How is online self-reported weight compared with image-captured weight? A comparative study using data from an online longitudinal study of young adults

Abstract

Background Accurate anthropometric measurement is important within epidemiological studies and clinical practice. Traditionally, self-reported weight is validated against in-person weight measurement. Objectives This study aimed to 1) determine the comparison of online self-reported weight against images of weight captured on scales in a young adult sample, 2) compare this across body mass index (BMI), gender, country, and age groups, and 3) explore demographics of those who did/did not provide a weight image. Methods Cross-sectional analysis of baseline data from a 12-mo longitudinal study of young adults in Australia and the UK was conducted. Data were collected by online survey via Prolific research recruitment platform. Self-reported weight and sociodemographics (for example, age, gender) were collected for the whole sample (n = 512), and images of weight for a subset (n = 311). Tests included Wilcoxon signed-rank test to evaluate differences between measures, Pearson correlation to explore the strength of the linear relationship, and Bland-Altman plots to evaluate agreement. Results Self-reported weight [median (interquartile range), 92.5 kg (76.7–112.0)] and image-captured weight [93.8 kg (78.8–112.8)] were significantly different (z = −6.76, P < 0.001), but strongly correlated (r = 0.983, P < 0.001). In the Bland-Altman plot [mean difference −0.99 kg (−10.83, 8.84)], most values were within limits of agreement (2 standard deviation). Correlations remained high across BMI, gender, country, and age groups (r > 0.870, P < 0.002). Participants with BMI in ranges 30–34.9 and 35–39.9 kg/m2 were less likely to provide an image. Conclusions This study demonstrates the method concordance of image-based collection methods with self-reported weight in online research.