Related Experiment Videos
Event history analysis and inference from observational epidemiology
1Department of Biostatistics, University of Copenhagen, Blegdamsvej 3, DK-2200 Copenhagen N, Denmark. N.Keiding@biostat.ku.dk
Statistics in Medicine
|September 4, 1999
Summary
Observational studies benefit from including time, but face challenges like time-dependent confounders. This research explores event history analysis and structural nested failure time models to assess the graft-versus-leukaemia effect in bone marrow transplantation.
Area of Science:
- Epidemiology
- Biostatistics
Background:
- Observational studies often lack temporal dynamics, potentially weakening causal inference.
- Time-dependent confounders, which change roles over time, present significant challenges in epidemiological research.
Purpose of the Study:
- To explore advanced statistical methods for strengthening causal inference in observational epidemiology.
- To address the complexities of time-dependent confounders in longitudinal studies.
- To empirically assess the graft-versus-leukaemia effect following bone marrow transplantation.
Main Methods:
- Application of event history analysis techniques.
- Utilisation of structural nested failure time models.
- Focus on observational epidemiological data from bone marrow transplant recipients.
Main Results:
- Demonstration of the utility of event history analysis in handling time-dependent covariates.
- Application of structural nested failure time models to disentangle complex temporal relationships.
- Empirical assessment of the graft-versus-leukaemia effect.
Conclusions:
- Systematic inclusion of time enhances epidemiological inference.
- Event history analysis and structural nested failure time models are valuable tools for addressing time-dependent confounding.
- These methods provide a robust framework for evaluating treatment effects in complex longitudinal settings like bone marrow transplantation.