Objective: To estimate the relative validity of a computerised dietary history instrument (DISHES 98).
Settings: Munich and Berlin.
Subjects: A total of 148 persons aged 19-59 y recruited from two research centres.
Design: A relative validation study. Energy and macronutrient intakes obtained with DISHES 98 were compared to those assessed with 3-day weighed dietary records and with a 24 h dietary recall.
Results: Intakes of energy, total, saturated and monounsaturated fat, polysaccharides and alcohol were significantly higher and intake of dietary fibre was significantly lower with the 3-day records than with DISHES 98. For intakes of total, animal and vegetable protein, total carbohydrates, mono- and disaccharides and cholesterol the mean difference between DISHES 98 and the 3-day dietary records was less than 5% of the intake with DISHES 98. Pearson's correlation coefficients between DISHES 98 and 3-day records varied from 0.34 for intake of polyunsaturated fat to 0.69 for intake of disaccharides and from 0.27 for polyunsaturated fat to 0.65 for total carbohydrates between DISHES 98 and the 24 h recall. The proportion of participants classified into the same or adjacent quintile of intake varied between 66.9% for polyunsaturated fat and 90.4% for alcohol comparing DISHES 98 and 3-day records and between 60.2% for polyunsaturated fat and 78.4% for total carbohydrates comparing DISHES 98 and 24 h recalls.
Conclusion: The observed differences between DISHES 98 and the other methods are in an acceptable range for assessing dietary intake in epidemiologic studies.