RECALL OF BODY-WEIGHT AND BODY-SIZE ESTIMATION IN WOMEN ENROLLED IN THE BREAST-CANCER DETECTION AND DEMONSTRATION PROJECT (BCDDP)

OBJECTIVE: To examine the relationship between body mass index (BMI) a nd pictorial representations of body size in women from young adulthoo d to late mid-life. DESIGN: Retrospective, descriptive study of BMI an d pictorial estimation of body size. SUBJECTS: 5,807 women age 33-77 y ears enrolled in the National Cancer Institute's and American Cancer S ociety's Breast Cancer Detection and Demonstration Project (BCDDP). ME ASUREMENTS: Body weight and height were measured in 1973. In 1977, a s ubset of the cohort recalled their usual height and weight at 10 y int ervals starting at age 20. In 1987, subjects reported their usual and current weight and selected one of nine pictorials best representing t heir body size at ages 15, 25, 40, 50 and +60 y. RESULTS: For the coho rt, and among White women, Pearson correlations between recalled BMI ( Kg/M(2)) and pictorials for each decade ranged from 0.62-0.67 and was 0.80 for current BMI and current pictorial. The range of correlations between pictorials and recalled BMI for other race/ethnic groups were 0.72-0.87 (Black), 0.53-0.75 (Hispanic) and 0.28-0.87 (Asian). Among a subset of women with data on measured BMI, recalled BMI and pictorial s at specific ages, the correlation between pictorials and measured BM I was 0.75, compared to the correlation between recalled BMI and measu red BMI which was 0.89. CONCLUSION: Correlations are higher between re called BMI and measured BMI compared to the correlation between pictor ials and measured BMI. Therefore, estimates of body size by pictorials alone may not be appropriate for epidemiological investigations. Alte rnate uses of pictorials may include assessment body weight in low lit erate populations or in instances where body weight is not or has not been measured.