Evaluation and Management After Acute Left-Sided Colonic Diverticulitis: A Systematic ReviewFREE

The Brown EPC used established systematic review methods (12). Detailed descriptions of our methods are presented in our companion article (11) and the full AHRQ report (10). In brief, for the full systematic review (10), we conducted literature searches in multiple databases restricted to 1 January 1990 through 1 June 2020, with updated searches through 16 November 2020. We included studies of adults with a history of acute left-sided colonic diverticulitis. Studies had to evaluate colonoscopy conducted after an episode of resolved acute diverticulitis, prophylactic nonsurgical treatments to prevent recurrence, or elective surgery.

We included comparisons of colonoscopy with other imaging or no imaging, as well as prospective or retrospective single-group studies. Randomized controlled trials (RCTs) needed at least 10 participants per group. Due to time and resource constraints, we included only the largest observational colonoscopy studies (n ≥ 200). On the basis of discussions with stakeholders, we included the following critical outcomes: CRC death, CRC, high-risk colonic premalignant lesions (including adenoma with high-grade dysplasia, adenoma ≥10 mm, villous adenoma, serrated polyp, and ≥3 adenomas per patient), and serious harms of colonoscopy (such as perforation). We also included 2 important clinical outcomes: colonoscopy tolerance and feasibility, and other harms (for example, bleeding).

We searched for pharmacologic treatments and nonpharmacologic interventions, including medical nutrition therapy. We included comparative studies (n ≥ 10 per group) versus other nonsurgical interventions or no intervention (or placebo). For elective surgery, we included comparisons of elective colonic surgery versus no surgery (n ≥ 30 per group). We excluded comparisons of different surgical techniques and delayed surgery for management of acute diverticulitis. Single-group studies were evaluated only for harms, and those assessing elective surgery needed at least 500 participants. We evaluated the following critical outcomes: recurrent diverticulitis, surgery for diverticulitis (for nonsurgical interventions), diverticulitis complications, hospitalization for diverticulitis, quality of life, functional outcomes, and harms or complications of interventions.

Abstracts and full-text articles were screened in duplicate. We extracted data and assessed for methodological quality using a customized form in the Systematic Review Data Repository (https://srdr.ahrq.gov/projects/1520). Study quality was assessed with items from the Cochrane Risk of Bias Tool for RCTs (13); the ROBINS-I (Risk Of Bias In Non-randomised Studies – of Interventions) tool (14); and the National Heart, Lung, and Blood Institute tool (15).

We conducted meta-analyses with the metaan or admetan packages in Stata, version 14.2 (StataCorp). We evaluated odds ratios (ORs) for categorical outcomes and proportions, which were meta-analyzed with the Freeman–Tukey arcsine transformation. We graded the strength of evidence as per the AHRQ Methods Guide (16) and assigned strength of evidence ratings as being high, moderate, low, or insufficient to estimate an effect.

ACP nominated this topic to AHRQ. ACP members joined panels of stakeholders who helped us refine the key questions and protocol. AHRQ program officers, ACP members, and other reviewers (both invited and public) provided comments on draft versions of the protocol and full evidence report. ACP and AHRQ did not participate in the literature search, determination of study eligibility criteria, data analysis, study evaluation, or interpretation of findings. After completion of the AHRQ report (10), we discussed draft versions of the manuscript with ACP members to ensure that it met the needs of the guideline development committee and properly conveyed our conclusions. ACP did not draft any portion of the manuscript or suggest alterations to conclusions.

Overall, 19 studies addressed use of colonoscopy after episodes of acute diverticulitis for the purpose of assessing risk for prevalent CRC. Two of these compared colonoscopy with no colonoscopy in patients with recent diverticulitis (17, 18), 2 compared colonoscopy in patients with recent diverticulitis versus healthy control participants (19, 20), 1 compared early (in-hospital) colonoscopy with later colonoscopy (21), 1 compared colonoscopy with flexible sigmoidoscopy (22), and 13 were single-group studies of patients who underwent colonoscopy (23–35). Details, including risk-of-bias assessments and study results, are provided in Supplements B and C.

High-strength evidence indicates that mesalamine does not reduce risk for recurrence but is not more harmful than placebo (Table 2). Evidence for other interventions (rifaximin, combination mesalamine and rifaximin, combination balsalazide [a 5-aminosalicylic acid prodrug] and probiotics, probiotics, and burdock tea) is too sparse to make conclusions (insufficient evidence). No studies evaluated medical nutrition therapy.

