Risk factors for severe delayed postpolypectomy bleeding

• Introduction
• Methods
• Data analysis
• Results
• Description and course of patients with postpolypectomy bleeding
• Risk factor assessment
• Discussion
• Financial support
• References

Background and study aims: Postpolypectomy bleeding is a rare but serious adverse event. The aim of this study was to identify factors associated with the risk of severe delayed postpolypectomy bleeding.

Patients and methods: This was a case-control study, comparing cases who developed hematochezia and required medical evaluation 6 hours to 14 days after colonoscopic polypectomy, and control patients who underwent polypectomy without delayed bleeding, and who were selected in approximately a 3 : 1 ratio. The following risk factors were specified a priori: resuming anticoagulation (within 1 week following polypectomy), aspirin use, hypertension, and polyp diameter.

Results: Of the 4592 patients who underwent colonoscopy with polypectomy, 41 patients (0.9 %) developed delayed postpolypectomy bleeding (cases), and 132 patients were selected as controls. The mean age was 64.3 years for cases and 65.4 years for controls. Cases presented on average 6 days after polypectomy (range 1 - 14 days), and 48 % required blood transfusion (average 4.2 units, range 0 - 17). Two patients required surgery. Anticoagulation was resumed following polypectomy in 34 % of cases compared with 9 % of controls (OR 5.2; 95 % CI 2.2 - 12.5; P < 0.001). For every 1 mm increase in polyp diameter, the risk of hemorrhage increased by 9 % (OR 1.09; 95 % CI 1.0 - 1.2; P = 0.008). Hypertension (OR 1.1) and aspirin use (OR 1.1) did not increase the risk of postpolypectomy bleeding. In exploratory analysis, diabetes (OR 2.5) and coronary artery disease (OR 3.0) were associated with postpolypectomy hemorrhage, but the association was no longer statistically significant once adjusted for the use of anticoagulation.

Conclusions: Resuming anticoagulation following polypectomy and polyp diameter were strongly associated with increased risk of severe delayed postpolypectomy bleeding.

Introduction

Colonoscopy with polypectomy is highly effective in reducing colon cancer incidence and mortality, and has become one of the most common medical procedures carried out in the USA [1] [2]. Although its benefits are undisputed, colonoscopy with polypectomy is associated with a small but finite rate of serious complications [3]. One of these complications is delayed postpolypectomy bleeding. Factors that increase the risk of immediate postpolypectomy bleeding have been well studied; however, available data leave many unanswered questions regarding factors that increase the risk of delayed postpolypectomy bleeding [4] [5] [6] [7] [8] [9] [10]. Does resuming anticoagulantion therapy after polypectomy increase the risk of delayed hemorrhage? This is an issue that has become increasingly important with the widespread use of these drugs. Although most experts agree that anticoagulation should be reversed prior to polypectomy, few data exist on resuming anticoagulation after polypectomy. Is hypertension a risk factor for postpolypectomy bleeding? A recently published study showed that it was but this was an unexpected finding (post hoc analysis), and therefore requires confirmation [11]. Our aim was to conduct a case-control study to determine risk factors associated with delayed postpolypectomy bleeding.

Methods

All patients who underwent colonoscopy with polypectomy at the Minneapolis Veterans Affairs Medical Center from 1999 to 2006 were identified. ‘Cases’ were defined as those patients who developed hematochezia between 6 hours and 14 days of polypectomy and any one of the following: visit to the emergency room, admission to the hospital, blood transfusion, repeat colonoscopy, or decrease in hemoglobin ≥ 1 gm/dL from baseline. ‘Controls’ were defined as patients who underwent colonoscopy with polypectomy but did not develop hematochezia within 14 days of the procedure. Controls were identified from the same cohort as the cases, using a string of computer-generated random numbers. Controls were selected in a 3 : 1 ratio compared with cases. In anticipation that complete data may not be available for all controls, we over-sampled controls by nine patients. Data on all study patients were obtained from endoscopy reports, electronic medical records, pharmacy records, and laboratory results, using an abstraction instrument that was created for the study. The study was approved by the institutional review board.

