The American Journal of Surgery (2015) 210, 852-858
Clinical Science
Predictive factors associated with adverse events in patients with toxic colitis: an analysis of the National Surgical Quality Improvement Project Anand Dayama, M.D.a,*, Dordaneh Sugano, B.S.b, Melvin E. Stone, M.D.b, John McNelis, M.D.b a
Department of Surgery, San Joaquin General Hospital, University of California Davis, French Camp, CA; bDepartment of Surgery, Jacobi Medical Center, Albert Einstein College of Medicine, 1400 Pelham Parkway South, Bronx, NY 10461, USA
KEYWORDS: Toxic colitis; Colectomy; Outcomes; NSQIP; Risk factors
Abstract BACKGROUND: Toxic colitis (TC) is a potentially lethal complication of inflammatory bowel disease and infectious colitis. METHODS: The National Surgical Quality Improvement Project dataset was queried to identify emergent colectomy for a primary diagnosis of TC using International Classification of Diseases, Ninth Revision codes. The study endpoints were 30-day mortality and 30-day morbidity. We performed multivariate logistic regression analyses to study factors associated with 30-day mortality. RESULTS: Of the 189 TC patients, mortality for colectomy was 26.5%, 42.3% experienced pulmonary complications, 20.6% experienced sepsis complications, 17.5% experienced cardiovascular complications, 12.7% experienced renal complications, and 14.8% experienced wound complications. On multivariate analysis, independent factors associated with mortality were age 70 to 80 years (odds ratio [OR] 3.5), age greater than 80 (OR 22.2), female sex (OR 4.1), uremia blood urea nitrogen greater than 40 (OR 4.1), coagulopathy international normalized ratio greater than 2 (OR 7.7), preoperative respiratory failure (OR 2.73), and preoperative steroid use (OR 3.9). CONCLUSIONS: Patients with TC are very ill. Poor outcome is associated with older age, female sex, preoperative azotemia, preoperative respiratory failure, and chronic steroid use. These factors will help acute care surgeons in preoperative risk assessment and could be an important addition to decision-making strategies. Ó 2015 Elsevier Inc. All rights reserved.
The authors declare no conflicts of interest. Presented at the Society of Critical Care Medicine’s 44th Critical Care Congress, January 17–21, 2015, Phoenix Convention Center, Phoenix, Arizona. * Corresponding author. Tel.: 11-404-817-1342; fax: 11-718-9185567. E-mail address: [email protected] Manuscript received December 2, 2014; revised manuscript March 24, 2015 0002-9610/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjsurg.2015.03.026
Toxic colitis is a potentially lethal complication of inflammatory bowel disease (IBD) and infectious colitis. It is defined as segmental or total dilatation of colon greater than 6 cm in the presence of acute colitis and sepsis.1 The incidence of toxic colitis is dependent on the underlying cause; for example, the lifelong incidence is 1% to 2.5% in ulcerative colitis patients2 and 3% to 5% in patients with Clostridium difficile.3,4 Although IBD is the most common etiology, almost any inflammatory condition of the colon can lead to toxic colitis: ischemic colitis, viral
A. Dayama et al.
Outcomes of colectomy in toxic colitis
colitis (such as CMV colitis, HIV/AIDS, Kaposi sarcoma), bacterial colitis (such as Salmonella, Shigella, Campylobacter), medication-induced colitis, and so on.5 Diagnosis of toxic colitis is made on the basis of a combination of clinical criteria (diarrhea, bloody diarrhea, obstipation, abdominal pain, tenderness, decreased bowel sounds, peritonitic signs, fever [.101.5 F], tachycardia [.120 beats/ minute], leukocytosis [.10.5 ! 103/mL], and anemia) and radiologic criteria (dilatation of colon .6 cm, diffuse colonic wall thickening, paracoloic fat stranding, perforation, and abscesses), along with any one of the following: dehydration, altered mental status, electrolyte abnormality, or hypotension.6 Colonoscopy is not usually performed because of high risk of perforation. Intensive medical management is generally the first line of treatment in toxic colitis. However, surgical salvage remains the cornerstone of treatment in medically refractory patients. This study aimed to assess 30-day surgical outcomes after colectomy for toxic colitis in the American College of Surgeons National Surgical Quality Improvement Project (ACS-NSQIP)da national datasetdand to evaluate factors independently associated with 30day mortality and complications.
Methods Data source The ACS-NSQIP is a risk-adjusted data collection mechanism that collects and analyzes clinical outcomes data. Participating hospitals use their collected data to develop quality initiatives that improve surgical care and to identify elements in provided health care that can be improved when compared with other institutions. The ACS-NSQIP collects data on a variety of clinical variables, including preoperative risk factors, intraoperative variables, and 30-day postoperative mortality and morbidity outcomes for patients undergoing major surgical procedures in both the inpatient and outpatient setting. A site’s surgical clinical nurse reviewer captures outcomes data using a variety of methods including medical chart abstraction.
Data selection Using the ACS-NSQIP database, we identified patients undergoing colectomies from 2005 to 2012 with a primary Table 1
853 diagnosis of toxic colitis using International Classification of Diseases, Ninth Revision (ICD-9) codes. The American Medical Association’s Current Procedural Terminology (CPT, Chicago, IL) codes that describe colectomies were used to identify the patients who had received the operation (eTable 1). Preoperative demographic data obtained included age, sex, race, history of smoking (within 1 year of surgery), and alcohol use (more than 2 drinks a day in the 2 weeks before surgery). Comorbidities measured included presence or absence of renal disease (dialysis dependence), coronary artery disease (angina within 30 days of surgery, myocardial infarction within 6 months of surgery, prior percutaneous coronary intervention, and prior cardiac surgery), congestive heart failure, hypertension, peripheral arterial disease requiring previous revascularization/amputation, rest pain in lower extremity, history of chronic obstructive pulmonary disease, neurologic event or disease (stroke with or without residual deficit, transient ischemic attack, and hemiplegia), diabetes mellitus, chronic corticosteroid use, weight loss (more than 10% in the 6 months before surgery), bleeding disorders, and presence of open wounds. Other factors considered were the American Society of Anesthesiologists’ (ASA) class admission status (from home vs facility), preoperative functional statusdability to perform activities of daily living in the 30 days before surgeryd(independent, partially dependent, totally dependent), dyspnea (none, moderate exertion, at rest), prior surgery within 30 days, do not resuscitate status before operation, disseminated cancer, preoperative sepsis, days from admission to operation, and body mass index. Preoperative laboratory variables analyzed included blood urea nitrogen, creatinine, albumin, bilirubin, hematocrit, platelet count, serum glutamic oxaloacetic transaminase, white blood cell count (WBC), partial thromboplastin time, and prothrombin time. Intraoperative data analyzed were intraoperative transfusion, type of operation (partial vs complete colectomy), and operative time.
Outcome The primary outcome was perioperative mortality and the secondary outcome was 30-day morbidity. We created composite endpoints to facilitate a better understanding of 30-day morbidity outcomes. These composite endpoints included the following: renal complications, pulmonary
Definitions of composite end points
1. 30-Day mortality: death within 30 days after surgery 2. Renal complications: progressive renal failure or acute renal failure 3. Pulmonary complications: pneumonia, unplanned intubation, or respiratory insufficiency requiring ventilation for greater than or equal to 48 hours postoperatively 4. Cardiovascular complications: pulmonary embolism, stroke/cerebrovascular event, cardiac arrest, myocardial infarction, or deep vein thrombosis/thrombophlebitis 5. Wound complications: superficial SSI, deep SSI, organ space SSI, and wound disruption SSI 5 surgical site infection.
