Accuracy of HAS-BLED and BleedMAP Scores in Predicting Postoperative Bleeding in Anticoagulated Patients Undergoing non-Cardiac Surgery

Abstract

Background

There are limited data on the performance of bleeding risk scores in predicting postoperative bleeding in anticoagulated patients undergoing elective non-cardiac surgery.

Objectives

To evaluate the predictive accuracy of the HAS-BLED and BleedMAP scores for postoperative bleeding risk in this population.

Methods

This retrospective study enrolled anticoagulated patients undergoing elective surgery referred to the Preoperative Medical Consultation clinic. The C-statistic was used to evaluate the discriminatory performance of the HAS-BLED and BleedMAP scores in predicting overall bleeding following non-cardiac surgery. The predictive performance of the two scores was compared using DeLong's method.

Results

Among 591 patients who had previously received anticoagulation therapy, 84.9% received warfarin, 9.5% enoxaparin, and 5.6% direct oral anticoagulants. Indications included atrial fibrillation (67.9%), venous thromboembolism (18.1%), and mechanical valve replacement (13.5%). Procedures were classified as minor (16.6%), low to moderate risk (46.4%), or high bleeding risk (37.1%). The mean HAS-BLED and BleedMAP scores were 2.35 ± 1.36 and 0.94 ± 0.81, respectively. At 1-month postoperative follow-up, the overall bleeding rate was 40.4%, with 2.5% classified as major bleeding events. The HAS-BLED score had a C-statistic of 0.512 (95% confidence intervals (CI): 0.434–0.591) for predicting overall bleeding, while the BleedMAP score showed moderate predictive ability (C-statistic: 0.581; 95% CI: 0.531-0.631). The two scores had no statistically significant difference (P = .13).

Conclusions

The HAS-BLED score showed limited discriminatory ability for postoperative bleeding in anticoagulated patients, though its negative predictive value was high. No significant difference in predictive performance was observed between the HAS-BLED and BleedMAP scores.

Keywords

anticoagulants BleedMAP HAS-BLED non-cardiac surgery postoperative bleeding

Introduction

Anticoagulants are effective in reducing the risk of thromboembolism, including preventing cardioembolic stroke in patients with non-valvular atrial fibrillation (AF) and decreasing recurrence and mortality in those with mechanical heart valves or venous thromboembolism.^1–3 An estimated 15% to 20% of anticoagulated patients undergo surgery or a procedure each year.⁴ For patients on anticoagulants undergoing elective non-cardiac surgery, anticoagulation management may involve temporarily interrupting therapy during the perioperative period, with some cases may require bridging using a short half-life parenteral regimen. This approach necessitates balancing the competing risks of thromboembolism and perioperative bleeding while considering multiple factors, including the thromboembolic risk during therapy interruption, such as CHA₂DS₂VASC score and the bleeding risk associated with patient- and procedure-related factors.^1,5–7 Bleeding is a major surgical complication that adversely affects clinical outcomes, including increased morbidities, the need for additional interventions, and mortality.⁸ Moreover, perioperative bleeding complications lead to prolonged hospital stays and incur higher hospitalization costs, with incremental increases of 0.7–6.3 days and approximately USD 3300–14 700 per patient, respectively, depending on the type of operation.^9,10 These impacts underscore the importance of accurate and reliable bleeding risk predictors.

Several bleeding risk scores have been developed to predict the risk of bleeding in patients receiving anticoagulation therapy. These scores are typically used to assess bleeding risk in patients with AF on anticoagulants. Commonly utilized risk assessment tools include the HEMORR2HAGES, HAS-BLED, ATRIA, and ORBIT scores.¹¹ Among these, the HAS-BLED score is a widely used and well-established tool for estimating the risk of major bleeding within one year for patients with AF receiving anticoagulation.¹² Risk categories are defined as low (score 0-1), moderate (score 2), and high (score ≥3).¹³ Whereas the BleedMAP score specifically predicts the risk of bleeding complications within three months of a procedure in patients on long-term anticoagulation who require temporary interruption of their medication. Risk levels are categorized as low (score 0), moderate (score 1-2), and high (score ≥3).¹⁴

Due to the limited number of established criteria for assessing bleeding in the post-procedural setting in anticoagulated patients and limited evidence to validate and compare these risk scores. In addition, there are a few proposed bleeding predictive scores with less popular use.

Our primary objective was to evaluate the predictive performance of the HAS-BLED and BleedMAP scores for post-procedural bleeding risk in anticoagulated patients. The secondary objective was to assess the predictive performance of both scores according to the type of anticoagulant used.

