Morbidity Prediction Using Pre- and Intraoperative Data

Abstract
Preoperative and intraoperative information about patient status was collected for 66 general surgical patients'' receiving general anesthesia who had independent objective measurements of postoperative morbidity. Data included information collected during preoperative patients'' visits, such as age, blood pressure and ASA [American Society of Anesthesiologists] physical status scores. Intraoperative information was of 2 types: manually recorded descriptors of the operative course, such as blood loss and duration of operation and heart rate and blood pressure recorded every 1 or 2 min by an automatic patient-monitoring system. The postoperative outcome was assessed by each patient''s senior nurse, who assigned a weighted morbidity score to each of 6 organ systems during a structured interview on the 7th postoperative day. No hospital death occurred in the series. A third of the patients were scored as having experienced some postoperative morbidity. Six of the 64 manually collected preoperative descriptors were significantly associated with positive morbidity scores. These variables related to preoperative hypertension and duration of operation. Of a series of 190 statistical descriptors of the anesthetic course, 50 were associated with morbid outcomes. Statistics that described the variability of intraoperative cardiovascular status were most often found to have significant associations with postoperative morbidity. Using all information available at the end of anesthesia, a discriminant function that correctly predicted outcome for 83% of the patients was developed. Variability of intraoperative physiologic status was shown to be an important predictor of morbid outcome. Eventually, outcome predictions such as these may be used to assign different postoperative therapeutic regimens to different patients, depending on need and expected outcome.