854 Table 2
The American Journal of Surgery, Vol 210, No 5, November 2015 Preoperative characteristics of patient with toxic colitis
Characteristics
Total, n (%)
Cohort without mortality within 30 days, n (%)
Age Female Race: black BMI Diabetes Dyspnea At rest Moderate exertion History of hypertension Chronic steroid use Functional status before surgery Partially dependent Dependent Ventilator support (respiratory insufficiency) History of congestive heart failure History of MI within 6 months Weight loss in last 6 months Bleeding disorder American Society of Anesthesiologists physical status classification II and III IV and V Do not resuscitate status Preoperative sepsis Preoperative WBC Preoperative platelet Preoperative INR Preoperative BUN Preoperative creatinine Preoperative bilirubin Preoperative SGOT Days from admission to operation Number of intraop RBC transfusion Operative time Preoperative acute renal failure Preoperative dialysis Wound classification 1 (clean) 2 (clean/contaminated) 3 (contaminated) 4 (dirty/infected) Emergency cases
60.17 6 17.85 95 (50.3) 21 (11.1) 25.47 6 9.72 34 (18)
56.63 6 17.37 65 (46.8) 13 (9.4) 25.29 6 10.05 21 (15.1)
Cohort with mortality within 30 days, n (%) 69.08 6 15.88 30 (60) 8 (16) 25.82 6 8.89 13 (26)
33 27 111 55
(17.5) (14.3) (58.7) (29.1)
18 17 78 35
(12.9) (12.2) (56.1) (25.2)
15 10 33 20
(30) (20) (66) (40)
34 54 49 18 10 16 44
(18) (28.6) (25.9) (9.5) (5.3) (8.5) (23.3)
26 29 25 6 7 14 27
(18.7) (20.9) (18) (4.3) (5) (10.1) (19.4)
8 25 24 12 3 2 17
(16) (50) (48) (24) (6) (4) (34)
71 (37.6) 118 (62.4) 5 (2.6) 154 (81.5) 20.07 6 13.04 251.55 6 148.43 1.48 6 .60 34.65 6 24.74 2.05 6 1.90 .88 6 1.07 48.26 6 64.34 5 6 7.7 1.34 6. 1.86 140.38 6 58.33 14 (21.7) 14 (7.4)
65 (46.8) 74 (53.2) 2 (1.4) 107 (77) 18.45 6 12.97 262.68 6 159.86 1.40 6. 45 30.40 6 24.26 1.82 6 1.79 .84 6 1.08 39.45 6 51.71 4.75 6 7.79 1.18 6 1.52 145.63 6 61.98 25 (18) 8 (5.8)
6 (12) 44 (88) 3 (6) 47 (94) 24.55 6 12.28 192.65 6 136.37 1.71 6. 87 46.45 6 22.29 2.69 6 2.06 1.05 72.75 6 86.42 5.56 6 7.73 1.79 6 2.55 125.78 6 44.04 16 (32) 6 (12)
6 60 70 53 148
5 47 52 35 102
1 13 18 18 46
(3.2) (31.7) (37) (28) (78.3)
(3.6) (33.8) (37.4) (25.2) (73.4)
(2) (26) (36) (36) (92)
P value ,.01 .10 .20 .74 .08 ,.01
.22 .04 ,.01 ,.01 ,.01 .36 .18 .03 , .01
.07 ,.01 ,.01 ,.01 .01 ,.01 ,.01 .33 .01 .52 .11 .01 .03 .148 .46
,.01
Data are expressed as mean 6 standard deviation. BMI 5 body mass index; BUN 5 blood urea nitrogen; INR 5 international normalized ratio; MI 5 myocardial infarction; RBC 5 red blood cell; SGOT 5 serum glutamic oxaloacetic transaminase; WBC 5 white blood cell.
complications, sepsis-related complications, cardiovascular complications, and wound complications. Events included in these composite endpoints are listed in Table 1. Other outcomes analyzed included wound infection, pneumonia, reintubation, mechanical ventilation greater than 48 hours, pulmonary embolus, deep venous thrombosis, renal insufficiency, acute renal failure, cardiac arrest, myocardial infarction, transfusion greater than 4 units of packed red
blood cells within 72 hours, sepsis, septic shock, and return to the operating room.
Statistical analysis We described categorical data as absolute numbers and percent prevalence (%) in the study cohort and continuous
A. Dayama et al. Table 3
Outcomes of colectomy in toxic colitis
855
30-day outcomes of colectomy for toxic colitis
Mortality Renal complications Cardiovascular complications Wound complications Pulmonary complications Sepsis complications Pneumonia Unplanned intubation On ventilator for .48 hours Pulmonary embolism Cardiac arrest requiring CPR Myocardial infarction Deep vein thrombosis Sepsis Septic shock Superficial SSI Deep SSI Organ space SSI Wound disruption Cerebrovascular accident Acute renal failure Progressive renal failure Length of stay Return to operating room
Total, n (%)
Cohort without mortality within 30 days, n (%)
Cohort with mortality within 30 days, n (%)
P value
189 24 (12.7) 33 (17.5) 28 (14.8) 80 (42.3) 39 (20.6) 30 (15.9) 19 (10.1) 66 (34.9) 5 (2.6) 10 (5.3) 2 (1.1) 16 (8.5) 24 (12.7) 15 (7.9) 12 (6.3) 2 (1.1) 10 (5.3) 6 (3.2) 4 (2.1) 23 (12.2) 2 (1.1) 21.59 6 23.39 28 (14.8)
– 9 (6.5) 21 (15.1) 27 (19.4) 46 (33.1) 26 (18.7) 19 (13.7) 7 (5) 38 (27.3) 5 (3.6) 1 (.7) 0 14 (10.1) 22 (15.8) 4 (2.9) 11 (7.9) 2 (1.4) 10 (7.2) 6 (4.3) 3 (2.2) 9 (6.5) 0 23.90 6 26.16 19 (13.7)
50 (26.5) 15 (30.) 12 (24) 1 (2) 34 (68) 13 (26) 11 (22) 12 (24) 28 (56) 0 9 (18) 2 (4) 2 (4) 2 (4) 11 (22) 1 (2) 0 0 0 1 (2) 14 (28) 2 (4) 15.16 6 10.68 9 (18)
– ,.01 .15 ,.01 ,.01 .27 .16 ,.01 ,.01 .17 ,.01 .01 .18 .03 ,.01 .14 .39 .05 .13 .94 ,.01 .01 ,.01 .46
CPR 5 cardiopulmonary resuscitation; SSI 5 surgical site infection.
variables as means 6 standard deviation. Categorical variables were compared using chi-square test or Fisher’s exact test for discrete values. Independent 2-sample t tests were used for normally distributed continuous variables, and the Wilcoxon rank-sum test was used for non-normally distributed continuous and ordinal variables. Multivariable logistic regression modeling was used to assess the predictive factors for perioperative mortality while controlling for possible confounders. Adjusted multivariable models were created by all candidate covariates potentially associated with postoperative mortality on univariate analysis with P value less than or equal to .2. Candidate covariates were evaluated for inclusion in adjusted models using forward selection with P less than .05 as the entry criterion. The model selection was based on a stepwise procedure, which alternates between dropping the least significant variable from the model and then reconsidering all potential variables for reintroduction into the model until no more variables can be added. The discriminatory capacity was tested using the Harrell’s concordance index. Harrell’s C statistic was used to estimate the proportion of correct predictions. Harrell’s C index ranges from .5 (no discrimination) to 1 (perfect discrimination). The overall model fit was obtained using the C statistic, the Hosmer–Lemeshow goodness-of-fit test, and the Pearson chi-square statistics. We analyzed data using IBM SPSS, Version 21.0 (SPSS, Inc, Armonk, NY).
