Evidence evaluation

The literature evaluation was conducted on the basis of the 2011 Oxford Centre for Evidence-Based Medicine Levels of Evidence for interventional, diagnostic and prognostic studies.² Experts from the respective Working Groups (WGs) assessed the methodological quality of each study according to checklists, using the ‘Critical Appraisal Tools’ of the Oxford CEBM³ or, in the case of non-randomised (cohort and case-control) studies, the Newcastle-Ottawa Scale.⁴

Recommendations

The recommendations and background information were drafted by the WG leaders based on the evidence, and adopted within the individual WGs by means of an e-mail circulation procedure. The grading of the recommendations was based on the formulation should, should, can (Table 1).

All recommendations were voted upon according to a Delphi procedure by all guideline participators using a 3-option decision scale (yes, abstention, no). In the second Delphi vote, all but 9 recommendations received 95% approval. The remaining recommendations also achieved a high level of agreement, at over 90%. In consultation with the Arbeitsgemeinschaft der Wisseenschaftlichen Medizinischen Fachgesellschaften (Working Group of the Scientific Medical Societies in Germany), it was decided to forego a consensus conference (SARS CoV-2 pandemic). The strength of consensus was defined as set out in Table 2. Following the second Delphi vote, the comments underwent final revision by the WGs and the guideline was editorially compiled by the coordinators.

Statements

“Statements” are explanations or observations regarding specific facts or questions without an immediate call for action. The statements have been adopted as part of a formal consensus procedure in accordance with that used for the recommendations, and can be based either on study results or on expert opinions.

Mortality

Comment on both statements

Data on mortality from diverticulitis are very heterogeneous and of relatively poor quality. Complicated diverticulitis, in particular, has a relevant mortality rate. This increases with age and the extent of comorbidity.^{5, 6} In addition, the presence of ascites in patients with liver cirrhosis is associated with increased perioperative mortality.⁷ Immunosuppression represents a special situation in which steroid therapy, especially, increases the rate of postoperative complications.^8-11 Similarly, in diverticular haemorrhage, mortality depends to a large extent on comorbidity. In most cases, the cause of death is not the bleeding per se.^12-17

Comment—Recommendation—4.6

Different studies have consistently shown clinical diagnosis (without imaging procedures) of diverticulitis to have a substantial error rate. The studies of Toorenvliet et al.¹⁸ and Laméris et al.¹⁹ reported a sensitivity of 68% and a positive predictive value of 65%, and a sensitivity of 71%, respectively. Laurell et al.²⁰ found a similar sensitivity (64%), despite the already mentioned limitations. Schwerk et al.²¹ report a false positive purely clinical assessment of “highly suspected diverticulitis” in 9/28 cases and 44/68 cases with a less clear clinical suspicion (“possible but equivocal diverticulitis”), as well as a false negative assessment in 9/34 cases (“diverticulitis very unlikely”).

Comment—Recommendation—4.7

Both sectional imaging methods (ultrasound, CT) illuminate the extraluminal structures, enabling a comprehensive differential diagnostic assessment of diverticulitis and related complications.

The colonic barium enema should no longer be used to diagnose diverticulitis.

Special technical preparations are not required for sonography in diverticulitis; in fact, acute diverticulitis is the most easily learnable ultrasound diagnosis of the intestinal tract.

The hypoechogenic diverticular protrusion with an echogenic centre has also been referred to as the dome sign (in patients with right-sided diverticulitis).²⁵

The sonographic criteria of an abscess are hypoechoic/anechoic paracolic or intramural foci formation with echogenic reverberation echoes or comet tail artefacts; on the other hand, gas reflexes within hypoechoic band-shaped structures are characteristic of fistulas. Key structures of overt perforation are evidence of free air and free, mixed echogenically reflecting fluids.

Using high-resolution sound frequencies (>/= 7.5 MHz), it is possible to reliably visualise the layers of the intestinal wall, which can be helpful for the differential diagnostic evaluation of diverticulitis. Muscle hypertrophy and elastosis, as well as nutritive vessels that thus run perpendicularly through the sigmoid wall, are regular findings that are prerequisite to (left-sided) diverticula formation. In about 85% of cases, endoscopically verified diverticulosis (without indication of the focus of pain) can be correctly detected sonographically, whereby the number of diverticula detected in ultrasound is always lower than in colonoscopy.²⁶

