Society of Cardiovascular Anesthesiologists Section Editor: Martin J. London
Preoperative Aspirin Use and Lung Injury After Aortic Valve Replacement Surgery: A Retrospective Cohort Study
Michael Mazzeffi, MD, MPH, Woderyelesh Kassa, BA, James Gammie, MD, Kenichi Tanaka, MD, MSc, Philip Roman, MD, MPH, Min Zhan, PhD, Bartley Griffith, MD, and Peter Rock, MD, MBA BACKGROUND: Acute respiratory distress syndrome (ARDS) occurs uncommonly after cardiac surgery but has a mortality rate as high as 80%. Aspirin may prevent lung injury in at-risk patients by reducing platelet-neutrophil aggregates in the lung. We hypothesized that preoperative aspirin use would be associated with a decreased risk of ARDS after aortic valve replacement surgery. METHODS: We performed a retrospective single-center cohort study that included all adult patients who had aortic valve replacement surgery during a 5-year period. The primary outcome variable was postoperative ARDS. The secondary outcome variable was nadir Pao2/Fio2 ratio during the first 72 hours after surgery. Both crude and propensity score–adjusted logistic regression analyses were performed to estimate the odds ratio for developing ARDS in aspirin users. Subgroups were analyzed to determine whether preoperative aspirin use might be associated with improved oxygenation in patients with specific risk factors for lung injury. RESULTS: Of the 375 patients who had aortic valve replacement surgery during the study period, 181 patients took aspirin preoperatively (48.3%) with most taking a dose of 81 mg (72.0%). There were 22 cases of ARDS in the cohort (5.5%). There was no significant difference in the rate of ARDS between aspirin users and nonusers (5.0% vs 6.7%, P = 0.52). There was also no significant difference in the nadir Pao2/Fio2 ratio between aspirin users and nonusers (P = 0.12). The crude odds ratio for ARDS in aspirin users was 0.725 (99% confidence interval, 0.229–2.289; P = 0.47), and the propensity score–adjusted odds ratio was 0.457 (99% confidence interval, 0.120–1.730; P = 0.13). CONCLUSIONS: Within the constraints of this analysis that included only 22 affected patients, preoperative aspirin use was not associated with a decreased incidence of ARDS after aortic valve replacement surgery or improved oxygenation. (Anesth Analg 2015;121:271–7)
A
cute respiratory distress syndrome (ARDS) occurs in as many as 20% of cardiac surgery patients and may have a mortality rate as high as 80%.1 There are a number of insults that occur during cardiac surgery that may contribute to lung injury, including exposure to cardiopulmonary bypass, transfusion of allogeneic blood products, and deflation of the lungs leading to atelectrauma.2 To date, few preventive strategies have been investigated in patients who are at high risk for acute lung injury, including patients having cardiac surgery. Identification of high-risk patients is possible, but, unfortunately, many risk factors are not modifiable. Gajic et al.3 proposed a lung injury prediction score that incorporates 8 predisposing conditions, including the presence of shock, sepsis, and emergency surgery, and 7 risk modifiers, including the use of high inspired oxygen concentrations, diabetes mellitus, and smoking, to
From the Department of Anesthesiology, University of Maryland School of Medicine, Baltimore, Maryland. Accepted for publication March 11, 2015. Funding: WK was funded by a National Institute of Aging T35 Grant. There was no other funding for the study. The authors declare no conflicts of interest. This report was previously presented, in part, at the Society for Cardiovascular Anesthesiologists annual meeting in Washington, DC, 2015. Reprints will not be available from the authors. Address correspondence to Michael Mazzeffi, MD, MPH, Department of Anesthesiology, University of Maryland School of Medicine, 22 S. Greene St., S11C00, Baltimore, MD 21201. Address e-mail to [email protected]. Copyright © 2015 International Anesthesia Research Society DOI: 10.1213/ANE.0000000000000793
August 2015 • Volume 121 • Number 2
