Although survival is traditionally modeled using Cox proportional hazards modeling, this approach may be inappropriate in sepsis, in which the proportional hazards assumption does not hold. Newer, more flexible models, such as Gray's model, may be more appropriate.