Future directions in the development of food-frequency questionnaires.

The clear documentation during the last decade that dietary intake can be measured by standardized food-frequency questionnaires has been fundamentally important for understanding the relation between dietary composition and health. Energy-adjusted correlation coefficients comparing nutrients assessed in this manner with those measured by dietary records or recalls have generally been in the range of 0.4-0.7 for a wide variety of populations and with use of questionnaires that have varied greatly in length and detail. The latter finding, however, suggests that major increments in validity are not likely to be obtained by further refinements, although even modest gains are worthy of continued efforts. Additional effort must be given to developing and evaluating questionnaires that extend the range of ages, populations, settings, and dietary factors that are studied. Comparisons with biochemical indicators of intake should be used when possible. Even well-established methods of assessment must be continually modified because of the rapidly changing patterns of food consumption and dietary composition within contemporary populations.