Elective versus emergency surgery for ulcerative colitis: a National Surgical Quality Improvement Program analysis

The American Journal of Surgery (2013) 205, 333-338 Midwest Surgical Association Elective versus emergency surgery for ulcerative colitis: a National Surgical Quality Improvement Program analysis Supriya S. Patel, M.D. a, Madhukar S. Patel, M.D., M.B.A., Sc.M. b, Melanie Goldfarb, M.D. a, Adrian Ortega, M.D. a, Glenn T. Ault, M.D. a, Andreas M. Kaiser, M.D. a, Anthony J. Senagore, M.D., M.B.A. a, * a Division of Colorectal Surgery, Keck School of Medicine, University of Southern California, 1441 Eastlake Avenue, Suite 7418, Los Angeles, CA 90033, USA; b Department of Surgery, Massachusetts General Hospital, Boston, MA, USA KEYWORDS: Ulcerative colitis; Emergency; Elective; Surgery; Medical therapy Abstract BACKGROUND: It is unclear whether advances in the medical management of ulcerative colitis (UC) have altered outcomes for medically intractable disease. Therefore, it is essential to understand the current impact of elective versus emergency surgery for UC. METHODS: The American College of Surgeons National Surgical Quality Improvement Program database was used to compare outcomes for elective versus emergency UC surgery between 2005 and 2010. RESULTS: Four thousand nine hundred sixty-two patients were eligible for study (94% elective and 6% emergent). Emergency surgery patients were significantly older and frequently underwent open surgery. Emergency cases were associated with a higher frequency of cardiac, pulmonary, and renal comorbidities; postoperative complications; longer hospital stays; and higher rates of return to the operating room. CONCLUSIONS: In the era of advanced UC medical therapy, the need for emergency surgery still exists and is associated with substantial morbidity and mortality. Data are needed to determine if earlier selection of surgery would be beneficial. Ó 2013 Elsevier Inc. All rights reserved. The mainstay of treatment for ulcerative colitis (UC) in the early stages remains medical management. The recent access to the new biologic immunosuppressive agents The American College of Surgeons National Surgical Quality Improvement Program and the hospitals participating in the ACS NSQIP are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors. The authors declare no conflicts of interest. * Corresponding author. Tel.: 11-323-865-3690; fax: 11-323-865-3671. E-mail address: anthony.senagore@med.usc.edu Manuscript received July 24, 2012; revised manuscript September 17, 2012 raises concern regarding the efficacy of these expensive therapies to safely avoid surgery versus simply delaying surgical intervention because of the use of multiple different drugs and cycles of treatment. 1 Recent evidence still suggests that approximately one quarter of patients will require a colectomy during the course of their disease. 2 Indications for surgery include medical intractability, bleeding, perforation, obstruction, cancer, and megacolon. Of patients who are medically managed for acute colitis, more than 50% will require a colectomy. 3,4 The acute physiologic status is a major determinant of the specific surgical options (ie, total colectomy/ileostomy vs ileoanal pouch) and the subsequent outcomes achieved. 5 Ideally, preoperative optimization and elective resection are desired. 0002-9610/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjsurg.2012.10.014

334 The American Journal of Surgery, Vol 205, No 3, March 2013 The goal of this study was to compare the current outcomes of patients undergoing elective versus emergency surgery for UC using the standardized assessment offered by the National Surgical Quality Improvement Program (NSQIP). As medical therapies continue to advance, it is important to define the current burden of elective and emergency surgical care. Methods Data source The American College of Surgeons NSQIP is a wellvalidated dataset enabling risk-adjusted comparisons of short-term surgical outcomes. 