Patients admitted to our hospital for overt LGIB between September 2009 and October 2013 and examined at the endoscopy unit of the National Center for Global Health and Medicine (NCGM) were enrolled. Patients from a previous study cohort[12], a newly diagnosed cohort (approval No. 750), and a randomized trial cohort (approval No. 765) were prospectively followed after discharge. Inclusion criteria were: (1) > 18 years old; (2) Japanese nationality; (3) overt LGIB examined within 1 wk of onset; and (4) being managed in hospital. Exclusion criteria were: (1) no informed consent provided; (2) not independent in activities of daily living; (3) not undergoing colonoscopy; (4) LGIB due to causes other than CDB; and (5) barium impaction therapy in the randomized controlled trial (approval No. 765). All inclusion and exclusion criteria were fulfilled before patients were enrolled. The NCGM is a large emergency hospital, with 900 beds, located in metropolitan Tokyo, Japan. All patients provided written informed consent prior to enrollment. The institutional review board at the NCGM approved this study (approval No. 810), and all clinical procedures conformed to Japanese and International ethical guidelines (Declaration of Helsinki).

Patients were asked about their lifestyle habits, medications, and comorbidities in a face-to-face interview on the pre-colonoscopy day.

Medication: The questionnaire survey form asked about the use of 9 kinds of NSAIDs, 2 kinds of low-dose aspirin, 10 other antiplatelet agents, 3 anticoagulants, acetaminophen, and oral corticosteroids. The questionnaire included photographs of all of these oral drugs, which are approved in Japan. Use of a medication was defined as oral administration within 1 mo of the interview. During hospitalization, low-dose aspirin, other antiplatelet drugs, and anticoagulants were temporarily stopped or other bridging drugs given whenever possible after consulting a cerebrocardiovascular specialist. Low-dose aspirin and other antiplatelet agents were resumed after hemostasis before discharge. Also, NSAIDs were temporarily stopped on admission and prescribed during hospitalization only on the advice of an orthopedist or rheumatology specialist. We suggested NSAID withdrawal to those who were intermittent users and to those who were suitable for alternative therapy such as acetaminophen, an antiinflammatory analgesic plaster, antigout drugs, antirheumatic drugs, or corticosteroids.

Comorbidities: Hypertension, diabetes mellitus, dyslipidemia, cerebrocardiovascular disease, chronic liver disease, and chronic kidney disease (CKD) were assessed. A history of cerebral infarction, cerebral hemorrhage, myocardial infarction, or angina pectoris was considered cerebrocardiovascular disease. CKD was considered present in patients on hemodialysis. Chronic liver disease included chronic viral hepatitis and alcoholic liver disease.

An electronic high-resolution video endoscope (model CFH260; Olympus Optical, Tokyo, Japan) was used after spontaneous cessation of bleeding. CDB was defined as either definite or presumptive on the basis of colonoscopy with multidetector computed tomography (MDCT)[13,14]. Definitive diagnosis was based on colonoscopic visualization of colonic diverticula with stigmata of recent hemorrhage such as active bleeding, adherent clot, or visible vessel[13,14]. A presumptive diagnosis was based on MDCT visualization of the extravasation of contrast medium in colonic diverticula and colonoscopy showing (1) a potential bleeding site in an area of positive MDCT findings; (2) fresh blood localized to colonic diverticula in the presence of a potential bleeding source on complete colonoscopy; or (3) bright red blood in the rectum confirmed by objective color testing and colonoscopy demonstrating a single potential bleeding source in the colon, complemented by negative upper endoscopy, or negative capsule endoscopy[13,14]. Patients in whom the bleeding source was identified received endoscopic treatment such as clipping or endoscopic ligation.

