Perioperative risk factors impact outcomes in emergency versus nonemergency surgery differently: Time to separate our national risk-adjustment models?

BACKGROUND

Emergency surgery (ES) is acknowledged to be riskier than nonemergency surgery (NES). Yet, little is known about the relative impact of individual perioperative risk factors on 30-day outcomes in ES versus NES.

METHODS

Using the 2011-2012 American College of Surgeons National Surgical Quality Improvement Program nationwide database, the 20 most common ES procedures were identified by Current Procedural Terminology code. Current Procedural Terminology codes with less than 300 observations in either ES or NES were excluded. Emergency surgery cases were defined as "emergent" and "nonelective" per American College of Surgeons National Surgical Quality Improvement Program criteria. Multivariable regression models were constructed to identify predictors of 30-day major morbidity and mortality (MMM) in each group, controlling for demographics, American Society of Anesthesiologists class, comorbidities, preoperative laboratory values, and procedure type. The odds ratios of independent predictors of MMM in ES and NES were derived then individually compared between the two groups; "effect modification" of procedure status (ES vs. NES) on each risk factor was subsequently calculated.

RESULTS

Of 986,034 patients, 170,131 met inclusion criteria (59,949 ES; 110,182 NES). The overall risk of MMM was significantly higher in ES versus NES (16.75% vs. 9.73%, p < 0.001; odds ratio, 1.18; 95% confidence interval, 1.12-1.24; p < 0.001). Of 40 ES- and 38 NES-identified independent risk factors, preoperative transfusion and white blood cell count of 4.5 × 10/μL or less carried significantly higher relative risk of MMM in ES versus NES. Conversely, ascites, preoperative anemia, and white blood cell count of 11 × 10/μL to 25 × 10/μL carried greater relative risk for MMM in NES. Four procedures (laparoscopic cholecystectomy, laparotomy, and umbilical and incisional herniorrhaphy) were inherently riskier in ES versus NES. The effect modification of ES (vs. NES) ranged between 0.68 (0.52-0.88) for ascites and 2.56 (1.67-3.92) for umbilical hernia repair.

CONCLUSIONS

Perioperative risk factors and procedure type impact postoperative morbidity and mortality differently in ES versus NES. Instead of using the same risk-adjustment model for both ES and NES, as currently practiced, our findings strongly suggest the need to benchmark emergent and elective surgeries separately.

LEVEL OF EVIDENCE

Prognostic/epidemiologic study, level III.