predict the likelihood of lung injury. The lung injury prediction score has high discriminative value and is well calibrated (area under the curve = 0.84 and Hosmer-Lemeshow Statistic = 0.88) but may not be applicable to all patient populations.4 Aspirin has pleiotropic effects because of its irreversible inhibition of cyclooxygenase leading to downstream inhibition of prostaglandin and leukotriene synthesis.5 For this reason, aspirin has been suggested as a potentially protective strategy against acute lung injury in at-risk patients.6 In a retrospective study of almost 4000 hospitalized medical patients who were at high risk for lung injury, Kor et al.7 found that aspirin was associated with a reduced rate of acute lung injury based on univariate analysis. However, when propensity matching was used to control for potential confounders in that analysis, the association between aspirin use and reduced risk for acute lung injury was negated. In another retrospective study of 575 critically ill hospitalized noncardiac surgery patients, prehospital use of aspirin and statins was associated with a reduced incidence of acute lung injury.8 Other studies have suggested that aspirin may reduce organ failure, other than lung injury, in critically ill patients.9 Currently, the Lung Injury Prevention Study with Aspirin, which is a randomized controlled trial funded by the National Heart, Lung, and Blood Institute, is underway to test the hypothesis that aspirin may prevent acute lung injury in at-risk hospitalized patients. We hypothesized that preoperative aspirin use would be associated with a decreased incidence of ARDS after aortic valve replacement surgery. Furthermore, we hypothesized www.anesthesia-analgesia.org
271
Copyright © 2015 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
Aspirin and Lung Injury After Aortic Valve Surgery
that preoperative aspirin use would be associated with improved oxygenation after surgery.
METHODS Subjects
The IRB of the University of Maryland, Baltimore, Maryland, approved the study and waived the requirement for informed consent. We performed a retrospective single-center cohort study that included all adult patients having aortic valve replacement surgery with cardiopulmonary bypass between July 1, 2008 and June 30, 2013. Patients having aortic valve replacement surgery were selected for the cohort because there was a relatively equal distribution of aspirin users and nonusers in this group. The study period was selected because electronic medical record data and data entered into our institutional Society for Thoracic Surgeons (STS) database were uniformly available for all patients who had cardiac surgery during that period, and a 5-year duration was thought to represent relatively homogenous perioperative care.
Definitions
Definitions for patient variables were based on the STS database definitions (versions 2.61 and 2.73; www.sts.org). ARDS was defined using the Berlin definition: onset of acute lung injury within 1 week of an identifiable insult, bilateral lung opacities on chest radiography, noncardiogenic pulmonary edema, and Pao2/Fio2 ratio of 201 to 300 mm Hg for mild ARDS, 101 to 200 mm Hg for moderate ARDS, and ≤100 mm Hg for severe ARDS. Pulmonary edema was assumed to be noncardiogenic if the most recent echocardiogram of the patients did not suggest left ventricular systolic or diastolic heart failure and the daily progress note of the intensive care physicians did not indicate a cardiogenic etiology. All cases of ARDS had sustained impairment in oxygenation represented by ≥48 hours of hypoxemia, defined by a Pao2/Fio2 ≤300 mm Hg, and were confirmed by the principal investigator and one other investigator. Pao2/Fio2 ratios were calculated using arterial blood gas measurements obtained at frequent intervals after surgery. The baseline Pao2/Fio2 ratio was calculated using the first arterial blood gas in the operating room. Some patients had this blood gas collected before anesthesia induction and others had it collected after the induction of general anesthesia. For patients who had their first blood gas collected before anesthesia induction, Fio2 was estimated using the number of liters per minute of nasal cannula oxygen that they were receiving (3% Fio2 added to 21% for each liter per minute of oxygen). Postoperative arterial blood gas measurements from the first 72 hours after surgery were used to determine the lowest Pao2/Fio2 ratio during that time period. Preoperative aspirin use was defined as having received aspirin within 5 days of surgery. All blood products given during the operative period were recorded.
Outcome Variables
The primary outcome variable was the occurrence of ARDS. The secondary outcome variable was nadir Pao2/Fio2 ratio during the first 72 hours after surgery.
