Abstract
Introduction:
This study aimed to assess the relationship between preoperative international normalized ratio (INR) levels and major postoperative bleeding events after total shoulder arthroplasty (TSA).
Methods:
The American College of Surgeons National Surgical Quality Improvement Program database was queried for TSA from 2011 to 2020. A final cohort of 2405 patients with INR within 2 days of surgery were included. Patients were stratified into four groups: INR ≤ 1.0, 1.0 < INR ≤ 1.25, 1.25< INR ≤ 1.5, and INR > 1.5. The primary outcome was bleeding requiring transfusion within 72 hours, and secondary outcome variables included complication, revision surgery, readmission, and hospital stay duration. Multivariable logistic and linear regression analyses adjusted for relevant comorbidities were done.
Results:
Of the 2,405 patients, 48% had INR ≤ 1.0, 44% had INR > 1.0 to 1.25, 7% had INR > 1.25 to 1.5, and 1% had INR > 1.5. In the adjusted model, 1.0 < INR ≤ 1.25 (OR 1.7, 95% CI 1.176 to 2.459), 1.25 < INR ≤ 1.5 (OR 2.508, 95% CI 1.454 to 4.325), and INR > 1.5 (OR 3.200, 95% CI 1.233 to 8.302) were associated with higher risks of bleeding compared with INR ≤ 1.0.
Discussion:
The risks of thromboembolism and bleeding lie along a continuum, with higher preoperative INR levels conferring higher postoperative bleeding risks after TSA. Clinicians should use a patient-centered, multidisciplinary approach to balance competing risks.
Total shoulder arthroplasty (TSA), including both anatomic and reverse shoulder arthroplasty, is an increasingly used surgical treatment. More than 800,000 people in the United States are estimated to be living with a type of shoulder arthroplasty.1,2 Approximately 31.8 per 100,000 people each year undergo TSA. The recent rise in TSA utilization is likely due in part to the expanding indications for the reverse prosthesis; initially devised for rotator cuff arthropathy, indications for reverse TSA now include some cases of osteoarthritis, proximal humerus fractures, and irreparable rotator cuff tears.3
More than six million people in the United States require chronic anticoagulation therapy for the prevention of thromboembolism, and these therapies are frequently interrupted for major invasive procedures.4,5 One measure of the coagulation pathway is the international normalized ratio (INR), which is calculated as the ratio of the patient's prothrombin time divided by the mean normal laboratory prothrombin time.6 Patients undergoing total joint arthroplasty have high bleeding risk according to the American Association of Orthopaedic Surgeons and the American College of Cardiology, and historically, preoperative INR ≤ 1.5 has been used as a threshold for these major surgeries.4,7 However, the safe INR threshold for orthopaedic procedures has been recently called into question, with some studies suggesting no increased risk of bleeding with preoperative INR levels above the historical cutoff,8-14 and others suggesting increased risk of bleeding and mortality at even lower INR levels.15-17 Only one previous study has investigated the association between preoperative INR levels and episode-of-care adverse events after TSA, and the optimal preoperative threshold INR level for TSA has not been well established.18
The primary objective of this study was to assess whether preoperative INR levels were associated with major postoperative bleeding events after TSA using a large national database over a recent 10-year period. Secondary objectives of this study were to assess for associations between preoperative INR levels and 30-day postoperative complications, revision surgery, readmission, and hospital length of stay after TSA.
Methods
Study Design and Cohort
A retrospective database study was done to determine the effect of preoperative INR on 30-day perioperative outcomes after TSA. Patients were identified using the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database. This database includes patients who undergo major surgical procedures at approximately 700 participating hospitals. Preoperative through 30-day postoperative data, including demographics, comorbidities, hospital length of stay, complications, readmission, and revision surgery, are collected in this database. The data are collected through medical record review and verified at the end of the 30-day postsurgery timeframe by telephone or written patient survey. Data accuracy is ensured by random audits and clinical reviews.19
The NSQIP database was queried for patients who underwent TSA from January 1, 2011, to December 31, 2020, by querying the current procedural terminology code 23472 (arthroplasty, glenohumeral joint; total shoulder [glenoid and proximal humeral replacement]). This procedural code is common to both anatomic and reverse shoulder arthroplasty, and therefore, the data were not sensitive to differences between these two types of shoulder arthroplasties; this procedural code includes primary rather than revision arthroplasty.
The initial database query yielded 33,489 patients who underwent TSA during the study period. Only patients with preoperative INR values within 2 days of the TSA were included in the study as done by Rudasill et al,17 and 31,084 patients were excluded for this reason. After exclusions, the study cohort consisted of 2,405 patients who underwent TSA.
