Reexploration for Bleeding After Coronary Artery Bypass Surgery: Risk Factors, Outcomes, and the Effect of Time Delay

Transcription

1 After Coronary Artery Bypass Surgery: Risk Factors, Outcomes, and the Effect of Time Delay Shishir Karthik, FRCS, Antony D. Grayson, BS, Emer E. McCarron, FRCS, D. Mark Pullan, FRCS(C-Th), and Michael J. Desmond, FRCA Departments of Cardiothoracic Surgery, Research and Development, and Cardiothoracic Anesthesia, the Cardiothoracic Centre Liverpool, Liverpool, United Kingdom Background. We aimed to identify risk factors for reexploration for bleeding after surgical revascularization in our practice. We also looked at the impact of resternotomy and the effect of time delay on mortality and other in-hospital outcomes. Methods. In all, 2,898 consecutive patients undergoing coronary artery bypass grafting between April 1999 and March 2002 were retrospectively analyzed from our cardiac surgery registry. Multivariate logistic regression analysis was used to identify risk factors for reexploration for bleeding. To assess the effect of preoperative aspirin and heparin, reexploration patients were propensity matched with unique patients not requiring reexploration. We carried out a casenote review to ascertain the timing and causes for bleeding in patients undergoing resternotomy. Results. Eighty-nine patients (3.1%) underwent reexploration for bleeding. Multivariate analysis revealed smaller body mass index (p 0.003), nonelective surgery (p 0.022), 5 or more distal anastomoses (p 0.035), and increased age (p 0.041) to have increased risks. Propensity-matched analysis showed that preoperative use of aspirin (p 0.004) and heparin (p 0.001) were associated with increased risk in the on-pump coronary surgery group only. Patients requiring resternotomy had a significantly greater need for inotropic agents (p < 0.001), and longer intensive care unit stay (p < 0.001) and postoperative stay (p < 0.001) than their propensity-matched controls. However, there was no significant difference in the mortality rate. Adverse outcomes were significantly higher when patients waited more than 12 hours after return to the intensive care unit for resternotomy. Conclusions. Risk factors for reexploration for bleeding after coronary artery bypass grafting include older age, smaller body mass index, nonelective cases, and 5 or more distal anastomoses. Preoperative aspirin and heparin were risk factors for the on-pump coronary artery surgery group. Patients needing reexploration are at higher risk of complications if the time to reexploration is prolonged. Policies that promote early return to the operating theater for reexploration should be encouraged. (Ann Thorac Surg 2004;78:527 34) 2004 by The Society of Thoracic Surgeons The proportion of patients undergoing coronary artery bypass graft surgery (CABG) that requires reexploration for bleeding has been reported to be between 2% and 6% [1 3]. Resternotomy for bleeding after CABG is an important source of morbidity in many cardiac units. We have previously shown that reexploration for bleeding is associated with deep and superficial wound infections [4]. Such patients are often hemodynamically unstable and require urgent or emergent resternotomy and are also at greater risk from the various hazards of blood and blood products [1]. Older age, smaller body mass index (BMI), longer cardiopulmonary bypass (CPB) times, greater number of distal anastomoses, and the use of internal mammary artery have been associated with greater risk of resternotomy for bleeding [1, 3, 5, 6]. There have been several changes in the practices associated with cardiac surgery over the last decade, and Accepted for publication Feb 20, Address reprint requests to Dr Karthik, Leeds General Infirmary, Great George St, Leeds LS17 8AP, UK; suchkats@yahoo. com. many of these are quite likely to have some impact on the rates of reexploration after CABG. Among the many changes that are likely to affect the hematological profile of the patient after cardiac surgery are the preoperative use of aspirin and clopidogrel, an increased number of unstable patients undergoing surgery, many of whom are receiving heparin infusions, off-pump CABG, a greater number of older and sicker patients having surgery, and varying practices with regard to the use of antifibrinolytic drugs. We performed a retrospective study at our center with a view to ascertain the risk factors that are associated with increased rates of reexploration after CABG. We also examined the effect of reexploration on adverse outcomes and the impact of delaying return to the operating theater. Material and Methods Patient Population and Data In all 2,898 consecutive patients undergoing CABG between April 1, 1999, and March 31, 2002, at the Cardiothoracic Center-Liverpool had data collected prospec by The Society of Thoracic Surgeons /04/$30.00 Published by Elsevier Inc doi: /j.athoracsur

