[Evaluation of an original tool for carbohydrate counting, aimed at facilitating the implementation of functional insulin therapy].

Objective: Carbohydrate counting is the most difficult component of functional insulin therapy. We thus designed a tool to facilitate carbohydrate counting of meals.
Research Design and Methods: The tool consists of an illustrated directory (16 x 10 cm, 119 pages) presenting 389 pictures of food, classified in 12 categories (breads, fruits, vegetables, etc.). For each photo, the name and mode of preparation of the foods are given, with the weight of the illustrated portion and its carbohydrate content as a multiple of 5 g. During the first phase of the study (3 days a week for 12 weeks), twelve patients with type 1 diabetes completed a precise food diary using a list and weight of all consumed foods. We were thus able to determine, for each of three meals (breakfast, lunch and dinner), the variability of their carbohydrate content. During the second phase of the study (2 weeks), the patients were given the possibility of using the illustrated food directory. We asked them first to estimate, from the photos, the global carbohydrate content of their meals, and then to weigh each food. This allowed us to calculate the true carbohydrate content of the meals from nutritional tables.
Results: During the first phase, the carbohydrate contents of breakfast, lunch and dinner were 67 +/- 29, 72 +/- 30 and 74 +/- 30 g, respectively (mean +/- SD, n = 12). For a given patient, the variability in the carbohydrate content of each meal was expressed by its standard deviation (SD). For the 12 patients, this variability was 18+/-+ 8, 25+/-+ 8 and 27+/-+ 11 g, respectively, for breakfast, lunch and dinner, and represented on the average about one-quarter of the total carbohydrate content. During the second phase, carbohydrate content, estimated by the patients using the illustrated food directory, correlated well with the retrospective evaluation based on nutritional tables (y = 0.95 x + 5 g, r2 = 0.8 ; n = 12, 235 meals).
Conclusion: This new illustrated food repertory allows accurate evaluation of the highly variable carbohydrate content of meals, and could thus facilitate functional insulin therapy.