The accuracy of body weight and height recall in middle-aged men

OBJECTIVE: To determine the accuracy of recalled height and weight, and cal culated body mass index (BMI), over a 27-37 y period. DESIGN: Comparison of measured height and weight with recalled height and w eight 27-37 y later. PARTICIPANTS: Two hundred and twenty-five men measured aged 18-24y as physi cal education students at Loughborough Training College, UK, between 1958 a nd 1967. RESULTS: Initial body weights were over-estimated by 3.1 +/- 4.5 kg and hei ghts by 1.1 +/- 1.8 cm, on average. Some 42% (95) of recalls were within 2. 5 kg and 79% (178) within 2.5 cm, resulting in 58% (130) of the differences in BMI calculated from recalled and actual heights and weights to be withi n 1 kg/m(2). However, 29% (66) of recalls were more than 5 kg and 8% (19) m ore than 10 kg from the measured values. Weight errors (actual-recalled) we re negatively related (r = -0.43, P < 0.001) to weight gain over the 27 -37 y interval. CONCLUSIONS: Middle-aged men who were formerly physical education students recalled their previous height and weight well, in most cases, 27 -37 y lat er. The bias from recalled data would be to underestimate weight gain by 3 kg and BMI by 1 kg/m(2), on average. Errors of more than 5 kg in 29% of par ticipants and of more than 19 kg in 8% would be expected to interfere serio usly with attempts to show epidemiological relations between early weight b ased on recall and subsequent outcomes.