Tory McGrath, MD, Colleen Gorman Koch, MD, MS, Meng Xu, MS, Liang Li, PhD, Tomislav Mihaljevic, MD, Priscilla Figueroa, MD, and Eugene H. Blackstone, MD Departments of Cardiothoracic Anesthesia, Quantitative Health Sciences, Thoracic and Cardiovascular Surgery, Laboratory Medicine and Clinical Pathology, and Outcomes Research, Cleveland Clinic, Cleveland, Ohio

Background. Platelet transfusion has been reported to confer increased morbidity after cardiac surgery but prior studies were limited by confounding variables including red blood cell (RBC) transfusions. Our objective was to examine the impact of platelet transfusion on outcomes in cardiac surgery controlling perioperative risk factors. Methods. A total of 32,298 patients underwent on-pump isolated coronary artery bypass grafting (CABG), an isolated valve, or a combined CABG and valve procedure between January 1, 1993 and January 1, 2006. Regression analysis and propensity methodology was employed to assess the association between platelet transfusion and morbidity. Results. Univariate comparisons demonstrated that patients who received platelet transfusions had increased morbidity. After risk adjustment with both multivariable regression and propensity methods, platelet transfusion was not significantly associated with in-hospital mortality: odds ratio (OR) 0.74 confidence limits 0.58, 0.95, p ⴝ 0.017 and 2.05% vs 3.06%, p ⴝ 0.017, respectively. Among

2,774 propensity matched-pairs, platelet transfusion was associated with similar or reduced morbidity, platelets versus no platelets: cardiac 2.42% vs 1.77%, p ⴝ 0.09; pulmonary 8.94% vs 9.88%, p ⴝ 0.23; renal 1.33% vs 1.48%, p ⴝ 0.65; neurologic 2.27% vs 3.21%, p ⴝ 0.033; serious infection 4.15% vs 5.34%, p ⴝ 0.037; and composite outcome 15.0% vs 17.2%, p ⴝ 0.024. Among a propensity-matched subgroup of patients never administered a concomitant RBC transfusion, platelet transfusion was not associated with increased morbidity: 4.49% vs 2.99%, p ⴝ 0.31. Conclusions. Platelet transfusion was not found to increase morbid risk after cardiac surgery. Our results should not be interpreted as advocating platelet transfusions in cardiac surgery; rather, platelet transfusion empirically in the setting of persistent microvascular bleeding is not associated with increased morbid risk.

O

was to examine the relationship between platelet transfusion and morbid outcomes after cardiac surgery, controlling for a number of patient, operative, and perioperative transfusion-related variables.

ne of the most common complications after cardiac surgery is bleeding that is often attributed to thrombocytopathy among innumerable causes [1]. Although approximately 20% of platelet transfusions today are administered to cardiovascular surgical patients [2], evidence-based guidelines for platelet transfusion in cardiac surgery are limited and point-of-care platelet function testing is typically not utilized in the clinical decision-making process whether to transfuse platelets [3, 4]. Our institution typically transfuses platelet concentrates empirically if microvascular bleeding persists after adequate reversal of heparin after cardiopulmonary bypass (CPB). The clinical threshold is further reduced for those patients on preoperative antiplatelet medications. Recently, disparate findings have been reported regarding whether platelet transfusions are associated with adverse outcomes after cardiac surgery [5–9]. In light of observations from our prior investigations [10, 11] and others [5–9], our objective Accepted for publication April 14, 2008. Address correspondence to Dr McGrath, Department of Cardiothoracic Anesthesia (G-3), Cleveland Clinic, 9500 Euclid Avenue, Cleveland, OH 44195; e-mail: [email protected].

© 2008 by The Society of Thoracic Surgeons Published by Elsevier Inc

(Ann Thorac Surg 2008;86:543–53) © 2008 by The Society of Thoracic Surgeons

Patients and Methods Patient Population From January 1, 1993 to January 1, 2006, 32,298 patients underwent isolated coronary artery bypass grafting (CABG), an isolated valve procedure, or a combined CABG and valve procedure requiring cardiopulmonary bypass at Cleveland Clinic. Perioperative variables and blood component utilization forms were prospectively collected concurrent with patient care and entered into Cleveland Clinic Department of Cardiothoracic Anesthesia Registry by trained database management personnel. The Cardiovascular Information Registry was accessed for additional operative variables. Institutional review Board approved and waived individual patient consent for performing research analyses on these validated, accurate databases. 0003-4975/08/$34.00 doi:10.1016/j.athoracsur.2008.04.051

ADULT CARDIAC

Platelet Transfusion in Cardiac Surgery Does Not Confer Increased Risk for Adverse Morbid Outcomes

544

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

Ann Thorac Surg 2008;86:543–53

Table 1. Baseline Variables for Unmatched Patients by Platelet Transfusion Variables

ADULT CARDIAC

Demographics: Age (years) Gender (female) BSA (kg/M2) Cardiac comorbidity: History of atrial fibrillation Heart failure LV ejection fraction % (normal 60% vs abnormal) Myocardial infarction Preoperative IABP Emergency surgery Cardiogenic shock Reoperation Coronary disease ⬎70%: Left main trunk LAD L circumflex RCA NYHA functional class: I II III IV Canadian angina class: 0 1 2 3 4 Comorbidity: Hypertension Diabetes Peripheral vascular disease Smoking COPD Stroke Renal disease Preoperative laboratory values: Hematocrit (%) Creatinine (mg/dL) Preoperative valve: Pathology: Aortic valve regurgitation Aortic valve stenosis Mitral valve regurgitation Operative factors: Procedure: CABG Valve procedure Combined CABG/ Valve Cardiopulmonary bypass time (min) Aortic clamp time (min) LITA or RITA

Platelet Transfusion (n ⫽ 3,599)

No Platelet Transfusion (n ⫽ 25,888)

p Valuea

69.2 (61.2, 75.2) 1,050 (29.2) 1.94 (1.78, 2.09)

64.7 (55.4, 72.4) 7,743 (29.9) 2.0 (1.83, 2.15)

