Laparoscopy Decreases Anastomotic Leak Rate in Sigmoid Colectomy for Diverticulitis

Background  Early studies comparing laparoscopic and open operations for diverticulitis failed to show any advantages of the laparoscopic approach. Our study compared the 30-day postoperative outcomes of laparoscopic and open sigmoid colectomy for diverticulitis by surgeons who had performed 20 or more laparoscopic colectomies before the study period.

Hypothesis  Patients who undergo an electivelaparoscopic operation for diverticulitis have reduced postoperativecomplications compared with patients who have a traditional openoperation.

Design  Retrospectiveanalysis.

Setting  Academic medicalcenter.

Patients  A total of 249patients who underwent elective open (n = 127) orlaparoscopic (n = 122) sigmoid colectomy with primaryanastomosis for diverticulitis between July 1, 2001, and February 1,2008.

Main Outcome Measures  Combined rates of free and contained anastomotic leaks. A logisticregression model was used to determine predictors of anastomoticleaks while controlling for significant differences between studygroups.

Results  Patients whounderwent laparoscopic or open operations were similar in age, sex,history of diagnosed intraabdominal abscess (9.4% vs 12.3%), andhistory of preoperative percutaneous abscess drainage (3.9% vs 4.9%).Patients who underwent the open procedure had a higher Charlsoncomorbidity index (1.6 vs 1.2; P = .04), and those wh underwentlaparoscopy more frequently underwent splenic flexure mobilization(82.8% vs 26.7%; P < .001).Patients who underwent a laparoscopy had lower rates of anastomoticleaks (2.4% vs 8.2%; P = .04).This finding held true on logistic regression analysis (odds ratio, 0.67; 95% confidence interval, 0.008-0.567; P = .01), even when controlling for age,Charlson comorbidity index, splenic flexure mobilization, and lengthof resected bowel.

Conclusion  Anastomotic leaks occurred less frequently after laparoscopicsigmoid colectomy performed by experienced laparoscopic colorectalsurgeons.

Diverticulitis remains a commondisease with a rising incidence in Western society.¹ Today, most attacks ofdiverticulitis can be managed without surgical intervention asantibiotic efficacy, percutaneous drainage, and diagnostic modalitiesimprove.^2,3As a result of the increase in nonoperative management, theindications for elective colectomy for diverticulitis have evolved as well. Infrequently, patients may still need traditional laparotomiesand colostomies to treat perforated feculent peritonitis. Far morecommonly, however, patients undergo elective procedures to preventfuture trips to the emergency department, minimize pain associatedwith attacks, and reduce anxiety associated with developing furtherattacks of diverticulitis.^4,5 The timing for elective operations for thesepatients, however, is a matter of heated debate.^3,5

Concomitant with theshift toward a more nonoperative approach to the treatment ofdiverticulitis, the surgical community is also witnessing rapidadvancement in the development of laparoscopy. Since the firstdescription of laparoscopic colectomy in 1991,⁶ numerous authors have published reports on theadvantages of laparoscopic colectomy for diverticular disease.^7-14 These and other studies documented multipleshort-term benefits in patients who underwent laparoscopiccolectomies, including shortened length of hospital stay, fewer wound infections, decreased pain, and better pulmonary function.^15,16 On the otherhand, investigators have not been able to show significantdifferences in the rates of serious complications such as deaths,anastomotic leaks, and reoperations—nor were they able to showany long-term advantages of laparoscopy.^7,14,17,18

Open sigmoid colectomy remains thecriterion standard for the surgical approach for patients with priorattacks of diverticulitis; however, this assertion has recently beenchallenged.¹⁹ Informationconcerning 8660 patients who underwent either open or laparoscopiccolectomies were entered into the database maintained by the AmericanCollege of Surgeon's National Surgical Quality Improvement Program(NSQIP). A review of these data found that laparoscopy for abdominalcolectomy decreased overall complications as well as individualcomplications, independent of the NSQIP morbidity probabilitystatistic, which is an NSQIP method for estimating a patient's riskfor postoperative complications based on that patient's preoperativecomorbidities.²⁰ Inaddition, the authors found a significant increase in wound andseptic complications in the open colectomy group. However, becausethe NSQIP database only documents deep infections in general, thisstudy was hindered by its inability to report one of the mostimportant complications pertinent to colorectal operations,specifically, the rate of anastomotic leaks. The authors also wereunable to report how many patients required diversion or how manypatients never received an anastomosis. Furthermore, their data contradicted the results of several randomized controlled studiesthat claim that laparoscopic and open techniques have similarpostoperative outcomes.^21,22

The purpose of this study was tosystematically compare the 30-day postoperative outcomes oflaparoscopic vs open sigmoid colectomies performed by surgeons whohad completed 20 or more laparoscopic colectomies before the study period. Our goal was to test the hypothesis that patients whounderwent elective laparoscopic sigmoid colectomy for diverticulitishave reduced postoperative complications compared with patients whohad traditional open sigmoid colectomy.

