Biondo S, Kreisler E, Millan M, Fraccalvieri D, Golda T, Frago R, Miguel B. Impact of Surgical Specialization on Emergency Colorectal Surgery Outcomes. Arch Surg. 2010;145(1):79-86. doi:10.1001/archsurg.2009.208
To evaluate the impact of surgeon specialization on emergency colorectal resection in terms of mortality, morbidity, and type of operation performed.
Observational study from January 1, 1993, through December 31, 2006.
Bellvitge University Hospital, Barcelona, Spain.
A total of 1046 patients underwent emergency colorectal resection. Patients were classified into 2 groups: those operated on by a colorectal surgeon (CS) and those operated on by a general surgeon (GS).
Main Outcome Measures
Preoperative variables studied were sex, age, American Society of Anesthesiologists grade, associated medical disease, presentation, reason for surgery, and type of operation. Univariate relations between predictors and outcomes were estimated, and multivariate logistic regression analysis was used to assess the prognostic effect of the combination of the variables.
Patients in the CS group underwent a significantly higher percentage of resection and primary anastomosis. The postoperative morbidity rate was 52.2% in the CS group and 60.5% in the GS group (P = .01). The anastomotic dehiscence rate was lower in the CS group (6.2%) than in the GS group (12.1%) (P = .01). Postoperative mortality decreased among patients in the CS group (17.9%) with respect to the patients in the GS group (28.3%) (P < .001). Being operated on by a CS was predictive in both the univariate and multivariate analyses for postoperative complications and mortality, and it was the only variable with predictive value for anastomotic dehiscence.
Specialization in colorectal surgery has a significant influence on morbidity, mortality, and anastomotic dehiscence after emergency operations.
Colonic emergencies remain major life-threatening conditions associated with high morbidity and mortality rates.1 Many factors have been reported as significant risk indicators for postoperative mortality in both perforation and obstruction.2- 7 The appropriate surgical treatment of colonic emergencies has been an issue of debate that continues to evolve.8,9
The development of colorectal surgery as a defined specialization has long been discussed. In many regions (eg, United States, Australasia), it exists as a surgical specialty.10 However, in Europe, colorectal surgery is not yet a formally accepted specialty. Nevertheless, there has been a trend in the past few decades, especially in tertiary referral hospitals, to organize clinical units of surgeons with a special interest in and professional activity focused specifically on coloproctology. Several studies11,12 have shown improved outcomes from fully trained colorectal surgeons (CSs). However, surgeon-related factors and outcomes have mostly been studied in the elective-surgery scenario of colorectal cancer, showing differences in terms of curative resection rate, local recurrence, and survival that reflect the degree of specialization.12,13 It has also been observed that specialization in colon and rectal surgery may contribute to the improvement of outcomes in the elective management of fistula-complicating diverticulitis.4 Nevertheless, limited conclusions can be drawn from the literature concerning the role of specialization with regard to outcomes in acute colonic emergencies because it has been studied only in 1 series dealing with patients with left-sided colorectal cancer and diverticulitis.14 The aims of the present observational study were (1) to evaluate the impact of surgeon specialization in emergency colorectal resection of all causes in terms of mortality, morbidity, type of operation performed, and hospital stay and (2) to assess the prognostic value of several factors (including surgeon specialization) for mortality, morbidity, and anastomotic dehiscence.
From January 1, 1993, through December 31, 2006, 1181 consecutive patients underwent emergency colorectal surgery at the University Hospital of Bellvitge in Barcelona, Spain. Inclusion criteria for the study were emergency colorectal resection for all causes. Exclusion criteria were any other colorectal procedure that did not include a resection.
Data were recruited prospectively from the computer database of the Colorectal Unit of the Department of Surgery at Bellvitge University Hospital, which was initially specific to research on emergency colorectal surgery. Collection of data was performed by surgeons from the department who collaborated in scientific projects under the supervision of a single surgeon (S.B.). From 2005 onward, a data manager served as a member of the unit to enter data on emergency and elective cases and manage the database. Data were collected on a daily basis by surgeons from the unit during admittance of patients to the hospital. Audits of the database were made before any study was started.
Patients were classified into 2 groups: patients operated on by a CS or by trainees assisted by a CS and patients operated on by a general surgeon (GS) or by trainees assisted by a GS. During the study period, 38 surgeons were involved: 8 CSs and 30 GSs.
A surgeon was defined as a CS if, after having completed general surgery training, he or she had undergone a training period of at least 1 year in the Colorectal Unit of the Bellvitge Hospital or in another national or international colorectal referral center. The rest of the surgeons were considered GSs independent of the Department of Surgery, where they usually worked. Surgeons with less than 1 year of practice after their training in general surgery were considered trainees.
