Periprocedural bleeding in patients undergoing percutaneous coronary intervention

INTRODUCTION — The advent of aggressive, multidrug antithrombotic therapy during percutaneous coronary intervention (PCI) has led to significant reductions in short- and long-term ischemic outcomes. (See "Antithrombotic therapy for elective percutaneous coronary intervention: General use".)

However, periprocedural bleeding remains a frequent complication of PCI, and its frequency and severity is related in part to the intensity of antithrombotic therapy. It is also a quality indicator of practice patterns. Initially, periprocedural bleeding was viewed as a relatively benign consequence, but data have linked the occurrence of bleeding and its treatment (ie, blood transfusion) to increased short- and long-term mortality.

This topic reviews most aspects of periprocedural bleeding among patients undergoing PCI. Gastrointestinal bleeding in this setting is discussed separately. (See "Gastrointestinal bleeding in patients undergoing percutaneous coronary intervention".)

Bleeding in relation to coronary artery bypass grafting or in association with diagnostic coronary angiography is discussed elsewhere. (See "Early noncardiac complications of coronary artery bypass graft surgery", section on 'Bleeding' and "Complications of diagnostic cardiac catheterization", section on 'Local vascular complications' and "Access-related complications of percutaneous access for diagnostic or interventional procedures".)

DEFINITION — For the purpose of this topic, periprocedural bleeding is any bleeding that occurs during or within 48 hours of the procedure. However, some studies have including episodes of bleeding that occur during the index hospitalization. Estimates of the incidence of periprocedural bleeding during percutaneous coronary intervention (PCI) lack precision due to the use of different definitions. (See 'Incidence' below.) In addition, relevant data come from differing patient populations who were treated with varying regimens of adjunctive pharmacology. Early studies of adjunctive pharmacology in PCI utilized definitions that were extrapolated from fibrinolytic trials (such as the TIMI and GUSTO scales). These early definitions focused on occurrence of intracranial bleeds or large drops in hemoglobin [1]. In other studies, the definitions of major bleeding have included less severe, but still clinically significant, bleeds (table 1A-B) such as the need for transfusion, bleeding requiring surgical intervention, prolonging hospital stay, or cardiac tamponade.

The heterogeneity in the definition of major bleeding at the time of PCI creates difficulty in conducting and interpreting the results of clinical research. For example, the relative safety (with regard to bleeding) of different antithrombotic agents may be difficult to ascertain when the combined safety end points across trials have different components. In an attempt to overcome this problem, there has been an attempt to develop a universal definition of bleeding. In 2011, the Bleeding Academic Research Consortium (BARC) published a consensus classification for bleeding (table 1B) [2]. BARC has been prospectively validated and BARC types 2, 3, or 5 bleeding identify patients at increased risk of death, bleeding, and ischemic events [3-5]. In addition, BARC has a comparable predictive ability to the TIMI, GUSTO, and REPLACE-2 scales [6]. While not universally accepted, BARC provides a contemporary standard and has been commonly used in clinical trials published after 2013.

INCIDENCE — Among patients enrolled in randomized trials and registries of adjunctive pharmacology for percutaneous coronary intervention (PCI), the incidence of major periprocedural bleeding ranged from 3 to 6 percent in reports published through 2008 [7-9]. More studies have found a rate closer to 2 percent [10,11]. The variability in frequency is significantly affected by access sites selected, namely radial versus femoral.

Femoral access site bleeding is the most frequent cause of periprocedural bleeding, accounting for approximately 30 to 60 percent of such events [7,8,12]. Other frequent sites of bleeding are gastrointestinal (10 to 15 percent) and retroperitoneal (5 to 12 percent). Intracranial bleeding is less common (2 to 3 percent) [7,8]. (See "Gastrointestinal bleeding in patients undergoing percutaneous coronary intervention".)

Access site bleeding is reduced by approximately one-half with use of the radial artery approach, although non-access related and overall bleeding may not be lower [13].

RISK FACTORS — Several studies have examined predictors of bleeding during and after percutaneous coronary intervention (PCI) [7-9,14,15]. Consistent baseline predictors are age greater than 75 years, female sex, lower body weight, and baseline comorbidities such as anemia and chronic kidney disease [16-18]. The reasons for the predilection among women are not clear; some studies have suggested this relation becomes less evident after correction for sex-specific baseline differences such as weight, while others demonstrate a strong relation between female sex and vascular complications [19-21].