Data analysis

Unconditional logistic regression was used to analyze the association between potential risk factors and postpolypectomy bleeding (dependent variable). We identified 41 cases of postpolypectomy bleeding. In order to avoid over-fitting our regression model by entering more than one independent variable for every eight ‘events’ in our dataset, we limited the a priori risk factor analysis to four potential risk factors [12]: anticoagulation (heparin or warfarin use within 1 week after a polypectomy); aspirin use (at least one dose of aspirin within 1 week prior and within 1 week after polypectomy); hypertension (systolic pressure > 140 mm Hg or diastolic > 90 mm Hg or use of anti-hypertensive medication); and polyp diameter. In post hoc analysis, the association between postpolypectomy bleeding and hypertension (modeled as a continuous variable), diabetes, chronic obstructive lung disease, coronary artery disease, and polyp type (sessile or pendunculated) was also computed.

Results

During the study period, 4592 patients underwent colonoscopy with polypectomy at the Minneapolis Veterans Affairs Medical Center. Of these, 41 patients (0.9 %, 95 % confidence interval [CI], 0.6 - 1.2) met study criteria for delayed postpolypectomy bleeding (cases), and 132 patients were selected, as described above, to serve as controls. Baseline characteristics of cases and controls were similar ([Table 1]). Of note, almost all study subjects were Caucasian and male.

Description and course of patients with postpolypectomy bleeding

Patients with postpolypectomy bleeding presented on average 6 days after polypectomy (range 1 - 14 days). All patients were admitted to the hospital or observed for at least 12 hours before discharge. The average length of in-hospital stay was 4.3 days (range 1 - 32). Blood was transfused in 48 % of patients (average 4.2 units, range 0 - 17). All but two patients (95 %) underwent a repeat colonoscopy. A polypectomy site was identified as the source of bleeding in 27 of 41 patients. Of these 27 patients, active bleeding from the polypectomy site was noted in five, an adherent clot in 11, ulcer with pigmented protuberance in two, ulcer with pigmented base in two, and a clean-based ulcer in six patients. In the remaining patients, either the polypectomy site could not be identified (n = 2), or the patient had had more than one polyp removed and none of the polypectomy sites showed obvious stigmata of recent hemorrhage. At colonoscopy, one patient was treated with thermal therapy, two with epinephrine injection alone, and 18 with epinephrine injection followed by placement of endoscopic clips. Four patients developed clinical signs of re-bleeding after repeat colonoscopy. All underwent another colonoscopy with endoscopic therapy. Two patients stopped bleeding, one patient underwent surgery to control bleeding, and another patient underwent surgery to repair a colonoscopy-related perforation.

Risk factor assessment

[Table 2] shows the measures of association between various factors and the risk of postpolypectomy bleeding. Anticoagulation with warfarin and/or heparin was resumed within 1 week of polypectomy in 34 % of cases compared with 9 % of controls (OR 5.2; 95 % CI 2.2 - 12.5; P < 0.001). Polyp diameter was also associated with increased risk of postpolypectomy hemorrhage. For every 1 mm increase in polyp diameter, the risk of hemorrhage increased by 9 % (OR 1.09; 95 % CI 1.0 - 1.2; P = 0.008). Hypertension (OR 1.1) or the use of aspirin (OR 1.1) were not found to be significant predictors of postpolypectomy bleeding.

Polyps were stratified by diameter (≥ 10 mm and < 10 mm), and the association between anticoagulation and risk of postpolypectomy bleeding was measured within each stratum. For polyps ≥ 10 mm, anticoagulation increased the risk of postpolypectomy bleeding by 1.5 times. For polyps < 10 mm, anticoagulation increased the risk of postpolypectomy bleeding by 6.75 times. However, a Breslow-Day test of homogeneity failed to show a statistically significant difference between these odds ratios (P = 0.25). The association between anticoagulation and risk of bleeding also did not differ between those using aspirin compared with those not on aspirin (Breslow-Day test of homogeneity P = 0.4).

