If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Reprint requests: Naoki Ishii, MD, PhD, MPH, Department of Gastroenterology, Tokyo Shinagawa Hospital, 6-3-22 Higashi-Ooi, Shinagawa-ku, Tokyo 140-8522, Japan.
Division of Endoscopy and Ultrasonography, Department of Clinical Pathology and Laboratory Medicine, Kawasaki Medical School General Medical Center, Okayama, Japan
Department of Gastroenterology and Hepatology, Center for Digestive Disease and Division of Endoscopy, University of Miyazaki Hospital, Miyazaki, Japan
Department of Emergency Medicine, Fujita Health University Hospital, Aichi, JapanEmergency and Critical Care Center, Saiseikai Yokohamashi Tobu Hospital, Kanagawa, Japan
Digestive and Lifestyle Diseases, and Hygiene and Health Promotion Medicine, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan
Department of Gastroenterological Endoscopy, Tokyo Medical University, Tokyo, JapanDepartment of Gastroenterology and Hepatology, National Center for Global Health and Medicine, Tokyo, Japan
We aimed to determine the optimal timing of colonoscopy and factors that benefit patients who undergo early colonoscopy for acute lower GI bleeding.
Methods
We identified 10,342 patients with acute hematochezia (CODE BLUE-J study) admitted to 49 hospitals in Japan. Of these, 6270 patients who underwent a colonoscopy within 120 hours were included in this study. The inverse probability of treatment weighting method was used to adjust for baseline characteristics among early (≤24 hours, n = 4133), elective (24-48 hours, n = 1137), and late (48-120 hours, n = 1000) colonoscopy. The average treatment effect was evaluated for outcomes. The primary outcome was 30-day rebleeding rate.
Results
The early group had a significantly higher rate of stigmata of recent hemorrhage (SRH) identification and a shorter length of stay than the elective and late groups. However, the 30-day rebleeding rate was significantly higher in the early group than in the elective and late groups. Interventional radiology (IVR) or surgery requirement and 30-day mortality did not significantly differ among groups. The interaction with heterogeneity of effects was observed between early and late colonoscopy and shock index (shock index <1, odds ratio [OR], 2.097; shock index ≥1, OR, 1.095; P for interaction = .038) and performance status (0-2, OR, 2.481; ≥3, OR, .458; P for interaction = .022) for 30-day rebleeding. Early colonoscopy had a significantly lower IVR or surgery requirement in the shock index ≥1 cohort (OR, .267; 95% confidence interval, .099-.721) compared with late colonoscopy.
Conclusions
Early colonoscopy increased the rate of SRH identification and shortened the length of stay but involved an increased risk of rebleeding and did not improve mortality and IVR or surgery requirement. Early colonoscopy particularly benefited patients with a shock index ≥1 or performance status ≥3 at presentation.
Colonoscopy plays both diagnostic and therapeutic roles in patients with acute lower GI bleeding (ALGIB). Although guidelines for the treatment of ALGIB have been published in the United States,
do not recommend early colonoscopy for patients with ALGIB as part of the usual clinical practice.
Several randomized controlled trials (RCTs) revealed that early (≤24 hours) colonoscopy neither improves stigmata of recent hemorrhage (SRH) identification nor reduces rebleeding or mortality.
Systematic reviews of studies on ALGIB have concluded that although early colonoscopy is associated with higher rates of endoscopic intervention, there is no improvement in rebleeding or mortality.
One possible reason for the discrepancy could be that background characteristics were balanced in the design phase and selection bias by indication was eliminated in RCTs. In contrast, this may be related to the restriction of patients enrolled in RCTs (eg, exclusion because of age, comorbidities, and severity of clinical status), which has contributed to the lower proportion of hemodynamically unstable patients in RCTs compared with that in observational studies. Considering the limited number of ALGIB patients recruited and the low incidences of rebleeding and mortality, it was considered difficult to observe significant differences for clinical outcomes in RCTs.
Although observational studies using insurance databases have included a large number of patients, detailed patient-level data were lacking, and it was not ideal in assessing emergency interventions such as colonoscopy and interventional radiology (IVR).
most of the elective group underwent colonoscopy within 24 to 48 hours. Outcomes may differ if the timing of colonoscopy is classified into 24 to 48 hours and 48 to 120 hours, and it may be valuable to determine whether to perform colonoscopy on weekends or holidays.
We hypothesized that we could elucidate the optimal timing of colonoscopy for ALGIB and specify the patients who benefited from early colonoscopy after presentation by using a large database of detailed patient data.