A retrospective analysis of the TRITON-TIMI 38 trial (see "Acute non-ST-elevation acute coronary syndromes: Early antiplatelet therapy") evaluated potential risk factors for serious bleeding (defined as both TIMI major and TIMI minor bleeding) [22]. In addition to those mentioned above, the following were identified: glycoprotein (GP) IIb/IIIa inhibitor use, duration of the pharmacologic intervention, ST-elevation myocardial infarction, and arterial hypertension.

The most common periprocedural factors are sheath size, procedural duration, intraaortic balloon counterpulsation, complexity of the intervention, and intensity and complexity of the antithrombotic regimen. Other risk factors include left main or three vessel disease, smoking, and International Normalized Ratio (INR) >2.6 in patients taking warfarin [23]. (See "Coronary artery disease patients requiring combined anticoagulant and antiplatelet therapy".)

Models to predict risk have been developed. A prognostic risk score for major bleeding during PCI has been developed with data from the REPLACE-1 and REPLACE-2 trials (table 2) [14,24]. Using a sum of weighted integers, examination of seven variables (age >55 years, female gender, glomerular filtration rate <60 mL/1.73 m², preexisting anemia, low-molecular-weight heparin with 48 hours of PCI, GP IIb/IIIa inhibitor use, and intraaortic balloon pump use) allows one to estimate the risk of bleeding with a range of 1.0 percent for no risk factors up to 5.4 percent for high-risk patients. While this prognostic score examined only patients with femoral access and has yet to be validated in less selected, "real world" registries, this assessment nevertheless does allow identification of high-risk patients who may benefit from enhanced observation after PCI.

The Academic Research Consortium (ARC) proposed clinical criteria for patients at high risk of bleeding that were validated in a study of 9623 patients undergoing PCI between 2014 and 2017 at a tertiary care center. The major criteria were, in order of descending importance: moderate or severe anemia, oral anticoagulation, malignancy, severe/end-stage chronic kidney disease, planned surgery, and thrombocytopenia. Age over 75 years, prior stroke, and prior bleeding were considered minor criteria. HBR was defined as one major or two minor criteria, with the rates of bleeding at one year being 9.1 percent in those with high bleeding risk [6].

Antithrombotic therapy — Antithrombotic agents are used during percutaneous coronary intervention to decrease the risk of ischemic events. However, they are associated with an increased risk of periprocedural bleeding:

●Oral anticoagulant therapy – Some patients scheduled for PCI are on long-term anticoagulant therapy. The approach to such patients is discussed separately. (See "Coronary artery disease patients requiring combined anticoagulant and antiplatelet therapy".)

●Glycoprotein IIb/IIIa platelet inhibitors – The use of a GP IIb/IIIa inhibitor is associated with a 1.4- to 2.7-fold increase in risk of periprocedural bleeding [7-9,25]. (See "Antithrombotic therapy for elective percutaneous coronary intervention: General use", section on 'GP IIb/IIIa inhibitors'.)

●Unfractionated heparin – The risk periprocedural bleeding with unfractionated heparin (UFH) is greatest when the activated clotting time is above 300 seconds. (See 'Bleeding parameters' below and "Acute ST-elevation myocardial infarction: Management of anticoagulation" and "Heparin and LMW heparin: Dosing and adverse effects", section on 'Other complications' and "Anticoagulant therapy in non-ST elevation acute coronary syndromes", section on 'Unfractionated heparin compared with enoxaparin'.)

●Low molecular weight heparin – The rate of bleeding with low molecular weight heparin (LMWH) has been compared with that with UFH in several clinical trials, with some showing an increased rate (9.1 percent versus 7.6 percent) [26] and others showing no difference [27,28]. (See "Antithrombotic therapy for elective percutaneous coronary intervention: Clinical studies", section on 'Low molecular weight heparin' and "Anticoagulant therapy in non-ST elevation acute coronary syndromes", section on 'Unfractionated heparin compared with enoxaparin'.)

●Fondaparinux – Fondaparinux is a selective antithrombin that has been compared with enoxaparin for treatment of non-ST-segment elevation acute coronary syndromes in the OASIS-5 study, in which 40 percent of patients had PCI. Patients randomized to fondaparinux had lower incidences of major bleeding (3.1 percent versus 5.0 percent) and death at 30 days (2.9 percent versus 3.5 percent). However, catheter-related thrombus was statistically more frequent in the fondaparinux group (0.9 percent versus 0.4 percent). There also has been concern that the dose of enoxaparin (1.0 mg/kg twice daily, adjusted for creatinine clearance) may have been excessive [29]. (See "Anticoagulant therapy in non-ST elevation acute coronary syndromes", section on 'Fondaparinux'.)