Our study population consisted of 249patients who underwent elective laparoscopic or open sigmoidcolectomy with primary anastomosis from July 1, 2001, throughFebruary 1, 2008, by 1 of 9 surgeons at the Massachusetts GeneralHospital. These patients were identified from the MassachusettsGeneral Hospital medical records database using the Research PatientDatabase Query tool to identify patients who underwent partialcolectomy for the diagnosis of diverticulitis using theInternational Classification of Diseases, NinthRevision²³ procedure code 562.11.The initial search generated 421 results. Theseelectronic records were then individually reviewed and 172 patientswere excluded because their operation was miscoded and they underwenta subtotal colectomy, diverticulitis was not the indication for surgical intervention, intestinal continuity had not been restored,or the operation was not performed on an elective basis. Theremaining 249 patients became our study population. Diverticulitis ofthe sigmoid colon was diagnosed in these patients based on the results of computed tomography imaging, and the disease wasconsidered asymptomatic at the time they were admitted for theirelective colectomy. Their medical records were then carefullyreviewed to document sex, age, comorbidities, preoperativemddications, preoperative disease history (eg, number of attacks ofdiverticulitis before resection, number of attacks with associatedabscesses, and number of attacks requiring percutaneous drainage),details of the ultimate operative procedures, and postoperativerecovery and complications. The age-adjusted Charlson comorbidityindex, initially developed in 1987 and extensively used in theliterature to demonstrate predictive validity for risk of mortality,^24,25 was calculated for each patient based on dataollected for preexisting comorbidities.

Our primary end point was the cumulative rate of postoperative complications within 30 days after the surgical procedure, including anastomotic leaks; postoperative wound infections; early small-bowel obstructions; cardiac (myocardial infarction or cardiac arrest), renal (rise in serum creatinine or new dialysis requirement), pulmonary (acute respiratory distress syndrome or respiratory distress requiring intubation), or neurologic (stroke or paralysis) complications; postoperative sepsis; reexploration in the operating room; intraabdominal abscess diagnosed by computed tomography scan; or readmission. We defined anastomotic leaks based on the criteria established by Damrauer et al.²⁶Free leaks were defined as diffuse gross contamination of the peritoneal cavity (peritonitis) as demonstrated on computed tomography scan or in the operating room during a reexploration. Contained leaks were defined as localized perianastomotic collections that demonstrated communication with the gastrointestinal lumen during their percutaneous drainage or during reexploration.

All statistical analysis was completed using SAS software (SAS Institute, Inc, Cary, North Carolina). Categorical variables are reported as percentages and frequencies; continuous variables are reported as mean (SD). We used intention-to-treat analysis to compare other differences between the groups, such as demographic factors, medical comorbidities, and ultimate surgical outcomes, using Fisher exact test, χ² test, or t test, as appropriate. Finally, a multivariate logistic regression model was fitted to determine predictors of anastomotic leaks while controlling for differences identified on univariate analysis. P < .05 was considered statistically significant.

One hundred twenty-seven patients underwent elective laparoscopic sigmoid colectomy and 122 patients underwent elective open sigmoid colectomy. Twenty-two patients (17.3%) who underwent a laparoscopic procedure eventually needed an open operation. The decision to convert was made at the discretion of the surgeon at the time of operation. Reasons for conversion included extensive adhesions, bleeding, and failure to visualize the ureter.

Patients who underwent laparoscopy were similar to those who had the open procedure in age, sex, history of intra-abdominal abscess during their prior episode of diverticulitis, or history of abscess requiring percutaneous drainage before the operation. However, patients who underwent open sigmoid colectomy had a higher Charlson comorbidity index (Table 1).

Intraoperatively, patientsundergoing laparoscopy were 4 times more likely to have a splenicflexure mobilization (82.8% vs 26.7%; P < .001) and nearly 5 times morelikely to have a double-stapled end-to-end anastomosis instead of anextracorporeal handsewn anastomosis (91.8% vs 21.3%; P < .001). However, the length of colon resection, derived from the pathology report (19.9 cm vs 19.1 cm), and the need for a diverting ileostomy (0.4% vs 0.1%) were similar.

Ultimately, on univariate analysis, our 30-day postoperative outcomes were largely similar between groups (Table 2). Patients had similar rates of wound infections, early small-bowel obstructions, ileus, and renal and cardiac complications. However, patients who underwent laparoscopy had lower rates of anastomotic leaks and intra-abdominal infections.

