Minimum colonoscopy observation time for colonic diverticular bleeding: A new benchmark based on the 5% plateau time

Abstract

BACKGROUND

Colonic diverticular bleeding (CDB) is a leading cause of gastrointestinal bleeding-related hospitalizations in Japan and is increasingly recognized as a significant burden in the United States. Identifying the stigmata of a recent hemorrhage (SRH) during colonoscopy enables targeted hemostasis and reduces rebleeding. However, no benchmark exists for an appropriate observation duration, resulting in operator-dependent variation. Short observation periods may lead to missed SRH, whereas unnecessarily prolonged procedures, particularly in older patients, can increase patient burden and limit endoscopy unit availability.

AIM

To establish a practical benchmark for minimum colonoscopy observation time to ensure reliable SRH detection in patients with CDB.

METHODS

We retrospectively analyzed patients with acute hematochezia who underwent an initial colonoscopy between January 2017 and December 2024 at a Japanese tertiary hospital. The Observation time was measured from scope insertion to SRH detection (excluding therapeutic time) or withdrawal. The primary outcome, the “5% plateau time”, was defined as the point when the proportion of patients newly identified with SRH in each 5-minute interval consistently dropped below 5%. Computed tomography (CT)-based stratified analyses were performed by endoscopists who conducted ≥ 10% of procedures.

RESULTS

Of the 1039 patients who underwent colonoscopy, 845 (mean age 77 ± 11 years; 64.5% male) were included. Nine board-certified endoscopists performed the procedures. SRH was detected in 286 patients (33.8%), with a median detection time of 19 minutes (interquartile range, 12-28 minutes). The overall 5% plateau time was 40 minutes and varied according to the CT findings: 40, 35, and 30 minutes for no extravasation, right-sided extravasation, and left-sided extravasation, respectively. This time point corresponded to when 80%-90% of SRH cases were detected. Despite variations in SRH detection rates and observation durations among endoscopists, the 5% plateau time was consistently approximately 40 minutes.

CONCLUSION

Although it varied according to the CT findings, the overall 5% plateau time was 40 minutes. This offers a practical benchmark for the minimum observation time without SRH detection.

Key Words: Diverticular hemorrhage; Lower gastrointestinal bleeding; Stigmata of recent hemorrhage; Observation time; Observation duration; Withdraw time; Bleeding source

Core Tip: This study introduces the “5% plateau time” as a novel benchmark for determining the minimum necessary colonoscopy observation time in colonic diverticular bleeding (CDB). By analyzing 845 cases, we found that stigmata of recent hemorrhage (SRH) detection significantly diminished beyond 40 minutes, with stratified times of 40, 35, and 30 minutes, depending on the computed tomography findings. This time-based metric offers practical guidance for balancing the diagnostic yield and procedural burden, particularly in older adults. By identifying the minimum observation time necessary for adequate SRH detection, the 5% plateau time supports a safer and more consistent endoscopic management of CDB.

INTRODUCTION

Colonic diverticular bleeding (CDB)’s prevalence has increased in the aging population and, by 2017, had surpassed hemorrhagic gastric ulcers as the leading cause of hospitalization for gastrointestinal bleeding in Japan[1,2]. A similar trend has been observed in the United States, where diverticular diseases are a major cause of hospitalization, accounting for over 210000 admissions and more than $2.2 billion in annual healthcare costs[3]. Although many cases resolve spontaneously[4-7], a substantial proportion of patients experience recurrent bleeding[6,8] requiring hospitalization, transfusion, endoscopic therapy, or occasionally surgery[2,9]. As the population ages, CDB is expected to pose a substantial clinical and healthcare burden.

Colonoscopy is the cornerstone of both the diagnosis and treatment of CDB. The detection of stigmata of recent hemorrhage (SRH) during colonoscopy enables definitive endoscopic hemostasis with a reduced risk of rebleeding[6,10]. Several technical and procedural strategies, including early colonoscopy, water-jet scopes, distal attachment caps, and performance by experienced endoscopists, have been explored to improve detection rates[11-14]. Observation time is also considered an important factor; however, few studies have evaluated it, and no clear benchmark exists. Consequently, the observation time remained operator-dependent. Brief observation times may lead to missed SRH and lost treatment opportunities, whereas unnecessarily prolonged procedures, particularly in older patients, may increase the risk of aspiration pneumonia, compound sedation-related burden[15], and be an inefficient use of endoscopy resources.