Material and Methods

Study Design and Population

This retrospective cohort study utilized data from the electronic medical records of anticoagulated patients who were referred to the Preoperative Medical Consultation (PROM) clinic at King Chulalongkorn Memorial Hospital (KCMH), a university hospital in Bangkok, Thailand. For peri-procedural anticoagulant management of vitamin K antagonists, parenteral anticoagulants, and direct oral anticoagulants (DOACs), the institutional protocol was developed in accordance with international practice guidelines.^15,16 Patients included in this study underwent procedures between November 1, 2015, and June 30, 2020. Each patient was evaluated for both thromboembolic risk during anticoagulant interruption and procedural bleeding risk. For patients with a low bleeding risk or an intermediate bleeding risk combined with a high thromboembolic risk, anticoagulant therapy was continued during the procedure. For patients with either high or intermediate bleeding risk combined with a low thromboembolic risk, anticoagulants were withheld without bridging prior to the procedure (warfarin was discontinued 5 days before, and DOACs were withheld 24-48 h preoperatively, depending on renal function and procedural bleeding risk). In patients whose thromboembolic risk clearly outweighed the increased bleeding risk, bridging therapy with parenteral anticoagulants was considered during interruption of oral anticoagulants. Post-procedurally, hemostasis was assessed before resuming therapy. Warfarin was usually resumed at the maintenance dose on the evening of, or the morning after, the procedure, whereas DOACs were resumed at 6–8 and 48–72 h after low-bleeding-risk and high-bleeding-risk procedures, respectively. In patients at high thromboembolic risk, bridging therapy was restarted 24–48 h after the procedure, depending on the bleeding risk. However, variations in practice that deviate from the institutional protocol might occur, depending on the attending physician's clinical judgment in providing care. The inclusion criteria were: patients aged 18 years or older at the time of enrollment; those scheduled for an elective non-cardiac surgery at KCMH and those who had been on long-term anticoagulant therapy for at least seven days prior to the preoperative assessment at the PROM clinic. The exclusion criteria included patients scheduled for cardiac surgery or cardiovascular interventions, pregnant patients, and those with incomplete data. Although several factors such as significant liver or renal impairment, concomitant use of antiplatelets or NSAIDs, improperly adjusted anticoagulant dosage, prior bleeding history, thrombocytopenia, and relevant comorbidities may influence perioperative bleeding, patients with these factors were eligible for the present study. The majority of these factors were accounted for in covariate analysis to verify their association with bleeding outcomes.

The protocol for this study was approved by the Institutional Review Board of KCMH and the Faculty of Medicine, Chulalongkorn University.

Bleeding Risk Scores and Bleeding Events

The HAS-BLED bleeding risk score includes the following factors: hypertension (systolic blood pressure >160 mm Hg); abnormal renal function, defined as chronic dialysis, renal transplantation, or a serum creatinine level ≥2.26 mg/dL; abnormal liver function, characterized by chronic liver disease, total bilirubin >2 times the upper limit of normal, or aspartate aminotransferase, alanine aminotransferase, or alkaline phosphatase levels >3 times the upper limit of normal; a history of stroke; a history of bleeding; labile international normalized ratio (INR), defined as a time-in-therapeutic range <60% in patients on warfarin; elderly (>65 years); drugs concomitantly (antiplatelet agents or nonsteroidal anti-inflammatory drugs); and excessive alcohol consumption.¹² The risk score is classified into three categories: low (score 0-1), moderate (score 2), and high (score ≥3). Major bleeding rates reported for these categories were 1.1%, 1.9%, and 4.9%, respectively.¹³ The BleedMAP bleeding risk score includes the following factors: a history of bleeding, a mitral mechanical valve prosthesis, active cancer, and a low platelet count (platelets <150 × 10³/µL). The risk score is classified into three levels: low (score 0), moderate (score 1-2), and high (score ≥3). Major bleeding rates reported for these levels were 0.81%, 2.67%, and 10%, respectively.¹⁴ Both scores were calculated during the pre-procedural setting. Bleeding events were evaluated for their procedural relevance and classified by severity based on the criteria established by the International Society on Thrombosis and Haemostasis (ISTH) for surgical patients.⁸ Major bleeding was defined as fatal bleeding, symptomatic bleeding in a critical area or organ, or surgical site bleeding that was unexpected, prolonged, and/or sufficiently extensive to cause hemodynamic instability. Minor bleeding was defined as any bleeding that did not meet the criteria for major bleeding. Further details on bleeding severity classifications are presented in Supporting Information Table S1. An independent adjudicator reviewed all events using medical charts documented by healthcare providers and recorded them in accordance with predefined outcome definitions. The follow-up period ranged from the procedure date to 30 days post-procedure, with only the first recorded bleeding episode included in the analysis.

Statistical Analysis

The required sample size was calculated to be 865 patients, based on a cohort study that reported a postoperative bleeding prevalence of 11%.¹⁷ Categorical variables were presented as frequencies and percentages, while continuous variables were presented as mean ± standard deviation (SD) for normally distributed data and as median [interquartile range, IQR] for non-normally distributed data. Comparisons of categorical variables were conducted using the Chi-square test or Fisher's exact test, as appropriate. Continuous variables were compared using the Mann-Whitney U test. The area under the receiver operating characteristic (ROC) curve, or C-statistic, was used to evaluate the discriminatory performance of the HAS-BLED and BleedMAP scores in predicting overall bleeding following non-cardiac operation in anticoagulated patients. An area under the ROC curve greater than 0.9 was considered to indicate high accuracy, 0.7–0.9 moderate accuracy, 0.5–0.7 low accuracy, and 0.5 a chance result.¹⁸ The predictive performances of these scores were compared using the method proposed by DeLong et al.¹⁹ Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were also reported for each bleeding score. Clinical implications and Youden's index were assessed to determine the appropriate cutoff points for the HAS-BLED and BleedMAP scores in predicting overall bleeding risk. As a secondary objective, we further examined the predictive performance of both scores according to the type of oral anticoagulant used, including vitamin K antagonists, and DOACs. Univariate and multivariate logistic regression analyses were conducted to identify factors associated with bleeding. Variables with a P-value <.1 in the univariate analysis were included in the multivariate analysis, and results were reported as odds ratios (OR) with 95% confidence intervals (CI). A P-value <.05 was considered statistically significant. All statistical analyses were performed using STATA version 16. Multiple imputation was used to handle missing data.