These differences held true on multivariate logisticregression analysis. Patients who underwent laparoscopy had a statistically significant decrease in their rates of anastomoticleaks compared with those who had the open procedure (odds ratio[OR], 0.67; 95% confidence interval, 0.008-0.567;P = .01), even when controlling for age (OR, 1.11; P = .90), Charlson comorbidity index (OR, 0.67; P = .20), splenic flexure takedown (OR, 0.94; P = .40), and length of resected bowel (OR,  0.94; P = .90).

Numerous studies^27-31 have shown laparoscopic colon operations to be equivalent to open operations in terms of safety profile, while also offering patients the advantage of shorter lengths of hospital stay, smaller incisions, less pain, and quicker return to full activity. Our data, taken together with the NSQIP database, challenge this assertion that postoperative complication rates are equivalent across surgical techniques.

Fortunately, anastomotic leaks are uncommon, but they are a serious complication after colectomy. On average, the occurrence of anastomotic leaks after sigmoid resection is between 0% and 8.2%.^15,16,32,33 Unfortunately, this means that surgical studies are either underpowered to detect differences in leak rates between competing surgical techniques or the incidence of leaks is underreported because authors fail to accurately establish a precise definition for their criteria for postoperative anastomotic leak. For example, a study by Dwivedi et al¹⁵ described 66 patients who underwent laparoscopic sigmoid colectomy and 88 patients who underwent open sigmoid colectomy for diverticulitis. One patient (1.5%) in the laparoscopic group had an anastomotic leak compared with 3 patients (3.4%) in the open group. The authors concluded that the rates of postoperative complications (including all subsets) were similar. A study by Hinojosa et al³² comparing 7239 patients who underwent sigmoid colectomy for benign disease reported a nearly equal leak rate (0.7% vs 0.8%) in their series. Although this was a very large study, the authors did not comment on their criteria for defining anastomotic leaks. Kasparek et al¹⁶ and Faynsod et al³³ reported no occurrences of anastomotic leaks in either group in their series, with both including 20 patients.

In our study, we defined anastomotic leaks in a more rigorous fashion. Patients were considered to have had an anastomotic leak if they had peritonitis requiring reexploration and if they had a postoperative abscess near their anastomosis, with communication to the anastomosis proved upon injection of contrast medium into the abscess cavity. Using this more precise definition, we were able to show that the laparoscopic procedure is superior to the open procedure in important postoperative outcomes. We found an 8.2% rate of anastomotic leaks in patients who underwent open colectomies. Patients who underwent laparoscopy had only a 2.4% rate of anastomotic leaks. Interestingly, the rates of leaks requiring reexplorations were similar between the 2 arms.

Our data have some limitations. Patients selected for laparoscopy may differ from those who are offered an open operation. As seen in our cohort, patients offered laparoscopy tended to be healthier; their Charlson comorbidity index was significantly lower. We attempted to control for differences that we could measure (ie, body mass index, age, sex, and comorbidities); however, subtle differences between the 2 groups might have been missed. Although the Charlson comorbidity index is not the ideal tool to control for comorbidities in this study population, we opted to use it given that it was the best validated instrument that we could find. Nevertheless, we believe that this study, along with the NSQIP database, supports the argument that laparoscopic sigmoid colectomy is associated with lower morbidity and may become a preferred approach for patients who are candidates for elective laparoscopic procedures.

Anastomotic leaks are less common after elective laparoscopic sigmoid resection than open sigmoid resection. Although patients treated with open operations had more comorbidities, the rate of anastomotic leaks remained higher in that group even when these differences were accounted for. Based on our study results, as well as the large NSQIP database, laparoscopic sigmoid resection is associated with lower morbidity.

Correspondence: Liliana Bordeianou, MD, Division of Gastrointestinal Surgery, Massachusetts General Hospital, 15 Parkman St, Bldg ACC 460, Boston, MA 02114 (lbordeianou@partners.org).

Accepted forPublication: January 25, 2010.

AuthorContributions:Study concept and design: Levack, Berger, and Bordeianou. Acquisition of data: Levack, Sylla, and Bordeianou. Analysis and interpretation of data: Levack, Rattner, and Bordeianou. Drafting of the manuscript: Levack, Rattner, and Bordeianou. Critical revision of the manuscript for important intellectual content: Levack, Berger, Sylla, and Bordeianou. Statistical analysis: Bordeianou. Administrative, technical, and material support: Levack, Sylla, Rattner, and Bordeianou. Study supervision: Berger, Rattner, and Bordeianou.

Financial Disclosure: None reported.

Previous Presentations: This study was presented as a poster at the 50th Annual Meeting of the Society for Surgery of the Alimentary Tract; June 1, 2009; Chicago, Illinois; and at the 56th Annual Meeting of the Massachusetts Chapter of the American College of Surgeons; December 5, 2009; Boston, Massachusetts.