Currently, no time-based rule for ending observation exists, and a pragmatic benchmark for the minimum observation time is lacking. The absence of such standards contributes to interoperator variability and may lead to both under- and over-observation. Therefore, defining a scientifically grounded lower bound, the minimum time that should be ensured for SRH detection, is essential for improving the diagnostic consistency and procedural efficiency of CDB colonoscopy.

Therefore, in this study, we proposed the “5% plateau time”, defined as the earliest time at which the incremental SRH detection rate falls below 5%, and evaluated it as a scientifically grounded and pragmatic benchmark for the minimum observation time for CDB colonoscopy.

MATERIALS AND METHODS

Study design and setting

This single-center, retrospective, observational study was conducted at the Shonan Kamakura General Hospital, a 669-bed tertiary care hospital in Japan. The hospital manages over 14000 emergency transportation cases and 42000 emergency department visits annually. More than 5000 colonoscopies are performed each year at the hospital, and all nine board-certified endoscopists have perform at least 1000 colonoscopies in their careers. A 24-hour emergency endoscopy service is available, which includes full coverage during nights and holidays.

Ethical considerations

This study was approved by the Institutional Review Board of the Future Medical Research Center Ethics Committee (Approval No. TGE1304-024). As this was a non-interventional, observational study using anonymized data, the requirement for informed consent was waived, and an opt-out approach was adopted. This study was conducted in accordance with the principles of the Declaration of Helsinki and is reported in line with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

Participants

Consecutive patients hospitalized for acute hematochezia who underwent colonoscopy during the same admission were included. CDB was defined as either: (1) Definite CDB, in which clear SRH were identified within a diverticulum[15-17]; or (2) Presumptive CDB, in which no SRH was observed and no alternative bleeding source was identified[16-18]. SRH was defined as active bleeding (Figure 1A), non-bleeding visible vessels (Figure 1B), or adherent clots (Figure 1C) that developed into active bleeding or non-bleeding visible vessels upon removal[17,18]. Patients were excluded if the observation time could not be assessed or if trainees were involved in the colonoscopy, as limited technical proficiency could not only prolong insertion and observation times but also lead to an unjustified reduction in the SRH detection rate.

Figure 1 Representative images of stigmata of recent hemorrhage in colonic diverticular bleeding. A: Active bleeding: Spurting or oozing blood observed from a diverticulum; B: Non-bleeding visible vessel: A clearly exposed vessel within a diverticulum without active bleeding; C: Adherent clot: A clot covering a diverticulum that reveals either active bleeding or a visible vessel upon removal.

Procedures

Only initial colonoscopies performed during hospitalization were included in the analysis. All procedures were performed using adult or pediatric water-jet variable-stiffness colonoscopes (Olympus, Tokyo, Japan) equipped with a distal attachment cap; the choice of scope was at the discretion of the endoscopist. Bowel preparation was also determined by the endoscopist based on contrast-enhanced computed tomography (CT) findings and the patient’s overall condition, and consisted of either ≥ 2 L of polyethylene glycol solution or enema preparation. Observation time was defined differently depending on the SRH status: In patients with definite CDB, it was measured from the time of scope insertion to SRH identification, and in patients with presumptive CDB, from the time of scope insertion to withdrawal. In all cases, the observation time was determined based on timestamps recorded on the preserved endoscopic still images. For definite CDB cases, the duration of the therapeutic intervention was excluded from the recorded observation time. The insertion phase was included because SRH can be detected during insertion and a total colonoscopy is not always performed in cases of CDB. There was no standardized institutional strategy regarding observation time, and the decision to conclude the examination was left to the discretion of each endoscopist.

