Association between Multiple Recurrent Events with Multivariate Modeling: A Retrospective Cohort Study

Recurrent event data are often encountered in biomedical research, for example, recurrent infections or recurrent hospitalizations for patients after renal transplant. In many studies, there are more than one type of events of interest. We aimed to identify the association between two types of events using multivariate joint modeling and then apply this statistical method in the clinical data set.

A retrospective cohort study.

Overall, 342 subjects with breast cancer whose records were registered for follow-up in a Cancer Research Center at Shohadaye Tajrish Hospital, Tehran, Iran from 2006 to 2015 were investigated. These patients were monitored for at least 6 months after diagnosis and their latest status were recorded. Joint frailty model was used for modeling the relationship between two types of recurrences with Frailty package in R software.

When the terminal event was considered as death, three-year and five-year survival rates for the patients were 0.79 and 0.68, respectively. Given the results obtained from a fitted joint frailty model, the risk of multiple recurrences (local and metastases) increased for the patients with tumor grades greater than I.

With regard to the significant variance of the frailty component of the metastases event, it can be inferred that patients with the same predictive variables are prone to different levels of metastases risk and, on the other hand, given the low frequency of types of recurrences, caution should be exercised when considering the obtained results.