Estimating and comparing adverse event probabilities in the presence of varying follow-up times and competing events.

Safety analyses in terms of adverse events (AEs) are an important aspect of benefit-risk assessments of therapies. Compared to efficacy analyses AE analyses are often rather simplistic. The probability of an AE of a specific type is typically estimated by the incidence proportion, sometimes the incidence density or the Kaplan-Meier estimator are proposed. But these analyses either do not account for censoring, rely on a too restrictive parametric model, or ignore competing events. With the non-parametric Aalen-Johansen estimator as the gold-standard, these potential sources of bias are investigated in a data example from oncology and in simulations, both in the one-sample and in the two-sample case. As the estimators may have large variances at the end of follow-up, the estimators are not only compared at the maximal event time but also at two quantiles of the observed times. To date, consequences for safety comparisons have hardly been investigated in the literature. The impact of using different estimators for group comparisons is unclear, as, for example, the ratio of two both underestimating or overestimating estimators may or may not be comparable to the ratio of the gold-standard estimator. Therefore, the ratio of the AE probabilities is also calculated based on different approaches. By simulations investigating constant and non-constant hazards, different censoring mechanisms and event frequencies, we show that ignoring competing events is more of a problem than falsely assuming constant hazards by use of the incidence density and that the choice of the AE probability estimator is crucial for group comparisons.