Articles. Funding Netherlands Organisation for Health Research and Development.

Transcription

1 Surgery versus conservative management for recurrent and ongoing left-sided diverticulitis (DIRECT trial): an open-label, multicentre, randomised controlled trial Bryan J M van de Wall*, Marguerite A W Stam*, Werner A Draaisma, R Stellato, Willem A Bemelman, Marja A Boermeester, Ivo A M J Broeders, Eric J Belgers, Boudewijn R Toorenvliet, Hubert A Prins, Esther C J Consten, on behalf of the DIRECT trial collaborators Summary Background Patients with recurrent or persisting complaints after an episode of left-sided diverticulitis are managed with either conservative measures or elective sigmoidectomy. To date, there are no data from randomised trials. We aimed to establish which leads to a better quality of life for patients with diverticulitis. Methods We did an open-label, multicentre, randomised controlled trial (DIRECT trial) in 24 teaching and two academic hospitals in the Netherlands. Patients aged years presenting with either recurrent (three or more presentations with clinical signs of acute diverticulitis within 2 years) or persistent abdominal complaints (ongoing lower left abdominal pain or persistent change in bowel habits for 3 months) after an episode of left-sided diverticulitis, confirmed by CT, ultrasound, or endoscopy, were included. Patients were excluded if they had previous elective or emergency surgery for acute sigmoid diverticulitis, an absolute operation indication, suspicion of a colorectal malignancy, with a preoperative or postoperative risk greater than III (on the American Society of Anesthesiologists classification), or were unable to complete questionnaire or follow-up. Patients were randomly assigned (3:3) to receive conservative management or elective (laparoscopic) sigmoidectomy using a digital randomisation system, stratified by type of disease and centre, with a block size of six. Patients, physicians, and researchers were not masked to allocation. Our primary endpoint was health-related quality of life, measured by the Gastrointestinal Quality of Life Index (GIQLI) at 6 months after inclusion or surgery, depending on randomisation group. This trial is registered with trialregister.nl, number NTR1478, and is closed for inclusion. Findings Between July 1, 2010, and April 1, 2014, we randomly assigned 109 patients to receive surgical (resection; n=53) or conservative management (n=56), after which the Data Safety and Monitoring Board prematurely terminated the trial because of increasing difficulties in recruitment. 47 (89%) of 53 patients received surgical and 43 (77%) of 56 patients received conservative management. The GIQLI score at 6 months follow-up was significantly higher in patients randomly assigned to receive surgical (mean [SD 22 3]) than conservative management (100 4 [22 7]; mean difference 14 2, 95% CI , p<0 0001). 43 (38%) of 109 patients had a severe adverse event in the first 6 months after (18 [34%] of 53 patients in the surgical group vs 23 [40%] of 57 patients in the conservative group). Seven (15%) patients who received surgical developed anastomotic leakage. Of the 56 patients assigned to be treated conservatively, 13 (23%) ultimately underwent elective resection due to ongoing abdominal complaints, with no anastomotic leakage. We recorded no patient deaths. Interpretation Elective sigmoidectomy, despite its inherent risk of complications, results in better quality of life than conservative management in patients with recurrent and persisting abdominal complaints after an episode of diverticulitis. Funding Netherlands Organisation for Health Research and Development. Lancet Gastroenterol Hepatol 2017; 2: Published Online October 19, S (16) See Comment page 2 *Contributed equally The DIRECT trial collaborators are listed in the appendix Department of Surgery, Meander Medical Centre, Amersfoort, Netherlands (B J M van de Wall PhD, M A W Stam PhD, W A Draaisma PhD, Prof I A M J Broeders PhD, E C J Consten PhD); Julius Center for Health Sciences and Primary Care, Utrecht, Netherlands (R Stellato PhD); Department of Surgery, Academic Medical Centre, Amsterdam, Netherlands (Prof W A Bemelman PhD, Prof M A Boermeester PhD); Department of Surgery, Atrium Medical Centre, Heerlen, Netherlands (E J Belgers PhD); Department of Surgery, Ikazia Hospital, Rotterdam, Netherlands (B R Toorenvliet PhD); and Department of Surgery, Jeroen Bosch Hospital, Den Bosch, Netherlands (H A Prins PhD) Correspondence to: Dr Esther C J Consten, Department of Surgery, Meander Medical Centre, Amersfoort, 3813 TZ, Netherlands ecj.consten@meandermc.nl Introduction Use of elective sigmoidectomy to treat diverticulitis is controversial. About 20% of patients develop recurrences after a conservatively treated episode of diverticulitis. 1 Traditionally, elective resection was advised after a second episode of diverticulitis. Patients with recurrent attacks were thought to be at a 60% risk of developing complications after resection and were less likely to respond to medical. 2 Studies 3,4 have now shown that complications mostly occur during a primary manifestation of diverticulitis. Elective resection as a prophylactic procedure to prevent further complications does not seem warranted. Guidelines 5 published by the American Society of Colon and Rectal Surgeons in 2006 recommend a tailored approach in which the effect of multiple recurrences and ongoing complaints on the quality of life of patients forms the most decisive factor. 6,7 Quality of life is becoming the most important factor in the decision of whether or not elective sigmoidectomy should be done. 8 In addition to recurrences, up to 33% of patients treated conservatively will continue to experience ongoing complaints, such as pain and discomfort, after See Online for appendix Vol 2 January

2 Research in context Evidence before this study We searched PubMed to identify relevant studies from Jan 1, 1980, to Jan 1, We examined studies published after 1980 in an attempt to account for changes in outcome during the past decades. We used the search terms diverticulitis, diverticular disease, medical, and surgical or operative. We searched the reference list of identified articles for further relevant publications. We included original studies that reported data for at least one outcome in medically (conservatively) or surgically treated patients, or both, with left-sided diverticulitis. Studies that focused on right-sided diverticulitis or other colonic diseases were excluded. Our search initially identified 378 studies; 37 studies matched our inclusion criteria and were included, which were mostly retrospective. Six of the 37 studies were prospective, however no randomised trial has yet been published. In the reported literature, both elective surgery and conservative for diverticulitis recurrences are assumed to be effective. However, the retrospective and heterogeneous nature of available studies impairs