Mortality and complications following surgery for diverticulitis: Systematic review and meta-analysis

Abstract

Background

The surgical treatment of diverticulitis is in a state of evolution. Clinicians across many disciplines need to counsel patients regarding surgical choices.

Objectives

A systematic review and meta-analysis was conducted to determine the mortality and complication rates following surgery for diverticulitis in both the emergent and elective setting.

Methods

We searched PubMed, Embase and the Cochrane Central Register of Controlled Trials (CENTRAL) for relevant articles published from 1980 to 2012. The primary outcome of interest was the point estimate of mortality, following surgery for diverticulitis.

Results

Of the 289 citations reviewed, we included 59 studies. Overall, the point estimate for mortality was 3.05%, with a 95% confidence intereval (CI) of 1.73–5.32 and p < 0.001. Mortality following emergent surgery was 10.64% (95% CI 7.95–14.11; p < 0.001), versus 0.50% (95% CI 0.46–0.54; p < 0.001) following elective operations. A laparoscopic approach had an estimated mortality of 0.75% (95% CI 0.35–1.58; p < 0.001), compared to an open surgical approach, which had a mortality of 4.69% (95% CI 2.29–9.36, p < 0.001). The mortality following a resection with primary anastomosis was 1.96% (95% CI 1.22–3.13; p < 0.001) and for the Hartmann’s procedure was 14.18% (95% CI 9.83–20.03; p < 0.001). A comparative analysis found that the risk of post-operative mortality was significantly higher following emergent surgery, compared to elective surgery (odds ratio (OR): 6.12 with 95% CI 1.62–23.10; p = 0.008; Q = 2.56, p = 0.46 and I²= 0); the open approach, compared to a laparoscopic approach (OR: 36.43 with 95% CI 9.94–133.6; p = 0.13; and Q = 2.79, p = 0.25 and I²= 28.26); and for Hartmann’s procedure, compared to primary anastomosis without diversion (OR: 25.45 with 95% CI 15.13–42.81, p < 0.001; and Q = 23.34, p = 0.14 and I²= 27.16). The overall reported post-operative complication rate was 32.64% (95% CI 27.43–38.32; p < 0.00). The overall surgical and medical complication rates were 18.96% and 13.93%, respectively.

Conclusions

Urgent surgical treatment of diverticulitis has a significant complication rate. Even elective surgery has a significant complication rate that needs to be considered when doing the clinical decision-making for recurrent diverticulitis.

Introduction

Diverticulitis is a common clinical condition and gastroenterologists are often involved in the decision-making regarding the need for surgery. Recent studies for the management of diverticulitis demonstrate a move away from routine sigmoid colectomy in patients, based on the number of episodes of diverticulitis, and suggest that elective sigmoid colectomy should be individualized, and should consider the risk of surgery and the impact of recurrent attacks on the patient’s lifestyle.¹ Despite this, the number of operative procedures for diverticulitis continues to grow in the USA; and a recent study that evaluated indications for surgery in the state of WA found that 70% of patients were treated with surgery for an episode-based indication and that 39% had ≤ 2 episodes.² There are a number of studies regarding surgical outcomes in diverticulitis, but no single source to guide patients, based on the risk of mortality and morbidity after surgical treatment for diverticulitis. The aim of our study was to perform a systematic review and meta-analysis of the literature, to determine the mortality and complication rate of surgery for diverticulitis in the emergent and elective settings.

Methods

Search strategy

The following databases were used to find relevant articles for this review: PubMed, Embase and Cochrane Central Register of Controlled Trials (CENTRAL). The keywords “diverticulitis”, “colon”, “colonic”, “incidence”, “prevalence”, “treatment”, “therapy” and “recurrence” were used. We used Boolean operators (AND, OR) to narrow and broaden the search. Our search was limited to studies published between 1980 and 2012, the English language and human subjects. We included studies or trials of any kind, including clinical trials, surveys, cohort studies, retrospective analyses, as well as meta-analyses and systematic reviews. Only published manuscripts were used. The initial search yielded a large number of studies. After the duplicate studies were removed, the remaining titles and abstracts were manually reviewed, removing all non-relevant studies. The studies that remained were reviewed in detail. Studies not meeting the inclusion criteria were removed. The cited references of the final set of studies were also reviewed and any relevant studies were included. Review articles were not included.