Results

Baseline Characteristics

The study included a total of 591 patients (Figure 1), of whom 53.6% were female (n = 317). The median age of the cohort was 69 years [IQR: 60–77] (Table 1). Among the anticoagulants used, 84.9% of patients received warfarin, 9.5% enoxaparin, and 5.6% DOACs. The main indications for anticoagulation were AF or atrial flutter (67.9%), venous thromboembolism (18.1%), and mechanical valve replacement (13.5%). Regarding procedural risk, 16.6% of patients underwent procedures classified as minor, 46.4% as low to moderate risk, and 37.1% as high risk for bleeding, with corresponding 30-day major bleeding risks of ∼0%, 0–2%, and >2%, respectively, according to Scientific and Standardization Committee guidance.²⁰ Bridging therapy was administered to 287 of 591 patients (46.4%) before the procedure and to 220 of 569 patients (38.7%) after the procedure. Other patient characteristics are detailed in Supporting Information Table S2. Procedural types are presented in Supporting Information Table S3.

Figure 1.

Study flow chart.

Table 1.

Patient Characteristics.

	All Patients n = 591	Bleeding Events		P-Value
	All Patients n = 591	Bleeding n = 239	Non-Bleeding n = 352	P-Value
Age (years), median [IQR]	69 [60, 77]	68 [57, 76]	70 [61, 77]	.132
Male, n (%)	274 (46.4)	103 (43.1)	171 (48.6)	.190
BMI (kg/m²), median [IQR]	24.2 [21.8, 27.3]	24.3 [21.5, 27.6]	24.1 [21.8, 27.1]	.828
Hypertension, n (%)	393 (66.5)	153 (64.0)	240 (68.2)	.292
Chronic kidney disease, n (%)	115 (19.5)	41 (17.2)	74 (21.0)	.244
Liver disease, n (%)	28 (4.7)	13 (5.4)	15 (4.3)	.508
AF/AFL, n (%)	401 (67.9)	155 (64.9)	246 (69.9)	.199
CHA₂DS₂-VASc (median [IQR])	4 [3, 5]	4 [3, 5]	4 [3, 5]	.210
VTE, n (%)	107 (18)	45 (18.8)	62 (17.6)	.707
Antiplatelets, n (%)	66 (11.2)	36 (15.1)	30 (8.5)	.013
Anticoagulants	-	-	-	-
Vitamin K antagonist, n (%)	502 (84.9)	200 (83.7)	302 (85.8)	.481
LMWH, n (%)	56 (9.5)	26 (10.9)	30 (8.5)	.337
Direct oral anticoagulants, n (%)	33 (5.6)	13 (5.4)	20 (5.7)	.900
NSAIDs, n (%)	24 (4.1)	9 (0.4)	15 (4.3)	.764
Indication for anticoagulant	-	-	-	-
AF/AFL, n (%)	401 (67.9)	155 (64.9)	246 (69.9)	.199
Venous thromboembolism, n (%)	107 (18.1)	45 (18.8)	62 (17.6)	.707
Systemic thromboembolism, n (%)	50 (8.5)	15 (6.3)	35 (9.9)	.116
Mechanical heart valve	-	-	-	-
- Mechanical AVR, n (%)	21 (3.6)	13 (5.4)	8 (2.3)	.041
- Mechanical MVR, n (%)	41 (6.9)	25 (10.5)	16 (4.5)	.005
- Others, n (%)	18 (3.0)	10 (4.2)	8 (2.3)	.184
HAS-BLED, mean ± SD	2.35 ± 1.36	2.37 ± 1.23	2.33 ± 1.46	.850
BleedMAP, mean ± SD	0.94 ± 0.81	1.09 ± 0.87	0.83 ± 0.75	<.001
Pre-procedural anticoagulant management	-	-	-	.093
Uninterrupted anticoagulation, n (%)	32 (5.4)	11 (4.6)	21 (6.0)	-
Interrupted without bridging, n (%)	272 (46.0)	99 (41.4)	173 (49.1)	-
Interrupted with bridging, n (%)	287 (48.6)	129 (54.0)	158 (44.9)	-
Post-procedural anticoagulant management^†	-	-	-	-
Anticoagulation resumed ≤24 h, n (%)	93 (16.3)	29 (12.2)	64 (19.3)	.025
Anticoagulation resumed >24 h, n (%)	476 (83.7)	208 (87.8)	268 (80.7)	.025
Bridging therapy, n (%)	220 (38.7)	111 (46.8)	109 (32.8)	<.001
Procedural bleeding risk^a	-	-	-	<.001
Minor risk, n (%) (30-d risk of major bleed ∼0%)	98 (16.6)	22 (9.2)	76 (21.6)	-
Low/moderate risk, n (%) (30-d risk of major bleed 0%-2%)	274 (46.4)	112 (46.9)	162 (46.0)	-
High risk, n (%) (30-d risk of major bleed >2%)	219 (37.1)	105 (43.9)	114 (32.4)	-