●Bivalirudin – Bivalirudin monotherapy, in place of heparin with a GP IIb/IIIa inhibitor, has been associated with a lower risk of periprocedural bleeding in several studies (see "Antithrombotic therapy for elective percutaneous coronary intervention: Clinical studies", section on 'Anticoagulant therapy'):

•In the REPLACE-2 trial of 6010 patients undergoing urgent or elective PCI, bivalirudin with provisional IIb/IIIa blockade reduced in-hospital major bleeding (2.4 percent versus 4.1 percent) in comparison with heparin with planned IIb/IIIa blockade [30].

•The ISAR-REACT 3 trial randomly assigned 4570 patients with stable or unstable angina who had been pretreated with 600 mg clopidogrel ≥2 hours before PCI in double-blind fashion to heparin or bivalirudin monotherapy. Bivalirudin led to a decrease in major bleeding at 30 days (3.1 percent versus 4.6 percent) [31]. Subsequent analysis showed that the decrease in bleeding occurred mainly in patients who were at low risk of bleeding (eg, age ≤75 years, men, body weight ≤70 kg, no renal insufficiency, single-vessel PCI), and that the patients with a lower bleeding risk were more likely to have myocardial infarction [15].

•The Acute Catheterization and Urgent Intervention Triage strategY (ACUITY) trial randomized 13,819 patients with unstable angina or non-ST-segment elevation myocardial infarction in open-label fashion to bivalirudin monotherapy, heparin with IIb/IIIa receptor inhibitor, or bivalirudin with IIb/IIIa receptor inhibitor. Bivalirudin monotherapy led to a statistically significant lower risk of major bleeding (3.0 percent versus 5.7 percent).

•In the HORIZONS-AMI study, 3602 patients with ST-segment elevation myocardial infarction were randomly assigned to either bivalirudin or heparin with IIb/IIIa receptor inhibitor. Bivalirudin use led to a statistically significantly, lower incidence of major bleeding (4.9 percent versus 8.3 percent), transfusion (2.1 percent versus 3.5 percent), thrombocytopenia (<100,000 platelets/mm³, 1.1 percent versus 2.9 percent), and 30-day death (2.1 percent versus 3.1 percent).

●In the MATRIX study, 7213 patients with acute coronary syndromes were randomly assigned to bivalirudin or unfractionated heparin; in a second randomization, patients in the bivalirudin group were also randomized to post-PCI bivalirudin or not [32]. The rate of major adverse events overall was not different according to bivalirudin therapy, but the BARC type 3 and type 5 bleeding was lower for the initial bivalirudin use versus heparin (30-day rate, 1.4 versus 2.5 percent; p<0.001), and for patients who received bivalirudin post-PCI versus no post-PCI infusion (1 versus 1.8 percent; p = 0.03).

●Platelet P2Y12 receptor blockers – Clopidogrel, prasugrel, ticagrelor, and cangrelor have been studied in patients with acute coronary syndromes, the majority of whom underwent PCI (see "Antithrombotic therapy for elective percutaneous coronary intervention: Clinical studies", section on 'antiplatelet therapy'):

•In the TRITON-TIMI 38 trial, 13,608 patients were randomly assigned to prasugrel or clopidogrel. Prasugrel was associated with a higher rate of major bleeding (2.4 percent versus 1.8 percent), including life-threatening bleeding (1.4 percent versus 0.9 percent) [33]. (See "Acute non-ST-elevation acute coronary syndromes: Early antiplatelet therapy".)

In the PLATO trial, 18,624 patients were randomly assigned to either ticagrelor or clopidogrel; approximately 65 percent underwent PCI. Ticagrelor led to a significantly higher rate of major bleeding (4.5 percent versus 3.8 percent) [34]. (See "Acute non-ST-elevation acute coronary syndromes: Early antiplatelet therapy".)

•Cangrelor, an intravenous, reversible adenosine diphosphate receptor antagonist, has been compared with 600 mg clopidogrel loading in two randomized trials. (See "Antithrombotic therapy for elective percutaneous coronary intervention: Clinical studies", section on 'Cangrelor'.)

In the CHAMPION PCI trial, cangrelor resulted in a trend for higher major bleeding using ACUITY criteria [35]. In CHAMPION PLATFORM, cangrelor was associated with a significantly higher incidence of major bleeding using the same criteria (5.5 percent versus 3.5 percent) [36].