Patients with postpolypectomy bleeding who received anticoagulation are detailed in [Table 3]. Of note, with the exception of one patient, INR at the time of bleeding was within or less than the therapeutic range for anticoagulation. Further, none of these patients received loading dosages of unfractionated heparin or warfarin.

A comparison was made between patients who received anticoagulation following polypectomy with those who were not. As expected, patients who received anticoagulants presented with a higher INR at the time of bleeding than those who did not receive anticoagulation therapy (1.95 vs. 1.05; P = 0.007). However, there was no significant difference between these two groups with regard to time of presentation (6.8 days following polypectomy vs. 5.8 days following polpectomy; P = 0.4) or the number of blood units transfused (1.8 vs. 2.2; P = 0.7).

In exploratory analysis, diabetes (OR 2.5), coronary artery disease (OR 3.0), and lung disease (OR 2.2) were found to be associated with postpolypectomy hemorrhage ([Table 2]). The association was attenuated once adjusted for the use of anticoagulation (diabetes OR 2.0, P = 0.09; coronary artery disease OR 2.1, P = 0.06; and lung disease OR 1.9, P = 0.1). An association between hypertension modeled as a continuous variable and the risk of postpolypectomy bleeding was also sought. The average of the three most recent blood pressure values within 1 year prior to polypectomy (OR 1.0; P = 0.3), or the mean arterial pressure (MAP) on the day of polypectomy (OR 1.0; P = 0.5) were not associated with the risk of postpolypectomy hemorrhage ([Table 2]).

Discussion

Resuming anticoagulation within 1 week of polypectomy and increasing polyp diameter significantly increased the risk for delayed postpolypectomy bleeding, whereas hypertension and the use of aspirin did not.

It is our practice to reverse anticoagulation prior to performing a colonoscopic polypectomy. As noted in [Table 1], the average INR at the time of polypectomy was 1.07 for cases and 1.04 for controls. The INR at the time of polypectomy for anticoagulated patients who went on to develop postpolypectomy bleeding was less than 1.3 (upper limit of normal at our institution) in all cases ([Table 3]). Therefore, it is unlikely that pre-procedure anticoagulation influenced the risk of delayed postpolypectomy bleeding in our study.

Our practice regarding resumption of anticoagulation following a polypectomy is to re-start warfarin on the night of the procedure, and heparin within 6 hours, as recommended by expert consensus [13]. As shown in [Table 3], anticoagulation was resumed within 24 hours in 10 of 14 patients, and within 72 hours in the remaining four patients. The delay in resuming anticoagulation beyond 24 hours was individualized, based on the endoscopist’s assessment of the patient’s risk of bleeding and thromboembolus. Our data suggest that resuming anticoagulation within 72 hours after a polypectomy was associated with a five-fold increased risk of postpolypectomy bleeding.

Although postpolypectomy bleeding remains a rare event, about half the patients in our study required blood transfusions, and two of 41 patients eventually underwent colonic surgery. As almost all reported cases of postpolypectomy bleeding have occurred within 2 weeks of the procedure, we speculate whether withholding anticoagulation for patients at high risk of postpolypectomy bleeding for up to 2 weeks may be advisable [4] [14] [15].

For most disease states for which anticoagulation is recommended, the absolute daily risk of thromboembolic events without anticoagulation is low. For example, patients with atrial fibrillation with moderate risk factors for stroke have an estimated daily risk of stroke without anticoagulation to be 0.01 %, which is substantially lower than the daily risk of postpolypectomy bleeding [16]. However, as a stroke may have a profoundly greater impact on morbidity and mortality, a formal decision analysis is needed in order to determine the optimum approach. Until such information is available, the decision to withhold anticoagulation for longer intervals will need to be individualized, balancing risk factors for bleeding with the risk factors for thromboembolic events.

