Related Experiment Videos
Modelling discrete time survival data with random slopes: evaluating haemodialysis centres.
Marilia Sá Carvalho1, Leonhard Knorr-Held
1National School of Public Health/FIOCRUZ, Rio de Janeiro, Brazil. carvalho@procc.fiocruz.br
Statistics in Medicine
|November 6, 2003
Summary
We developed a new survival model for haemodialysis patients to analyze dialysis center performance and patient outcomes. The model revealed significant variations in patient survival trends across different dialysis centers.
Area of Science:
- Biostatistics
- Public Health
- Nephrology
Background:
- Registry data from haemodialysis patients in Rio de Janeiro, Brazil, is crucial for understanding treatment variations.
- Dialysis center performance can significantly impact patient survival rates.
- Existing models may not fully capture center-specific temporal trends in patient outcomes.
Purpose of the Study:
- To propose a hierarchical discrete time survival model for analyzing haemodialysis registry data.
- To estimate hazard ratio differences attributed to dialysis center performance.
- To adjust for individual and center-level covariates and model residual calendar time trends.
Main Methods:
- A hierarchical discrete time survival model was developed.
- The model incorporated a random slope approach to estimate time-varying trends across centers.
- Registry data from multiple dialysis centers in Rio de Janeiro was analyzed.
Main Results:
- Significant variations in residual calendar time trends were observed across dialysis centers.
- These variations persisted after adjusting for important observed covariates.
- The model successfully estimated differences in hazard ratios linked to center performance.
Conclusions:
- The proposed hierarchical discrete time survival model effectively analyzes registry data for haemodialysis patients.
- Dialysis center performance exhibits significant, time-varying effects on patient survival.
- This modeling technique is adaptable for survival time analyses in other diseases.