Outcome Variables
Our primary outcome variable was postoperative bleeding requiring blood transfusion within 72 hours of surgery. Our secondary outcome variables included measured 30-day postoperative complications within the NSQIP database, specifically, superficial surgical site infection, deep surgical site infection, organ/space surgical site infection, wound dehiscence, pneumonia, unplanned reintubation, pulmonary embolism, persistent ventilator requirement, renal failure, urinary tract infection, stroke, cardiac arrest, myocardial infarction, blood transfusion, deep vein thrombosis, and sepsis/septic shock. Detailed definitions for each complication are available from NSQIP.19 In addition, our secondary outcome variables included a composite variable for 30-day postoperative complication, scored positively if a patient had one or more complications as measured by the NSQIP database, as well as a composite variable for 30-day infectious complication, scored positively if a patient had one or more of superficial surgical site infection, deep surgical site infection, organ/space surgical site infection, and wound dehiscence. Finally, our secondary outcome variables also included hospital length of stay, 30-day revision surgery, and 30-day hospital readmission.
Explanatory Variables
Our explanatory variables included INR within 2 days of TSA, stratified into four groups as per the work of Rudasill et al and Sivasundaram et al: (1) INR ≤ 1.0, (2) 1.0 < INR ≤ 1.25, (3) 1.25 < INR ≤ 1.5, and (4) INR > 1.5.16-18 Moreover, the following patient-related variables were collected: age, sex, race, BMI, diabetes mellitus, current smoking status, functional status (independent, partial dependent, or totally dependent), chronic obstructive pulmonary disease, congestive heart failure, hypertension requiring medication, disseminated cancer, chronic steroid or immunosuppressive therapy, bleeding disorder, and American Society of Anesthesiologists (ASA) classification.
Statistical Analysis
Descriptive statistics were calculated for the final cohort. All variables had greater than 98% complete data, except race, which had 87% complete data. Statistical analyses were done using complete data sets only, and missing data are given in Table 1. The Kruskal-Wallis tests were used for continuous explanatory variables and the chi-squared test for categorical explanatory variables. Unadjusted simple logistic regression was used to assess the effect of INR level on the risk of dichotomous outcomes such as bleeding, complication, and readmission in the 30-day postoperative period. Then, multivariable logistic regression was used to assess the effect of INR level adjusted for sex, age, diabetes mellitus, smoking, and ASA classification on the risks of these outcomes. Unadjusted simple linear regression was used to assess the effect of INR level on hospital length of stay. Then, multivariable linear regression was used to assess the effect of INR level adjusted for sex, age, diabetes mellitus, smoking, and ASA classification on hospital stay duration. α = 0.05 was used as the criterion for statistical significance. Data curation was done with R, and statistical tests were done with SAS.
Table 1.
Baseline Patient Characteristics of the Study Cohort Stratified by the INR Level (n = 2,405)
| Variable | INR Level | P | |||
| INR ≤ 1.0 (n = 1,147) | 1.0 < INR ≤ 1.25 (n = 1,047) | 1.25 < INR ≤ 1.5 (n = 176) | INR > 1.5 (n = 35) | ||
| Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | ||
| Age (years)a | 70.1 (9.6) | 72.5 (9.2) | 74.8 (8.2) | 73.3 (8.0) | <0.0001 |
| BMI | 30.5 (7.6) | 31.3 (7.7) | 31.3 (7.9) | 30.0 (9.5) | 0.1 |
| n (%) | n (%) | n (%) | n (%) | ||
| Female sex | 744 (64.9) | 556 (53.1) | 102 (58.0) | 22 (62.9) | <0.0001 |
