Survival analysis for AdVerse events with VarYing follow‐up times (SAVVY): Rationale and statistical concept of a meta‐analytic study

Abstract

The assessment of safety is an important aspect of the evaluation of new therapies in clinical trials, with analyses of adverse events being an essential part of this. Standard methods for the analysis of adverse events such as the incidence proportion, that is the number of patients with a specific adverse event out of all patients in the treatment groups, do not account for both varying follow‐up times and competing risks. Alternative approaches such as the Aalen–Johansen estimator of the cumulative incidence function have been suggested. Theoretical arguments and numerical evaluations support the application of these more advanced methodology, but as yet there is to our knowledge only insufficient empirical evidence whether these methods would lead to different conclusions in safety evaluations. The Survival analysis for AdVerse events with VarYing follow‐up times (SAVVY) project strives to close this gap in evidence by conducting a meta‐analytical study to assess the impact of the methodology on the conclusion of the safety assessment empirically. Here we present the rationale and statistical concept of the empirical study conducted as part of the SAVVY project. The statistical methods are presented in unified notation, and examples of their implementation in R and SAS are provided.

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

“…Therefore, the use of the incidence proportion may often be justified if censoring is low and the CEs have been specified appropriately. This emphasizes the need to consider thoroughly how the CEs are defined in the specific trial as this directly impacts the amount of censoring, see Stegherr et al 18 for guidance on this aspect. As outlined earlier, the type of CEs will also have an impact on the type of estimands, typically either while on treatment or treatment policy, and the type of CEs may be influenced by the design of AE follow‐up.…”

“…This is reflected by the fact that AE follow‐up is often shorter than follow‐up for primary endpoints in oncology such as death. For instance, in oncology, another CE could be progression, in which case we would consider the intention‐to‐treat during progression‐free survival, corresponding to the while‐on‐treatment estimand when progression leads to end of treatment 18 . These considerations also have a technical aspect, at least for non‐parametric survival analyses.…”

Assessing the Impact of COVID-19 on the Clinical Trial Objective and Analysis of Oncology Clinical Trials—Application of the Estimand Framework

“…Such events might be declared as competing and estimators other than the simple incidence proportion, for example, the Aalen–Johansen estimator, may potentially give more accurate estimates of the absolute AE risk of the actual treatment (Aalen and Johansen 1978 ; Stegherr et al. 2019 ), with the same limitation of patients potentially only being at risk during the pandemic, see Section 3.2. In particular, in the absence of randomization in single-arm trials or, if there is an overlap between known risks of the treatment and COVID-19 related AEs, the assessment of trial treatment’s or COVID-19 contribution may be challenging.…”

Clinical Trial Drug Safety Assessment for Studies and Submissions Impacted by COVID-19

“…See Section 10.9 of PHUSE ( 2017 ) and Stegherr et al. ( 2019 ) (and the references therein) for a discussion of some pitfalls of percentage-based methods and possible alternative methods. With the COVID-19 pandemic, it can be expected that more patients will discontinue early or have periods in which study drug has been interrupted.…”

“… 2019 ) if not already incorporated, and/or to use alternative methods proposed in recent literature (e.g., Stegherr et al. 2019 and references therein). While we acknowledge the need for future research on methods for safety analyses and implementation of the estimand framework for safety in clinical trials, to assist cross-functional study teams with the most immediate needs, we have chosen to focus on the safety analyses that are likely already planned and the associated purpose without introducing estimand-related language.…”