The Department of Surgery is structured into specialized units with exclusive dedication to colorectal, hepatobiliary, pancreatic, liver transplantation, endocrine, bariatric, and gastroesophageal surgery. Twenty-six specialists in general surgery are distributed into these clinical units on a permanent basis. The Department of Surgery provides 2 on-call, in-house staff surgeons per day. All patients who undergo emergency digestive surgery and abdominal trauma surgery are attended to without referrals to specialized surgeons, except in the case of a patient who requires hepatic surgery.
Preoperative variables studied were sex, age, American Society of Anesthesiologists (ASA) grade, associated major medical disease (cardiovascular, respiratory or renal disease, diabetes mellitus, immunocompromised status), indication for surgery, cause of disease, type of operation, and type of surgeon. The severity of peritonitis was labeled according to the Hinchey classification.15 Preoperative evaluation of patients included clinical examination, blood tests, abdominal ultrasonography, computed tomography or hydrosoluble contrast enema as needed, and abdominal and chest radiography.
Emergency presentation syndromes included the following: (1) abdominal distention, nausea, vomiting, and absence of flatus or bowel movements (in patients with any of these conditions, preoperative bowel preparation was not given and surgery was performed less than 24 hours from time of admittance); (2) clinical signs of peritonitis, septic status with fever, and elevated white blood cell count; and (3) massive low gastrointestinal tract hemorrhage confirmed by angiography.
The reasons for surgery were classified into 4 main groups: tumor, diverticular disease, ischemic colitis, and other (including inflammatory bowel disease, iatrogenic foreign body traumatism, fecaloma, complicated hernia, volvulus, rectal prolapse, and unknown causes). Outcome variables included postoperative complications, development of anastomotic dehiscence, subsequent operation, length of postoperative stay, and mortality. According to the UK Surgical Infection Study Group,16 anastomotic dehiscence was defined as the leak of luminal content from a surgical join between 2 hollow viscera. Postoperative mortality was defined as death that occurs during the first 30 postoperative or in-hospital days, regardless of the interval between primary operation and death.
The choice of the operation depended on the site of the lesion. If the complicated lesion was localized at or proximal to the splenic flexure, right or extended right colectomy with ileocolic anastomosis was performed. In the presence of lesions distal to the splenic flexure, a left hemicolectomy or anterior rectal resection was performed. Following our strategy for the management of left-sided colonic emergencies described elsewhere,17,18 the first choice of operation was resection and primary anastomosis with intraoperative colonic lavage. Dependent on the judgment of the surgeon, protective ileostomy was performed in some high-risk patients (eg, steroid-dependent treatment, fecal peritonitis in fit patients). Subtotal colectomy with ileorectal anastomosis was performed in cases of associated proximal colon damage or if a synchronous tumor was observed on the proximal colon. In high-risk patients (septic shock, fecal peritonitis, preoperative organ failure) or in the presence of unresectable cancer, alternative interventions used were right or subtotal colectomy with terminal ileostomy, Hartmann procedure, bowel bypass, or colostomy. No laparoscopic procedures had been performed in the present series. Patients were followed up by the operating surgeon and surgeons from the department to which he or she was assigned. Some surgeons assigned to the emergency department admitted their patients to the colorectal ward, although the operating surgeon was also involved in follow-up.
The continuous variables are presented as mean, median, and range. Categorical variables are presented as absolute numbers and percentages. Comparative analyses of the quantitative data were performed by means of the t test or nonparametric test when needed (Mann-Whitney test). The χ2 test for proportions or Fisher exact test was used in the analysis as appropriate. For all analyses the P values were 2-tailed, and P < .05 was considered to indicate statistical significance.
A range of factors to estimate the probability of death, overall complications, and anastomotic dehiscence were investigated: sex, age, ASA grade, presentation, reason for surgery, type of operation, and type of surgeon (CS or GS). Univariate relations between predictors and outcomes were estimated by means of the χ2 test for proportions or the Fisher exact tests. The odds ratio was used as the measure of association. Predictors were considered to have statistically significant effects if the 95% confidence interval for the odds ratio did not include the value 1. Multivariate logistic regression analysis was used for the same factors even if not significant in the univariate analysis because of their clinical relevance to assess the prognostic effect of the combination of the variables. The SPSS software statistical package was used for the analysis (SPSS, version 13.0 for Windows; SPSS Inc, Chicago, Illinois).
During the study period, of the 1181 patients who had colorectal emergencies, 135 patients were excluded from the analysis. Reasons for exclusion were unresectable tumors with colostomy in 102 patients, unresectable tumors with bowel bypass in 19 patients, and missing data in 14 patients. A total of 1046 patients underwent colorectal resection. Three hundred sixty-eight patients were included in the CS group and 678 patients in the GS group. Surgeons in the CS group operated on a mean (SD) of 46.0 (56.0) patients, and surgeons in the GS group operated on a mean (SD) of 22.6 (22.8) (P = .12).
