Intensive care medicine

Surgery

Discontinuation

Political science

Internal medicine

MEDLINE

Medicine

Mayo Clinic Proceedings

Medical decisions require balancing expected benefits and harms of alternative management strategies, weighted according to effect on patient outcomes. In practice, however, clinical decisions frequently appear also to be driven by other factors. This issue is illustrated by clinical guidelines and practice patterns for interruption of chronic oral anticoagulation when patients undergo procedures or operations. Each year, more than 250,000 patients in North America undergo procedures or operations for which prescribed anticoagulation is interrupted.1Douketis J.D. Spyropoulos A.C. Spencer F.A. et al.Perioperative management of antithrombotic therapy: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines.Chest. 2012; 141: e326S-e350SAbstract Full Text Full Text PDF PubMed Scopus (1280) Google Scholar Deciding whether to interrupt anticoagulation requires weighing the risk and resultant consequences of experiencing a thromboembolic event during therapy interruption against those of bleeding from the procedure without anticoagulation interruption. Clinical practice guidelines exist to assist with these decisions.2Acosta R.D. Abraham N.S. Chandrasekhara V. et al.The management of antithrombotic agents for patients undergoing GI endoscopy.Gastrointest Endosc. 2016; 83: 3-16Abstract Full Text Full Text PDF PubMed Scopus (375) Google Scholar However, for many procedures, assumptions about the dangers of periprocedural bleeding often lead to a recommendation for anticoagulation interruption, despite limited evidence of the incremental risks of bleeding while continuing anticoagulation therapy. Colonoscopy is one of the most common reasons for anticoagulation interruption. The American Society for Gastrointestinal Endoscopy (ASGE) 2016 guidelines for anticoagulation interruption are based on the estimated relative risk of experiencing a bleeding or embolic complication.2Acosta R.D. Abraham N.S. Chandrasekhara V. et al.The management of antithrombotic agents for patients undergoing GI endoscopy.Gastrointest Endosc. 2016; 83: 3-16Abstract Full Text Full Text PDF PubMed Scopus (375) Google Scholar These guidelines classify diagnostic colonoscopy without biopsy as low risk for bleeding and, thus, do not recommend anticoagulation interruption. In contrast, colonoscopy with polypectomy is classified as high risk, and anticoagulation interruption is recommended (5 days for warfarin and a varying interval for the novel oral anticoagulant drugs—dabigatran, rivaroxaban, apixaban, and edoxaban—adjusted for renal function). Many guidelines seek to balance number needed to treat (NNT) with number needed to harm (NNH). Based on published studies, the ASGE guidelines estimate a 1% absolute risk of an embolic event (primarily stroke) in patients with warfarin interruption for 4 to 7 days.3Garcia D.A. Regan S. Henault L.E. et al.Risk of thromboembolism with short-term interruption of warfarin therapy.Arch Intern Med. 2008; 168: 63-69Crossref PubMed Scopus (285) Google Scholar, 4Blacker D.J. Wijdicks E.F.M. McClelland R.L. Stroke risk in anticoagulated patients with atrial fibrillation undergoing endoscopy.Neurology. 2003; 61: 964-968Crossref PubMed Scopus (146) Google Scholar This surprisingly high number has been consistently confirmed in other studies.5Wahl M.J. Pinto A. Kilham J. Lalla R.V. Dental surgery in anticoagulated patients: stop the interruption.Oral Surg Oral Med Oral Pathol Oral Radiol. 2015; 119: 136-157Abstract Full Text Full Text PDF PubMed Scopus (55) Google Scholar, 6Wysokinski W.E. McBane R.D. Daniels P.R. et al.Periprocedural anticoagulation management of patients with nonvalvular atrial fibrillation.Mayo Clin Proc. 2008; 83: 639-645Abstract Full Text Full Text PDF PubMed Scopus (73) Google Scholar Although data are limited, there seems to be an equal thromboembolic risk with temporary interruption of novel oral anticoagulant drugs for procedures.7Healey J.S. Eikelboom J. Douketis J. et al.Periprocedural bleeding and thromboembolic events with dabigatran compared with warfarin: results from the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) randomized trial.Circulation. 