Study selection

Studies were included into this meta-analysis if mortality and/or complications were reported following surgical intervention for complicated or uncomplicated diverticulitis. The following surgical complications were included: intra-abdominal abscess, anastomotic leak, bleeding, wound infection, wound dehiscence, pulmonary embolism, pneumonia, urinary tract infection and myocardial infarction. These complications were chosen for analysis because they were the most frequently reported. A number of studies were excluded for the following reasons: missing or incomplete data, surgical resection of the sigmoid colon for a reason other than diverticulitis (i.e. cancer or bleeding) and studies focusing on or including diverticular bleeding or right-sided diverticulitis.

Data extraction

The remaining studies were thoroughly reviewed. Data extracted from each study included: first author; year of publication; country of study origin; study design; number of patients in study; patient characteristics, including Hinchey class at presentation; number of patients undergoing surgical intervention; timing of surgery (emergent versus elective); surgical technique and method of surgery (laparoscopic versus open). The following surgical outcomes were extracted: mortality, intra-abdominal abscess, anastomotic leak, bleeding, wound infection, wound dehiscence, pulmonary embolism, pneumonia, urinary tract infection and myocardial infarction. Only the reported complications were included. The post-operative complications were divided into two categories: surgical complications and medical complications. To reduce errors, each study was reviewed in detail three times.

Primary and secondary outcomes of interest

The primary outcome of interest for this analysis was the point estimate of mortality, following emergent, elective, laparoscopic, open, resection with primary anastomosis, and Hartmann’s surgery for complicated or uncomplicated diverticulitis. This also included several subgroup comparative analyses (emergent versus elective, laparoscopic versus open, primary anastomosis without diversion versus Hartmann’s surgery). Secondary outcomes of interest included the point estimate complications after emergent, elective, laparoscopic, open, primary anastomosis and Hartmann’s surgery.

Definitions

Mortality was defined as death occurring within 30 days of surgery. Surgical intervention included patients undergoing sigmoid resection, with or without a proximal diversion. We did not include patients undergoing intra-abdominal lavage alone. Emergency surgery was defined as surgery within 48 hours of admission. Elective surgery was defined as surgery performed after discharge from the initial hospitalization for acute diverticulitis. Patients whom were converted to open surgery from an initial laparoscopic approach were included in the open surgical subgroup. Regarding primary resection with anastomosis, we did not include the patients whom had had a diverting ostomy in the subgroup comparative analysis versus Hartmann’s; however, we did include these patients in the point estimate mortality data. Intra-operative complications were not analyzed. A number of studies used the Hinchey classification system to grade the severity of diverticulitis, and this is as follows: Class I disease is a localized pericolic inflammation, Class II is a localized pericolic abscess, Class III is diffuse purulent peritonitis and Class IV is diffuse feculent peritonitis.³ Subgroup analysis was performed according to the Hinchey grade, when this was reported.

Statistical analysis

Statistical analysis for the category variables only with the proportions of events was performed using counts for the event and the total sample in the studies. We computed the effect size of the proportions for all the studies that had complete data for the events and for the total sample size. Heterogeneity among studies was computed as I². For category variables across more than one group, the effect size for odds ratios (ORs) with 95% CIs were computed along with I². Similarly, for continuous variables, we computed the effect size for the point estimate of the mean and standard error (SE). For all computations, an alpha level of 0.05 was used. In the present study, we computed both the fixed and random effects. The statistical analyses and computations were done using Comprehensive Meta Analysis version 2.0.

Results

The initial search strategy yielded 289 studies. Of these articles, 59 met the inclusion criteria (Figure 1).^4–62 The mean age of patients presenting with acute diverticulitis was 62.4 ± years, SE 1.38 years. Males represented 46% of the population.

Figure 1.

Search results: Initial search and final selection of articles.

Post-op: After the operation.

