SAS and R functions to compute pseudo-values for censored data regression

Area of Science:

• Biostatistics
• Survival Analysis
• Medical Informatics

Background:

• Direct regression modeling of survival data, including the survival function, restricted mean, and cumulative incidence function, is essential for analyzing right-censored outcomes.
• Existing methods often require complex statistical techniques or specialized software.
• Pseudo-value-based approaches offer a promising alternative for direct modeling.

Purpose of the Study:

• To present user-friendly SAS macros and R functions for computing pseudo-values.
• To facilitate direct regression modeling of survival functions, restricted mean, and cumulative incidence functions using standard generalized estimating equation software.
• To demonstrate the application of these tools in a bone marrow transplant patient study.

Main Methods:

• Development and implementation of SAS macros and R functions to compute pseudo-values from right-censored survival data.
• Utilizing generalized estimating equation (GEE) software for fitting regression models after pseudo-value computation.
• Application of the developed routines to a real-world dataset from bone marrow transplant patients.

Main Results:

• The provided SAS macros and R functions successfully compute pseudo-values.
• Regression estimates for survival analysis can be readily obtained using standard GEE software with the computed pseudo-values.
• The methodology is effectively illustrated using data from a bone marrow transplant study.

Conclusions:

• The developed SAS macros and R functions provide accessible tools for pseudo-value computation.
• This approach simplifies direct regression modeling of survival outcomes with right-censored data.
• The routines are practical for researchers in biostatistics and clinical epidemiology, particularly those analyzing transplant data.