In acute diverticulitis, at the hands of an experienced examiner, the sensitivity and specificity of abdominal sonography with directed questions and prospective evaluation are each 98%.²¹ Direct visualisation of the inflamed diverticulum is possible with a sensitivity of 96% in uncomplicated acute diverticulitis, but noticeably more difficult in the case of complicated findings (overall sensitivity 77%, specificity 99%).²⁴ Whereas ultrasound is mostly primarily focussed on the (painful) inflamed diverticulum, the detection of an inflamed diverticulum as sole criterion in computed tomography (CT) diagnostics achieves a sensitivity of only 43%.²⁷

An early systematic prospective comparative study from France shows an accuracy of 84% for both sonography and CT; the sensitivity was 85 versus 91%, specificity 84 versus 77%, Positive Predictive Value (PPV) 85 versus 81%, and Negative Predictive Value 84 versus 88%. With regard to other, alternative diagnoses, the sensitivity of CT was higher, at 50% versus 33% (ultrasound), as was also the case for the detection of pericolic abscesses.²⁷ A retrospective analysis from Spain shows a sensitivity of 86% in operated patients with acute diverticulitis, but 94% sensitivity in all patients with acute diverticulitis. The difference shows that uncomplicated acute diverticulitis, in particular, is a domain of sonography; however, this older study also found that 10 of 34 patients who underwent emergency surgery had false negative ultrasound findings (sensitivity 70%).²⁸

Due to developments in equipment, techniques and thematic know-how standards, both investigations must be regarded as no longer representative.

In a comparative prospective study from Germany with 4 experienced ultrasound examiners and the CT facilities of a university clinic, sonography showed a sensitivity of 100% (CT 98%), while the specificity of both methods was 97%. In cases of extensive peridiverticulitis and covert perforations, CT showed a clear tendency towards overstaging, whereas sonography showed a somewhat less pronounced tendency towards understaging. Overt perforations or abscesses were not missed by either procedure.²⁹

Like sonography, CT is a practicable and valuable examination where there is suspicion of acute diverticulitis. Both are suitable techniques to visualise the diagnosis and the severity of diverticulitis, to identify important differential diagnoses, and to guide the surgical approach in a stratified manner.

Diagnostic criteria for diverticulitis are the direct detection of inflamed diverticula, thickening of the intestinal wall to over 3 (5) mm and increased contrast medium absorption in CT and MRI (and, where appropriate, in contrast-enhanced ultrasound/CEUS). Indirect signs are perifocal mesenteric injection and free abdominal fluid as an expression of inflammation. Covert or overt perforations or evidence of abscesses detected by any imaging procedure are signs of complicated diverticulitis.

For CT, older studies still using a single-line detector configuration showed sensitivities and specificities of 87%–100% and 90%–100%, respectively.^{27, 30-32} The technique was found to be highly suited as a means of determining disease severity and, if necessary, of initiating further surgical consequences.^{33, 34} In initially conservatively treated patients, the severity of changes in the CT scan is an indicator of the likely necessity for surgery in the further disease course; however, even in the case of severe CT findings (pericolic air, abscess), there was no indication for surgery in the majority of patients during the course of disease.³³ Complications such as abscesses and covert or overt perforations can be evaluated by CT with a high degree of certainty.³⁵ Early studies showed CT to be superior to sonography.²⁸ CT-guided, interventional abscess relief can improve patient outcomes prior to surgery.^36-38

Comment—Recommendation 4.8

Computed tomography is currently performed in most German clinics as an examination with positive intravenous and oral contrast using diluted iodine-containing contrast agents. In addition, for better assessment of the rectum and sigmoid colon, rectal contrast via enema with a water-soluble contrast agent is recommended. The examination is carried out as a regular abdomen CT in the portal venous phase, with a tube voltage of 100–120 kVp and a tube current of around 120 mAs.

In recent years, studies have been conducted that dispensed with not only intravenous, but also oral and rectal contrasting; in addition, the use of modern multi-line CTs, which employ a low-dose technique with 30 mAs, can achieve the same diagnostic results as regular CT.³⁹ Theoretically, this could reduce radiation exposure from an average of 10 mSv to around 3 mSv, which would broaden the usability of the investigation. A comparison between single-line and multi-line technology in CT has not been conducted. Since studies using the single-line technique already achieved sensitivities and specificities of almost 100%, there is no relevant benefit to be expected from multi-line diagnostics.

As regards technology, therefore, it can be asserted that all modern CTs appear to be adequate and suitable for diagnosis. On the question of contrast, be it intravenous, oral or rectal, there are few publications suggesting that it may be possible to dispense with any type of contrast agent.³⁹ As yet, the evidence level of these few studies is insufficient for general application. Currently, if there are no contraindications, the technique with contrast should be used.