definitive conclusions. Furthermore, no studies, to our knowledge, took into consideration all of the important factors ie, persisting abdominal symptoms, quality of life, total costs, recurrence, and complication rate. Added value of this study Our study is the first, to our knowledge, to compare both strategies directly with one another and in a randomised fashion. We showed a significantly higher quality of life after sigmoidectomy than after conservative management in patients with chronic abdominal complaints after diverticulitis. Furthermore, almost a quarter of patients initially treated conservatively needed elective resection due to severe complaints. The percentage of anastomotic leakages in our study was quite high; however, this adverse event is expected in patients with diverticular disease. One should note, however, that our trial was terminated prematurely, because of problems with patient accrual. Implications of all the available evidence In keeping with low-quality data, the results from our randomised trial suggest that elective resection for patients with recurrent diverticulitis or with ongoing complaints is significantly better than conservative management. Although procedure-related risks should be taken into account and should be thoroughly discussed with the patient, we believe that, in view of our findings, these risks should not outweigh the benefits of surgery. a first episode of left-sided diverticulitis. 9 Although resection minimises the risk of recurrent disease and relieves most patients from ongoing complaints, the actual effect on quality of life is not well understood. Andeweg and colleagues 9 analysed data from 21 studies of 1858 patients. Their search yielded no direct comparisons of gastrointestinal symptoms or general quality of life between elective resection and conservative management of recurrent diverticulitis. Only low-quality cohort studies were available, indicating a potential benefit for resection. Several studies have suggested an overlap between functional bowel disorders such as irritable bowel syndrome and symptomatic diverticular disease. 10 These uncertainties combined with the risk of complications and mortality have led to reluctance in surgeons and gastroenterologists to offer resection. 11 Both conservative and operative s are used without high-level evidence being available to support either of these strategies. A multicentre, randomised clinical trial on this topic would provide evidence towards a definitive view on the actual effects on quality of life and risks of both strategies nationally. It would create the opportunity to do cost-effectiveness and cost-use analyses, providing health authorities with much needed evidence for decision making. We thus aimed to do a randomised trial to compare conservative management to elective sigmoidectomy in patients with recurring or ongoing complaints after an episode of left-sided diverticulitis. Our primary aim was to determine which strategy results in better quality of life. Methods Study design We did an open-label, multicentre, randomised clinical trial (the DIRECT trial) in 24 teaching hospitals and two academic centres in the Netherlands. All participating centres did at least 50 laparoscopic colorectal resections annually. The study was completed and approved in concordance to the declaration of Helsinki, Good Clinical Practice guidelines, and local ethical standards. Our study protocol was published before the start of the study. 12 Participants We included patients aged years who presented to trial centres with either ongoing abdominal complaints or frequently recurring left-sided diverticulitis after a confirmed (ie, seen with CT scan, ultrasonography, or endoscopy) episode of diverticulitis. Ongoing abdominal complaints were defined as patients with ongoing, lower left-sided abdominal pain, or persistent change in bowel habits for more than 3 months after an episode of diverticulitis. These ongoing complaints had to be accompanied by signs of persistent inflammation on either CT scan (bowel-wall thickening) or colonoscopy (hyperaemia) during outpatient visits. We defined frequently recurring diverticulitis as three or more presentations with clinical signs of acute diverticulitis within 2 years. Radiological examination was not 14 Vol 2 January 2017

3 mandatory to confirm recurrences. A minimal symptomfree interval of 3 months between presentations was mandatory for it to be counted as a recurrence. Exclusion criteria were previous elective or emergency surgery for acute sigmoid diverticulitis, an absolute operation indication (perforation with purulent or fecal peritonitis, symptomatic bowel stenosis, or fistula), suspicion of a colorectal malignancy, and patients classified with a preoperative or postoperative risk of greater than III on the American Society of Anesthesiologists (ASA) classification, 13 as decided by the surgeon or patient specialists. Furthermore, patients unable to fill out a questionnaire or complete the follow-up (eg, due to a psychiatric disease) were excluded. Patients were recruited by local surgeons and gastroenterologists at the participating trial centres. Patients were educated about the trial and had a 3-day reflective period, after which participants provided written informed consent. The Medical Research Ethics Committees United, located in Nieuwegein, Netherlands, provided ethical approval. Randomisation and masking Patients were randomly assigned (3:3) by an independent data manager using a stratified (by type of complaints [ongoing or recurrent] and by centre) digital en-block randomisation system to receive conservative or surgical management. The block size was six. As the trial is open label, allocation was not masked to patients, physicians, or researchers at any timepoint. Procedures Patients randomly assigned to receive elective surgery underwent an elective sigmoidectomy within 3 months after randomisation. Laparoscopic surgery was the preferred approach in all cases. Most guidelines advise that the surgeon place the distal margin of resection where the taenia coli splay out onto the upper rectum. For the proximal margin it was advised to resect in the area of pliable colon without hypertrophy or inflammation. Not all the diverticula-bearing colon had to be removed. A primary anastomosis was intended in all patients who underwent surgery (ie, both patients who were randomly assigned for resection and patients in the conservative group who underwent surgery), but if considered necessary a (temporary) loop ileostomy was created. Patients randomly assigned to receive conservative management were treated according to the current daily practice in the Netherlands; this might consist of intensive monitoring, lifestyle measures, supplementary dietary fibre, and analgesics or laxatives if indicated to be needed by the treating physician. 