●Dual antiplatelet therapy – Duration of dual antiplatelet therapy (aspirin plus P2Y12 receptor blockers) has been linked to risk of bleeding [37-39].

A meta-analysis of nine randomized clinical trials, consisting of 25,907 patients, showed no overall difference in net adverse clinical events between patients who received one and three months versus 6 to 12 months of dual antiplatelet therapy (hazard ratio [HR] 0.92; 95% CI 0.79-1.07). The shorter duration was associated with lower risk of any bleeding (HR 0.55; 95% CI 0.46-0.66). However, there was a nonsignificant but increasing trend of net adverse clinical events for those with more left main and left anterior descending lesions and those with ST-elevation myocardial infarction on presentation [40].

OUTCOMES AFTER BLEEDING — Periprocedural bleeding is associated with an increased risk of early (in-hospital and 30-day) and late mortality, major adverse cardiovascular events (MACE), and readmission for bleeding [16,41]. The discriminatory power of bleeding for predicting one-year mortality was comparable to nonfatal myocardial infarction in one study [42] and stronger than nonfatal myocardial infarction as a predictor of 30-day death in multiple studies [24,43,44].

The following studies have evaluated the relationship between periprocedural bleeding and outcome:

Among patients enrolled in randomized trials and various clinical registries, there is an approximate 3- to 10-fold increase in in-hospital and 30-day mortality for bleeding versus no bleeding [8-10,22,45-47].

For example, in a 2012 report from the (United States) National Cardiovascular Data Registry of over 3,000,000 procedures performed between 2004 and 2011, major bleeding was associated with increased in-hospital mortality (5.26 versus 1.87 percent; p<.001) in a matched cohort [10]. In the entire cohort of individuals, the adjusted population attributable risk for mortality related to major bleeding was 12.1 percent. The increased risk of these events has been observed for both major and minor bleeding events, with a direct relation between severity and likelihood of poor outcomes [9,42].

The increase in the risk of early death exists for both non-access and access site bleeding. In a meta-analysis of 25 PCI studies (n = 2,400, 645), bleeding was an independent predictor of death at both sites (risk ratios [RR] 4.06, 95% confidence interval [CI] 3.21-5.14 and 1.71, 95% CI 1.37-2.13, respectively) [48]. The increase in risk was also dependent on the site of bleeding (eg, gastrointestinal [RR = 2.78], retroperitoneal [RR = 5.87], and intracranial [RR = 22.71]).

●Periprocedural bleeding also is associated with prolonged hospital stay, intraprocedural complications (coronary perforation, embolization, dissection, ventricular fibrillation), and increased risk of myocardial infarction, stroke, renal failure, rehospitalization, and increased hospital costs [8,10,47,49].

●Whether the mortality risk attributable to bleeding persists in late follow-up is unclear, with some studies suggesting a 2- to 4.5-fold increased risk of death at one year [7,8,42,45], with others finding no significant increase [22].

In some cases, it is not known whether bleeding is causative or simply a marker of heightened patient risk (figure 1) [50]. When a local effect (eg, intracranial hemorrhage with mass effect) or profound bleeding (eg, blood loss leading to severe hypotension) is present, bleeding is causative. Nevertheless, comorbidity frequently is present in patients who bleed, and these morbidities often independently correlate with risk of death and nonfatal adverse events. Although the relation of bleeding to adverse outcomes remains significant after adjustment for these morbidities, statistical modeling cannot entirely account for baseline differences between patients with and without bleeding.

Moreover, steps in management taken to address bleeding carry the potential to aggravate the risk of ischemic complications and subsequent mortality. Antithrombotic or antiplatelet therapy may be withheld in patients with bleeding, augmenting the risk of thrombotic complications such as coronary artery stent thrombosis. In the GRACE registry, patients with acute myocardial infarction who bled had a higher frequency of discontinuation of aspirin, thienopyridines, and heparin. Patients with bleeding who had discontinuation of these medications had a higher likelihood of in-hospital death in contrast to patients in whom these medications were continued despite bleeding [50]. In addition, blood transfusion may worsen prognosis [51].

Bleeding leads to platelet activation and initiation of the clotting cascade to cause rapid local hemostasis. Mechanisms for preventing the systemic amplification of this response may be deficient in patients with vascular disease because such pathways are predominantly active within endothelial cells, potentially leading to a hypercoagulable state [52]. Erythropoietin, released in response to anemia, also may promote a prothrombotic state through activation of platelets and inducing plasminogen activator inhibitor-1, a procoagulant cytokine [53,54]. This prothrombotic state may extend beyond the acute phase of erythropoietin release.

