The risk factors for post-polypectomy bleeding and establishment of a risk-scoring model for small colorectal polyps (<1.5cm) in an ambulatory surgery center: a retrospective analysis


 Background: As one of the most common complications of colonoscopy, the risk factors of post-polypectomy bleeding (PPB) has been rarely explored in an ambulatory surgery unit. We aim to develop a risk-scoring model to predict the risk of PPB forsmall colorectal polyps (<1.5cm) in an ambulatory surgery unit. Methods: The patients with single small colorectal polyps (<1.5cm) who underwent endoscopic polypectomy in the Ambulatory Surgery Center of our hospital between January 2014 and June 2017 were included and retrospectively reviewed. We analyzed patient’s clinical characteristics, morphological and pathological characteristics of polyps, polypectomy techniques, and the occurrence of PPB. Risk factors of PPB were identified with a multivariable logistic regression model. In addition, a risk-scoring system was developed and validated eventually. Results: Among the 771 patients enrolled, 26 (3.4%) patients suffered PPB. The male gender, elderly age (≥ 60 years), using hot biopsy forceps as polypectomy technique adenoma in histopathology, complicated withhypertension, use of anticoagulant or antiplatelet agents, and early excessive activities significantly increased the risk of PPB (P＜0.05) as indicated by the results of multivariable logistic regression analysis. The area under the ROC curve (AUC) in the model group (0.890) and validation group (0.924) indicated that the risk-scoring model could predict the occurrence of PPB effectively. Conclusions: This risk-scoring method may help to predict the risk of PPB forsmall colorectal polyps, fit well in the Ambulatory Surgery Center, and provide a new approach to help reduce the incidence of hemorrhage after colorectal polypectomy.Trial registration: This study was retrospectively registered and approved by the Ethics Committee of West China Hospital of Sichuan University (IRB number: ChiCTR1800020201).


Background
Colorectal polypectomy is currently considered an effective strategy to reduce the incidence of colorectal cancer. Colonoscopic polypectomy is routinely believed to be a safe procedure; however postpolypectomy bleeding (PPB), which is one of the most frequent complications after endoscopic operations, may cause serious problems and adverse consequences.
Ambulatory surgery provides high quality and e cient care for a wide variety of surgical procedures.
During the last decades, ambulatory surgery has grown rapidly and now accounts for the majority of operations performed in endoscopic therapy in China. On the other hand, the safety of day surgery must be emphasized. Thus, the safety of colonoscopic polypectomy in the Day Surgery Unit should be taken into account when discussing cost-effective economy and rapid recovery.
Previous studies mostly involved multiple or large colorectal polyps among inpatients.The patients underwent the polypectomy in the Day Surgery Unit and were characterized with smaller size polyps (≤1.5 cm), younger population and limited complications. They were encouraged to perform appropriate activities and diets after the polypectomy as soon as possible. Consequently, it is essential to balance the safety and e ciency via investigating the risk factors of PPB based on day surgery. Furthermore, it would be valuable to establish a risk-scoring model of ambulatory surgery to predict the occurrence of PPB, as little research has described it ever before.

Patients
The records of 2,744 patients who presented with colorectal polyps and underwent an endoscopic colorectal polypectomy in the Day Surgery Unit of West China Hospital, Sichuan University from January 2014 to June 2017 (total 42 months) were reviewed and analyzed. Inclusion criteria were (1) patients with single colorectal polyp, (2) a polyp size ≤ 15mm and (3) aged between 14 and 80 years old. In addition, (4) all patients were required to meet the American Association of Anesthesiologists (ASA) score of less than 3. Patients ( ) with multiple-colorectal polyps, ( ) a laterally spreading tumor (LST), and ( ) a history of in ammatory bowel disease (IBD) or ( ) hemorrhagic disease were excluded. We also excluded (v) the cases of carcinoma which were pathologically con rmed after polypectomy, as well as (vi) the patients with incomplete clinical data.. Consequently, 771 patients were ultimately enrolled. We divided the patients into bleeding and non-bleeding groups according to the occurrence of PPB. A total of another 198 patients with colorectal polyps were included in the Day Surgery Unit from July 2017 to December 2017 as a validation cohort. The study ow is shown in Figure 1. Complete medical records of the patient-related characteristics, polyp-related characteristics, and polypectomy techniques, as well as the use of prophylactic clips during the endoscopic procedure, were collected. This study was approved by the Ethics Committee of West China Hospital of Sichuan University (IRB number: ChiCTR1800020201).

