Urgent Colonoscopy in Lower GI Bleeding: Not So Fast

Abstract

The role of urgent colonoscopy in lower gastrointestinal bleeding (LGIB) remains controversial. Although some studies have shown that examinations performed within 12–24 h of admission improve diagnostic yield and reduce rebleeding and surgery, others have not. In this issue of the American Journal of Gastroenterology, Laine and Shah present a randomized trial of urgent ( < 12 h from admission) vs. elective (36–60 h from admission) colonoscopy in 72 patients with LGIB. A total of 15% of patients with presumed LGIB were found to have upper gastrointestinal bleeding, highlighting the importance of excluding a gastroduodenal source in patients with severe hematochezia. The majority of patients with LGIB (72%) stopped bleeding spontaneously, and there were no differences in rebleeding, blood transfusions, diagnostic or therapeutic interventions, length of hospital stay, or hospital charges in patients undergoing urgent vs. elective colonoscopy. However, the limited number of patients in this study and the fact that patients in the urgent colonoscopy arm appeared to have more severe bleeding than those undergoing elective examinations make it difficult to draw conclusions regarding the utility of urgent vs. elective colonoscopy in LGIB.

Lower gastrointestinal bleeding (LGIB) is a common gastrointestinal emergency. The incidence of LGIB is increasing, and, in the elderly, may surpass that of upper gastrointestinal bleeding (UGIB) (1). However, the literature regarding LGIB lags considerably behind UGIB. Endoscopy is the standard of care in the management of UGIB, and urgent esophagogastroduodenoscopy (EGD) within 12–24 h of admission has been shown to provide valuable prognostic information, facilitate the treatment of high-risk lesions, and improve patient outcomes and resource utilization.

In contrast, the role of urgent colonoscopy in LGIB remains controversial. Historically, colonoscopy in LGIB was performed electively due to the need for colon preparation and concern regarding complications. Over the last two decades, a number of studies have indicated that urgent colonoscopy, defined as colonoscopy performed within 12–24 h of admission following a rapid colon purge, is safe and may facilitate the identification and treatment of bleeding lesions (2-4). However, studies comparing this approach to delayed colonoscopy or to other interventions for LGIB are limited. Moreover, urgent colonoscopy is logistically complicated, stigmata of hemorrhage are arguably difficult to identify, and there are a number of other potential management options to choose from, including flexible sigmoidoscopy, angiography, radionuclide scintigraphy, and multidetector computed tomography scanning.

In this issue of the American Journal of Gastroenterology, Laine and Shah present a randomized trial of urgent vs. elective colonoscopy in patients with LGIB (5). After undergoing an EGD to exclude an upper gastrointestinal source, 72 patients with severe LGIB were randomized 1:1 to colonoscopy within 12 h of admission or elective colonoscopy within 36–60 h. Colonoscopies were performed by the on-call or attending gastroenterologist, and management decisions outside the endoscopic encounter were left to the admitting team. The authors found no differences in the primary outcome, rebleeding during the hospitalization, or secondary outcomes, including number of units of blood transfused, number of diagnostic or therapeutic interventions, length of hospital stay, or hospital charges, and conclude that urgent colonoscopy does not improve outcomes in patients with serious LGIB.

This study represents a significant effort to advance the study of LGIB and highlights several important points. First, 15% of patients with presumed LGIB were subsequently diagnosed with UGIB via urgent upper endoscopy, underscoring the need to exclude an upper source in patients presenting with hematochezia and hemodynamic compromise. Urgent upper endoscopy followed by a rapid colon preparation if negative is the optimal approach to patients with severe hematochezia. Nasogastric lavage is a less sensitive test for bleeding in the stomach or duodenum, but, if placed, can be used to facilitate subsequent bowel preparation.

Second, most patients with LGIB will stop bleeding spontaneously and therefore will not benefit from urgent therapeutic interventions. Even in this study of patients with significant hematochezia at presentation, only 18% had further bleeding while hospitalized. Therefore, risk stratification tools that reliably identify high-risk patients who are likely to require hemostasis would be invaluable. There have been three published studies aiming to systematically predict outcomes in LGIB (6-9). These studies have identified a number of risk factors in LGIB, including hemodynamic instability, ongoing bleeding, and comorbid illness, and have been able to predict poor outcome with reasonably good to very good accuracy. However, only one study evaluated the need for bleeding control as a specific outcome (6), and the ability of these tools to improve the triage of patients with LGIB in routine practice is uncertain. Large, prospective studies in diverse clinical settings are needed to develop and test decision aids in LGIB.

Third, early performance of colonoscopy appears to improve diagnostic yield. In the Laine trial, 78% of colonoscopies were diagnostic in the urgent group compared with 67% in the elective group, although this difference was not statistically significant. Notably, the only two stigmata of hemorrhage that were identified were in patients with diverticulosis undergoing urgent colonoscopy. Similarly, in a trial by Green et al. (2), a definitive bleeding source was identified significantly more often in patients with LGIB undergoing urgent colonoscopy vs. elective colonoscopy (42 vs. 22%, P = 0.03), and no stigmata of hemorrhage were identified in patients undergoing elective colonoscopy following a positive tagged red blood cell scan (mean time to colonoscopy was 38 h). Other series have found that the yield of colonoscopy is higher when performed in the setting of active bleeding vs. electively (10,11), suggesting that the timing of colonoscopy relative to bleeding is more important than the interval since admission. However, diagnosis alone will have little impact on major clinical outcomes. Stigmata of hemorrhage need to be identified and treated in order to justify urgent interventions.