In this multicenter cohort study, we evaluated the effects of colonoscopy timing on the clinical outcomes of patients with ALGIB. Further, we investigated the interaction between colonoscopy timing and baseline characteristics for the primary outcome and aimed to identify which patients with ALGIB benefited from early colonoscopy.
Methods
Patients and study design
This multicenter, retrospective, cohort study was conducted in Japan using real-world data collected by gastroenterologists who were directly involved in the treatment of hematochezia between January 2010 and December 2019. The CODE-BLUE-J study has been described previously (Supplementary Table 1, available online at www.giejournal.org).
We identified 10,342 patients hospitalized for acute hematochezia and excluded those who did not undergo colonoscopy (4 patients died within 24 hours after admission, before they underwent colonoscopy, because of multiple organ failure caused by liver cirrhosis, hepatoma, or sepsis), underwent colonoscopy >120 hours after presentation, experienced postprocedural bleeding, or were diagnosed with upper GI bleeding and colorectal tumors. A few studies have noted that tumor bleeding is associated with substantially higher mortality
; hence, we excluded patients with tumor bleeding in the present study. Additionally, because we intended to compare the outcomes using the propensity score, we excluded patients with second or subsequent episodes of acute hematochezia. Ultimately, a cohort of 6270 patients with ALGIB was included and divided into the early (≤24 hours), elective (24-48 hours), and late (48-120 hours) colonoscopy groups (Fig. 1). Considering the insufficient evidence of colonoscopy after 24 hours and the management of patients with ALGIB on weekends or holidays, we set the elective and late groups.
Certain terms have been defined for this study. SRH was defined as the presence of active bleeding or detection of a nonbleeding visible vessel or an adherent clot on colonoscopy.
Rebleeding was defined as a significant quantity of fresh blood loss or passage of wine-colored stools after colonoscopy, associated with any of the following: systolic blood pressure <100 mm Hg, pulse rate ≥100 beats/min, or >2 g/dL decrease in hemoglobin levels.
Furthermore, the requirement for IVR or surgery after colonoscopy during admission was evaluated. Blood transfusion was measured by the amount of packed red blood cells (PRBCs) received during admission, and the length of stay was measured in days.
The study protocol was approved by the Institutional Ethics Committee of Tokyo Medical University (T2019- 0244). A single institutional review board review was applied to this study, and approval from the ethical committees and institutional review boards of all 49 participating hospitals was obtained through the opt-out method (Supplementary Table 1). The requirement for acquiring informed consent from patients was waived because of the retrospective nature of this study.
Outcomes
The primary outcome was 30-day rebleeding rate. Secondary outcomes were the rate of SRH identification, the requirement for corrective IVR or surgery, 30-day mortality, amount of PRBCs, and length of stay. In the subgroup analyses, we aimed to clarify the characteristics of patients who benefited from early colonoscopy for 30-day rebleeding. The requirements of IVR or surgery, mortality, blood transfusion, and length of stay were considered to be located downstream of rebleeding.
The characteristics of 6270 patients with ALGIB are shown in Supplementary Table 2 (available online at www.giejournal.org). Three data sets were derived from the 6270 cases: early versus elective, elective versus late, and early versus late.
Missing data were observed in this study; a missing-data analysis was performed using Little’s test of missing completely at random test in each data set.
The probability that the data were missing completely at random was <.0001 in the 3 data sets. Because complete-case analysis is valid only under missing completely at random,
fully conditional specification, or sequential generalized regression. No missing data were noted in colonoscopy timing and clinical outcomes including 30-day rebleeding, the requirement of IVR or surgery, and mortality. The outcome of 30-day rebleeding and patient characteristics were included in the imputation models, and 20 imputed sheets were created for each set.
A propensity score was constructed in each sheet of imputed data using age ≥75 years; sex
and colorectal surgery, hypertension, diabetes mellitus, dyslipidemia, and cerebral and cardiovascular diseases; malignancy; Charlson comorbidity index score ≥3
; creatinine level; C-reactive protein level; prothrombin time-international normalized ratio; platelet count; and bleeding causes for colonoscopy timing. The inverse probability of treatment weighting (IPTW) method was used to adjust for the baseline characteristics among groups, and the average treatment effect of colonoscopy timing was evaluated for outcomes on the multiply imputed data in each comparison group. The average treatment effect is an average effect on all subjects in the population. The average treatment effect of colonoscopy timing is considered the effect of the experimental arm of colonoscopy timing on all subjects in the population if patients were forced to receive the experimental arm of colonoscopy timing. A robust sandwich variance estimator was used to correctly estimate the variance and confidence interval (CI) for the pseudo-population.