2012; 126: 343-348Crossref PubMed Scopus (467) Google Scholar In comparison, the references provided in the ASGE guideline report only an average 0.6% (range, 0%-1.4%; references 61, 101, 103, and 105) rate of significant bleeding (generally defined as requiring transfusion or hospitalization) with polypectomy and uninterrupted anticoagulation (Table).8Friedland S. Soetikno R. Colonoscopy with polypectomy in anticoagulated patients.Gastrointest Endosc. 2006; 64: 98-100Abstract Full Text Full Text PDF PubMed Scopus (65) Google Scholar, 9Howell DA, Eswaran SL, Loew BJ, et al. Use of hemostatic clips in patients undergoing colonoscopy in the setting of coumadin anticoagulation therapy cecal intubation rate cecal intubation time the cost effectiveness of colonic stenting as a bridge to curative surgery in patients with acute lef. 2006;63(5):2006.Google Scholar, 10Friedland S. Sedehi D. Soetikno R. Colonoscopic polypectomy in anticoagulated patients.World J Gastroenterol. 2009; 15: 1973-1976Crossref PubMed Scopus (48) Google Scholar, 11Horiuchi A. Nakayama Y. Kajiyama M. Tanaka N. Sano K. Graham D.Y. Removal of small colorectal polyps in anticoagulated patients: a prospective randomized comparison of cold snare and conventional polypectomy.Gastrointest Endosc. 2014; 79: 417-423Abstract Full Text Full Text PDF PubMed Scopus (202) Google Scholar Thus, the NNT and NNH seem to favor uninterrupted anticoagulation.TableRisks of Periprocedural Bleeding or Major Emboli/Stroke From References in ASGE Guidelines.aAfib = atrial fibrillation; INR = international normalized ratio; MV = mechanical valve; NA = not applicable; NR = not reported; TE = thromboembolism.Reference, yearbFrom Acosta et al.2Patients (No.)Anticoagulation indicationDuration of follow-up (d)INR at procedureBleeding risk with uninterrupted anticoagulation (%)Major embolic/stroke risk with interruption of anticoagulation (%)Bleeding riskFriedland et al,8Friedland S. Soetikno R. Colonoscopy with polypectomy in anticoagulated patients.Gastrointest Endosc. 2006; 64: 98-100Abstract Full Text Full Text PDF PubMed Scopus (65) Google Scholar 200621NR21-562.30.0Howell et al,9Howell DA, Eswaran SL, Loew BJ, et al. Use of hemostatic clips in patients undergoing colonoscopy in the setting of coumadin anticoagulation therapy cecal intubation rate cecal intubation time the cost effectiveness of colonic stenting as a bridge to curative surgery in patients with acute lef. 2006;63(5):2006.Google Scholar 200671NR301.51.4Friedland et al,10Friedland S. Sedehi D. Soetikno R. Colonoscopic polypectomy in anticoagulated patients.World J Gastroenterol. 2009; 15: 1973-1976Crossref PubMed Scopus (48) Google Scholar 2009123Afib 65%21-56NA0.8TE 16%Valve 9%Other 13%Horiuchi et al,11Horiuchi A. Nakayama Y. Kajiyama M. Tanaka N. Sano K. Graham D.Y. Removal of small colorectal polyps in anticoagulated patients: a prospective randomized comparison of cold snare and conventional polypectomy.Gastrointest Endosc. 2014; 79: 417-423Abstract Full Text Full Text PDF PubMed Scopus (202) Google Scholar 201435Afib 74%142.40TE 20%Other 6%Average0.6Stroke riskGarcia et al,3Garcia D.A. Regan S. Henault L.E. et al.Risk of thromboembolism with short-term interruption of warfarin therapy.Arch Intern Med. 2008; 168: 63-69Crossref PubMed Scopus (285) Google Scholar 20081024Afib 54%30NR0.7TE 14%MV 13%Blacker et al,4Blacker D.J. Wijdicks E.F.M. McClelland R.L. Stroke risk in anticoagulated patients with atrial fibrillation undergoing endoscopy.Neurology. 2003; 61: 964-968Crossref PubMed Scopus (146) Google Scholar 20031137Afib301.31.1Average0.9a Afib = atrial fibrillation; INR = international normalized ratio; MV = mechanical valve; NA = not applicable; NR = not reported; TE = thromboembolism.b From Acosta et al.2Acosta R.D. Abraham N.S. Chandrasekhara V. et al.The management of antithrombotic agents for patients undergoing GI endoscopy.Gastrointest Endosc. 2016; 83: 3-16Abstract Full Text Full Text PDF PubMed Scopus (375) Google Scholar Open table in a new tab However, such a simple numerical comparison of possible complications with and without anticoagulation interruption is not sufficient to guide clinical decision making. The outcomes of these potential adverse sequelae often are of markedly different magnitude. This is especially important because strokes associated with atrial fibrillation (the most common indication for anticoagulation use) frequently are large, devastating events.12Lin H.J. Wolf P.A. Kelly-Hayes M. et al.Stroke severity in atrial fibrillation: the Framingham Study.Stroke. 