Imprecise methods may both obscure and aggravate a relation between fat and breast cancer

Abstract

Bias in diet reporting may be both random and non-random, and may be found with simpler as well as more advanced dietary instruments. A random bias will contribute to obscure relations between diet and disease. A systematic bias may obscure or aggravate such associations. Underreporting of non-protein energy has been found to be substantial, particularly among those who are obese or have high dietary intakes. Such a non-random bias on the group level would tend to aggravate associations between dietary non-protein and disease. Whether the net result of the random and non-random bias aggravates or obscures relations depends on the relative magnitude of the two.