POSTOPERATIVE HEMORRHAGE in cardiac surgery is a

Transcription

1 Increased Chest Tube Drainage Is Independently Associated With Adverse Outcome After Cardiac Surgery Michael C. Christensen, MSc, MPA, DrPH,* Frank Dziewior, MD, Angela Kempel, MSc, and Christian von Heymann, MD, DEAA From *Regulatory Affairs, Novo Nordisk A/S, Søborg, Denmark; Klinik für Herzchirurgie, Herzzentrum Augsburg, Augsburg, Germany; Department of Pharmametrics, Institute for Health Economics and Epidemiology, Freiburg, Germany; and Klinik für Anästhesiologie mit Schwerpunkt operative Intensivmedizin, Charité-Universitätsmedizin Berlin, Berlin, Germany. Supported by Novo Nordisk A/S. M.C.C. is an employee of Novo Nordisk, A/S. A Kempel A.K. has received consulting fees from Novo Nordisk in regards to this study. Address reprint requests to Michael C. Christensen, MSc, MPA, DrPH, Regulatory Affairs Novo Nordisk A/S, Vandtårnsvej 114, 2860 Søborg, Denmark. mcrc@novonordisk.com 2012 Elsevier Inc. All rights reserved / $36.00/0 doi: /j.jvca Objective: To investigate the clinical relevance of specific volume criteria for hemorrhage in a patient population undergoing cardiac surgery with cardiopulmonary bypass (CPB). Design: A retrospective analysis; postoperative hemorrhage was defined by a fixed set of criteria >200 ml/h in any 1 hour or part thereof, or (2) >2 ml/kg/h for 2 consecutive hours in the first 6 hours after surgery. Classification and regression tree (CART) analysis were used to validate the results of the specific volume criteria. Multivariate regression analysis was applied to investigate the association of specific volume criteria for hemorrhage with clinical outcomes. Setting: A university hospital. Participants: All adult cardiac surgery patients undergoing surgery with CPB at the authors center in Interventions: None. Measurements and Main Results: A total of 1,188 patients underwent cardiac surgery, and 76 patients (6.4%) experi- POSTOPERATIVE HEMORRHAGE in cardiac surgery is a serious complication that independently is associated with significant postoperative adverse events. 1 The exact etiology has not been elucidated completely, yet existing studies identify a wide range of risk factors, including advanced age, nonelective surgery, low body surface area, prolonged cardiopulmonary bypass (CPB) time ( 150 minutes), combined valve and coronary artery bypass graft surgery, number of bypass grafts ( 5), re-exploration, and preoperative use of antiplatelet agents. 2-4 In clinical practice, hemodynamic instability caused by ongoing hemorrhage may be the most important reason for consecutive organ failure. The need for increased blood product replacement to compensate for blood loss has been associated with a range of postoperative complications, including renal failure, acute respiratory distress syndrome, sepsis, atrial fibrillation, severe infection, and a higher risk of death. 2,5-9 In addition to the clinical consequences, massively transfused patients and patients undergoing re-exploration place significant demands on hospital resources, including intensive care management and overall hospital length of stay enced postoperative hemorrhage according to the fixed criteria for blood loss. Blood loss as measured by these criteria was associated with a higher 30-day mortality (odds ratio [OR] 2.9, p < 0.001), incidence of stroke (OR 3.3, p ), re-exploration (OR , p < ), intensive care unit stay >72 hours (OR 1.3, p < ), and mechanical ventilation >24 hours (OR 3.4, p ). The clinical relevance of these criteria is supported by CART analysis. Conclusions: Postoperative hemorrhage (drainage loss) exceeding 200 ml/h in 1 hour or 2 ml/kg for 2 consecutive hours occurring within 6 hours after cardiac surgery is associated with higher 30-day mortality and other postoperative complications. Further research is needed to validate these results Elsevier Inc. All rights reserved. KEY WORDS: cardiopulmonary bypass, hemorrhage, coagulation, anticoagulation, outcomes, cardiac surgery Between 3% and 10% of all patients undergoing cardiac surgery with CPB experience (excessive) postoperative hemorrhage, 1 whereas 3% to 7% require a second operation to control hemorrhage. 5,6,12-15 The ability to identify patients at risk for complications clearly is important to avoid adverse outcomes and an unnecessary use of hospital resources. Interestingly, postoperative hemorrhage rarely has been defined and examined using specific volume criteria, but rather it has been characterized through surrogate markers like re-exploration because of hemorrhage or massive transfusion during and after cardiac surgery. 