Outcome prediction after moderate and severe traumatic brain injury: Why two models if we could do with one?