⬍0.001 0.37 ⬍0.001

325 (9.03) 1,285 (1285 (35.7) 1,581 (44.7)

1681 (6.49) 7,567 (29.2) 12,685 (49.7)

⬍0.001 ⬍0.001 ⬍0.001

1,900 (52.8) 146 (4.12%) 141 (3.92) 20 (0.565) 1,375 (38.2)

11,185 (43.2) 295 (1.16) 263 (1.02) 51 (0.2) 5,200 (20.1)

⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001

439 (12.2) 2,428 (67.5) 2,095 (58.2) 2,267 (63.0)

20,778 (8.02) 14,475 (55.9) 1,222 (47.2) 13,042 (50.4)

⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001

503 (14.0) 1,656 (46) 670 (18.6) 770 (21.4)

4,523 (17.5) 13,332 (51.5) 4,537 (17.5) 3,492 (13.5)

256 (7.11) 1,401 (38.9) 1,439 (40.0) 434 (12.1) 69 (1.92)

5,450 (21.1) 8,086 (31.2) 9,372 (36.2) 2,418 (9.34) 557 (3.15)

2,340 (68.4) 811 (23.5) 1,346 (37.4) 2,104 (59.4) 287 (8.11) 377 (10.5) 314 (8.72)

1,6547 (66.1) 5977 (23.7) 8,697 (33.6) 14,741 (57.8) 2,061 (8.08) 1,916 (7.4) 1,074 (4.15)

⬍0.001

39.8 (35.5, 43.0) 1.1 (0.9, 1.4)

40.8 (37.5, 43.5) 1.0 (0.9, 1.2)

1,019 (28.3) 746 (20.7) 1,628 (45.2)

6,950 (26.9) 4,780 (18.5) 13,143 (50.8)

2,030 (57.3) 626 (17.7) 884 (25.0) 118 (91, 148) 83 (63, 105) 1,553 (43.2)

13,381 (52.5) 7,995 (31.3) 4,133 (16.2) 91 (70, 116) 69 (53, 88) 11,629 (44.9)

0.008 0.77 ⬍0.001 0.06 0.96 ⬍0.001 0.77 ⬍0.001 ⬍0.001

0.06 0.001 ⬍0.001 ⬍0.001

⬍0.001 ⬍0.001 0.045

Ann Thorac Surg 2008;86:543–53

545

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

Table 1. Continued

Surgery before 2003 Blood product usage: RBC (units) a

No Platelet Transfusion (n ⫽ 25,888)

p Valuea

3,326 (92.4)

20,374 (78.7)

⬍0.001

4 (2, 8)

0 (0 ,2)

⬍0.001

Comparisons with Wilcoxon’s rank sum and ␹ where appropriate. 2

BSA ⫽ body surface area; CABG ⫽ coronary artery bypass grafting; COPD ⫽ chronic obstructive pulmonary disease; FFP ⫽ fresh frozen plasma; IABP ⫽ intraaortic balloon pump; ITA ⫽ internal thoracic artery; LAD ⫽ left anterior descending; LVEF ⫽ left ventricular ejection fraction; NYHA ⫽ New York Heart Association; RBC ⫽ red blood cells; RCA ⫽ right coronary artery.

Endpoints Outcome variables included in-hospital mortality, cardiac, pulmonary, renal, and neurologic morbidities, serious infection, and re-exploration for bleeding. Cardiac morbidity was defined as a low cardiac index (⬍1.8 L · min⫺1 · m⫺2) despite adequate fluid replacement and inotropic support for greater than 4 hours or a postoperative myocardial infarction with at least one of the following: preoperative or perioperative intraaortic balloon pump (IABP) placement, intraoperative ventricular assist device (VAD) placement, or return to the operating room for IABP, VAD, or extracorporeal membrane oxygenator (ECMO) placement. Myocardial infarction was diagnosed based on electrocardiogram and cardiac isoenzymes. Pulmonary morbidity was characterized as mechanical ventilatory support requiring intubation for 72 or greater hours postoperatively. Renal morbidity was defined as new-onset renal failure requiring dialysis. Neurologic morbidity included new global or focal neurologic deficits or death without awakening. Serious infection morbidity was defined as pneumonia, mediastinitis, or sepsis with the diagnosis of sepsis requiring organisms isolated from culture(s) in conjunction with elevated temperature and white blood cell counts. Reexploration for bleeding was defined as a return to the operating room specifically for bleeding. A composite outcome of adverse events consisted of in-hospital mortality, cardiac morbidity, respiratory insufficiency, renal morbidity, serious infection, and neurologic morbidity. There were less than 3% missing values for the data set; missing values were deleted from the data set prior to analysis.

Statistical Analysis We examined the impact of platelet transfusion on morbid outcomes after cardiac surgery with three approaches. Our first approach involved use of propensity methodology. Demographics, comorbidity, operative, and transfusion variables noted in Table 1 were used for development of a propensity score. The propensity score was obtained by calculating the predicted probability of receiving platelet transfusion for each patient from a logistic model [12]. Greedy 1:1 matching techniques were then used to select patients who received a platelet transfusion with those that did not by choosing the patients with the nearest propensity score [13]. Comparisons of morbid outcomes were made between the propensity-matched patients with the ␹2 or Fisher exact test where appropriate. Among the 32,298

patients, 2,811 received fresh frozen plasma (FFP) or cryoprecipitate and were excluded from propensity analysis due to an inability to adequately propensity match these patients. Hence, propensity methodology was applied to the data set of 29,487 patients. A separate multivariable logistic regression model on in-hospital mortality was applied to the entire data set (n ⫽ 32,298 including patients who received FFP and cryoprecipitate). Backward variable selection procedure was used with retention criteria of p less than 0.05 for all variables contained in Table 1. The balancing score and platelet transfusion status were forced into the final regression model. Finally, similar propensity methodology was used to further examine morbid outcomes by platelet transfusion status only for the subset of patients who did not receive RBC transfusion.

Results Unmatched Patient Population Among 29,487 patients, 3,599 (12%) received a platelet transfusion and 25,888 (88%) did not. Distribution of baseline demographics, comorbidity, operative variables, and transfusion status for the unmatched cohort of patients is presented by platelet transfusion status in Table 1.