Sensitivity analyses were performed by IPTW for the complete-case data (n = 4148) and multivariate logistic and linear regression for the multiply imputed data (n = 6270). SRH identification, 30-day rebleeding, corrective IVR or surgery requirement, 30-day mortality, amount of PRBCs, and length of stay were compared depending on the timing of colonoscopy.
The interaction between the timing of colonoscopy (early vs elective vs late) and baseline characteristics was investigated for 30-day rebleeding using multivariate logistic regression on multiply imputed data. If an interaction was observed, 3 data sets (early vs elective, elective vs late, and early vs late) were divided depending on the characteristics on which the interaction was observed. Multiple imputation was performed by chained equations in each divided data set. A propensity score was constructed in each sheet of the multiply imputed data, and the average treatment effect of colonoscopy timing was investigated for outcomes using the IPTW method.
A 2-tailed P < .05 was considered statistically significant. Multiple subgroup analyses can lead to false-positive results.
A P value was not demonstrated, and the magnitude and direction of treatment differences and a corresponding 95% CI were described in the subgroup analyses. Statistical analyses were performed using the Stata software, version 16 (Stata Corp LP, College Station, Tex, USA).
Results
Patient characteristics
Detailed characteristics of patient with acute hematochezia were published previously in the CODE BLUE-J study report.
We studied 6270 patients with hematochezia who met the inclusion criteria. The characteristics of the study population are shown in Supplementary Table 2. More than 10% of the information on current smoking, current drinking, and prothrombin time-international normalized ratio was missing. However, the data of colonoscopy timing were all included in the present study. The predominant causes of ALGIB were colonic diverticular bleeding, ischemic colitis, rectal ulcer, inflammatory bowel disease, hemorrhoids, and small-bowel bleeding.
Outcomes compared among the 3 groups on weighting multiply imputed data
Of 6270 patients, 4133 (66%), 1137 (18%), and 1000 (16%) patients were categorized into the early (≤24 hours), elective (24-48 hours), and late (48-96 hours) groups, respectively (Table 1). The baseline characteristics were not balanced in observed and multiply imputed data. Therefore, weighting multiply imputed data was conducted, and most absolute standardized differences became <.1. Outcomes among the groups using observed data (Supplementary Table 3, available online at www.giejournal.org) and the IPTW method are demonstrated (Table 2). The rate of SRH identification was significantly higher in the early group: early versus elective (ref) (odds ratio [OR], 1.785; 95% CI, 1.521-2.094), elective versus late (ref) (OR, 1.505; 95% CI, 1.188-1.906), and early versus late (ref) (OR, 2.562; 95% CI, 2.089-3.143). More endoscopic therapies were performed in the early group (Supplementary Table 4, available online at www.giejournal.org). However, the results of 30-day rebleeding were diametrically opposite (Table 2). Early colonoscopy was associated with higher 30-day rebleeding: early versus elective (ref) (OR, 1.347; 95% CI, 1.119-1.622), elective versus late (ref) (OR, 1.712; 95% CI, 1.287-2.277), and early versus late (ref) (OR, 2.259; 95% CI, 1.752-2.914). No significant differences were noted in the requirement for IVR or surgery, mortality, and transfused PRBCs among the groups. Length of stay was significantly shorter in the early group than in the elective and late groups.
Table 1Patient characteristics of observed, multiply imputed, and weighting multiply imputed data between colonoscopy timing (n = 6270)
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
171 (4.1)
96 (8.4)
77 (7.7)
.138
.032
.170
.256
.157
.103
.062
.145
.043
Values are n (%). The covariates used for the construction of the propensity score were well balanced in the weighting multiply imputed data.
∗ Malignancy included solid tumors and malignant lymphomas.
† Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Our analyses did not use the Oakland score as a covariate because we created a propensity score using the covariates used to estimate the Oakland score.
For reference, we also included the Oakland score in our analyses, but the results were similar.
Sensitivity analyses
We evaluated the robustness of our results by using the IPTW method of complete-case data and multivariate regression (Table 3). Baseline characteristics of weighting complete data are demonstrated in Supplementary Table 5 (available online at www.giejournal.org). The baseline characteristics were well balanced among the 3 groups in the weighting complete data. The higher rates of SRH identification and 30-day rebleeding were observed in the early group, similar to the results on weighting multiply imputed data. There were no significant differences in the requirement for IVR or surgery and transfused PRBCs among the groups. Mortality was significantly higher in the early group than in the late group (OR, 4.584; 95% CI, 1.027-20.466). Length of stay was significantly shorter in the early group than in the late group (Table 3).