Results Patient characteristics From 2005 to 2012, a total of 189 patients underwent colectomy for toxic colitis. Of these, 148 colectomies were performed emergently. The patients’ mean age was 60.17 6 17.85 years, and 41.2% of the patients were between 18 and 60 years, 29.4% between 60 and 70 years, 16.6% between 70 and 80 years, and 12.8% of the patients were older than 80 years. About 50.3% of the patients were female and 11.1% black. Subgroup analysis comparisons based on 30day mortality of preoperative patient demographics and comorbidity data are shown in Table 2. Compared with patients in the 30-day survivor cohort, the 30-day mortality cohort had a greater prevalence of women (60% vs 46%), higher mean age (69.08 6 15.88 vs 56.63 6 17.37), higher prevalence of patients on chronic steroids (40% vs 25.2%), higher number of patients in ASA class IV and V (88% vs 53.2%), more patients presenting with preoperative sepsis (94% vs 77%), and had a higher mean leukocytosis count (24.55 6 12.28 vs 18.45 6 12.97). Mean time from admission to operation was 4.75 6 7.79 days for survivors as compared with 5.5 6 7.73 in mortality cohorts. This difference was not statistically significant.
856
The American Journal of Surgery, Vol 210, No 5, November 2015
Table 4 Multivariate model showing predictive factors for 30-day mortality following colectomy for toxic colitis with concordance index of 88% Variable
Adjusted odds ratio
95% CI Lower
Age group 18–60 Reference 60–70 2.329 .716 70–80 3.507 .957 .80 22.222 5.716 BUN ,40 Reference .40 4.099 1.626 ASA class II and III Reference IV and V 2.723 .802 INR ,2 Reference .2 7.738 1.909 Sex Male Reference Female 4.069 1.512 Respiratory insufficiency No Reference Yes 2.713 1.052 Preoperatively on chronic steroids No Reference Yes 3.943 1.554
Upper
P value
7.576 12.847 86.386
.000 .160 .058 .000
10.335
.003
9.251
.108
31.365
.004
10.948
.005
of steroids, uremia, coagulopathy, and respiratory insufficiency before surgery. Women undergoing colectomy for toxic colitis had 4.1 times higher odds for mortality as compared with men. Patients with preoperative uremia (blood urea nitrogen .40) also had a 4.1 odds ratio for mortality. Patients on chronic steroids had 3.9 times higher odds for mortality as compared with those not on steroids. Preoperative respiratory insufficiency carried 2.7 times higher odds of mortality. Compared with age less than 60, patients aged 60 to 70 years had 2.3 times higher odds of mortality, patients aged 70 to 80 years had 3.5 times higher odds, and patients above 80 years had 22.2 times higher odds of 30-day mortality. The models were robust and fit the observed data well as tested using the C statistic (C 5 .88) and the Hosmer–Lemeshow test (P 5 .457). The C statistic represents the areas under the receiver operating curve. In the Hosmer–Lemeshow test, P greater than .05 signifies that the model fits the observed data well, which was confirmed by the Pearson chi-square statistic (.9), indicating appropriate goodness-of-fit for the model.
Comments 7.001
.039
10.005
.004
ASA 5 American Society of Anesthesiologists; CI 5 confidence interval; BUN 5 blood urea nitrogen; INR 5 international normalized ratio.
Perioperative outcomes The 30-day mortality following colectomy for toxic colitis was 26.5%, 42.3% experienced pulmonary complications, 20.6% sepsis complications, 17.5% cardiovascular complications, 12.7% renal complications, and 14.8% wound complications. Comparisons of perioperative adverse events categorized by 30-day mortality cohort are shown in Table 3. Compared with patients in the 30-day survivor cohort, those in the 30-day mortality cohort had a greater incidence of prolonged mechanical ventilation (56% vs 27.3%), return to operating room (18% vs 13.7%), acute renal failure (28% vs 6.5%), and cardiac arrest requiring cardiopulmonary resuscitation (18% vs .7%). Total length of stay for survivors was 23.90 6 26.16 days.
Multivariate regression The independent variables associated with 30-day mortality at the P less than .2 level in univariate analysis were used to construct a multivariate model. Table 4 reports the results from the multivariate regression analysis. After adjusting, the factors independently associated with mortality were female sex, advanced age, preoperative chronic use
This study uses a national, multicenter, prospective clinical database to report contemporary surgical outcomes following colectomy for toxic colitis. In the analysis of this national dataset, we found that patients with toxic colitis are very ill irrespective of cause, with 81% patients experiencing preoperative sepsis and two thirds of patients in ASA class IV and V. Toxic colitis itself is associated with high postoperative morbidity and mortality. We also found that female sex, advancing age, preoperative chronic use of steroids, uremia, coagulopathy, and respiratory insufficiency before surgery are independently associated with postcolectomy 30-day mortality. Although there are numerous studies reporting outcomes after colectomies, this study is unique in that it reports outcomes of colectomy for toxic colitis. This is the largest report of patients undergoing surgery for toxic colitis. The high morbidity and mortality of toxic colitis requires early and intensive medical management with IV steroids, antibiotics, decompressive maneuvers, and other resuscitative measures to treat the underlying cause. If there is no sign of improvement within 7 days or if there are any signs of deterioration, urgent surgical intervention should be considered.7–9 Goligher et al10 reported a reduction in mortality from 20% to 7% in severe episodes of ulcerative colitis because of early colectomy. In our analysis, mean time from admission to operation was 4.75 6 7.79 days for survivors as compared with 5.5 6 7.73 days in mortality cohorts; however, this difference was not statistically significant. Specific clinical indications for colectomy were not available for analysis because of dataset limitations. Deciding the timing of a colectomy for fulminant colitis remains difficult. The mortality rate of 26.5% in our analysis is consistent with the mortality rate of 20% to 57% reported
A. Dayama et al.
Outcomes of colectomy in toxic colitis
in the literature.3,6,11,12 Greenstein et al11 reported 13 deaths out of 59 patients (22%) who underwent surgery for fulminant colitis secondary to IBD, whereas Dallal et al3 reported 57% deaths following colectomy for fulminant colitis secondary to C. difficile. In a recent study, Lee et al12 reported 33% mortality for colectomy performed for C. difficile colitis. In the further evaluation of risk factors, independent factors associated with mortality in the NSQIP cohort were nonmodifiable patient-related factors like age, female sex, and chronic steroid use, along with modifiable factors like uremia, coagulopathy, and respiratory insufficiency. Age is a well-recognized risk factor for mortality after colectomy for fulminant colitis and is previously reported by Pepin et al13; however, the optimal cutoffs for classifying age were not identified. In a meta-analysis, Bhangu et al14 reported a 2.3 times higher risk of death among patients greater than 75 years old which resembled our model in which the risk of death was 2.3 times higher in 60 to 70 year olds, 3.5 times higher in 70 to 80 year olds, and 22 times higher in greater than 80 year olds (compared with patients ,60). We identified female sex as an independent risk factor of mortality. This had not been previously identified as a predictive factor for mortality in patients undergoing colectomy for toxic colitis. We found that women have 4 times higher risk of mortality after colectomy for toxic colitis than men. The majority of women in our study population were postmenopausal with a mean age of 61.22 years (male mean age was 59.10 years). In our opinion, this was because of a lack of protective factors normally present in premenopausal women. In the cardiac surgery literature, where there have also been higher rates of female inpatient mortality, there has been research hypothesizing that hormonal pathways, such as menopause and proinflammatory effects of exogenous hormone therapy, are responsible for this discrepancy.15,16 We believe that menopause-induced hormonal changes may have increased the risk of female inpatient mortality in the postsurgical period. However, we are hesitant to propose this hypothesis without an analysis of the hormonal status or use of hormonal replacement therapy in the female patients. As neither of these conditions is recorded in the NSQIP database, it would be difficult to suggest a direct impact of hormonal status on surgical outcomes. Although sex is a nonmodifiable factor, we identify it as an important parameter for preoperative risk stratification and patient and family counseling. We also found chronic steroid use to be independently associated with postoperative death, which is highly relevant as IBD patients are frequently on chronic steroid therapy. Per clinical guidelines ‘‘Toronto Consensus Statements 2012,’’ patients with acute severe episodes of ulcerative colitis should receive a trial of IV steroids. Patients should be assessed for nonresponse to steroids within 3 days in order to proceed promptly to second-line medical therapy or surgery, as inappropriate delays to surgery may lead to higher risks of postoperative complications.17