Table 2. Morbid Outcomes for Unmatched Patients by Platelet Transfusion

Outcome Composite outcome Hospital death Morbidity: Cardiac Pulmonary Renal Neurologic Serious infection Return to OR for bleeding

a

Platelet Transfusion (n ⫽ 3,599)

No Platelet Transfusion (n ⫽ 25,888)

656 (18.2) 121 (3.36)

1,778 (6.87) 207 (0.8)

121 (3.42) 408 (11.5) 75 (2.12) 103 (2.91) 184 (5.2) 256 (7.11)

172 (0.67) 726 (2.85) 108 (0.42) 334 (1.31) 358 (1.4) 335 (1.29)

p Value ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001

a Includes a composite outcome of adverse events consisted of in-hospital mortality, cardiac morbidity, pulmonary morbidity, renal morbidity, serious infection, and neurologic morbidity.

OR ⫽ operating room.

ADULT CARDIAC

Platelet Transfusion (n ⫽ 3,599)

Variables

546

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

Ann Thorac Surg 2008;86:543–53

Table 3. Baseline Variables for Propensity-Matched Patients by Platelet Transfusion Variables

ADULT CARDIAC

Demographics: Age (years) Gender (female) BSA (kg/M2) Cardiac comorbidity: History of atrial fibrillation Heart failure LVEF % (normal vs abnormal) Myocardial infarction Preoperative IABP Emergency surgery Cardiogenic shock Reoperation Coronary disease ⬎70%: Left main trunk LAD L circumflex RCA NYHA functional class: I II III IV Canadian angina class: 0 1 2 3 4 Comorbidity: Hypertension Diabetes Peripheral vascular disease Smoking COPD Stroke Renal disease Preoperative laboratory values: Hematocrit (%) Creatinine (mg/dL) Preoperative valve pathology: Aortic valve regurgitation Aortic valve stenosis Mitral valve regurgitation Operative Factors: Procedure: CABG Valve procedure Combined CABG/ Valve Cardiopulmonary bypass time (min) Aortic clamp time (min) LITA or RITA Surgery before 2003

Platelet Transfusion (n ⫽ 2,774)

No Platelet Transfusion (n ⫽ 2,774)

68.8 (61.1, 75.1) 824 (29.7) 1.95 (1.79, 2.10)

68.9 (61.4, 75.0) 833 (30) 1.95 (1.80, 2.10)

p Value 0.92 0.79 0.38

243 (8.76) 983 (35.4) 1,251 (45.1) 1,466 (52.8) 97 (3.5) 86 (3.1) 16 (0.58) 970 (35.0)

287 (9.26) 975 (35.1) 1,274 (45.9) 1,445 (52.1) 91 (3.28) 73 (2.63) 17 (0.61) 962 (34.7)

0.51 0.82 0.54 0.57 0.66 0.30 0.86 0.82

330 (11.9) 1,864 (67.2) 1,627 (58.7) 1,727 (62.3)

324 (11.7) 1,914 (69.0) 1,644 (59.3) 1,736 (62.6)

0.80 0.15 0.64 0.80 0.92

370 (13.3) 1,309 (47.2) 519 (18.7) 576 (20.8)

368 (13.3) 1,321 (47.6) 529 (19.1) 556 (20.0)

234 (8.44) 1,001 (36.1) 1,142 (41.2) 337 (12.1) 60 (2.16)

242 (8.72) 961 (34.6) 1,161 (41.9) 354 (12.8) 56 (2.02)

1,904 (68.6) 672 (24.2) 1,061 (38.2) 1,651 (59.5) 220 (7.93) 291 (10.5) 236 (8.51)

1,913 (69.0) 673 (24.3) 1,099 (39.6) 1,632 (58.8) 219 (7.89) 298 (10.7) 236 (8.51)

0.80

0.79 0.98 0.30 0.60 0.96 0.76 ⬎0.99

39.8 (35.4, 43.0) 1.1 (0.9, 1.3)

39.6 (35.6, 42.6) 1.1 (0.9, 1.4)

0.29 0.68

781 (28.2) 591 (21.3) 1266 (45.6)

786 (28.3) 621 (22.4) 1237 (44.6)

0.88 0.33 0.43

1,593 (57.4) 499 (18.0) 682 (24.6) 115 (89, 144.5)

1,615 (58.2) 495 (17.8) 664 (23.9) 114 (89, 140)

0.18

82 (62, 103) 1,254 (45.2) 2,514 (90.6)

81 (62, 101) 1,277 (46) 2,499 (90.1)

0.29 0.54 0.50

0.82

Ann Thorac Surg 2008;86:543–53

547

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

Table 3. Continued

Blood product usage: RBC (units)

Platelet Transfusion (n ⫽ 2,774) 4 (2, 6)

No Platelet Transfusion (n ⫽ 2,774)

p Value

4 (2, 7)

0.74

BSA ⫽ body surface area; CABG ⫽ coronary artery bypass grafting; COPD ⫽ chronic obstructive pulmonary disease; FFP ⫽ fresh frozen plasma; IABP ⫽ intraaortic balloon pump; ITA ⫽ internal thoracic artery; LAD ⫽ left anterior descending; LVEF ⫽ left ventricular ejection fraction; NYHA ⫽ New York Heart Association; RBC ⫽ red blood cells; RCA ⫽ right coronary artery.

Patients who received platelets were significantly different from those who did not receive platelets. Unmatched patients who received platelets were older, had smaller body surface area measurements, had more comorbidity, more frequently underwent reoperations and emergencies, and had a more unstable presentation. In addition, they received more red blood cells (RBC) and had longer aortic cross-clamp and cardiopulmonary bypass times (Table 1). Outcomes among unmatched patients who received a platelet transfusion demonstrated significantly higher morbidity for all postoperative morbid events (Table 2).

