Background: We recently modified our prediction scoring system "Estimation of Physiologic Ability and Surgical Stress" and have designated the current version mE-PASS. This scoring system has been designed to obtain predicted postoperative mortality rates before surgery and this study was performed to assess its usefulness in elective surgery for gastric carcinoma.
Methods: We investigated seven variables for mE-PASS and evaluated the postoperative course in 3,449 patients who underwent elective surgery for gastric carcinoma in Japan between August 20, 1987 and April 9, 2007, in order to quantify the predicted in-hospital mortality rates (R). The calibration and discrimination power of R were assessed using the Hosmer-Lemeshow test and the area under the receiver operating characteristic curve (AUC), respectively. The ratios of observed-to-estimated mortality rates (OE ratios) were quantified as a measure of quality.
Results: The overall postoperative morbidity and mortality rates were 19.0 and 2.0%, respectively. R demonstrated good power in calibration (χ(2) value, 12.5; df 8; P = 0.89) as well as discrimination (AUC, 95% confidence intervals: 0.80, 0.75-0.85). The OE ratios between hospitals ranged from 0.44 to 1.8. Overall, the OE ratios seemed to improve with time (OE ratio, 95% confidence intervals: 1.3, 0.73-2.4 for the early period between 1987 and 2000; 1.0, 0.59-1.7 for the middle period between 2001 and 2004; and 0.65, 0.36-1.2 for the late period between 2005 and 2007).
Conclusion: Based on these findings, mE-PASS might be useful for medical decision-making and for assessing the quality of care in elective surgery for gastric carcinoma.