In the older, surgically dominated literature, CT is often the only cross-sectional imaging method that is used. A study from the Netherlands gives cause for critical consideration in this context: The study examined the validity of preoperative CT in all patients (n = 75) who underwent emergency surgery for perforated diverticulitis, after having been examined by CT within 24 h prior to the operation. The assessment was carried out retrospectively by two independent radiologists on the basis of the CT data sets, taking no account of the clinical presentation. The accuracy of the CT for different stages of perforation unexpectedly turned out to be only 71%–92% (PPV 45%–89%), and thus considerably lower than generally assumed. In 42% of patients with Hinchey stage 3, the study showed understaging in the CT (Hinchey stage 1 or 2) (indicating that CT has a PPV of only 61% for Hinchey stages 1 and 2).⁴⁰

In a comparable study from Germany, the preoperative CT was compared with intraoperative findings and histology in 204 patients. In patients with Hansen & Stock (HS) stage IIa (phlegmon), correct detection was found in 52% (intraoperative findings) and 56% (histology). Understaging was found in 12 (11)% and overstaging in 36 (33)%. The accuracy of staging for abscessing (HS IIb, Hinchey I/II) was 92% (intraoperative findings) and 90% (histology), with understaging in 3% and 0%, respectively, and overstaging in 5 (10)%. Overt perforation (HS IIc, Hinchey III/IV) was recorded correctly in 100%, yielding a PPV for CT of 52 (56)%, 92 (90)% and 100 (100)% for HS IIa, HS IIb and HS IIc, respectively.⁴¹ The value of the radiological assessment thus seems to be clearly examiner-dependent in the (important) HS stage IIa/IIb (understaging in the Netherlands, overstaging in Germany). For the preoperative differential diagnosis of phlegmonous diverticulitis (HS IIa) versus perforated diverticulitis (HS IIb/IIc), CT cannot universally be considered the gold standard.

Comment- Recommendation—4.9

The use of MRI to assess colonic diverticulitis is not yet widespread, either in practice or in studies. There are several problems concerning its practical implementation: Severe abdominal pain during the long procedure required for data acquisition often results in motion artefacts. Occasionally, claustrophobia prevents the examination from being adequately conducted. MRI is also associated with higher costs than CT and, in many clinics, MRI is not available around the clock for emergency examinations. What is more, the clinically and therapeutically very important issue of small air pockets around the colon when diagnosing overt or covert perforation is especially difficult to assess with MRI; as a result, its usefulness in complicated diverticulitis is very limited. To date, a systematic evaluation of the limit of detectability of small quantities of abdominal air is lacking in the literature. The technique has only been evaluated in small, usually specially chosen patient collectives.^42-45 Based on the results of these studies, it can only be concluded that similar results are achievable using MRI with oral or rectal contrast, or with intravenous contrast agent administration, as can be achieved by CT. However, it must be noted that there are no available studies dedicated to complicated diverticulitis or the detection of small pockets of free air in covert perforation.

In the absence of study data, it is not possible to give a definitive recommendation for the technical implementation of MRI in diverticulitis. Currently, analogous to CT imaging, a contrast-enhanced MRI examination with intravenous, oral and rectal contrast should be performed. The protocol should include high-resolution T1 weighted 3D gradient echo sequences as well as T2 sequences to allow assessment of acute inflammatory situations. The question as to whether intraluminal contrast using the dark-lumen technique⁴⁴ or T1 positive contrast can achieve a better differential diagnosis of abscesses⁴³ has not yet been answered in the literature.

Colon MRI for diagnosis of diverticulitis should therefore only be carried out in centres conducting controlled studies and in certain specific cases (e.g., examinations in pregnant women or paediatric patients, for reasons of radiation reduction).

Comment—Recommendation 4.10

Colonoscopy can explain abdominal complaints, is able to detect lower Gastrointestinal (GI) bleeding and to rule out tumours. It is suited for the differentiaton between diverticula and mucosal-inflammation and polypoid findings or diverticulosis with an atypical or symptomatic course.⁴⁶

Colonoscopy is not required to detect acute diverticulitis⁴⁷; an increased risk of perforation, although unproven, cannot be ruled out.

