Repeatability of three-day dietary records in the EURODIAB IDDM complications study

Repeatability of a dietary method is important in determining the quality of nutritional data. It should be assessed in the population of interest. This study evaluated the repeatability of nutritional data from standardized three-day dietary records, from the clinic-based, cross-sectional multi-centre EURODIAB IDDM Complications Study. 15% of the total EURODIAB cohort was randomly selected to test the repeatability of nutritional intake data. Two three-day records, completed three weeks apart, were available for 216 diabetic patients (7.5%) representative of the total cohort. All records were analysed centrally, for intakes of protein (animal and vegetable), fat (saturated fat and cholesterol), carbohydrate, fibre, alcohol and energy. Repeatability was measured comparing mean intakes, determining the proportion of patients classified into the same/opposite quartile by the two three-day records and assessing mean differences with standard deviations (s.d.d). There were no significant differences in mean energy and nutrient intakes between the first and second records. Classification of individuals into the opposite quartile occurred only in 0-4% of patients and overall about 50% (range 44-74%) of the subjects were classified into the same quartiles of intakes. Only small mean differences were found for energy intake (-156 (1633) kJ; 95% confidence limits -375, 63 kJ) and nutrients with s.d.ds comparable to intra-individual variations in the general population. The differences in energy intake were randomly distributed over the range of intakes. The present study demonstrates that standardized three day dietary records show a high degree of repeatability within a short period of time in a sample of European IDDM patients. The good repeatability strengthens the conclusions drawn from the nutritional data in the EURODIAB IDDM Complications Study.