2-6 Because not all patients who show large transfusion requirements experience postoperative hemorrhage, 16 and not all patients with postoperative hemorrhage undergo re-exploration, surrogate markers may not quantify precisely the clinical outcomes associated with postoperative hemorrhage. Using specific volume criteria for hemorrhage may allow for the early detection of patients at high risk for adverse outcomes and thereby facilitate the use of appropriate multimodal treatment protocols to improve outcomes. The objective of this study was to identify which volume of postoperative drainage loss was associated with adverse outcomes in a patient population undergoing cardiac surgery with CPB. This was done by (1) examining the validity of the predefined volume criteria through multivariable logistic regression analysis, and (2) by classification and regression tree (CART) analysis to identify threshold levels for postoperative drainage loss associated with adverse clinical outcomes. METHODS The Quality Assurance Database of the Bundesgeschäftsstelle Qualitätssicherung (BQS) in Germany was established in 2001 to enhance the quality of all surgical interventions performed in the country. All cardiac surgery units in Germany annually submit a standard set of perioperative data related to all cardiac surgeries performed. In this study, data were obtained for all patients 18 years undergoing cardiac surgery at the Heart Center of the Klinikum Augsburg between January 46 Journal of Cardiothoracic and Vascular Anesthesia, Vol 26, No 1 (February), 2012: pp 46-51

2 CHEST TUBE DRAINAGE AND CARDIAC SURGERY 47 1, 2006, and December 31, A total of 1,233 patients underwent cardiac surgery with CPB at this hospital that year, and after checking for missing data, 1,188 patients (96%) were eligible for inclusion in the study. Aprotinin was used routinely at this institution, with the exception of patients with known symptomatic intolerance to this drug. The dosage of aprotinin administered was 1 million KIU before CPB, 1 million KIU during CPB, and 1 million KIU post-cpb (in total: 3 million KIU). In patients with a high risk of bleeding (eg, patients undergoing a second cardiac surgery, patients with endocarditis, and patients with an assumed CPB run 120 minutes), double the dose was administered (in total: 6 million KIU). The Institutional Review Board at Klinikum Augsburg approved the study (Institutional Review Board Number 45033). In patients undergoing cardiac surgery with CPB, postoperative hemorrhage was defined as bleeding into drains placed in the cardiothoracic cavity after admission to the intensive care unit (ICU). Drainage loss was assessed on an hourly basis after the completion of a 30-minute stabilization period. was defined by at least one of the following criteria: (1) 200 ml/h in any 1 hour or part thereof or (2) 2 ml/kg/h for 2 consecutive hours occurring in the first 6 hours after surgery. These criteria are similar to those previously applied in a study examining the safety and efficacy of recombinant activated factor VII for bleeding after cardiac surgery. 17 Blood loss during the stabilization period was not included in the definition of postoperative hemorrhage. Such loss may be caused by postural changes when transferring the patient from the operating room table to the bed or because of fluid in the pleural or mediastinal cavity, which may have arisen from the rinsing with water as an attempt to achieve surgical hemostasis. Drainage loss after this period (ie, first 6 hours after surgery) was not considered to be of microvascular origin because it most likely originates from a surgical source (eg, detached arterial clip). All perioperative data were collected from the BQS database for the study population at Klinikum Augsburg. Preoperative data included age, sex, body mass index (BMI), admitting diagnosis, the administration of preoperative anticoagulants, the administration of preoperative platelet inhibitors or vasoactive medication (nitrate intravenously and inotrope intravenously), the preoperative hematocrit level, preoperative renal dysfunction, comorbidities (eg, liver disease, gastrointestinal disease, endocrine disease, malignancy, or other), previous surgery, and the urgency of surgery (ie, elective, urgent, emergent, and resuscitation). Intraoperative variables included surgery