| Racea | 0.6 | ||||
| White | 895 (91.8) | 874 (93.2) | 148 (93.7) | 28 (90.3) | |
| Black or African American | 46 (4.7) | 46 (4.9) | 8 (5.1) | 2 (6.5) | |
| Native American | 16 (1.6) | 10 (1.1) | 2 (1.3) | 1 (3.2) | |
| Asian | 17 (1.7) | 6 (0.6) | 0 (0) | 0 (0) | |
| Pacific Islander | 1 (0.1) | 2 (0.2) | 0 (0) | 0 (0) | |
| ASA classificationa | <0.0001 | ||||
| 1 | 13 (1.1) | 10 (1.0) | 0 (0) | 0 (0) | |
| 2 | 386 (33.7) | 224 (21.4) | 20 (11.4) | 5 (14.3) | |
| 3 | 697 (60.8) | 728 (69.6) | 134 (76.1) | 22 (62.9) | |
| 4 | 51 (4.5) | 83 (7.9) | 22 (12.5) | 8 (22.9) | |
| 5 | 0 (0) | 1 (0.1) | 0 (0) | 0 (0) | |
| Diabetes mellitus | 235 (20.5) | 230 (22.0) | 34 (19.3) | 11 (31.4) | 0.4 |
| Current smoker | 135 (11.8) | 84 (8.0) | 10 (5.7) | 5 (14.3) | 0.005 |
| COPD | 104 (9.1) | 78 (7.5) | 19 (10.8) | 4 (11.4) | 0.3 |
| CHF | 8 (0.7) | 27 (2.6) | 5 (2.8) | 2 (5.7) | 0.001 |
| Hypertension | 771 (67.2) | 779 (74.4) | 137 (77.8) | 28 (80) | 0.0002 |
| Disseminated cancer | 7 (0.6) | 7 (0.7) | 3 (1.7) | 0 (0) | 0.4 |
| Steroids/immunosuppression | 65 (5.7) | 56 (5.4) | 6 (3.4) | 2 (5.7) | 0.7 |
| Bleeding disorder | 80 (7.0) | 131 (12.5) | 39 (22.2) | 13 (37.1) | <0.0001 |
| Functional statusa | 0.07 | ||||
| Independent | 1102 (96.7) | 976 (94.0) | 164 (94.3) | 34 (100) | |
| Partially dependent | 35 (3.1) | 57 (5.5) | 10 (5.8) | 0 (0) | |
| Totally dependent | 3 (0.3) | 5 (0.5) | 0 (0) | 0 (0) | |
COPD = chronic obstructive pulmonary disease, CHF = congestive heart failure, INR = international normalized ratio
Complete data sets were used for statistical analyses. Data were partially missing for the variables of race (n = 303), age (n = 42), functional status (n = 19), and ASA classification (n = 1).
Italics denote statistical significance.
Results
Cohort Characteristics
A total of 2,405 patients who underwent TSA with preoperative INR values within 2 days of surgery were included in the study cohort. The cohort had an average age of 71.5 years, and 40.8% of patients were male sex. The cohort exhibited average BMI of 30.9. Most of the patients were ASA 3 classification (65.8%), followed by ASA 2 classification (26.4%).
When stratified by INR into four groups (INR ≤ 1.0, 1.0 < INR ≤ 1.25, 1.25 < INR ≤ 1.5, and INR > 1.5), the groups were significantly different in age, sex, ASA classification, smoking status, comorbid congestive heart failure, comorbid hypertension, and comorbid bleeding disorder (Table 1). Mean preoperative INR was similar from 2011 to 2020 (Figure 1).
Figure 1.
Diagram showing the yearly mean preoperative international normalized ratio values from 2011 to 2020.
Bleeding Requiring Transfusion
The risk of postoperative bleeding requiring blood transfusion within 72 hours of surgery significantly increased with increasing INR levels (P < 0.0001) (Table 2). In the unadjusted logistic regression analysis, when compared against patients with INR ≤ 1.0 as the reference, patients with 1.0 < INR ≤ 1.25 had 1.9 times the odds of bleeding, patients with 1.25 < INR ≤ 1.5 had 3.2 times the odds of bleeding, and patients with INR > 1.5 had 4.2 times the odds of bleeding (Table 3). In the multivariable logistic regression analysis adjusted for sex, age, diabetes mellitus, smoking, and ASA classification, when compared against patients with INR ≤ 1.0 as the reference, patients with 1.0 < INR ≤ 1.25 had 1.7 times the odds of bleeding, patients with 1.25 < INR ≤ 1.5 had 2.5 times the odds of bleeding, and patients with INR > 1.5 had 3.2 times the odds of bleeding. Older age was associated with increased odds of bleeding (OR 1.028, 95% CI 1.007 to 1.049), and male sex was associated with decreased odds of bleeding (OR 0.583, 95% CI 0.409 to 0.830) (Table 4).
Table 2.