PREVENTION — The prevention of periprocedural bleeding begins with the identification of patients at high bleeding risk. (See 'Risk factors' above.)

Access site — Radial artery access, compared with the femoral artery, is associated with fewer bleeding complications [55-57], and in patients with acute coronary syndromes, has been found to improve survival in randomized clinical trials [58,59]. For patients at increased risk of access site bleeding, it should be considered strongly. Its role in preventing periprocedural bleeding is discussed elsewhere. (See "Periprocedural complications of percutaneous coronary intervention", section on 'Vascular complications' and "Access-related complications of percutaneous access for diagnostic or interventional procedures".)

Because the vast majority of periprocedural bleeds occur at the femoral artery access site, meticulous attention to the methods for vascular access (eg, avoidance of posterior wall puncture and puncture too high or too low) by ideally using ultrasound guidance and fine needle entry, proper hemostasis, and appropriate sheath management (ie, use of smaller size, shorter duration of insertion) reduces the incidence of bleeding. (See "Complications of diagnostic cardiac catheterization", section on 'Hemostasis at the femoral access site' and "Percutaneous arterial access techniques for diagnostic or interventional procedures", section on 'Hemostasis at the access site'.)

Control of hemostasis is necessary to minimize the risk of bleeding from the femoral artery puncture site. The use of vascular closure devices has been shown to reduce the incidence of periprocedural bleeding in some [11] but not all studies [9,60,61]. We do not believe that any one device is superior to the others. The addition of ultrasound guidance to femoral access has been shown to reduce access time but not bleeding [62]. (See "Complications of diagnostic cardiac catheterization", section on 'Hemostasis at the femoral access site' and "Percutaneous arterial access techniques for diagnostic or interventional procedures", section on 'Hemostasis at the access site'.)

Gastrointestinal tract — The prevention of gastrointestinal tract bleeding in patients undergoing percutaneous coronary intervention (PCI) is discussed separately. We recommend the use of proton pump inhibitors in patients at high risk of gastrointestinal bleeding. (See "Gastrointestinal bleeding in patients undergoing percutaneous coronary intervention", section on 'Proton pump inhibitors'.)

Bleeding parameters — Monitoring of activated clotting time (ACT) is important for minimizing the risk of such bleeding in patients receiving unfractionated heparin. For PCI without glycoprotein (GP) IIb/IIIa inhibitor, goal ACTs generally are 300 to 350 seconds. For patients receiving an IIb/IIIa receptor inhibitor, goal ACTs are lower (200 to 250 seconds) because these drugs suppress thrombin activity [63]. Of note, post-procedural unfractionated heparin (UFH) in PCI patients is associated with a ≥2.5-fold increase the risk of periprocedural bleeding [60,64]. In contrast, use of platelet reactivity parameters has not been found to associated with reduction in bleeding [65,66].

Patients on oral anticoagulants — The approach to patients on oral anticoagulants at the time of PCI is discussed separately. (See "Periprocedural management of antithrombotic therapy in patients receiving long-term oral anticoagulation and undergoing percutaneous coronary intervention", section on 'Elective patients'.)

CLINICAL MANIFESTATIONS AND DIAGNOSIS — The clinical manifestations and diagnosis of periprocedural bleeding depend on the affected anatomic site. As mentioned above, femoral artery, gastrointestinal tract, and retroperitoneal space are the most common locations for major bleeding after percutaneous coronary intervention (PCI). Closed cavity bleeding, such as in the cranium or pericardial space, is less common but may be immediately life-threatening. (See "Complications of diagnostic cardiac catheterization", section on 'Local vascular complications' and 'Incidence' above and "Access-related complications of percutaneous access for diagnostic or interventional procedures".)

In many patients, clinical manifestations are obvious: hematemesis or large groin hematoma. However, occult bleeding, especially retroperitoneal, should be suspected in patients with unexplained altered neurologic findings, hypotension, clinical or hemodynamic instability, or significant drops in hemoglobin or hematocrit.

In many patients, clinical manifestations are obvious: hematemesis or large groin hematoma. However, occult bleeding should be suspected in patients with unexplained altered neurologic findings, hypotension, clinical or hemodynamic instability, or significant drops in hemoglobin or hematocrit.