Twelve studies (10 RCTs, 1 NRCS, and 1 single-group study) evaluated nonsurgical interventions to prevent recurrent diverticulitis (details are in Supplement C) (37–46). With few exceptions, all participants had a documented prior episode of acute diverticulitis. Five RCTs were funded by industry (39, 40, 42, 43), and 1 was explicitly not funded by industry (44).

High-strength evidence indicates that elective surgery reduces risk for recurrence of diverticulitis among patients with prior complicated or frequently recurrent uncomplicated diverticulitis but no evidence regarding which of these patients may benefit most from surgery (Table 3). There was low- to moderate-strength evidence that serious adverse events are uncommon with elective surgery, except for required procedures for anastomotic leakage.

Two small RCTs (48–51) and 1 large NRCS (52) with mostly unadjusted analyses evaluated elective surgery (laparoscopic sigmoid colectomy [48], laparoscopic sigmoidectomy [49–51], and colectomy [52]) compared with nonoperative management. Nonoperative management was described as conservative management (49–51), observation (48), or simply nonoperative management (52). Details, including risk-of-bias assessments and study results, are provided in Supplements B and C. The DIRECT trial (49) included patients with uncomplicated disease who had either smoldering symptoms (persisting >3 months) or frequent recurring symptoms (≥3 within 2 years), whereas You and colleagues (48) included patients with a history of complicated diverticulitis manifesting as extraluminal air with or without abscess. The NRCS (52) included 7072 patients with a history of an acute diverticular abscess (complicated diverticulitis). Participant ages were similar across studies, with participants in their mid-50s, and 28% to 54% were male. One RCT was funded by industry (48), the other was not, and the NRCS did not report the funding source. Both RCTs had low risk of bias for randomization, incomplete outcome data, and selective reporting but high risk of bias for blinding. The NRCS reported unadjusted analyses of critical outcomes and was thus at high risk of confounding and selection bias.

The RCTs had only a single death (in a nonoperative group) at 3 and 5 years. The NRCS (52) reported an unadjusted analysis of diverticulitis-related death but found substantially fewer deaths in the elective surgery group (0.2%) at 5 years than the nonsurgical treatment group (1.9%), with an unadjusted OR of 0.13 (CI, 0.03 to 0.29). However, there were important clinical differences between the groups, including that the surgical patients were younger, more likely to be White, less likely to have Medicaid insurance, and less likely to have various comorbidities.

Risk for recurrence was much lower among participants who had elective surgery. The RCT by You and colleagues found an OR for recurrence of 0.18 (CI, 0.04 to 0.80) at 3 years, the DIRECT trial found an OR of 0.29 (CI, 0.11 to 0.79) at 5 years, and the NRCS found an unadjusted OR of 0.13 (CI, 0.10 to 0.17), also at 5 years (Supplement Figure C-12).

The DIRECT trial reported on quality of life and pain in 109 participants. Across 4 scales (Gastrointestinal Quality of Life Index, Short Form-36 mental and physical, and EuroQol-5D), people in the elective surgery group had statistically and clinically greater improvements in quality of life and pain measures at 6 months and 5 years compared with baseline (Supplement Table C-4c-2).

Notably, none of the eligible studies reported subgroup (or similar) analyses to evaluate potential heterogeneity of treatment effect of the benefits of elective surgery.

Serious adverse events associated with elective surgery were reported in 19 studies (2 RCTs [48–51], 1 NRCS [52], and 16 eligible single-group studies [53–69]). In general, composite adverse events were common, but individual adverse event rates were low. However, each specific adverse event was reported by only a small subset of the 19 included studies. Adverse events are summarized in Supplement Table C-4c-4. The summary risks were 0.7% (CI, 0.3% to 1.4%) for 30-day mortality (10 studies; n = 199 915), 4.3% (CI, 2.2% to 6.9%) for anastomotic leakage requiring procedure (6 studies; n = 15 367), 1.6% (CI, 1.0% to 2.3%) for sepsis (7 studies; n = 82 597), 0.7% (CI, 0.1% to 1.6%) for myocardial infarction (5 studies; n = 65 459), and 0.3% (CI, 0.1% to 0.6%) for pulmonary embolism (5 studies; n = 43 818). One study (63) of Medicare beneficiaries, with an overall 30-day death rate of 1.22%, found that the OR for the oldest (≥85 years) versus the youngest (65 to 69 years) age groups was 10.2 (CI, 6.49 to 16.0), and the odds increased for every age group in between.