6,7 For this study, the NSQIP participant user files from 2005 to 2010 were compiled. All patients undergoing surgery for UC were identified via postoperative International Classification of Diseases, Ninth Revision edition codes (556 556.9). Patients who received preoperative chemotherapy or radiation or who presented with a preoperative diagnosis of disseminated cancer were excluded from analysis. Patients were then divided into elective and emergency surgery groups based on whether or not their case was classified as an emergency. By definition, emergency status was assigned by the surgeon or anesthesiologist and resulted in a surgical procedure within 12 hours of admission. Variables The following NSQIP-defined patient demographics were compared between the 2 groups: age (mean), sex, smoking history, alcohol use, ventilator dependence, chronic obstructive pulmonary disease history, preoperative pneumonia, preoperative congestive heart failure, myocardial infarction history, previous percutaneous coronary intervention, previous cardiac surgery, angina history, preoperative hypertension, preoperative renal failure, preoperative dialysis, impaired sensorium, steroid use, weight loss (.10% loss in body weight in the 6 months preceding surgery), preoperative transfusion (of.4 U whole blood/packed red blood cells within the 72 hours preceding surgery), prior operation (within 30 days), and American Society of Anesthesiology class. The mean body mass index was calculated using height and weight data. The following variables were reclassified based on their absence or presence: diabetes (no vs oral hypoglemic/insulin dependence), dyspnea (no vs moderate exertion/dyspnea at rest), and preoperative sepsis (none vs systemic inflammatory response syndrome/sepsis/severe sepsis/septic shock). Operative procedures performed were identified using Current Procedural Terminology (CPT) codes. Outcome measures included mean operative time, mean length of hospital stay, mean length of surgical stay (defined as the hospital stay starting from the date of surgery), postoperative morbidities, and 30-day mortalities. System-specific postoperative morbidities were reclassified as follows: respiratory complications (eg, pneumonia, unplanned intubation, pulmonary embolus, and failure to wean from ventilator for.48 h), renal complications (eg, progressive renal insufficiency and acute renal failure), cardiac complications (eg, cardiac arrest and myocardial infarction), hematologic complications (eg, transfusion requirement, deep venous thrombosis, and thrombophlebitis), infectious complications (eg, sepsis, septic shock, and urinary tract infection), and wound complications (eg, superficial/ deep/organ space surgical site infection, open wound, and wound dehiscence). 8 A separate category entitled any major complication was created and defined as positive by the presence of any system-specific (eg, respiratory, renal, cardiac, hematologic, infectious, and wound) complication. Statistical analysis Categoric variables were compared using Pearson chisquare or Fisher exact tests. Continuous variables were compared with independent sample t tests or analysis of variance. Univariate predictors of mortality were identified using a P value of,.1; these variables were entered into a logistic regression model to identify multivariate predictors of mortality. The concordance index (or c-statistic) and Hosmer-Lemeshow goodness-of-fit tests were used to confirm the validity of the multivariate model. Statistical analyses were performed using Statistical Package for Social Sciences software (version 18.0; SPSS, Chicago, IL). P values,.05 were considered statistically significant. Results Demographics A total of 4,962 patients meeting the inclusion criteria were identified during the study period (Table 1). Of these patients, 4,664 (94%) underwent elective resection, whereas 298 (6%) underwent emergency surgery. Patients in the elective surgery group were younger (44 vs 53 years, P,.001) and more likely to be male (57% vs 51%, P 5.04). Patients in the emergency surgery group presented with a higher rate of pulmonary (32% vs 6%, P,.001), cardiac (43% vs 21%, P,.001), and renal comorbidities (10% vs.2%, P,.001). No significant difference in preoperative alcohol consumption or steroid use was observed between the 2 groups. Operative characteristics Patients undergoing emergency surgery had significantly shorter anesthesia (216 vs 283 minutes, P,.001) and operative (154 vs 215 minutes, P,.001) times. The most common procedure performed in the emergency surgery group was an open total abdominal colectomy without proctectomy with ileostomy or ileoproctostomy (43%, CPT 44150), whereas the most common procedure performed in the elective surgery group was laparoscopic total abdominal colectomy with proctectomy and ileoanal anastomosis (12%, CPT 44211).

