Surgical management of complicated diverticulitis: systematic review and individual patient data network meta-analysis : An EAES/ESCP collaborative project.

BACKGROUND

We performed a systematic review and network meta-analysis (NMA) of individualized patient data (IPD) to inform the development of evidence-informed clinical practice recommendations.

METHODS

We searched MEDLINE, Embase, and Cochrane Central in October 2023 to identify RCTs comparing Hartmann's resection (HR), primary resection and anastomosis (PRA), or laparoscopic peritoneal lavage (LPL) among patients with class Ib-IV Hinchey diverticulitis. Outcomes of interest were prioritized by an international, multidisciplinary panel including two patient partners. Article screening, data extraction for IPD, and risk of bias appraisal were performed by two reviewers. We used a random-effects NMA to synthesize direct and indirect evidence. Heterogeneity was evaluated using the I statistic. The panel appraised the certainty of the evidence using GRADE and CINeMA.

RESULTS

Fourteen reports of seven RCTs were derived from 4,659 articles. IPD data were available for 595/678 patients (88.8%) across trials. Patients had a mean age ± SD of 64.61 ± 13.64 years and a mean BMI ± SD of 26.12 ± 5.20 kg/m, representing Hinchey classes I (1.2%), II (1.0%) III (76.3%), and IV (12.1%), respectively. Using minimal important difference thresholds, in-hospital/30-day mortality was higher among patients receiving LPL versus HR [42 more per 1000, 95% CI (41 fewer to 331 more), moderate effect; low certainty] as well as PRA [45 more per 1000 patients, 95% CI (33 fewer to 340 more) moderate effect; low certainty] without heterogeneity (I = 0%). Among 417 patients from four trials, there was a lower stoma rate among patients receiving PRA versus LPL [539 fewer per 1000, 95% CI (647 fewer to 306 fewer), large effect; low certainty].

CONCLUSION

PRA likely confers a lower stoma rate at 1 year compared to HR, while there may be no difference in 30-day/in-hospital mortality. LPL likely confers a higher in-hospital/30-day mortality rate compared to HR and PRA.