Outcome measures

The primary outcome was the “5% plateau time”, defined as the time point at which the incremental detection rate of SRH per 5-minute interval consistently dropped below 5%, indicating a practical lower bound for adequate observation (Figure 2). Secondary outcomes included stratified analyses of the 5% plateau time based on contrast-enhanced CT findings. Patients were classified into three groups: (1) The no extravasation group, defined as those who either underwent contrast-enhanced CT with no extravasation or did not undergo contrast-enhanced CT (including those who received non-contrast CT or no CT); (2) The right-sided extravasation group; and (3) The left-sided extravasation group. Extravasations were defined as right-sided if they occurred between the cecum and the transverse colon, and left-sided if they occurred between the descending colon and sigmoid colon.

Figure 2 Concept of 5% plateau time. The incremental stigmata of recent hemorrhage detection rates are assessed at 5-minute intervals. The 5% plateau time is defined as the point at which the incremental detection rate consistently fell below 5%. SRH: Stigmata of recent hemorrhage.

Sensitivity analysis

Because the 5% threshold was chosen arbitrarily, we performed sensitivity analyses using 3% and 10% thresholds to assess the robustness and clinical relevance of our findings. For each threshold, we identified the corresponding plateau time and calculated the cumulative SRH detection rate achieved by that time point. This allowed for a comparison of the diagnostic yield and procedural duration across different threshold values.

Statistical analysis

The patient demographics and clinical characteristics were summarized for the entire cohort. The observation times were described for both the overall population and the subset with SRH identification, and histograms were generated for visualization. Continuous variables are reported as means with standard deviations or medians with interquartile ranges (IQR), and categorical variables as frequencies and percentages.

The cumulative SRH detection rates were plotted over time for definite CDB cases. To assess the plateau in detection, SRH detection rates were calculated in 5-min increments, and both cumulative and interval-based detection trends were visualized to identify the 5% plateau time. Time points corresponding to 50%, 80%, 90%, and 95% of cumulative SRH detections were also determined. Sensitivity analyses with 3% and 10% thresholds were conducted using the same approach to evaluate the robustness and clinical relevance of the 5% threshold.

Additionally, we described the 30-day rebleeding rates according to SRH detection status. Among patients without SRH, the rates were compared between those with observation times above and below the 5% plateau.

Subgroup analyses were conducted based on CT-based localization by individual endoscopists. Categorical variables were analyzed using the Fisher’s exact test for comparison across endoscopists. Parametric continuous variables were evaluated using one-way analysis of variance, and non-parametric continuous variables were compared using the Kruskal-Wallis test. All statistical tests were two-sided, with a P-value < 0.05 considered statistically significant. All analyses were performed using R software (version 4.4.3).

RESULTS

Patient flow

The patient selection process is illustrated in Figure 3. A total of 1039 patients were diagnosed with CDB at initial colonoscopy during the study period. After excluding 194 patients based on the exclusion criteria, 845 patients were included in the final analysis. SRH was detected in 286 patients, yielding a detection rate of 33.8%.

Figure 3 Patient flow diagram of the study. A total of 1039 patients were diagnosed with colonic diverticular bleeding. Finally, 845 patients were included in the analysis.

Baseline demographics

The baseline characteristics of the 845 patients are summarized in Table 1. The mean age of the patients was 77 ± 11 years, and 64.5% were male. The mean body mass index (BMI) of the cohort was 22.9 ± 3.8 kg/m². Antiplatelet and anticoagulant agents were used in 34.0% and 16.3% of the patients, respectively. A shock index ≥ 1 on admission was observed in 8.0% of the patients. Colonoscopy was performed within 12 hours in 26.6% of patients, and 9.2% of patients underwent enema preparation. Based on contrast-enhanced CT findings, 72.9% had either no extravasation or did not undergo contrast-enhanced CT, 15.3% had right-sided extravasation, and 11.8% had left-sided extravasation. The median observation time for the overall cohort was 28 minutes (IQR, 18-44 minutes). Successful endoscopic hemostasis was achieved in 272 patients (32.0%). Among patients with SRH, the success rate was 95.1% (272/286), whereas endoscopic hemostasis was not attempted in patients without SRH.

Table 1 Baseline characteristics of patients with colonic diverticular bleeding, n (%).