272 www.anesthesia-analgesia.org
Statistical Analysis
All statistical analyses were performed using SAS 9.3 (Cary, NC). The cohort was separated into aspirin users and nonusers. Categorical variables were reported as number (%) and continuous variables were examined for normality using histograms and the Shapiro-Wilk test. Normally distributed continuous variables were reported as mean ± SD and nonnormally distributed continuous variables were reported as median (Q1, Q3). Comparisons were made between patient characteristics and outcomes for aspirin users and nonusers using the Fisher exact test for categorical variables, t tests for normally distributed continuous variables, and Wilcoxon rank sum tests for nonnormally distributed continuous variables. Ninety-nine percent confidence intervals (CIs) were calculated for risk differences in outcome variables. Logistic regression analysis was performed with aspirin as the independent variable and ARDS as the dependent variable to determine the crude odds ratio for ARDS in patients taking aspirin. A 99% CI was calculated for the crude odds ratio. In addition to estimating a crude odds ratio, we used the following preoperative variables to create a propensity score for the probability of receiving aspirin: age, cerebral vascular disease, congestive heart failure, diabetes mellitus, dyslipidemia, end-stage renal disease, sex, height, hypertension, infectious endocarditis, international normalized ratio, left ventricular ejection fraction, peripheral vascular disease, and body weight. The propensity score methods used were based on previously published methods.10,11 Propensity scores were calculated using logistic regression with preoperative variables entered into the model as independent variables and aspirin use entered as the dependent variable. Independent variables were selected for inclusion in the model based on whether they were thought to be associated with aspirin use and the outcome variable. To assess the balance of propensity scores across the aspirin users and nonusers, the distributions of propensity scores were plotted using histograms. Also, a receiver operating characteristic curve was created, and the area under the curve was calculated as a measure of model discrimination. Propensity scores were used to control for confounding by entering them into a logistic regression model as a second independent variable with aspirin and ARDS as the dependent variable. A 99% CI was calculated for the propensity score–adjusted odds ratio of ARDS in aspirin users.
Subgroup Analyses
A priori we decided to perform 3 subgroup analyses based on biological plausibility. The subgroups were stratified by age, predicted prolonged (>24 hours) mechanical lung ventilation, and the number of units of fresh-frozen plasma transfused intraoperatively. The strata in the subgroups were based on the distributions of the data and what we perceived as logical increments. In the subgroup analyses, we examined only the secondary outcome variable (nadir Pao2/Fio2 ratio during the 72 hours after surgery) because the incidence of ARDS was low in the cohort, and cell counts would otherwise be very low for individual strata. Ninety-nine percent CIs were calculated for risk differences between aspirin users and nonusers at different levels of oxygenation impairment.
anesthesia & analgesia
Copyright © 2015 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
Table 1. Patient Medical and Demographic Characteristics Variable Age (years), median (Q1, Q3) Male sex, n (%) Weight (kg), median (Q1, Q3) Height (cm), mean ± SD Length of stay (days), median (Q1, Q3) Diabetes mellitus, n (%) Dyslipidemia, n (%) Dialysis dependent, n (%) Hypertension, n (%) Endocarditis, n (%) Chronic lung disease, n (%) None Mild Moderate Severe Peripheral vascular disease, n (%) Cerebrovascular disease, n (%) Heart failure within 2 weeks, n (%) Aortic insufficiency, n (%) None Trace Mild Moderate Severe Aortic stenosis, n (%) Left ventricular ejection fraction (percentage), median (Q1, Q3) Preoperative Pao2/Fio2 ratio (mm Hg), median (Q1, Q3) Preoperative international normalized ratio median (Q1, Q3) Cardiopulmonary bypass duration (minutes), median (Q1, Q3) Intraoperative RBC transfusion (units), median (Q1, Q3) Intraoperative plasma transfusion (units), median (Q1, Q3) Predicted mechanical lung ventilation >24 hours (probability), median (Q1, Q3)