857 In-hospital factors recorded in the NSQIP database identified coagulopathy, uremia, and respiratory insufficiency as factors associated with mortality in our multivariate analysis. Derangement of these factors may reflect underlying sepsis or organ dysfunction. Previous reports have identified various risk factors that are independently associated with mortality in the treatment of fulminant colitis such as elevated lactate level, CRP level, and WBC greater than 50,000.13,14 However, lactate and CRP level were not available in the ACS-NSQIP for the analysis. The mean preoperative WBC count in this cohort was 20.07 6 13.04, with a mean of 18.5 in survivors vs 24.5 in nonsurvivors (significant on univariate analysis but not on multivariate regression). The findings of this study must be interpreted in the context of the study design. The use of ACS-NSQIP data has numerous advantages, as it includes a robust sample of patients from more than 300 hospitals across the nation. Because of the diversity of hospitals that contribute data to the ACS-NSQIP, this study is more representative of ‘‘realworld’’ outcomes of colectomy for toxic colitis. ACSNSQIP includes preoperative laboratory values as well as information regarding preoperative health status and comorbidities. These are not available in administrative databases such as the Healthcare Cost and Utilization Project Nationwide Inpatient Sample and the Medicare database. The data are collected prospectively with rigorous attention to detail and with standardized definitions for preoperative variables and complications. However, this dataset is not without limitations. Case identification was based on postoperative diagnosis ICD-9 codes, which have the potential to miss some cases. The type of colectomy performed for each patient was identified by CPT codes, which are subjected to coding variability. These operations are often coded as either a partial or total colectomy. However, no operation type (partial/complete) or specific CPT code was associated with increased mortality. This finding was in line with the Bhangu meta-analysis, which also found no significant difference in overall mortality following total colectomy and end ileostomy as compared with segmental colonic resection for fulminant C. difficile colitis.14 The variables that were analyzed were limited to those that could be captured by the ACS-NSQIP dataset; specifically, the cause of toxic colitis, variables regarding preoperative medical treatment, indications for colectomy, and reasons for reoperation were not captured. However, this study presents much needed national data on surgical outcomes, results of colectomy for toxic colitis, and risk factors associated with adverse events postoperatively. This study provides broad, nationally relevant results that can aid acute care surgeons, patients, and family members in making informed decisions regarding the surgical care of toxic colitis. In addition, these results may help identify patients at high risk of postoperative mortality, a vital aspect of goal of care discussion, which is ethically warranted in the care of severely ill patients. In a case such as multiorgan failure with toxic colitis, the decision whether or not to
858
The American Journal of Surgery, Vol 210, No 5, November 2015
operate can be an immensely difficult one. This study can help in making informed decision in order to avoid the medicolegal ramifications either of performing an unnecessary procedure or of withholding a lifesaving one.
Conclusions Our study provides real-world postoperative outcomes data on colectomy for toxic colitis. Patient with toxic colitis are very ill. The high morbidity and mortality of toxic colitis require early diagnosis, intense medical treatment, and early involvement of acute care surgeons. Advanced age, female sex, and chronic steroid use are nonmodifiable risk factors associated with a significant increase in mortality, and the presence of these factors may affect decisions regarding the surgical management of toxic colitis. Coagulopathy, uremia, and respiratory insufficiency are modifiable risk factors associated with mortality, and indicate underlying organ failure and poor outcomes, emphasizing the need for timely intervention. Future studies using alternative data sources may provide additional valuable information involving the timing of surgery for toxic colitis.
Acknowledgment We acknowledge the NSQIP surgical clinical nurse reviewers and administrators for their dedication to ensure the integrity of NSQIP data.
Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.amjsurg.2015.03.026.
References 1. Strong SA. Management of acute colitis and toxic megacolon. Clin Colon Rectal Surg 2010;23:274–84.
2. Gan SI, Beck PL. A new look at toxic megacolon: an update and review of incidence, etiology, pathogenesis, and management. Am J Gastroenterol 2003;98:2363–71. 3. Dallal RM, Harbrecht BG, Boujoukas AJ, et al. Fulminant Clostridium difficile: an underappreciated and increasing cause of death and complications. Ann Surg 2002;235:363–72. 4. Sailhamer EA, Carson K, Chang Y, et al. Fulminant Clostridium difficile colitis: patterns of care and predictors of mortality. Arch Surg 2009;144:433–9; discussion, 439–40. 5. Ausch C, Madoff RD, Gnant M, et al. Aetiology and surgical management of toxic megacolon. Colorectal Dis 2006;8:195–201. 6. Jalan KN, Sircus W, Card WI, et al. An experience of ulcerative colitis. I. Toxic dilation in 55 cases. Gastroenterology 1969;57:68–82. 7. Katzka I, Katz S, Morris E. Management of toxic megacolon: the significance of early recognition in medical management. J Clin Gastroenterol 1979;1:307–11. 8. Present DH, Wolfson D, Gelernt IM, et al. Medical decompression of toxic megacolon by ‘‘rolling’’. A new technique of decompression with favorable long-term follow-up. J Clin Gastroenterol 1988;10: 485–90. 9. Neal MD, Alverdy JC, Hall DE, et al. Diverting loop ileostomy and colonic lavage: an alternative to total abdominal colectomy for the treatment of severe, complicated Clostridium difficile associated disease. Ann Surg 2011;254:423–7; discussion, 427–9. 10. Goligher JC, Hoffman DC, de Dombal FT. Surgical treatment of severe attacks of ulcerative colitis, with special reference to the advantages of early operation. Br Med J 1970;4:703–6. 11. Greenstein AJ, Sachar DB, Gibas A, et al. Outcome of toxic dilatation in ulcerative and Crohn’s colitis. J Clin Gastroenterol 1985;7:137–43. 12. Lee DY, Chung EL, Guend H, et al. Predictors of mortality after emergency colectomy for Clostridium difficile colitis: an analysis of ACSNSQIP. Ann Surg 2014;259:148–56. 13. Pepin J, Vo TT, Boutros M, et al. Risk factors for mortality following emergency colectomy for fulminant Clostridium difficile infection. Dis Colon Rectum 2009;52:400–5. 14. Bhangu A, Nepogodiev D, Gupta A, et al, West Midlands Research Collaborative. Systematic review and meta-analysis of outcomes following emergency surgery for Clostridium difficile colitis. Br J Surg 2012;99:1501–13. 15. Hu FB, Grodstein F, Hennekens CH, et al. Age at natural menopause and risk of cardiovascular disease. Arch Intern Med 1999;159:1061–6. 16. Vaccarino V, Krumholz HM, Yarzebski J, et al. Sex differences in 2year mortality after hospital discharge for myocardial infarction. Ann Intern Med 2001;134:173–81. 17. Bitton A, Buie D, Enns R, et al. Treatment of hospitalized adult patients with severe ulcerative colitis: Toronto consensus statements. Am J Gastroenterol 2012;107:179–94. author reply 195.
