The primary assumption of the hazard ratio is the proportional hazards assumption. This means that the ratio of the hazard rates between the two groups remains constant over time. If this assumption is violated, the HR may not be a valid measure of the effect. Other assumptions include the independence of survival times and the absence of time-dependent covariates.