Entire cohort 69 (59, 79) 213 (56.8) 84 (71, 99) 168.2 + 14.2 6 (5, 8) 111 (29.6) 236 (62.9) 14 (3.7) 286 (76.3) 30 (8.0)
No aspirin, n = 194 65 (54, 76) 114 (58.8) 83 (69, 101) 167.3 + 16.5 6 (5, 8) 53 (27.3) 101 (52.1) 9 (4.6) 144 (74.2) 19 (9.8)
321 (85.6) 38 (10.1) 11 (2.9) 5 (1.3) 25 (6.7) 68 (18.1) 176 (46.9) 122 (32.8) 28 (7.5) 87 (23.4) 64 (17.2) 71 (19.1) 307 (81.9) 59 (50, 60) 317 (243, 414) 1.3 (1.2, 1.4) 104 (86, 125) 1 (0, 3) 0 (0, 2) 0.09 (0.06, 0.14)
Aspirin, n = 181 72 (64, 80) 99 (55.0) 84 (73, 98) 169.1 + 11.2 7 (5, 9) 58 (32.0) 135 (74.6) 5 (2.8) 142 (78.5) 11 (6.1)
162 (83.5) 24 (12.4) 6 (3.1) 2 (1.0) 10 (5.2) 29 (15.0) 86 (44.3)
159 (87.9) 14 (7.7) 5 (2.8) 3 (1.6) 15 (8.3) 39 (21.6) 90 (49.7)
59 (30.7) 9 (4.7) 37 (19.3) 38 (19.8) 49 (25.5) 146 (75.3) 60 (60, 63) 322.5 (243, 412) 1.3 (1.2, 1.4) 101 (84, 120) 2 (0, 3) 0 (0, 2) 0.08 (0.06, 0.14)
63 (35.0) 19 (10.6) 50 (27.8) 26 (14.4) 22 (12.2) 161 (89.0) 58 (50, 60) 303 (245, 418) 1.3 (1.2, 1.5) 105 (87, 128) 1 (0, 3) 0 (0, 2) 0.10 (0.06, 0.15)
P
Preoperative Aspirin Use and Lung Injury After Aortic Valve Replacement Surgery: A Retrospective Cohort Study
Michael Mazzeffi, MD, MPH, Woderyelesh Kassa, BA, James Gammie, MD, Kenichi Tanaka, MD, MSc, Philip Roman, MD, MPH, Min Zhan, PhD, Bartley Griffith, MD, and Peter Rock, MD, MBA BACKGROUND: Acute respiratory distress syndrome (ARDS) occurs uncommonly after cardiac surgery but has a mortality rate as high as 80%. Aspirin may prevent lung injury in at-risk patients by reducing platelet-neutrophil aggregates in the lung. We hypothesized that preoperative aspirin use would be associated with a decreased risk of ARDS after aortic valve replacement surgery. METHODS: We performed a retrospective single-center cohort study that included all adult patients who had aortic valve replacement surgery during a 5-year period. The primary outcome variable was postoperative ARDS. The secondary outcome variable was nadir Pao2/Fio2 ratio during the first 72 hours after surgery. Both crude and propensity score–adjusted logistic regression analyses were performed to estimate the odds ratio for developing ARDS in aspirin users. Subgroups were analyzed to determine whether preoperative aspirin use might be associated with improved oxygenation in patients with specific risk factors for lung injury. RESULTS: Of the 375 patients who had aortic valve replacement surgery during the study period, 181 patients took aspirin preoperatively (48.3%) with most taking a dose of 81 mg (72.0%). There were 22 cases of ARDS in the cohort (5.5%). There was no significant difference in the rate of ARDS between aspirin users and nonusers (5.0% vs 6.7%, P = 0.52). There was also no significant difference in the nadir Pao2/Fio2 ratio between aspirin users and nonusers (P = 0.12). The crude odds ratio for ARDS in aspirin users was 0.725 (99% confidence interval, 0.229–2.289; P = 0.47), and the propensity score–adjusted odds ratio was 0.457 (99% confidence interval, 0.120–1.730; P = 0.13). CONCLUSIONS: Within the constraints of this analysis that included only 22 affected patients, preoperative aspirin use was not associated with a decreased incidence of ARDS after aortic valve replacement surgery or improved oxygenation. (Anesth Analg 2015;121:271–7)
A
cute respiratory distress syndrome (ARDS) occurs in as many as 20% of cardiac surgery patients and may have a mortality rate as high as 80%.1 There are a number of insults that occur during cardiac surgery that may contribute to lung injury, including exposure to cardiopulmonary bypass, transfusion of allogeneic blood products, and deflation of the lungs leading to atelectrauma.2 To date, few preventive strategies have been investigated in patients who are at high risk for acute lung injury, including patients having cardiac surgery. Identification of high-risk patients is possible, but, unfortunately, many risk factors are not modifiable. Gajic et al.3 proposed a lung injury prediction score that incorporates 8 predisposing conditions, including the presence of shock, sepsis, and emergency surgery, and 7 risk modifiers, including the use of high inspired oxygen concentrations, diabetes mellitus, and smoking, to