time, CPB time, aortic cross-clamp time, the incidence of circulatory arrest, body temperature, the use of hemofiltration, the use of an intra-aortic balloon pump, the use of extracorporeal membrane oxygenation, the use of other ventricular or cardiac assist devices, and any intraoperative complications including low cardiac output and hypoxemia. Postoperative variables included any complications (eg, myocardial infarction [MI], stroke [ischemic and hemorrhagic], low cardiac output, cardiac tamponade, and coagulopathy), postoperative blood loss, re-exploration, survival up to day 30 after surgery, postoperative transfusion of packed red blood cells (termed blood transfusion throughout the article), ventilation support, the length of stay in the ICU, manpower requirements while in the ICU (measured by the Therapeutic Intervention Scoring System [TISS]-28 score), and the total length of hospital stay. Information on hourly blood loss was extracted manually from the patients medical records. The clinical outcomes examined in relation to postoperative hemorrhage were mortality at day 30 after surgery and postoperative complications defined as MI, stroke, and re-exploration. The definition of a myocardial infarction required new ECG changes, including new Q waves, a reduced R wave, and an increase of the creatine kinase (CK), or a persistent left or right branch block with an 8-fold elevation of the CK and a percentage of the myocardial isoenzyme of the CK-MB of 8%, or persistent changes of the ST segment 1 mm and an 8-fold elevation of the CK and a percentage of the myocardial isoenzyme of the CK-MB of 8%. Stroke was defined according to the modified Rankin scale 18 and re-exploration as reoperation for bleeding or for other reasons. Mechanical ventilation 24 hours and ICU stay 72 hours also were examined as surrogate markers for short-term clinical outcome. All postoperative complications were defined in compliance with the national definitions of the BQS as the Quality Assurance database at Klinikum Augsburg follows the same data structure as the national BQS database. The statistical analysis of the data was as follows. First, the perioperative characteristics of the study population were examined, and differences in patients with and without postoperative hemorrhage as defined by the fixed set of volume criteria were compared. Noncontinuous variables were compared using the chi-square test or the Fisher exact test as appropriate. Continuous variables were described by the mean standard deviation. The Mann-Whitney rank sum test was used to compare medians when a normality test failed. Second, CART analysis was applied to identify a threshold level for hemorrhage with an adverse outcome. CART analysis is a nonparametric decision tree technique that can be used to provide a simple and reliable model for risk stratification. 19 The technique has been used in a variety of clinical research settings for risk prediction of various clinical outcomes. 20,21 In the context of this study, CART analysis was applied for risk prediction of mortality at day 30 in relation to the postoperative blood loss. The association of postoperative hemorrhage with clinical outcomes (eg, postoperative complications, re-exploration, and death at day 30) and surrogate markers for clinical outcome (mechanical ventilation 24 hours and ICU stay 72 hours) were examined using multivariate logistic regression analysis. Two variants of analyses were performed to determine the association of blood loss with clinical outcome: (1) blood loss defined according to the fixed set of volume criteria ( 200 ml/h in any 1 hour or part thereof or 2 ml/kg/h for 2 consecutive hours occurring in the first 6 hours after surgery) and (2) threshold levels of blood loss identified through CART analysis. Adjustment was done for all pre- and intraoperative variables potentially relevant for the outcomes observed, including age, sex, BMI, the admitting diagnosis (eg, unstable angina, acute MI, valve disorder, or other underlying disorder), the use of preoperative antiplatelet medication, the use of intravenous nitrate and inotropes, preoperative hematocrit and creatinine levels, comorbidities, preoperative coagulopathy, type of surgery (ie, elective, urgent, and emergent), surgery time, CPB duration, aortic cross-clamp time, body temperature, the use of an intra-aortic balloon pump, extracorporeal membrane oxygenation or other assist device at the end of surgery, and any intraoperative complications such as low cardiac output and hypoxemia. Adjustment also was done for any postoperative blood transfusions because blood transfusion has been identified as a risk factor for adverse clinical outcome after cardiac surgery. 