Endoscopically visible inflammatory changes at the diverticular neck are detected in about 0.8% of colonoscopies without the presence of acute diverticulitis.⁴⁸

Luminal changes are secondary in the pathogenesis of diverticulitis, since the disease begins as a bacterial penetration into the depths of the diverticulum, and crucial complications (phlegmon, microperforation, fistula, abscess) are transmural. If sonography indicates the intestinal wall to be thickened by >11 mm, colonoscopy shows the spontaneous drainage of pus from inflamed diverticula.⁴⁹

Comment—Recommendation—4.11

Due to insufficient data, opinions regarding the safety and importance of colonoscopy differ considerably.

In a series of 54 patients with diverticulitis, perforation occurred during colonoscopy in 1.9%; however, in a further 39 patients in whom covert perforation or abscess had been excluded by CT, no perforations were observed. A total of two CT-negative adenocarcinomas and a bone fragment in the inflamed diverticulum were detected as relevant findings.⁵⁰ The colonoscopies were carried out 4–12 days after hospital admission (median 5.8 days). The rate of total colonoscopies (reaching the coecum or tumour stenosis in 81.7% of cases) was lower than in an elective situation.

In the same clinic, a study was conducted to investigate early (in the hospital stay) versus postponed (after 6 weeks) colonoscopy for CT-confirmed diverticulitis. The authors identified neither perforations, nor any diagnostic gain.⁴⁷ They did, however, recognise a benefit for patients with an atypical disease course who had persistent symptoms after a week of antibiotic therapy or a relapse within 2 months. In this situation (23/224 patients), a therapeutically relevant diagnosis was made by colonoscopy in 4/23 cases (17%): in 3 cases, an adenocarcinoma, and in one case, a chicken bone lodged in a diverticulum, which was successfully endoscopically removed.⁵¹

Comment—Recommendation—4.12

Until now, colonoscopy has often been recommended (a) in principle after conservatively treated acute diverticulitis and (b) before sigmoid resection. This recommendation is based firstly on the differential diagnosis of other diseases with similar symptoms, and secondly, on coincidence of synchronous carcinoma or adenoma in predominantly older patients.

However, the importance and necessity of colonoscopy has been called into question by several studies in differing healthcare systems, due to the quality of consistent CT diagnostics of diverticulitis (and doubtless also on the grounds of health-economic considerations).

In a retrospective longitudinal study of 205 patients with a CT-guided diagnosis of acute uncomplicated diverticulitis, colonoscopy revealed adenomas in 9.3% of patients, 5.4% of which were advanced neoplasms.⁵² One patient was diagnosed with sigmoid carcinoma and one with Inflammatory Bowel Disease (IBD) (however, these two patients reported symptoms that would in any case have prompted colonoscopy). This rate of adenoma and carcinoma detection is somewhat lower than would be statistically expected based on the evaluation of data from screening colonoscopies.

In 100 patients four to six weeks after hospital treatment for acute diverticulitis (CT-based diagnosis), colonoscopy revealed at least one polyp in 32%, advanced adenoma in only one case and not a single malignancy; therefore, there were only a few (directly) relevant findings in only a small number of cases.⁵³

Though, prognostically, even findings of non-advanced adenoma should generally be considered a relevant pathology of the colon, other investigations allow detection of coincidental colon carcinoma on a larger scale. A widely cited retrospective study from the USA found that 5 out of 73 (7%) patients who underwent surgery for acute diverticulitis at the University Hospital of St. Louis between 1992 and 2001 had a previously undetected colon carcinoma.⁵⁴

In addition, a database analysis from Australia⁵⁵ found a slightly increased rate (2.1%) of colon carcinoma within one year after CT-based diagnosis of left-sided diverticulitis (evaluation of 1088 patients; comparison with the national cancer registry). In 319 patients, colonoscopy was performed within one year after diverticulitis was diagnosed: In nine of these patients, a colon carcinoma was identified (2.8%).

A systematic literature search on the usefulness of colonoscopy with respect to colon carcinoma detection up to 24 weeks after CT diagnosis of diverticulitis identified only 10 studies, with 771 documented patients.⁵⁶ The rate of colorectal cancer was 2.1% (95% Confidence Interval 1.2%–3.2%), and thus well above the expected prevalence (0.68%) in US citizens aged >55 years.

In another meta-analysis⁵⁷ that included 1796 patients after resolution of diverticulitis, the prevalence of carcinoma was 1.6% and the rate of detected polyps 20.2%. A systematic review by Meyer et al.⁵⁸ showed almost identical results, with a 1.9% prevalence of colorectal cancer (polyps 22.7%, advanced adenomas 4.4%, adenomas 14.2%). This review also showed that Colorectal Carcinoma was found significantly more frequently in patients with complicated than with uncomplicated diverticulitis (7.9% vs. 1.3%).

