Related Experiment Videos
Analysis of a clinical trial involving a combined mortality and adherence dependent interval censored endpoint
L A Moyé1, B R Davis, C M Hawkins
1University of Texas Health Science Center, Houston 77025.
Insights
Analyzing combined clinical trial endpoints with mixed censoring is challenging. This study introduces a novel rank-based statistic to effectively analyze these complex time-to-event measures, enhancing statistical power.
Area of Science:
- Biostatistics
- Clinical Trial Methodology
- Statistical Analysis
Background:
- Clinical trials frequently utilize combined endpoints, merging multiple measures for comprehensive assessment.
- Analysis of combined time-to-event endpoints is standard when both components are regularly measured.
- Challenges arise when one component is right-censored and the other is interval-censored.
Purpose of the Study:
- To develop and evaluate a statistical method for analyzing combined time-to-event endpoints with mixed censoring.
- To address the analytical difficulties posed by right-censored and interval-censored data within a single endpoint.
Main Methods:
- A novel test statistic is proposed, based on the rank ordering of events for the combined endpoint.
- The statistical power of the developed test statistic is explored and evaluated.
Main Results:
- The paper introduces a new rank-based statistic for analyzing combined endpoints with mixed censoring.
- The study investigates the power of this novel statistical approach.
Conclusions:
- The proposed rank-based statistic offers a viable method for analyzing complex combined endpoints in clinical trials.
- This approach can improve the statistical rigor and power when dealing with mixed right and interval censoring.
Abstract:
Clinical trials often involve a variety of clinical and laboratory measures that are used as endpoints and sometimes two of these measures are combined in one endpoint. When the individual components of such a combined endpoint are 'time to event' measurements, the analysis is straightforward if each of the components is measured frequently and regularly over time. However, the analysis of the combined endpoint is more difficult when one component of the endpoint is right censored and the other is interval censored. This paper describes a statistic, based on a rank ordering of events for such a combined measure. The power of the test statistic is explored.