Accounting for Competing Events in Longitudinal Studies of Geriatric Outcomes

In time-to-event analyses of geriatric outcomes, competing events such as mortality are common but are often handled inappropriately, leading to biased risk estimates and potentially incorrect inferences. We present a case study illustrating the consequences of failing to account for competing events and review how such analyses are conducted in recent JAGS publications.

For the case study, we analyzed data from the Precipitating Events Project to estimate the risk and rate of two geriatric outcomes-injurious falls and restricted activity-using methods that appropriately account for competing events and, for comparison, methods that assume competing events are absent. For the literature review, we searched PubMed for eligible JAGS articles in 2024 and assessed whether and how competing events were addressed in analyses of risk and rate.

In the case study, we demonstrate that methods that failed to account for competing events produce inflated estimates of risk and potentially misleading differences across strata of independent variables. Also, when competing events are present, the statistical significance of the association between an independent variable and outcome rate and risk, respectively, was not consistently concordant. In the literature review, only 47% of articles reported analyses of both risk and rate, and terminology used to describe risk analyses often limited interpretation.

Failure to properly account for competing events in longitudinal studies of geriatric outcomes can result in biased risk estimates and incorrect inferences. Clearer reporting of time-to-event analyses would strengthen the rigor and interpretability of scientific reports in the geriatric literature.