1996; 27: 1760-1764Crossref PubMed Scopus (1110) Google Scholar Although there is variability, patients also generally express a clear preference to avoid stroke even at the cost of significant bleeding—assessing the outcomes of stroke to be much more severe than those of a significant gastrointestinal bleed.13Alonso-Coello P. Montori V.M. Díaz M.G. et al.Values and preferences for oral antithrombotic therapy in patients with atrial fibrillation: physician and patient perspectives.Health Expect. 2015; 18: 2318-2327Crossref PubMed Scopus (45) Google Scholar However, the ASGE guidelines implicitly weight procedural bleeding as the greater harm. Thus, despite emphasis on a patient-centric approach to medicine, with patient preferences helping guide decisions when no clearly superior medical approach exists, current guidelines for anticoagulation discontinuation for colonoscopy not only do not consider but actually conflict with patient preferences. In clinical practice, anticoagulation discontinuation deviates ever further from optimal care. In an effort to avoid the potential of having to reschedule a repeated procedure, many practitioners in our communities discontinue anticoagulation for all colonoscopies because of the possibility of having to perform a polypectomy. Because only approximately one-third of all screening and diagnostic colonoscopy patients undergo a polypectomy, only approximately 0.2% (0.6% ÷ 3) of such colonoscopy patients interrupting anticoagulation therapy potentially benefit via a reduced bleeding risk compared with a 1% chance of experiencing a stroke or other serious thromboembolic event with anticoagulation interruption. We acknowledge that bleeding risks of polypectomy higher than those mentioned previously herein have been reported and likely are related to the anatomical complexity of the polyps. Yet the low bleeding risks noted previously herein, cited in the endoscopy guidelines, show that many polyps, especially most that are smaller than 10 mm, can be safely removed without anticoagulation interruption.11Horiuchi A. Nakayama Y. Kajiyama M. Tanaka N. Sano K. Graham D.Y. Removal of small colorectal polyps in anticoagulated patients: a prospective randomized comparison of cold snare and conventional polypectomy.Gastrointest Endosc. 2014; 79: 417-423Abstract Full Text Full Text PDF PubMed Scopus (202) Google Scholar This fact is rarely discussed when the risks and benefits of anticoagulation interruption for colonoscopy are considered. Delayed bleeding occurs more often in patients receiving oral anticoagulation after polypectomy even when anticoagulation was interrupted for the procedure.14Witt D.M. Delate T. Mccool K.H. et al.Incidence and predictors of bleeding or thrombosis after polypectomy in patients receiving and not receiving anticoagulation therapy.J Thromb Haemost. 2009; 7: 1982-1989Crossref PubMed Scopus (60) Google Scholar But this bleeding is generally not associated with significant morbidity or mortality, and it typically occurs at a time when anticoagulation would have been resumed regardless of an interruption strategy. The current ASGE guidelines generally are evidence based and clinically nuanced (eg, including recommendations for heparinoid “bridging” in patients at high thromboembolic risk). However, evidence for bridging is weak, with most studies showing no benefit and often increased periprocedural bleeding risk. And bridging, if it does not decrease periprocedural thromboembolic risk, will likewise offer no benefit regarding the delayed bleeding risk after polypectomy at 7 to 10 days because almost uniformly full oral anticoagulation would have been resumed by this time. Avoiding any anticoagulation until this period has passed would be problematic given the stroke risks outlined previously herein for even shorter interruptions. But we think that the more appropriate question is whether anticoagulation should be interrupted rather than whether bridging might reduce the inherent thromboembolic