ABSTRACT
OBJECTIVES: This study aimed to estimate the validity of self-reported information on ever-night shift work among women with and without breast cancer and illustrate the consequences for breast cancer risk estimates. METHODS: During 2015-2016, 225 women diagnosed with breast cancer and 1800 matched controls without breast cancer employed within the Danish hospital regions during 2007-2016 participated in a questionnaire-based survey. Their reported night shift work status was linked with objective payroll register day-by-day working hour data from the Danish Working Hour Database and the Danish Cancer Registry. For the breast cancer patients and their matched controls, we estimated sensitivity and specificity for ever-working night shifts using the payroll data as the gold standard. We also used quantitative bias analysis to estimate the impact on relative risk estimates for a hypothetical population. RESULTS: For breast cancer patients, we observed a sensitivity of ever-night shifts of 86.2% and a specificity of never-night shifts of 82.6%. For controls, the sensitivity was 80.6% and the specificity 83.7%. Odds ratio for breast cancer in a hypothetical population decreased from 1.12 [95% confidence interval (CI) 1.03-1.21] to 1.05 (95% CI 0.95-1.16) when corrected by the sensitivity and specificity estimates. CONCLUSION: This study shows that female breast cancer patients had slightly better recall of previous night shift work than controls. Additionally, both breast cancer patients and controls recalled previous never-night shift work with low specificity. The net effect of this misclassification is a small over-estimation of the relative breast cancer risk due to night shift work.