After discharge, all patients were requested to visit the hospital in the case of bloody stools or on a scheduled day within the first month and then every three months during the observation and follow-up periods. Telephone interviews were conducted with patients who did not visit the hospital on scheduled days, and a hospital visit was recommended. Rebleeding was defined as a significant amount of fresh bloody or wine-colored stools (> 200 mL) during the follow-up period and was evaluated by both anoscopy and MDCT within 12 h of onset. Clinically suspected rebleeding should prompt further colonoscopy when possible, but no routine second-look colonoscopy was performed when rebleeding occurred during hospitalization or within 1 mo of discharge. Colonoscopy was performed to confirm rebleeding and determine the need for intervention when frequent or massive bleeding occurred along with unstable vital signs, systolic blood pressure ≤ 90 mmHg or pulse ≥ 110 beats/min, and a non-response to ≥ 2 units of transfused blood during a 24-h period. We distinguished between rebleeding and remaining blood from the index bleeding episode.

The primary outcome was rebleeding due to CDB. The secondary outcome was a cerebrocardiovascular or thrombotic event during the follow-up period. Patients lost to follow-up or death were censored at the time of their last visit. The Kaplan-Meier method was used to estimate the recurrence of CDB at 1, 6, 12, and 24 mo after discharge. Risk factors and the effect of NSAID withdrawal on rebleeding were analyzed using the log-rank test. Cox proportional hazards regression was used to analyze the independent risk factors for rebleeding by considering factors significant (P < 0.1) in the log-rank test and adjusting for factors such as age > 70 years and hypertension, as previously reported[5-7].

A value of P < 0.05 was considered significant. All statistical analysis was performed using Stata version 10 software (StataCorp, College Station, TX, United States).

During the study period, 337 patients were admitted to our hospital for acute LGIB. After exclusion, 132 patients with CDB were enrolled in this study. Baseline characteristics at discharge are shown in Table 1. Diverticula were located predominantly in the bilateral colon in 48%, in the right side in 32%, and in the left side in 20% of cases. Of these, 17% (23/132) had colonoscopic evidence of stigmata of recent hemorrhage diverticula and were treated by endoscopic procedures during hospitalization, in which 37 patients (28%) received a mean transfusion of 8 units of packed red blood cells.

Among the patients using NSAIDs on admission (n = 41), 26 (63%) discontinued NSAID use. Many patients who could withdraw NSAIDs were intermittent users (n = 23), and could be switched to alternative therapies included acetaminophen (n = 11), an antiinflammatory analgesic plaster (n = 10), narcotic analgesics (n = 2), antigout drugs (n = 3), anti-rheumatic drugs (n = 1), and corticosteroids (n = 1), while the remaining patients (n = 7) took nothing. Some of these 26 patients took more than one of these drugs. No patients were taking NSAIDs for cardiovascular disease. Fifteen patients (11%) on NSAIDs and 13 patients (8.5%) on acetaminophen continued to take them after discharge. No patients on low-dose aspirin, non-aspirin antiplatelet agents, or corticosteroids discontinued use or started taking them during the follow-up period (Table 1).

The most common comorbidities were hypertension (63%), cardiovascular disease (25%), dyslipidemia (22%), diabetes mellitus (20%), hemodialysis (5.3%), chronic liver disease (3.8%), and cerebrovascular disease (1.5%) (Table 1).

After discharge, 39 patients (30%) experienced rebleeding with a mean follow-up of 15 ± 13 mo. The probability of rebleeding was 3.1% at 1 mo, 19% at 6 mo, 27% at 12 mo, and 38% at 24 mo (Figure 1A). No patients required angiographic embolization or surgical resection within the follow-up period.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Probability of bleeding recurrence during the follow-up period (n = 132). A: Incidence of rebleeding with a mean follow-up period of 15 mo; B: Probability of rebleeding was significantly higher in the [non-steroidal antiinflammatory drug (NSAID) group than in the non-NSAID group (P < 0.01)]; C: Probability of rebleeding was significantly higher in the hemodialysis group than in the non-hemodialysis group (P < 0.01).