2 528 KARTHIK ET AL Ann Thorac Surg REEXPLORATION FOR BLEEDING 2004;78: Abbreviations and Acronyms BMI body mass index CABG coronary artery bypass grafting CI confidence intervals CPB cardiopulmonary bypass IABP intra-aortic balloon pump ICU intensive care unit IMA internal mammary artery OR odds ratio tively into our cardiac surgery registry. All patients undergoing CABG at our center, irrespective of the use of CPB, are fully heparinized, namely, they receive 300 U per kilogram body weight of heparin, irrespective of whether they are having off-pump or on-pump surgery. Patients undergoing CABG along with heart valve repair or replacement, resection of a ventricular aneurysm or other surgical procedure were not included. Data collection methods and definitions have been described in detail previously [7]. The preoperative and operative characteristics collected are listed in Tables 1 and 2, respectively. The primary outcome measure for our study was reexploration for bleeding, which was defined as bleeding that required surgical reexploration after initial departure from the operating theater. The decision to perform resternotomy was made by the consultant surgeon responsible based on the protocol promulgated by Kirklin and Barratt-Boyes [8]. Although these criteria were followed for a majority of the cases, it was not so for all of them. The criteria are as follows: (1) drainage of more than 500 ml during the first hour, more than 400 ml during each of the first 2 hours, more than 300 ml during each of the first 3 hours, or more than 1000 ml in total in the first 4 hours (2) sudden massive bleeding; (3) obvious signs of cardiac tamponade; (4) excess bleeding despite correction of coagulopathies; and (5) cardiac arrest in a patient who continues to bleed. Outcomes collected, after reexploration, are listed in Table 3. In-hospital mortality was defined as death within the same hospital admission regardless of cause. All patients transferred from the base hospital to another hospital were followed up to confirm their status at discharge. Criteria for defining sternal wound infections were in accord with the published evidence-based guidelines by the Centers for Disease Control and Prevention [9]. Postoperative stroke was defined as a new focal neurologic deficit or comatose state occurring postoperatively that persisted for more than 24 hours after its onset. We excluded confused states, transient events, and intellectual impairment from our study to avoid any subjective bias. Renal failure was defined as patients with a postoperative creatinine level greater than 200 mol/l or patients requiring dialysis. A retrospective case note review was carried out to collect data not routinely collected in our cardiac surgery Table 1. Preoperative Characteristics Patients (Number) (Number) p Value Age at operation (years) (917) 2.0 (18) (636) 3.3 (21) (620) 3.7 (23) (483) 3.3 (16) (242) 4.6 (11) Sex Female 19.9 (577) 3.8 (22) Male 80.1 (2321) 2.9 (67) Body mass index (kg/m 2 ) (633) 4.7 (30) (1427) 2.7 (38) (665) 2.7 (18) (173) 1.7 (3) Angina class IV No 67.1 (1944) 2.8 (55) Yes 32.9 (954) 3.6 (34) Diabetes No 82.3 (2385) 3.0 (71) Yes 17.7 (513) 3.5 (18) Respiratory disease No 67.9 (1969) 3.1 (62) Yes 32.1 (929) 2.9 (27) Peripheral vascular disease No 85.8 (2487) 3.0 (74) Yes 14.2 (411) 3.7 (15) Cerebrovascular disease No 91.6 (2653) 3.1 (81) Yes 8.4 (245) 3.3 (8) Renal dysfunction No 97.2 (2818) 3.1 (87) Yes 2.8 (80) 2.5 (2) Preoperative serum creatinine (mmol/l) (1504) 3.2 (48) (1316) 3.0 (39) (78) 2.6 (2) Prior cardiological intervention No 94.9 (2751) 3.2 (87) Yes 5.1 (147) 1.4 (2) Ejection fraction 30% No 91.9 (2662) 3.0 (81) Yes 8.1 (236) 3.4 (8) database in all patients who required reexploration for bleeding and a propensity-matched group of patients who did not (see Statistical Analysis). This case note review provided information on preoperative use of aspirin and heparin. Preoperative aspirin or heparin use was defined as continuing the medications until the day of surgery. This casenote review also provided informa-