Procedural bleeding risk classification based on the SSC guidance document for periprocedural anticoagulant management.¹⁸

†

A total of 569 patients received post-procedural anticoagulants; 237 experienced bleeding events, whereas 332 did not.

Abbreviations: AF, Atrial Fibrillation; AFL, Atrial Flutter; AVR, Aortic Valve Replacement; CHA₂DS₂-VASc, Congestive heart failure, Hypertension, Age ≥75 years, Diabetes mellitus, Stroke, Vascular disease, Age 65-74 years, Sex category (female); IQR, Interquartile Range; LMWH, Low Molecular Weight Heparin; MVR, Mitral Valve Replacement; NSAIDs, Nonsteroidal Anti-inflammatory Drugs; SSC, Scientific and Standardization Committee Communication.

The HAS-BLED score was calculated for a cohort of 208 patients, with a mean score of 2.35 ± 1.36. Of these, 26.9% were categorized as low risk, while 73.1% were at moderate to high risk for bleeding. For the BleedMAP score, 524 patients were assessed, with a mean score of 0.94 ± 0.81. Based on this assessment, 33.1% were categorized as low risk, while 66.9% were at moderate to high risk for bleeding (Table 2).

Table 2.

HAS-BLED Score and BleedMAP Score.

	No. of Patients with Complete Data	Bleeding Events		P-Value
	No. of Patients with Complete Data	Bleeding	Non-Bleeding	P-Value
HAS-BLED, n	208	90	118
Low risk (0-1 points)	56	23 (41.1%)	33 (58.9%)	.698
Moderate risk (2 points)	60	25 (41.7%)	35 (58.3%)	.766
High risk (≥3 points)	92	42 (45.7%)	50 (54.3%)	.537
BleedMAP, n	524	221	303
Low risk (0 points)	173	62 (35.8%)	111 (64.2%)	.039
Moderate risk (1-2 points)	337	149 (44.2%)	188 (55.8%)	.205
High risk (≥3 points)	14	10 (71.4%)	4 (28.6%)	.025

Event Rates of Bleeding and Thromboembolism

At the 1-month follow-up, the overall bleeding rate was 40.4% (n = 239), with 2.5% (n = 15) of these cases classified as major bleeding events. Among the patients who experienced bleeding, 50.2% (n = 120) had bleeding related to the procedure, 14.2% (n = 34) had bleeding unrelated to the procedure, and 35.6% (n = 85) had bleeding with an unknown relation to the procedure. There were no fatal bleeding events. Overall, 54% of these patients received bridging therapy before the procedure, with a median anticoagulant interruption of 5 days [IQR: 4–6]. Antiplatelet therapy was continued in 5 of 36 patients (14%), and the median CHA₂DS₂-VASc score was 4 [IQR: 3–5]. For thromboembolism events, there were 5 patients (0.8%), including 2 patients with ischemic stroke, 1 patient with pulmonary embolism, and 2 patients with other thromboembolism complications. Among these patients, 40% received bridging therapy, and the median anticoagulant interruption was 5 days [IQR: 2.5–7.5]. Only one patient had been receiving oral antiplatelet therapy prior to the procedure, which was withheld. For those with AF, the CHA₂DS₂-VASc scores were 3, 6, and 7, respectively.

Discrimination of Bleeding Risk Scores

In the overall cohort, the HAS-BLED score had a C-statistic of 0.512 (95% CI: 0.434-0.591) for predicting overall bleeding. The BleedMAP score showed moderate predictive ability, with a C-statistic of 0.581 (95% CI: 0.531-0.631). The difference between the two scores was not statistically significant (P = .13) (Figure 2). In the multiple imputation datasets, the HAS-BLED score demonstrated a C-statistic of 0.526 (95% CI: 0.479-0.573), while the BleedMAP score had a C-statistic of 0.590 (95% CI: 0.543-0.637), consistent with the findings observed in the original dataset.

Figure 2.

Receiver operating curve of the HAS-BLED and BleedMAP risk classification in predicting bleeding among the overall study cohort.

The sensitivity, specificity, PPV, and NPV for the HAS-BLED and BleedMAP scores are presented in Table 3.

Table 3.

Sensitivity, Specificity, PPV, and NPV of the HAS-BLED and BleedMAP Scores in Predicting Overall Bleeding Following non-Cardiac Surgery in Anticoagulated Patients.

