Applying competing risks analysis to recurrences of primary breast cancer

Introduction: in time-to-event analysis, competing risks are observed when a subject is at risk of multiple types of events. A competing risk is an event that prevents the event of interest from happening or modifies its chance to occur. Objectives: the purposes of this article are to provide an overview of competing risks analysis and to illustrate its application to the follow-up of breast cancer patients in order to estimate the risk of disease recurrence. Design: cohort study. Setting and participants: population-based random sample of patients with stage I-III primary female breast cancer diagnosed in 2000-2013, aged 15 years or older, resident in the Forli health care district ( northern Italy), treated surgically and with available information regarding disease laterality. Main outcome measures: subhazard ratio, hazard ratio and cumulative incidence function for loco-regional recurrences and distant metastasis during 10 years of follow-up. Methods: breast cancer recurrences were evaluated taking into account death from any cause and occurrence of a second primary breast cancer as competing risks. Recent developments in competing risks methodology were used. The results obtained were compared with those from the Cox regression model, in which the competing risks were not accounted for. Conclusions: estimating the risk of recurrence without accounting for the competing risks resulted in a divergence of the hazard/subhazard functions. Competing risks analysis is preferable when the statistical assumption of independence of end-points may be violated.