3 Ann Thorac Surg KARTHIK ET AL 2004;78: REEXPLORATION FOR BLEEDING 529 Table 1. Continued Patients (Number) (Number) p Value Three-vessel disease No 19.5 (565) 2.8 (16) Yes 80.5 (2333) 3.1 (73) Left main stem stenosis 50% No 79.3 (2298) 3.3 (75) Yes 20.7 (600) 2.3 (14) Priority of surgery Elective 82.0 (2376) 2.6 (63) Urgent 16.6 (482) 4.8 (23) Emergency 1.4 (40) 7.5 (3) Prior cardiac surgery No 97.4 (2823) 3.1 (87) Yes 2.6 (75) 2.7 (2) Data shown as percentages and comparisons made with chi-square tests for trend. tion on indications for reexploration, the source of bleeding, time from initial departure from theater to reexploration, and blood loss before reexploration. Table 2. Operative Characteristics Patients p Value Use of CPB No 25.6 (743) 2.6 (19) Yes 74.4 (2155) 3.3 (70) CPB duration (min), CPB only (48) 2.1 (1) (423) 2.6 (11) (837) 3.2 (27) (587) 3.7 (22) (260) 3.5 (9) Cross-clamp duration (min), CPB only (1750) 3.2 (56) (405) 3.5 (14) Use of LIMA No 8.9 (259) 1.5 (4) Yes 91.1 (2639) 3.2 (85) Number of grafts (387) 2.6 (10) (1137) 2.9 (33) (998) 2.8 (28) (376) 4.8 (18) Data shown as percentages and comparisons made with chi-square tests for trend. CPB cardiopulmonary bypass; LIMA left internal mammary artery. Table 3. Outcomes in Propensity-Matched Groups (n 84) Control (n 84) p Value Mortality (%) 1.2 (1) 3.6 (3) Sternal wound infection (%) 9.5 (8) 4.8 (4) Stroke (%) 3.6 (3) 6 (5) Renal failure (%) 9.5 (8) 2.4 (2) IABP support (%) 8.3 (7) 2.4 (2) Inotrope support (%) 71.4 (60) 35.7 (30) Mechanical ventilation 32.1 (27) 7.1 (6) hours (%) ICU stay (days) Postoperative stay (days) Categorical data shown as percentages. Continuous data shown as median values. IABP intra-aortic balloon pump; ICU intensive care unit. Statistical Analysis Due to nonnormal distributions (tested using the Shapiro-Wilk test), continuous variables are shown as median with 25th and 75th percentiles and comparisons were made with Wilcoxon rank-sum tests as appropriate. Categorical variables are shown as a percentage and comparisons were made with 2 tests. Standard statistical methods were used to calculate odds ratios (OR) and 95% confidence intervals (CI). Multivariate logistic regression analysis was used, with forward stepwise techniques, to identify risk factors for reexploration for bleeding and to adjust for potentially confounding variables, including surgeon, anesthetist and time of operation (by year) [10]. If continuous variables were found to be significant, they were then assessed as dichotomous variables (eg, BMI 25, 25 to 30, and so forth). Continuous variables were either included as dichotomous variables or in their original form in the final model depending on the strength of association. In all cases a p value less than 0.05 was considered significant. All statistical analysis was performed retrospectively with SAS for Windows Version 8.2 (SAS Institute, Cary, NC). In order to assess the effect of preoperative aspirin and heparin use, plus the impact of reexploration for bleeding on outcomes, we performed a case note review on propensity-matched patients. Patients requiring reexploration for bleeding were propensity-matched with unique patients not reexplored. To do this, logistic regression [10] was used to develop a propensity score for reexploration for bleeding for all patients who underwent CABG [11]. The propensity score was constructed using all the variables listed in Table 4. The C statistic, which is equivalent to the receiver-operating characteristic curve, for this model was 0.76 [12]. Patients reexplored were matched with patients not reexplored who had an identical 5-digit propensity score. If this could not be done, we then proceeded to a 4-, 3-, 2-, or 1-digit match [13]. 5-digit matches were achieved in 31 of the patients reexplored, while 46 were matched with a 4-digit propensity score and 7 were matched with a 3-digit