In a pre-specified analysis, we identified polyp diameter as the only other factor associated with an increased risk of bleeding. Other investigators have found age [5], hypertension [11], polyp location [5] [10], and sessile polyps [5] [10] to be significant risk factors.

In our study, as in others, there was no incremental increase in the risk of bleeding with aspirin use, either alone, or in combination with anticoagulants [8] [9]. Our study was not sufficiently powered to determine if anticoagulation with a combination of warfarin and heparin increased the risk of bleeding when compared with warfarin alone. More studies are needed to better define risk factors for postpolypectomy bleeding, and to test the hypothesis that delaying anticoagulation therapy for 2 weeks will further reduce the bleeding risk.

In a recently published article, Watabe et al. found that patients with hypertension were at a 5.6-fold (95 % CI, 1.8 - 17.2) increased risk of developing delayed postpolypectomy bleeding [11]. We used the same definition of hypertension as these authors, but failed to validate hypertension as a significant risk factor for postpolypectomy bleeding. In post hoc analysis, we modeled hypertension as a continuous variable, first using a mean of the last three blood pressure measures within the year prior to polypectomy, and then using the blood pressure on the day of polypectomy. Again, we failed to find an association between hypertension and postpolypectomy bleeding.

Most study subjects in both studies were male, and the average age was approximately 65 years. In the study by Watabe et al. the majority of polypectomies were carried out using a hot biopsy forceps, and only one of 38 patients required a blood transfusion. Also, hypertension was not a pre-specified risk factor in the analysis. In our study, all polypectomies were carried out using a snare, and 48 % of patients required a blood transfusion. As we were aware of the results of the Watabe study, we specified hypertension as a potential risk factor prior to collecting or analyzing the data. Further, we restricted the number of independent variables in our logistic regression model to four, as we had only 41 ‘events’ in our dataset. In post hoc univariate analysis, we found that diabetes, coronary artery disease, and lung disease were statistically significant risk factors for postpolypectomy bleeding. However, in multivariate analysis, when these were adjusted for the use of anticoagulation, the measures of association were attenuated and no longer statistically significant at the traditional alpha of 0.05. We speculate that like diabetes, coronary artery disease, and lung disease, hypertension may be a marker for patients who require anticoagulation and who perhaps have general poor health, and not an independent risk factor for postpolypectomy bleeding. Although blood pressure control has undisputed health benefits, better blood pressure control may not lower the risk of postpolypectomy bleeding.

Our study had several limitations. First, some patients included in the control group may have presented to other hospitals with postpolypectomy bleeding that was not recorded in their medical charts. This would result in misclassification, and a decrease in the power of our study. Second, most subjects in our study were male and Caucasian, reflecting the patient population seen at our institution. Thus, the results of our study are directly applicable only to this group. Third, the sample size for subgroup analysis was small, and the resulting estimates were therefore imprecise.

In summary, we found that anticoagulation following polypectomy and increasing polyp size were strong and independent predictors of postpolypectomy bleeding. Preventative strategies incorporating these risk factors should be developed to lower this risk.