Variable	All cases (n = 845)	SRH detected (n = 286)	SRH not detected (n = 559)
Age, mean ± SD (years)	77 ± 11	77 ± 11	77 ± 11
Male	545 (64.5)	182 (63.6)	363 (64.9)
BMI, mean ± SD (kg/m2)	22.9 ± 3.8	22.8 ± 3.8	22.9 ± 3.8
Antiplatelet agent use	287 (34.0)	99 (34.6)	188 (33.6)
Anticoagulant agent use	138 (16.3)	47 (16.4)	91 (16.3)
Shock index ≥ 1 on admission	68 (8.0)	25 (8.7)	44 (7.9)
Syncope episode	59 (7.0)	19 (6.6)	40 (7.2)
Colonoscopy within 12 hours	225 (26.6)	94 (32.9)	131 (23.4)
Colonoscopy within 24 hours	520 (61.5)	198 (69.2)	322 (57.6)
History of colectomy	50 (5.9)	12 (4.2)	38 (6.8)
Enema preparation	78 (9.2)	34 (11.9)	44 (7.9)
Laboratory data
White blood cell count, mean ± SD (109/L)	7.5 ± 2.6	7.5 ± 2.6	7.5 ± 2.6
Hemoglobin, mean ± SD (g/dL)	11.2 ± 2.4	11.3 ± 2.5	11.2 ± 2.4
Hematocrit, mean ± SD (%)	34 ± 7.0	34 ± 7.0	34 ± 7.0
CE-CT findings
No extravasation/no CT performed	616 (72.9)	172 (60.1)	444 (79.4)
Right-sided extravasation	129 (15.3)	64 (22.4)	65 (11.6)
Left-sided extravasation	100 (11.8)	50 (17.5)	50 (8.9)
Observation time (minutes), median (IQR)	28 (18-44)	19 (12-28)	33 (22-48)
Successful endoscopic hemostasis	272 (32.0)	272 (95.1)	-

SRH: Stigmata of recent hemorrhage; IQR: Interquartile range; BMI: Body mass index; SRH: Stigmata of recent hemorrhage; CE-CT: Contrast enhanced computed tomography; CT: Computed tomography.

Outcomes

Among the 845 patients included in the final analysis, SRH was detected during the initial colonoscopy in 286 (33.8%). The median observation time until SRH detection was 19 minutes (IQR, 12-28 minutes). A histogram showing the distribution of observation time in the overall cohort and patients with definite SRH is shown in Figure 4. In patients with definite SRH, the frequency of SRH detection increased in the first approximately 15 minutes of observation and subsequently declined. The cumulative detection curve showed that 50%, 80%, 90%, and 95% of SRH cases were detected at 19, 32, 43, and 51.8 minutes, respectively (Figure 5A). The 5% plateau time was at 40 minutes (Figure 6A). These findings indicate that, although longer observations increase the likelihood of SRH detection, the detection rate does not increase linearly with time. Instead, it plateaued beyond a certain point, suggesting a clinically reasonable minimum duration for an adequate evaluation.

Figure 4 Distribution of observation times during colonoscopy. The overall cohort (n = 845) is shown in light green, and patients with definite stigmata of recent hemorrhage (SRH) (n = 286) are shown in dark green. Observation time includes insertion time but excludes therapeutic intervention time in definite SRH cases. SRH: Stigmata of recent hemorrhage.

Figure 5 Cumulative stigmata of recent hemorrhage detection curves for each group. A: Overall; B: No extravasation; C: Right-sided extravasation; D: Left-sided extravasation. Dashed lines represent the time points at which 50%, 80%, 90%, and 95% of stigmata of recent hemorrhage cases were detected.

Figure 6 Stigmata of recent hemorrhage detection trends per 5-minute interval. A: Overall; B: No extravasation; C: Right-sided extravasation; D: Left-sided extravasation. The 5% plateau time was defined as the time point at which the incremental detection rate consistently fell below 5% per 5-minute interval. Vertical dashed lines indicate the threshold. SRH: Stigmata of recent hemorrhage; CE-CT: Contrast enhanced computed tomography.

We also evaluated the association between SRH and 30-day rebleeding. Patients with SRH had a significantly lower rebleeding rate than those without SRH (21.3% vs 30.2%, P = 0.007). Among the 559 patients without SRH detection, there was no significant difference in rebleeding rates between those observed for ≥ 5% plateau time and those observed for < 5% plateau time (33.0% vs 28.4%, P = 0.26; Table 2).

