Michael C Brumm1,2, Michele M West3, Charles F Lynch3,4, Brian J Smith5,6. 1. Department of Biostatistics, University of Iowa College of Public Health, Iowa City, IA, USA. michael-brumm@uiowa.edu. 2. Holden Comprehensive Cancer Center, University of Iowa, Iowa City, IA, USA. michael-brumm@uiowa.edu. 3. Iowa Cancer Registry, University of Iowa College of Public Health, Iowa City, IA, USA. 4. Department of Epidemiology, University of Iowa College of Public Health, Iowa City, IA, USA. 5. Department of Biostatistics, University of Iowa College of Public Health, Iowa City, IA, USA. 6. Holden Comprehensive Cancer Center, University of Iowa, Iowa City, IA, USA.
Abstract
PURPOSE: Overweight and obesity are risk factors for several cancers; however, population-based cancer registries do not routinely collect data on body mass index (BMI). This study evaluated the utility of supplementing cancer registry data with BMI data derived from driver's license records. METHODS: We linked self-reported height and weight data from driver's license records to directly measured values, obtained via medical record abstraction, in a sample of 712 adult Iowa residents with cancer diagnosed during 2007-2012. Matched BMI values were subjected to a comprehensive evaluation of quantitative and categorical measures of agreement between data sources. RESULTS: Driver's license issue dates preceded diagnosis dates in 60.7% of cases, with time lags ranging from 3.0 years pre-diagnosis to 2.9 years post-diagnosis. Statistical analysis of agreement between continuous BMI values and ordinal BMI categories yielded an overall intraclass correlation estimate of 0.79 (95% confidence interval [CI] 0.77, 0.82) and an overall weighted kappa estimate of 0.63 (95% CI 0.59, 0.68), respectively. Subgroup analyses indicated reduced reliability among obesity-related cancers, particularly multiple myeloma, ovarian cancer, and pancreatic cancer. Neither measurement order nor time lag significantly affected agreement between BMI values. CONCLUSIONS: These findings suggest that self-reported driver's license data provide a reasonable approximation of BMI, but are less precise than interview- and questionnaire-based methods. Furthermore, the degree of bias is seemingly unaffected by measurement order and time lag, but appears to become more pronounced as BMI itself increases.
PURPOSE: Overweight and obesity are risk factors for several cancers; however, population-based cancer registries do not routinely collect data on body mass index (BMI). This study evaluated the utility of supplementing cancer registry data with BMI data derived from driver's license records. METHODS: We linked self-reported height and weight data from driver's license records to directly measured values, obtained via medical record abstraction, in a sample of 712 adult Iowa residents with cancer diagnosed during 2007-2012. Matched BMI values were subjected to a comprehensive evaluation of quantitative and categorical measures of agreement between data sources. RESULTS: Driver's license issue dates preceded diagnosis dates in 60.7% of cases, with time lags ranging from 3.0 years pre-diagnosis to 2.9 years post-diagnosis. Statistical analysis of agreement between continuous BMI values and ordinal BMI categories yielded an overall intraclass correlation estimate of 0.79 (95% confidence interval [CI] 0.77, 0.82) and an overall weighted kappa estimate of 0.63 (95% CI 0.59, 0.68), respectively. Subgroup analyses indicated reduced reliability among obesity-related cancers, particularly multiple myeloma, ovarian cancer, and pancreatic cancer. Neither measurement order nor time lag significantly affected agreement between BMI values. CONCLUSIONS: These findings suggest that self-reported driver's license data provide a reasonable approximation of BMI, but are less precise than interview- and questionnaire-based methods. Furthermore, the degree of bias is seemingly unaffected by measurement order and time lag, but appears to become more pronounced as BMI itself increases.
Authors: Béatrice Lauby-Secretan; Chiara Scoccianti; Dana Loomis; Yann Grosse; Franca Bianchini; Kurt Straif Journal: N Engl J Med Date: 2016-08-25 Impact factor: 91.245
Authors: Farhad Islami; Ann Goding Sauer; Kimberly D Miller; Rebecca L Siegel; Stacey A Fedewa; Eric J Jacobs; Marjorie L McCullough; Alpa V Patel; Jiemin Ma; Isabelle Soerjomataram; W Dana Flanders; Otis W Brawley; Susan M Gapstur; Ahmedin Jemal Journal: CA Cancer J Clin Date: 2017-11-21 Impact factor: 508.702
Authors: Sze Lin Yoong; Mariko Leanne Carey; Catherine D'Este; Robert William Sanson-Fisher Journal: BMC Med Res Methodol Date: 2013-03-13 Impact factor: 4.615