What to do if Your Study Contradicts Conventional Wisdom

I just read a recent paper from the British Journal of Sports Medicine, "Daily Energy Expenditure and Cardiovascular Disease Risk in Masai, Ruran and Urban Bantu Tanzanians". The study caught my eye because I think we have a lot to learn from healthy traditionally-living populations worldwide.

The Masai have a very unique diet consisting almost exclusively of whole cow's milk, cow's blood and meat. As you might imagine, they eat a lot of fat, a lot of saturated fat and a modest amount of carbohydrate (from lactose). They also have low total cholesterol, low blood pressure, and virtually no overweight. They have been a thorn in the side of the lipid hypothesis for a long time.

The Bantu are an agricultural population that traditionally eat a diet low in fat and high in carbohydrate. Their staples are root vegetables, corn, beans, fish and wild game. The paper also describes a group of urban Bantu, which eats a diet intermediate in fat and carbohydrate. Incidentally, the investigators describe it as a "high-fat diet", despite the fact that the percentage fat is about the same as what Americans and Europeans eat, shamelessly exposing their bias.

The investigators recorded the three groups' diets, activity levels, physical characteristics and various markers of cardiovascular disease risk. Here's what they found: only 3% of Masai were obese, compared to 12% of rural Bantu and 34% of urban Bantu (they'd fit right in here!). The Masai, despite smoking like chimneys, had generally lower CVD risk factors than the other two populations, with the urban Bantu being significantly worse off than the rural Bantu.

Overall, the Masai came out looking really good, with the rural Bantu not too far behind. The urban Bantu look almost as bad as Americans. How do we make sense of these two conflicting facts? 1) The urban Bantu eat an amount of fat and saturated fat that's right in the middle of what the Masai and the rural Bantu eat, yet they seem the most likely to keel over spontaneously. 2) Saturated fat KILLS!! Answer: keep digging until you find something else to blame your results on.

They certainly did find something, and it's the reason the study was published in the British Journal of Sports Medicine rather than the American Journal of Clinical Nutrition. The Masai exercise more than either of the other two groups. I don't have too much trouble believing that. However, the authors used a dirty trick to augment their result: they normalized calorie expenditure to body weight. They present their data as kcal/kg/day. In other words, the fatter you are, the lower your apparent energy expenditure! It makes no sense to me. But it does inflate the apparent exercise of the Masai, simply because of the fact that they're thinner than the other two groups.

Due to this unscrupulous number massaging, here's what they got (data re-plotted by me):


I'm going to try to un-massage the data. Here's what it looks like when I factor bodyweight out of the equation. Calories expended (above resting metabolic rate) is on the Y-axis. The bars look a bit closer together...



Here's what it looks like when you add back resting metabolic rate. I assumed 1500 kcal/day. This graph is an approximation of their total energy expenditure per day:



Hmm, the differences keep getting smaller, don't they? I'm not challenging the fact that the Masai exercise more than the other two groups, but I do have a problem with this kind of manipulation of the data in misleading ways.

Their conclusion is that exercise is protecting the Masai from the deadly saturated fats in their diet. A more parsimonious explanation is that saturated fat per se doesn't cause heart disease. It's also more consistent with other healthy cultures that ate high-fat diets like the Inuit, certain Australian aboriginal groups, and some American Indian groups. It's also consistent with the avalanche of recent trials of low-carbohydrate diets, in which people consistently see improvements in weight, blood pressure, and CVD markers, among other things. Not that I have much faith in blood lipid markers of CVD.

My conclusion, from this study and others, is that macronutrients don't determine how healthy a diet is. The specific foods that compose the diet do. The rural Masai are healthy on a high-fat diet, the rural Bantu are fairly healthy on a low-fat, high carbohydrate diet. Only the urban Bantu show a pattern really consistent with the "disease of civilization", despite a daily energy expenditure very similar to the rural Bantu. They're unhealthy because they eat too much processed food: processed vegetable oil, processed grain products, refined sugar.

Thanks to kevinzim for the CC photo