Endoscopic Colorectal Polypectomy
Written consent was obtained before the operation. If anticoagulant or antiplatelet agents were needed, such as aspirin, warfarin or clopidogrel, they were required to discontinue these at least 5 days before the operation. Endoscopic colorectal polypectomy was performed by electronic endoscopes (JIF-H260Z; Olympus Optical Co, Ltd, Tokyo, Japan) by experienced endoscopists who had performed at least 500 cases of endoscopic polypectomies. We routinely applied argon plasma coagulation (APC) (ERBE Co, Ltd, Germany) and hot biopsy forceps (HBF) (Stericlin Co, Ltd, Germany) for diminutive polyps (d≤5mm) and small polyps (d≤10mm), while larger sessile polyps (10mm d≤15mm) were resected by endoscopic mucosal resection (EMR) and pedunculated polyps were removed by snares (SAS-1-S; COOK Co, Ltd, US). Hemostatic clips were selected when bleeding occurred during the operation or to prevent delayed hemorrhage.

Post-polypectomy Bleeding
In our study, PPB was con rmed with the presence of hematochezia, while melena or hemorrhoids were excluded. We also de ned early PPB (EPPB) as hemorrhage within 24 hours of the colorectal polypectomy and delayed PPB (DPPB) was referred to as hemorrhage during 24 hours to 4 weeks after the endoscopic operation. Follow-up telephone calls within 4 weeks were conducted regularly on the 2nd day, 7th day, 14th day, and 28th day after discharge from the hospital.

Patient-related Factors
We collected the patients demographic characteristics, including gender and age. In addition, the factors of antithrombotic agents, history of smoking and alcohol consumption, postoperative activities and diet were compared between two groups. Smoking was de ned as a continuous or cumulative smoking habit for 6 months or more in one's lifetime. Alcohol consumption referred to drinking 10 grams per day on average. Comorbidities of hypertension, diabetes mellitus, cerebrovascular disease, coronary heart disease, hyperlipidemia, chronic obstructive pulmonary disease (COPD), and rheumatoid diseases were also reviewed. Improper postoperative activities referred to intense exercise or heavy physical activity within the 2 weeks after endoscopic operations. Inappropriate diet was de ned as starting oral feeding within 6 hours or having spicy or greasy food within 1 week after the operation.

Polyp-related Factors
The size, location, gross morphology and the histopathology of the colorectal polyps were carefully documented. An open-biopsy forcep of 6mm was used as standard to measure the polyp size. The polyp location contained ascending colon (cecum was included), transverse colon (hepatic exure and splenic exure were included), descending colon, sigmoid colon, and rectum. The morphology of the polyp was categorized into four types (Yamada , Yamada , Yamada , and Yamada ) according to the criteria of Japanese Yamada Classi cation. Polyps were classi ed histopathologically as adenomatous (tubular, tubulovillous, and villous) or hyperplastic, in ammatory, and others (hamartomatous, retentional, etc.).

Statistical Analysis
All statistical analysis were performed with SPSS software version 24.0 (SPSS Institute, Chicago, IL).
Categorical variables were compared with the Fisher exact test or the χ 2 test. Continuous variables were compared with either the unpaired Student t test or the Mann-Whitney U test. The odds ratios for delayed post-polypectomy hemorrhage were calculated by unconditional logistic regression. In addition, multivariable logistic regression analysis was performed to identify independent variables associated with PPB. The score-based prediction rule was generated from the new logistic regression equations by using a regression coe cient-based scoring method []. The total score for each patient represented the sum of the score for each independent risk factor. The calibration was evaluated with the Hosmer-Lemeshow (H-L) goodness-of-t test. To evaluate the predictive performance of the scoring model, the receiver operating characteristic (ROC) curve and the area under the ROC (AUC) were adopted. An AUC of 1.0 indicated perfect concordance, while an AUC of 0.5 indicated no relationship. Meanwhile, external validation of the model was performed by measuring the discriminatory ability with AUC.

Baseline Characteristics
During the study period, a total of 771 single-polyp patients underwent colorectal polypectomy with 771 polyps being completely removed. The baseline characteristics of the patients are shown in Table 1 Overall, PPB occurred in 26 patients (3.4%) while DPPB developed in 23 patients. PPB appeared to take placed as on day 7 (5, 10) after the polypectomy, of which 15 (57.7%) patients had received endoscopic hemostatic procedures. Only one patient underwent a blood transfusion, and none required selective arterial embolization or surgery. The results showed that gender, age, early postoperative activity, and hypertension were statistically different between the bleeding and non-bleeding groups (P<0.05).

Risk Factors of PPB
As revealed in the univariate analyses, the gender and age of the patients, complication of hypertension, history of alcohol, and early postoperative activity differed signi cantly (P<0.05) between the bleeding and non-bleeding groups. In terms of multivariate logistic regression analysis, gender of male, age older than 60, histopathology of adenomatous polyps, polypectomy technique of HBF, complication of hypertension, use of anticoagulant/antiplatelet agents, and early excessive postoperative activities appeared to be the independent risk factors, which were associated with PPB, whereas location, size, morphology, applicaton of prophylactic clips, other comorbidities, history of smoking or alcohol, as well as inappropriate diet were not statistically signi cant ( with a signi cantly increasing trend of risk from the low to high risk groups. The predictive accuracy of the risk score for PPB was 0.890 (95%CI, 0.806-0.960) measured by AUC (Figure 2). In addition, our prediction model calibrated well with the Hosmer-Lemeshow goodness-of-t test (χ 2 = 1.030, P = 0.794).