Tabulated 30-day Postoperative Complications, Reoperations, Readmissions, and Hospital Lengths of Stay Stratified by the INR Level (n = 2405)
| Outcome | INR Level | P | |||
| INR ≤ 1.0 (n = 1147) | 1.0 < INR ≤ 1.25 (n = 1047) | 1.25 < INR ≤ 1.5 (n = 176) | INR > 1.5 (n = 35) | ||
| n (%) | n (%) | n (%) | n (%) | ||
| Bleeding requiring transfusion | 54 (4.7) | 88 (8.4) | 24 (13.6) | 6 (17.1) | <0.0001 |
| Complicationa | 86 (7.5) | 122 (11.7) | 33 (18.8) | 7 (20.0) | <0.0001 |
| Infectious complicationa | 7 (0.6) | 8 (0.8) | 1 (0.6) | 0 (0) | 0.9 |
| Superficial surgical site infection | 4 (0.4) | 3 (0.3) | 0 (0) | 0 (0) | 0.9 |
| Deep surgical site infection | 1 (0.1) | 0 (0) | 0 (0) | 0 (0) | 0.8 |
| Organ/space surgical site infection | 1 (0.1) | 5 (0.5) | 1 (0.6) | 0 (0) | 0.3 |
| Wound dehiscence | 1 (0.1) | 1 (0.1) | 0 (0) | 0 (0) | 0.9 |
| Pneumonia | 4 (0.4) | 13 (1.2) | 7 (4.0) | 2 (5.7) | <0.0001 |
| Unplanned reintubation | 4 (0.4) | 4 (0.4) | 3 (1.7) | 0 (0) | 0.09 |
| Pulmonary embolism | 4 (0.4) | 4 (0.4) | 0 (0) | 0 (0) | 0.9 |
| Ventilator requirement | 4 (0.4) | 1 (0.1) | 2 (1.1) | 0 (0) | 0.1 |
| Renal insufficiency | 2 (0.2) | 1 (0.1) | 1 (0.6) | 0 (0) | 0.6 |
| Urinary tract infection | 15 (1.3) | 18 (1.7) | 2 (1.1) | 0 (0) | 0.7 |
| Stroke | 1 (0.1) | 0 (0) | 2 (0.1) | 0 (0) | 0.001 |
| Cardiac arrest | 2 (0.2) | 1 (0.1) | 0 (0) | 0 (0) | 0.9 |
| Myocardial infarction | 4 (0.4) | 5 (0.5) | 1 (0.6) | 1 (2.9) | 0.2 |
| Deep vein thrombosis | 3 (0.3) | 5 (0.5) | 2 (1.1) | 0 (0) | 0.4 |
| Sepsis/septic shock | 4 (0.4) | 3 (0.3) | 0 (0) | 0 (0) | 0.9 |
| Revision Surgeryb | 19 (1.7) | 27 (2.6) | 3 (1.7) | 1 (2.9) | 0.5 |
| Readmissionb | 48 (4.2) | 59 (5.7) | 17 (9.7) | 1 (2.9) | 0.02 |
| Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | ||
| Hospital length of stayb | 2.4 (2.3) | 2.8 (3.0) | 3.0 (3.2) | 5.5 (7.7) | 0.003 |
INR = international normalized ratio
The figures shown denote the number of unique patients with one or more of the class of complications in the 30-day postoperative period. Patients may have had more than one of such complications, counted only once for analytical statistics to maintain the assumption of independent events.
Complete data sets were used for statistical analyses. Data were partially missing for the outcomes of revision surgery (n = 2), readmission (n = 10), and hospital length of stay (n = 2).
Italics denote statistical significance.
Table 3.
Simple Logistic Regression Analysis on the Effect of INR Level on the Risk of Bleeding, Complication, and Readmission in the 30-day Postoperative Period
| Variable | Bleeding requiring transfusion | Complication | Readmission | |||
| OR (95% CI) | P | OR (95% CI) | P | OR (95% CI) | P | |
| INR level | ||||||
| INR ≤ 1.0 | Reference | Reference | Reference | |||
| 1.0 < INR ≤ 1.25 | 1.857 (1.309, 2.635) | 0.0005 | 1.627 (1.218, 2.174) | 0.001 | 1.361 (0.921, 2.012) | 0.1 |
| 1.25 < INR ≤ 1.5 | 3.196 (1.919, 5.322) | <0.0001 | 2.847 (1.838, 4.411) | <0.0001 | 2.448 (1.374, 4.363) | 0.002 |
| INR > 1.5 | 4.188 (1.668, 10.514) | 0.002 | 3.085 (1.309, 7.267) | 0.01 | 0.669 (0.090, 4.991) | 0.7 |
OR = odds ratio, CI = confidence interval, INR = international normalized ratio
Italics denote statistical significance.
Table 4.