Table 2 Thirty-day rebleeding rates according to stigmata of recent hemorrhage detection and observation time, n (%).

Status	30-day rebleed	P value
SRH detected	61/286 (21.3)	0.007a
SRH not detected	169/559 (30.2)
Observation time ≥ 5% plateau time	74/224 (33.0)	0.26
Observation time < 5% plateau time	95/335 (28.4)

^aP < 0.01 vs stigmata of recent hemorrhage not detected.

SRH: Stigmata of recent hemorrhage.

When stratified by contrast-enhanced CT findings, SRH detection rates were 27.9% (172/616), 49.6% (64/129), and 50.0% (50/100) in the no extravasation, right-sided extravasation, and left-sided extravasation groups, respectively.

In the no extravasation group, the cumulative detection times for 50%, 80%, 90%, and 95% of SRH cases were 18, 32, 39, and 49 minutes, respectively (Figure 5B), and the 5% plateau time was 40 minutes, with a cumulative detection rate of 90.1% at that time point (Figure 6B). In the right-sided extravasation group, the 50%, 80%, 90%, and 95% time points were 23, 35.8, 52.5, and 56.7 minutes, respectively (Figure 5C), with a 5% plateau time of 35 minutes and a corresponding cumulative detection rate of 79.7% (Figure 6C). In the left-sided extravasation group, 50%, 80%, 90%, and 95% of SRH cases were detected by 17.5, 28.4, 38.6, and 50.5 minutes, respectively (Figure 5D), and the 5% plateau time was 30 minutes with a cumulative detection rate of 82.0% (Figure 6D).

A sensitivity analysis using alternative thresholds yielded similar results. The 10%, 5%, and 3% plateau times were 25, 40, and 45 minutes, respectively (Figure 7), corresponding to cumulative SRH detection rates of 71.0%, 89.2%, and 92.3%, respectively.

Figure 7 Sensitivity analysis of incremental stigmata of recent hemorrhage detection using alternative plateau thresholds. The horizontal dashed lines indicate the thresholds of 3%, 5%, and 10%, and the vertical dashed lines mark the corresponding plateau times. SRH: Stigmata of recent hemorrhage.

Among the four endoscopists who performed ≥ 10% of procedures, SRH detection rates ranged from 23.0% to 51.8%, while median observation times ranged from 20 minutes (IQR, 14-33 minutes) to 39 minutes (IQR, 24-51 minutes) (Table 3). Endoscopist C had the longest observation time but the lowest SRH detection rate (26.1%), whereas Endoscopist B achieved the highest detection rate (51.8%) with a relatively long observation time [27 minutes (IQR, 18-45 minutes)]. Patient characteristics, including age, sex, BMI, colonoscopy timing, and enema preparation, were comparable among the four groups. However, the proportion of patients receiving antiplatelet agents was significantly higher in groups A and B (P = 0.01), and contrast-enhanced CT findings differed modestly but significantly between the groups (P = 0.04). The 5% plateau time matched that of the main analysis (40 minutes) for endoscopists A and B, who also had the highest SRH detection rates. Endoscopists C and D showed plateau times of 45 and 35 minutes, respectively, which deviated slightly from the 40-min benchmark but remained generally comparable.

Table 3 Comparison of patient characteristics and outcomes across endoscopists performing ≥ 10% of procedures, n (%).