A. Dayama et al. eTable 1
44140: 44141: 44143: 44144: 44150: 44151: 44160: 44210:
Outcomes of colectomy in toxic colitis
Current Procedural Terminology codes for colectomy procedures included in the study Colectomy, partial; with anastomosis Colectomy, partial; with skin level cecostomy or colostomy Colectomy, partial; with end colostomy and distal closure of segment (Hartmann type procedure) Colectomy, partial; with resection, with colostomy or ileostomy, and creation of mucous fistula Colectomy, total, abdominal, without proctectomy; with ileostomy or ileoproctostomy Colectomy, total, abdominal, without proctectomy; with continent ileostomy Colectomy, partial, with removal of terminal ileum with ileocolostomy Laparoscopy, surgical; colectomy, total, abdominal, without proctectomy, with ileostomy or ileoproctostomy
858.e1
Clinical Science
Predictive factors associated with adverse events in patients with toxic colitis: an analysis of the National Surgical Quality Improvement Project Anand Dayama, M.D.a,*, Dordaneh Sugano, B.S.b, Melvin E. Stone, M.D.b, John McNelis, M.D.b a
Department of Surgery, San Joaquin General Hospital, University of California Davis, French Camp, CA; bDepartment of Surgery, Jacobi Medical Center, Albert Einstein College of Medicine, 1400 Pelham Parkway South, Bronx, NY 10461, USA
KEYWORDS: Toxic colitis; Colectomy; Outcomes; NSQIP; Risk factors
Abstract BACKGROUND: Toxic colitis (TC) is a potentially lethal complication of inflammatory bowel disease and infectious colitis. METHODS: The National Surgical Quality Improvement Project dataset was queried to identify emergent colectomy for a primary diagnosis of TC using International Classification of Diseases, Ninth Revision codes. The study endpoints were 30-day mortality and 30-day morbidity. We performed multivariate logistic regression analyses to study factors associated with 30-day mortality. RESULTS: Of the 189 TC patients, mortality for colectomy was 26.5%, 42.3% experienced pulmonary complications, 20.6% experienced sepsis complications, 17.5% experienced cardiovascular complications, 12.7% experienced renal complications, and 14.8% experienced wound complications. On multivariate analysis, independent factors associated with mortality were age 70 to 80 years (odds ratio [OR] 3.5), age greater than 80 (OR 22.2), female sex (OR 4.1), uremia blood urea nitrogen greater than 40 (OR 4.1), coagulopathy international normalized ratio greater than 2 (OR 7.7), preoperative respiratory failure (OR 2.73), and preoperative steroid use (OR 3.9). CONCLUSIONS: Patients with TC are very ill. Poor outcome is associated with older age, female sex, preoperative azotemia, preoperative respiratory failure, and chronic steroid use. These factors will help acute care surgeons in preoperative risk assessment and could be an important addition to decision-making strategies. Ó 2015 Elsevier Inc. All rights reserved.
The authors declare no conflicts of interest. Presented at the Society of Critical Care Medicine’s 44th Critical Care Congress, January 17–21, 2015, Phoenix Convention Center, Phoenix, Arizona. * Corresponding author. Tel.: 11-404-817-1342; fax: 11-718-9185567. E-mail address: [email protected] Manuscript received December 2, 2014; revised manuscript March 24, 2015 0002-9610/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjsurg.2015.03.026
Toxic colitis is a potentially lethal complication of inflammatory bowel disease (IBD) and infectious colitis. It is defined as segmental or total dilatation of colon greater than 6 cm in the presence of acute colitis and sepsis.1 The incidence of toxic colitis is dependent on the underlying cause; for example, the lifelong incidence is 1% to 2.5% in ulcerative colitis patients2 and 3% to 5% in patients with Clostridium difficile.3,4 Although IBD is the most common etiology, almost any inflammatory condition of the colon can lead to toxic colitis: ischemic colitis, viral
A. Dayama et al.
Outcomes of colectomy in toxic colitis
colitis (such as CMV colitis, HIV/AIDS, Kaposi sarcoma), bacterial colitis (such as Salmonella, Shigella, Campylobacter), medication-induced colitis, and so on.5 Diagnosis of toxic colitis is made on the basis of a combination of clinical criteria (diarrhea, bloody diarrhea, obstipation, abdominal pain, tenderness, decreased bowel sounds, peritonitic signs, fever [.101.5 F], tachycardia [.120 beats/ minute], leukocytosis [.10.5 ! 103/mL], and anemia) and radiologic criteria (dilatation of colon .6 cm, diffuse colonic wall thickening, paracoloic fat stranding, perforation, and abscesses), along with any one of the following: dehydration, altered mental status, electrolyte abnormality, or hypotension.6 Colonoscopy is not usually performed because of high risk of perforation. Intensive medical management is generally the first line of treatment in toxic colitis. However, surgical salvage remains the cornerstone of treatment in medically refractory patients. This study aimed to assess 30-day surgical outcomes after colectomy for toxic colitis in the American College of Surgeons National Surgical Quality Improvement Project (ACS-NSQIP)da national datasetdand to evaluate factors independently associated with 30day mortality and complications.
Methods Data source The ACS-NSQIP is a risk-adjusted data collection mechanism that collects and analyzes clinical outcomes data. Participating hospitals use their collected data to develop quality initiatives that improve surgical care and to identify elements in provided health care that can be improved when compared with other institutions. The ACS-NSQIP collects data on a variety of clinical variables, including preoperative risk factors, intraoperative variables, and 30-day postoperative mortality and morbidity outcomes for patients undergoing major surgical procedures in both the inpatient and outpatient setting. A site’s surgical clinical nurse reviewer captures outcomes data using a variety of methods including medical chart abstraction.
Data selection Using the ACS-NSQIP database, we identified patients undergoing colectomies from 2005 to 2012 with a primary Table 1
853 diagnosis of toxic colitis using International Classification of Diseases, Ninth Revision (ICD-9) codes. The American Medical Association’s Current Procedural Terminology (CPT, Chicago, IL) codes that describe colectomies were used to identify the patients who had received the operation (eTable 1). Preoperative demographic data obtained included age, sex, race, history of smoking (within 1 year of surgery), and alcohol use (more than 2 drinks a day in the 2 weeks before surgery). Comorbidities measured included presence or absence of renal disease (dialysis dependence), coronary artery disease (angina within 30 days of surgery, myocardial infarction within 6 months of surgery, prior percutaneous coronary intervention, and prior cardiac surgery), congestive heart failure, hypertension, peripheral arterial disease requiring previous revascularization/amputation, rest pain in lower extremity, history of chronic obstructive pulmonary disease, neurologic event or disease (stroke with or without residual deficit, transient ischemic attack, and hemiplegia), diabetes mellitus, chronic corticosteroid use, weight loss (more than 10% in the 6 months before surgery), bleeding disorders, and presence of open wounds. Other factors considered were the American Society of Anesthesiologists’ (ASA) class admission status (from home vs facility), preoperative functional statusdability to perform activities of daily living in the 30 days before surgeryd(independent, partially dependent, totally dependent), dyspnea (none, moderate exertion, at rest), prior surgery within 30 days, do not resuscitate status before operation, disseminated cancer, preoperative sepsis, days from admission to operation, and body mass index. Preoperative laboratory variables analyzed included blood urea nitrogen, creatinine, albumin, bilirubin, hematocrit, platelet count, serum glutamic oxaloacetic transaminase, white blood cell count (WBC), partial thromboplastin time, and prothrombin time. Intraoperative data analyzed were intraoperative transfusion, type of operation (partial vs complete colectomy), and operative time.
Outcome The primary outcome was perioperative mortality and the secondary outcome was 30-day morbidity. We created composite endpoints to facilitate a better understanding of 30-day morbidity outcomes. These composite endpoints included the following: renal complications, pulmonary
Definitions of composite end points
1. 30-Day mortality: death within 30 days after surgery 2. Renal complications: progressive renal failure or acute renal failure 3. Pulmonary complications: pneumonia, unplanned intubation, or respiratory insufficiency requiring ventilation for greater than or equal to 48 hours postoperatively 4. Cardiovascular complications: pulmonary embolism, stroke/cerebrovascular event, cardiac arrest, myocardial infarction, or deep vein thrombosis/thrombophlebitis 5. Wound complications: superficial SSI, deep SSI, organ space SSI, and wound disruption SSI 5 surgical site infection.
