Dietary methods in cardiovascular disease critique.

It is planned that a wide variety of questions related to nutrient intake and cardiovascular disease will be addressed in NHANES III. Given the scope of these questions and the number of different types of designs that may be utilized--cross-sectional, cohort, and case control--several different dietary methodologies will probably be needed within NHANES III. Assessment and ranking of current intake is a primary objective to the overall NHANES III design. The 24-hour recall will provide this assessment and maintain comparability with the previous extensive NHANES. In many instances, however, information on usual intake or past intake will be necessary to better elucidate a nutrient-disease relationship, either because of temporal exposure concerns or because of the variability of the nutrient intake. A food history or quantitative frequency method for the specific nutrient(s) would best address these needs. Information on specific foods or other diet constituents (such as fish, dairy foods, alcohol, or water intake) in addition to nutrient levels will be of interest not only for cardiovascular disease but, with some food items, for cancer and osteoporosis as well. A history or quantitative frequency method would be appropriate for this purpose. The question of serial and multiple measures needs to be addressed. For adequate classification of individuals, several nonconsecutive measures would be most appropriate. The actual number of measures would be determined by the estimate of the most acceptable reduction in variance that could be achieved with multiple measures, balanced by the feasibility and cost of such measures. The use of multiple measures would necessitate the use of alternative interview techniques, most likely by telephone. It is clear that, if this type of methodology is adopted, further validation and extensive pretesting would need to be conducted. A subsample approach could also be utilized. Automated coding of dietary data and computer-prompted immediate data entry interview techniques may facilitate the use of these dietary data collection methods. Whichever methods are utilized, standardization of the coding procedures and additions to the nutrient data base, particularly for sodium, are essential. Finally, there is considerable interest in evaluating any deleterious effects of dietary modification on health and mortality outcome. For example, there is concern that a fat-modified diet may increase cancer mortality in some or that a low-sodium diet may be unpalatable and lead to a lowered dietary intake and potential nutrient deficiencies. It would be useful in this regard for NHANES III to be able to ascertain if altered dietary levels were due to self-selected dietary practices, to prescription of a therapeutic diet, or to a disease state or illness that lowered overall food intake.