A risk-adjusted tool based on three preoperative variables from the National Surgical Quality Improvement Program had a high rate of efficacy in predicting inpatient mortality, which suggests it may be useful in resource-limited settings such as small rural hospitals or low- and middle-income countries.
"By offering a simplified risk-adjustment tool, we can compare surgical outcomes among hospitals on a global scale, regardless of the spectrum of surgical procedures offered or hospital resources," Jamie E. Anderson and associates in the department of surgery, University of California, San Diego, wrote in Archives of Surgery.
"Although participation in programs such as the NSQIP offers administrative support and comparison of outcomes among participating hospitals, the low-cost options reported can expand the number of hospitals that participate in risk-adjustment outcomes analysis and quality improvement programs," they added.
The American College of Surgeons’ NSQIP risk-adjusted tool uses more than 130 variables plus a 30-day patient follow-up, and thus is not affordable for use in settings where resources are limited, including small, rural hospitals, the authors pointed out (Arch. Surg. 2012;147:798-803).
Using data from more than 600,000 patients in the 2005-2009 NSQIP database, they developed different models to predict inpatient mortality and validated the models based on data on 239 patients from a 110-bed hospital in California with a level IV trauma center. They calculated that the "area under the receiver operator characteristic curve (AUROC)" for each model as a measure of how well the model separated the two groups of interest (survivors vs. nonsurvivors) with a value of 1.0 (or 100%) would mean that the model was able to completely separate the two groups.
The model using three preoperative NSQIP variables – age, American Society of Anesthesiologists (ASA) physical status classification, and functional status – had AUROC values that were more than 0.90, or more than 90%, which was similar to the value achieved for the model that used four or six variables. The model that used 66 variables was about the same as the value achieved with the model that used 4 variables (about 91%).
Considering that an AUROC value of 0.5 indicates that the model cannot distinguish between two groups any better than chance and that an AUROC value of 1.0 indicates that the model completely discriminates between the two groups, the authors said that an AUROC value of greater than 90% is substantial. Therefore, based on their results, "3 or 4 variables may be sufficient for adequate risk adjustment to measure surgical outcomes," they added.
"Future risk-adjustment tools [should] be based on 6 or fewer variables to allow for surgical outcomes to be measured and compared within and among hospitals in resource-limited settings," they concluded.
The authors had no disclosures.