Patients receiving platelets returned more often to the operating room for reexploration for bleeding (7% vs 2.5%, p ⬍ 0.001). Despite this, many risk-adjusted morbid outcomes remained significantly less for patients who received platelets. Among propensity-matched pairs, platelet transfusion was associated with reduced inhospital mortality, fewer neurologic events, reduced postoperative serious infectious complications, less composite adverse outcome events, and equivocal occurrence for cardiac, pulmonary, and renal morbidity (Table 4).

Propensity-Matched Population

Patients who received platelets had reduced in-hospital mortality (odds ratio [OR] 0.74, confidence limits [CL] 0.58 and 0.95, p ⫽ 0.017). Of note, patients who received transfusion of RBC had increased mortality (OR 1.64, CL 1.54 and 1.75, p ⬍ 0.001) as did transfusion of FFP (OR 3.29, CL 2.16 and 5.01, p ⬍ 0.001) and use of cryoprecipitate (OR 2.47, CL 1.79 and 3.41, p ⬍ 0.001) (Table 5).

Propensity modeling yielded 2,774 propensity-matched pairs by platelet transfusion status. Matched pairs had a similar distribution of demographic variables, comorbidity, clinical presentation, operative factors, and RBC transfusions (Table 3). Figure 1 displays the distribution of RBC units transfused by platelet transfusion status, which is similar between the propensity-matched groups. The remaining 825 patients who received platelet transfusions were not amenable to propensity modeling primarily due to inability to balance RBC transfusion unit status. The Appendix lists baseline variables among those matched and unmatched patients who received platelet transfusions.

Multivariable Logistic Regression on Mortality for Entire Patient Population

Composite Morbid Outcome for Propensity-Matched Patients by Platelet Transfusion Who Did Not Receive a RBC Transfusion Among patients who never received a RBC transfusion, 452 patients (3.2%) received a platelet transfusion and 13,890 (96.8%) did not. Propensity methods resulted in Fig 1. Distribution of red blood cell (RBC) unit transfusion by platelet transfusion status for the propensity-matched patient groups. ( ⫽ no platelet; s ⫽ platelet.)

ADULT CARDIAC

Variables

548

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

Ann Thorac Surg 2008;86:543–53

Table 4. Morbid Outcomes for Propensity-Matched Patients by Platelet Transfusion

ADULT CARDIAC

Outcome Composite outcomea Hospital death Morbidity: Cardiac Pulmonary Renal Neurologic Serious infection Return to OR for bleeding

Platelet No Platelet Transfusion Transfusion (n ⫽ 2,774) (n ⫽ 2,774) p Value 416 (15.0) 57 (2.05)

478 (17.2) 85 (3.06)

0.024 0.017

67 (2.42) 248 (8.94) 37 (1.33) 63 (2.27) 115 (4.15) 195 (7.03)

49 (1.77) 274 (9.88) 41 (1.48) 89 (3.21) 148 (5.34) 69 (2.49)

0.09 0.23 0.65 0.033 0.037 ⬍0.001

Baseline univariate comparisons were made between those who received a platelet transfusion and those who did not with the use of Wilcoxon rank sum and ␹2 tests where appropriate. a Includes a composite outcome of adverse events consisted of in-hospital mortality, cardiac morbidity, pulmonary morbidity, renal morbidity, serious infection, and neurologic morbidity. OR ⫽ operating room.

387 matched-pairs who were similarly matched on baseline and operative variables (Table 6). Comparison between the 387 propensity-matched pairs of patients who received and did not receive platelet transfusions demonstrated a similar composite outcome (15 [4.99%] vs 10 [2.99%], p ⫽ 0.31).

Comment We report that administration of platelets in the perioperative period was not associated with increased risk for postoperative mortality, infectious complications, and neurologic events, and did not confer increased risk for other morbid outcomes even though these patients returned to the operating room for bleeding more frequently. While the unadjusted relationship reported higher risk for every morbid outcome in patients receiving platelets, after matching and, in particular, adjustment for RBC transfusion status, platelet transfusion no longer conferred higher risk. Our prior investigations [10, 11] and that of others [14] have suggested that perioperative RBC transfusion is the single risk factor most reliably associated with increased incidence of postoperative morbid events after isolated CABG. While platelet transfusion status alone in our study appears to identify a group of cardiac surgical patients at very high risk for morbid events (see Table 2), results of this current investigation suggest that platelet transfusions are associated with similar or lower postoperative morbidity or mortality when RBC and other confounding variables are appropriately adjusted (see Table 2 versus Table 4). Our ability to adjust for units of RBC transfused as well as other preoperative, comorbid, and operative variables in our propensity-matched pairs serves as one of the strengths of this study evaluating the impact of platelet transfusions on morbidity and mortality in cardiac surgery.

Five prior retrospective investigations evaluated the impact of platelet transfusions on various adverse outcomes in cardiac surgery. The results of these studies yielded conflicting results concerning the association between platelet transfusions and morbid events. Vamvakas and Carven [5] examined the association between volumes of various allogeneic blood transfusion (ABT) products and prolonged postoperative ventilation in 416 low-risk patients undergoing CABG. After risk adjustment, the authors concluded that while the volume of administered RBC supernatant correlated with an increased likelihood of prolonged postoperative mechanical ventilation, the volume of platelet, plasma (including FFP and cryoprecipitate), or total blood product supernatant did not. Spiess and colleagues [6] published a retrospective multicenter examination of platelet transfusions and adverse outcomes in 1,720 CABG patients. These patients Table 5. Multivariable Logistic Regression on Hospital Mortality for Entire Patient Population (n ⫽ 32,298) Variable Transfusion: Platelets (yes/no) RBC (units) FFP (yes/no) Cryo (yes/no) Demographics: Age (years) Laboratory Values: Hematocrit (%) Cardiac morbidity: NYHA functional class: 1 vs 4 2 vs 4 3 vs 4 Heart failure Emergency surgery Left main trunk ⱖ 70% Cardiogenic shock Comorbidity: COPD Renal disease Diabetes Operative variables: ITA use Cardiopulmonary bypass time (log transformation) Aortic clamp time (log transformation) Preoperative valve pathology: Mitral valve regurgitation Balancing score