In a prospective, multicentric study,⁵⁹ no differences were observed in the prevalence of carcinoma or adenoma in patients who had had diverticulitis compared with a group undergoing routine screening.

Thus, the recommendation for total colonoscopy in patients >50 years of age with clinically conspicuous diverticular disease who have not undergone colonoscopy <5 years previously equates to a special situation of preventive colonoscopy; that is, colonoscopy is useful, despite reports of discrepant views from other health care systems.

This recommendation also serves to decisively counter the patients' subjective view that ultrasound or CT examination carried out due to the diverticulitis might be sufficient to rule out malignancy or dysplasia.

Indisputably, colonoscopy makes an essential contribution to further diagnostic clarification of CT-detected thickening of the colon wall.^{60, 61} Likewise, in the case of bowel stenosis, that is, including patients with recurrent diverticulitis with an indication for surgery, colonoscopy should generally be performed to ascertain the dignity (malignant vs. benign) of the stenosis. Since diverticulitis can also occasionally mask IBD, in patients with persistent pain, blood and/or mucous in the stool and signs of inflammation, it seems appropriate to confirm the diagnosis by colonoscopy, regardless of the patient's age.⁶²

Comment—Recommendation—4.13 and 4.14

Painless lower GI bleeding is predominantly ascribable to arterial diverticular bleeding (35%) and angiodysplasia (21%)⁶³; in elderly patients with diverticula, diverticular bleeding accounts for up to 50% of lower GI bleeding,^{64, 65} whereby diverticular bleeding is, however, usually a complication of diverticulosis rather than diverticulitis.

The aims of diagnosis and therapy of arterial diverticular haemorrhage are to clearly localise the source of bleeding, assess its severity and the probability of recurrence, and stop the bleeding - if possible as a definitive therapy, that is, also in respect of subsequent rebleeding.

Details of earlier bleeding severity are based on the patient's (only limitedly reliable) description of the amount of blood. Blood pressure and pulse rate (shock index) indicate the circulatory impact of the bleeding.⁶⁴ Validated scores like those used in upper GI bleeding (Rockall, Glasgow Blatchford) have not been reported. While spontaneous descriptions of the colour of lower GI bleeding are often questionable, a colour comparison chart can be helpful.⁶⁶

Recurrent bleeding is more often found in patients with endoscopically detected active bleeding or evidence of a vascular stump and coagulum-covered bleeding site, as well as those with arterial hypertension (RR 4.2), platelet aggregation inhibition (RR 2.4) or NSAIDs (RR 2.6).⁶⁷

Anticoagulant drugs also constitute a risk for more severe bleeding and rebleeding.⁶⁸

In accordance with the S2k guideline “Gastrointestinal Bleeding”,⁶⁹ if diverticular bleeding is suspected, gastroscopy should be performed early to rule out severe upper GI bleeding as the cause of haematochezia.⁶⁴

Comment- Recommendation—4.15a and 4.15b

In the case of acute peranal bleeding, upper GI bleeding must be considered as a differential diagnosis; thus, gastroscopy should be performed as early as possible. If the gastroscopy findings fail to explain the bleeding, a sigmoidoscopy should be carried out in order to rule out an anorectal source of bleeding.

Haematochezia with fresh blood arouses high suspicion of a bleeding source in the lower GI tract. However, peranal passage of fresh blood can also be a manifestation of heavy upper GI bleeding with a rapid transit time.

Diverticular bleeding is clinically indistinguishable from severe colonic bleeding of other origins; a priori, therefore, the situation is one of lower (i.e., colonic) GI bleeding. In this context, it should be noted that upper and mid-GI bleeding describes bleeding not only from within the gastroscopically visible segment, but also from anywhere within the whole small bowel. Heavy bleeding from the upper and mid-GI tract can mimic lower GI bleeding by causing the passing of brighter-coloured blood. Therefore, alongside colonoscopy, esophagogastroduodenoscopy is also part of the diagnostic concept, and additionally, if no evidence is found of a probable source of bleeding, (capsule) endoscopy of the small bowel (in haemodynamically stable patients) or angiography (in unstable patients).

