Comparison of nutrient calculation systems.

A set of 54 24-hour dietary recalls collected in 1975-1976 from males aged 35-57 years who were participating in a cardiovascular risk factor intervention program was submitted to three different nutrient calculation systems to investigate how much of a difference exists among systems in calculating nutrient intakes. The three computerized systems were of varying levels of sophistication. Among differences found, one system reported 1.4% more calories derived from polyunsaturated fat than the other two. For studies investigating the effects of dietary fat intake, this difference between systems may be important. Other significant group differences were seen for carbohydrate and alcohol. Although mean differences among the three systems were not great, dramatic differences were encountered when evaluating individual recalls. Nutrient intake data obtained from dietary recalls for individuals and for groups for whatever purpose are subject to the bias of the nutrient calculation system used. These biases should be considered when interpreting results, comparing results with other studies, and when developing treatment plans in the clinical setting. Recommendations for enhanced standardization include: 1) thorough descriptions in research reports of the particular system used; 2) exchange of standard menus between systems; 3) enhanced quality control of the coding process; 4) periodic updating of the nutrient data base to accommodate new food products and changes in composition of foods.