4 530 KARTHIK ET AL Ann Thorac Surg REEXPLORATION FOR BLEEDING 2004;78: Table 4. Patient Characteristics in Propensity-Matched Groups (n 84) Control (n 84) p Value Age at operation (years) Female sex (%) 23.8 (20) 27.4 (23) Prior cardiac surgery (%) 1.2 (1) 3.6 (3) Left main stem stenosis 11.9 (10) 13.1 (11) % (%) Three-vessel disease (%) 82.1 (69) 79.8 (67) Ejection fraction 30% (%) 9.5 (8) 13.1 (11) Nonelective surgery (%) 29.8 (25) 20.2 (17) Peripheral vascular disease (%) 16.7 (14) 15.5 (13) Diabetes (%) 21.4 (18) 16.7 (14) Body mass index (kg/m 2 ) Renal dysfunction (%) 1.2 (1) 3.6 (3) Respiratory disease (%) 30.9 (26) 17.9 (15) Number of grafts (graft per patient) Off-pump CABG (%) 22.6 (19) 14.3 (12) Categorical data shown as percentages. Continuous data shown as median values. CABG coronary artery bypass grafting. match. 5 patients who were reexplored for bleeding could not be matched and hence, they were not included in the propensity-matched study. Results Incidence Overall, of the 2,898 patients in the study, 89 (3.1%) patients required reexploration for bleeding after isolated CABG. The incidence of reexploration has increased from 2.9% in April 1999% to 4.3% in March 2002 (p 0.001). Of the 89 patients reexplored, the primary indication for reexploration was bleeding (n 75). Other indications included 10 cases of tamponade, 2 cardiac arrests with bleeding, and 2 others. The major sites of bleeding were as follows: 42 graft/anastomoses, 23 sternal/left internal mammary artery bed, 6 extra-pericardial, 2 cannulation sites, and 16 unspecified. The median time to reexploration after initial departure from theater was 8.5 hours (25th and 75th percentiles: 5 to 12 hours), with the longest time being 168 hours. Median blood loss before reexploration was 1,680 ml (25th and 75th percentiles: 1,050 ml to 2,340 ml), with the greatest blood loss being 6,480 ml. Eight patients required more than one reexploration. Patient Characteristics Patient preoperative and operative characteristics are shown in Tables 1 and 2, respectively. In the univariate analyses, only BMI (p 0.045) and priority of surgery (p 0.021) were significantly associated with the need for reexploration for bleeding. Off-pump CABG had no association with the need for reexploration (p 0.346). Risk Factors The results of the multivariate logistic regression analysis (Table 5) showed that the independent predictors of reexploration for bleeding were BMI less than 25 kg/m 2 (p 0.003), nonelective surgery (p 0.022), five or more grafts (p 0.035), and increased age (p 0.041). The discriminatory ability of the logistic model, as measured by the C statistic, was 0.65, indicating a reasonable, although weak, ability to discriminate between patients who required reexploration for bleeding and those who did not. The predicted risks of individual patients were rank ordered and divided into deciles. Within each decile of estimated risk, the number of reexplorations predicted was accumulated against the number of observed reexplorations. The Hosmer-Lemeshow goodness-of-fit statistic across deciles of risk was not statistically significant (p 0.893), indicating minimal departure from a perfect fit. Aspirin and Heparin The propensity-matched analysis provided 84 patients who required reexploration for bleeding successfully matched to 84 who did not. The patient characteristics of the propensity-matched groups are in Table 4, which shows both groups were well matched with respect to major preoperative characteristics such as age, sex distribution, extent of disease, patients with poor left ventricular function, presence of comorbidities such as diabetes, peripheral vascular disease, respiratory and renal dysfunction, previous cardiac surgery, BMI, and number of grafts performed. Patient case notes were reviewed to identify the use of preoperative aspirin or heparin, and the results are shown in Figure 1. Patients who required reexploration for bleeding had a higher proportion of patients on preoperative aspirin (p 0.004) and aspirin/ heparin (p 0.001) in patients undergoing on-pump CABG. No association was found between reexploration for bleeding and preoperative aspirin and heparin use in off-pump CABG. Table 3 shows the hospital outcomes of the 168 propensity-matched patients. Patients who were reexplored were more likely to require inotrope support, mechanical ventilation longer than 24 hours, and have longer stays on the intensive care unit (ICU) and in the hospital postoperatively (p in all cases). Table 5. Multivariate Risk Factors for for Bleeding Risk Factor Odds Ratio 95% Confidence Intervals p Value BMI 25 kg/m to Nonelective surgery to Number of grafts to Age at operation a to a For every additional year. C Lemeshow-Hosmer goodness-of-fit BMI body mass index.