22,23 Finally, adjustment was done for any postoperative MIs observed as well as postoperative low cardiac output because both of these events also are known to adversely affect the clinical outcome after cardiac surgery 23,24 ; p 0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 16.0 (SPSS, Inc, Chicago, IL). RESULTS Of the 1,188 patients identified, 76 patients (6.4%) were identified with postoperative hemorrhage using the fixed volume criteria for blood loss. Descriptive statistics of the study population are provided in Tables 1 and 2. The mean ( standard deviation) blood loss during the first 24 hours after surgery was 1,669 ml ( 1,170) in the patients classified as patients with postoperative hemorrhage according to the fixed criteria compared with 299 ml ( 145 ml) in patients classified as without postoperative hemorrhage. Postoperative hemorrhage defined according

3 48 CHRISTENSEN ET AL Table 1. Preoperative Patient Characteristics Patient Characteristics Total Cohort (N 1,188) No Postoperative (N 1,112) Postoperative (N 76) p Value Blood loss vs No Blood Loss Mean age ( SD) 67.2 ( 10.8) 67.2 ( 10.8) 66.9 ( 2.1) Sex (n, %), Male 865 (72.8) 803 (72.2) 62 (81.6) Mean BMI ( SD) 27.6 ( 4.9) 27.6 ( 4.9) 27.1 ( 4.9) Admitting Diagnosis (n, %) Unstable angina 79 (6.6) 75 (6.7) 4 (5.3) Acute myocardial infarction 583 (49.1) 551 (49.6) 32 (42.1) Valve disorder 290 (24.4) 270 (36.8) 20 (26.3) Other underlying disorder 236 (19.9) 216 (19.7) 20 (26.3) Preoperative medication Preoperative antiplatelet medication (aspirin and clopidogrel) 663 (55.8) 623 (56.0) 40 (52.6) Preoperative anticoagulants (n, %) 263 (22.1) 246 (22.1) 17 (22.4) Nitrate IV (n, %) 55 (4.6) 52 (4.7) 3 (3.9) Inotrope IV (n, %) 22 (1.9) 19 (1.7) 3 (3.9) Comorbidities (n, %) Liver disease 61 (5.1) 53 (4.8) 8 (10.5) GI disease 68 (5.7) 65 (5.8) 3 (3.9) Endocrine disease 85 (7.2) 76 (6.8) 9 (11.8) Malignancy 35 (2.9) 33 (3.0) 2 (2.6) Acute infections 83 (7) 78 (7.0) 5 (6.6) Preoperative coagulopathy 17 (1.4) 13 (1.2) 4 (5.3) Preoperative HCT, mean ( SD) 40.0 ( 5.14) 40.1 ( 5.12) 39.1 ( 5.2) Preoperative renal dysfunction (n, %) 302 (25.4) 274 (24.6) 28 (36.8) Preoperative creatinine (mg/dl) 1.15 ( 0.6) 1.13 ( 0.6) 1.3 ( 0.6) Previous OR (n, %) 87 (7.3) 77 (6.9) 10 (13.2) Urgency (n, %) Elective 450 (37.9) 431 (38.8) 19 (25.0) Urgent 485 (40.8) 452 (40.6) 33 (43.4) Emergent 241 (20.3) 220 (19.8) 21 (27.6) Resuscitation 12 (1.0) 9 (0.8) 3 (3.9) NOTE. Numbers marked in bold indicate significant p values at a 5% significance level. Abbreviations: IV, intravenously; HCT, hematocrit; GI, gastrointestinal; OR, operations; SD, standard deviation. to the fixed volume criteria was significantly associated with a higher incidence of postoperative complications and death at day 30 after surgery (Table 3). Postoperative hemorrhage also was significantly associated with a higher level of re-exploration, prolonged ventilation, longer ICU stay, and greater manpower requirements while in the ICU (TISS-28 score). Using CART analysis, the threshold level for hemorrhage that most significantly predicted death at day 30 was 495 ml. Table 2. Intraoperative Patient Characteristics Surgery Characteristics Total Cohort (N 1,188) No Postoperative (N 1,112) Postoperative (N 76) p Value vs No Surgery time (min) ( 50.2) ( 46.2) ( 84.7) CPB duration (min) 78.3 ( 33.3) 77.1 ( 31.5) 96.3 ( 50.8) Aortic cross-clamp time (min) 51.7 ( 22.4) 51.7 ( 21.2) 61.9 ( 33.9) Deep hypothermic circulatory arrest (%) 24 (2.0) 20 (3.9) 4 (5.3) Body temperature* 34.8 ( 2.9) 34.9 ( 2.8) 33.8 ( 3.5) Use of IABP (%) 40 (3.4) 34 (3.1) 6 (7.9) Use of ECMO (%) 3 (0.3) 2 (0.2) 1 (1.3) Any intraoperative complications (%) 38 (3.2) 30 (2.7) 8 (10.5) Intraoperative low cardiac output (%) 16 (1.3) 13 (1.2) 3 (3.9) Intraoperative other complications (%) 20 (1.7) 16 (1.4) 4 (5.3) NOTE. Numbers marked in bold indicate significant p values at a 5% significance level. Abbreviations: CPB, cardiopulmonary bypass; OR, operating room; IABP, intra-aortic balloon pump; ECMO, extracorporeal membrane oxygenation. *Nadir temperature during CPB. The source of temperature monitoring was a Foley catheter. Intraoperative OR complications included low cardiac output, hypoxemia, and other intraoperative OR complications.