14 Antibiotics or the use of mesalazine was not routinely advised for use by our study group and guidelines. If patients repeatedly presented to the hospital with severe abdominal complaints, resection could be offered irrespective of assignment. No absolute criteria for severe abdominal complaints were used to define whether a patient was allowed to undergo resection. Instead, we chose a pragmatic approach in which it was left at the patient s and physician s discretion whether or not the symptoms were deemed too unbearable to pursue a conservative strategy during the follow-up period. All data regarding patient demographics and disease characteristics were collected at baseline (ie, at recruitment). The Hinchey classification 15 was used to classify the primary episode of diverticulitis and was based on findings of either CT scan or ultrasonography. Outcomes Our primary endpoint was health-related quality of life, measured by the Gastrointestinal Quality of Life Index (GIQLI; appendix p 1). 16 The GIQLI is a questionnaire containing 36 questions on gastrointestinal complaints. All questions are attributed a score from no points for the least desirable option, to 4 points for the most desirable option. The maximal obtainable score is 144, reflecting an optimal quality of life without any symptoms. Patients were asked to fill out the questionnaires at inclusion. At 3 months and 6 months after inclusion or surgery (if randomly assigned to the elective surgery group), patients were sent these questionnaires by postal mail to complete. Secondary endpoints included additional quality-of-life assessments using the EuroQol five dimensions questionnaire (EQ-5D), Visual Analogue Score for pain (VAS-pain), and the 36-item Short Form health survey (SF-36) sent to participants via postal mail at the same timepoints as the GIQLI questionaire. The EQ-5D is a standardised instrument to measure generic health status. The health state is converted to one summary index with application of a formula that essentially attaches weights to each of the levels in each dimension. This formula is based on the valuation of EQ-5D health states from general population samples. The score ranges from 0, reflecting worst health state, to 1, corresponding with the best possible health state. The VAS-pain is a visual analogue scale that measures the millimetres of pain. A score of 0 corresponds with no pain and a score of 100 represents the worst possible pain imaginable. The SF-36 consists of eight scaled scores, which are the weighted sums of the questions in their section. Each scale is directly transformed into a scale on the assumption that each question carries equal weight. The lower the score the more disability, and the higher the score the less disability ie, a score of 0 is equivalent to maximum disability and a score of 100 is equivalent to no disability. Other secondary outcomes included mortality (30-day mortality and mortality associated with development of complications of diverticulitis during follow-up) and morbidity (defined as recurrence, perforation, fistula, symptomatic stenosis, abscess, stoma formation, emergency surgery or re-operation, and perioperative and Vol 2 January

4 postoperative complications). Short-term and long-term outcomes of surgery were recorded at hospital discharge, outpatient visits during follow-up, and after 6 months of follow-up, including perioperative complications and any additional admissions to hospital and health-care consumption of all included patients. The Clavien-Dindo classification 17 was used to rank perioperative in-hospital complications in an objective and reproducible manner. Direct and indirect health costs were additional secondary outcomes; these data will be presented elsewhere. Statistical analysis We based the sample size calculation on the minimum clinically important difference (MID) of the GIQLI score. The MID can be estimated by taking half an SD of a quality-of-life instrument. 18,19 On the basis of studies by Forgione and colleagues 20 and Zdichavsk and colleagues, 21 the MID of the GIQLI score was estimated at 10 points. Both studies also showed that GIQLI scores improved by around 10 points between 1 month and 1 year after elective resection; 20,21 thus, a difference of 10 points corresponds with the MID and the expected improvement after elective resection. To show this difference by use of an independent t test (α=0 05, δ=10, ϛ=21, 1 β=0 9), we needed about 97 patients per group (194 in total) for this study. To compensate for a potential loss to follow-up of 10%, we calculated that 214 patients was the optimal sample size needed. Source data verification was done by an independent data monitor affiliated with our data management bureau (the Factory, Bilthoven, the Netherlands). At the interim analysis, data for 25% of control participants were verified, and for final analysis, 100% of data were verified. The monitor verified informed consent provided by patients, inclusion criteria, adverse events, and adherence to Good Clinical Practice guidelines, using the resources available (eg, patient charts at the participating hospital) according to a predefined monitoring plan. An independent Data and Safety Monitoring Board monitored the progress of the study and ensured that the safety of patients enrolled in the study was not compromised. Safety was tested at regular intervals according to predefined stopping rules. We analysed data according to the intention-to-treat principle. We analysed the difference in quality-of-life outcomes between the two groups using a mixed model with repeated measures over time and included all available data from patients for the first 6 months after randomisation. The variance-covariance matrix was modelled as unstructured. The fixed effects were time after randomisation (categorical), group, a group time interaction, and baseline GIQLI score. We tested the difference in GIQLI scores at 6 months after randomisation using a linear contrast from this model. If at least 75% of the items were filled in, we calculated a score based on the mean of the completed items. To test whether the effect of on the GIQLI score was different for the recurrent and ongoing groups, a fully stratified model (with interactions for time, group, and time with stratification group) was compared with the previous model. A likelihood ratio test (LRT) was used to compare the models with and without interactions for stratification group. We did an additional per-protocol analysis (in the population who received assigned ) to test whether protocol violations had any effect on the primary outcome. All analyses were done with SAS (version 9.2). This trial is registered with trialregister.nl, number NTR1478, and is closed for inclusion. Role of the funding source The funder of the study reviewed the study protocol but had no role in study design, data collection, data analysis, data interpretation, or writing of the report. BJMW, MAW, WAD, and ECJC had direct access to all the data in the study. All authors had final responsibility for the decision to submit for publication. Results The recruitment period was from July 1, 2010, to the premature termination date of April 1, The trial was terminated early on the basis of the advice of the Data and Safety Monitoring Board. The main reason for termination was the increasing difficulty of recruitment of patients. At termination, we had screened 431 patients, including patients found by case-finding exercises (ie, chart review in all participating centres to identify eligible and missed patients) implemented when difficulties in recruitment were noted. 