854 Table 2
The American Journal of Surgery, Vol 210, No 5, November 2015 Preoperative characteristics of patient with toxic colitis
Characteristics
Total, n (%)
Cohort without mortality within 30 days, n (%)
Age Female Race: black BMI Diabetes Dyspnea At rest Moderate exertion History of hypertension Chronic steroid use Functional status before surgery Partially dependent Dependent Ventilator support (respiratory insufficiency) History of congestive heart failure History of MI within 6 months Weight loss in last 6 months Bleeding disorder American Society of Anesthesiologists physical status classification II and III IV and V Do not resuscitate status Preoperative sepsis Preoperative WBC Preoperative platelet Preoperative INR Preoperative BUN Preoperative creatinine Preoperative bilirubin Preoperative SGOT Days from admission to operation Number of intraop RBC transfusion Operative time Preoperative acute renal failure Preoperative dialysis Wound classification 1 (clean) 2 (clean/contaminated) 3 (contaminated) 4 (dirty/infected) Emergency cases
60.17 6 17.85 95 (50.3) 21 (11.1) 25.47 6 9.72 34 (18)
56.63 6 17.37 65 (46.8) 13 (9.4) 25.29 6 10.05 21 (15.1)
Cohort with mortality within 30 days, n (%) 69.08 6 15.88 30 (60) 8 (16) 25.82 6 8.89 13 (26)
33 27 111 55
(17.5) (14.3) (58.7) (29.1)
18 17 78 35
(12.9) (12.2) (56.1) (25.2)
15 10 33 20
(30) (20) (66) (40)
34 54 49 18 10 16 44
(18) (28.6) (25.9) (9.5) (5.3) (8.5) (23.3)
26 29 25 6 7 14 27
(18.7) (20.9) (18) (4.3) (5) (10.1) (19.4)
8 25 24 12 3 2 17
(16) (50) (48) (24) (6) (4) (34)
71 (37.6) 118 (62.4) 5 (2.6) 154 (81.5) 20.07 6 13.04 251.55 6 148.43 1.48 6 .60 34.65 6 24.74 2.05 6 1.90 .88 6 1.07 48.26 6 64.34 5 6 7.7 1.34 6. 1.86 140.38 6 58.33 14 (21.7) 14 (7.4)
65 (46.8) 74 (53.2) 2 (1.4) 107 (77) 18.45 6 12.97 262.68 6 159.86 1.40 6. 45 30.40 6 24.26 1.82 6 1.79 .84 6 1.08 39.45 6 51.71 4.75 6 7.79 1.18 6 1.52 145.63 6 61.98 25 (18) 8 (5.8)
6 (12) 44 (88) 3 (6) 47 (94) 24.55 6 12.28 192.65 6 136.37 1.71 6. 87 46.45 6 22.29 2.69 6 2.06 1.05 72.75 6 86.42 5.56 6 7.73 1.79 6 2.55 125.78 6 44.04 16 (32) 6 (12)
6 60 70 53 148
5 47 52 35 102
1 13 18 18 46
(3.2) (31.7) (37) (28) (78.3)
(3.6) (33.8) (37.4) (25.2) (73.4)
(2) (26) (36) (36) (92)
P value ,.01 .10 .20 .74 .08 ,.01
.22 .04 ,.01 ,.01 ,.01 .36 .18 .03 , .01
.07 ,.01 ,.01 ,.01 .01 ,.01 ,.01 .33 .01 .52 .11 .01 .03 .148 .46
,.01
Data are expressed as mean 6 standard deviation. BMI 5 body mass index; BUN 5 blood urea nitrogen; INR 5 international normalized ratio; MI 5 myocardial infarction; RBC 5 red blood cell; SGOT 5 serum glutamic oxaloacetic transaminase; WBC 5 white blood cell.
complications, sepsis-related complications, cardiovascular complications, and wound complications. Events included in these composite endpoints are listed in Table 1. Other outcomes analyzed included wound infection, pneumonia, reintubation, mechanical ventilation greater than 48 hours, pulmonary embolus, deep venous thrombosis, renal insufficiency, acute renal failure, cardiac arrest, myocardial infarction, transfusion greater than 4 units of packed red
blood cells within 72 hours, sepsis, septic shock, and return to the operating room.
Statistical analysis We described categorical data as absolute numbers and percent prevalence (%) in the study cohort and continuous
A. Dayama et al. Table 3
Outcomes of colectomy in toxic colitis
855
30-day outcomes of colectomy for toxic colitis
Mortality Renal complications Cardiovascular complications Wound complications Pulmonary complications Sepsis complications Pneumonia Unplanned intubation On ventilator for .48 hours Pulmonary embolism Cardiac arrest requiring CPR Myocardial infarction Deep vein thrombosis Sepsis Septic shock Superficial SSI Deep SSI Organ space SSI Wound disruption Cerebrovascular accident Acute renal failure Progressive renal failure Length of stay Return to operating room
Total, n (%)
Cohort without mortality within 30 days, n (%)
Cohort with mortality within 30 days, n (%)
P value
189 24 (12.7) 33 (17.5) 28 (14.8) 80 (42.3) 39 (20.6) 30 (15.9) 19 (10.1) 66 (34.9) 5 (2.6) 10 (5.3) 2 (1.1) 16 (8.5) 24 (12.7) 15 (7.9) 12 (6.3) 2 (1.1) 10 (5.3) 6 (3.2) 4 (2.1) 23 (12.2) 2 (1.1) 21.59 6 23.39 28 (14.8)
– 9 (6.5) 21 (15.1) 27 (19.4) 46 (33.1) 26 (18.7) 19 (13.7) 7 (5) 38 (27.3) 5 (3.6) 1 (.7) 0 14 (10.1) 22 (15.8) 4 (2.9) 11 (7.9) 2 (1.4) 10 (7.2) 6 (4.3) 3 (2.2) 9 (6.5) 0 23.90 6 26.16 19 (13.7)
50 (26.5) 15 (30.) 12 (24) 1 (2) 34 (68) 13 (26) 11 (22) 12 (24) 28 (56) 0 9 (18) 2 (4) 2 (4) 2 (4) 11 (22) 1 (2) 0 0 0 1 (2) 14 (28) 2 (4) 15.16 6 10.68 9 (18)
– ,.01 .15 ,.01 ,.01 .27 .16 ,.01 ,.01 .17 ,.01 .01 .18 .03 ,.01 .14 .39 .05 .13 .94 ,.01 .01 ,.01 .46
CPR 5 cardiopulmonary resuscitation; SSI 5 surgical site infection.
variables as means 6 standard deviation. Categorical variables were compared using chi-square test or Fisher’s exact test for discrete values. Independent 2-sample t tests were used for normally distributed continuous variables, and the Wilcoxon rank-sum test was used for non-normally distributed continuous and ordinal variables. Multivariable logistic regression modeling was used to assess the predictive factors for perioperative mortality while controlling for possible confounders. Adjusted multivariable models were created by all candidate covariates potentially associated with postoperative mortality on univariate analysis with P value less than or equal to .2. Candidate covariates were evaluated for inclusion in adjusted models using forward selection with P less than .05 as the entry criterion. The model selection was based on a stepwise procedure, which alternates between dropping the least significant variable from the model and then reconsidering all potential variables for reintroduction into the model until no more variables can be added. The discriminatory capacity was tested using the Harrell’s concordance index. Harrell’s C statistic was used to estimate the proportion of correct predictions. Harrell’s C index ranges from .5 (no discrimination) to 1 (perfect discrimination). The overall model fit was obtained using the C statistic, the Hosmer–Lemeshow goodness-of-fit test, and the Pearson chi-square statistics. We analyzed data using IBM SPSS, Version 21.0 (SPSS, Inc, Armonk, NY).
