First mammography screening participation and breast cancer incidence and mortality in the subsequent 25 years: population based cohort study

Abstract Objective To determine whether women who did not attend their first mammography screening invitation have a long term risk of poor screening adherence and breast cancer outcomes. Design Population based cohort study. Setting Stockholm, Sweden. Participants 432 775 women who received invitations to the Swedish Mammography Screening Programme between 1991 and 2020 and were initially invited at either 50 years of age or 40 years of age. Main outcome measures Screening adherence, breast cancer incidence, tumour characteristics, and breast cancer mortality tracked through linkage to multiple Swedish national registers, with follow-up until 2023 (up to 25 year follow-up period). Cumulative breast cancer incidences were calculated from first screening participation. Cox proportional hazards models estimated hazard ratios for breast cancer mortality; logistic regression models assessed associations with tumour characteristics by odds ratios. Results During a total of 4 940 375 person years of follow-up, 16 059 new cases of breast cancer were documented. Among women invited to their first mammography screening, 32.1% (n=138 760) did not participate. These non-participants were persistently less likely to attend subsequent screenings and were more likely have symptom detected, advanced stage breast cancer diagnosed. Specifically, compared with first screening participants, non-participants had an odds ratio of 1.53 (95% confidence interval 1.24 to 1.88) for stage III cancer (160 (4.1%) v 266 (2.9%) cases) and 3.61 (2.79 to 4.68) for stage IV cancer (150 (3.9%) v 105 (1.2%) cases). During a total of 6 818 686 person years of follow-up, 1603 deaths from breast cancer were documented. Non-participation at first screening was also associated with significantly higher breast cancer mortality, with a 25 year cumulative mortality of 9.9 per 1000 versus 7.0 per 1000 for participants (adjusted hazard ratio 1.40, 95% confidence interval 1.26 to 1.55). By contrast, the 25 year breast cancer incidence was similar between groups (7.8% in participants versus 7.6% in non-participants), suggesting that the elevated mortality among first screening non-participants likely reflects delayed detection rather than increased incidence. Conclusions This study shows that first screening non-participants represent a large population at long term risk of dying from breast cancer, providing an opportunity for targeted interventions to improve adherence to screening and thereby decrease mortality risk.