4 CHEST TUBE DRAINAGE AND CARDIAC SURGERY 49 Table 3. Clinical Outcomes: Postoperative Defined According to Fixed Volume Criteria Versus No Postoperative Outcomes Total Cohort (N 1,188) No Postoperative (N 1,112) Postoperative (N 76) p Value Postoperative complications (overall), n (%) 75 (6.3) 60 (5.4) 15 (19.7) < Postoperative MI 17 (1.5) 14 (1.3) 3 (3.9) Postoperative stroke 58 (4.9) 46 (4.1) 12 (15.8) Postoperative low cardiac output 65 (5.5) 54 (4.9) 11 (14.5) Postoperative tamponade 43 (3.6) 34 (3.1) 9 (11.8) Postoperative coagulopathy 16 (1.3) 7 (0.6) 9 (11.8) < day mortality 78 (6.6) 61 (5.5) 17 (22.4) < Re-exploration 84 (7.1) 38 (3.4) 45 (59.2) Re-exploration for bleeding 47 (4.0) 9 (0.8) 38 (50) < Re-exploration for other causes 37 (3.0) 29 (2.6) 7 (9.2) Postoperative ventilation 24 h 147 (12.4) 120 (10.8) 27 (35.5) Hemofiltration 188 (15.8) 165 (14.8) 23 (30.3) Mean postoperative blood loss/chest tube drainage 549 ( 941) 472 ( 873) 1,669 ( 1,170) < Postoperative transfusion of blood products 463 (39.0) 388 (34.9) 64 (84.2%) < RBC 452 (38.0) 388 (24.9) 64 (84.2%) < FFP 95 (8.0) 56 (5.0) 39 (51.3%) < Platelets 39 (3.3) 19 (1.7) 19 (25.0%) < ICU stay 72 h (n,%) 290 (24.4) 251 (22.6) 39 (51.3%) < Total LOS ICU (d) 4.1 ( 4.7) 3.1 ( 5.0) 5.7 ( 5.3) < TISS-28 Score 66.9 ( 69.5) 44.7 ( 52.8) 85.9 ( 76.5) < Total Hospital LOS (d) 17.4 ( 11.6) 17.2 ( 11.2) 19.2 ( 15.8) NOTE. Numbers marked in bold indicate significant p values at a 5% significance level. Abbreviations: MI, myocardial infarction; RBC, packed red blood cells; FFP, fresh frozen plasma; ICU, intensive care unit; LOS, length of stay; TISS-28, Therapeutic Intervention Scoring System-28. This implies that continuous hemorrhage 495 ml during the first 24 hours after surgery was significantly associated with adverse clinical outcomes compared with postoperative blood loss below this level. Applying the same multivariate regression analysis to the predefined criteria for hemorrhage and the threshold identified by CART analysis ( 495 ml), postoperative hemorrhage was significantly associated with higher mortality at day 30, incidence of re-exploration for bleeding, ICU stay 72 hours, and mechanical ventilation 24 hours (Table 4). Postoperative hemorrhage according to the fixed volume criteria also was significantly associated with a higher risk of stroke during hospitalization. DISCUSSION Postoperative hemorrhage exceeding 200 ml/h in any 1 hour or part thereof or 2 ml/kg in 2 consecutive hours occurring within the first 6 hours after cardiac surgery is significantly associated with higher mortality at day 30, incidence of reexploration, and prolonged mechanical ventilation ( 24 hours). To the authors knowledge, this is the first study that examines the association between specific volume criteria for blood loss and clinical outcomes. Only one very small study previously has examined the consequences of postoperative hemorrhage in a population defined by volume criteria. 11 In this study, Herwaldt et all 11 found that 75 coronary artery bypass graft surgery patients with postoperative hemorrhage stayed in the hospital for a significantly longer period at significantly higher hospital costs. No significant difference in survival was observed. In comparison with the present study, this investigation used a case-control study design and considered a small study population constructed from different time periods in the early 1990s, which may have biased the results. More importantly, however, their definition of postoperative hemorrhage included patients undergoing re-exploration for bleeding as well as patients identified according to volume criteria. In contrast, the present analysis investigated a much larger patient population using specific volume criteria for inclusion only. Specific volume criteria of postoperative blood loss should be considered a more reliable definition of bleeding because surrogate markers, such as re-exploration for bleeding, may not capture patients who respond to the usual standard of care. The present authors aimed to validate the clinical relevance of the fixed-volume criteria by CART analysis. The CART