Rahmatollah Moradzadeh1, Mohammad Ali Mansournia2, Reza Ghiasvand3, Taban Baghfalaki4, Haidar Nadrian5, Kourosh Holakouie-Naieni2. 1. Department of Epidemiology, School of Health, Arak University of Medical Sciences, Arak, Iran. 2. Department of Epidemiology and Biostatistics, School of Public Health, Tehran University of Medical Sciences, Tehran, Iran. 3. Oslo Center for Biostatistics and Epidemiology, Institute of Basic Medical Sciences, University of Oslo, Norway. 4. Department of Statistics, Faculty of Mathematical Sciences, Tarbiat Modares University, Tehran, Iran. 5. Department of Health Education & Promotion, Faculty of Health Sciences, Tabriz University of Medical Sciences, Tabriz. Iran.
Abstract
BACKGROUND: Our aim was to determine the association between age at menarche (AAM) and breast cancer adjusted for recall bias (misclassification) in AAM. METHODS: We have used data provided from a case-control study conducted in Iran from 2005 to 2009. The cases and controls were frequency matched based on 5-year age groups and region of residence. First, logistic regression was conducted to estimate the odds ratio (OR) and second, Bayesian analysis was applied to estimate the ORs adjusted for misclassification. RESULTS: The study was conducted on 880 cases and 998 controls. In the assumption of no correction for recall bias on self-reported AAM, the OR was 1.36 (95% Credible Interval (0.98, 1.90). Based on a sensitivity value = 71% and a specificity value = 81% (the indices about the ratio of true recall of AAM) for the case and control groups (as the first scenario), the AAM ≤ 12 years of age was associated with a lower OR for breast cancer by 1.23 (95% Credible Interval: 0.50, 3.13). In the other scenario, with consideration of 100% sensitivity and specificity of self- reported AAM in the case group, and 71% and 81% sensitivity and specificity of the item in the control group, the related OR between breast cancer and AAM was found increased to 2.96 (95% Credible Interval: 0.75, 7.66). CONCLUSION: After adjustment for misclassification related to recall bias, this study provides evidence that the self-reported mode of AAM has a moderate impact on calculation of the OR.
BACKGROUND: Our aim was to determine the association between age at menarche (AAM) and breast cancer adjusted for recall bias (misclassification) in AAM. METHODS: We have used data provided from a case-control study conducted in Iran from 2005 to 2009. The cases and controls were frequency matched based on 5-year age groups and region of residence. First, logistic regression was conducted to estimate the odds ratio (OR) and second, Bayesian analysis was applied to estimate the ORs adjusted for misclassification. RESULTS: The study was conducted on 880 cases and 998 controls. In the assumption of no correction for recall bias on self-reported AAM, the OR was 1.36 (95% Credible Interval (0.98, 1.90). Based on a sensitivity value = 71% and a specificity value = 81% (the indices about the ratio of true recall of AAM) for the case and control groups (as the first scenario), the AAM ≤ 12 years of age was associated with a lower OR for breast cancer by 1.23 (95% Credible Interval: 0.50, 3.13). In the other scenario, with consideration of 100% sensitivity and specificity of self- reported AAM in the case group, and 71% and 81% sensitivity and specificity of the item in the control group, the related OR between breast cancer and AAM was found increased to 2.96 (95% Credible Interval: 0.75, 7.66). CONCLUSION: After adjustment for misclassification related to recall bias, this study provides evidence that the self-reported mode of AAM has a moderate impact on calculation of the OR.