From the Department of Anesthesiology, University of Maryland School of Medicine, Baltimore, Maryland. Accepted for publication March 11, 2015. Funding: WK was funded by a National Institute of Aging T35 Grant. There was no other funding for the study. The authors declare no conflicts of interest. This report was previously presented, in part, at the Society for Cardiovascular Anesthesiologists annual meeting in Washington, DC, 2015. Reprints will not be available from the authors. Address correspondence to Michael Mazzeffi, MD, MPH, Department of Anesthesiology, University of Maryland School of Medicine, 22 S. Greene St., S11C00, Baltimore, MD 21201. Address e-mail to [email protected]. Copyright © 2015 International Anesthesia Research Society DOI: 10.1213/ANE.0000000000000793
August 2015 • Volume 121 • Number 2
predict the likelihood of lung injury. The lung injury prediction score has high discriminative value and is well calibrated (area under the curve = 0.84 and Hosmer-Lemeshow Statistic = 0.88) but may not be applicable to all patient populations.4 Aspirin has pleiotropic effects because of its irreversible inhibition of cyclooxygenase leading to downstream inhibition of prostaglandin and leukotriene synthesis.5 For this reason, aspirin has been suggested as a potentially protective strategy against acute lung injury in at-risk patients.6 In a retrospective study of almost 4000 hospitalized medical patients who were at high risk for lung injury, Kor et al.7 found that aspirin was associated with a reduced rate of acute lung injury based on univariate analysis. However, when propensity matching was used to control for potential confounders in that analysis, the association between aspirin use and reduced risk for acute lung injury was negated. In another retrospective study of 575 critically ill hospitalized noncardiac surgery patients, prehospital use of aspirin and statins was associated with a reduced incidence of acute lung injury.8 Other studies have suggested that aspirin may reduce organ failure, other than lung injury, in critically ill patients.9 Currently, the Lung Injury Prevention Study with Aspirin, which is a randomized controlled trial funded by the National Heart, Lung, and Blood Institute, is underway to test the hypothesis that aspirin may prevent acute lung injury in at-risk hospitalized patients. We hypothesized that preoperative aspirin use would be associated with a decreased incidence of ARDS after aortic valve replacement surgery. Furthermore, we hypothesized www.anesthesia-analgesia.org
271
Copyright © 2015 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
Aspirin and Lung Injury After Aortic Valve Surgery
that preoperative aspirin use would be associated with improved oxygenation after surgery.
METHODS Subjects
The IRB of the University of Maryland, Baltimore, Maryland, approved the study and waived the requirement for informed consent. We performed a retrospective single-center cohort study that included all adult patients having aortic valve replacement surgery with cardiopulmonary bypass between July 1, 2008 and June 30, 2013. Patients having aortic valve replacement surgery were selected for the cohort because there was a relatively equal distribution of aspirin users and nonusers in this group. The study period was selected because electronic medical record data and data entered into our institutional Society for Thoracic Surgeons (STS) database were uniformly available for all patients who had cardiac surgery during that period, and a 5-year duration was thought to represent relatively homogenous perioperative care.