Since diverticular bleeding stops spontaneously in 90% of cases, early colonoscopy is generally recommended in order to precisely identify the source of bleeding. Early colonoscopy (here, <24 h; OR 8.4), an experienced endoscopist (>1000 colonoscopies; OR 3.0), use of an Endo-Cap (OR 3.4) and use of a water jet rinse (OR 5.8) were proven to be prognostically favourable factors in lower GI bleeding.⁷⁰ Reliable identification of the bleeding source is achieved in 22% of early elective colonoscopies, that is, 7.5 times more frequently than after 24 h (p < 0.01) and 22 times more frequently than after 48 h (p < 0.01). In addition to the detection and localisation of the bleeding source, however, the application of endoscopic therapy should also be a primary aim.⁷¹ In patients with active haematochezia and diverticula, early colonoscopy (<12 h) with antegrade irrigation allows bleeding diverticula to be identified and interventionally treated in at least 20% of cases.⁷² Therefore, in patients with haemodynamical instability, having ruled out upper GI and anorectal bleeding, it seems necessary to perform colonoscopy after shortened bowel prep (4–6 L polyethylene glycol solution; if required, via a gastric tube over up to 12 h), applying additional cleaning methods such as enemas, and using an endowasher, as required. In stable patients, it is sufficient to use conventional preparation (split dosage) and perform the examination within 12–24 h. The detection rate of certain or probable sources of bleeding decreases over time.

Comment—Statement—4.16

Poncet et al.⁷⁴ reported spontaneous cessation of bleeding in 92.4% of a population of 133 patients with definitive or probable diverticular haemorrhage (among 1145 patients undergoing colonoscopy due to lower GI bleeding) over a period of 8½ years. An intervention was required in only 10/133 patients, 3 endoscopic, 4 radiological, and 3 surgical; additionally, 4 of the 7 patients initially receiving an endoscopic/radiological intervention required surgery as a secondary measure.

Although, in the light of these data, diverticular haemorrhage may appear prognostically favourable, it must not be trivialised, since there is a considerable tendency for recurrence, risk factors for rebleeding (including age, hypertension, low-dose American Society of Anesthesiologists (ASA), and NSAIDs) are widespread, and emergency surgery without exact localisation of the bleeding is associated with relevant morbidity and mortality.

The current literature, predominantly from Asia, shows that endoscopic band ligation (EBL) is superior to endoscopic clipping in terms of the rebleeding rate (6% vs. 33%; p = 0.018), while both procedures achieve an initial haemostasis rate of 100% without complications due to the respective techniques⁷⁵ In another Japanese multicentre study, the rebleeding rate under EBL was 10%, compared with 31% after endoscopic clipping (p < 0.01).⁷⁶ Early rebleeding was shown to emanate mainly from the same diverticulum as the initial bleeding. The main risk factor for earlier recurrence was the localisation of diverticular bleeding in the right colon, a peculiarity in Asia; it is thus unclear whether these results are transferable to the sigmoidal pseudodiverticula that are predominant in Western Europe.

As an alternative to rubber band ligation, the application of a so-called “over-the-scope clip” (OTSC) may be considered. In some case series, this procedure also seems to have a favourable effect on the further course of disease.^{77, 78}

Although afflicted with many uncertainties, the questionnaire-based retrospective study of Mizuki et al.⁷⁹ at least suggests that left-sided diverticula are less likely to bleed recurrently than right-sided or bilateral diverticula. The finding of this study, that non-interventionally treated patients had fewer rebleeding episodes (38.7% vs. 61.5%, p < 0.05) than those in the endoscopic intervention arm (clipping or adrenaline injection at the diverticular neck) should not be understood as an indication that the intervention increases the bleeding risk, but rather as an expression of the limitations of such an analysis. The detection of definitive bleeding stigmata was linked to therapeutic intervention, while the absence of a clear source of bleeding was associated with conservative treatment. In addition, no statement was made concerning the severity of bleeding (shock index, transfusions, haematocrit); therefore, despite having similar epidemiological data, the groups do not appear by any means comparable.

In Asia, bilateral diverticula increase the risk of acute diverticular haemorrhage (p = 0.0021), as do obesity, arterial hypertension, coronary sclerosis and low-dose ASA.⁸⁰

Comment—Recommendation—4.17

Today, it is generally accepted that in the therapeutic management of GI bleeding, the possibilities of endoscopic diagnostics and therapy should be first exploited.^{72, 81-84} In the case of repeated or persistent bleeding without an endoscopically clearly identifiable source, a CT angiography (or angiography) should be performed for localisation diagnostics during the suspected active bleeding.