5 Ann Thorac Surg KARTHIK ET AL 2004;78: REEXPLORATION FOR BLEEDING 531 Fig 1. Effect of preoperative aspirin and heparin use in propensitymatched groups. (Top) On-pump coronary artery bypass graft surgery. for bleeding: solid bars aspirin (p 0.004); shaded bars aspirin/heparin (p 0.001). (Bottom) Off-pump coronary artery bypass graft surgery. for bleeding: solid bars aspirin (p 0.626); shaded bars aspirin/heparin (p 0.394). Analysis of all 89 patients who required reexploration for bleeding and the association between time to reexplore and postoperative outcomes is shown in Table 6. The incidence of major adverse outcomes after CABG was greater when there was a greater delay in taking the patients back for resternotomy, especially if this was greater than 12 hours after the initial return to ICU. All the patients who died after retsernotomy also belonged to this group. Comment Of 2,898 patients undergoing CABG in our study, 89 patients (3.1%) required resternotomy for bleeding. This figure compares quite well with incidence of 2% to 6% mentioned in the literature [1 3]. We did notice an increase in the incidence of reexploration for bleeding from 2.9%, 1999 to 2000, to 4.3%, 2001 to This is different from the experience of Munoz and coworkers [14], who noted a decline in the rates of reexploration of 3.6% from 1992 to 1994, to 2.0% from 1995 to An obvious site for bleeding was found in 73 of these 89 cases (82%). This number is substantially higher than that of Unsworth-White and coworkers [2], who found an obvious cause in 67% of their patients. The most common site was from the conduit or the anastomosis in 42 (57.5%), from the sternum or internal mammary artery bed in 23 (31.5%), from extrapericardial bleeders in 6 cases (8.2%), and from cannulation and cardiotomy sites in 2 cases (2.8%). In their study, Unsworth-White and coworkers found the conduit and anastomosis to be the site for bleeding in 26%, but found a significantly higher incidence of bleeding from cannulation and cardiotomy sites (21%) than we did. We believe that most cases of bleeding after CABG have a surgical cause, and hence, there is a stronger case for early reexploration of patients who continue to bleed after their return to the ICU. Univariate analysis of patient characteristics revealed patients with smaller BMI to have a significantly higher risk of reoperation for bleeding (p 0.045). The other group of patients at a significant risk of needing reoperation for bleeding were the nonelective surgery patients. Off-pump CABG was not associated with any significant risk for reoperation. On multivariate analysis, BMI less than 25 kg/m 2 (p 0.003), nonelective surgery (p 0.022), patients requiring 5 or more grafts (p 0.035), and elderly patients were at a greater risk of needing reexploration for bleeding. Most studies looking at risk factors for bleeding after cardiac surgery have found increased risk in older patients, smaller body surface area, longer CPB times, greater number of distal anastomosis, dialysis dependant renal failure, and the use of the internal mammary artery [1, 3, 5, 6]. Our finding of an association between BMI less than Table 6. Time to Reexplore and Postoperative Outcomes 3 (n 14) Hours to 3 6 (n 16) 6 12 (n 28) 12 (n 31) Mortality (%) a (4) Sternal wound infection (%) 14.3 (2) (3) 9.7 (3) Stroke (%) a (6) Renal failure (%) a (3) 22.6 (7) IABP support (%) (2) 7.1 (2) 16.1 (5) Inotrope support (%) 85.7 (12) 75.0 (12) 60.7 (17) 77.4 (24) Mechanical ventilation 24 hours (%) a 14.3 (2) 25.0 (4) 25.0 (7) 54.8 (17) ICU stay (days) a Postoperative stay (days) Blood loss in ICU (mls) a a p Categorical data shown as percentages. Continuous data shown as median values. IABP intra-aortic balloon pump; ICU intensive care unit.