Unfortunately, based on the results of Laine and Shah (5), it is difficult to draw conclusions regarding the utility of urgent vs. elective colonoscopy in LGIB. The lack of statistically significant differences between the two study groups may be explained by insufficient statistical power. The authors estimated that 134 patients would be needed in each arm to detect a 15% difference in the primary outcome (rebleeding). However, after 6 years of recruitment, only 27% of the specified sample size was enrolled. The authors nevertheless argue that based on the 95% confidence interval of the difference in rebleeding between the two arms (−9 to 26%) urgent colonoscopy would at most reduce rebleeding a modest 9% compared with elective colonoscopy. The only other randomized trial of urgent colonoscopy in LGIB was also underpowered, but did find a trend toward decreased rebleeding in patients undergoing urgent examinations (22 vs. 30%) (2). Indeed, it seems unlikely that any single-center study would be able to recruit a sufficient number of patients to determine whether urgent colonoscopy improves outcomes in LGIB.

Furthermore, despite randomization, the patients in the urgent colonoscopy arm appeared to have more severe bleeding at presentation. The mean admission hemoglobin concentrations were significantly lower in the urgent arm, and these patients received significantly more pre-randomization blood transfusions. In addition, they appeared to be more likely to have diverticular bleeding (31 vs. 19%) and angiodysplasia (6 vs. 0%) vs. other less severe bleeding sources, and to have undergone tagged red blood cell scan (6 vs. 1) and angiography (4 vs. 0). More severe bleeding in the urgent colonoscopy group may account for the somewhat higher rebleeding rate in the urgent vs. elective colonoscopy group (22 vs. 13%), and the negative results of the study.

In the Laine study, urgent colonoscopy did not reduce hospital length of stay compared with elective colonoscopy even though performance of elective examinations was delayed 36–60 h after admission. In contrast, several observational studies have shown that earlier performance of colonoscopy is a strong independent predictor of length of stay, likely due to efficient triage of low-risk patients and not due to treatment of high-risk stigmata (12,13). However, all patients in the Laine trial had severe bleeding and may not have been judged safe candidates for early discharge regardless of colonoscopy timing or findings. In addition, studies of UGIB have shown that when triage is left to the discretion of the admitting team, patients with low-risk findings are frequently not discharged despite the evidence that this is safe and efficient (14). Therefore, care protocols are needed to facilitate discharge based on low-risk clinical and endoscopic findings. In this setting, early colonoscopy is likely to result in earlier discharge and therefore reduced costs of care.

Endoscopies in Laine and Shah (5) were performed by attending or on-call gastroenterologists in patients with all sources of hematochezia after a standard 4 l polyethylene glycol preparation. The quality of the colon preparations was not noted except that five patients (7%, two in the urgent arm and three in the routine arm) required repeat colonoscopies secondary to inadequate preparation. In contrast, a specialized bleeding team performed all colonoscopies in a study by Jensen et al. (4) in which urgent colonoscopy was found to reduce rebleeding and the need for surgery. In addition, patients in the Jensen study were a highly selected group with stigmata of diverticular bleeding who underwent very aggressive bowel preparation (5–6 l of polyethylene glycol). The exemplary results in the Jensen study indicate that careful patient selection, experienced endoscopists, and scrupulous colon preparation may be necessary for successful urgent colonoscopy, and these requirements may limit the effectiveness of this strategy in routine clinical practice.

Although most patients with LGIB will do well without urgent intervention, LGIB is by no means either benign or straightforward. Out of the 72 patients in this study, there were 41 subsequent diagnostic or therapeutic interventions including 7 tagged red blood cell scans and 4 angiographies, and yet, no source of bleeding was identified in 15 patients (21%). Importantly, 4 of 7 diagnoses made on repeat examinations were in patients undergoing repeat colonoscopy and none of the 11 radiographic tests provided a diagnosis or therapy. These data support colonoscopy as an efficient strategy for LGIB both initially and for recurrent or undiagnosed bleeding. However, studies are needed to compare colonoscopy to radiographic interventions for severe LGIB.

In summary, the optimal timing of colonoscopy in LGIB remains to be determined. As Laine and Shah (5) demonstrate, most patients will stop bleeding spontaneously and will not benefit from urgent colonoscopy. However, a small subset will have ongoing bleeding requiring timely intervention. Data suggest that colonoscopy performed near the time of active bleeding is more likely to identify and treat stigmata. Therefore, early examinations in these patients may improve outcomes, but to demonstrate this would require a much larger trial. In addition, the ability to identify high-risk patients at the time of presentation would substantially improve the utility of urgent colonoscopy in practice. Finally, it is critical to exclude an upper gastrointestinal source of bleeding in patients presenting with severe hematochezia.