Endoscopists	A (n = 113)	B (n = 141)	C (n = 88)	D (n = 161)	P value
SRH detected	49 (43.4)	73 (51.8)	23 (26.1)	37 (23.0)	< 0.001a
Observation time (minutes), median (IQR)	20 (14-33)	27 (18-45)	39 (24-51)	23 (16-36)	< 0.001b
5% plateau time (minutes)	40	40	45	35	-
Age, mean ± SD (years)	79 ± 10	77 ± 11	78 ± 10	76 ± 13	0.42
Male	73 (64.6)	89 (63.1)	45 (51.1)	107 (66.5)	0.10
BMI, mean ± SD (kg/m2)	22.2 ± 3.5	22.7 ± 3.5	23.1 ± 4.2	22.7 ± 3.6	0.28
Antiplatelet agent use	42 (37.2)	53 (37.6)	20 (22.7)	38 (23.6)	0.01c
Anticoagulant agent use	22 (19.5)	17 (12.1)	17 (19.3)	30 (18.6)	0.31
Shock index ≥ 1 on admission	11 (9.7)	7 (5.0)	6 (6.8)	17 (10.6)	0.29
Colonoscopy within 12 hours	33 (29.2)	31 (22.0)	16 (18.2)	41 (25.5)	0.28
History of colectomy	6 (5.3)	6 (4.3)	5 (5.7)	9 (5.6)	0.95
Enema preparation	14 (12.4)	11 (7.8)	5 (5.7)	14 (8.7)	0.38
CE-CT findings					0.04d
No extravasation/no CT performed	76 (67.3)	100 (70.9)	68 (76.1)	123 (76.4)
Right-sided extravasation	26 (23.0)	17 (12.1)	9 (10.2)	25 (15.5)
Left-sided extravasation	11 (9.7)	24 (17.0)	11 (12.5)	13 (8.1)

^aP < 0.001 for comparison of stigmata of recent hemorrhage detection rate among endoscopists.

^bP < 0.001 for comparison of observation time among endoscopists.

^cP < 0.05 for comparison of antiplatelet agent use among endoscopists.

^dP < 0.05 for comparison of contrast-enhanced computed tomography findings among endoscopists.

IQR: Interquartile range; BMI: Body mass index; SRH: Stigmata of recent hemorrhage; CE-CT: Contrast enhanced computed tomography; CT: Computed tomography.

DISCUSSION

The incremental SRH detection rate did not increase linearly with time; rather, it declined after approximately 15 minutes and gradually plateaued, indicating diminishing returns with prolonged observation. Based on a large single-center dataset of 845 patients, the overall 5% plateau time was 40 minutes. When stratified by contrast-enhanced CT findings, the plateau time was 40, 35, and 30 minutes in the no extravasation, right-sided extravasation, and left-sided extravasation groups, respectively. Sensitivity analyses using alternative thresholds demonstrated similar trends; at the 10% plateau threshold, the cumulative SRH detection rate was 71.0%, suggesting that nearly 30% of the cases would have been missed. In contrast, the 3% plateau threshold was reached at 45 minutes, capturing 92.3% of the cases, with only a 3.1% increase over the 5% threshold, despite an additional 5 minutes of observation. Collectively, these findings support the utility of the 5% plateau time as a clinically relevant benchmark for determining the minimum observation time necessary to ensure adequate SRH detection in CDB.

The only prior study to quantitatively assess the relationship between observation time and SRH detection in CDB was conducted by Watanabe et al[19], who used receiver operating characteristic curve analysis to identify a 19-minute cutoff and reported that prolonged observation was an independent predictor of SRH detection (odds ratio 10.3, 95% confidence interval: 3.84-27.9, P < 0.001). Although the Watanabe et al’s study[19] highlights observation time as a modifiable factor, it has several limitations. Receiver operating characteristic curve analysis treats SRH detection as a binary outcome and optimizes sensitivity and specificity at a single threshold, which may oversimplify the gradual, time-dependent nature of detection. Clinically, a practical minimum observation time is needed that ensures sufficient diagnostic yield before termination instead of a statistically optimal time point. The 5% plateau time addresses this by defining the point of diminishing returns in the cumulative SRH detection. In addition, the Watanabe et al’s study[19] measured the time from cecal intubation, which may not reflect real-world workflows, in which colonoscopy is often targeted based on CT findings. Moreover, the therapeutic time was included in the observation period, potentially overestimating the actual inspection time. By contrast, our study measured the observation time from scope insertion and excluded the therapeutic duration, providing a more accurate and clinically relevant benchmark. Thus, the 5% plateau time offers a standardized and pragmatic threshold for the minimum inspection time in CDB management.

