Abstract
Introduction: We investigated an alternative for self-reported past adult body weight resulting in less missing data, but still providing a valid measure for ranking participants.
Methods: We collected information from 2,647 women aged ≥45 years on their height and weight (to calculate body mass index (BMI)) when they were in their early thirties. Participants were also asked how their weight in their early thirties compared with that of other women of the same age and height (comparative weight), categorised as ‘a lot less’, ‘a little less’, ‘about the same’, ‘a little more’ and ‘a lot more’. These two weight measures were compared using a Spearman rank correlation coefficient.
Results: Data was missing less often for comparative weight (1.2%) than self reported weight (12.5%). Of the 2,304 women with both BMI and comparative weight data, the two measures fell in the same category for 54.3%. Comparative weight underestimated self-reported BMI for 36.4%. The correlation coefficient was 0.64 (0.61–0.66), indicating that comparative weight is in overall good correlation with past BMI.
Conclusion: We showed that measurement of past comparative weight (1) reduced missing data compared to recalled BMI, and (2) a good correlation between measures indicating that comparative weight may be a useful alternative for ranking on self-reported past BMI.
Methods: We collected information from 2,647 women aged ≥45 years on their height and weight (to calculate body mass index (BMI)) when they were in their early thirties. Participants were also asked how their weight in their early thirties compared with that of other women of the same age and height (comparative weight), categorised as ‘a lot less’, ‘a little less’, ‘about the same’, ‘a little more’ and ‘a lot more’. These two weight measures were compared using a Spearman rank correlation coefficient.
Results: Data was missing less often for comparative weight (1.2%) than self reported weight (12.5%). Of the 2,304 women with both BMI and comparative weight data, the two measures fell in the same category for 54.3%. Comparative weight underestimated self-reported BMI for 36.4%. The correlation coefficient was 0.64 (0.61–0.66), indicating that comparative weight is in overall good correlation with past BMI.
Conclusion: We showed that measurement of past comparative weight (1) reduced missing data compared to recalled BMI, and (2) a good correlation between measures indicating that comparative weight may be a useful alternative for ranking on self-reported past BMI.
Original language | English |
---|---|
Pages (from-to) | 46-48 |
Number of pages | 3 |
Journal | Australasian Epidemiologist |
Volume | 23 |
Issue number | 1 |
Publication status | Published - Jul 2016 |