109 (25%) were included and randomly assigned to receive surgical (n=53 [49%]) or conservative (n=56 [51%]; figure). Main reasons for not participating in the study were personal preference for one of the strategies (among potential participants and physicians, 34 had a preference for conservative management whereas 132 had a preference for surgical management) and not meeting the definitions for ongoing or recurring complaints (156 patients). In the elective surgical group, six (11%) of 53 patients ultimately declined surgery and were treated conservatively. In the conservative management group, 13 (23%) of 56 patients underwent surgery due to severe ongoing abdominal complaints during the study period. However, all patients were analysed in their original randomisation group, according to the intention-to-treat analysis. In accordance with the amended protocol, the α-spending approach proposed by Lan and DeMets 22 was applied to O Brien-Fleming stopping boundaries. 23 At the time of analysis, 140 GIQLI measurements at 3 months or 6 months after randomisation were available from 85 participants, whereas the expected 16 Vol 2 January 2017

5 431 patients assessed for eligibility 109 randomly assigned 53 allocated to receive surgical 47 recevied surgical 6 did not receive surgical 2 preference 2 family circumstances 2 administrative* 53 included in intention-to-treat analysis 23 missing 3 months follow-up data 13 missing 6 months follow-up data 232 excluded 166 preference 156 did not meet ongoing or recurring diverticulitis criteria 56 allocated to receive conservative 43 recevied conservative 13 did not receive conservative 13 severe complaints requiring surgery 56 included in intention-to-treat analysis 4 missing 3 months follow-up data 6 missing 6 months follow-up data Figure: Trial profile *During analysis of the results we identified two patients who had been randomly assigned to receive elective resection, but had received a conservative. total information from all 214 patients to be included was 428 GIQLI measurements (two completed questionnaires per patient for both timepoints). Therefore, the information fraction (τ) available for the interim analysis was (140/428), resulting in an O Brien-Fleming alpha (α OB(τ) ) of No statistically significant difference between randomised groups was found in terms of the number of GIQLI measurements available at 6 months. Due to α spending at the interim analysis, all tests for the primary and secondary outcomes in the final analysis were completed at a (ie, ) level of significance. Baseline characteristics of patients are shown in table 1. The median age of the analysed patients was 55 2 years (IQR ) with 39 (36%) of 109 being men (table 1). The primary episode of diverticulitis was seen by use of CT scan in 87 patients. Ultrasonography was used for 15 (14%) patients at presentation and one received an endoscopy to confirm the primary episode. On the basis of the findings of the CT scans and ultrasonography, a Hinchey I episode was found in 84 patients and Hinchey II episode in 16 patients (table 1). One patient had a Hinchey III episode, who received a peritoneal lavage. No resection was done. One patient had a Hinchey IV episode, and received resection of the transverse colon. The sigmoid colon was not resected in this patient. Four (4%) patients could not be classified because either an endoscopy was undertaken (n=1) or the Hinchey classification was not described in the radiological report (n=3). Surgical group (n=53) Conservative management group (n=56) Sex Male 15 (28%) 24 (43%) Female 38 (72%) 32 (57%) Age (years) 54 1 ( ) 56 5 ( ) Body-mass index 28 7 (4 7) 27 8 (4 9) ASA classification I 26 (49%) 26 (46%) II 27 (51%) 29 (52%) III 0 1 (2%) Type of complaint Recurring diverticulitis 17 (32%) 23 (41%) Ongoing complaints 36 (68%) 33 (59%) Number of episodes 3 1 (1 0) 4 1 (2 0) Duration of symptoms (weeks)* 31 0 ( ) 32 0 ( ) Hinchey classification of primary episode H1 35 (66%) 49 (88%) H2 10 (19%) 6 (11%) H3 1 (2%) 0 H4 1 (2%) 0 Imaging studies CT 40 (75%) 47 (84%) Ultrasound 9 (17%) 6 (11%) Colonoscopy 41 (77%) 43 (77%) Medications Pain medication 13 (25%) 14 (25%) Laxatives 22 (42%) 28 (50%) Quality of life VAS-pain 63 3 (21 7) 69 3 (13 6) GIQLI 92 6 (22 8) 92 2 (21 3) Data are n (%), mean (SD), or median (IQR). ASA=American Society of Anesthesiologists. VAS-pain=Visual Analogue Scale for pain (score range 0 100). GIQLI=Gastrointestinal Quality of Life Index (score range 0 144). *Applies only to patients with ongoing complaints. Six patients in the surgical group and one patient in the conservative management group could not be classified because either an endoscopy (n=4) was done or the Hinchey classification was not described in the radiological report (n=3). Hinchey I includes both patients with pericolic inflammation (modified Hichey Ia) and patients with pericolic abscess (modified Hinchey Ib). 24 Although we did not discriminate between the two in our study, most patients only had pericolic inflammation without abscess. Some patients underwent more than one diagnostic examination. Some patients did no receive medication. Table 1: Baseline characteristics More patients had ongoing complaints than recurrent complaints in both groups (table 1). Patients with recurrent diverticulitis had a mean of 4 0 episodes (SD 1 4) before inclusion in the study. Patients with ongoing complaints had a median duration of 31 weeks (IQR 16 81) of complaints before inclusion. In patients with ongoing complaints, as stated in the protocol, it was necessary to find proof of ongoing infection. Among patients with ongoing abdominal complaints, persistent wall thickening and infection was noted by CT scan in 18 of 33 patients in the conservative group and 20 of Vol 2 January