External Validation of the Risk Scoring Model
A total of another 198 patients with colorectal polyps were included in the Day Surgery Unit from July 2017 to December 2017 as an external validation cohort. The percentage of the risk for PPB in each calculated score and AUC were 0.924 in the validation dataset ( Figure 2).

Discussion
The current study of a large cohort established a novel, simple-to-use risk-scoring system of colorectal PPB, which comprises various aspects of clinical features. Our nding is one of the largest cohorts to investigate the frequent but serious adverse event of PPB in the Ambulatory Surgery Center. We have dedicated to risk strati cation based on these signi cant clinical risk factors, and established a predictive model that can be applied to the safety of polypectomy in the Ambulatory Surgery Center.
We found that advanced age and hypertension in patients were independent risk factors for PPB. In recent years, although patients with colorectal polyps tended to be younger, PPB continued to occur in the elder patients because of poor vascular compliance [,,]. In addition, the various cardiovascular and cerebrovascular diseases associated with advanced age and hypertension might bring about impaired blood vessel wall and abnormal coagulation function [].
Based on the ndings of previous studies, large size was a signi cant polyp-related factor that has been unequivocally proven to increase the risk of delayed bleeding. Buddingh KT [3] thought that the risk increased by 13% for every 1mm increase in polyp diameter (OR:1.13, 95%CI 1.05-1.20, P<0.001). However, we did not come to a similar conclusion as we believe that a limited colorectal polyp size of 15mm would not play such a vital role. The signi cant associations between colorectal adenomatous polyps and postoperative bleeding in our study, in particular, have been previously corroborated [].
Adenomatous polyps, reported by Uno Y [], had more blood vessels exposed upon removal of the polyp, which was probably related to PPB.
Moreover, we also demonstrated that HBF had an obviously higher risk of PPB compared with EMR or snare excision. In addition, the placement of prophylactic clips had no bene t on reducing the incidence of PPB in the light of our ndings.
According to risk strati cation of endoscopic procedures, endoscopic polypectomy is de ned as high risk procedure. What is more, for patients receiving anticoagulation or antiplatelet therapy, the risk of haemorrhage sharply increases. The challenge is to weigh the bene ts against risks of thromboembolism and PPB. We found that the use of anticoagulant or antiplatelet drugs led to a higher risk of bleeding, which was consistent with several reported studies [,,]. In particular, bridge anticoagulation is necessary in patients with high thromboembolic risks who are undergoing polypectomy [,]. The recent BRIDGE trial, which showed that bridging anticoagulation is associated with a signi cantly higher risk of hemorrhage []. Regrettably, we did not classify the categories of antithrombotic drugs, such as warfarin, aspirin or clopidogrel. In future, we may pay close attention to the potential risk of PPB in patients receiving antithrombotic therapy although these drugs has been strictly discontinued and restarted. We consider that it is necessary to delay discharge or extend the follow-up time to ensure the safety of these special patients who received colorectal polypectomy in ambulatory surgery unit.
We also demonstrated that intense exercise or heavy physical activity within 2 weeks after polypectomy was associated with delayed bleeding. In clinical practice, we emphasize the importance of postpolypectomy recovery at home, especially the guidance of discharge instructions within 7 days. We advocate the e cient and safe hospital clinical pathway of day surgery, meanwhile greater emphasis needs to be placed on standardized discharge criteria and high quality discharge follow-up based on clinical risk assessment.
Although our study provides a reliable risk-scoring model of PPB, it has several limitations. An inherent potential bias was inevitable as the study is retrospective despite parital data being collected prospectively. Furthermore, bridge anticoagulation and the time to restart the antithrombotic therapy should be considered as signi cant variables associated with PPB for a more complete risk-scoring system.

Conclusions
In summary, this was the rst study to clarify the risk factors of PPB in the certain populaton of ambulatory surgery. We aimed to promote the development and growth of high quality ambulatory surgery. To this end, this risk-scoring model has resulted in the founding of new association of ambulatory surgery and endoscopic therapy. The signi cance of this study is that a predictive model was developed, which could provide more valuable clinical information for making a better decision about revising the access criteria and follow-up after discharge. Furthermore, high e ciency, remarkable safety and cost reduction of ambulatory surgery have been improving the access of the general population to utilize of endoscopic treatment with colorectal polyps. " †" : Mann-Whitney U-test was performed; "*" : The difference was statistically significant.