Risk scores	HAS-BLED	BleedMAP
Sensitivity (95% CI)	94.4 (91.3-97.6)	31.7 (27.7-35.7)
Specificity (95% CI)	11.9 (7.5-16.3)	82.2 (78.9-85.5)
PPV (95% CI)	45.0 (38.2-51.7)	56.5 (52.2-60.7)
NPV (95% CI)	73.7 (67.7-79.7)	62.3 (58.1-66.4)

Abbreviations: CI, confidence intervals; NPV, negative predictive value; PPV, positive predictive value.

As a secondary objective, we assessed the performance of both scores according to the type of oral anticoagulant used. For the HAS-BLED score, the C-statistic for predicting overall bleeding was 0.483 (95% CI: 0.388-0.577) in patients receiving warfarin, and 0.636 (95% CI: 0.394-0.879) in those receiving DOACs. For the BleedMAP score, the corresponding C-statistics were 0.579 (95% CI: 0.524-0.633) for warfarin, and 0.656 (95% CI: 0.452-0.860) for DOACs.

Risk Factors for Bleeding

In the univariable analysis, eight factors were identified as being associated with bleeding (P < .1). In the multivariable analysis, five factors were independently associated with overall bleeding (P < .05). These included a history of bleeding (OR: 1.74; 95% CI: 1.22–2.48; P = .002), prior use of oral antiplatelet therapy (OR: 2.12; 95% CI: 1.23–3.67; P = .007), having a mitral mechanical heart valve (OR: 2.13; 95% CI: 1.11–4.10; P = .024), undergoing a ‘low-to-moderate bleeding risk’ procedure (OR: 1.92; 95% CI: 1.08–3.41; P = .026), and undergoing a ‘high bleeding risk’ procedure (OR: 2.55; 95% CI: 1.39–4.68; P = .002). The results of both univariable and multivariable analyses are presented in Table 4.

Table 4.

Risk Factors for Overall Bleeding: Univariate and Multivariate Analyses.

Characteristics	Univariate Model		Multivariate Model
Characteristics	Odds Ratio (95%CI)	P-Value	Odds Ratio (95%CI)	P-Value
Abnormal liver function	0.62 (0.24-1.55)	.301
Abnormal renal function	0.96 (0.53-1.72)	.884
Active cancer	1.26 (0.84-1.87)	.264
Age >65 years	0.74 (0.53-1.04)	.085*
Bridging therapy: before or after the procedure	1.28 (0.82-2.00)	.283
Bridging therapy: before and after the procedure	1.88 (1.29-2.73)	<.001*
History of bleeding	1.82 (1.30-2.54)	<.001*	1.74 (1.22-2.48)	.002
History of alcohol use	0.56 (0.25-1.23)	.149
Labile INRs	0.95 (0.62-1.48)	.832
Mitral mechanical heart valve	2.35 (1.36-4.06)	.002*	2.13 (1.11-4.10)	.024
Prior use of NSAIDs	0.88 (0.38-2.04)	.765
Prior use of oral antiplatelet	1.90 (1.14-3.19)	.014*	2.12 (1.23-3.67)	.007
Procedural bleeding risk: Low-moderate	2.39 (1.40-4.07)	.001*	1.92 (1.08-3.41)	.026
Procedural bleeding risk: High	3.18 (1.85-5.48)	<.001*	2.55 (1.39-4.68)	.002
Restart the anticoagulant therapy within 24 h	0.58 (0.37-0.94)	.026*
Stroke	1.10 (0.74-1.65)	.641
Systolic blood pressure >160 mm Hg	1.19 (0.79-1.79)	.417
Thrombocytopenia	1.18 (0.68-2.03)	.554

* P-value <.1.

Abbreviations: INR, International Normalized Ratio; NSAIDs, Non-Steroidal Anti-Inflammatory Drugs.

Discussion

The main principal findings of this study indicate that the HAS-BLED score demonstrated limited predictive performance for post-procedural bleeding in anticoagulated patients. On the other hand, the BleedMAP score showed moderate predictive performance. The BleedMAP score exhibited a slightly higher discriminatory ability than the HAS-BLED score. This may be attributable to the similarity between our study population and the BleedMAP development cohort, both of which included patients receiving anticoagulants for various indications. However, this difference was not statistically significant (P = .13). This may be because the high prevalence of AF (68%) in our cohort, the population for which the HAS-BLED score was initially developed, may have preserved the predictive performance of the HAS-BLED score despite its limitations. The optimal cutoff point for the BleedMAP score, determined using Youden's index, was ≥2, with a sensitivity of 31.7% (95% CI: 27.7-35.7), specificity of 82.2% (95% CI: 78.9-85.5), PPV of 56.5% (95% CI: 52.2-60.7), and NPV of 62.3% (95% CI: 58.1-66.4). The HAS-BLED score was not good enough for discriminatory ability in predicting post-procedural bleeding in anticoagulated patients, with a low C-statistic of 0.51 (95% CI: 0.43-0.59). The study also evaluated the optimal cutoff point for the HAS-BLED score in predicting post-procedural bleeding. The study also evaluated the optimal cutoff point for the HAS-BLED score in predicting post-procedural bleeding. ROC curve analysis identified a cutoff of ≥1, yielding a sensitivity of 94.4% (95% CI: 91.3-97.6), specificity of 11.9% (95% CI: 7.5-16.3), PPV of 45.0% (95% CI: 38.2-51.7), and NPV of 73.7% (95% CI: 67.7-79.7). This high NPV indicates potential utility in identifying patients at low risk of bleeding.