Multivariable Logistic Regression Analysis on the Effect of INR Level Adjusted for Sex, Age, Diabetes Mellitus, Smoking, and ASA Classification on the Risk of Bleeding, Complication, and Readmission in the 30-day Postoperative Period
| Variable | Bleeding requiring transfusion | Complication | Readmission | |||
| OR (95% CI) | P | OR (95% CI) | P | OR (95% CI) | P | |
| INR level | ||||||
| INR ≤ 1.0 | Reference | Reference | Reference | |||
| 1.0 < INR ≤ 1.25 | 1.700 (1.176, 2.459) | 0.005 | 1.491 (1.099, 2.023) | 0.01 | 1.354 (0.902, 2.032) | 0.1 |
| 1.25 < INR ≤ 1.5 | 2.508 (1.454, 4.325) | 0.0009 | 2.178 (1.364, 3.477) | 0.001 | 1.955 (1.048, 3.648) | 0.04 |
| INR > 1.5 | 3.200 (1.233, 8.302) | 0.02 | 1.977 (0.778, 5.019) | 0.2 | 0.549 (0.072, 4.172) | 0.5 |
| Male sex | 0.583 (0.409, 0.830) | 0.003 | 0.652 (0.486, 0.874) | 0.004 | 0.768 (0.522, 1.129) | 0.2 |
| Age | 1.028 (1.007, 1.049) | 0.008 | 1.026 (1.008, 1.043) | 0.004 | 0.989 (0.969, 1.010) | 0.3 |
| Diabetes mellitus | 1.405 (0.977, 2.022) | 0.07 | 1.292 (0.944, 1.768) | 0.1 | 0.764 (0.480, 1.219) | 0.3 |
| Current smoker | 1.560 (0.899, 2.708) | 0.1 | 1.369 (0.849, 2.209) | 0.2 | 1.265 (0.699, 2.290) | 0.4 |
| ASA classification | ||||||
| 1 | Reference | Reference | Reference | |||
| 2 | >999.999 (<0.001, >999.999) | 0.9 | >999.999 (<0.001, >999.999) | 0.9 | 0.422 (0.052, 3.396) | 0.4 |
| 3 | >999.999 (<0.001, >999.999) | 0.9 | >999.999 (<0.001, >999.999) | 0.9 | 1.242 (0.163, 9.446) | 0.8 |
| 4 | >999.999 (<0.001, >999.999) | 0.9 | >999.999 (<0.001, >999.999) | 0.9 | 2.330 (0.288, 18.845) | 0.4 |
OR = odds ratio, CI = confidence interval, INR = international normalized ratio
Reference group for age is minus 1 unit.
Italics denote statistical significance.
Complications, Revision Surgery, and Readmission
The increasing INR level was significantly associated with increased risk of postoperative complications (P < 0.0001), pneumonia (P < 0.0001), urinary tract infection (P = 0.001), and hospital readmission (P = 0.02). Revision surgery was not associated with the INR level (P = 0.5) (Table 2).
The unadjusted logistic regression analysis demonstrated a stepwise increase in the odds of complications with increasing INR levels. The risk of hospital readmission was higher for the 1.25 < INR ≤ 1.5 group compared with the INR ≤ 1.0 group (Table 3). In the multivariable logistic regression analysis adjusted for sex, age, diabetes mellitus, smoking, and ASA classification, when compared against patients with INR ≤ 1.0 as the reference, patients with 1.0 < INR ≤ 1.25 had 1.5 times the odds of complications and patients with 1.25 < INR ≤ 1.5 had 2.2 times the odds of complications. Older age was associated with increased odds of complications (OR 1.026, 95% CI 1.008 to 1.043), and male sex was associated with decreased odds of complications (OR 0.652, 95% CI 0.486 to 0.874). In the multivariable logistic regression analysis, patients with 1.0 < INR ≤ 1.25 compared with patients with INR ≤ 1.0 had 2.0 times the odds of hospital readmission (Table 4).
Hospital Stay Duration
Increasing INR levels were significantly associated with longer hospital lengths of stay in a stepwise manner (P = 0.003) (Table 2). The unadjusted linear regression analysis showed that each higher tier of INR level was associated with a half-day increase in total hospital length of stay (P < 0.0001) (Table 5). Higher INR level, older age, female sex, higher ASA classification, diabetes mellitus, and current smoking were associated with greater hospital length of stay in the multivariable linear regression analysis (Table 6).
Table 5.
Simple Linear Regression Analysis on the Effect of INR Level on Hospital Length of Stay
| Variable | Hospital length of stay | |
| β (95% CI) | P | |
| INR level | 0.511 (0.345, 0.677) | <0.0001 |
β = β regression coefficient, CI = confidence interval, INR = international normalized ratio
Italics denote statistical significance.
Table 6.
Multivariable Linear Regression Analysis on the Effect of INR Level Adjusted for Sex, Age, Diabetes Mellitus, Smoking, and ASA Classification on Hospital Length of Stay
| Variable | Hospital length of stay | |
| β (95% CI) | P | |
| INR level | 0.358 (0.195, 0.521) | <0.0001 |
| Male sex | −0.741 (−0.964, −0.519) | <0.0001 |
| Age | 0.026 (0.013, 0.038) | <0.0001 |
| Diabetes mellitus | 0.536 (0.268, 0.804) | <0.0001 |
| Current smoker | 0.513 (0.135, 0.891) | 0.008 |
| ASA classification | 0.565 (0.367, 0.764) | <0.0001 |
β = β regression coefficient, CI = confidence interval, INR = international normalized ratio
Italics denote statistical significance.