Table 3Outcomes between colonoscopy timing using inverse probability of treatment weighting on the complete data and multivariate logistic regression on the multiply imputed data
Identification of stigmata of recent hemorrhage
30-day rebleeding
Interventional radiology or surgery
30-day mortality
Blood transfusion (packs)
Length of stay (days)
OR, 95% CI
P value
OR, 95% CI
P value
OR, 95% CI
P value
OR, 95% CI
P value
Coefficient, 95% CI
P value
Coefficient, 95% CI
P value
inverse probability of treatment weighting on complete data
Elective
Reference
Reference
Reference
Reference
Reference
Reference
Early
1.914, 1.566-2.337
<.001
1.340, 1.070-1.677
.011
1.127, .683-1.860
.638
1.252, .408-3.842
.694
–.246, –.733 to .240
.321
–.844, –1.765 to .077
.073
Late
Reference
Reference
Reference
Reference
Reference
Reference
Elective
1.285, .963-1.713
.088
1.652, 1.174-2.324
.004
1.006, .541-1.868
.986
2.297, .412-12.825
.343
.563, –.070 to 1.196
.081
–.215, –1.689 to 1.260
.776
Late
Reference
Reference
Reference
Reference
Reference
Reference
Early
2.361, 1.848-3.015
<.001
2.047, 1.506-2.783
<.001
.960, .575-1.603
.876
4.584, 1.027-20.466
.046
.132, –.265 to .529
.515
–1.044, –1.200 to –.088
.032
Multivariate logistic and linear regression on multiply imputed data
Late
Reference
Reference
Reference
Reference
Reference
Reference
Elective
1.595, 1.252-2.031
<.001
1.755, 1.320-2.333
<.001
.966, .569-1.639
.897
1.336, .378-4.721
.653
.405, .046-.765
.027
–.640, –1.450 to .170
.121
Early
2.934, 2.396-3.593
<.001
2.409, 1.889-3.072
<.001
.968, .628-1.493
.883
1.148, .392-3.360
.801
.041, –.256 to .338
.787
–1.578, –2.248 to –.908
<.001
Elective
Reference
Reference
Reference
Reference
Reference
Reference
Early
1.840, 1.562-2.166
<.001
1.372, 1.140-1.652
.001
1.003, .666-1.510
.990
.860, .331-2.232
.756
–.364, –.642 to –.087
.010
–.937, –1.563 to –.312
.003
P < .05 was considered to indicate statistical significance. CI, Confidence interval; OR, odds ratio.
Multivariate logistic regression using the multiply imputed data revealed that the rates of SRH identification and 30-day rebleeding were significantly higher in the early group (Table 3). No significant differences were observed in the requirement for IVR or surgery and mortality among the groups. Length of stay was shorter in the early group than in the elective and late groups.
Subgroup analyses based on shock index and PS
The interaction between the timing of colonoscopy (early vs elective vs late) and baseline patient characteristics was investigated for 30-day rebleeding. Interactions were observed between early and late colonoscopy and shock indices of <1 and ≥1 (P for interaction = .038) and PS 0 to 2 and PS ≥3 (P for interaction = .022) regarding 30-day rebleeding (Table 4). Although there was a higher OR of early and late colonoscopy to 30-day rebleeding for both a shock indices of <1 and ≥1 (OR, 2.097; 95% CI, 1.553-2.832 and OR, 1.095; 95% CI, .541-2.218, respectively), the direction of OR of early and late colonoscopy to 30-day rebleeding was opposite in the PS 0 to 2 and PS ≥3 cohorts (OR, 2.481; 95% CI, 1.898-3.244 and OR, .458; 95% CI, .164-1.278, respectively). In addition, the early group had a significantly decreased IVR or surgery requirement in the shock index ≥1 cohort compared with the late group (OR, .267; 95% CI, .099-.721).
Table 4Subgroup analyses depending on shock index and performance status on the weighting multiply imputed data
This large retrospective study was conducted to determine the optimal timing of colonoscopy for patients with ALGIB. Although early colonoscopy was associated with higher SRH identification allowing more frequent endoscopic hemostasis and a shorter length of stay, the 30-day rebleeding rate was significantly higher than elective and late colonoscopy. Conversely, the rates of 30-day rebleeding and IVR or surgery requirement were lower in early colonoscopy for patients with shock index ≥1 or PS ≥3 than in late colonoscopy.
Previous RCTs showed that early (≤24 hours) colonoscopy for ALGIB did not improve SRH identification
Moreover, a single-center RCT revealed that urgent (≤12 hours) compared with nonurgent (>36 hours) colonoscopy for ALGIB did not improve further bleeding.