The HAS-BLED score was mainly focused on predict major bleeding in patients with AF.¹² Santise et al extended its application to population undergoing cardiac surgery, finding that the score was associated with an increased risk of major bleeding events, which were documented in 5.4% of cases.²¹ Relative to these findings, the incidence of major bleeding events in our study was lower (2.5%). This inconsistency may be attributed to differences in procedural bleeding risk: patients in our cohort primarily underwent low- to moderate-risk procedures, whereas cardiac surgeries were excluded in our study. Additionally, operational definitions of bleeding events varied across studies. However, the HAS-BLED score has not been widely validated for non-cardiac surgery. Kuo et al.²² reported that in patients undergoing transurethral resection of the prostate, a high-risk HAS-BLED category demonstrated significant predictive value for clinically significant hematuria (C-statistic 0.62). For tooth extraction procedures, however, the findings have been inconsistent. Iwata et al.²³ found that HAS-BLED scores >3 were significantly associated with post-extraction bleeding in patients receiving warfarin, while Kataoka et al.²⁴ concluded that the HAS-BLED score was insufficient for predicting such bleeding in similar patients. As our study included procedures with varying bleeding risks, the HAS-BLED score may have had limited discriminatory ability in predicting bleeding.

We found a major bleeding rate of 2.5% in our cohort, which was approximately similar to those reported in the original BleedMAP cohort of 2% and 5% for major and overall bleeding, respectively. However, the score has not been validated.¹⁴

Among patients in this cohort, anticoagulation therapy had included both vitamin K antagonists and DOACs. Specifically, 84.9% of patients had received warfarin, while 5.6% had been treated with DOACs. Among those who had received warfarin, bleeding events occurred in 40% of patients, with a time-in-therapeutic range of 58%. In the DOACs group, 39% experienced bleeding events. No patients received an inappropriate renal dose; however, two patients were prescribed a dose lower than recommended for their renal function. The performance of the HAS-BLED score across the two anticoagulant groups was consistent with that observed in our overall cohort. The C-statistic was 0.483 (95% CI: 0.388-0.577) in warfarin-treated patients and 0.636 (95% CI: 0.394-0.879) in those who had received DOACs. In contrast, the BleedMAP score demonstrated variability in predictive performance between the two groups. It showed moderate predictive ability for bleeding in warfarin-treated patients, with a C-statistic of 0.579 (95% CI: 0.524-0.633), while in patients who had received DOACs, the C-statistic was 0.656 (95% CI: 0.452-0.860). This discrepancy may be attributed to the BleedMAP score being initially developed using data from patients treated with warfarin.¹⁴ Moreover, the number of patients who received DOACs was relatively small.

The rate of bridging therapy with parenteral anticoagulants in real practice of our cohort may be considered high, as bridging was over prescribed in patients who had significant bleeding risk. Current guidelines therefore recommend against the routine use of bridging therapy and suggest its consideration only in selected patients at high risk of thromboembolism without significant bleeding risk.^4,7

A history of bleeding was one of the common factors included in the calculation of both the HAS-BLED¹² and BleedMAP¹⁴ scores to predict bleeding. Similarly, our study found that this factor was associated with post-procedural bleeding in the multivariable analysis. Additionally, prior use of oral antiplatelet agents has been identified as a risk factor for bleeding and is a component of the HAS-BLED score.¹² As antiplatelet agents inhibit platelet aggregation, patients receiving these medications have an increased risk of bleeding.

Our study found that a mechanical mitral heart valve is a predictor of post-procedural bleeding complications, consistent with the findings of Tafur et al.¹⁴ Patients with mechanical mitral valves are at a high risk of thromboembolism, requiring higher target INR levels than those with mechanical valves in other positions. Additionally, they often require bridging anticoagulation during warfarin interruption,¹⁵ which further predisposes them to a higher risk of bleeding.

In addition to patient-related risk factors, the type of procedure affects post-procedural bleeding. Our study found that patients undergoing high-bleeding-risk and low- to moderate-bleeding-risk procedures had a greater risk of bleeding than those undergoing minor-bleeding-risk procedures. Consistently, Clark et al.⁵ demonstrated that anticoagulated patients undergoing high-bleeding-risk procedures have an increased risk of major bleeding relative to those undergoing low-bleeding-risk procedures.

The HAS-BLED and BleedMAP scores may assist healthcare providers in the pre-procedural assessment of bleeding risk, specifically among anticoagulated patients undergoing non-cardiac surgery. These tools could support perioperative decision-making by identifying patients at increased risk of postoperative bleeding, especially those receiving warfarin. The HAS-BLED score demonstrated a high NPV, indicating that patients with a score of 0 are at low risk of bleeding and may not require intensive monitoring. In contrast, the BleedMAP score showed moderate predictive performance, with patients scoring ≥2 being at increased risk of bleeding. These findings suggest that both scores may help inform individualized bleeding risk mitigation strategies in the perioperative setting. Furthermore, reducing bleeding complications may avoid additional costs related to prolonged hospitalization and corrective therapies.