analysis identified a threshold level for blood loss that is slightly higher than that implied by the fixed-volume criteria. The fixed-volume criteria (200 ml/h in any 1 hour or any part thereof or 2 ml/kg for 2 consecutive hours in any hour in the first 6 hours after surgery) is stricter than the threshold identified through CART analysis because it requires that patients at, for example, lb of body weight must lose a minimum of 320 to 400 ml of blood within the first 6 hours to be classified as a patient with postoperative hemorrhage according to this definition. However, the threshold level identified by the CART analysis does indicate that the specific volume criteria applied in this analysis are indeed an acceptable definition of postoperative hemorrhage and may serve as a clinical prediction rule. In clinical practice, the early identification of bleeding patients using the threshold of 200 ml in any 1 of 6 hours after

5 50 CHRISTENSEN ET AL Table 4. Multivariate Analysis of Postoperative Outcome p Value OR 95% CI 30-day mortality* criteria ml MI* criteria ml Stroke* criteria ml Re-exploration for bleeding* criteria ml ICU 72 h* criteria ml Ventilation 24 h* criteria ml *Adjustment was done for the following variables in the multivariate analysis: age, sex, BMI, admitting diagnosis (unstable angina, acute MI, valve disorder, other underlying disorder), use of preoperative antiplatelet medication, use of intravenous nitrate and inotropes, preoperative hematocrit and creatinine levels, comorbidities, preoperative coagulopathy, type of surgery (ie, elective, urgent, emergent), surgery time, CPB duration, aortic cross-clamp time, body temperature, use of IABP, ECMO or other assist device at the end of surgery, any intraoperative complications such as low cardiac output and hypoxemia, postoperative blood transfusion, and postoperative MIs and postoperative low cardiac output. admission to the ICU (or alternatively 495 ml in the first 24 hours) may allow for the early detection of patients at high risk for adverse outcomes and facilitate the use of appropriate multimodal treatment protocols to improve outcomes after cardiac surgery. 3,4 The present analysis was not designed to draw definitive conclusions on the clinical impact of an elevated drainage loss. However, the authors believe that these relatively low thresholds of blood loss may draw the attention of the attending physician toward the higher risk of these patients to experience an adverse outcome. The therapeutic measures to be taken may include an earlier use of blood products that should be assisted by laboratory support. (For example, the administration of an antifibrinolytic in case thromboelastography shows elevated fibrinolytic activity or the use of platelet concentrates in case the platelet count is low.) Earlier re-exploration may be of benefit for these patients because this may sufficiently reduce blood loss. The indication for taking patients back to the operating room relies completely on the cardiac surgeon s view if he/she is convinced that there is any benefit from extended surgical hemostasis. Compared with surrogate markers for postoperative hemorrhage, the use of specific volume criteria for postoperative blood loss also offers a more precise measurement of the actual risk associated with postoperative hemorrhage in cardiac surgery. A number of studies using transfusion requirements as a surrogate marker have reported odds ratios of 8.1 for in-hospital mortality and 6.7 for mortality at day 30 after surgery compared with patients without the need for transfusion. 2,8 Transfusion requirements are not necessarily associated with excessive blood loss because anemia, low BMI (associated with a higher degree of hemodilutional anemia during CPB), and a high variability in transfusion practice may influence the actual number of blood products transfused. 4,16,25-27 Studies using re-exploration as a surrogate marker for postoperative hemorrhage have reported odds of death on the order of 2.0 to 3.0 compared with patients not undergoing re-exploration. 