Testing the proportional hazards assumption in case-cohort analysis

Table 1 Proportion of simulations (out of 1000) with P-values < 0.05 for a single variable

Variable type	Shape of hazards functions & proportionality	Corr with time	Corr with rank	Corr with KM estimate
Binary	Non-monotonic proportional	0.059	0.056	0.054
	Non-monotonic non-proportional	0.874	0.877	0.857
	Increasing non-proportional	0.445	0.409	0.407
	Decreasing non-proportional	0.805	0.931	0.928
	One constant one increasing non-proportional	0.657	0.666	0.646
Continuous	Non-monotonic proportional	0.052	0.059	0.060
	Increasing proportional	0.041	0.045	0.042
	Non-monotonic non-proportional	0.997	0.996	0.995

Note: Abbreviation: Corr correlation, KM Kaplan-Meier Survival Curve.
The full-cohort sample size for all models was 2000; group 1 and group 2 had equal sample size; and the subcohort sample size was 500 using simple random sampling.

ISSN: 1471-2288