Can we use plasma hyperosmolality as a predictor of mortality for ST-segment elevation myocardial infarction?