Computerized analysis of food records: role of coding and food composition database

Reported dietary intake records of 30 subjects (26 men and 4 women) were analysed by three different centres using their own computerized nutrient database systems. The agreement between systems was evaluated by different statistical criteria (the correlation coefficient, the mean difference and the proportion of individuals placed in the same thirds of distribution). Significant differences between the three systems were found in the calculation of alcohol, polyunsaturated fatty acids, saturated fatty acids, linoleic acid, linolenic acid, cholesterol, magnesium, sodium and water. To ascertain the extent of mean differences that could be attributed to the coding process or to the database used, coding forms of each centre were forwarded to the other two centres. Analysis of variance showed that differences in the data obtained by the three systems were mainly due to the food composition database used.