a1 University of Cambridge, Institute of Public Health, Strangeways Research Laboratory, Wort's Causeway, Cambridge CB1 8RN, UK
a2 JDRF/WT Diabetes and Inflammation Laboratory, University of Cambridge, Cambridge CS2 2XY, UK
a3 Dunn MRC Human Nutrition Unit, Cambridge, CB2 2XY, UK
Background and objective: A new data-entry system (DINER – Data Into Nutrients for Epidemiological Research) for food record methods has been devised for the European Prospective Investigation into Cancer (EPIC) cohort study of 25 000 men and women in Norfolk. DINER has been developed to address the problems of efficiency and consistency of data entry, comparability of data, maximising information and future flexibility in large long-term population studies of diet and disease that use record methods to assess dietary intakes. DINER captures more detail than traditional systems and enables provision of new variables for specific food types or groups. The system has been designed to be fully flexible and easy to update. Analysis of consistency of data entry was tested in a group of 3525 participants entered by 25 coders.
Results: A food list of 9000 food items and values for 24 000 portion sizes have been incorporated into the database, using information from the 5979 diaries coded since 1995. Analysis of consistency of entry indicated that this has largely been achieved. The effect of coders in a multivariate regression model was significant only if the three coders involved in early use of the program were included (P<0.013).
Conclusions: The development of DINER has facilitated the use of more accurate record methods in large-scale epidemiological studies of diet and disease. Furthermore, the retention of original information as an extensive food list allows greater flexibility in later analyses of data of multiple dietary hypotheses.
(Received March 30 2001)
(Accepted May 22 2001)