Odds Ratio and 95% CI

p Value

0.74 (0.58,0.95) 1.64 (1.54, 1.75) 3.29 (2.16, 5.01) 2.47 (1.79, 3.41)

0.017 ⬍0.001 ⬍0.001 ⬍0.001

1.02 (1.01, 1.03)

⬍0.001

1.03 (1.009, 1.05)

0.005 0.002

0.85 (0.62, 1.16) 0.66 (0.51, 0.85) 0.99 (0.77, 1.28) 1.42 (1.15, 1.76) 1.92 (1.30, 2.81) 1.45 (1.12, 1.87) 1.88 (1.11, 3.19)

0.001 ⬍0.001 0.005 0.020

1.69 (1.31, 2.17) 1.61 (1.26, 2.07) 1.36 (1.11, 1.67)

⬍0.001 ⬍0.001 0.003

0.78 (0.63, 0.95) 2.72 (1.84, 4.02)

0.016 ⬍0.001

0.51 (0.36, 0.71)

⬍0.001

1.31 (1.06, 1.64) 0.30 (0.11, 0.83)

0.014 0.020

CI ⫽ confidence interval; COPD ⫽ chronic obstructive pulmonary disease; CPB ⫽ cardiopulmonary bypass time; FFP ⫽ fresh frozen plasma; ITA ⫽ internal thoracic artery; NYHA ⫽ New York Heart Association; RBC ⫽ red blood cells.

Ann Thorac Surg 2008;86:543–53

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

549

Table 6. Baseline Variables for Propensity-Matched Patients Who Did Not Receive a Red Cell Transfusion by Platelet Transfusion

Demographics: Age (years) Gender (female) BSA (kg/M2) Cardiac comorbidity: History of atrial fibrillation Heart failure LVEF (normal vs abnormal) Myocardial infarction Preoperative IABP Emergency surgery Reoperation Coronary disease ⱖ70%: Left main Left anterior descending Left circumflex RCA NYHA functional class: I II III IV Canadian angina class: 0 1 2 3 4 Comorbidity: Hypertension Diabetes Peripheral vascular disease Smoking COPD Stroke Renal disease Preoperative laboratory values: Hematocrit (%) Creatinine (mg/dL) Preoperative valve pathology: Aortic valve regurgitation Aortic valve stenosis Mitral valve regurgitation Operative Factors: Procedure CABG Valve procedure Combined CABG/ Valve Cardiopulmonary bypass time Aortic clamp time (min) LITA or RITA use Surgery before 2003

Platelet Transfusion (n ⫽ 334)

No Platelet Transfusion (n ⫽ 334)

63.1 (55.3, 71.3) 23 (6.89) 2.09 (1.95, 2.23)

63.4 (55.8, 70.7) 25 (7.49) 2.09 (1.98, 2.22)

0.81 0.77 0.82

23 (6.89) 79 (23.7) 145 (43.4) 170 (50.9) 5 (1.5) 5 (1.5) 92 (27.5)

22 (6.59) 77 (23.1) 150 (44.9) 159 (47.6) 3 (0.90) 6 (1.8) 79 (23.7)

0.88 0.86 0.70 0.40 0.48 0.76 0.25

31 (9.28) 232 (69.5) 204 (61.1) 197 (59.0)

37 (11.1) 228 (68.3) 205 (61.4) 207 (62.0)

0.44 0.74 0.94 0.43 0.28

46 (13.8) 181 (54.2) 44 (13.2) 63 (18.9)

50 (15.0) 179 (53.6) 30 (8.98) 75 (22.5)

26 (7.78) 109 (32.6) 166 (49.7) 28 (8.38) 5 (1.5)

18 (5.39) 132 (39.5) 156 (46.7) 23 (6.89) 5 (1.5)

223 (66.8) 66 (19.8) 95 (28.4) 224 (67.1) 14 (4.19) 19 (5.69) 13 (3.89)

226 (67.7) 74 (22.2) 108 (32.3) 222 (66.5) 14 (4.19) 13 (3.89) 11 (3.29)

42.7 (40.1, 45.0) 1.0 (0.9, 1.2)

42.6 (40.0, 44.9) 1.0 (0.9, 1.2)

0.80 0.79

83 (24.9) 54 (16.2) 114 (34.1)

78 (23.4) 46 (13.8) 110 (32.9)

0.65 0.39 0.74

225 (67.4) 57 (17.1) 52 (15.6) 111 (87.5, 133) 80.0 (64.5, 99.0) 170 (50.9) 289 (86.5)

235 (70.4) 54 (16.2) 45 (13.4) 107 (87, 133) 80.0 (65.0, 98.5) 173 (51.8) 284 (85.0)

0.40 0.76 0.51 0.67 0.89 0.82 0.58

p Value

0.35

0.81 0.45 0.27 0.87 ⬎0.99 0.28 0.68

BSA ⫽ body surface area; CABG ⫽ coronary artery bypass grafting; COPD ⫽ chronic obstructive pulmonary disease; FFP ⫽ fresh frozen plasma; IABP ⫽ intraaortic balloon pump; ITA ⫽ internal thoracic artery; LAD ⫽ left anterior descending; LVEF ⫽ left ventricular ejection fraction; NYHA ⫽ New York Heart Association; RBC ⫽ red blood cells; RCA ⫽ right coronary artery.

