Reproducibility of self-reported past body weight

Objective: To examine the reproducibility of self-reported past body weight for ages 25 and 40 y and effects influencing recall of body weight.

Design and Subjects: Repeated administration of a computer-guided interview including questions on past body weight for 25 and 40 y of age to 120 middle-aged participants of a validation study on dietary assessment methods at the EPIC-Potsdam study centre. Recall error was defined as difference in body weight reported in the second compared to the first interview. Reproducibility of recall of body weight was assessed by plotting the difference between the interviews against the mean of the recalled weight. Possible influences of sex, age, measured current body weight, current body mass index, and highest education attained on recall error were examined.

Results: Mean difference in recalled body weight was small, but variation of differences was considerable. Classification into recall error strata demonstrated reproducibility of weight recall within +/- 3 kg for 75.8% for weight recall for age 25 y and 81.7% for age 40 y, respectively. Absolute recall error was not influenced by age, sex, current body weight, current BMI, and educational attainment.

Conclusions: We observed a high degree of reproducibility of self-reported past body weight for ages 25 and 40 y between the two interviews administered With respect to reproducibility administration of a single questionnaire or interview to obtain information on past body weight thus seems to be sufficient.