Discussion
TSA is a commonly performed major total joint arthroplasty.1-3 A substantial subset of the population receives chronic anticoagulation therapy for the prevention of thromboembolism,4,5 and INR levels are germane to the preoperative work-up for these patients. However, controversy remains regarding the optimal preoperative INR threshold for orthopaedic procedures.16,17 Tamim et al15 performed a large database study of 636,231 patients who underwent major surgery and demonstrated an increased risk of postoperative bleeding at INR levels above 1.1; however, approximately 90% of the surgeries in this study were nonorthopaedic procedures, with the majority being general surgical procedures. Rudasill et al performed large database studies of 21,239 patients who underwent total knee arthroplasty and 17,567 patients who underwent total hip arthroplasty, demonstrating increased bleeding risks at INR levels above 1.25.16,17 Conversely, multiple case series suggest that INR levels elevated above 1.5 may be safe for patients undergoing ambulatory procedures, such as knee arthroscopy, shoulder arthroscopy, and hand surgery.9,10,12 The effect of preoperative INR levels on postoperative bleeding after TSA has not been previously well described. In this study, we have demonstrated that increasing INR levels are associated with stepwise increasing risk of postoperative bleeding requiring transfusion within 72 hours of TSA, even when adjusted for relevant medical comorbidities. Even below the historical INR threshold of 1.5, we have demonstrated an association between INR levels and 30-day postoperative complications, 30-day hospital readmission, and hospital length of stay after TSA.
Acute blood loss anemia remains a relevant concern during and after TSA. Previous studies have demonstrated an average intraoperative blood loss of approximately 350 mL in TSA procedures.20,21 When accounting for postoperative blood loss, the overall blood loss after TSA is approximately 850 mL.21 Blood transfusions carry associated risks of allergic reaction, infection, cardiovascular dysfunction, and disease transmission.22 Our study demonstrates a direct relationship between increasing preoperative INR levels and the risk of postoperative bleeding requiring transfusion after TSA. The prevalence of postoperative bleeding requiring transfusion was 4.7% in patients with preoperative INR ≤ 1.0 compared with 17.1% in patients with preoperative INR > 1.5. Unsurprisingly, patients with preoperative INR greater than the historical threshold of 1.5 had more than 3 times the odds of postoperative bleeding compared with the reference group. However, it is interesting to note that even patients with mildly elevated preoperative INR levels, below the historical threshold, had markedly higher risk of postoperative bleeding. After adjusting for relevant comorbidities, patients with 1.0 < INR ≤ 1.25 had 1.7 times the odds of postoperative bleeding and patients with 1.25 < INR ≤ 1.5 had 2.5 times the odds of postoperative bleeding compared with the reference group. Our findings contrast with those of Sivasundaram et al,18 who reported on 1014 TSA procedures in the NSQIP database from 2006 to 2016, and found that only INR levels greater than 1.5 were associated with postoperative bleeding requiring transfusion and 30-day complications in their adjusted multivariate analysis. The difference in our findings may be due to our more recent study period extending to 2020 and our larger sample size of 2,405 patients.
In addition to our primary outcome, we have demonstrated that patients with mildly elevated preoperative INR levels are at higher risk for 30-day complications and readmission after TSA. After adjusting for relevant comorbidities, patients with 1.0 < INR ≤ 1.25 had 1.5 times the odds of all-cause complications and patients with 1.25 < INR ≤ 1.5 had 2.2 times the odds of all-cause complications compared with the reference group; patients with 1.25 < INR ≤ 1.5 had 2.0 times the odds of hospital readmission compared with the reference group. Moreover, higher preoperative INR levels were associated with longer lengths of stay. Preoperative INR > 1.5 was not associated with all-cause complications or readmission in our study, which we suspect was due to the small number of patients in this group limiting our power to detect a notable difference. Our findings support previous research that advanced age and female sex are associated with complications, specifically blood transfusions, after shoulder arthroplasty.22
Our findings must be interpreted in the context of several limitations, some of which are common to the use of a large, pooled database. First, conclusions able to be drawn from the retrospective analysis of a large national database are limited by the data available. We were able to study preoperative INR levels, but not patients' anticoagulation regimens and how these were managed perioperatively. Since patients were identified by query of the procedural code, we were not able to compare bleeding risks of anatomic TSA and reverse TSA. Moreover, we were not able to account for differences among preoperative diagnoses, such as osteoarthritis, rotator cuff arthropathy, or proximal humerus fracture, nor the effects of prior ipsilateral shoulder surgery. Second, a large number of patients who underwent TSA in the NSQIP database during the study period were excluded for lack of an INR value within 2 days of surgery. It is likely that these excluded patients represent a healthier cohort such that the rates of bleeding, complications, and readmissions found in our study are higher than those of the general population. Among the study cohort, there were notable baseline differences in comorbidities, which may serve as confounding variables. The relatively smaller number of patients in higher INR cohorts may limit our power to detect notable differences among groups. Third, INR is commonly used for the monitoring of warfarin, but not all antithrombotic medications. INR is not routinely used to monitor direct thrombin inhibitors, direct factor Xa inhibitors, and antiplatelet medications, and therefore, our findings are not applicable for the risk stratification of patients on these therapies. Finally, there is a growing trend toward outpatient shoulder arthroplasty.23-25 The NSQIP database captures major surgeries performed at participating hospitals, but not ambulatory surgery centers; therefore, our study is vulnerable to selection bias if patients who undergo TSA at hospitals have higher comorbidity burdens placing them at increased risk of postoperative bleeding.