5,6 The present data clearly show that re-exploration was not performed in every patient with postoperative bleeding, which may be explained by the fact that only patients with refractory and persistent bleeding require surgical hemostasis. 17 Therefore, using re-exploration for bleeding is a rather imprecise marker for postoperative hemorrhage. Although the association between postoperative blood transfusion and death in patients undergoing cardiac surgery with CPB has been established in several studies, it is noteworthy that the association between postoperative hemorrhage and death at day 30 in the present study using specific volume criteria for postoperative blood loss was significant regardless of the postoperative blood transfusion administered. In order words, in the present study, the higher mortality observed in patients with postoperative hemorrhage was not driven by a higher use of packed red blood cells postoperatively. These results underscore that an even relatively moderate blood loss of 200 ml in any hour or part thereof in the first 6 hours after surgery may be significantly associated with adverse or fatal outcomes and require close monitoring and/or early treatment. Although the pathophysiologic explanation for the results was not investigated by the present analysis, it might be hypothesized that a drainage loss of 495 ml in 24 hours is associated with more blood transfusions, more hypovolemia, and systemic hypotension resulting in secondary organ failure, such as prolonged respiratory support, the need for renal replacement therapy, and a higher incidence of systemic inflammatory response syndrome. This study has a number of limitations. First, retrospective analysis is subject to usual investigator bias and to errors in manual extraction from the medical records. Therefore, not all the relevant factors for total and postoperative blood loss may have been available for this analysis, including data on the level of intraoperative transfusion of blood and blood components. Nevertheless, the measurement of postoperative blood loss relies on an hourly documentation in the medical records and is a much more precise measure of postoperative hemorrhage than any of the surrogates markers (ie, postoperative blood transfusion or re-exploration) so far applied in similar clinical research. Second, the results cannot prove causality between the volumes of blood loss and postoperative outcomes given the retrospective nature of the analysis. Third, the authors did not differentiate the results in regard to the causes of bleeding (ie, examined whether the cause of bleeding was surgical or caused by coagulopathy). Because previous research has indi-

6 CHEST TUBE DRAINAGE AND CARDIAC SURGERY 51 cated a significantly higher morbidity and mortality in patients suffering from coagulopathy than in patients with surgical bleeding, 7 it would have been interesting to quantify this effect in the present results. Fourth, results from the primary analysis originate from a relatively small number of patients (n 76) although these results have been shown to be robust in CART analysis. Fifth, data for this study were obtained from only one cardiac surgery center, which limits the generalizability of the findings. Therefore, the findings should be considered hypothesis generating, prompting re-evaluation in an international and prospective setting. Sixth, it should be noted that the data collection was performed in patients treated with aprotinin, which may have influenced the threshold of blood loss. This implies that the present results are not to be generalized to other antifibrinolytics and may yield different results using other antifibrinolytic medications. Future studies should take such treatment into account and re-evaluate the present findings with adjustment for different antifibrinolytic treatment (eg, tranexamic acid and -aminocaproic acid). CONCLUSIONS Postoperative hemorrhage exceeding 200 ml/h in 1 hour or 2 ml/kg for 2 consecutive hours occurring within 6 hours after cardiac surgery is significantly associated with higher 30-day mortality and other postoperative complications. Further research is needed to validate these results. 