenic microorganisms so Snow clearly couldn’t have used culture techniques to verify the organisms presence. He could have asked persons living away from the pump in question to drink the water in question but even in Snow’s time, the unwritten ethical position was: “First do no harm”. And so, armed with a dot map and his 71 case histories, the pump was disabled. This may have been a leap of faith in that definitive proof of cause and effect wasn’t to hand, but the quality of Snow’s data was excellent and compelling.

So in the field of human nutrition, how much of our public health nutrition policy is based on fact and how much on faith. In the early 1950’s. Ancel Keys studies the relationship between dietary fats, plasma cholesterol and the rate of heart disease in seven countries across the globe. Keys and his colleague Paco Grande then completed a series of human intervention studies in which patients in a psychiatric hospital were fen on reconstituted milk with a wide variety of fats and oils. The net outcome was definitive proof that the effects of dietary fat on plasma cholesterol could be accurately predicted using equations derived in their studies. But was there proof that lowering cholesterol would reduce the risk of heart disease. Endless studies ensued and all showed that high levels of plasma cholesterol were , at a population level, predictive of a higher risk of heart disease. Thus the dietary lipid hypothesis was upheld and entered the policy arena of public health nutrition.

A contrasting story is that of antioxidant micronutrients, particularly vitamins C and E. In the early 1980s, there was a widespread belief that plasma antioxidant levels played a major role in cardiovascular disease, in cancer and in ageing. The data was dominated by associations studies linking published levels of plasma antioxidant status in different countries with national disease rates. The relationships were most impressive. Animal studies also added to the theory and in vitro studies abounded showing how anti-oxidant vitamins could protect fractions such as low-density lipoprotein from oxidative damage which would otherwise render them very atherogenic. As often happens in the field of health research, someone wanted to cut to the chase and head for glory with a human intervention study. And so the ATBC (alpha tocopherol {vitamin E} beta carotene) study was designed and implemented. It failed to uphold the hypothesis and many reasons were put forward as to why the study was “unsuccessful”. Based on knowledge of these flaws, more intervention studies were rushed along and, all in all, the antioxidant theory was abandoned.

We can look at some other “successes” and “failures” in nutritional epidemiology. The protective role of folic acid in reducing the risk of a neural tube defect birth was shown in a randomized controlled trial leading to a major initiative in public health nutrition with the fortification of flour with folic acid. Trans fatty acids were removed from the food chain wherever possible on the basis of a strong human intervention study. In contrast, notwithstanding the strong evidence from correlational studies in humans of a link between fish oil fatty acids and cognitive decline, endless intervention studies have failed to show a protective effect.

Thus we have an excellent track record in human nutrition in translating observational studies that show an association between some aspect of diet and some health attribute into dietary intervention studies and then basing our policy interventions on food of those intervention studies.

Taxing sugar-sweetened beverages to reduce the incidence of obesity requires data from human intervention studies that show a direct link between weight gain and consumption of such beverages at rates that correspond to reality before anything is done. There are intervention studies and meta-analyses of such studies and the evidence is very weak if non-existent. To suggest that we trust these data so much and that we are fully confident that fiscal measures will be always positive and rarely negative is simply wrong and is bad science. To argue that sometimes we need to take leaps of faith in nutrition policy flies in the face of 6 decades of rigorous research on which nutrition policy has been built. Facts and not faith should drive policy. Facts are universal but faith is a subjective value.