Financial support

M. S. Sawhney is supported, in part, by VA Clinical Science R&D Service (Grant no. 04S-CRCOE-001).

Competing interests: None

References

• 1 Winawer S J, Zauber A G, O’Brien M J. et al . Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps.  N Engl J Med. 1993;  328 901-906
  CrossrefPubMedSearch in Google Scholar
• 2 Winawer S J, Zauber A G, Ho M N. et al . Prevention of colorectal cancer by colonoscopic polypectomy.  N Engl J Med. 1993;  329 1977-1981
  CrossrefPubMedSearch in Google Scholar
• 3 Nelson D B, McQuaid K R, Bond J H. et al . Procedural success and complications of large-scale screening colonoscopy.  Gastrointest Endosc. 2002;  55 307-314
  CrossrefPubMedSearch in Google Scholar
• 4 Rex D K, Lewis B S, Waye J D. Colonoscopy and endoscopic therapy for delayed post-polypectomy hemorrhage.  Gastrointest Endosc. 1992;  38 127-129
  CrossrefPubMedSearch in Google Scholar
• 5 Rosen L, Bub D S, Reed J F, Nastasee S A. Hemorrhage following colonoscopic polypectomy.  Dis Colon Rectum. 1993;  36 1126-1131
  CrossrefPubMedSearch in Google Scholar
• 6 Gibbs D H, Opelka F G, Beck D E. et al . Postpolypectomy colonic hemorrhage.  Dis Colon Rectum. 1996;  39 806-810
  CrossrefPubMedSearch in Google Scholar
• 7 Sorbi D, Norton I, Conio M. et al . Postpolypectomy lower GI bleeding: descriptive analysis.  Gastrointest Endosc. 2000;  51 690-696
  CrossrefPubMedSearch in Google Scholar
• 8 Hui A, Wong R MY, Ching J YL. et al . Risk of colonoscopic polypectomy bleeding with anticoagulants and antiplatelet agents: analysis of 1657 cases.  Gastrointest Endosc. 2004;  59 44-48
  CrossrefPubMedSearch in Google Scholar
• 9 Yousfi M, Gostout C, Baron T. et al . Postpolypectomy lower gastrointestinal bleeding: potential role of aspirin.  Am J Gastroenterol. 2004;  99 1785-1789
  CrossrefPubMedSearch in Google Scholar
• 10 Heldwein W, Rosch T, Meining A. et al . The Munich Polypectomy Study (MUPS): prospective analysis of complications and risk factors in 4000 colonic snare polypectomies.  Endoscopy. 2005;  37 1116-1122
  Thieme ConnectPubMedSearch in Google Scholar
• 11 Watabe H, Yamaji Y, Okamoto M. et al . Risk assessment for delayed hemorrhagic complication of colonic polypectomy: polyp-related factors and patient-related factors.  Gastrointest Endosc. 2006;  64 73-78
  CrossrefPubMedSearch in Google Scholar
• 12 Cepeda M, Boston R, Farrar J, Strom B. Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders.  Am J Epidemiol. 2003;  158 280-287
  CrossrefPubMedSearch in Google Scholar
• 13 Eisen G M, Baron T H, Dominitz J A. et al . Guideline on the management of anticoagulation and antiplatelet therapy for endoscopic procedures.  Gastrointest Endosc. 2002;  55 775-779
  CrossrefPubMedSearch in Google Scholar
• 14 Garbay J R, Suc B, Rotman N. et al . Multicentre study of surgical complications of colonoscopy.  Br J Surg. 1996;  83 42-44
  CrossrefPubMedSearch in Google Scholar
• 15 Waye J D, Lewis B S, Yessayan S. Colonoscopy: a prospective report of complications.  J Clin Gastroenterol. 1992;  15 347-351
  CrossrefPubMedSearch in Google Scholar
• 16 Gage B, Waterman A, Shannon W. et al . Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation.  JAMA. 2001;  285 2864-2870
  CrossrefPubMedSearch in Google Scholar