ADULT CARDIAC

Variables

550

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

ADULT CARDIAC

were part of randomized, double-blinded, placebocontrolled studies in the phase III trial for licensure of aprotinin. Among these patients, 284 (14%) of the 1,720 patients received a platelet transfusion, 88% of patients transfused with platelets also received RBC transfusions, and patients receiving three or more RBC transfusions also received more platelet transfusions. Various CABGrelated adverse events relative to platelet transfusion were analyzed with the finding that platelet transfusion was associated with increased operative time and bleeding, more RBC transfusions, more surgical reoperations for bleeding, prolonged hospital duration, and an increased likelihood of postoperative infection, stroke, and death. Limitations of the authors’ reported findings included their inability to adjust for confounding relative to preoperative variables (eg, patients receiving platelet transfusions had significantly more prior MIs and worse systolic dysfunction), concomitant RBC transfusions, and aprotinin administration, which the authors acknowledged by noting platelet transfusion may rather serve as a surrogate marker for more critically ill patients and subsequent adverse outcomes. In 2005, Sreeram and colleagues [7] examined the rate of infectious complications in 2,657 adult primary CABG patients who received perioperative non-leukoreduced RBCs, FFP, and platelets. The authors concluded that while infectious complications were not increased by non-leukoreduced platelet or FFP transfusions, nonleukoreduced RBC transfusions conferred an increased risk of infectious morbidity. In 2006, Karkouti and colleagues [8] assessed whether leukoreduced platelet transfusions were associated with morbid events (low cardiac output syndrome, MI, stroke, renal failure, and sepsis) or mortality in their review of 11,459 patients undergoing a range of cardiac surgical procedures from 1999 to 2004. Despite measured adverse events being higher in the patients who received more platelet transfusions, there was no evident association between leukoreduced platelet transfusions and increased morbidity or mortality after risk adjustment. Finally, in 2007, Vamvakas [9] reported observational outcome data on the impact of non-leukoreduced platelet transfusions relative to postoperative infections on their previously published cohort of 416 low-risk patients undergoing CABG noted above. In contrast to Spiess and colleagues’ study, which suggested non-leukoreduced platelet transfusions were associated with an increased likelihood of postoperative infection, Vamvakas’ findings did not imply a relationship between non-leukoreduced platelet transfusions and increased postoperative infections. Our results suggest that platelet transfusion in cardiac surgery does not confer increased risk for morbidity or mortality congruent with four of these prior investigations [5, 7–9] and in contrast to the results of Spiess and colleagues [6]. Although our study is similar to previous investigations examining the role of platelet transfusions and adverse events in cardiac surgery, there are multiple key differences to highlight. First, in contrast to four of these prior studies, which limited their cohort to CABG patients only, our investigation involves the largest di-

Ann Thorac Surg 2008;86:543–53

verse cohort undergoing CABG and valve procedures to date. Second, a greater number and diverse set of potential confounders were appropriately adjusted for in our analysis. The number of RBC units transfused was not only appropriately balanced between the propensitymatched pairs by platelet transfusion status but also adjusted for in the multivariable modeling. Potential confounders of FFP and cryoprecipitate transfusions demonstrated significant risk for mortality in our multivariable regression model; these variables were eliminated from the propensity matching due to an inability to match patients on FFP and cryoprecipitate status. Given recent findings suggesting aprotinin use is associated with increased risk of renal dysfunction, MI or heart failure, stroke or encephalopathy, and long-term mortality after CABG surgery [15–17], our investigation is not confounded by the use of aprotinin compared with previous platelet transfusion outcomes studies [6, 8] because aprotinin was administered in less than one percent of the entire population under investigation. Third, more morbid outcomes in the setting of platelet transfusion status were examined in our investigation. Lack of an increased serious infection morbidity based on platelet transfusion status is an intriguing finding in our study. Negating the known potential infectious risks of platelets due to bacterial contamination, platelets play an integral role not only in hemostasis but also a prominent if not decisive role in wound healing. For example, upon activation, platelets secrete a myriad of proteins (eg, platelet-derived growth factor, transforming growth factor-␤) that cause a complex interplay of intracellular and extracellular events to occur including cellular chemotaxis, proliferation and differentiation, tissue debris elimination, angiogenesis, and collagen synthesis [18 – 21]. All of these cellular events in turn promote adequate wound healing long after fibrinolysis has occurred. In a review highlighting the uses of platelet-rich plasma, application of platelet-rich plasma or platelet releasate to oral, maxillofacial, plastic, and orthopedic surgery patients has been demonstrated to enhance significantly hard- and soft-tissue healing, and potentially even decrease postoperative infection rates [22]. Prior studies [23, 24] assessing use of platelet-rich plasma or gel in cardiac surgery focused not on infectious outcomes but rather on the impact of reducing bleeding and blood transfusions with conflicting results. While we clearly are not able to demonstrate a direct salutary effect of platelet transfusions in reducing postoperative infection in this nonrandomized study, platelet transfusions do not appear to increase the overall risk of postoperative infection. In contrast to RBC transfusions being the single most important risk factor associated with morbidity after isolated CABG [10, 11, 14], results of our study suggest platelet transfusions in cardiac surgery do not appear to represent a risk factor for adverse outcomes although the possibility of platelet transfusions serving as a surrogate marker for more critically ill patients cannot be excluded. Whether platelet transfusions confer equivocal adverse outcomes versus an actual benefit relative to reducing morbid outcomes remains indeterminate. It also remains

unclear whether leukoreduction of platelets versus RBCs may also modify the transfusion risks of these respective ABT products: conflicting data in the cardiac surgical population regarding the effects of leukoreduction has been reported [25–27] and the purported benefits of leukoreduction decreasing infections, organ dysfunction, or mortality have yet to be proven clinically [28]. Our center instituted universal leukoreduction in 2002. Due to the ethical concerns of randomizing platelet transfusions to cardiac surgical patients at risk for, or experiencing, active bleeding, this cohort investigation served as the next best approach to examine potential relationships between platelet transfusion and adverse clinical outcomes in cardiac surgery. However, inherent to nonrandomized studies, the inability to capture every intraoperative and postoperative process-of-care events and the potential for unobserved or unknown confounders may have influenced our reported findings. Many events and outcomes, however, are captured and adjusted for with “surrogate” variables. For example, increasing CPB times may reflect more challenging surgery; massive RBC transfusion may be associated with problems with sternal reentry or dissection. Platelet transfusion is not typically administered in response to complications per se but rather to microvascular bleeding after heparin reversal. Without an ability to randomize patients to platelet transfusion, we have attempted to control for variables known to be associated with adverse outcomes in the cardiac surgery population. Although unidentified variables may have changed over the 13year period of this patient population under investigation, date of surgery was adjusted for in both propensitymatched pairs and multivariable regression model. Another limitation intrinsic to all cohort investigations concerns that association or correlation does not prove causality. Conversely, lack of association between platelet transfusions and the outcomes we examined does not prove platelet transfusions are risk free. We also did not have sufficient detailed information regarding preoperative platelet counts, preoperative antiplatelet medication administration and specific sources, types, and volume of platelets administered to each individual patient (eg, single versus pooled donors, strictly leukoreduced, non-leukoreduced, or a combination thereof). Whether there exists a dose-dependent phenomenon effect of platelet transfusion volume or donor exposures relative to beneficial versus adverse outcomes remains unknown. In general, patients with thrombocytopenia and those receiving antiplatelet medications have a “lowered threshold” for platelet transfusion in the setting of microvascular bleeding in the post-CPB period. Platelet transfusion is not based on specific measure of platelet function testing in our institution. A low platelet count with normal function is rarely a problem but rather on-going microvascular bleeding is the primary reason for platelet transfusion in our institution. Nevertheless, even if platelet transfusions were administered more often to patients with thrombocytopenia or thrombocytopathies secondary to antiplatelet medications in our