Post-operative mortality

Overall, there were 982 deaths following 134,428 surgical operations for diverticulitis. This yielded a point estimate for mortality of 3.05%, with 95% confidence interval (CI) 1.73–5.32, p < 0.001. We further broke down mortality into the timing of the surgery (emergent or elective), modality (laparoscopic or open), technique (resection with primary anastomosis or Hartmann’s), severity of perforation (Hinchey Class 1 and 2, or Hinchey Class 3 and 4) and the calculated point estimate data.

There were 2079 emergent operations performed, resulting in 260 deaths (mortality of 10.64% with 95% CI 7.95–14.11; p < 0.001); and following 127,169 elective operations, there were 624 deaths (mortality of 0.50% with 95% CI 0.46–0.54; p < 0.001).

A laparoscopic approach was used in 16,170 procedures that had 24 deaths (mortality 0.75%, with 95% CI 0.35–1.58; p < 0.001). An open surgical approach was used in 113,701 cases, with 840 deaths (mortality 4.69%, with 95% CI 2.29–9.36; p < 0.001).

Resection with primary anastomosis was successfully completed in 5546 patients, and 59 of these patients died (mortality 1.96%, with 95% CI 1.22–3.13; p < 0.001). The Hartmann’s procedure was used in 952 patients, resulting in 158 deaths (mortality 14.18%, with 95% CI 9.83–20.03; p < 0.001).

Of the 595 patients presenting with Hinchey Class 3–4, there were 44 deaths post-operatively (mortality 7.87%, with 95% CI 5.86–10.48, p < 0.001). Of the 1186 patients presenting with Hinchey Class 1–2, there were 22 deaths post-operatively (mortality 2.05%, with 95% CI 0.63–6.48; p < 0.001).

Comparative analysis of emergent and elective surgery

There were four studies that reported mortality following emergent and elective surgery separately, within the same cohort, and these were used for a comparative analysis between these two clinical situations.^4–7 There were 493 patients in the emergent surgery group and 199 patients in the elective surgery group. The mortality in the emergent surgery group was 9.78% (95% CI 4.13–21.43) and in the elective surgery group mortality was 1.93% (95% CI 0.53–6.70). The risk of post-operative mortality was significantly higher following emergent surgery, compared to elective surgery (OR: 6.12 with 95% CI 1.62–23.10; p = 0.008; Q = 2.56; p = 0.46; I²= 0).

There were three studies that reported mortality separately, following a laparoscopic approach and an open approach.^8–10 There were 14,671 patients in the laparoscopic group and 110,295 patients in the open group. The mortality in the open surgery group was 0.54% (95% CI 0.49–0.59) and in the laparoscopic surgery group mortality was 0.47% (95% CI 0.07–3.26). Post-operative mortality was significantly higher following an open approach, compared to a laparoscopic approach (OR: 36.43; with 95% CI 9.94–133.6; p < 0.001; Q = 2.79; p = 0.25; I²= 28.26).

There were 18 studies that reported mortality following resection with primary anastomosis without diversion and Hartmann’s procedure.^4,6,11–26 There were 829 patients in the primary anastomosis group and 766 patients in the Hartmann’s procedure group. Post-operative mortality was significantly higher following the Hartmann’s procedure, compared to resection with primary anastomosis without diversion (OR: 25.45; with 95% CI 15.13–42.81; p < 0.001; Q = 23.34; p = 0.14; I²= 27.16).

Post-operative complications

There were 40 studies that reported post-operative complications and these were reviewed in detail.^{8–11,14–16,18–23,25,27–52} Post-operative complications were divided into two categories: surgical and medical. The most commonly reported surgically related complications included: intra-abdominal abscess, anastomotic leak, bleeding, wound infection and wound dehiscence. The most commonly reported medically related complications were: pulmonary embolism, pneumonia, urinary tract infection and myocardial infarction.

Overall and individual complication rates

The overall reported post-operative complication rate was 32.64%; with 95% CI 27.43–38.32; p < 0.001. The overall surgical and medical complication rates were 18.96% and 13.93%, respectively.