S.S. Patel et al. UC: elective versus emergency surgery 335 Table 1 Preoperative demographics of the study cohort (N 5 4,962) Elective surgery (n 5 4,664) Emergency surgery (n 5 298) P value Age (mean) 43.95 6 15.98 53.10 6 19.73,.001 Male (%) 56.7 50.5.038 Comorbidities BMI (mean) 25.94 6 6.43 24.44 6 9.03.005 DM (%) 7.5 13.4,.001 Smoking history (%) 8 15.1,.001 Alcohol use (%) 1.7 2.7.207 Ventilator dependence (%).2 14.8,.001 Dyspnea (%) 4.9 19.5,.001 COPD (%) 1.5 12.1,.001 Pneumonia (%).3 5,.001 CHF (%).2 5,.001 MI (%).2 1.3.004 PCI (%) 2.4 5.006 Cardiac surgery (%) 1.9 5.4,.001 Angina (%).2 1.3.003 HTN (%) 19.8 41.6,.001 Renal failure (%).1 6.4,.001 Dialysis (%).1 5,.001 Impaired sensorium (%).1 12.4,.001 Steroid use (%) 39.9 40.3.889 Weight loss.10% (%) 9.5 18.1,.001 Transfusion (.4 U preoperatively) (%).7 3.4,.001 Sepsis (%) 5.1 66.6,.001 Prior operation within 30 days (%).7 12.7,.001 ASA class I (%) 3.3 1.031 ASA class II (%) 67.1 25.3,.001 ASA class III (%) 28.2 34.7.016 ASA class IV (%) 1.4 34.3,.001 ASA class V (%) 0 4.7,.001 Pulmonary comorbidities (%) 5.9 31.5,.001 Cardiac comorbidities (%) 21.1 43.3,.001 Renal comorbidities (%).2 10.1,.001 ASA 5 American Society of Anesthesiology; BMI 5 body mass index; CHF5congestive heart failure; COPD5chronic obstructive pulmonary disease; DM 5 diabetes mellitus; HTN5hypertension; MI 5 myocardial infarction; PCI 5 percutaneous coronary intervention. Outcomes Postoperative outcomes were compared between the 2 cohorts (Table 2). Patients in the emergency surgery group had significantly longer mean lengths of hospital (21 vs 9 days, P,.001) and surgical (18 vs 8 days, P,.001) stays. A higher rate of surgical site infection/wound (24% vs 18%, P 5.005), respiratory (31% vs 4%, P,.001), renal (6% vs 1%, P,.001), cardiac (5% vs 0%, P,.001), hematologic (17% vs 5%, P,.001), and infectious complications (24% vs 11%, P,.001) was observed in the emergency surgery cohort. Patients in the emergency surgery group also had higher frequencies of return to the operating room (21% vs 7%, P,.001) and significantly increased 30-day mortalities (13% vs 1%, P,.001). Univariate and multivariate analysis Univariate and multivariate analysis was performed to identify independent predictors of increased mortality. On univariate analysis, patient age; a history of diabetes; smoking or alcohol use; preoperative impaired sensorium; weight loss; sepsis; prior operation within 30 days; the presence of pulmonary, cardiac, or renal comorbidities; and emergency status were each associated with increased mortality. These variables were entered into multivariate analysis, and the following factors were identified as independent predictors of 30-day mortality: Age (odds ratio [OR] 5 1.06; 95% confidence interval [CI], 1.04 1.09) preoperative sepsis (OR 5 4.5; 95% CI, 2.0 9.9), preoperative pulmonary comorbidities (OR 5 2.7; 95% CI, 1.4 5.1), and emergency status (OR 5 2.6; 95% CI, 1.2 5.7). Comments The optimal timing of surgical intervention in patients with UC should be determined based on a clear and comprehensive understanding of the short- and long-term outcomes for the patient including quality of life. The use of the newer biologic agents for rescue treatment in the

336 The American Journal of Surgery, Vol 205, No 3, March 2013 Table 2 Comparison of postoperative outcomes Elective surgery (n 5 4,664) Emergency surgery (n 5 298) P value Length of hospital stay 8.86 6 8.92 20.75 6 17.82,.001 Length of surgical stay 8.17 6 8.07 18.22 6 15.75,.001 Superficial SSI (%) 7.7 8.1.834 Open wound/wound infection (%) 1.8 7.4,.001 Deep incisional SSI (%) 2.1 1.208 Organ space SSI (%) 6.7 10.4.014 Wound disruption (%) 1.5 3.050 Pneumonia (%) 1.8 9.7,.001 Unplanned intubation (%) 1.1 6,.001 PE (%).8 2.3.015 Failure to wean from vent for.48 h (%) 1.4 26.5,.001 Progressive renal insufficiency (%).8 2.050 Acute renal failure (%).3 4,.001 UTI (%) 4.5 6.4.130 Cardiac arrest requiring CPR (%).2 3.4,.001 MI (%).2 2,.001 Bleeding requiring transfusion (%) 2.9 11.7,.001 DVT, thrombophlebitis (%) 2.4 5.7.001 Sepsis (%) 6.5 12.8,.001 Septic shock (%) 1.4 7,.001 Return to OR (%) 7.4 20.5,.001 30-day mortality (%).6 13.1,.001 Any SSI or wound disruption (%) 17.7 24.2.005 Any respiratory complication (%) 3.8 31.2,.001 Any renal complication (%) 1.1 6,.001 Any cardiac complication (%).3 5.4,.001 Any hematologic complication (%) 5.3 17.1,.001 Any infectious complication (%) 10.9 23.5,.001 Any major complication (%) 27.3 (1,273/4,664) 57.7 (172/298),.001 CPR 5 cardiopulmonary resuscitation; DVT 5 deep vein thrombosis; MI 5 myocardial infarction; OR 5 operating room; PE 5 pulmonary embolus; SSI 5 surgical site infection; UTI 5 urinary tract infection. setting of acute UC flares has complicated the decisionmaking algorithm. Delays in recognition of medical treatment failures can result in the need for emergent surgical intervention. In this study, we showed that emergency surgery for UC is associated with a significant risk of postoperative