Computed tomography angiography and conventional angiography (+/−Digital Substraction Angiography [DSA]) are valid options for localising diverticular haemorrhage during active bleeding. In practice, however, their use is rarely required. Computed tomography angiography enables reliable localisation of a haemorrhage if bleeding is still sufficiently active at the time of examination.⁸⁵ The same applies to conventional angiography, which offers the additional advantage of possible intervention (haemostasis through arterial embolisation: transcatheter arterial embolisation, TAE).

In a retrospective study, transarterial embolisation was performed in 52 patients with lower GI bleeding. The source of bleeding could only be clearly localised in 32/52 cases. Technical success was reported in 100%; however, there was a 30-day rebleeding rate of 27% and a 30-day mortality of 29%. In two patients, postinterventional intestinal ischaemia occurred.⁸⁶

Thus, the availability of technical equipment and personnel with sufficient expertise are of greater importance in acute severe haemorrhage than in less severe cases; this is particularly relevant when considering whether to transport patients if the appropriate equipment/expertise is lacking. In this situation - although there are no studies to confirm this - experience has shown that if endoscopic therapy is insufficiently effective, emergency surgery, as an option that is both reliable and well proven, should be given preference.

In a retrospective study, the mortality rate associated with emergency colectomy for diverticular haemorrhage was 17% and the rate of non-fatal complications 20%.⁸⁷

Comment—Recommendation—4.18

There are no data available from clinical studies for this scenario. The individual decision must be made by the attending physician after detailed discussion with the patient, taking into account the perioperative risk of elective surgery compared with the perioperative risk of emergency surgery in the event of diverticular bleeding that cannot be stopped endoscopically.

Comment—Statement—4.19

There are no clinical data to show what is the most suitable surgical procedure. If the bleeding cannot be endoscopically and interventionally localised and brought under control, there is a vital indication for urgent surgical therapy. Since these patients are critically ill and often multimorbid, laparotomy should be followed by colectomy, this being the fastest procedure. Whether to perform an anastomosis by means of ileorectostomy, or a discontinuity resection with closure of the anorectal stump and creation of a terminal ileostomy, is an individual decision in which the bleeding activity and intensity (previous transfusions) and the patient's comorbidity must be taken into account. In view of the fact that patients in the emergency situation are predominantly critically ill, discontinuity resection is usually the procedure of choice. In the study by Plummer et al., for instance, anastomotic leak was the most common cause of postoperative mortality.⁸⁷

Likewise, in very rare cases of non-localised recurrent haemorrhage requiring repeated transfusions, subtotal resection can be indicated. If this intervention is performed electively, the choice of procedure should be made individually, based on the characteristics of the patient and the experience of the surgeon. Studies comparing laparoscopic and conventional resection in this situation are not available.^{88, 89}

Comment—Recommendation 4.20

Few clinical data are available on the extent of surgical resection in patients with localisable diverticular bleeding. In a retrospective study of 42 consecutive patients with diverticular haemorrhage in the years 1993–2000, bleeding was localisable in 6 patients by colonoscopy (n = 2) or angiography (n = 4). Ten patients were treated by segmental resection and 32 patients underwent colectomy. In 5 of the 10 patients with segmental colon resection, the bleeding was localisable by colonoscopy, whereas this was achieved in only one of the 32 subtotally colectomised patients. The patients who underwent segmental resection were 10 years younger (65+/−13 vs. 75+/−12 years; p = 0.03), while there was no difference in the duration of surgery (208+/−77 vs. 212+/−58 min). Intraoperative blood loss was higher for subtotal resection (578+/−347 ml) than for segmental resection (305+/−146 ml; p = 0.02). No difference was found with regard to in-hospital morbidity (20 vs. 19%), mortality (10 vs. 3%), rebleeding (12.5 vs. 0%), stool frequency (2.4+/−1 vs. 3.5+/−2), the Cleveland Clinic incontinence score (0.6+/−1 vs. 2+/−3.6) or patient satisfaction over a mean follow-up period of 4.1 (0.5–7.4) years (p in each case >0.05).⁸⁸ Older studies essentially confirm these results.⁸⁹ Against this background, segmental and total colectomy are justifiable procedures in certain individual cases.

In surgical practice, if the site of the bleeding is clear, segmental colectomy is more frequently performed.

Comment - Statement 4.21

Fistulas to the urinary bladder or vagina are a relevant complication of diverticulitis. About 90% of fistulas in diverticulitis involve these two entities, while fistulas to the small bowel, skin, uterus or ovaries, psoas muscles or hip joints are less common findings.⁹⁰ The vast majority of patients (ca. 85%) with a sigmovesical fistula are male.