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

551

study, there was no evident significant increased adverse outcome with platelet transfusion despite those patients transfused with platelets undergoing an increased frequency of surgical reexploration for bleeding. After risk adjustment, platelet transfusions did not confer any additional morbidity or mortality in this large cardiac surgical cohort. Our results should not be interpreted as advocating platelet transfusions in cardiac surgery; rather, platelet transfusions are not associated with morbidity in cardiac surgical patients when transfused empirically in the setting of persistent abnormal microvascular bleeding after heparin reversal. Whether there are potential benefits of platelet transfusion remains unknown and requires cautious and systematic exploration. Moreover, given that platelet transfusions are expensive and continue to be utilized often empirically in significant quantities in the cardiac surgical setting where a number of coagulation disorders can contribute to persistent postoperative bleeding, future studies should also entail standardizing perioperative point-of-care platelet function testing in order to establish more objective evidence-based platelet transfusion guidelines in the setting of abnormal microvascular bleeding and to direct more efficient platelet concentrate utilization in the cardiac surgical population.

All financial support was obtained within the Department of Cardiothoracic Anesthesia. There was no external source of funding for this project.

References 1. Harker LA, Malpass TW, Branson HE, Hessel EA, Slichter SJ. Mechanism of abnormal bleeding in patients undergoing cardiopulmonary bypass: acquired transient platelet dysfunction associated with selective alpha-granule release. Blood 1980;56:824 –34. 2. Cobain TJ, Vamvakas EC, Wells A, Titlestad K. A survey of the demographics of blood use. Transfusion Medicine 2007;17:1–15. 3. Goodnough LT, Johnston MF, Toy PT. The variability of transfusion practice in coronary artery bypass surgery. Transfusion Medicine Academic Award Group. JAMA 1991; 265:86 –90. 4. Ferraris VA, Ferraris SP, Saha SP, et al, Society of Thoracic Surgeons Blood Conservation Guideline Task Force; Spiess BD, Shore-Lesserson L, Stafford-Smith M, et al., Society of Cardiovascular Anesthesiologists Special Task Force on Blood Transfusion. Perioperative blood transfusion and blood conservation in cardiac surgery: the Society of Thoracic Surgeons and the Society of Cardiovascular Anesthesiologists clinical practice guideline. Ann Thorac Surg 2007; 83(5 suppl):S27– 86. 5. Vamvakas EC, Carven JH. Allogeneic blood transfusion and postoperative duration of mechanical ventilation: effects of red cell supernatant, platelet supernatant, plasma components, and total transfused fluid. Vox Sang 2002;82:141–9. 6. Spiess BD, Royston D, Levy JH, et al. Platelet transfusions during coronary artery bypass graft surgery are associated with serious adverse outcomes. Transfusion 2004;44:1143– 8. 7. Sreeram GM, Welsby IJ, Sharma AD, et al. Infectious complications after cardiac surgery: lack of association with fresh frozen plasma or platelet transfusions. J Cardiothorac Vasc Anesth 2005;19:430 – 4.

ADULT CARDIAC

Ann Thorac Surg 2008;86:543–53

552

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

Ann Thorac Surg 2008;86:543–53

ADULT CARDIAC

8. Karkouti K, Wijeysundera DN, Yau TM, et al. Platelet transfusions are not associated with increased morbidity or mortality in cardiac surgery. Can J Anesth 2006;53:279 – 87. 9. Vamvakas EC. Platelet transfusion and postoperative infection in cardiac surgery. Transfusion 2007;47:352– 4. 10. Koch CG, Li L, Duncan AI, et al. Morbidity and mortality risk associated with red blood cell and blood-component transfusion in isolated coronary artery bypass grafting. Crit Care Med 2006;34:1608 –16. 11. Koch CG, Li L, Duncan AI, et al. Transfusion in coronary artery bypass grafting is associated with reduced long-term survival. Ann Thorac Surg 2006;81:1650 –7. 12. Rosenbaum P, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41–55. 13. Parson L. Reducing bias in a propensity score matched-pair sample using greedy matching techniques. Proceedings of 26th Annual SAS Users Group International Conference 2001:214 – 6. 14. Michalopoulos A, Tzelepis G, Dafni U, Geroulanos S. Determinants of hospital mortality after coronary artery bypass grafting. Chest 1999;115:1598 – 603. 15. Mangano DT, Tudor JC, Dietzel C, for the Multicenter Study of Perioperative Ischemia (McSPI) Research Group and the Ischemia Research and Education Foundation. The risk associated with aprotinin in cardiac surgery. N Engl J Med 2006;354:353– 65. 16. Karkouti K, Beattie WS, Dattilo KM, et al. A propensity score case-control comparison of aprotinin and tranexamic acid in high-transfusion-risk cardiac surgery. Transfusion 2006;46: 327–38. 17. Mangano DT, Miao Y, Vuylsteke A, et al; for the Investigators of the Multicenter Study of Perioperative Ischemia Group and the Ischemia Research and Education Foundation. Mortality associated with aprotinin during 5 years following coronary artery bypass graft surgery. JAMA 2007; 297:471–9.