morbidity and mortality. In fact, emergency status itself is an independent risk factor for increased 30-day mortality. Our findings are consistent with previous studies that have shown worse outcomes after emergency surgery for UC. Using administrative databases, de Silva et al 9 observed that emergency colectomy for UC was associated with an increased risk of overall and infectious postoperative complications. Further examination of the emergency surgery cohort showed that a delay in surgical intervention of.14 days from the time of admission resulted in an increased frequency of postoperative complications. In a single-center experience involving 72 patients undergoing emergency surgery for UC, Pal et al 4 observed a significant mortality difference between patients who were operated on before versus after 5 days of intensive medical therapy (0% vs 24%, respectively; P 5.03). Similarly, Randall et al 10 studied 80 patients with severe UC who were treated with steroids and/or cyclosporine before surgery and showed that an increased length of in-hospital medical treatment before surgical intervention was associated with an increased risk of postoperative complications. Although our assessment is limited by a lack of access data on the duration of medical therapy before surgical therapy, we were able to show a significantly increased incidence of complications in all systems (ie, surgical site infection/wound, respiratory, renal, cardiac, hematologic, and infectious) after emergency surgery. In addition, patients undergoing emergency surgery had a higher frequency of return to the operating room and a 30-day mortality rate of 13% versus 1% (P,.001) when compared with patients undergoing elective surgery. Consistent with previously published literature, we observed that the use of laparoscopy in the emergency setting was limited when compared with the elective setting, which may contribute to the shortand long-term complications observed in the acute setting. 5,11 However, we did not include the surgical approach (ie, laparoscopic vs open) in our univariate and multivariate models and are hence unable to comment on whether the surgical approach is an independent predictor of morbidity and/or mortality.

S.S. Patel et al. UC: elective versus emergency surgery 337 Our study was limited by its retrospective nature, the constraints of the dataset, and the lack of any high-quality information regarding outcomes on risk-adjusted medically managed patients. Unfortunately, all patients admitted with colitis who received surgery more than 12 hours after being admitted were categorized in the elective group based on the NSQIP definition of emergency status. Although we were able to provide data on postoperative outcomes, the specific indications for surgical intervention as well as the duration and types of same admission and preoperative medical therapy were not available for analysis. Additionally, we are unable to comment on the long-term outcomes of patients undergoing emergency surgery and the results of subsequent restorative procedures. Future studies comparing the surgical outcomes of patients before and after the advent of biological therapies, such as infliximab, are merited. Nonetheless, on the basis of our findings, it is clear that the immediate postoperative course of patients undergoing emergency surgery is associated with a significantly higher risk of morbidity and mortality than patients undergoing elective resections. Despite advances in medical therapy, a significant number of patients with UC continue to present in the emergency setting and require emergency surgical intervention. It is difficult to quantify what percentage of these patients represent a population of escalating degrees of failed medical therapy that might have been recognized earlier. In turn, emergency surgery is associated with an increased risk of postoperative complications and mortality. This analysis does suggest that a more global dataset, which includes the impact of antecedent medical therapy as well as the relative functional results after restorative proctectomy, could provide a value-based analysis and risk-adjusted protocol of therapy. Early recognition of medical treatment intractability and the impact this may have on long-term quality of life may result in earlier surgical intervention and more favorable outcomes for certain populations. Collaboration between the gastroenterologist and surgeon is critical in determining the optimal timing of surgical intervention. References 1. Biondi A, Zoccali M, Costa S, et al. Surgical treatment of ulcerative colitis in the biologic therapy era. World J Gastroenterol 2012;18: 1861 70. 2. Metcalf AM. Elective and emergent operative management of ulcerative colitis. Surg Clin North Am 2007;87:633 41. 3. Bojic D, Radojicic Z, Nedeljkovic-Protic M, et al. Long-term outcome after admission for acute severe ulcerative colitis in Oxford: the 1992-1993 cohort. Inflamm Bowel Dis 2009;15:823 8. 