6 532 KARTHIK ET AL Ann Thorac Surg REEXPLORATION FOR BLEEDING 2004;78: kg/m 2 and increased reexploration for bleeding is similar to the findings from Dacey and colleagues [1] from the Northern New England Cardiovascular Disease Study Group. Those investigators demonstrated that reexplorations increased as body surface area decreased. The reasons behind this relationship are unclear but may be due to dilution of coagulation factors [1]. It is likely that both the mediastinal and pericardial fat have a tamponading effect on the small bleeders in the mediastinum. We also think that, as a group, patients with smaller BMI are less likely to tolerate major blood losses. Hence, the threshold for resternotomy is usually lower in these patients. We were unable to find a significant relationship between CPB time and reexplorations for bleeding, although increasing number of grafts was a significant predictor irrespective of the use and duration of CPB. This association between number of grafts and reexploration can partly be explained by longer times on CPB, but may more realistically be due to an increased number of potential bleeding sites and to the overall complexity of the procedure. Dacey and coworkers [1] found that both CPB duration and number of grafts were associated with reexploration for bleeding. This difference between both studies may be due to sample size, with the Dacey study having more than 8,000 patients. Aspirin (in a dose of 75 mg to 300 mg daily) is a strong inhibitor of the platelet cycloxygenase, thus inhibiting the production of thromboxane A2. Thromboxane A2 is essential for the initiation of clot formation by platelets. This loss of function due to aspirin is permanent for the lifetime of the platelet. Because the average lifespan of platelets is about 7 days, the usual practice of stopping aspirin a week before surgery was considered appropriate. The association between bleeding and clotting abnormalities and various components of CPB and heparinization are very well known and documented. Cardiopulmonary bypass also tends to affect platelet function quite severely. These could be the reasons for the increased risk of bleeding seen in many studies of patients undergoing on-pump CABG [15, 16]. We undertook a propensity score match study to evaluate the effect of preoperative aspirin and heparin on the need for reexploration for bleeding. This analysis revealed that the preoperative use of aspirin and heparin were more common in patients reexplored for bleeding compared with a propensity-matched group who were not reexplored. After taking into account the use of CPB, this relationship still existed in on-pump CABG but disappeared for CABG patients who avoided CPB. This finding concurs with previous work from our institution, by Srinivasan and coworkers [17], which showed that continued aspirin does not increase postoperative blood loss, resternotomy rates, and blood or blood product requirements in patients who undergo off-pump CABG. This would suggest that the interaction of preoperative aspirin or heparin with CPB is a significant cause for increased risk of resternotomy for bleeding. The debate around aspirin use in conventional onpump CABG is still undecided. Dacey and colleagues [18], in a study by the Northern New England Cardiovascular Disease Study Group, showed that preoperative aspirin significantly reduced mortality without increasing bleeding-related morbidity. However, the Veterans Administration Cooperative Study found higher rates of reoperation and bleeding when using preoperative aspirin [15]. The propensity score matching gave us the ability to look at outcomes of patients needing resternotomy. Although there was a trend toward a greater incidence of mortality and major morbidity in the resternotomy group compared with the propensity-matched control group, we did not find any significant difference in the mortality rates and major morbidity, including sternal wound infection, stroke, renal failure, and need for intraaortic balloon support, in these two groups. These patients were more likely to need inotopes postoperatively and had longer average ICU and postoperative stays. Moulton and coworkers [5] found a significantly higher mortality rate and increased risk of renal failure, sepsis, and need for prolonged ventilation in the group needing reexploration. They concluded that reexploration is a significant multivariate indicator of increased morbidity and mortality in all patients, but more so in a selected low-risk subgroup of patients. They also felt that the amount of bleeding is a significant risk factor rather than the act of reexploration. Unsworth-White and coworkers [2] also found an association between reexploration and increased mortality and morbidity. A possible reason for the difference in outcomes of patients needing resternotomy for bleeding in our study compared with those of Moulton and Unsworth-White could be the time-delay factor. In our study, 58 of the 89 patients (65.2%) needing resternotomy were taken back within 12 hours of returning to the ICU. While no details regarding this time delay are available from the Moulton and Unsworth-White study, that is, as our study shows, one of the main causes for adverse outcomes after resternotomy for bleeding (Table 6). There was a significantly greater incidence of mortality and major complications including stroke, renal failure, and longer ICU stay among patients who faced a greater delay (longer than 12 hours) in return to the operating theater for reexploration. That indicates the continued instability of these patients for a longer time, greater blood loss, and greater need for blood products. Although not statistically significant, the incidence of sternal wound infection is slightly greater in the group reexplored early (ie, within 3 hours). Many of the patients from this group were hemodynamically unstable and required emergency reopening in the ICU, rather than in the operating theater in a better environment, and that could explain the higher sternal wound infection rate in this group of patients. Since conducting this study, we have strictly followed the protocol promulgated by Kirklin and Barratt-Boyes at our institution to promote early return to the operating theater for resternotomy for bleeding. Although this was