Femoral artery — Significant bleeding in the groin may be evident by the complaint of local pain and the finding of a large hematoma, particularly in thin patients. However, in obese patients, a large femoral artery bleed may be difficult to diagnose by physical examination. Ultrasound normally suffices for evaluation and can be used to guide therapy (eg, thrombin injection for pseudoaneurysm). Retroperitoneal bleeding usually is associated with femoral access above the inguinal ligament, posterior wall femoral puncture, or inadvertent perforation of the aortoiliac vessels with guidewire or catheter advancement. Retroperitoneal bleeding may present as hypotension and bradycardia during sheath removal (it may be mistaken for hypervagotonia).

Important remote site bleeding can occur within the cranium, thorax, or abdomen. Early diagnosis requires considering the possibility. For suspected bleeding at any of these sites, computed tomography (CT) imaging of the body or head (without contrast) is recommended.

Retroperitoneal space — Due to the size of the retroperitoneal space, bleeding may be clinically significant before clinical manifestations develop. Back pain and hypotension are common. We prefer angiography using femoral access (if possible) to secure the diagnosis and to potentially facilitate catheter-based treatment. Angiography can be used to facilitate the diagnosis but is also very important to control the hemorrhage with balloon inflation proximal to the site of hemorrhage. In life-threatening bleeding, inflation of a balloon placed from the contra lateral groin into the proximal iliac artery above the site of bleeding is life-saving. CT is an alternative for the diagnosis of retroperitoneal bleeding if the diagnosis is considered after the femoral catheters have been removed.

Gastrointestinal tract — Significant periprocedural bleeding from the gastrointestinal tract usually occurs in patients with prior gastrointestinal lesions, such as gastric or duodenal ulcers, gastritis, or colonic polyps. Hematemesis is the most common clinical manifestation. The source of the bleed is determined by a workup of lower or upper gastrointestinal bleeding. (See "Approach to acute upper gastrointestinal bleeding in adults" and "Approach to acute lower gastrointestinal bleeding in adults".)

Intracranial bleeding — Acute intracranial bleeding is a life-threatening complication. Patients may complain of acute onset of headache or vomiting or they may be observed to have mental status changes or focal deficits. (See "Spontaneous intracerebral hemorrhage: Pathogenesis, clinical features, and diagnosis", section on 'Clinical presentation'.)

Urgent noncontrast cranial CT is the study most widely used to evaluate for the presence of acute intracerebral hemorrhage, which is evident almost immediately. In patients with neurologic findings post-angiography, urgent CT can be used to rule out bleeding and facilitate early attempts at intervention to diagnose and then treat embolic occlusion. (See "Spontaneous intracerebral hemorrhage: Pathogenesis, clinical features, and diagnosis", section on 'Head CT'.)

MANAGEMENT — Prompt recognition, timely diagnostic evaluation, achievement of hemostasis, and, when appropriate, reversal of anticoagulation, are fundamental in the management of major periprocedural bleeding. There are no established approaches to the management of patients at high risk and we use an individualized approach. We monitor high-risk patients closely. We admit unstable patients to an intensive or coronary care unit [17].

The section below discussed the general approach to major bleeding after percutaneous coronary intervention (PCI). Management of specific complications of periprocedural bleeding is discussed elsewhere. (See "Complications of diagnostic cardiac catheterization".)

Reversal of anticoagulation — Reversal of the effect of anticoagulation should be considered in patients with life-threatening bleeding, with need for open surgery to correct the etiology of bleeding, or in whom hemostasis cannot be achieved. Because reversal of anticoagulation theoretically may promote thrombogenicity, reversal should be sought only with careful consideration of the potential benefits and risks to the patient. Except in rare cases, patients with recent intracoronary stent implantation should continue to receive dual antiplatelet therapy to ameliorate the risk of acute stent thrombosis [67].

●UFH – Intravenous protamine rapidly reverses the effects of unfractionated heparin and, to a lesser extent, that of low molecular weight heparin. Protamine has been shown to facilitate early sheath removal after PCI, although one report has reported potential harm after placement of drug-eluting stents [63,64,68]. (See "Heparin and LMW heparin: Dosing and adverse effects", section on 'Reversal'.)

●Bivalirudin – For bivalirudin, there is no agent for reversal of its effects, but the therapeutic half-life of the drug is relatively short (25 min in patients with normal renal function).

●Warfarin – For patients with bleeding on warfarin, either fresh frozen plasma (2 to 3 units) and/or vitamin K should be administered. Vitamin K, when given intravenously (0.5 to 3 mg), should be infused slowly with doses adjustment according to urgency of the bleeding, and with awareness of the potential for anaphylaxis when administered via this route. Of note, there is no difference in effectiveness of vitamin K given orally or intravenously 24 hours after administration. Thus, oral vitamin K (1 to 3 mg) can be used in patients who do not need immediate reversal of anticoagulation. (See "Management of warfarin-associated bleeding or supratherapeutic INR".)