Definitions
Definitions for patient variables were based on the STS database definitions (versions 2.61 and 2.73; www.sts.org). ARDS was defined using the Berlin definition: onset of acute lung injury within 1 week of an identifiable insult, bilateral lung opacities on chest radiography, noncardiogenic pulmonary edema, and Pao2/Fio2 ratio of 201 to 300 mm Hg for mild ARDS, 101 to 200 mm Hg for moderate ARDS, and ≤100 mm Hg for severe ARDS. Pulmonary edema was assumed to be noncardiogenic if the most recent echocardiogram of the patients did not suggest left ventricular systolic or diastolic heart failure and the daily progress note of the intensive care physicians did not indicate a cardiogenic etiology. All cases of ARDS had sustained impairment in oxygenation represented by ≥48 hours of hypoxemia, defined by a Pao2/Fio2 ≤300 mm Hg, and were confirmed by the principal investigator and one other investigator. Pao2/Fio2 ratios were calculated using arterial blood gas measurements obtained at frequent intervals after surgery. The baseline Pao2/Fio2 ratio was calculated using the first arterial blood gas in the operating room. Some patients had this blood gas collected before anesthesia induction and others had it collected after the induction of general anesthesia. For patients who had their first blood gas collected before anesthesia induction, Fio2 was estimated using the number of liters per minute of nasal cannula oxygen that they were receiving (3% Fio2 added to 21% for each liter per minute of oxygen). Postoperative arterial blood gas measurements from the first 72 hours after surgery were used to determine the lowest Pao2/Fio2 ratio during that time period. Preoperative aspirin use was defined as having received aspirin within 5 days of surgery. All blood products given during the operative period were recorded.
Outcome Variables
The primary outcome variable was the occurrence of ARDS. The secondary outcome variable was nadir Pao2/Fio2 ratio during the first 72 hours after surgery.
272 www.anesthesia-analgesia.org
Statistical Analysis
All statistical analyses were performed using SAS 9.3 (Cary, NC). The cohort was separated into aspirin users and nonusers. Categorical variables were reported as number (%) and continuous variables were examined for normality using histograms and the Shapiro-Wilk test. Normally distributed continuous variables were reported as mean ± SD and nonnormally distributed continuous variables were reported as median (Q1, Q3). Comparisons were made between patient characteristics and outcomes for aspirin users and nonusers using the Fisher exact test for categorical variables, t tests for normally distributed continuous variables, and Wilcoxon rank sum tests for nonnormally distributed continuous variables. Ninety-nine percent confidence intervals (CIs) were calculated for risk differences in outcome variables. Logistic regression analysis was performed with aspirin as the independent variable and ARDS as the dependent variable to determine the crude odds ratio for ARDS in patients taking aspirin. A 99% CI was calculated for the crude odds ratio. In addition to estimating a crude odds ratio, we used the following preoperative variables to create a propensity score for the probability of receiving aspirin: age, cerebral vascular disease, congestive heart failure, diabetes mellitus, dyslipidemia, end-stage renal disease, sex, height, hypertension, infectious endocarditis, international normalized ratio, left ventricular ejection fraction, peripheral vascular disease, and body weight. The propensity score methods used were based on previously published methods.10,11 Propensity scores were calculated using logistic regression with preoperative variables entered into the model as independent variables and aspirin use entered as the dependent variable. Independent variables were selected for inclusion in the model based on whether they were thought to be associated with aspirin use and the outcome variable. To assess the balance of propensity scores across the aspirin users and nonusers, the distributions of propensity scores were plotted using histograms. Also, a receiver operating characteristic curve was created, and the area under the curve was calculated as a measure of model discrimination. Propensity scores were used to control for confounding by entering them into a logistic regression model as a second independent variable with aspirin and ARDS as the dependent variable. A 99% CI was calculated for the propensity score–adjusted odds ratio of ARDS in aspirin users.