Results Patient characteristics From 2005 to 2012, a total of 189 patients underwent colectomy for toxic colitis. Of these, 148 colectomies were performed emergently. The patients’ mean age was 60.17 6 17.85 years, and 41.2% of the patients were between 18 and 60 years, 29.4% between 60 and 70 years, 16.6% between 70 and 80 years, and 12.8% of the patients were older than 80 years. About 50.3% of the patients were female and 11.1% black. Subgroup analysis comparisons based on 30day mortality of preoperative patient demographics and comorbidity data are shown in Table 2. Compared with patients in the 30-day survivor cohort, the 30-day mortality cohort had a greater prevalence of women (60% vs 46%), higher mean age (69.08 6 15.88 vs 56.63 6 17.37), higher prevalence of patients on chronic steroids (40% vs 25.2%), higher number of patients in ASA class IV and V (88% vs 53.2%), more patients presenting with preoperative sepsis (94% vs 77%), and had a higher mean leukocytosis count (24.55 6 12.28 vs 18.45 6 12.97). Mean time from admission to operation was 4.75 6 7.79 days for survivors as compared with 5.5 6 7.73 in mortality cohorts. This difference was not statistically significant.
856
The American Journal of Surgery, Vol 210, No 5, November 2015
Table 4 Multivariate model showing predictive factors for 30-day mortality following colectomy for toxic colitis with concordance index of 88% Variable
Adjusted odds ratio
95% CI Lower
Age group 18–60 Reference 60–70 2.329 .716 70–80 3.507 .957 .80 22.222 5.716 BUN ,40 Reference .40 4.099 1.626 ASA class II and III Reference IV and V 2.723 .802 INR ,2 Reference .2 7.738 1.909 Sex Male Reference Female 4.069 1.512 Respiratory insufficiency No Reference Yes 2.713 1.052 Preoperatively on chronic steroids No Reference Yes 3.943 1.554
Upper
P value
7.576 12.847 86.386
.000 .160 .058 .000
10.335
.003
9.251
.108
31.365
.004
10.948
.005
of steroids, uremia, coagulopathy, and respiratory insufficiency before surgery. Women undergoing colectomy for toxic colitis had 4.1 times higher odds for mortality as compared with men. Patients with preoperative uremia (blood urea nitrogen .40) also had a 4.1 odds ratio for mortality. Patients on chronic steroids had 3.9 times higher odds for mortality as compared with those not on steroids. Preoperative respiratory insufficiency carried 2.7 times higher odds of mortality. Compared with age less than 60, patients aged 60 to 70 years had 2.3 times higher odds of mortality, patients aged 70 to 80 years had 3.5 times higher odds, and patients above 80 years had 22.2 times higher odds of 30-day mortality. The models were robust and fit the observed data well as tested using the C statistic (C 5 .88) and the Hosmer–Lemeshow test (P 5 .457). The C statistic represents the areas under the receiver operating curve. In the Hosmer–Lemeshow test, P greater than .05 signifies that the model fits the observed data well, which was confirmed by the Pearson chi-square statistic (.9), indicating appropriate goodness-of-fit for the model.
Comments 7.001
.039
10.005
.004
ASA 5 American Society of Anesthesiologists; CI 5 confidence interval; BUN 5 blood urea nitrogen; INR 5 international normalized ratio.
Perioperative outcomes The 30-day mortality following colectomy for toxic colitis was 26.5%, 42.3% experienced pulmonary complications, 20.6% sepsis complications, 17.5% cardiovascular complications, 12.7% renal complications, and 14.8% wound complications. Comparisons of perioperative adverse events categorized by 30-day mortality cohort are shown in Table 3. Compared with patients in the 30-day survivor cohort, those in the 30-day mortality cohort had a greater incidence of prolonged mechanical ventilation (56% vs 27.3%), return to operating room (18% vs 13.7%), acute renal failure (28% vs 6.5%), and cardiac arrest requiring cardiopulmonary resuscitation (18% vs .7%). Total length of stay for survivors was 23.90 6 26.16 days.
Multivariate regression The independent variables associated with 30-day mortality at the P less than .2 level in univariate analysis were used to construct a multivariate model. Table 4 reports the results from the multivariate regression analysis. After adjusting, the factors independently associated with mortality were female sex, advanced age, preoperative chronic use
This study uses a national, multicenter, prospective clinical database to report contemporary surgical outcomes following colectomy for toxic colitis. In the analysis of this national dataset, we found that patients with toxic colitis are very ill irrespective of cause, with 81% patients experiencing preoperative sepsis and two thirds of patients in ASA class IV and V. Toxic colitis itself is associated with high postoperative morbidity and mortality. We also found that female sex, advancing age, preoperative chronic use of steroids, uremia, coagulopathy, and respiratory insufficiency before surgery are independently associated with postcolectomy 30-day mortality. Although there are numerous studies reporting outcomes after colectomies, this study is unique in that it reports outcomes of colectomy for toxic colitis. This is the largest report of patients undergoing surgery for toxic colitis. The high morbidity and mortality of toxic colitis requires early and intensive medical management with IV steroids, antibiotics, decompressive maneuvers, and other resuscitative measures to treat the underlying cause. If there is no sign of improvement within 7 days or if there are any signs of deterioration, urgent surgical intervention should be considered.7–9 Goligher et al10 reported a reduction in mortality from 20% to 7% in severe episodes of ulcerative colitis because of early colectomy. In our analysis, mean time from admission to operation was 4.75 6 7.79 days for survivors as compared with 5.5 6 7.73 days in mortality cohorts; however, this difference was not statistically significant. Specific clinical indications for colectomy were not available for analysis because of dataset limitations. Deciding the timing of a colectomy for fulminant colitis remains difficult. The mortality rate of 26.5% in our analysis is consistent with the mortality rate of 20% to 57% reported
A. Dayama et al.
Outcomes of colectomy in toxic colitis
in the literature.3,6,11,12 Greenstein et al11 reported 13 deaths out of 59 patients (22%) who underwent surgery for fulminant colitis secondary to IBD, whereas Dallal et al3 reported 57% deaths following colectomy for fulminant colitis secondary to C. difficile. In a recent study, Lee et al12 reported 33% mortality for colectomy performed for C. difficile colitis. In the further evaluation of risk factors, independent factors associated with mortality in the NSQIP cohort were nonmodifiable patient-related factors like age, female sex, and chronic steroid use, along with modifiable factors like uremia, coagulopathy, and respiratory insufficiency. Age is a well-recognized risk factor for mortality after colectomy for fulminant colitis and is previously reported by Pepin et al13; however, the optimal cutoffs for classifying age were not identified. In a meta-analysis, Bhangu et al14 reported a 2.3 times higher risk of death among patients greater than 75 years old which resembled our model in which the risk of death was 2.3 times higher in 60 to 70 year olds, 3.5 times higher in 70 to 80 year olds, and 22 times higher in greater than 80 year olds (compared with patients ,60). We identified female sex as an independent risk factor of mortality. This had not been previously identified as a predictive factor for mortality in patients undergoing colectomy for toxic colitis. We found that women have 4 times higher risk of mortality after colectomy for toxic colitis than men. The majority of women in our study population were postmenopausal with a mean age of 61.22 years (male mean age was 59.10 years). In our opinion, this was because of a lack of protective factors normally present in premenopausal women. In the cardiac surgery literature, where there have also been higher rates of female inpatient mortality, there has been research hypothesizing that hormonal pathways, such as menopause and proinflammatory effects of exogenous hormone therapy, are responsible for this discrepancy.15,16 We believe that menopause-induced hormonal changes may have increased the risk of female inpatient mortality in the postsurgical period. However, we are hesitant to propose this hypothesis without an analysis of the hormonal status or use of hormonal replacement therapy in the female patients. As neither of these conditions is recorded in the NSQIP database, it would be difficult to suggest a direct impact of hormonal status on surgical outcomes. Although sex is a nonmodifiable factor, we identify it as an important parameter for preoperative risk stratification and patient and family counseling. We also found chronic steroid use to be independently associated with postoperative death, which is highly relevant as IBD patients are frequently on chronic steroid therapy. Per clinical guidelines ‘‘Toronto Consensus Statements 2012,’’ patients with acute severe episodes of ulcerative colitis should receive a trial of IV steroids. Patients should be assessed for nonresponse to steroids within 3 days in order to proceed promptly to second-line medical therapy or surgery, as inappropriate delays to surgery may lead to higher risks of postoperative complications.17