Our study has several limitations. First, as a retrospective study, some laboratory data to fill in the calculator were missing, which limited the analysis of certain patients. Among those with missing data, most did not experience bleeding. However, we applied multiple imputation to manage the missing data, and the results remained consistent. In addition, there were limitations in classifying bleeding events. Nonetheless, we used the ISTH criteria for surgical patients to categorize the severity of bleeding. Second, despite collecting data over an extended period, our study population included a relatively small number of anticoagulated patients. However, the higher-than-anticipated prevalence of postoperative bleeding made the 591-patient sample sufficient to support the study conclusions. Third, the development of modern surgical techniques may affect the bleeding outcomes. Fourth, this study was conducted at KCMH, a single-center and a teaching hospital, where potential confounding factors—such as surgeon experience, training status, and variations in bridging regimens—may have influenced the outcomes. Fifth, this study focused exclusively on postoperative bleeding, thereby excluding intraoperative bleeding events from the analysis. This distinction may limit the generalizability of our findings to overall perioperative bleeding risk. Therefore, the variable performance and limited discriminatory ability of these risk scores warrant evaluation in prospective studies. Lastly, the protocol recorded only the first post-procedural bleeding episode, potentially underestimating cumulative bleeding events and limiting analysis of re-bleeding patterns. Future studies should prospectively capture all bleeding episodes to better characterize recurrence risk.

Conclusions

The HAS-BLED score demonstrated limited predictive performance for postoperative bleeding in anticoagulated patients undergoing elective non-cardiac surgery. Nevertheless, its NPV was high. There was no statistically significant difference in discriminatory ability between the HAS-BLED and BleedMAP scores. Additionally, procedural bleeding risk—not currently included in either score—emerged as an independent and potentially valuable predictor of postoperative bleeding.

Supplemental Material

sj-docx-1-cat-10.1177_10760296251415186 - Supplemental material for Accuracy of HAS-BLED and BleedMAP Scores in Predicting Postoperative Bleeding in Anticoagulated Patients Undergoing non-Cardiac Surgery

Supplemental material, sj-docx-1-cat-10.1177_10760296251415186 for Accuracy of HAS-BLED and BleedMAP Scores in Predicting Postoperative Bleeding in Anticoagulated Patients Undergoing non-Cardiac Surgery by Thanaporn Poohirunkul, PharmD, Yotsaya Kunlamas, PharmD, Sarawut Siwamogsatham, MD, and Krittin Bunditanukul, PhD in Clinical and Applied Thrombosis/Hemostasis

Footnotes

Acknowledgments

We would like to thank King Chulalongkorn Memorial Hospital, the Faculty of Pharmaceutical Sciences, and the Faculty of Medicine, Chulalongkorn University for providing essential information. We also extend our thanks to Dr Thanapoom Rattananupong for his expert guidance in statistical analysis.

ORCID iDs

Thanaporn Poohirunkul

Yotsaya Kunlamas

Sarawut Siwamogsatham

Krittin Bunditanukul

Ethical Approval and Informed Consent Statements

Ethical approval for this study was obtained from the Institutional Review Board of King Chulalongkorn Memorial Hospital and the Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand, on 20 August 2021. The requirement for informed consent was waived due to the retrospective nature of the study (Reference No. 664/64).

Author Contributions

TP, SS, and KB designed the study. TP collected data. TP, YK, and KB performed the statistical analysis. TP wrote the first draft of manuscript. All authors discuss data, revised the manuscript, and approved the final version for publication.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Ajarn Kasem Pangsriwong Foundation,

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.

Supplemental Material

Supplemental material for this article is available online.

References

Doherty

Gluckman

Hucker

, et al. 2017 ACC expert consensus decision pathway for periprocedural management of anticoagulation in patients with nonvalvular atrial fibrillation: A report of the American College of Cardiology clinical expert consensus document task force. J Am Coll Cardiol. Feb 21 2017;69(7):871–898. doi:10.1016/j.jacc.2016.11.024.

Otto

Nishimura

Bonow

, et al. 2020 ACC/AHA guideline for the management of patients with valvular heart disease: A report of the American College of Cardiology/American Heart Association joint committee on clinical practice guidelines. Circulation. Feb 2 2021;143(5):e72–e227. doi:10.1161/cir.0000000000000923.

Becattini

Agnelli

. Treatment of venous thromboembolism with new anticoagulant agents. J Am Coll Cardiol. Apr 26 2016;67(16):1941–1955. doi:10.1016/j.jacc.2016.01.072.

Douketis

Spyropoulos

Murad

, et al. Perioperative management of antithrombotic therapy: An American college of chest physicians clinical practice guideline. Chest. Nov 2022;162(5):e207–e243. doi:10.1016/j.chest.2022.07.025.