However, this RCT was discontinued during patient enrollment because of the difficulty in reaching the prespecified sample size, which was a significant limitation of the study. The findings were associated with a potential risk of Type II error and lack of generalizability. Contrarily, 5 retrospective studies
Safety and effectiveness of early colonoscopy in management of acute lower gastrointestinal bleeding on the basis of propensity score matching analysis.
suggested that early (≤24 hours) colonoscopy may increase the yield of endoscopic intervention, decrease the length of stay, and reduce the requirement for blood transfusion. However, none has demonstrated benefits in reducing rebleeding rates or mortality, a finding corroborated by our study. This differences between RCTs and observational studies may be associated with the restriction of patients enrolled in RCTs (eg, exclusion because of high age, comorbidities, and unstable status) and selection bias by indication.
Our study potentially helps estimate the optimal timing of colonoscopy by comparing outcomes among early, elective, and late colonoscopy groups. In addition to our large database with detailed patient data, no studies have been done on colonoscopy conducted after 24 hours of presentation. The early group experienced a higher rate of SRH identification, whereas a higher 30-day rebleeding rate was observed. This may be explained by the fact that bleeding stops spontaneously in most ALGIB patients,
and ALGIB patients may be triaged as those requiring interventions through subsequent colonoscopy. Although the requirements for IVR or surgery and mortality were not different based on the colonoscopy timing, early colonoscopy was significantly associated with a shorter length of stay in the present study. Because this design was a multicenter observational study and an indication for extending the length of stay was not prespecified, the length of stay may be dependent on the timing of colonoscopy and the preference of each gastroenterologist. Outcomes such as 30-day rebleeding, IVR or surgery requirement, and mortality did not differ between elective and late colonoscopy. These results may be valuable, especially in the areas in which there is no timely availability of staff and other support services after work hours on weekends and holidays.
We observed interactions between early and late colonoscopy and patient characteristics (PS and shock index) for 30-day rebleeding, and a subgroup analyses based on PS 0 to 2 and PS ≥3 and shock indices of <1 or ≥1 were performed. The direction and the magnitude of ORs for 30-day rebleeding were different between patients with PS 0 to 2 and PS ≥3. Therefore, PS was determined to be a modifier of the qualitative effect of colonoscopy timing on 30-day rebleeding. Although the magnitude of ORs for rebleeding was different between patients with shock indices of <1 and ≥1, the direction of ORs was the same. The shock index was the quantitative effect modifier between early and late colonoscopy on 30-day rebleeding. However, the direction and magnitude of ORs were different between shock indices of <1 and ≥1 for IVR or surgery requirement. Early colonoscopy significantly decreased the risk of requiring IVR or surgery in patients with shock index ≥1 compared with late colonoscopy.
To find the reason for the interaction, we compared the causes of bleeding by subgroups on shock index and PS. The results of the comparison of bleeding causes regarding shock index and PS subgroups are demonstrated in Supplementary Table 6 (available online at www.giejournal.org). The proportion of rectal ulcer was significantly higher in the PS ≥3 subgroup than in the PS 0 to 2 subgroup. Significant differences were observed in colonic diverticular bleeding, ischemic colitis, rectal ulcer, small-bowel bleeding, colorectal varices, and Dieulafoy’s ulcer between shock indices of <1 and ≥1. The different causes of bleeding may contribute to the effectiveness of colonoscopy timing through baseline characteristics of shock index and PS. Because the guideline of the British Society of Gastroenterology recommends the use of a shock index as a risk stratification tool for patients with ALGIB, and IVR and endoscopic therapy were considered the treatment options depending on the results of enhanced CT in patients with shock index ≥1,3 our results support the management algorithm of this guideline.
The difference in analyses between our study and other studies may be explained by adjusting the bleeding causes. We used the IPTW method as previously conducted by Guo et al,
who performed an observational study of endoscopy timing for upper GI bleeding. Considering the onset of GI bleeding, the causes of bleeding may precede the allocation of endoscopy timing. Bleeding causes were included for the construction of the propensity score to eliminate the confounding effect of bleeding causes in our study (Fig. 2). In the previous observational study, the proportion of bleeding causes was significantly different among the timing of endoscopy.
Certainly, in our study, when baseline characteristics except for causes of bleeding were balanced on our weighted data (Supplementary Table 7, available online at www.giejournal.org), the absolute standardized differences were >.2 regarding the cause of bleeding (Supplementary Table 8, available online at www.giejournal.org). When observational studies of endoscopy timing for GI bleeding are being planned as quasi-RCTs, it may be better to manage the causes of bleeding as covariates to be balanced.