Subgroup Analyses
A priori we decided to perform 3 subgroup analyses based on biological plausibility. The subgroups were stratified by age, predicted prolonged (>24 hours) mechanical lung ventilation, and the number of units of fresh-frozen plasma transfused intraoperatively. The strata in the subgroups were based on the distributions of the data and what we perceived as logical increments. In the subgroup analyses, we examined only the secondary outcome variable (nadir Pao2/Fio2 ratio during the 72 hours after surgery) because the incidence of ARDS was low in the cohort, and cell counts would otherwise be very low for individual strata. Ninety-nine percent CIs were calculated for risk differences between aspirin users and nonusers at different levels of oxygenation impairment.
anesthesia & analgesia
Copyright © 2015 International Anesthesia Research Society. Unauthorized reproduction of this article is prohibited.
Table 1. Patient Medical and Demographic Characteristics Variable Age (years), median (Q1, Q3) Male sex, n (%) Weight (kg), median (Q1, Q3) Height (cm), mean ± SD Length of stay (days), median (Q1, Q3) Diabetes mellitus, n (%) Dyslipidemia, n (%) Dialysis dependent, n (%) Hypertension, n (%) Endocarditis, n (%) Chronic lung disease, n (%) None Mild Moderate Severe Peripheral vascular disease, n (%) Cerebrovascular disease, n (%) Heart failure within 2 weeks, n (%) Aortic insufficiency, n (%) None Trace Mild Moderate Severe Aortic stenosis, n (%) Left ventricular ejection fraction (percentage), median (Q1, Q3) Preoperative Pao2/Fio2 ratio (mm Hg), median (Q1, Q3) Preoperative international normalized ratio median (Q1, Q3) Cardiopulmonary bypass duration (minutes), median (Q1, Q3) Intraoperative RBC transfusion (units), median (Q1, Q3) Intraoperative plasma transfusion (units), median (Q1, Q3) Predicted mechanical lung ventilation >24 hours (probability), median (Q1, Q3)
Entire cohort 69 (59, 79) 213 (56.8) 84 (71, 99) 168.2 + 14.2 6 (5, 8) 111 (29.6) 236 (62.9) 14 (3.7) 286 (76.3) 30 (8.0)
No aspirin, n = 194 65 (54, 76) 114 (58.8) 83 (69, 101) 167.3 + 16.5 6 (5, 8) 53 (27.3) 101 (52.1) 9 (4.6) 144 (74.2) 19 (9.8)
321 (85.6) 38 (10.1) 11 (2.9) 5 (1.3) 25 (6.7) 68 (18.1) 176 (46.9) 122 (32.8) 28 (7.5) 87 (23.4) 64 (17.2) 71 (19.1) 307 (81.9) 59 (50, 60) 317 (243, 414) 1.3 (1.2, 1.4) 104 (86, 125) 1 (0, 3) 0 (0, 2) 0.09 (0.06, 0.14)
Aspirin, n = 181 72 (64, 80) 99 (55.0) 84 (73, 98) 169.1 + 11.2 7 (5, 9) 58 (32.0) 135 (74.6) 5 (2.8) 142 (78.5) 11 (6.1)
162 (83.5) 24 (12.4) 6 (3.1) 2 (1.0) 10 (5.2) 29 (15.0) 86 (44.3)
159 (87.9) 14 (7.7) 5 (2.8) 3 (1.6) 15 (8.3) 39 (21.6) 90 (49.7)
59 (30.7) 9 (4.7) 37 (19.3) 38 (19.8) 49 (25.5) 146 (75.3) 60 (60, 63) 322.5 (243, 412) 1.3 (1.2, 1.4) 101 (84, 120) 2 (0, 3) 0 (0, 2) 0.08 (0.06, 0.14)
63 (35.0) 19 (10.6) 50 (27.8) 26 (14.4) 22 (12.2) 161 (89.0) 58 (50, 60) 303 (245, 418) 1.3 (1.2, 1.5) 105 (87, 128) 1 (0, 3) 0 (0, 2) 0.10 (0.06, 0.15)
P