The individual surgical complication rates were: intra-abdominal abscess, 3.59% with 95% CI 2.69–4.78, p < 0.001; anastomotic leak, 3.99% with 95% CI 3.07–5.18, p < 0.001; bleeding, 2.13% with 95% CI 1.5–3.02, p < 0.001; wound infection, 6.78% with 95% CI 4.17–10.85, p < 0.001; and wound dehiscence, 2.16% with 95% CI 1.06–4.36, p < 0.001.

The individual medical complications were: pulmonary embolism, 0.83% with 95% CI 0.51–1.35, p < 0.001; pneumonia, 3.7% with 95% CI 1.98–6.81, p < 0.001; urinary tract infection, 3.12% with 95% CI 1.67–5.76, p < 0.001; and myocardial infarction, 1.4% with 95% CI 0.81–2.4, p < 0.001.

The post-operative complication rate following emergent surgery was 53.6% with 95% CI 38.82–67.79, p = 0.046; and following elective surgery was 22.52% with 95% CI 18.54–27.07, p < 0.001. The post-operative complication rate following a laparoscopic approach was 22.48% with 95% CI 17.1–28.96, p < 0.001; and following an open approach, was 41.26% with 95% CI 31.32–51.96, p < 0.001. The overall post-operative complication rate following resection with primary anastomosis was 27.62%, with 95% CI 21.56–34.63, p < 0.001; and following the Hartmann’s procedure, it was 40.55% with 95% CI 6.62–87.72, p = 0.092.

We further analyzed the post-operative surgical complications for each surgical procedure (Table 1).

Table 1.

Surgical variable and complication rate.

Surgical variable and complication	Rate (%)	95% CI	p value
Emergent
Abdominal abscess	5.92	4.36–7.98	< 0.001
Anastomotic leak	8.08	3.41–17.97	< 0.001
Bleeding	2.78	1.50–5.10	< 0.001
Wound infection	16.0	8.71–27.54	< 0.001
Pulmonary embolism	1.66	0.66–4.12	< 0.001
Pneumonia	8.41	4.74–14.49	< 0.001
Urinary tract infection	4.36	1.60–11.36	< 0.001
Elective
Abdominal abscess	1.69	0.96–2.95	< 0.001
Anastomotic leak	3.21	2.18–4.71	< 0.001
Bleeding	2.85	1.89–4.27	< 0.001
Wound infection	4.38	2.30–8.16	< 0.001
Pulmonary embolism	0.62	0.25–1.54	< 0.001
Pneumonia	2.27	1.45–3.55	< 0.001
Urinary tract infection	2.87	1.21–6.66	< 0.001
Laparoscopic
Abdominal abscess	1.72	0.99–2.97	< 0.001
Anastomotic leak	2.38	2.15–2.63	< 0.001
Bleeding	2.50	1.35–4.60	< 0.001
Wound infection	3.74	1.53–8.85	< 0.001
Pulmonary embolism	0.68	0.24–1.93	< 0.001
Pneumonia	0.99	0.84–1.17	< 0.001
Urinary tract infection	2.21	1.32–3.66	< 0.001
Open
Abdominal abscess	4.74	3.48–6.42	< 0.001
Anastomotic leak	5.71	3.56–9.04	< 0.001
Bleeding	2.10	1.41–3.10	< 0.001
Wound infection	8.63	4.01–17.62	< 0.001
Pulmonary embolism	1.56	0.85–2.83	< 0.001
Pneumonia	4.96	2.05–11.55	< 0.001
Urinary tract infection	3.41	1.21–9.22	< 0.001
Primary anastomosis
Abdominal abscess	2.12	1.15–3.87	< 0.001
Anastomotic leak	3.99	3.07–5.18	< 0.001
Bleeding	2.72	1.85–3.98	< 0.001
Wound infection	6.34	4.27–9.33	< 0.001
Wound dehiscence	2.01	0.96–4.18	< 0.001
Pulmonary embolism	0.82	0.41–1.64	< 0.001
Pneumonia	3.33	2.29–4.81	< 0.001
Urinary tract infection	2.57	1.55–4.22	< 0.001
Myocardial infarction	1.46	0.82–2.60	< 0.001
Hartmann’s
Abdominal abscess	8.55	6.44–11.27	< 0.001
Bleeding	2.05	0.81–5.07	< 0.001
Wound infection	17.05	8.11–32.39	< 0.001
Wound dehiscence	6.93	3.66–12.76	< 0.001
Pulmonary embolism	2.83	1.38–5.69	< 0.001
Pneumonia	8.29	3.68–17.64	< 0.001
Urinary tract infection	3.22	0.62–15.03	< 0.001
Myocardial infarction	4.76	2.53–8.78	< 0.001

