Related Experiment Videos
Meta-analysis of failure-time data with adjustment for covariates
1Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts.
Summary
This study introduces a method to combine medical study data by adjusting for patient differences, improving accuracy in failure-time analysis. Applying this to angioplasty data reveals significant variations in artery patency based on lesion type and patient factors.
Area of Science:
- Medical Statistics
- Biostatistics
- Health Services Research
Background:
- Combining failure-time data from multiple studies is crucial for robust medical research.
- Heterogeneity in patient populations (case-mix) across studies can bias pooled results.
- Existing methods may not adequately account for case-mix differences, leading to inaccurate conclusions.
Purpose of the Study:
- To present and illustrate a novel technique for combining failure-time data from diverse sources.
- To adjust for case-mix variations among studies before pooling results.
- To apply the technique to femoropopliteal angioplasty patency data.
Main Methods:
- Utilized a proportional-hazards model and actuarial life-table approach.
- Assumed study variations are partly due to case-mix heterogeneity.
- Adjusted for prognostic factors (lesion type, runoff) before pooling patency data.
Main Results:
- Unadjusted pooled five-year patency for femoropopliteal angioplasty was 45%.
- Adjusted five-year patency varied significantly: 60% (stenosis, good runoff) to 24% (occlusion, poor runoff).
- Case-mix adjustment revealed substantial variability not apparent in unadjusted pooled data.
Conclusions:
- Pooling studies without case-mix adjustment yields misleading average results with narrow confidence intervals.
- The presented technique effectively combines failure-time data while accounting for case-mix.
- Accurate meta-analysis of medical studies requires adjustment for patient-level prognostic factors.