6 Surgical group (n=53) Conservative management group (n=56) p value* GIQLI 0 months 92 6 (22 8) 92 2 (21 3) 3 months (23 8) (22 1) 6 months (22 3) (22 7) VAS-pain 0 months 63 3 (21 7) 69 3 (13 6) 3 months 37 4 (27 7) 48 2 (25 9) 6 months 23 9 (23 4) 48 3 (22 9) < SF-36 physical 0 months 37 0 (7 1) 36 9 (6 7) 3 months 40 0 (8 6) 38 3 (7 7) 6 months 43 5 (8 8) 39 5 (7 0) SF-36 mental 0 months 41 6 (14 5) 43 3 (9 5) 3 months 44 4 (11 1) 44 5 (10 3) 6 months 47 7 (12 4) 45 3 (10 3) EQ-5D 0 months 0 69 (0 22) 0 74 (0 20) 3 months 0 75 (0 24) 0 75 (0 18) 6 months 0 84 (0 20) 0 73 (0 19) Data are mean (SD), unless otherwise specified. GIQLI=Gastrointestinal Quality of Life Index (score range 0 144). VAS-pain=Visual Analogue Scale for pain (score range 0 100). SF-36=Short Form 36 (score range 0 100). EQ-5D=EuroQol-5D (score range 0 1). *The p value is from the difference between the two groups, based on a contrast from a mixed model. Table 2: Quality of life during 6 months follow-up 36 patients in the surgical group. At colonoscopy, among patients with ongoing abdominal complaints, persistent wall thickening and infection were present in five of the 33 patients in the conservative group and eight of 36 patients in the surgical group, whereas redness and hyperaemia were present in ten of the 33 patients in the conservative group and eight of 36 patients in the surgical group. Mean GIQLI score at 6 months follow-up was significantly higher in the surgical group (114 4 [SD 22 3]) than the conservative group (100 4 [SD 22 7]) with a mean difference of 14 2 (95% CI , p<0 0001; table 2). The GIQLI score improved by more than 10 points (indicating a clinical measurable difference) in 37 (70%) of 53 patients in the surgical group versus 19 (34%) of 56 patients in the conservative management group. However, in 13 (25%) of patients in the surgical group versus 31 (55%) patients in the conservative management group the score improved or deteriorated by fewer than 10 points, indicating no clinically measurable difference. Three (6%) patients in the surgical group versus six (11%) in the conservative management group had a lower (10 points or more) score at 6 months follow-up than at baseline (data not shown). After stratification for ongoing and recurrent complaints, the effect of sigmoidectomy on GIQLI score for the recurrent group was an increase from baseline of 15 2 points (95% CI ) and for the ongoing group an increase of 13 5 points ( ). We found no evidence for a difference in effect between the ongoing and recurrent group (LRT p=0 9861; data not shown). In our analysis of the primary endpoint in the per-protocol population (surgical group [n=47] vs conservative group [n=43]), mean GIQLI score at 6 months in the surgical group was (SD 21 7) and was (20 7) in the conservative management group (mean difference 15 8 [95% CI ]). An improvement of more than 10 points was reported in 34 (72%) patients in the surgical group versus 13 (30%) patients for the conservative group. More patients in the conservative management group than in the surgical group had a GIQLI score that improved or deteriorated by less than 10 points (26 [60%] in conservative management group vs 11 [23%] in surgical group) and deteriorated by more than 10 points at 6 months follow-up (four [9%] vs two [5%]). The secondary endpoint of EQ-5D (p=0 0013), VAS-pain score (p<0 0001), and SF-36 physical score (p=0 016) were all significantly better in the surgical group than the conservative management group at 6 months (table 2). Only the SF-36 mental health score (p=0 2631) was not significantly different between the two groups, with a lower score in the conservative management (45 3 [10 3]) than the surgical group (47 7 [12 4]) after 6 months (table 2). 41 (38%) of 109 patients had one or more severe adverse events within the first 6 months in the study: 18 (34%) of 53 patients had one or more severe and non-severe adverse events in the surgical group, compared with 23 (40%) of 57 patients in the conservative management group. No deaths occurred during the 6 months of follow-up in either group. 38 severe and non-severe adverse events occurred in the surgical management group (appendix p 6). Perioperative complications among patients in the surgical group included anastomotic leakage (n=7), ileus (n=4), intra-abdominal abscess (n=2), wound infection (n=2), fistula (n=1), cardiac arrest during surgery (n=1), intraluminal bleeding (n=1), and wound dehiscence (n=1). Other complications in this group included electrolyte deficiencies (n=3), delirium (n=1), urinary tract infection (n=4), respiratory insufficiency (n=1), and hospital presentations un related to disease (n=6). Table 3 describes in-hospital operative complications according the Clavien-Dindo classification. One patient in the surgical group had a recurrence shortly after inclusion but before the scheduled resection. During follow-up in the surgical management group, one patient developed a trocar site hernia and two patients had an anastomotic stenosis Vol 2 January 2017

7 A total of 30 severe and non-severe adverse events occurred in the conservative management group (appendix p 6). Seven (13%) patients developed a diverticulitis recurrence during the 6 months of follow-up. All recurrences were treated conservatively. 13 (23%) patients did not respond to conservative management and ultimately required elective sigmoidectomy due to severe abdominal complaints. Perioperative complications in this group included acute coronary syndrome (n=1) and a negative diagnostic re-laparoscopy for suspicion of anastomotic leakage (n=1; table 3). Other adverse events during follow-up included urinary tract infection (n=2), anal blood loss treated conservatively (n=2), pulmonary embolism (n=1), and other hospital presentations not related to disease (n=3). Overall 60 patients underwent elective sigmoid resection during the study period (47 in surgical group and 13 in conservative management group; table 3). In 57 (95%) of 60 patients a laparoscopic approach was used. Conversion to an open procedure was needed in three (5%) patients; one patient because of adhesions and in the other two patients no reason was noted. A stoma was created in ten (21%) of the 47 operated patients in the surgical group versus one (8%) of 13 operated patients in the conservative management group. During the 6 months follow-up, nine (82%) of 11 stomas were reversed. All resected specimens from the 60 patients who underwent resection were postoperatively sent for pathological examination (table 3). Active inflammation was found in 36 (60%) patients. In seven (10%) patients other inflammation-related parameters (eg, abscesses, covered perforation, and fibrosis) were clearly present. Discussion Although prematurely terminated because of problems with recruitment, our randomised trial showed significantly better disease-specific and general quality of life after elective sigmoidectomy than with conservative management for patients with multiple recurrences or ongoing complaints after a first episode of diverticulitis. Recurrences and severe abdominal complaints occurred in more patients who were randomly assigned to receive conservative management than in those who underwent surgical resection. As a result, 13 (23%) patients who were initially managed conservatively ultimately required sigmoidectomy. The effect of elective resection and conservative management on patient-reported outcomes in patients with diverticulitis has previously been poorly studied. In 2016 a systematic review and meta-analysis was published by