Figure 2Schema of colonoscopy timing for acute lower GI bleeding. Considering the onset of GI bleeding, causes of bleeding precede the allocation of endoscopy timing and are not changed by colonoscopy timing. The causes of bleeding were considered the confounders between colonoscopy timing and clinical outcomes from the viewpoint of causal inference. The causes of bleeding and confounders were not omitted in the analysis of the association between colonoscopy timing and clinical outcomes in the cohort study that matched patient characteristics.
The notable strength of this study was the large population that was analyzed for the influence of pertinent confounding factors on colonoscopy timing such as clinical status, blood investigation findings, comorbidities, and concomitant medications.
Essentially, the proportion of hemodynamically unstable patients is higher in observational studies, including the present study, than in RCTs (6.0%-47.2%
Safety and effectiveness of early colonoscopy in management of acute lower gastrointestinal bleeding on the basis of propensity score matching analysis.
). Furthermore, we investigated the characteristics of patients who benefit from early colonoscopy.
The limitations of this study are its retrospective and observational design. However, we applied robust inclusion and exclusion criteria to eliminate selection bias as much as possible and maintained internal validity. Conversely, the generalizability of results from RCTs may sometimes be reduced because of the differences in patient characteristics between RCTs and observational studies. Second, unmeasured confounders may still exist in the IPTW method. Finally, our study only included patients who received colonoscopy. Patients who received colonoscopy had relatively greater ALGIB severity, and this may be a limitation of generalizability. Nonetheless, by using interaction analysis, we were able to identify the patient characteristics that were considered as indications for early colonoscopy.
In summary, we have demonstrated that early colonoscopy improved SRH identification and shortened the length of stay. However, early colonoscopy was associated with a higher 30-day rebleeding rate and did not improve IVR or surgery requirement or mortality. Early colonoscopy was beneficial for those with a shock index ≥1 or PS ≥3. Therefore, most ALGIB patients do not need to receive colonoscopy immediately but rather vitals and PS can be an indication of the requirement for early colonoscopy. Further RCTs are warranted to clarify these findings.
Acknowledgments
We thank Kazuyo Jo, Shiho Kamimura, Sanae Habu, Akiko Takamatsu, Minako Kajihara, and Kenko Yoshida for their help with data collection and analysis. Access to the data supporting the results of this study can be requested and will be reviewed with the principal investigator of this study through the corresponding author. The data are not available to the public because of privacy and ethical restrictions.
Appendix
Supplementary Table 1Affiliations of and ethics committee approval numbers from 49 participating hospitals
Institution
Prefecture
Department
Ethics committee approval no.
Period of data collection
No. of cases
1
Tokyo Medical University
Tokyo
Department of Gastroenterological Endoscopy
T20190244
2013.01.02 to 2019.11.13
134
2
National Center for Global Health and Medicine
Tokyo
Department of Gastroenterology and Hepatology
3539
2010.01.19 to 2019.06.18
1375
3
Tokyo Shinagawa Hospital
Tokyo
Department of Gastroenterology
20-A-04
2014.07.28 to 2019.12.30
250
4
Nippon Medical School, Graduate School of Medicine
Tokyo
Department of Gastroenterology
B-2020-147
2014.10.31 to 2019.12.23
538
5
Chiba Hokusoh Hospital, Nippon Medical School
Chiba
Department of Gastroenterology
802
2011.08.28 to 2019.12.23
188
6
Saga Medical Center Koseikan
Saga
Department of Gastroenterology
20-01-01-03
2013.01.02 to 2019.12.30
394
7
St Luke’s International University
Tokyo
Department of Gastroenterology
20-R012
2014.01.03 to 2019.12.29
536
8
Kawasaki Medical School
Okayama
Division of Endoscopy and Ultrasonography, and Department of Clinical Pathology and Laboratory Medicine
3890
2014.03.27 to 2018.2.18
156
9
Kawasaki Medical School General Medical Center
Okayama
Division of Endoscopy and Ultrasonography, and Department of Clinical Pathology and Laboratory Medicine
3890
2012.02.07 to 2017.12.27
88
10
University of Tsukuba
Ibaraki
Department of Gastroenterology, and Division of Endoscopic Center
R02-030
2013.01.02 to 2019.12.30
134
11
Tokyo Metropolitan Bokutoh Hospital