18. Bhanot S, Alex JC. Current applications of platelet gels in facial plastic surgery. Facial Plast Surg 2002;18:27–33. 19. Anitua E, Andia I, Ardanza B, Nurden P, Nurden AT. Autologous platelets as a source of proteins for healing and tissue regeneration. Thromb Haemost 2004;91:4 –15. 20. Buckwalter JA, Einhorn TA, Bolander ME, et al. Healing of musculoskeletal tissues. In: Rockwood CA Jr, Bucholz RW, Green DP, eds. Fractures in adults. Philadelphia, PA: Lippincott-Raven; 1996:261–304. 21. Marx RE. Platelet-rich plasma: evidence to support its use. J Oral Maxillofac Surg 2004;62:489 –96. 22. Eppley BL, Pietrzak WS, Blanton M. Platelet-rich plasma: a review of biology and applications in plastic surgery. Plast Reconstr Surg 2006;118:147e–59e. 23. DelRossi AJ, Cernaianu AC, Vertrees RA, et al. Platelet-rich plasma reduces postoperative blood loss after cardiopulmonary bypass. J Thorac Cardiovasc Surg 1990;100:281– 6. 24. Wajon P, Gibson J, Calcroft R, Hughes C, Hughes C, Thrift B. Intraoperative plateletpheresis and autologous platelet gel do not reduce chest tube drainage or allogeneic blood transfusion after reoperative coronary artery bypass graft. Anesth Analg 2001;93:536 – 42. 25. Bilgin YM, van de Watering LM, Eijsman L, et al. Doubleblind, randomized controlled trial on the effect of leukocytedepleted erythrocyte transfusion in cardiac valve surgery. Circulation 2004;109:2755– 60. 26. Wallis JP, Chapman CE, Orr KE, Clark SC, Forty JR. Effect of WBC reduction of transfused RBCs on post-operative infection rates in cardiac surgery. Transfusion 2002;42:1127–34. 27. van de Watering LM, Hermans J, Houbiers JG, et al. Beneficial effects of leukocyte depletion of transfused blood on postoperative complications in patients undergoing cardiac surgery: a randomized clinical trial. Circulation 1998;97: 562– 8. 28. Blajchman MA. The clinical benefits of the leukoreduction of blood products. J Trauma 2006;60(6 suppl):S83–90.

Appendix Baseline Variables for Unmatched and Propensity-Matched Patients Who Received Platelet Transfusions

Variables Demographics: Age (years) Gender (female) BSA (kg/M2) Cardiac comorbidity: History of atrial fibrillation Heart failure LV ejection fraction (normal vs abnormal) Myocardial infarction Preoperative IABP Emergency surgery Reoperation Coronary disease ⱖ70%: Left main Left anterior descending Left circumflex RCA

Matched Platelet Transfusion (n ⫽ 2,774)

Not Matched Platelet Transfusion (n ⫽ 825)

68.8 (61.1, 75.1) 824 (29.7) 1.95 (1.79, 2.10)

70.1 (62.2, 75.3) 226 (27.4) 1.92 (1.78, 2.07)

243 (8.76) 983 (35.4) 1251 (45.1) 1446 (52.8) 97 (3.5) 86 (3.1) 970 (35.0)

82 (9.94) 302 (36.6) 330 (43.1) 434 (52.6) 49 (6.4) 55 (6.67) 405 (49.1)

0.30 0.54 0.32 0.90 ⬍0.001 ⬍0.001 ⬍0.001

330 (11.9) 1964 (67.2) 1627 (58.7) 1727 (62.3)

109 (13.2) 564 (68.4) 468 (56.7) 540 (65.5)

0.31 0.53 0.33 0.09

p Value 0.07 0.20 0.029

Ann Thorac Surg 2008;86:543–53

McGRATH ET AL PLATELET TRANSFUSION AND MORBIDITY

553

Appendix Continued

NYHA functional class: I II III IV Canadian angina class: 0 1 2 3 4 Comorbidity: Hypertension Diabetes Peripheral vascular disease Smoking COPD Stroke Renal disease Preoperative laboratory values: Hematocrit (%) Creatinine (mg/dL) Preoperative valve pathology: Aortic valve regurgitation Aortic valve stenosis Mitral valve regurgitation Operative Factors: Procedure CABG Valve procedure Combined CABG/ Valve Cardiopulmonary bypass time Aortic clamp time (min) LITA or RITA use Surgery before 2003

Matched Platelet Transfusion (n ⫽ 2,774)

Not Matched Platelet Transfusion (n ⫽ 825)

p Value 0.024

370 (13.3) 1309 (47.2) 519 (18.7) 576 (20.8)

133 (16.1) 347 (42.1) 151 (18.3) 194 (23.5)

234 (8.44) 1,001 (36.1) 1,142 (41.2) 337 (12.1) 576 (20.8)

22 (2.67) 400 (48.5) 297 (36.0) 97 (11.8) 194 (23.5)

1,904 (68.6) 672 (24.2) 1,061 (38.2) 1,651 (59.5) 220 (7.93) 291 (10.5) 236 (8.51)

436 (67.4) 139 (20.5) 285 (34.5) 766 (59.1) 67 (8.75) 86 (10.4) 78 (9.45)

⬍0.001

39.8 (35.4, 43.0) 1.1 (0.9, 1.3)

39.8 (35.8, 43.0) 1.2 (1.0, 1.4)

781 (28.2) 591 (21.3) 1266 (45.6)

238 (28.8) 155 (18.8) 362 (43.9)

1593 (57.4) 499 (18) 682 (24.6) 115 (89, 144.5) 82 (62, 103) 1254 (45.2) 2514 (90.6)

437 (57) 127 (16.6) 202 (26.4) 126 (98.3, 161) 85 (64, 109) 299 (36.2) 812 (98.4)

0.54 0.039 0.054 0.85 0.46 0.96 0.40 0.71 ⬍0.001 0.70 0.12 0.37

0.85 0.37 0.32 ⬍0.001 0.003 ⬍0.001 ⬍0.001

ADULT CARDIAC

Variables