Factors associated with non-participation in breast cancer screening: Analysis of the 2016 and 2019 Comprehensive Survey of Living Conditions in Japan

Abstract Background The participation rate for breast cancer screening remains to be suboptimal in Japan. Therefore, it is important to identify factors associated with non-participation and to identify people at high risk for non-participation. Methods We carried out a cross-sectional study using the data of women aged 40–74 years from the 2016 and 2019 Comprehensive Survey of Living Conditions. We selected candidate predictor variables from the survey sheets and conducted a multivariable logistic regression for non-participation in breast cancer screening for the past 2 years. In addition, using data from 2016, we created an integer risk score for non-participation and tested its predictive performance in 2019. Results The proportion of participants in breast cancer screening in 2016 and 2019 were 46.7% (50,177/107,513) and 48.7% (49,498/101,716), respectively. In multivariable logistic regression analysis, age over 50 years, single/divorced/widowed, lower education level, lower household expenditure, being insured for National Health Insurance, employed to small/middle scale company, non-regularly employed, current smoker, never/quit drinking or middle/high-risk drinking, lower self-rated health status, higher Kessler Psychological Distress Scale score, non-participation in the annual health checkups for diseases other than cancer, not constantly visiting hospitals/clinics showed a positive association with non-participation. The 9-item risk score (age, marital status, education, health insurance plan, employment, smoking, drinking, non-participation in the annual health checkups for diseases other than cancer, and not constantly visiting hospitals/clinics) and 3-item risk score (age, health insurance plan, non-participation in the annual health checkups for diseases other than cancer) showed the area under the receiver operating characteristic curve of 0.744 and 0.720, respectively. Conclusion We identified factors associated with non-participation in breast cancer screening. The simple risk score would be useful for public health sectors to identify people at risk for non-participation.