Predicting Postoperative Adverse Events of Common Surgical Procedures in the Medicare Population

摘要

The purpose of the report was to develop severity of illness models using chart-abstracted data from hospital admissions to predict adverse outcomes following four procedures performed frequently in the Medicare population: coronary artery bypass graft, coronary angioplasty, cholecystectomy, and prostatectomy. These models are useful for evaluating quality of care. Combinations of clinical variables indicating severity of both the index disease and co-morbid conditions were strongly associated with, but only weakly predictive of, the occurrence of postoperative adverse events within each surgical procedure. The models developed from the study are useful in adjusting adverse event rates and in measuring quality of care among surgical patients. An additional finding was that there was substantial disagreement between the chart-abstracted data and Health Care Financing Administration (HCFA) claims data regarding the presence of adverse outcomes.