Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome

Table 5 Comparison of four measures of the treatment effect in a trial

Criterion	Measure
	log HR	Median^a	RMST^a	ADS^a
1. Is easily interpreted	no	yes	yes	yes
2. Does not assume proportional hazards	no	yes	yes	yes
3. Reflects entire survival history	yes	no	yes	no
4. Is a measure of survival time	no	yes	yes	no
5. Can be used with all models	no	yes	yes	yes
6. Can be calculated in any dataset	yes	no	yes	yes
7. Does not require a time point to be specified	yes	yes	no	no
8. Does not change with extended follow-up	no	yes	yes	yes
9. Is routinely associated with a clinically meaningful time point	no	no	yes	yes

^aThe measure is the difference in the given statistic between trial arms. ADS = absolute difference in survival.

ISSN: 1471-2288