Log-rank tests revealed NSAID use and hemodialysis as significant risk factors for rebleeding, while the other factors were not found to be significant (Table 2). Patients without right-sided diverticula had a risk of rebleeding, but this did not reach statistical significance (Table 2). The probability of rebleeding was 2.6% (95%CI: 0.86%-8.0%) at 1 mo and 21% (95%CI: 14%-30%) at 12 mo in non-NSAID users, and 7.1% (95%CI: 1.0-41) at 1 mo and 77% (95%CI: 52-94) at 12 mo in NSAID users (Figure 1B). The probability of rebleeding was 2.5% (95%CI: 1.0-7.6) at 1 mo and 24% (95%CI: 17-34) at 12 mo in non-hemodialysis patients, and 14% (95%CI: 2.1%-67%) at 1 mo and 71% (95%CI: 39%-96%) at 12 mo in hemodialysis patients (Figure 1C). Multivariate analysis revealed NSAID use and hemodialysis as independent risk factors for rebleeding (Table 3).

Of the NSAID users (n = 41), 15 continued and 26 discontinued NSAID use after discharge. During the follow-up period, no patient developed a thrombotic event, such as pulmonary embolism, deep vein thrombosis, or a cerebrocardiovascular event. No patient who discontinued NSAID use on discharge resumed NSAID use during the follow-up period. The probability of rebleeding was 0% at 1 mo and only 9.4% (95%CI: 2.4%-33%) at 12 mo in discontinuing NSAID users compared with 7.1% (95%CI 1.0%-41%) at 1 mo and 77% (95%CI: 52%-94%) at 12 mo in the continuing users (P < 0.01 by the log-rank test; Figure 2). The hazard ratio (HR) for rebleeding in the discontinuing NSAIDs group was 0.06 (95%CI: 0.01%-0.31%) after adjusting for age > 70 years, right-sided diverticula, history of hypertension, and hemodialysis in the Cox proportional hazards model.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Probability of recurrent bleeding in patients with non-steroidal antiinflammatory drugs on admission (n = 41). The probability of rebleeding was higher in subjects continuing non-steroidal antiinflammatory drugs (NSAIDs) than in those discontinuing NSAIDs (P < 0.01).

Even though intensive management for diverticular bleeding is successful during the hospital stay, patients are at high risk of recurrent bleeding following discharge. Unlike upper gastrointestinal (GI) bleeding, there are no preventive therapies for lower GI bleeding. Thus, preventing recurrence requires risk factors to be identified.

The use of non-steroidal anti-inflammatory drugs (NSAIDs) is a significant risk factor for recurrence, but the risks and benefits of their discontinuation on diverticular bleeding have not been examined.

In total, 39 patients (30%) had rebleeding during a median follow-up of 15 mo. Of the 41 NSAID users on admission, 26 (63%) discontinued NSAID use at discharge. Multivariate analysis revealed continued NSAID use (HR = 4.6, P < 0.01) and hemodialysis (HR = 3.0, P < 0.01) as independent risk factors for rebleeding. Discontinuation of NSAIDs significantly reduced rebleeding risk compared with continued use (log-rank test, P < 0.01).

Discontinuation of NSAIDs is an effective preventive measure against rebleeding in diverticular bleeding. However, medication withdrawal should be considered in view of its risks and benefits, particularly in patients of advanced age or with comorbidities.

The precise mechanisms of NSAID-induced diverticular bleeding has been suggested to be due to the inhibition of platelets, cyclooxygenase-1, and prostaglandin synthesis in the lower GI tract. Mucosal injury due to the inhibition of intestinal prostaglandin synthesis might lead to the development of intestinal erosion and ulcers. When mucosal ulceration induced by NSAIDs occurs at the neck or dome of the diverticula, nutrient-providing arteries rupture into the colonic lumen with the potential to cause sudden bleeding.

The manuscript investigates the impact of discontinuing NSAIDs and the long-term recurrence in colonic diverticular bleeding. The authors tried to bring new ideas about preventive measure against colonic diverticular re-bleeding.