The variation in plateau times across contrast-enhanced CT findings is likely attributable to differences in the anatomical route and the time required to reach the bleeding site during colonoscope insertion. In left-sided colonic lesions, early access to the antegrade insertion from the rectum allows shorter observation times. This aligns with the 30-minute plateau time observed in the left-sided extravasation group. By contrast, right-sided lesions, which are anatomically distant, may require longer insertion and detection times. Additionally, SRH detection patterns in the right-sided group exhibited a bimodal distribution, possibly reflecting two distinct scenarios: Cases with ongoing active bleeding and those in which bleeding temporarily ceased, necessitating meticulous inspection. In the no extravasation group, comprehensive observation of the entire colon was often required, which resulted in the longest plateau time.

This study is the first to propose a quantitative indicator that addresses both the quality and duration of colonoscopic observations in CDB, providing a clinically interpretable threshold for adequate SRH detection. The derived 5% plateau time roughly corresponded to the time point at which 80%-90% of SRH cases were detected, supporting its validity as a clinically meaningful benchmark. Rather than aiming to maximize detection indefinitely, this indicator offers a practical guide for determining when additional observation time is unlikely to provide substantial additional benefits, thus helping clinicians decide when to conclude an examination. This is particularly important in CDB, a condition that primarily affects older adults who are more vulnerable to complications associated with prolonged sedation-assisted procedures, including aspiration pneumonia[15]. Although we did not conduct a formal cost-effectiveness analysis, the 5% plateau time may offer utility in resource-limited settings by supporting a more standardized and efficient use of endoscopic resources. Future prospective studies incorporating health economic evaluations are needed to clarify the cost-benefit profile. Despite advances in hemostatic techniques such as the shift from indirect to direct clipping and the adoption of endoscopic band ligation, which have reduced rebleeding rates from approximately 40% to below 20%[20-23], SRH detection during initial colonoscopy remains suboptimal at approximately 30%[16]. To further improve outcomes, future development should focus on enhancing SRH detectability, including the integration of artificial intelligence-assisted diagnostic systems and advanced imaging technologies.

Limitations

First, this was a retrospective, single-center study, and institutional factors may have influenced the findings. However, rather than imposing a strict observation protocol, this study allowed procedural discretion to reflect the variation observed in actual practice and identify the minimum observation time applicable to diverse operator behaviors. Notably, we observed considerable interendoscopist differences in both SRH detection rates and observation times, reflecting the inherent heterogeneity of clinical practice. Despite this variation, the 5% plateau time remained relatively consistent (35-45 minutes) among endoscopists, suggesting that the proposed threshold may serve as a practical lower bound for adequate observation in typical settings. Second, our definition of observation time included the insertion phase, which may have been affected by endoscopist proficiency and patient anatomy. However, in real-world practice, particularly when contrast-enhanced CT is used to locate the bleeding site, total colonoscopy is often omitted, and lesion identification frequently occurs during insertion. Given the clinical context, separating insertion time from observation time is not always practical. Therefore, inclusion of the insertion phase is considered clinically appropriate and relevant to how colonoscopy is often performed in patients with CDB. Moreover, all procedures were performed by endoscopists with experience exceeding 1000 colonoscopies; cases involving trainees were excluded to ensure technical consistency. Third, the study focused on a Japanese population, in which right-sided diverticula are more prevalent. The applicability to other populations, particularly in Western countries where left-sided diverticula predominate, remains uncertain and warrants further validation. Fourth, the SRH detection rates at the 5% plateau time were 79.7% and 82.0% in the right- and left-sided extravasation groups, respectively, indicating that approximately 20% of the SRH cases may not have been detected within this timeframe. Therefore, although the plateau time may serve as a practical minimum benchmark, it should not be regarded as a definitive endpoint for terminating the examination.

CONCLUSION

We introduced a novel concept, the “5% plateau time”, defined as the earliest time at which the incremental SRH detection rate falls below 5%. The overall plateau time was 40 minutes, with subgroup-specific values of 40, 35, and 30 minutes for the no, right-sided, and left-sided extravasation groups, respectively. The 5% plateau time provides a scientifically grounded and practical minimum observation time for colonoscopy in patients with CDB, offering a benchmark to standardize practice, reduce the risk of underdetection, and support consistent and efficient management.

ACKNOWLEDGEMENTS

We thank the team members of the Gastroenterology Medicine Center and Emergency Department at Shonan Kamakura General Hospital for their dedication to daily clinical practice.