Risk of Bleeding Following Percutaneous Endoscopic Gastrostomy (PEG): A Five Year Retrospective Study

Abstract

Percutaneous Endoscopic Gastrostomy (PEG), the primary procedure for the long term nutrition of patients with swallowing disorders, is considered a high risk endoscopic procedure for bleeding complications. This retrospective review aims to determine predictive factors of bleeding complications associated with PEG, including the risk of anticoagulation and anti-platelet therapy. Methods: During a 5 year period (Jan 1, 2002 to Dec 31, 2006), 956 PEG were performed using a standard pull-through technique. Complete electronic medical records were available in 883 cases, and were reviewed for predictive factors of bleeding, bleeding complications, and interventions for hemostasis. Bleeding was defined as blood per PEG (bright blood, coffee ground material, heme-positive aspirate), hematemesis, melena, hematochezia, peri-stomal bleeding, drop in hemoglobin or occult blood in stool. Results: Of the 883 cases, 54.9% (n=495) were male, 45.1% (n=398) were female, with a mean age of 67.8 ± 16.6 years. Gastrointestinal bleeding following PEG occurred in 28 cases (3.2%), of which only 5 cases (0.57%) were found at endoscopy to be a result of PEG placement; 3 developed immediate peri-stomal bleeding, one of which required intra-dermal injection of epinephrine. Endoscopy on the other 2 patients revealed PEG related bleeding gastric ulcers treated with injection therapy. Blood transfusion was required in 1 patient. Cases were reviewed for the use of anti-platelet (aspirin, clopidogrel) or anticoagulants (heparin, enoxaparin, warfarin). Among cases where PEG related bleeding was identified, 100% (n=5) were on at least 1 anti-platelet or anticoagulant drug, with none having received clopidogrel. In the non-bleeding group, 62.53% (n=549) were on at least one anti-platelet or anticoagulant drug administered in proximity to PEG insertion, and 9.7% (n=85) were treated with clopidogrel. The mean cessation period for clopidogrel prior to PEG was 2.3 ± 2.14 days. A multiple logistic regression analysis of variables including gender, indication, or the use of anti-platelet or anticoagulant medications showed that none of these variables were statistically significant predictors of bleeding (p≥ 0.127). Conclusions: Bleeding complications associated with PEG placement are rare, even in the face of recent use of anticoagulants and anti-platelet medications. Specifically, clopidogrel did not contribute to an increase in bleeding outcomes, despite cessation of the drug a mean of 2.3 days prior to PEG. Prospective studies are needed to establish evidence-based guidelines for optimal cessation of anti-platelet or anticoagulant medications prior to PEG.