In this study, we have demonstrated a stepwise increase in the risk of postoperative bleeding requiring transfusion with increasing preoperative INR levels in patients undergoing TSA. Moreover, higher preoperative INR levels are associated with longer hospital length of stay. Finally, even at levels below the historical threshold of 1.5, higher preoperative INR levels are associated with increased risk of 30-day complications and hospital readmission, after adjusting for relevant medical comorbidities. Our conclusions are relevant for patient counseling, risk stratification, and the decision for surgery. Increased rates of adverse events at higher preoperative INR levels have cost implications insofar because healthcare expenditures can be expected to increase. Our findings do not suggest that a single numerical INR value can be defined as a safe preoperative threshold for all patients. Rather, it is important for patients and clinicians to understand that the risks of thromboembolism and bleeding comprise a continuum, with higher preoperative INR levels conferring higher postoperative bleeding risks after TSA. However, the decision to hold and/or reverse anticoagulation therapy before TSA must be balanced against the risk of thromboembolism. This clinical judgment may be complex in patients with multiple medical comorbidities, may involve shared decision making with the patient, and should involve multiple disciplines, such as the orthopaedic surgeon, anesthesiologist, primary care physician, and cardiologist. We do not endorse an absolute preoperative INR threshold for TSA, but rather, we encourage clinicians to use a multidisciplinary approach for the individual patient to best balance the risks of thromboembolism and bleeding.
Footnotes
None of the following authors or any immediate family member has received anything of value from or has stock or stock options held in a commercial company or institution related directly or indirectly to the subject of this article: Dr. Zhang, Dr. Dyer, and Dr. Earp.
Contributor Information
George S. M. Dyer, Email: gdyer@mgh.harvard.edu.
Brandon E. Earp, Email: bearp@partners.org.
References
- 1.Farley KX, Wilson JM, Kumar A, et al. : Prevalence of shoulder arthroplasty in the United States and the increasing burden of revision shoulder arthroplasty. JB JS Open Access 2021;6:e20.00156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wagner ER, Farley KX, Higgins I, Wilson JM, Daly CA, Gottschalk MB: The incidence of shoulder arthroplasty: Rise and future projections compared with hip and knee arthroplasty. J Shoulder Elbow Surg 2020;29:2601-2609. [DOI] [PubMed] [Google Scholar]
- 3.Best MJ, Aziz KT, Wilckens JH, McFarland EG, Srikumaran U: Increasing incidence of primary reverse and anatomic total shoulder arthroplasty in the United States. J Shoulder Elbow Surg 2021;30:1159-1166. [DOI] [PubMed] [Google Scholar]
- 4.Barlow BT, Hannon MT, Waldron JE: Preoperative management of antithrombotics in arthroplasty. J Am Acad Orthop Surg 2019;27:878-886. [DOI] [PubMed] [Google Scholar]
- 5.Baron TH, Kamath PS, McBane RD: Management of antithrombotic therapy in patients undergoing invasive procedures. N Engl J Med 2013;368:2113-2124. [DOI] [PubMed] [Google Scholar]
- 6.Eckman MH, Levine HJ, Pauker SG: Effect of laboratory variation in the prothrombin-time ratio on the results of oral anticoagulant therapy. N Engl J Med 1993;329:696-702. [DOI] [PubMed] [Google Scholar]
- 7.Doherty JU, Gluckman TJ, Hucker WJ, 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 2017;69:871-898. [DOI] [PubMed] [Google Scholar]
- 8.Blasdale C, Lawrence CM: Perioperative international normalized ratio level is a poor predictor of postoperative bleeding complications in dermatological surgery patients taking warfarin. Br J Dermatol 2008;158:522-526. [DOI] [PubMed] [Google Scholar]
- 9.Edmunds I, Avakian Z: Hand surgery on anticoagulated patients: A prospective study of 121 operations. Hand Surg 2010;15:109-113. [DOI] [PubMed] [Google Scholar]
