Comparison of two scoring systems used to monitor diets in outpatient clinical trials

Dietary stability and compliance are crucial to the proper interpretation of the results of clinical trials evaluating the efficacy of lipid-lowering drugs since dietary variations can obscure the true effects of the drugs being tested. Documentation of compliance to National Cholesterol Education Program (NCEP) dietary guidelines can be difficult to obtain, however, especially since many diets may meet one or two but not all the criteria for a Step 1 or 2 classification. The purpose of this study was to compare the ability of two diet scoring systems (the Food Record Rating [FRR] and the Ratio of Ingested Saturated fat and Cholesterol to Calories [RISCC]) to classify these ambiguous diets.

Three-day diet diaries (n = 622) were obtained from patients participating in a multicenter, clinical trial testing the lipid-lowering effects of a fiber supplement. The FRR score of each diary was calculated; the diary was then computer analyzed for nutrient composition, and the RISCC score was calculated. Based upon the NCEP dietary criteria for total fat, saturated fat, and cholesterol each diet was classified as either Step 1 or Step 2. Diets exceeding Step 1 criteria were classified as typical American (Step 0). Diets not meeting all 3 criteria for any given Step were considered 'NCEP unclassifiable'. Using the FRR and RISCC scores of only the NCEP-classifiable diets, the optimal RISCC and FRR cutoff points to distinguish between Step 0 and 1 diets and Step 1 and 2 diets were determined.

Only 50% of the diets were NCEP-classifiable. Using these diets, a RISCC of 20 best distinguished a Step 0 from a Step 1 diet, and 13 segregated Step 1 from 2 diets. The FRR cutoff points were 14 and 8, respectively. Using these values, the RISCC was able correctly to classify 92-97% of the diets, whereas the FRR correctly classified only 73-80%. Variability of scores within each Step was twice as high for the FRR as for the RISCC. The FRR was more biased by total kilocalories than was the RISCC.

We conclude that the RISCC scoring system was more accurate and precise than the FRR system for diet classification, and was a superior tool for classifying the ambiguous diets. Since the RISCC also requires (and therefore provides) quantitative nutrient data and the FRR does not, the former is a better dietary monitoring tool for clinical trials.