Some errors inherent in a longitudinal dietary survey revealed by the urine nitrogen test.

In the longitudinal population study 'Women in Gothenburg', subgroups of 154, 205 and 331 women in 1968/69, 1974/75 and 1980/81 respectively, were subjected to a dietary interview and collected a 24-h urine specimen for nitrogen analyses. The dietary data demonstrated a continuous increase in intake of energy and protein with time (2030, 2150 and 2350 kcal/d and 73, 77 and 90 g protein/d), while the 'true' protein intake (calculated from 24-h urine nitrogen) was unchanged (75, 73 and 75 g/d). The discrepancies were interpreted as being due to the changes in the dietary questionnaires made before the interviews in 1974/75, and again in 1980/81, changes expected to improve the method. In 1968/69 overweight women (BMI 24-30), as well as obese women (BMI greater than 30), reported a significantly smaller intake of energy and protein than lean women. Their reported protein intake was significantly lower than their 'true' protein intake (69 versus 77 g/d and 73 versus 94 g/d, respectively). Significant underreporting of dietary intake by the overweight and obese women in the studies 1974/75 and 1980/81 was revealed by the urine nitrogen test after correction for the methodological errors of the questionnaires. The results illustrate the necessity of using an independent validation test in dietary surveys.