Can standardized mortality ratio be used to compare quality of intensive care unit performance?