In patients with sigmovesical fistula, sonography or CT often shows a focal wall thickening of the (filled) bladder; evidence of air in the bladder in this situation confirms the fistula. Affected patients often report the presence of air bubbles in the urine (“champagne urine”) only when questioned; on the other hand, recurrent or therapy-refractory urinary tract infections and dysuria are characteristic and show the diagnostic path forward. No matter which tomographic technique is used, direct fistula detection is only realisable in a certain proportion of cases. If the symptoms are clear (pneumaturia, recurrent urinary tract infection), suspicion of an enterovesical fistula in the cross-sectional imaging is a sufficient indication for sigmoid resection.

While colonoscopy can detect residual inflammatory activity, Crohn's disease as an important differential diagnosis, and stenosis, endoscopic diagnosis of fistulas succeeds only rarely (<10%⁹⁰). Similarly, detection rates of cystoscopy (10%), cystography (17%), colonic barium contrast imaging (36%), MRI (60%) and CT (61%) are disappointing. Qualitative detection of a fistula is best performed (sensitivity 95%) by the so-called poppy seed test, in which 250 g natural poppy seeds are taken in the evening and the urine is examined for the appearance of poppy seeds over the next 48 h.^{91, 92}

The extent to which urological diagnostics are useful or necessary before sigmoid resection and fistula excision must therefore be decided on a case-by-case basis, and is consequently more often determined by local factors.

In another modification, 35 g poppy seeds were consumed in 160 g yoghurt or with 340 ml liquid; here, too, the poppy seed test, with a sensitivity of 100%, was significantly (p = 0.03) superior to CT examination (70% sensitivity) — at 8.2% of the cost.⁹³

In principle, the poppy seed test is also suitable for detecting a colovaginal fistula; it is recommended to insert a tampon or cotton wool pad for detection after ingestion of the test substance. In individual cases, colposcopy and vaginal transrectal endosonography can be useful supplementary techniques in addition to sonography and CT; general or comparable information on the respective detection rates of these methods is not available.

Comment—Recommendation 4.22

The Classification of diverticular disease (CDD) allows different degrees of severity and different situations to be categorised. This is useful if it is linked with different diagnostic and/or therapeutic pathways and recommendations, the provision and use of different measures, and the possibility to improve safety for both patient and physician.

In principle, therefore, a classification should cover all facets of diverticular disease without becoming impractical by being excessively detailed and structurally emphasising rare situations. In addition, it should allow the disease course to be easily and correctly described according to the specific situation.

Comment—Recommendation 4.22b

In 2014, the DGVS and DGAV adopted a new classification, the CDD (Table 3), which has since found its way into the literature and clinical practice.⁹⁴

As evidenced by diverse classifications that accommodate national circumstances and take into account not only diagnostics and therapy, but also new aspects of aetiology, pathogenesis and the nosological understanding of the disease, the classifications of Hinchey (including the modifications by Sher and Wasvary), Ambrosetti and Hansen/Stock (and their modifications by Köhler and Siewert) can now be considered outdated in terms of their practical relevance and/or content.

The scope of the more recent (since 2011) guidelines and classifications, and the relative weighting of different content, is presented in a thorough review by Galetin et al. (2018),⁹⁵ in which the previous S2k guideline of the DGVS/DGAV (2014).⁹⁶

Numerous classifications and modifications describe the various stages of diverticular disease. For current critical reviews, see^{97, 98} and.⁹⁵

Both of these goals are achieved primarily by the Hinchey classification as modified by Wasvary,^{99, 100} and by the classification of HS.¹⁰¹ However, the former includes only the different manifestations of diverticulitis with a category of mild clinical diverticulitis (relevant for outpatient treatment), while the classification by HS does not further differentiate perforated disease (micro/macroperforation, abscess size and site).

An advantage of the HS classification was the inclusion of chronic relapsing (recurrent) disease. However, it does not differentiate between chronic recurrent disease without complications (individual indication for elective surgery) and chronic recurrent disease with complications (obligatory indication for surgical therapy). In the Hansen-Stock classification, acute diverticulitis with accompanying phlegmon falls under the category of complicated diverticulitis.

The CDD classification correctly classifies this disease type as uncomplicated, with a good prognosis under conservative therapy. In particular, the sonographic finding of a hyperechoic mesenteric cap as a correlate of peridiverticular changes is found in both stages HS I and HS IIa (without being categorised as complicated diverticulitis). The boundary between HS I and HS IIa is difficult to visualise with CT (or with sonography) and a differentiation between microperforation and macroperforation, which would be desirable, is lacking.