Estimands framework for adverse-event analysis with varying follow-up (benefit assessment)
This method helps researchers correctly assess the safety of medical treatments by accounting for differences in how long patients are followed in clinical trials. It uses the concept of 'estimands'—clear definitions of what should be measured—to ensure that adverse events are analyzed in a way that supports fair and accurate benefit-risk evaluations.
At a glance
Use when
Analyzing safety data in randomized clinical trials where follow-up times vary between patients; conducting benefit-risk assessments requiring precise safety estimands; preparing submissions for HTA agencies with rigorous safety evaluation standards
Avoid when
When adverse event data are sparse or follow-up is uniformly short across all patients; when only descriptive summaries of adverse events are needed without causal or comparative interpretation
Inputs
Clinical trial data on adverse events, including timing of events and patient follow-up durations; definition of the safety estimand of interest (e.g., treatment policy, hypothetical, composite)
Outputs
Clearly defined safety estimands and corresponding statistical estimates (e.g., hazard ratios, cumulative incidence) that account for varying follow-up times
How it works
The framework applies the estimands concept, originally developed for efficacy, to safety outcomes in clinical trials, particularly focusing on time-to-first adverse event of a specific type. It addresses challenges arising from varying follow-up times across patients and proposes appropriate statistical estimators for defined safety estimands. The paper reviews current HTA agency practices, highlights issues in meta-analyses of adverse event data, and provides recommendations for analysis methods that align with the target estimand, ensuring valid inference in benefit-risk assessment.
- HTA domains
- Clinical Effectiveness, Safety
- Categories
- AppraisalHeterogeneity
- Assumptions
- Adverse event occurrence is accurately recorded and follow-up times are known; the chosen estimand aligns with the clinical question; censoring mechanisms are appropriately handled in the analysis
- Strengths
- Promotes alignment between clinical objectives and statistical analysis in safety assessment; reduces risk of misleading conclusions due to inappropriate methods; enhances transparency and interpretability of safety results; supports consistent evaluation by HTA bodies
- Limitations
- Requires careful pre-specification of estimands; may be complex to implement in trials with complex treatment patterns or high dropout rates; limited guidance on handling recurrent events within this specific framework
- Also known as
- Estimands for safety analysis, Safety estimands framework, Adverse event estimands with varying follow-up
Questions this answers
- › What should be estimated when analyzing adverse events in clinical trials with differing follow-up times?
- › How can the estimands framework be applied to safety endpoints?
- › Which statistical estimators are appropriate for time-to-first adverse event under varying follow-up?
- › How do varying follow-up times bias safety assessments if not properly addressed?
- › What are current HTA agency practices in evaluating safety data?
- › What challenges arise in meta-analyses of adverse event data and how can they be mitigated?
References & sources
Similar by meaning
- Biometrical methods for analysis of adverse events in benefit assessment
- SAVVY (Survival analysis for AdVerse events with VarYing follow-up times)
- Change-point survival models for relative treatment effect extrapolation
- Structured expert elicitation for long-term survival outcomes (NICE DSU TSD 26)
- NICE DSU Technical Support Document 14
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.