M. S. Sawhney, MD

Section of Gastroenterology
Beth Israel Deaconess Medical Center
RABB-ROSE 101

330 Brookline Avenue
Boston, MA 02215
USA

Fax: +1-617-667-1171

Email: msawhney@bidmc.harvard.edu

References

• 1 Winawer S J, Zauber A G, O’Brien M J. et al . Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps.  N Engl J Med. 1993;  328 901-906
  CrossrefPubMedSearch in Google Scholar
• 2 Winawer S J, Zauber A G, Ho M N. et al . Prevention of colorectal cancer by colonoscopic polypectomy.  N Engl J Med. 1993;  329 1977-1981
  CrossrefPubMedSearch in Google Scholar
• 3 Nelson D B, McQuaid K R, Bond J H. et al . Procedural success and complications of large-scale screening colonoscopy.  Gastrointest Endosc. 2002;  55 307-314
  CrossrefPubMedSearch in Google Scholar
• 4 Rex D K, Lewis B S, Waye J D. Colonoscopy and endoscopic therapy for delayed post-polypectomy hemorrhage.  Gastrointest Endosc. 1992;  38 127-129
  CrossrefPubMedSearch in Google Scholar
• 5 Rosen L, Bub D S, Reed J F, Nastasee S A. Hemorrhage following colonoscopic polypectomy.  Dis Colon Rectum. 1993;  36 1126-1131
  CrossrefPubMedSearch in Google Scholar
• 6 Gibbs D H, Opelka F G, Beck D E. et al . Postpolypectomy colonic hemorrhage.  Dis Colon Rectum. 1996;  39 806-810
  CrossrefPubMedSearch in Google Scholar
• 7 Sorbi D, Norton I, Conio M. et al . Postpolypectomy lower GI bleeding: descriptive analysis.  Gastrointest Endosc. 2000;  51 690-696
  CrossrefPubMedSearch in Google Scholar
• 8 Hui A, Wong R MY, Ching J YL. et al . Risk of colonoscopic polypectomy bleeding with anticoagulants and antiplatelet agents: analysis of 1657 cases.  Gastrointest Endosc. 2004;  59 44-48
  CrossrefPubMedSearch in Google Scholar
• 9 Yousfi M, Gostout C, Baron T. et al . Postpolypectomy lower gastrointestinal bleeding: potential role of aspirin.  Am J Gastroenterol. 2004;  99 1785-1789
  CrossrefPubMedSearch in Google Scholar
• 10 Heldwein W, Rosch T, Meining A. et al . The Munich Polypectomy Study (MUPS): prospective analysis of complications and risk factors in 4000 colonic snare polypectomies.  Endoscopy. 2005;  37 1116-1122
  Thieme ConnectPubMedSearch in Google Scholar
• 11 Watabe H, Yamaji Y, Okamoto M. et al . Risk assessment for delayed hemorrhagic complication of colonic polypectomy: polyp-related factors and patient-related factors.  Gastrointest Endosc. 2006;  64 73-78
  CrossrefPubMedSearch in Google Scholar
• 12 Cepeda M, Boston R, Farrar J, Strom B. Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders.  Am J Epidemiol. 2003;  158 280-287
  CrossrefPubMedSearch in Google Scholar
• 13 Eisen G M, Baron T H, Dominitz J A. et al . Guideline on the management of anticoagulation and antiplatelet therapy for endoscopic procedures.  Gastrointest Endosc. 2002;  55 775-779
  CrossrefPubMedSearch in Google Scholar
• 14 Garbay J R, Suc B, Rotman N. et al . Multicentre study of surgical complications of colonoscopy.  Br J Surg. 1996;  83 42-44
  CrossrefPubMedSearch in Google Scholar
• 15 Waye J D, Lewis B S, Yessayan S. Colonoscopy: a prospective report of complications.  J Clin Gastroenterol. 1992;  15 347-351
  CrossrefPubMedSearch in Google Scholar
• 16 Gage B, Waterman A, Shannon W. et al . Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation.  JAMA. 2001;  285 2864-2870
  CrossrefPubMedSearch in Google Scholar

M. S. Sawhney, MD

Section of Gastroenterology
Beth Israel Deaconess Medical Center
RABB-ROSE 101

330 Brookline Avenue
Boston, MA 02215
USA

Fax: +1-617-667-1171

Email: msawhney@bidmc.harvard.edu