1. Whitlock R, Crowther MA, Ng HJ: Bleeding in cardiac surgery: Its prevention and treatment An evidence-based review. Crit Care Clin 21: , Karkouti K, Wijeysundera DN, Yau TM, et al: The independent association of massive blood loss with mortality in cardiac surgery. Transfusion 44: , Sellman M, Intonti MA, Ivert T: Reoperations for bleeding after coronary artery bypass procedures during 25 years. Eur J Cardiothorac Surg 11: , Romagnoli S, Bevilacqua S, Gelsomino S, et al: Small-dose recombinant activated factor VII (NovoSeven) in cardiac surgery. Anesth Analg 102: , Dacey LJ, Munoz JJ, Baribeau YR, et al: Reexploration for hemorrhage following coronary artery bypass grafting: Incidence and risk factors. Northern New England Cardiovascular Disease Study Group. Arch Surg 133: , Moulton MJ, Creswell LL, Mackey ME, et al: Reexploration for bleeding is a risk factor for adverse outcomes after cardiac operations. J Thorac Cardiovasc Surg 111: , Hall TS, Brevetti GR, Skoultchi AJ, et al: Re-exploration for hemorrhage following open heart surgery differentiation on the causes of bleeding and the impact on patient outcomes. Ann Thorac Cardiovasc Surg 7: , Murphy GJ, Reeves BC, Rogers CA, et al: Increased mortality, postoperative morbidity, and cost after red blood cell transfusion in patients having cardiac surgery. Circulation 116: , 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 34: , Christensen MC, Krapf S, Kempel A, et al: Costs of excessive postoperative hemorrhage in cardiac surgery. J Thorac Cardiovasc Surg 138: , Herwaldt LA, Swartzendruber SK, Zimmerman MB, et al: after coronary artery bypass graft procedures. Infect Contr Hosp Epidemiol 24:44-50, Karkouti K, O Farrell R, Yau TM, et al: Prediction of massive blood transfusion in cardiac surgery. Can J Anaesth 53: , Daly DJ, Myles PS, Smith JA, et al: Anticoagulation, bleeding and blood transfusion practices in Australasian cardiac surgical practice. Anaesth Intensive Care 35: , 2007 REFERENCES 14. Karthik S, Grayson AD, McCarron EE, et al: Reexploration for bleeding after coronary artery bypass surgery: Risk factors, outcomes, and the effect of time delay. Ann Thorac Surg 78: , Charalambous CP, Zipitis CS, Keenan DJ: Chest reexploration in the intensive care unit after cardiac surgery: A safe alternative to returning to the operating theater. Ann Thorac Surg 81: , Karkouti K, Wijeysundera DN, Beattie WS, et al: Risk associated with preoperative anemia in cardiac surgery: A multicenter cohort study. Circulation 117: , Gill R, Herbertson M, Vuylsteke A, et al: Safety and efficacy of recombinant activated factor VII: A randomized placebo-controlled trial in the setting of bleeding after cardiac surgery. Circulation 120: 21-27, Bonita R, Beaglehole R: Modification of Rankin Scale: Recovery of motor function after stroke. Stroke 19: , Breiman L, Friedman JH, Olshen RA: Classification and Regression trees. Belmont, CA, Wadsworth, Altamirano J, Zapata L, Agustin S, et al: Predicting 6-week mortality after acute variceal bleeding: Role of Classification and Regression tree analysis. Ann Hepatol 8: , Takahashi O, Cook EF, Nakamura T, et al: Risk stratification for in-hospital mortality in spontaneous intracerebral haemorrhage: A Classification and Regression tree analysis. QJM 99: , Engoren MC, Habib RH, Zacharias A, et al: Effect of blood transfusion on long-term survival after cardiac operation. Ann Thorac Surg 74: , Koch CG, Li L, Duncan AI, et al: Transfusion in coronary artery bypass grafting is associated with reduced long-term survival. Ann Thorac Surg 81: , Surgenor SD, Kramer RS, Olmstead EM, et al: The association of perioperative red blood cell transfusions and decreased long-term survival after cardiac surgery. Anesth Analg 108: , Rogers MA, Blumberg N, Saint S, et al: Hospital variation in transfusion and infection after cardiac surgery: A cohort study. BMC Med 7:37, Snyder-Ramos SA, Möhnle P, Weng YS, et al: The ongoing variability in blood transfusion practices in cardiac surgery. Transfusion 48: , Bennett-Guerrero E, Zhao Y, O Brien SM, et al: Variation in use of blood transfusion in coronary artery bypass graft surgery. JAMA 304: , 2010