Figure 2 shows the funnel plot for publication bias. The point estimate of the mortality rate out of the number of surgeries in 57 studies showed that there was significant publication bias. The Begg-Mazumdar Test of interdependence of variance and effect size, using the Kendall method, showed bias (Kendall's tau = −0.4825; p < 0.0001). The test for asymmetry of the funnel plot also showed bias (Egger: bias = 0.96343 (95% CI 1.67338–5.53488); p = 0.00044).

Figure 2.

Funnel plot of published articles.

logit: natural logarithm of odds ratio; SE: Standard error.

Discussion

The results of this study showed that there was a substantial risk for mortality after emergent surgery for diverticulitis, which needs to be considered in clinical decision-making. In emergency situations, the estimated mortality was 10%, with the highest rates seen in patients with disease rated at Hinchey 3 and Hinchey 4. On the other hand, elective surgery for diverticulitis was associated with a significantly lower mortality rate of 0.5%. In that latter group, the morbidity after surgery may be a more important factor in clinical decision-making.

Recent guidelines recommend urgent sigmoid colectomy for patients with acute diverticulitis in whom non-operative therapy fails.⁶³ The literature is replete with uncontrolled studies. Given the shortcomings in the literature, the clinician must weigh the risk of surgery with a protracted post-operative course, with a primary colorectal anastomosis performed at the time of the initial procedure; versus the alternative, a proximal diversion procedure that could well become permanent.

Surgical treatment of diverticulitis is in a state of evolution. In 2006, the guidelines of the American College of Colon and Rectal Surgeons stopped recommending surgery after two episodes of diverticulitis.⁶⁴ The most recently published guidelines now recommend that the decision should be individualized.⁶⁵ The decisions that are based on the severity of symptoms, disruption of quality of life and the number of attacks have significant limitations. The guidelines do not provide estimates of how many patients remain symptomatic despite surgery; and the risk of surgical complications is often poorly understood by patients and their caregivers. Therefore, patients are often confused about their choices and some studies suggest that patients expect surgery to cure them of symptoms. Many patients remain symptomatic after surgery; it is unclear whether their symptoms are due to diverticulosis, irritable bowel syndrome or other factors.

Our analysis has some limitations, due to the variable nature of reporting in the studies reviewed. The impact of individual patient variables (age, co-morbidities, gender and number of previous attacks of diverticulitis) could not be assessed, as they were not reported in detail. Similarly, we cannot make any conclusions in this paper about the quality or training of the surgeon, nor the hospital setting (community hospitals versus academic centers), nor the impact of high-volume centers, compared to low volume ones; as these data were not reported.

In the past, surgery was considered for diverticulitis when the patient had suffered two or more episodes of acute diverticulitis. Recent data on the natural history of diverticulitis suggests that only one-third of patients with an attack of diverticulitis will have another attack; and roughly 30% of those whom have two will have another episode later. Most of these episodes retain the same degree of severity as the initial episode. Our study provided a summary of the risks associated with emergent and elective surgery for diverticulitis. While surgery may be inevitable in the complicated cases with diverticulitis, the ambulatory patient with a history of one or two episodes of diverticulitis presents a more difficult challenge in clinical practice. We agree with the views expressed in a recent Italian consensus on diverticulitis, in which the authors state that a recommendation for surgery has to balance the severity of symptoms, the risk of recurrence and the morbidity of surgery.⁶⁶ We hope that the quantitative data on the complications that we have provided in our study will help a clinician managing a patient with diverticulitis to evaluate the risks and benefits of a planned procedure. Quantitative data may also help patients to determine if they wish to undertake the risk of surgery, compared to the risk of another attack of diverticulitis.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest

None declared.