Background: Nutritional epidemiology relies largely on dietary assessment methods for the estimation of the "exposure' variables which may be related to disease risk.
Methods: This paper describes a methodological study conducted in Malmö, Sweden, to compare nutrient intake--estimated by two alternative dietary assessment methods--with a reference method consisting of 18 days of weighed food records. The two candidate methods were an extensive food frequency questionnaire with portion size to be estimated from a booklet of 120 sets of photos (method A) and a method involving the combination of a shorter questionnaire and a two-week food record (method B).
Results: In absolute values, both methods overestimated nutrient intake by 20-40%, with method B closer to the reference for most nutrients. Both crude and energy-adjusted correlations between A-reference and B-reference were of the order of 0.50-0.60 for energy, energy-providing nutrients and most vitamins and minerals. Correlations were in the same range for most of the 14 fatty acids considered in the analyses. Protein intake, estimated from the analyses of urinary nitrogen on 6-8 repeated 24-hour urine collections per subject, was almost identical to the reference method values. Correlation between nitrogen-derived values and dietary measurement was 0.75.
Conclusions: Overall, the study indicated that both methods A and B had good ranking validity compared to the reference and that in most cases the combined method (B) performed slightly better than the extensive food frequency method (A).