Andeweg and colleagues 9 on this matter. No comparative studies had generic or disease-specific quality of life as endpoints. The remaining studies published so far were non-comparative cohort studies. As summarised by Andeweg and colleagues, 9 patients treated surgically reported a higher mean SF-36 score (73 4 [ ] vs 58 2 [ ]) and a lower prevalence of Surgical group (n=47)* Conservative management group (n=13) Surgical technique Laparoscopy 46 (98%) 12 (92%) Open surgery 1 (2%) 1 (8%) Conversion 3 (6%) 0 Stoma 10 (21%) 1 (8%) 3 months follow-up 5 (11%) 1 (8%) 6 months follow-up 1 (2%) 1 (8%) Time inclusion to surgery Month 1 13 (28%) 3 (23%) Month 2 21 (45%) 6 (46%) Month 3 12 (26%) 1 (8%) Month (15%) Month 5 1 (2%) 0 Month (8%) Postoperative care Surgical ward 45 (96%) 12 (92%) Duration 6 0 ( ) 5 0 ( ) Intensive care unit 2 (4%) 0 Duration 0 3 ( ) 0 0 ( ) Perioperative complications Re-interventions 13 (28%) 0 Clavien-Dindo Grade I 13 (28%) 0 Grade II 5 (11%) 1 (8%) Grade IIIa 2 (4%) 0 Grade IIIb 9 (19%) 1 (8%) Grade IV 2 (4%) 0 Grade V 0 0 Grade d 0 0 Leakage 7 (15%) 0 Pathological evaluation Inflammation 30 (64%) 6 (46%) Abscesses 1 (2%) 0 Covered perforation 0 1 (8%) Fibrosis 4 (9%) 1 (8%) Data are n (%) or median (IQR). *47 (89%) of 53 patients received surgery in the surgical group. 13 (23%) of 56 patients received surgery in the conservative management group. Absolute number of events per group. In-hospital perioperative complications. Table 3: Perioperative data of patients who required resection during follow-up chronic abdominal pain (11% [95% CI 1 21] vs 38% [19 56]) than those treated conservatively. Additionally, all studies included in the review 9 had a high risk of bias. Our study is the first, to our knowledge, to directly compare both strategies in a randomised fashion. Our primary endpoint, the GIQLI score, is a commonly used and a well-validated questionnaire for gastrointestinal complaints. A 10-point difference in quality of life on this scale is considered to be clinically relevant. 16,18,19 We showed a difference of 14 points, Vol 2 January

8 indicating a clinically measurable difference in quality of life in favour of resection. The difference in quality of life between these two groups is further backed up by the significant differences we reported in the other questionnaires used, except for the SF-36 mental health questionnaire. On the basis of our results and the existing evidence described in the systematic review, 9 elective resection seems to be more effective than conservative measures for patients with persisting or recurring complaints after an episode of diverticulitis. Chronic abdominal complaints after an episode of diverticulitis show great overlap with irritable bowel syndrome. Studies have shown a potential role of low-grade inflammation, altered motility, visceral hypersensitivity, and altered intestinal microbiota in patients with persistent complaints after diverticulitis. Irritable bowel syndrome has many of the same underlying pathophysiological mechanisms. Some 10 have suggested that diverticulitis might cause irritable bowel syndrome in a similar way to post-infectious irritable bowel syndrome. Obviously, surgery is not the appropriate alternative for this functional disorder because it is unlikely to improve symptoms. However, the results of our study undermine this rationale. We reported a significantly higher quality of life after sigmoidectomy in patients with chronic abdominal complaints after diverticulitis than in those treated with conservative management. The results of our study suggest that ongoing symptoms after an episode of diverticulitis are not caused by functional bowel disorders. Symptoms seem to have a more organic cause, originating from the diseased colonic segment. Therefore, these patients are suitable for resection. Inflammation was noted in only 60% of all resected sigmoid specimens at pathological evaluation. This figure could partly be explained by the fact that among the resected specimens, there were also specimens of patients categorised as having recurring diverticulitis. Resection was probably completed when no active inflammation was present in other words in between recurrences. Furthermore, in patients with chronic complaints, some are likely to have had resection during a point when their chronic inflammation was at its lowest grade and therefore not detectable at pathological examination. We reported an incidence of 12% anastomotic leakages in all operated patients and a protective stoma rate of 18%. These figures are rather high compared with published literature 30,31 rates of 1 4% for anastomotic leakage and 5 3% for protective stoma. The high incidence can only be explained by the fact that we did a real-life study without specific selection of surgeons or centres. 30,31 We could not do a formal regression analysis to establish which factors are associated with the high leakage and protective stoma rate, because the absolute number of patients was too small. Taking into account the results of our study and available literature, should elective resection be undertaken in patients with recurrent and persistent complaints after an episode of diverticulitis? Is an improvement of 14 points on the GIQLI score worth the risks of surgery? Conservative management is often successful for treating primary and recurrent episodes of diverticulitis. However, recurrences occur in 20% of patients and 36% have persistent abdominal complaints, both having an substantial effect on patient quality of life, as shown in our study and the existing literature. 1,9 So far, there are no s aimed at preventing these recurrences and persistent complaints. Conservative management consists mainly of supportive care. 14 Sigmoidectomy might be a viable option. Both literature 9,20 and this trial show a significant improvement in quality of life after surgery. Risks of recurrences are minimised, but are not fully excluded. Persistent abdominal complaints after surgery are less common than with conservative management (9% vs 36%). 9 Surgery inevitably has several disadvantages. A potential harmful procedure is done in a non-lethal disease. Results of surgery have greatly improved in terms of morbidity and mortality since the introduction of laparoscopic surgery. However, as noted above, anastomotic leakage still occurs. Mortality is rare after surgery and ranges from 0% to 0 2%. 