No differences were observed between the 2 groups of patients in clinical characteristics and reason for surgery. Obstruction was more frequent in the CS group, whereas perforation was the presentation with a higher rate among patients in the GS group. No differences were found in the grade of peritonitis between the 2 groups. Resection and primary anastomosis and subtotal colectomy with ileorectal anastomosis were more frequently performed by surgeons in the CS group, whereas patients underwent a higher percentage of Hartmann operations when operated on by surgeons in the GS group (P < .001) (Table 1). Twenty-five patients underwent protective ileostomy: 17 patients in the CS group and 8 patients in the GS group (P = .08).
The overall postoperative morbidity rate was 57.6% (602 of 1046 patients). One or more complications were observed in 192 patients in the CS group (52.2%) and in 410 patients in the GS group (60.5%) (P = .01). Table 2 gives the type and number of postoperative complications and their differences between the 2 groups. The anastomotic dehiscence rate of the whole series was 9.9% (71 of 720 patients with primary anastomosis). It occurred in 17 patients in the CS group (6.2%) and in 54 patients in GS group (12.1%) (P = .01). When anastomotic dehiscence was analyzed stratifying for the reason for surgery, differences were observed only in patients operated on for complicated colorectal cancer: 6 of 159 patients (3.8%) in the CS group and 36 of 265 patients (13.6%) in the GS group (P = .001).
The overall postoperative mortality rate was 24.7% (258 patients). Differences were observed between the 2 groups. Sixty-six of 368 patients in the CS group (17.9%) died, whereas 192 of 678 patients (28.3%) in GS group died (P < .001). Reasons for mortality were heart complications in 31 patients, respiratory failure in 39 patients, uncontrolled sepsis in 61 patients, multiorgan failure in 123 patients, and advanced oncologic disease at diagnosis in 4 patients.
One hundred fifty-two patients in all the series (14.5%) needed additional operations. No differences were observed between the 2 groups: 54 patients (14.7%) in the CS group and 98 patients (14.5%) in the GS group (P = .92). Reasons for additional operation were anastomotic dehiscence in 60 patients, intra-abdominal abscess in 42 patients (only 15 patients had had an anastomosis performed previously, 10 in the ascending colon and 5 in the descending colon), abdominal wound dehiscence in 40 patients, and postoperative intra-abdominal hemorrhage in 10 patients. In all patients treated for abscess, with or without surgery, anastomotic dehiscence was excluded either radiologically (computed tomography and enema) or at the time of operation. Patients with anastomotic dehiscence, with or without abscess, were not included among these 67 patients with abscess. The median postoperative hospital stay was 16 days (range, 5-153 days) in the CS group and 16 days (range, 4-235 days) in the GS group (P = .91).
Tables 3, 4, and 5 list the predictors considered in the univariate and multivariate analyses and give their relation to overall postoperative complications, anastomotic dehiscence, and mortality, respectively. The type of surgeon was predictive for postoperative complications and mortality in both the univariate and multivariate analyses, and it was the only variable with predictive value for anastomotic dehiscence.
Table 6 gives a subgroup analysis of outcomes according to the syndrome displayed at presentation. The GSs had a significantly higher anastomotic dehiscence rate among patients with obstruction and higher complication and mortality rates among patients with peritonitis when compared with the CSs.
This study shows that surgical specialization influences the outcomes of emergency colorectal surgery in terms of lower postoperative morbidity and mortality rates and a higher percentage of single-stage surgery. The weaknesses of the present study are related to the unequal case load with which both groups of surgeons handled and the potential bias in the distribution of patients in the 2 groups. However, the mean of operations performed per surgeons in the 2 groups, even if different, did not reach statistical significance. A higher percentage of peritonitis was observed in the GS group, whereas a higher percentage of obstructions were operated on by CSs. The reason for this distribution is that, in our hospital, emergency general surgery care is not specialized. Although it has not been an established policy, patients waited overnight to be operated on the next morning by a CS in some cases of obstruction. This would explain the fact that obstruction cases are more common in the CS group.
Several aspects may clarify this issue. It was observed in the multivariate analysis that peritonitis was not predictive of mortality, morbidity, or anastomotic dehiscence. Also, as reported in results, neither severity in peritonitis grade nor ASA classification showed any difference in both groups of patients. The absence of differences in these comparisons lessens the impact of the unequal distribution of obstructions and peritonitis in the 2 groups. Moreover, the subgroup analysis (peritonitis and obstruction) favors the argument that, beyond presentation and patient characteristics, it is the surgeon factor that makes the difference.