Tokyo
Department of Gastroenterology
02-024
2013.04.02 to 2019.12.31
808
12
Saiseikai Yokohamashi Tobu Hospital
Kanagawa
Emergency and Critical Care Center
20200030
2011.01.22 to 2019.12.17
131
13
The University of Tokyo
Tokyo
Department of Gastroenterology
2020067NI
2010.01.04 to 2018.08.10
542
14
Toranomon Hospital
Tokyo
Department of Gastroenterology
2021
2011.04.02 to 2019.12.04
134
15
Nagoya University Hospital
Aichi
Department of Endoscopy
2020-0152
2016.01.02 to 2019.12.27
206
16
Hiroshima City Asa Citizens Hospital
Hiroshima
Department of Gastroenterology
2002/1/24
2010.01.02 to 2019.12.27
517
17
National Hospital Organization Fukuokahigashi Medical Center
Fukuoka
Department of Gastroenterology and Hepatology
2020-臨-2
2017.04.02 to 20.9.12.3
107
18
Nara City Hospital
Nara
Department of Gastroenterology and Hepatology, and Center for Digestive and Liver Diseases
NCH倫20-8
2013.05.02 to 2019.12.25
264
19
Graduate School of Medical and Dental Sciences, Niigata University
Niigata
Division of Gastroenterology
2020-0052
2010.8.28 to 2019.12.18
142
20
St Marianna University School of Medicine
Kanagawa
Division of Gastroenterology and Hepatology, Department of Internal Medicine
4802
2013.08.12 to 2019.12.13
343
21
Oita University
Oita
Department of Gastroenterology
1845
2010.03.06 to 2019.12.03
199
22
Tokyo Saiseikai Central Hospital
Tokyo
Department of Internal Medicine
2020-015-01
2017.01.09 to 2019.12.29
120
23
Fukuoka University Hospital
Fukuoka
Department of Gastroenterological Endoscopy
U20-05-016
2016.03.30 to 2019.12.12
80
24
Fukuoka University Chikushi Hospital
Fukuoka
Department of Gastroenterology
C20-052
2010.01.05 to 2017.03.04
199
25
Kitano Hospital, Tazuke Kofukai Medical Research Institute
Osaka
Department of Gastroenterology and Hepatology
P200500400
2010.1.24 to 2019.12.21
586
26
Graduate School of Medical Sciences, Kyushu University
Fukuoka
Department of Medicine and Clinical Science
2020-289
2010.05.30 to 2019.12.25
101
27
University of Miyazaki Hospital
Miyazaki
Department of Gastroenterology and Hepatology, and Center for Digestive Disease and Division of Endoscopy
0-0734
2010.07.16 to 2019.10.20
142
28
University of the Ryukyus Hospital
Okinawa
Department of Endoscopy
1656
2016.01.04 to 2019.12.08
86
29
Naha City Hospital
Okinawa
Department of Gastroenterology
2004a4
2018.11.02 to 2019.12.26
121
30
Kagoshima University Graduate School of Medical and Dental Sciences
Kagoshima
Digestive and Lifestyle Diseases
200041疫
2019.02.26 to 2019.10.07
7
31
Kagoshima City Hospital
Kagoshima
Department of Gastroenterology
2020-25
2019.01.27 to 2019.12.27
45
32
Kagoshima Kouseiren Hospital
Kagoshima
Department of Gastroenterology
215
2019.01.03 to 2019.12.25
28
33
Kagoshima Medical Center
Kagoshima
Department of Gastroenterology
2020-22
2019.01.01 to 2019.12.25
40
34
Izumi General Medical Center
Kagoshima
Department of Gastroenterology
60
2019.01.08 to 2019.12.26
34
35
Kirishima City Medical Association Medical Center
Kagoshima
Department of Gastroenterology
202005
2019.01.16 to 2019.10.12
19
36
Kagoshima Prefectural Oshima Hospital
Kagoshima
Department of Gastroenterology
97
2018.12.25 to 2019.12.27
23
37
National Hospital Organization Kyoto Medical Center
Kyoto
Department of Gastroenterology
20-020
2011.05.31 to 2019.12.20
205
38
Fukushima Medical University
Fukushima
Department of Gastroenterology
一般2020-112
2013.12.07 to 2019.12.26
100
39
Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital
Tokyo
Department of Gastroenterology
2503
2010.06.15 to 2019.12.16
205
40
Kitasato University, School of Medicine
Kanagawa
Department of Gastroenterology
C20-174
2010.10.07 to 2019.12.26
76
41
Suita Municipal Hospital
Osaka
Department of Gastroenterology and Hepatology
2020-研2
2011.03.04 to 2019.12.30
88
42
Akita University Graduate School of Medicine
Akita
Department of Gastroenterology and Neurology
2491
2010.09.22 to 2019.05.11
69
43
Japanese Red Cross Shizuoka Hospital
Shizuoka
Department of Gastroenterology
Jun-20
2010.01.08 to 2019.12.30
173
44
Hirosaki University Hospital
Aomori
Division of Endoscopy
2020-32
2010.04.16 to 2019.12.28
112
45
Graduate School of Medical Sciences, Kumamoto University
Kumamoto
Department of Gastroenterology and Hepatology
2040
2010.01.03 to 2019.12.22
111
46
National Hospital Organization Kyushu Medical Center
Values are n (%) unless otherwise defined. Colonic diverticular bleeding is the most common cause (71.7%). More than 10% of the information of current smoking, current drinking, and prothrombin time-international normalized ratio was missed.
