Dietary fat and obesity: an unconvincing relation

Abstract

Bray and Popkin (1) attempted to refute my earlier conclusion that diets high in fat do not appear to be the primary cause of excess body fat in our society and that a reduction in dietary fat is not the solution (2). The topic is important because replacement of dietary fat with carbohydrate has been the predominant nutritional advice in the past decade (3). In a sedentary population with widespread insulin resistance, this change in diet induces hyperinsulinemia, hypertriglyceridemia, and low HDLcholesterol concentrations (4, 5). Thus, if weight is not substantially reduced, rates of coronary artery disease may not be reduced and could even increase. Bray and Popkin criticize the epidemiologic data that I cited on time trends and on geographic comparisons relating dietary fat to body fat (2). Although these data provide some useful evidence, I concluded that the relation between dietary fat and body fat is particularly difficult to evaluate in nonexperimental studies and that the best evidence should come from long-term randomized trials. Bray and Popkin devote considerable attention to nonexperimental data, which they deem to be superior. Their alternative geographic analysis, which they characterize as more “representative” of world populations, compares poor developing countries such as India, Mali, and China with the United States and other affluent countries. The differences in wealth and lifestyle among these countries are so extreme that their analysis hardly constitutes evidence for a causal relation between dietary fat and body fat. However, some of the data for individual countries depicted in their Figure 1 are informative. The fact that nearly 60% of the South African population is overweight, with an intake of <22% of energy from fat, indicates that a massive obesity problem can occur even with fat intakes that are generally considered to be low. The same applies to Saudi Arabia. Bray and Popkin specifically criticize the ecologic study of 65 Chinese counties because there was little variation in body fat. However, these data are actually more informative because confounding by extreme variations in affluence and physical activity is less problematic among these counties than among countries worldwide. The lack of variation in body weight despite substantial variation in dietary fat intakes (range: 7‐22% of energy from fat) is just the point.