Here's to Your Health

The popular media is replete with claims of the effectiveness of various diets on weight, cholesterol, or other outcomes (e.g., Pritikin, Atkins, Ornish). Some of these authors advocate a high-fat, lowcarbohydrate diet, blaming increased carbohydrate consumption for the increasing rates of obesity and diabetes. Others advocate exactly the opposite: a low-fat, high-carbohydrate diet. The evidence for many dietary claims is based on epidemiological studies that may be confounded by other factors. For example, vegetarians are likely to differ from nonvegetarians in ways other than just diet; they may be more likely to exercise or take vitamins. Consequently, it may be difficult to adequately adjust for possible confounding in observational studies. How can we reliably test dietary claims? The clinical trial is considered the gold standard of medical evidence because it eliminates many of the biases possible with other types of studies. Random assignment ensures, to the extent possible, the comparability of participants across diets on other factors that might influence the outcome. This column reviews some issues arising from clinical trials comparing diets. Surrogate Outcomes