SD, Standard deviation.
∗ Malignancy included solid tumors and malignant lymphomas.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
115 (4.3)
59 (7.6)
58 (8.4)
0
.007
0
Values are n (%). The covariates used for the construction of the propensity score were well balanced in the weighting complete data.
∗ Malignancy included solid tumors and malignant lymphomas.
† Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
CDB included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
CDB included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
CDB included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
CDB included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
CDB included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
318 (5.7)
22 (4.3)
.18
315 (5.4)
26 (7.0)
.18
Values are n (%). The proportions of CDB and rectal ulcer were significantly higher in the PS ≥3 subgroup. The significant differences were observed in CDB, ischemic colitis, rectal ulcer, small-bowel bleeding, colorectal varices, and Dieulafoy's ulcer between the shock index <1 and shock index ≥1 cohorts. The data of PS and shock index were missed in 78 and 125 patients. CDB, Colonic diverticular bleeding; PS, performance status.
∗ CDB included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Malignancy included solid tumors and malignant lymphomas.
.020
.020
.000
.002
.001
.018
Charlson comorbidity index ≥3
.008
.001
.008
.004
.002
.016
Medications
Antiplatelet
.071
.007
.078
0
.001
.026
Anticoagulant
.039
.066
.104
.013
.003
.001
Nonsteroidal anti-inflammatory drugs
.036
.017
.054
.014
.001
.009
Initial measurements
Shock index ≥1
.136
.133
.003
.003
.007
.022
Syncope
.100
.056
.156
.009
.001
.002
Abdominal pain
.129
.290
.418
.001
.003
.002
Diarrhea
.101
.163
.262
.005
.001
.014
Hemoglobin ≤11 g/dL
.046
.032
.014
.004
.004
0
Albumin ≤3.0 g/dL
.028
.067
.039
.004
.001
.005
Creatinine ≤1.5 mg/dL
.047
.118
.071
.001
.002
.005
C-reactive protein ≥5 mg/dL
.079
.037
.116
.004
.009
.003
Prothrombin time-international normalized ratio ≥1.5
.076
.036
.040
.007
.002
.008
Platelets ≤15 104/μL
.003
.011
.008
.002
.004
.003
The propensity scores were constructed by demographic characteristics, medications, and initial measurements. The causes of bleeding were not used for the propensity score.
∗ Malignancy included solid tumors and malignant lymphomas.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
.178
.027
.151
.120
.027
.198
The proportion of colonic diverticular bleeding was not well balanced before and after the inverse probability of treatment weighting.
∗ Colonic diverticular bleeding included definitive and presumptive. Enterocolitis included infectious colitis, inflammatory bowel disease, drug-induced colitis, nonspecific ulcer, and nonspecific colitis. Anal bleeding included hemorrhoids and anal lesions such as anal laceration. Small-bowel bleeding included definitive small-bowel bleeding, presumptive small-bowel bleeding, and Meckel's diverticulum. Other diseases included miscellaneous and unknown lesions.
Safety and effectiveness of early colonoscopy in management of acute lower gastrointestinal bleeding on the basis of propensity score matching analysis.
DISCLOSURE: All authors disclosed no financial relationships. Researchsupportfor this study was provided by the Ministry of Health, Labour and Welfare, Japan (grantno. 19HB1003), JSPS KAKENHI (grantnos. JP17K09365 and 20K08366), Smoking Research Foundation, Takeda Science Foundation, andGrants-in-Aid for Research from the National Center for Global Health and Medicine (grantnos. 29-2001, 29- 2004, 19A1011, 19A1022, 19A-2015, 29-1025, and 30-1020).
DIVERSITY, EQUITY, AND INCLUSION: We worked to ensure gender balance in the recruitment of human subjects. We worked to ensure ethnic or other types of diversity in the recruitment of human subjects. We worked to ensure that the language of the study questionnaires reflected inclusion. The author list of this paper includes contributors from the location where the research was conducted who participated in the data collection, design, analysis, and/or interpretation of the work.
01863-6&id=fx1.jpg){kind=link}
01863-6&id=gr1.jpg){kind=link}
01863-6&id=gr2.jpg){kind=link}