Surgeons need to have mastered many technically demanding operations for both the elective and emergency scenarios. Patients with colonic obstruction, complicated colonic cancer,19,20 or perforated diverticulitis are critically ill patients who require surgical confidence and experience to achieve the best operative decision and results. In a study that used decision analytic techniques to determine the optimal strategy for perforated diverticulitis, resection and primary anastomosis with defunctioning stoma were proposed as the optimal strategy for selected patients with low risk of postoperative complications.21 In that context, it has recently been observed that CSs are more likely to perform a single-stage operation for all scenarios examined. Moreover, the attitude of surgeons in the United Kingdom is changing, and the concept of specialist management in complex colonic diseases is gaining acceptance.9
In the present series, all the surgeons but 8 (3 in the CS group and 5 in the GS group) were trained in the department itself, which makes their basic performance homogeneous. It can be observed that despite the existence of a protocol of management for colorectal emergencies17,22,23 and even if patients are similar in clinical characteristics, there are significant differences in the type of operation performed, dependent on the type of surgeon. Colorectal surgeons perform a higher rate of 1-stage surgery with resection and primary anastomosis compared with GSs. Moreover, although with small numbers, surgeons in the CS group performed a higher number of primary resections with derivative ileostomy than did surgeons in the GS group. These results could reflect the trend among CSs to extend criteria for resection and primary anastomosis when a diverting stoma is associated with patients who would have been treated in the past by the Hartmann operation. The differences between the 2 groups could reflect judgment more than technical skill. In this respect, we strongly agree with O’Connell, in the commentary on the article by Zorcolo et al,14 that “specialization means much more than simple familiarity with technical aspects of surgical resection; it implies a concentration of multidisciplinary expertise and the application of established patient care pathways in a framework that allows regular reappraisal and outcome audit.”14(pp1467-1468)
Many factors, such as age, associated medical problems, ASA grade, sex, colonic perforation, and ischemic colitis, have been reported as significant predictors of morbidity and mortality after emergency surgery.3,5,19,24,25 Published postoperative complication rates range from 20% to 50%,19,25 whereas reported mortality rates vary between 5% and 38%.26,27 We report an overall morbidity rate of 57.6% and a mortality rate of 24.7%. The mortality rate in the CS group was 17.9%, higher than the mortality rates observed in the Scottish study.14 The inclusion of patients at high risk for mortality (ischemic colitis) could have influenced our results.
In the present series, 14.5% of the patients needed additional operations. The similar number of additional operations in the 2 groups despite the higher number of complications among patients in the GS group could be related to the higher number of postoperative deaths observed among complicated patients in this group.
Several factors that were considered to be of clinical relevance for morbidity and mortality were analyzed. Sex, ASA grade, Hartmann operation, and type of surgeon were significant predictors of morbidity in multivariate analysis. When the study was performed with a focus on postoperative mortality, all the variables considered except syndromes at presentation and reasons for surgery (except ischemic colitis) were predictive in the logistic regression analysis.
Anastomotic leak continues to be one of the most feared complications in colorectal surgery. There is discrepancy in the reported incidence of anastomotic dehiscence after elective or emergency colonic resection, ranging from 4.3% to 13.0% for all resections.28 In addition, there is no clear consensus about the risk factors for anastomotic failure after elective or emergency colorectal surgery.29,30 Moreover, anastomotic leak is associated with poor survival and higher tumor recurrence rates after curative resection of colorectal cancer, and efforts should be undertaken to avoid this complication to improve the long-term outcome.24
In the present series, the overall anastomotic dehiscence rate was 9.9%, with a significant difference between the 2 groups of surgeons (6.2% in the CS group and 12.1% in the GS group). According to the reason for surgery, differences were observed in patients with complicated neoplasia. Moreover, when a multivariate analysis was performed, operation by a CS was the only predictive variable for anastomotic dehiscence.
In conclusion, this study shows that specialization in colorectal surgery improves mortality, morbidity, and anastomotic dehiscence rates after colorectal emergencies. Also, it raises the percentage of single-stage procedures. It may be difficult to control many disease-related factors, but, as a consequence of these results, we may need to encourage health authorities to organize emergency procedures in such a way that CSs are in charge of handling colorectal emergencies.
Correspondence: Sebastiano Biondo, MD, PhD, Department of Surgery, Colorectal Unit, Bellvitge University Hospital, C/ Feixa Llarga s/n, L’Hospitalet de Llobregat, 08907 Barcelona, Spain (firstname.lastname@example.org).
Accepted for Publication: December 15, 2008.
Author Contributions:Study concept and design: Biondo. Acquisition of data: Millan, Fraccalvieri, Golda, and Frago. Analysis and interpretation of data: Biondo, Kreisler, Millan, and Miguel. Drafting of the manuscript: Biondo, Kreisler, and Fraccalvieri. Critical revision of the manuscript for important intellectual content: Biondo, Kreisler, Millan, Golda, and Miguel. Statistical analysis: Miguel. Administrative, technical, and material support: Miguel. Study supervision: Biondo.
Financial Disclosure: None reported.