857 In-hospital factors recorded in the NSQIP database identified coagulopathy, uremia, and respiratory insufficiency as factors associated with mortality in our multivariate analysis. Derangement of these factors may reflect underlying sepsis or organ dysfunction. Previous reports have identified various risk factors that are independently associated with mortality in the treatment of fulminant colitis such as elevated lactate level, CRP level, and WBC greater than 50,000.13,14 However, lactate and CRP level were not available in the ACS-NSQIP for the analysis. The mean preoperative WBC count in this cohort was 20.07 6 13.04, with a mean of 18.5 in survivors vs 24.5 in nonsurvivors (significant on univariate analysis but not on multivariate regression). The findings of this study must be interpreted in the context of the study design. The use of ACS-NSQIP data has numerous advantages, as it includes a robust sample of patients from more than 300 hospitals across the nation. Because of the diversity of hospitals that contribute data to the ACS-NSQIP, this study is more representative of ‘‘realworld’’ outcomes of colectomy for toxic colitis. ACSNSQIP includes preoperative laboratory values as well as information regarding preoperative health status and comorbidities. These are not available in administrative databases such as the Healthcare Cost and Utilization Project Nationwide Inpatient Sample and the Medicare database. The data are collected prospectively with rigorous attention to detail and with standardized definitions for preoperative variables and complications. However, this dataset is not without limitations. Case identification was based on postoperative diagnosis ICD-9 codes, which have the potential to miss some cases. The type of colectomy performed for each patient was identified by CPT codes, which are subjected to coding variability. These operations are often coded as either a partial or total colectomy. However, no operation type (partial/complete) or specific CPT code was associated with increased mortality. This finding was in line with the Bhangu meta-analysis, which also found no significant difference in overall mortality following total colectomy and end ileostomy as compared with segmental colonic resection for fulminant C. difficile colitis.14 The variables that were analyzed were limited to those that could be captured by the ACS-NSQIP dataset; specifically, the cause of toxic colitis, variables regarding preoperative medical treatment, indications for colectomy, and reasons for reoperation were not captured. However, this study presents much needed national data on surgical outcomes, results of colectomy for toxic colitis, and risk factors associated with adverse events postoperatively. This study provides broad, nationally relevant results that can aid acute care surgeons, patients, and family members in making informed decisions regarding the surgical care of toxic colitis. In addition, these results may help identify patients at high risk of postoperative mortality, a vital aspect of goal of care discussion, which is ethically warranted in the care of severely ill patients. In a case such as multiorgan failure with toxic colitis, the decision whether or not to
858
The American Journal of Surgery, Vol 210, No 5, November 2015
operate can be an immensely difficult one. This study can help in making informed decision in order to avoid the medicolegal ramifications either of performing an unnecessary procedure or of withholding a lifesaving one.
Conclusions Our study provides real-world postoperative outcomes data on colectomy for toxic colitis. Patient with toxic colitis are very ill. The high morbidity and mortality of toxic colitis require early diagnosis, intense medical treatment, and early involvement of acute care surgeons. Advanced age, female sex, and chronic steroid use are nonmodifiable risk factors associated with a significant increase in mortality, and the presence of these factors may affect decisions regarding the surgical management of toxic colitis. Coagulopathy, uremia, and respiratory insufficiency are modifiable risk factors associated with mortality, and indicate underlying organ failure and poor outcomes, emphasizing the need for timely intervention. Future studies using alternative data sources may provide additional valuable information involving the timing of surgery for toxic colitis.
Acknowledgment We acknowledge the NSQIP surgical clinical nurse reviewers and administrators for their dedication to ensure the integrity of NSQIP data.
Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.amjsurg.2015.03.026.
References 1. Strong SA. Management of acute colitis and toxic megacolon. Clin Colon Rectal Surg 2010;23:274–84.
2. Gan SI, Beck PL. A new look at toxic megacolon: an update and review of incidence, etiology, pathogenesis, and management. Am J Gastroenterol 2003;98:2363–71. 3. Dallal RM, Harbrecht BG, Boujoukas AJ, et al. Fulminant Clostridium difficile: an underappreciated and increasing cause of death and complications. Ann Surg 2002;235:363–72. 4. Sailhamer EA, Carson K, Chang Y, et al. Fulminant Clostridium difficile colitis: patterns of care and predictors of mortality. Arch Surg 2009;144:433–9; discussion, 439–40. 5. Ausch C, Madoff RD, Gnant M, et al. Aetiology and surgical management of toxic megacolon. Colorectal Dis 2006;8:195–201. 6. Jalan KN, Sircus W, Card WI, et al. An experience of ulcerative colitis. I. Toxic dilation in 55 cases. Gastroenterology 1969;57:68–82. 7. Katzka I, Katz S, Morris E. Management of toxic megacolon: the significance of early recognition in medical management. J Clin Gastroenterol 1979;1:307–11. 8. Present DH, Wolfson D, Gelernt IM, et al. Medical decompression of toxic megacolon by ‘‘rolling’’. A new technique of decompression with favorable long-term follow-up. J Clin Gastroenterol 1988;10: 485–90. 9. Neal MD, Alverdy JC, Hall DE, et al. Diverting loop ileostomy and colonic lavage: an alternative to total abdominal colectomy for the treatment of severe, complicated Clostridium difficile associated disease. Ann Surg 2011;254:423–7; discussion, 427–9. 10. Goligher JC, Hoffman DC, de Dombal FT. Surgical treatment of severe attacks of ulcerative colitis, with special reference to the advantages of early operation. Br Med J 1970;4:703–6. 11. Greenstein AJ, Sachar DB, Gibas A, et al. Outcome of toxic dilatation in ulcerative and Crohn’s colitis. J Clin Gastroenterol 1985;7:137–43. 12. Lee DY, Chung EL, Guend H, et al. Predictors of mortality after emergency colectomy for Clostridium difficile colitis: an analysis of ACSNSQIP. Ann Surg 2014;259:148–56. 13. Pepin J, Vo TT, Boutros M, et al. Risk factors for mortality following emergency colectomy for fulminant Clostridium difficile infection. Dis Colon Rectum 2009;52:400–5. 14. Bhangu A, Nepogodiev D, Gupta A, et al, West Midlands Research Collaborative. Systematic review and meta-analysis of outcomes following emergency surgery for Clostridium difficile colitis. Br J Surg 2012;99:1501–13. 15. Hu FB, Grodstein F, Hennekens CH, et al. Age at natural menopause and risk of cardiovascular disease. Arch Intern Med 1999;159:1061–6. 16. Vaccarino V, Krumholz HM, Yarzebski J, et al. Sex differences in 2year mortality after hospital discharge for myocardial infarction. Ann Intern Med 2001;134:173–81. 17. Bitton A, Buie D, Enns R, et al. Treatment of hospitalized adult patients with severe ulcerative colitis: Toronto consensus statements. Am J Gastroenterol 2012;107:179–94. author reply 195.
A. Dayama et al. eTable 1
44140: 44141: 44143: 44144: 44150: 44151: 44160: 44210:
Outcomes of colectomy in toxic colitis
Current Procedural Terminology codes for colectomy procedures included in the study Colectomy, partial; with anastomosis Colectomy, partial; with skin level cecostomy or colostomy Colectomy, partial; with end colostomy and distal closure of segment (Hartmann type procedure) Colectomy, partial; with resection, with colostomy or ileostomy, and creation of mucous fistula Colectomy, total, abdominal, without proctectomy; with ileostomy or ileoproctostomy Colectomy, total, abdominal, without proctectomy; with continent ileostomy Colectomy, partial, with removal of terminal ileum with ileocolostomy Laparoscopy, surgical; colectomy, total, abdominal, without proctectomy, with ileostomy or ileoproctostomy
858.e1