risk of interruption. Thus, patient safety may be best served by not interrupting anticoagulation for colonoscopy procedures but rather scheduling a second procedure with a brief interruption of anticoagulation for patients discovered to require very high-risk polypectomy, such as size greater than 10 mm or broad base. We developed a decision flow chart balancing the risks and benefits of anticoagulation interruption to assist in this process (Figure). Examination of other procedures may produce similar assessments of the risks vs benefits of anticoagulation interruption. Pacemaker implantation and atrial fibrillation ablation have been performed safely without anticoagulation interruption,15Birnie D.H. Healey J.S. Wells G.A. et al.Pacemaker or defibrillator surgery without interruption of anticoagulation.N Engl J Med. 2013; 368: 2084-2093Crossref PubMed Scopus (417) Google Scholar, 16Di Biase L. Burkhardt J.D. Santangeli P. et al.Periprocedural stroke and bleeding complications in patients undergoing catheter ablation of atrial fibrillation with different anticoagulation management: results from the Role of Coumadin in Preventing Thromboembolism in Atrial Fibrillation (AF) Patients Undergoing Catheter Ablation (COMPARE) randomized trial.Circulation. 2014; 129: 2638-2644Crossref PubMed Scopus (378) Google Scholar for example, with no increased bleeding compared with usual care. Why do physicians continue to make these decisions both in clinical guidelines and in clinical practice that seem to be disadvantageous to patients on more thorough analysis? In part, physician behavior is a reflection of common decision-making heuristics and biases, including Prospect Theory, especially loss aversion, whereby individuals respond more strongly to potential losses than potential gains.17Kahneman D. Thinking, Fast and Slow. Vol 1. Doubleday Canada, Toronto2011Google Scholar Procedure-associated bleeding is immediate and its cause obvious, and the loss to the patient and physician of bleeding (including, possibly, to physician reputation) may be more compelling and salient than the gain of continued protection against stroke, a complication that usually occurs outside of the immediate time frame of the procedure and may not even become known to the physician performing the procedure. Thus, anticoagulation recommendations may demonstrate the unintended bias of physicians writing guidelines and in practice to minimize immediate procedural complications, which we refer to as procedural myopia. How best to assist our patients with these decisions? Given patient preference variability, we believe that these decisions require an individualized conversation regarding a patient's perceptions of the probability and impact of potential benefits, harms, and inconveniences. So although it is not certain that a patient-centric approach will always result in stroke avoidance, the path forward needs to be chosen after shared decision making with a patient who has been well informed about the potential consequences of each course of action. Such discussions often do not take place today. In summary, the evidence base does not support current clinical practice guidelines for the common question of whether to interrupt anticoagulation for colonoscopy and probably other low-risk procedures. In addition, clinical practice guidelines often do not adequately incorporate patient preferences concerning alternative potential outcomes. Given that stroke seems to be a much more serious outcome than local bleeding complications, interruptions of anticoagulation recommended by clinical practice guidelines may be inappropriate. Thus, we recommend (1) careful evidence-based review of the likelihood and clinical impact of possible adverse effects of continuing or interrupting anticoagulation, which should precede any interruption decision; (2) recognition of the unintended bias of physicians writing guidelines to minimize immediate procedural complications, which must be addressed; and (3) shared decision making with patients that includes informed individualized weighting and discussion of possible benefits and harms of medical actions.

Should Procedures or Patients Be Safe? Bias in Recommendations for Periprocedural Discontinuation of Anticoagulation