Accuracy of body mass index estimated from self-reported height and weight in mid-aged Australian women

Objective: To assess the accuracy of body mass index (BMI) estimated from self-reported height and weight from a mailed survey, in a population-based sample of mid-aged Australian women. Methods: One hundred and fifty nine women (age 54-59 years) were recruited from the Australian Longitudinal Study on Women's Health (ALSWH). Participants provided height and weight data in a mailed survey and were then measured (Brisbane, Australia 2005). Differences between self-reported and measured data were examined by plotting against the measured values and using paired t-tests and kappa statistics. Factors associated with biased reporting were assessed using regression models. Results: Both self-reported height and weight tended to be underestimated, with a mean difference of 0.67 cm (95% CI 0.26 to 1.08 cm) and 0.95 kg (95% CI 0.44 to 1.47 kg) respectively. Reported height and derived BMI was more accurate among married women than single women (average difference of 1.28 cm, 95% CI 0.19 to 2.37 cm and -1.00 kg/m(2), 95% CI -1.69 to -0.30, respectively). Women with BMI 18.5-24.9 kg/m(2) reported weight more accurately than obese women (average difference of 2.26 kg, 95% CI 0.14 to 4.38 kg). There was 84% agreement between BMI categories derived from self-reported and measured data, with 85%, 73% and 94% of women correctly classified as obese, overweight, and healthy BMI using self-reported data and kappa=0.81. Conclusions: There is substantial agreement between self-reported and measured height and weight data for mid-aged women, especially among married and healthy weight women. Implications: Population-based studies among mid-aged women in Australia can use self-reported data obtained from mailed surveys to derive BMI estimates.