7 Ann Thorac Surg KARTHIK ET AL 2004;78: REEXPLORATION FOR BLEEDING 533 the guiding protocol even before this study, since performing this study, we have adhered to its principles even more strictly to minimize the time delay and the risk that this delay poses. There are some limitations, which may affect the conclusions drawn from our study. The primary limitation of the study is its retrospective nature. Propensity score adjustment is no substitute for a properly designed, randomized, controlled trial. The retrospective nature of the study cannot account for the unknown variables affecting the outcome that are not correlated strongly with measured variables. However, retrospective comparisons with propensity score adjustment are more versatile and offer a useful way of interpreting large amounts of audit data and seeking answers to questions that may present insuperable difficulties in the design of randomized, controlled trials. Another limitation of the study is the small number of patients who required reexploration, especially while analyzing the impact of preoperative aspirin or heparin and the effect of time delay. We run the potential risk of type II errors. That is highlighted when looking at sternal wound infections in the resternotomy and control groups, where we have a p value more than 0.05, but the infection rate is almost double in the patients who required resternotomy. Previous work from our institution would suggest that reexploration for bleeding is associated with sternal wound infection [4]. A further limitation is that we only assessed the use of preoperative aspirin and heparin. The continued use of clopidogrel until the day of surgery has been a relatively recent phenomenon at our center, and during the study time period there were just 3 patients in the resternotomy group receiving clopidogrel. Hence, we were unable to evaluate its impact on resternotomy. However, this will be of interest in the future. A final limitation is that we do not have complete data on the use of antifibrinolytic agents. None of our patients undergoing CABG received aprotinin; however, a small proportion (about 5%) have been given tranexamic acid, 2 g, after being weaned from CPB, which was a reflection of the individual practice of two anesthetists. In conclusion, our study reemphasizes the known risk factors of reexploration after CABG, including age, smaller BMI, greater number of grafts, and nonelective cases. It also highlights the role of preoperative aspirin and its association with a greater risk of bleeding after on-pump but not off-pump CABG. This finding is of importance as a greater number of patients are on aspirin until the day of surgery, and hence, there may be an associated increase in the incidence of resternotomy for bleeding. That may also represent one of the potential advantages of avoiding CPB in patients undergoing CABG, especially if they are at a higher risk for bleeding and continue to take aspirin until the day of surgery. Our study shows that in most cases a surgical explanation for bleeding is apparent on reexploration. Also, on comparing similar groups of patients, resternotomy itself does not seem to add to the mortality and morbidity after CABG. One must remember that often these patients have several characteristics that put them at a greater risk of adverse outcomes after surgery. Our major finding is the direct link between the incidence of major complications and mortality after reexploration and the time delay to reexploration. We found that the outcomes are significantly worse and the mortality rate is greater if there is a delay of more than 12 hours to resternotomy after CABG. On this basis, we would strongly recommend a policy of early reexploration for bleeding, as it does not seem to add to the risk and a cause is usually found. The major cause for poor outcome after resternotomy for bleeding is the associated time delay and not the resternotomy itself. We would like to acknowledge the cooperation given to us by all the Consultant Cardiac Surgeons at the Cardiothoracic Center- Liverpool: Mr John A. C. Chalmers, Mr Walid C. Dihmis, Mr Brian M. Fabri, Ms Elaine M. Griffiths, Mr Neeraj K. Mediratta, Mr D. Mark Pullan, and Mr Abbas Rashid. We would also like to thank Janet Deane, who maintains the quality and ensures completeness of data collected in our Cardiac Surgery Registry. References 1. Dacey LJ, Munoz JJ, Baribeau YR, et al. Re-exploration for haemorrhage following coronary artery bypass grafting. Incidence and risk factors. Arch Surg 1998;133: Unsworth-White MJ, Herriot A, Valencia O, et al. Resternotomy for bleeding after cardiac operation: a marker for increased morbidity and mortality. Ann Thorac Surg 1995; 59: Sellman M, Intonti MA, Ivert T. Reoperations for bleeding after coronary artery bypass procedures during 25 years. Eur J Cardiothorac Surg 1997;11: Lu J, Grayson AD, Jha P, Srinivasan AK, Fabri