4. Pal S, Sahni P, Pande GK, et al. Outcome following emergency surgery for refractory severe ulcerative colitis in a tertiary care centre in India. BMC Gastroenterol 2005;5:39. 5. Cima RR. Timing and indications for colectomy in chronic ulcerative colitis: surgical consideration. Dig Dis 2010;28:501 7. 6. Fink AS, Campbell Jr DA, Mentzer Jr RM, et al. The National Surgical Quality Improvement Program in non-veterans administration hospitals: initial demonstration of feasibility. Ann Surg 2002;236:344 53. 7. Khuri SF. The NSQIP: a new frontier in surgery. Surgery 2005;138: 837 43. 8. Alavi K, Sturrock PR, Sweeney WB, et al. A simple risk score for predicting surgical site infections in inflammatory bowel disease. Dis Colon Rectum 2010;53:1480 6. 9. de Silva S, Ma C, Proulx MC, et al. Postoperative complications and mortality following colectomy for ulcerative colitis. Clin Gastroenterol Hepatol 2011;9:972 80. 10. Randall J, Singh B, Warren BF, et al. Delayed surgery for acute severe colitis is associated with increased risk of postoperative complications. Br J Surg 2010;97:404 9. 11. Fowkes L, Krishna K, Menon A, et al. Laparoscopic emergency and elective surgery for ulcerative colitis. Colorectal Dis 2008;10:373 8. Discussion Dr Conor P. Delaney (Cleveland, OH): 1. Did you notice any difference in outcomes between the earlier years and the more recent years? I think many of us who do this type of surgery feel that with increasing use of biologic therapy, patients are actually getting to surgery sicker now than 2 years ago. 2. Because it has taken 2 decades and laparoscopic colectomy is still less prevalent than open colectomy, I think it is fascinating that the commonest elective case is laparoscopic total colectomy and proctectomy, although this appeared to only account for 12% of cases. Did this include patients having ileo-anal anastomosis? 3. Many patients come in with toxicity or acute colitis and are in the hospital longer than 12 hours before a decision is made to perform surgery. Presumably these had to be included with the elective group? 4. Finally, from the NSQIP dataset, you can get albumin and white blood cell count. Can these be used as markers of toxicity to further evaluate your dataset? Supriya S. Patel, M.D. (Los Angeles, CA): With respect to your first question, we compiled the NSQIP participation user files from 2005 to 2010 and are unable to comment on any differences between the individual years. It would be interesting to compare outcomes before and after these dates as more NSQIP participation user files become available. With respect to your second question, the 12% of patients in the elective cohort who underwent laparoscopic colectomies did include patients with an ileal pouch anal anastomosis; furthermore, there was an additional 12% of patients in the elective group who underwent laparoscopic colectomies, but these patients received a stoma rather than ileal pouch anal anastomosis. Unfortunately, we did have to include those patients who presented with acute colitis and received surgical intervention after 12 hours in the elective group. By NSQIP definitions, emergency status is assigned by the surgeon or anesthesiologist and is only given to those patients who receive a surgical intervention within 12 hours of admission. With respect to your last question, the preoperative white blood cell count and albumin are both accessible with the NSQIP dataset, and it would be interesting to see whether these laboratory values help to identify patients who are at an increased risk for treatment failure. We did not compare these preoperative laboratory values between our elective and emergency surgery groups, but previous

338 The American Journal of Surgery, Vol 205, No 3, March 2013 studies have suggested that an increased white blood count is associated with disease activity and that hypoalbuminemia is an indicator of an increased risk for colectomy in patients with ulcerative colitis. All these associations were drawn based on datasets other than NSQIP and because these parameters are readily available from the NSQIP, as you point out, it would be interesting to see whether we can use NSQIP to validate these observations. William C. Cirocco, M.D. (Grosse Pointe, MI): Based on your findings, are we making a case, maybe, for doing less at that initial emergent operation? You know, if the patient is stable, I would favor doing a pouch at the time. That is the best time I believe unless you are going to just divert and come back and leave the rectum untouched, but I was wondering where you were leading us with these results; if that is something that you had considered? Dr Patel: I think that especially in the emergency setting there is a strong indication based on our observations for staged procedures, but the point that we wanted to bring across with this study is that given all the new agents and advancements in medical therapies that are available, the discussion as to when to intervene needs to be had earlier.