Methods for dealing with fats of unknown type in dietary surveys: the Lipid Research Clinic Population Study

The design of the nutrition analysis tables adopted by the Lipid Research Clinics program of the National Heart Lung Blood Institute provides a specialized opportunity for assessing the fat component of 1-day dietary recalls. When the fat added during preparation of foods was unknown, rules were adopted for imputing composition. In order to investigate the effect that unknown fats produce on the apparent fatty acid composition of the diet, a detailed examination was made of 923 one-day dietary recalls from three clinics within the program. Total fat per recall from this sample averaged 107.6 g as calculated from visit 2 information. The average amount of unknown fat per recall at these three clinics was about 15.5%. About half the unknown fat at one clinic (Toronto-McMaster) could be identified and characterized through a procedure elected under Lipid Research Clinics protocol, "postinterview enquiry." Using this information, hypothetical "composite values," based on weighted means derived from the relative frequency of known use of fats and oils, were substituted for the unknown fat designations. Comparison of these composite values with values for known fats supports this approach and suggests the recommendation that postinterview information be collected for each region at the time of the inception of a new nutrition trial.