BM. Risk factors for sternal wound infection and mid-term survival following coronary artery bypass surgery. Eur J Cardiothorac Surg 2003;23: Moulton MJ, Creswell LL, Mackey ME, Cox JL, Rosenbloom M. Re-exploration for bleeding is a risk factor for adverse outcomes after cardiac operations. J Thorac Cardiovasc Surg 1996;111: Despotis GJ, Filos KS, Zoys TN, Hogue CW Jr, Spitznagel E, Lappas DG. Factors associated with excessive postoperative blood loss and haemostatic transfusion requirements: a multivariate analysis in cardiac surgical patients. Anaesth Analg 1996;82: Wynne-Jones K, Jackson M, Grotte G, Bridgewater B, for the North West Regional Cardiac Surgery Audit Steering Group. Limitations of the Parsonnet score for measuring risk stratified mortality in the north west of England. Heart 2000;84: Kirklin JW, Barratt-Boyes BG. Cardiac surgery. New York: John Wiley & Sons, 1986: Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR, for the Hospital Infection Control Practices Advisory Committee. Guideline for prevention of surgical site infection, Infect Control Hosp Epidemiol 1999;20: Hosmer D, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons, Blackstone E. Comparing apples and oranges. J Thorac Cardiovasc Surg 2002;123: Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143: Parsons LS. Reducing bias in a propensity score matched-

8 534 KARTHIK ET AL Ann Thorac Surg REEXPLORATION FOR BLEEDING 2004;78: pair sample using greedy matching techniques. Proceedings of the Twenty-Sixth Annual SAS Users Group International Conference. Cary, NC: SAS Institute, Munoz JJ, Birkmeyer NJO, Dacey LJ, et al. Trends in rates of re-exploration for haemorrhage after coronary artery bypass surgery. Ann Thorac Surg 1999;68: Goldman S, Copeland J, Moritz T, et al. Improvement in early saphenous vein graft patency after coronary artery bypass surgery with antiplatelet therapy: results of a Veterans Administration Cooperative Study. Circulation 1988;77: Sethi GK, Copeland JG, Goldman S, Moritz T, Zadina K, INVITED COMMENTARY Excessive bleeding immediately following cardiac surgery can be due to either surgical error or to impaired hemostasis. When excessive bleeding leads to reexploration of the operative area, morbidity and mortality have shown to be substantially increased. Immediate resternotomy is sometimes necessary and is often performed within hours after the operation. This decision to return to theater (or perform the reexploration in the intensive care unit sometimes) is almost always made by the attending surgeon. He or she can use guidelines most commonly indicating reexploration when bleeding exceeds 500 ml/hour the first 2 hours, or if bleeding continues at a rate of more than 200 ml/hour, or if clinical or echocardiographic signs of pericardial tamponade are present. Risk factors for bleeding leading to reexploration have been identified in many studies, and include: patient age ( 80 years old), small body surface area, longer cardiopulmonary bypass times, larger number of anastomoses, diabetes, and usage of an internal mammary artery. The same risk factors are identified in the current study reported by Karthik and colleagues. However, a remarkable new finding in their study is that the nonelective patient had a significantly higher risk for reexploration due to excessive bleeding. Further insight into the pathophysiologic mechanism of this observation is failing. In addition, in a subanalysis of their study, using an upcoming new statistical technique for nonrandomized comparisons called propensity scoring, they found that the use of aspirin in combination with heparin Henderson WG, and the Department of Veterans Affairs Cooperative Study on Antiplatelet Therapy. Implications of preoperative administration of aspirin in patients undergoing coronary artery bypass grafting. Department of Veterans Affairs Cooperative Study on Antiplatelet Therapy. J Am Coll Cardiol 1990;15: Srinivasan AK, Grayson AD, Pullan DM, Fabri BM, Dihmis WC. Effect of preoperative aspirin use in off-pump coronary artery bypass surgery. Ann Thorac Surg 2003;6: Dacey LJ, Munoz JJ, Johnson ER, et al. Effect of preoperative aspirin use on mortality in coronary artery bypass grafting patients. Ann Thorac Surg 2000;70: is also a risk factor for reexploration in on-pump coronary artery bypass grafting (CABG) patients. One of the possible elements playing a role in the increased incidence of reexploration for bleeding after CABG in nonelective patients could be of an iatrogenic nature. For it is well known that there are more drugs that affect postoperative hemostasis of such patients, such as abciximab, warfarin, ticlopidin, and clopidogrel. These drugs are playing an increasing role in the treatment of acute coronary syndromes by interventional cardiologists. Mainly used in combinations, these drugs are indicated as standard therapy for prevention of coronary stent thrombosis. As a consequence of the widespread use of the aforementioned drugs, more patients are undergoing urgent or emergent CABG procedures while, and maybe because of, having these drugs on board. The results of this present study and those of upcoming studies focused on the effects of such drugs on hemostasis after cardiac surgery, whether performed on-pump or off-pump, need to help cardiac surgeons in reducing morbidity and mortality of their patients. Piet Boonstra, MD, PhD Department of Cardiothoracic Surgery University Hospital Groningen Hanzeplein EZ Groningen, the Netherlands p.w.boonstra@thorax.azg.nl 2004 by The Society of Thoracic Surgeons /04/$30.00 Published by Elsevier Inc doi: /j.athoracsur