- 10.Flanigan DC, Muchow R, Orwin J, Graf B: Arthroscopy on anticoagulated patients: A retrospective evaluation of postoperative complications. Orthopedics 2010;33:82-86. [DOI] [PubMed] [Google Scholar]
- 11.Hshieh TT, Kaung A, Hussain S, Curry MP, Sundaram V: The international normalized ratio does not reflect bleeding risk in esophageal variceal hemorrhage. Saudi J Gastroenterol 2015;21:254-258. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wallace DL, Latimer MD, Belcher HJ: Stopping warfarin therapy is unnecessary for hand surgery. J Hand Surg Br 2004;29:203-205. [DOI] [PubMed] [Google Scholar]
- 13.Segal JB, Dzik WH, Transfusion Medicine/Hemostasis Clinical Trials Network: Paucity of studies to support that abnormal coagulation test results predict bleeding in the setting of invasive procedures: An evidence-based review. Transfusion 2005;45:1413-1425. [DOI] [PubMed] [Google Scholar]
- 14.Sølbeck S, Ostrowski SR, Johansson PI: A review of the clinical utility of INR to monitor and guide administration of prothrombin complex concentrate to orally anticoagulated patients. Thromb J 2012;10:5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Tamim H, Habbal M, Saliba A, et al. : Preoperative INR and postoperative major bleeding and mortality: A retrospective cohort study. J Thromb Thrombolysis 2016;41:301-311. [DOI] [PubMed] [Google Scholar]
- 16.Rudasill SE, Liu J, Kamath AF: Revisiting the international normalized ratio (INR) threshold for complications in primary total knee arthroplasty: An analysis of 21,239 cases. J Bone Joint Surg Am 2019;101:514-522. [DOI] [PubMed] [Google Scholar]
- 17.Rudasill SE, Liu J, Kamath AF: Revisiting the international normalized ratio threshold for bleeding risk and mortality in primary total hip arthroplasty: A national surgical quality improvement Program analysis of 17,567 patients. J Bone Joint Surg Am 2020;102:52-59. [DOI] [PubMed] [Google Scholar]
- 18.Sivasundaram L, Raji Y, Mengers SRP, et al. : Investigating the international normalized ratio thresholds for complication in shoulder arthroplasty. J Am Acad Orthop Surg 2021;29:131-137. [DOI] [PubMed] [Google Scholar]
- 19.American College of Surgeons: ACS National Surgical Quality Improvement Program. Available at https://www.facs.org/quality-programs/acs-nsqip, Accessed January 15, 2022. [Google Scholar]
- 20.Gruson KI, Lo Y, Stallone S, Qawasmi F, Lee S, Shah P: A comparison of operative time and intraoperative blood volume loss between stemless and short-stem anatomic total shoulder arthroplasty: A single institution's experience. J Am Acad Orthop Surg Glob Res Rev 2022;6:e22.00141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kim MS, Kim JD, Ro KH, Park JJ, Rhee YG: Hematologic profile in reverse total shoulder arthroplasty: Perioperative and postoperative blood loss. J Shoulder Elbow Surg 2019;28:1737-1742. [DOI] [PubMed] [Google Scholar]
- 22.Park C, Blevins KM, Paul AV, Long JS, Meyer LE, Anakwenze OA: Perioperative management in shoulder arthroplasty: A review of current practice. Orthop Clin North Am 2022;53:483-490. [DOI] [PubMed] [Google Scholar]
- 23.Avant-Garde Health and Codman Shoulder Society Value Based Care Group, Khan AZ, Best MJ, Fedorka CJ, et al. : Impact of the COVID-19 pandemic on shoulder arthroplasty: Surgical trends and postoperative care pathway analysis. J Shoulder Elbow Surg 2022;31:2457-2464. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kramer JD, Chan PH, Prentice HA, Hatch J, Dillon MT, Navarro RA: Same-day discharge is not inferior to longer length of in-hospital stay for 90-day readmissions following shoulder arthroplasty. J Shoulder Elbow Surg 2020;29:898-905. [DOI] [PubMed] [Google Scholar]
- 25.Leroux TS, Zuke WA, Saltzman BM, et al. : Safety and patient satisfaction of outpatient shoulder arthroplasty. JSES Open Access 2018;2:13-17. [DOI] [PMC free article] [PubMed] [Google Scholar]