─ Session info ───────────────────────────────────────────────────────────────
setting value
version R version 4.4.2 (2024-10-31)
os Ubuntu 24.04.1 LTS
system x86_64, linux-gnu
ui X11
language (EN)
collate C.UTF-8
ctype C.UTF-8
tz UTC
date 2024-12-16
pandoc 3.4 @ /opt/quarto/bin/tools/ (via rmarkdown)
─ Packages ───────────────────────────────────────────────────────────────────
package * version date (UTC) lib source
survival 3.7-0 2024-06-05 [2] CRAN (R 4.4.2)
[1] /home/runner/work/CAMIS/CAMIS/renv/library/linux-ubuntu-noble/R-4.4/x86_64-pc-linux-gnu
[2] /opt/R/4.4.2/lib/R/library
─ External software ──────────────────────────────────────────────────────────
setting value
SAS 9.04.01M7P080520
──────────────────────────────────────────────────────────────────────────────R vs SAS - Estimating and Testing Cause-Specific Hazard
Comparison of R and SAS
The following table shows the options available in R and SAS for estimating and testing cause-specific hazard in a competing risk analysis, especially the capabilities and whether the results match.
| Analysis | Supported in R package survival |
Supported in SAS PROC PHREG |
Results Match |
|---|---|---|---|
| Cause-specific hazard ratio estimates | Yes: with coxph() |
Yes | Yes |
| Stratified cause-specific hazard ratio estimates | Yes: with the stratification variable x specified as strata(factor(x)) on the right-hand side of the input formula |
Yes: with strata statement |
Yes |
| Variance estimates for the parameter estimates with robust sandwich estimator | Yes: default (robust = TRUE) |
Yes: with covsandwich or covs option in proc phreg statement |
Yes |
| Confidence intervals for hazard ratio estimates | Yes: Wald’s method by default | Yes: Wald’s method by default | Yes |
| Estimating cause specific hazard for multiple events | Yes | Yes | Depends (see note below) |
Additional details for using survival in R are given here and for SAS PROC PHREG here .
Estimating cause specific hazard ratios for multiple events
R and SAS have different approach when it comes to estimating the hazard ratios for multiple events. Results for the hazard ratio estimates are the same between the two; what is different is the global hypothesis:
The global hypothesis per
coxph()in this case is “There is no difference in the hazards of experiencing any of the events.”In
PROC PHREG, one syntax allows the hazard ratio estimates to be generated for all events. However, there is no corresponding global hypothesis as incoxph()in R. In SAS, there are only individual global hypotheses, one for each event. In addition, currently, when this syntax is used in SAS, stratified analysis cannot be implemented.
Summary
Most of the functionality of
survival::coxph()andproc phregalso apply to estimating cause-specific hazards in competing risks settings.Due to the different internal numerical estimation methods of R and SAS, results only match up to the 4th decimal places. However, overall consistency can be established between the two for estimating and testing cause-specific hazard ratio using Cox’s PH model.
Session Info