●Direct oral anticoagulants – A discussion of the emergency treatment of bleeding in patients on anticoagulants other than heparin or warfarin is found elsewhere. (See "Management of bleeding in patients receiving direct oral anticoagulants".)

Reversal of GP IIb/IIIa inhibitors — Glycoprotein (GP) IIb/IIIa inhibitors increase the risk of periprocedural bleeding. Platelet transfusion is effective for reversing the effects of abciximab, but is less effective against the small molecule GP IIb/IIIa inhibitors (eg, eptifibatide, tirofiban). For the small molecule agents, the biologic half-life with dosing adjusted for creatinine clearance is approximately 2.5 hours. The management of thrombocytopenia induced by GP IIb/IIIa inhibitors is discussed elsewhere. (See "Drug-induced immune thrombocytopenia".)

Blood transfusion — In patients who have undergone PCI, the rate of and risk factors for bleeding and outcomes after blood transfusion have been well studied. The optimal strategy for transfusion in these individuals has not.

In a retrospective cohort study of all patient visits from the United States (US) CathPCI Registry (2009 to 2013), the overall rate of transfusion was 2.14 percent [69]. Risk factors for transfusion included older age, female sex, hypertension, diabetes, advanced renal dysfunction, prior myocardial infarction or prior heart failure. In other studies, the use of a potent antithrombotic regimen has also been identified as a risk factor. (See "Periprocedural complications of percutaneous coronary intervention", section on 'Access site bleeding' and "Periprocedural complications of percutaneous coronary intervention", section on 'Anticoagulation-associated bleeding' and "Access-related complications of percutaneous access for diagnostic or interventional procedures", section on 'Access site bleeding'.)

A large number of studies have linked the need for transfusion to adverse outcomes, including mortality, after percutaneous coronary intervention (PCI). The following are representative studies:

●In the British Columbia registry of PCI patients, transfused patients, in comparison to nontransfused, had significantly increased 30-day (12.6 percent versus 1.3 percent) and one-year mortality (22.3 percent versus 3.2 percent). Notably, the rates of transfusion and mortality were 40 to 50 percent lower when radial access was used in place of femoral access [70]. (See "Periprocedural complications of percutaneous coronary intervention", section on 'Radial artery access' and "Percutaneous arterial access techniques for diagnostic or interventional procedures", section on 'Radial artery'.)

●Two separate registry studies (one of 17,901 patients from Mayo Clinic and one of 10,974 patients from Washington Hospital) demonstrated incremental risk of death with increasing transfusion requirements after PCI [8,9]. In the Mayo Clinic study, increased mortality was demonstrated in patients who received only one or two units of blood in comparison to nontransfused patients, with follow-up extending to six years after PCI.

●In the US CathPCI Registry (2009 to 2013) referred to above, receipt of transfusion was associated with significantly increased risks of myocardial infarction, stroke, and in-hospital death (4.5 versus 1.8, 2.0 versus 0.2, and 12.5 versus 1.2 percent, respectively) [69].

Potential mechanisms of adverse outcome after blood cell transfusion — Patients in need of blood cell transfusion after PCI have been found to be older and have greater comorbidity than nontransfused patients. Adjustment for these baseline characteristics using either multivariate models and/or propensity analyses has shown transfusion to be an independent predictor of both in-hospital and one-year survival. While statistical methods cannot fully account for retrospective bias, several explanations, in addition to bias, have been proposed to help explain the link between blood transfusion and excess mortality:

●Rises in hemoglobin level with blood cell transfusion increase oxygen delivery, but measures of tissue oxygenation do not change or, in fact, may decrease. Storage of red blood cells may lead to less deformability and more fragility of the cell, reduced nitric oxide activity (resulting in less vasodilatation during decoupling of oxygen delivery), and depletion of 2,3-diphosphoglyceric acid (leading to reduced tissue extraction of oxygen) [71-75].

●Transfusion may be prothrombotic, due to enhancement of procoagulant protein plasminogen activator inhibitor (PAI)-1 content during cell storage, acute platelet release of CD40 ligand during transfusion, release of adenosine diphosphate from stored red blood cells that leads to platelet activation, and relative deficiencies in nitric oxide in the transfused blood products [76].