30 Recurrences and persistent abdominal complaints still occur in a small proportion of patients. Additionally, the translational step from quality-of-life scores to a clinical setting remains problematic. Knowledge of the minimum important difference can aid interpretation of quality-oflife results in terms of clinical utility, as in our study. However, we cannot fully grasp the true implications of quality-of-life studies for patients in a real-life setting. Nevertheless, many studies 1,9 have previously investigated surgery for recurrent and persistent complaints, with most having described a beneficial effect. We therefore strongly feel that patients, as described in this study, should be considered for resection. Nevertheless, procedure-related risks should be taken into account and thoroughly discussed with the patient during the decision-making process. Our study has several limitations that need consideration. We had to stop the trial prematurely and were not able to reach the targeted sample size. Published scientific literature suggests that termination of a study might lead to overestimation of effects. 32 This problem frequently affects trials that have been terminated early due to benefit. Although our study was stopped due to issues with patient recruitment (and not due to benefit) and results have been corrected for overestimation by adjustment of p value threshold to determine whether results were significant, overestimation cannot be fully ruled out. A further limitation is that there might be a possible placebo effect of surgery. Jonas and colleagues 33 did a meta-analysis on trials comparing invasive surgery to sham procedures. They concluded that the placebo effect accounted for 78% of improvement in patients with chronic pain conditions including endometriosis, 20 Vol 2 January 2017

9 back pain, arthritis, angina, and migraine. 33 However, establishing which of the observed outcomes from sigmoidectomy in this study are due to the specific efficacy of the or to other factors is difficult. Additionally, there were differences in baseline characteristics between groups in terms of the Hinchey classification of the primary episode, most likely due to chance by randomisation. Another potential limitation is the nature of our control group. Controversy exists regarding non-surgical management for prevention of recurrent and persistent complaints after an episode of diverticulitis. Although increasing evidence shows that antibiotic has no beneficial effect on recovery, complications, and recurrences, the use of mesalazine remains a viable option. 34 We did not include the standard use of mesalazine in the conservative management group because its use is not advocated in the Dutch guidelines for of diverticulitis. 7 However, many randomised trials assessed the role of mesalazine for of diverticulitis with variable success. 35,36 In the future, mesalazine (or any other non-surgical measure) might have a more dominant role in preventing recurrent and persistent complaints after an episode of diverticulitis. Furthermore, our inclusion period was relatively long. As regimens and surgical techniques are developed, this inclusion period might affect applicability of results. To the best of our knowledge, however, no major changes to guidelines 7,37 have been made since the start of this trial. Although our study showed a significant difference of 14 2 points (95% CI ) on the GIQLI score between groups, the MID of 10 points still lies within the boundaries of the confidence interval. This result implies that the difference between the groups is statistically significant, but it might not be clinically relevant. However, our study was prematurely terminated after including about only half (109 patients) of the targeted 214 patients, which may have increased the size of the confidence intervals. Of note, the 13 patients who were randomly assigned to receive conservative management but ultimately underwent resection had a smaller interval between resection and their quality of life assessment at 6 months because these assessments were taken at 3 months and 6 months after inclusion, irrespective of whether they ultimately required surgery. There might have been a transient effect of surgery on abdominal symptoms and quality of life, however the number of patients involved is small. We had much discussion on whether to analyse data at 6 months or 1 year s follow-up. Differences in quality of life are best measured at 6 months follow-up to balance and mitigate the confounding effects of surgery on quality of life before this timepoint and time itself after this timepoint. After this point, time may have a substantial effect on the quality-of-life scores, making effects less detectable. For this reason, we chose to analyse quality of life at 6 months. Several issues exist with regard to the generalisability of our results. First, the inclusion criteria were quite strict, so results are therefore not applicable to all patients with frequently recurring diverticulitis or persistent complaints after an episode in general. Second, a pragmatic approach, adhering to the Dutch guidelines, was used to define conservative management and to confirm the primary episode of diverticulitis by use of CT scan, ultrasonography, or endoscopy. Discrepancies exist between different countries with regard to these topics, which should be taken into account in interpretation of our results. Further, one could debate whether patients with recurrent diverticulitis and those with ongoing complaints should both have been eligible for this study. Some authors state that these are different entities and therefore cannot be compared. 11,38 However, in our trial both patients with ongoing complaints and those with recurrences had significant benefit in terms of quality-of-life outcomes. Stratification on the nature of complaints did not affect the primary outcome. Trials comparing surgery with non-surgical management are notoriously difficult. Patients and clinicians both have their own individual preferences, making counseling and inclusion difficult. Patient recruitment for this study was time consuming and challenging due to the elaborate patient information necessary to explain the randomisation between the surgery and conservative management groups. We feel that this aspect makes this study unique and difficult to repeat. We conclude that elective sigmoidectomy, despite the inherent risk of complications, leads to improved quality of life compared with conservative management for patients with recurrent and persisting abdominal complaints after an episode of left-sided diverticulitis. Futhermore, we noted that it is likely that around a quarter of patients initially treated conservatively will require elective resection due to severe complaints. Delay to choose a surgical option did not negatively affect surgical outcome. About one in five patients who underwent surgery needed a second operation for stoma reversal. Taken together, these new data suggest that elective resection is an effective option for patients with recurrent and persisting abdominal complaints after an episode of left-sided diverticulitis. Long-term data are not yet available, but will be published in the future. Contributors BJMW was involved in study design, completion of the trial, data interpretation, and writing of the report. MAWS was involved in completion of the trial, data analysis, data interpretation, and writing of the report. WAD, IAMJB, and ECJC were involved in study design, data interpretation, and writing of the report. RS was involved in data analysis, interpretation, and writing of the report. WAB and MAB were involved in study design and the writing of the report. EJB, BRT, and HAP were involved in significant data acquisition. Declaration of interests We declare no competing interests. Vol 2 January

