Can prolonged intensive care therapy be predicted in cardiac surgical patients?

The commonly used severity scores for critically ill patients are poorly related to postoperative mortality and morbidity in patients undergoing cardiac surgery. Specific scoring systems have been developed for cardiac surgical patients. Among these, the modified Parsonnet model and the Society of Thoracic Surgeons model are the most reliable models for risk stratification in cardiac surgery. Prediction of clinical outcome is possible in the vast majority of cardiac surgical patients, but it can vary considerably among individual patients. Furthermore, interinstitutional and surgical differences such as surgeon experience, case mix, risk adjustment, quality of anesthesia, and postoperative management, as well as environmental factors (well-staffed stepdown units, fast-track policy, early withholding or withdrawing therapy) may influence patient outcome and length of intensive care unit stay. Despite increased preoperative comorbidity partially related to increasing age, the mean length of intensive care unit stay is decreasing owing to gains in expertise, new technologies, and improved quality of care. Therefore, the weight of different risk variables is likely to change over time and must be adapted in a regular manner. Finally, any risk stratification system must be simple, accurate, adaptable, verifiable, and inexpensive.