●Hemolytic reactions and the consequences of large-volume transfusion (eg, coagulopathy, electrolyte disturbances) may lead to increased mortality [77].

The physiologic changes in red blood cells that occur with storage are discussed elsewhere. (See "Practical aspects of red blood cell transfusion in adults: Storage, processing, modifications, and infusion", section on 'RBC age/storage duration effect on clinical outcomes'.)

Recommendations for blood transfusion — General recommendations for transfusion in patients with cardiovascular disease are found elsewhere. Briefly, we suggest transfusion for stable patients with coronary artery disease when the hemoglobin falls below 7 to 8 g/dL; for patients with acute coronary syndrome, we suggest transfusion when the hemoglobin falls below 8 to 10 g/dL. (See "Indications and hemoglobin thresholds for RBC transfusion in adults", section on 'Acute and chronic cardiovascular disease' and "Overview of the nonacute management of ST-elevation myocardial infarction", section on 'Red cell transfusion'.)

The optimal transfusion strategy for patients who have undergone PCI has not been well-studied, but the general recommendations presented above are a reasonable starting point. The rate and expected duration of bleeding, as well as the cardiovascular instability of the patient, should be considered in decision making. We feel that there is insufficient evidence to recommend specific hemoglobin (or hematocrit) targets for patients who have undergone PCI.

For patients with periprocedural bleeding and who have severe symptoms of angina or dyspnea (class ≥III or lifestyle impairment) or have evidence of myocardial ischemia despite revascularization, it is reasonable to transfuse when the hemoglobin falls below 10 g/dL. Transfusion may be reasonable at higher hemoglobin levels in patients for whom the rate of bleeding is fast and likely to continue. The clinical threshold for transfusion may be lower in those with acute coronary syndromes, as the risk of anemia appears to be greater in these patients [14,78]. In the absence of such findings, the net benefit may be minimal with significant potential for harm. One exception may be an asymptomatic patient who is at high risk from rebleeding from a noncompressible site, in whom a greater reserve of red cell mass may be desired.

In a Canadian study of 357 critically ill patients with cardiovascular disease, a restrictive transfusion policy (transfusion for hgb <7.0 g/dL), in comparison to a liberal policy (transfusion for hgb <10.0 g/dL), was associated with less multiorgan dysfunction but no difference in 30-day mortality (23 percent for both groups) [79,80]. The authors did report a trend toward less 30-day survival in the restrictive policy group among the patients with acute myocardial infarction or unstable angina (21 percent versus 26 percent), but this difference was not significant. In a separate study of patients with acute coronary syndromes, the predicted probability of 30-day death was higher with transfusion at nadir hematocrit values above 25 percent [81]. Conversely, another study of older adult patients with acute myocardial infarction demonstrated mortality benefit for transfusion for hematocrit less than 30 percent [78].

The general discussion of indications for red cell transfusion is found elsewhere. (See "Indications and hemoglobin thresholds for RBC transfusion in adults".)

SUMMARY AND RECOMMENDATIONS

●Incidence and risk factors – Periprocedural bleeding occurs in 3 to 6 percent of cases. Risk factors include age greater than 75 years, female gender, baseline anemia, and chronic kidney disease, sheath size, procedural duration, intraaortic balloon pump placement, complexity of the intervention, and intensity and complexity of the antithrombotic regimen. (See 'Incidence' above and 'Risk factors' above.)

●Outcomes – Periprocedural bleeding is associated with adverse outcomes including a fourfold or greater increase in early and late mortality (figure 1). The cause of adverse outcomes is often multifactorial and includes the local effects of the bleed, excess patient morbidity that is frequently present in those who bleed, and potential harm from therapy taken to address occurrence of bleeding. (See 'Outcomes after bleeding' above.)

●Prevention – This includes identification of patients at high risk; meticulous attention to the access site including consideration of radial access; and proper use of the activated clotting time for patients receiving heparin. (See 'Prevention' above.)

●Management – Management of periprocedural bleeding must be individualized as each intervention aimed at terminating the bleeding may adversely impact outcome. Interventions include reversal of anticoagulation and glycoprotein (GP) IIb/IIIa inhibitors, and blood transfusion. While blood transfusion is often considered for patients who bleed, it has been linked to increased mortality after percutaneous coronary intervention (PCI). Therefore, blood transfusion should only be undertaken after careful consideration of the potential benefit and risks to the patient. (See 'Management' above.)

●Periprocedural and long-term gastrointestinal bleeding are discussed separately. (See "Gastrointestinal bleeding in patients undergoing percutaneous coronary intervention".)