10 Acknowledgments We thank the Netherlands Organisation for Health Research and Development (ZonMw) for supplying the grant that made this trial possible. We thank all staff of the participating hospitals of the DIRECT trial for their efforts, and especially the patients for their participation. The DIRECT trial is part of a national consortium, the Dutch Diverticular Disease Collaborative Study group. This group consists of researchers of the Ladies, DIABOLO, and DIRECT trials (three diverticulitis trials in the Netherlands). References 1 Peppas G, Bliziotis LA, Oikonomaki D, Falagas ME. Outcomes after medical and surgical of diverticulitis: a systematic review of the ava ilable evidence. J Gastroenterol Hepatol 2007; 22: Köhler L, Sauerland S, Neugebauer E. Diagnosis and of diverticular disease: results of a consensus development conference. The Scientific Committee of the European Association for Endoscopic Surgery. Surg Endosc 1999; 13: Anaya DA, Flum DR. Risk of emergency colectomy and colostomy in patients with diverticular disease. Arch Surg 2005; 140: Moreno AM, Wille-Jørgensen P. Long-term outcome in 445 patients after diagnosis of diverticular disease. Colorectal Dis 2007; 9: Rafferty J, Shellito P, Hyman NH, et al. Standards Committee of American Society of Colon and Rectal Surgeons. Practice parameters for sigmoid diverticulitis. Dis Colon Rectum 2006; 49: Vennix S, Morton DG, Hahnloser D, Lange JF, Bemelman WA, on behalf of the Research Committee of the European Society of Coloproctocology. Systematic review of evidence and consensus on diverticulitis: an analysis of national and international guidelines. Colorectal Dis 2014; 16: Andeweg CS, Felt-Bersma R, Verbon A, Stoker J, Boermeester M, Bleichrodt R. Summary of the practice guideline on diverticulitis in the colon: diagnostics and in specialty care. Ned Tijdschr Geneeskd 2013; 157: A6124 (in Dutch). 8 Pasternak I, Wiedemann N, Basilicata G, Melcher GA. Gastrointestinal quality of life after laparoscopic-assisted sigmoidectomy for diverticular disease. Int J Colorectal Dis 2012; 27: Andeweg CS, Berg R, Staal JB, Ten Broek RP, van Goor H. Patient-reported outcomes after conservative or surgical management of recurrent and chronic complaints of diverticulitis: systematic review and meta-analysis. Clin Gastroenterol Hepatol 2016; 14: Cohen E, Fuller G, Bolus R, et al. Increased risk for irritable bowel syndrome after acute diverticulitis. Clin Gastroenterol Hepatol 2013; 11: Senapati A. The surgeon s view. Dig Dis 2012; 30: Van de Wall B, Draaisma W, Consten E, et al. DIRECT trial. Diverticulitis recurrences or continuing symptoms: operative versus conservative. A multicenter randomised clinical trial. BMC Surg 2010; 10: Fitz-Henry J. The ASA classification and peri-operative risk. Ann R Coll Surg Engl 2011; 93: Andeweg CS, Felt-Bersma R, Verbon A, Stoker J, Boermeester M, Bleichrodt R. Summary of the practice guideline on diverticulitis in the colon: diagnostics and in specialty care. Ned Tijdschr Geneeskd 2013; 157: A6124 (in Dutch). 15 Kaiser AM, Jiang JK, Lake JP, et al. The managment of complicated diverticulitis and the role of computed tomography. Am J Gastroenter 2005; 100: Eypasch E, Williams JI, Wood-Dauphinee S, et al. Gastrointestinal Quality of Life Index: development, validation and application of a new instrument. Br J Surg 1995; 82: Dindo D, Demaetines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004; 240: Cohen J. Statistical power analysis for the behavioural sciences, 2nd edn. New Jersey: Lawrence, Erlbaum Associates, Norman GR, Sloan JA, Wyrwich KW. Interpretation of changes in health-related quality of life. The remarkable universality of half a standard deviation. Med Care 2003; 41: Forgione A, Leroy J, Cahill RA, et al. Prospective evaluation of functional outcome after laparoscopic sigmoid colectomy. Ann Surg 2009; 249: Zdichavsk M, Granderath FA, Blumenstock G, et al. Acute laparoscopic intervention for diverticular disease (AIDD): a feasible approach. Arch Surg 2008; 395: Lan KG, Demets DL. Discrete sequential boundaries for clinical trials. Biometrika 1983; 70: O Brien PC, Fleming TR. Multiple testing procedure for clinical trials. Biometrics 1979; 35: Hinchey EJ, Schaal PG, Richards GK. Treatment of perforated diverticular disease of the colon. Adv Surg 1978; 12: Humes DJ, Simpson J, Smith J, et al. Visceral hypersensitivity in symptomatic diverticular disease and the role of neuropeptides and low grade inflammation. Neurogastroenterol Motil 2012; 24: 318 e Horgan AF, McConnell EJ, Wolff BG, The S, Paterson C. Atypical diverticular disease: surgical results. Dis Colon Rectum 2001; 44: Bassotti G, Sietchiping-Nzepa F, De Roberto G, Chistolini F, Morelli A. Colonic regular contractile frequency patterns in irritable bowel syndrome: the spastic colon revisited. Eur J Gastroenterol Hepatol 2004; 16: Clemens CH, Samsom M, Roelofs J, van Berge Henegouwen GP, Smout AJ. Colorectal visceral perception in diverticular disease. Gut 2004; 53: Tursi A, Brandimarte G, Giorgetti GM, Elisei W. Assessment of small intestinal bacterial overgrowth in uncomplicated acute diverticulitis of the colon. World J Gastroenterol 2005; 11: Jones OM, Stevenson AR, Clark D, et al. Laparoscopic resection for diverticular disease: follow-up of 500 consecutive patients. Ann Surg 2008; 248: Royds J, O Riordan JM, Eguare E, O Riordan D, Neary PC. Laparoscopic surgery for complicated diverticular disease: a single-centre experience. Colorectal Dis 2012; 14: Bassler D, Briel M, Montori VM, et al. Stopping randomized trials early for benefit and estimation of effects: systematic review and meta-regression analysis. JAMA 2010; 303: Jonas WB, Crawford C, Colloca L, et al. To what extent are surgery and invasive procedures effective beyond a placebo response? A systematic review with meta-analysis of randomised, sham controlled trials. BMJ Open 2015; 5: e Chabok A, Påhlman L, Hjern F, Haapaniemi S, Smedh K, for the AVOD Study Group. Randomized clinical trial of antibiotics in acute uncomplicated diverticulitis. Br J Surg 2012; 99: Stollman N, Magowan S, Shanahan F, Quigley EM, for the DIVA Investigator Group. A randomized controlled study of mesalamine after acute diverticulitis: results of the DIVA trial. J Clin Gastroenterol 2013; 47: Raskin JB, Kamm MA, Jamal MM, et al. Mesalamine did not prevent recurrent diverticulitis in phase 3 controlled trials. Gastroenterology 2014; 147: van Dijk S, Boermeester MA. Current for diverticulitis: state of affairs in Ned Tijdschr Geneeskd 2016; 160: D69 (in Dutch). 38 Lembcke B. The gastroenterologist s view. Dig Dis 2012; 30: Vol 2 January 2017