Clark

Douketis

Hasselblad

Schulman

Kindzelski

Ortel

. Predictors of perioperative major bleeding in patients who interrupt warfarin for an elective surgery or procedure: Analysis of the BRIDGE trial. Am Heart J. Jan 2018;195:108–114. doi:10.1016/j.ahj.2017.09.015.

Spyropoulos

Al-Badri

Sherwood

Douketis

. Periprocedural management of patients receiving a vitamin K antagonist or a direct oral anticoagulant requiring an elective procedure or surgery. J Thromb Haemost. May 2016;14(5):875–885. doi:10.1111/jth.13305.

Thompson

Fleischmann

Smilowitz

, et al. 2024 AHA/ACC/ACS/ASNC/HRS/SCA/SCCT/SCMR/SVM guideline for perioperative cardiovascular management for noncardiac surgery: A report of the American College of Cardiology/American Heart Association joint committee on clinical practice guidelines. J Am Coll Cardiol. Nov 5 2024;84(19):1869–1969. doi:10.1016/j.jacc.2024.06.013.

Schulman

Angeras

Bergqvist

Eriksson

Lassen

Fisher

. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in surgical patients. J Thromb Haemost. Jan 2010;8(1):202–204. doi:10.1111/j.1538-7836.2009.03678.x.

Johnston

Afolabi

Tewari

Danker

. Clinical and economic burden associated with disruptive surgical bleeding: A retrospective database analysis. Clinicoecon Outcomes Res. 2023;15:535–547. doi:10.2147/ceor.S411778.

10.

Stokes

Shah

, et al. Impact of bleeding-related complications and/or blood product transfusions on hospital costs in inpatient surgical patients. BMC Health Serv Res. May 31 2011;11(1):135. doi:10.1186/1472-6963-11-135.

11.

Kirchhof

Benussi

Kotecha

, et al. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Eur Heart J. Oct 7 2016;37(38):2893–2962. doi:10.1093/eurheartj/ehw210

12.

Pisters

Lane

Nieuwlaat

de Vos

Crijns

Lip

. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: The euro heart survey. Chest. Nov 2010;138(5):1093–1100. doi:10.1378/chest.10-0134.

13.

Lip

Andreotti

Fauchier

, et al. Bleeding risk assessment and management in atrial fibrillation patients. Executive summary of a position document from the European heart rhythm association [EHRA], endorsed by the European Society of Cardiology [ESC] working group on thrombosis. Thromb Haemost. Dec 2011;106(6):997–1011. doi:10.1160/th11-10-0690.

14.

Tafur

McBane

2nd Wysokinski

, et al. Predictors of major bleeding in peri-procedural anticoagulation management. J Thromb Haemost. Feb 2012;10(2):261–267. doi:10.1111/j.1538-7836.2011.04572.x.

15.

Douketis

Spyropoulos

Spencer

, 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. Feb. 2012;141(2 Suppl):e326S–e350S. doi:10.1378/chest.11-2298.

16.

Heidbuchel

Verhamme

Alings

, et al. Updated European heart rhythm association practical guide on the use of non-vitamin K antagonist anticoagulants in patients with non-valvular atrial fibrillation. Europace. Oct 2015;17(10):1467–1507. doi:10.1093/europace/euv309.

17.

Patterns of Ambulatory Bridging of Anticoagulant among Patients Undergoing Planned Surgery in King Chulalongkorn Memorial Hospital (KCMH).

18.

Akobeng

. Understanding diagnostic tests 3: Receiver operating characteristic curves. Acta Paediatr. May 2007;96(5):644–647. doi:10.1111/j.1651-2227.2006.00178.x.

19.

DeLong

Clarke-Pearson

. Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics. Sep 1988;44(3):837–845.

20.

Spyropoulos

Brohi

Caprini

, et al. Scientific and standardization committee communication: Guidance document on the periprocedural management of patients on chronic oral anticoagulant therapy: Recommendations for standardized reporting of procedural/surgical bleed risk and patient-specific thromboembolic risk. J Thromb Haemost. Nov 2019;17(11):1966–1972. doi:10.1111/jth.14598.

21.

Santise

Nardella

Migliano

Testa

Maselli

. The HAS-BLED score is associated with Major bleeding in patients after cardiac surgery. J Cardiothorac Vasc Anesth. Jun. 2019;33(6):1601–1606. doi:10.1053/j.jvca.2019.01.021.

22.

Kuo

Lim

ZLT

Letch

Silverman

Kim

JJY

McClintock

. Evaluating the utility of the HAS-BLED bleeding-estimator tool for transurethral resection of prostate. BJUI Compass. Jan 2025;6(1):e480. doi:10.1002/bco2.480.

23.

Iwata

Tachibana

Kusumoto

, et al. Risk factors associated with post-extraction bleeding in patients on warfarin or direct-acting oral anticoagulants: A retrospective cohort study. Oral Maxillofac Surg. Dec 2022;26(4):641–648. doi:10.1007/s10006-022-01039-0.

24.

Kataoka

Hoshi

Ando

. Is the HAS-BLED score useful in predicting post-extraction bleeding in patients taking warfarin? A retrospective cohort study. BMJ Open. Mar 2 2016;6(3):e010471. doi:10.1136/bmjopen-2015-010471.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB