Deep Sternal Wound Infection: Risk Factors and Outcomes

Transcription

1 Deep Sternal Wound Infection: Risk Factors and Outcomes Michael A. Borger, MD, Vivek Rao, MD, Richard D. Weisel, MD, Joan Ivanov, MSc, Gideon Cohen, MD, Hugh E. Scully, MD, and Tirone E. David, MD Division of Cardiovascular Surgery, The Toronto Hospital, and Centre for Cardiovascular Research and the Collaborative Program in Cardiovascular Sciences, University of Toronto, Toronto, Ontario, Canada Background. Deep sternal wound infection (DSWI) is a serious complication of cardiac operations performed by median sternotomy. We attempted to define the predictors of DSWI and to describe the outcomes of two treatment strategies used at our institution. Methods. Retrospective review was performed using prospectively gathered data on 12,267 consecutive cardiac surgical patients from 1990 to Chart review was performed on all patients in whom DSWI developed, and follow-up was obtained on 100% of these patients. Results. Deep sternal wound infections developed in 92 patients (incidence 0.75%). Multivariable predictors for development of DSWI in all patients were (odds ratios and 95% confidence intervals in parentheses) (1) diabetes mellitus (2.6; 1.7 to 4.0) and (2) male sex (2.2; 1.3 to 3.9). In patients receiving coronary artery bypass grafting alone, independent predictors were (1) bilateral internal thoracic artery grafts (3.2; 1.1 to 8.9), (2) diabetes (2.7; 1.6 to 4.3), and (3) male sex (1.8; 0.9 to 3.7). For all other patients, predictors were (1) age more than 74 years (3.3; 1.1 to 10.1), (2) male sex (3.0; 1.1 to 8.1), and (3) diabetes (2.3; 0.9 to 5.8). Bilateral internal thoracic artery grafts increased the risk of DSWI in all subgroups of coronary artery bypass graft patients, particularly in diabetics who had a 14.3% incidence of DSWI after bilateral internal thoracic artery grafting. Patients with DSWIs received either sternal debridement with primary closure (n 45) or sternectomy with flap reconstruction (n 46). The 6-month freedom from adverse event rate (ie, readmission, reoperation, or death) was 76% for both groups of patients. Conclusions. Male sex and diabetes are predictors of DSWI in all cardiac surgical patients. Bilateral internal thoracic artery grafting may be contraindicated in diabetic patients. (Ann Thorac Surg 1998;65:1050 6) 1998 by The Society of Thoracic Surgeons Deep sternal wound infection (DSWI) is a serious complication of cardiac surgical procedures performed by sternotomy. It results in increased mortality, morbidity, and hospital costs [1]. Several studies have examined risk factors for development of DSWI [2 4]. Commonly quoted risk factors include obesity, chronic obstructive pulmonary disease, elderly age, peripheral vascular disease, reoperation, use of internal thoracic artery (ITA) conduits, operation time, low cardiac output, ventilation time, and reexploration for bleeding. Several studies have also examined treatment options for DSWI [1, 5 8]. Therapeutic options include debridement with early or delayed closure, closed continuous irrigation, and partial or complete sternectomy with flap reconstruction. The purpose of this study was to (1) determine the risk factors for DSWI in a cohort of cardiac surgical patients; (2) evaluate the potential for risk factor modification; and (3) analyze the outcomes of the two treatment protocols for DSWI used in this population, sternal debridement with primary closure and sternectomy with flap reconstruction. Accepted for publication Nov 7, Address reprint requests to Dr Weisel, Division of Cardiovascular Surgery, The Toronto Hospital, EN , 200 Elizabeth St, Toronto, ON, Canada M5G 2C4. Patients and Methods Population From January 1990 to December 1995, 12,267 consecutive cardiac surgical procedures through a midline sternotomy were performed at the Toronto Hospital. Our computerized registry was surveyed for those patients with a diagnosis of sternal wound infection, then a manual chart review was performed on these patients. Deep sternal wound infection was defined according to the guidelines of the Centers for Disease Control and Prevention [9], with patients meeting at least one of the following criteria: (1) an organism is isolated from culture of mediastinal tissue or fluid; (2) evidence of mediastinitis is seen during operation; or (3) one of the following, chest pain, sternal instability, or fever ( 38 C), is present and there is either purulent discharge from the mediastinum or an organism isolated from blood culture or culture of drainage of the mediastinal area. Patients who underwent operative therapy for the wound infection (n 92) were the subject of follow-up and evaluation. Perioperative Management Patients were showered and shaved the day of their operation. All patients received intravenous cefazolin, or vancomycin if allergic to penicillins, preoperatively and 1998 by The Society of Thoracic Surgeons /98/$19.00 Published by Elsevier Science Inc PII S (98)

2 Ann Thorac Surg BORGER ET AL 1998;65: DEEP STERNAL WOUND INFECTION 1051 for 48 hours postoperatively. The operative field was painted with povidone-iodine solution and the skin was covered with an iodoform-impregnated adhesive plastic sheet (Ioban; 3M, Brookings,. The skin was incised with a scalpel and electrocautery was used to open the presternal layers and pericardium. Bone wax was used only if sternal bleeding was profuse. Internal thoracic arteries were harvested as pedicled in situ grafts when used for coronary bypass. Sump drains were placed in the mediastinum, and chest tubes were inserted into the pleural spaces if opened. The sternum was closed with stainless steel wires. The presternal space was obliterated with two layers of absorbable suture, and the skin was closed with a subcuticular absorbable suture. Patients were extubated when they were hemodynamically stable, normothermic, and ventilating spontaneously. All drains were removed the morning after operation or when drainage was less than 25 ml/h. Management of Infection Seventy-one of the 92 patients (77%) were diagnosed with DSWI on the same admission as their cardiac surgical procedure. All discharged patients with wound infections are readmitted to our institution under the attending cardiovascular surgeon. Patients were started on broad-spectrum antibiotics when the diagnosis of sternal wound infection was made. Ninety-one of the 92 patients with DSWI underwent surgical therapy (the remaining patient had a cardiac arrest and died on the way to the operating room). Patients underwent either sternal debridement with primary closure (n 45) or sternectomy with muscle flap reconstruction (n 46) at the discretion of the attending surgeon. In general, however, patients underwent prompt, simple sternal debridement, without closed irrigation, if the infection was thought to be localized to a small section of the sternum and there was little or no purulent drainage. Patients received flap reconstruction if there was evidence of mediastinitis, if a large amount of purulent drainage was present, or if the majority of the wound was thought to be involved. All flap reconstructions were performed by the Toronto Hospital plastic surgery consultants. Study Variables Multiple variables were recorded prospectively and analyzed retrospectively as predictors of DSWI. Patients were divided into two groups: those with DSWI (n 92) and those without DSWI (control patients, n 12,175). Comparisons of baseline demographics and intraoperative and postoperative variables were made between the two groups of patients to identify predictors of DSWI. These variables and their definitions are listed in Appendix 1. Follow-up Postoperative follow-up was obtained on 100% of patients. Follow-up was performed to identify events that occurred within 6 months of the surgical treatment for the sternal wound infection. Follow-up was completed first by review of the surgeon s office chart and then by telephone interview with the patient or his or her family physician. Statistical Methods Univariate analysis was performed using the 2 test (or Fisher s exact test, where appropriate) for categorical variables and two-tailed Student s t test for continuous variables. All variables suggested by the univariate analysis (p 0.25) or those judged to be clinically important were entered into a stepwise multiple logistic regression analysis model. That model with the best Hosmer- Lemeshow goodness of fit statistic and receiver-operator characteristic curve were chosen, as previously described [10]. All analyses were performed using the SAS program (SAS Institute, Cary, NC). Results During the study period of 1990 to 1995, 309 of the 12,267 patients (2.5%) undergoing cardiac surgery at the Toronto Hospital developed a sternal wound infection, with 92 patients (0.75%) undergoing operative intervention for DSWI. The remaining 217 patients received antibiotics with resolution of the infection, and these patients will not be considered any further; this study pertains only to those patients undergoing operative therapy for DSWI. Isolated aortocoronary artery bypass grafting (CABG) was performed in 8,050 (65.6%) patients, valvular procedures with or without CABG were performed in 2,542 (20.7%) patients, and other procedures including left ventricular aneurysmectomies, ascending aorta replacements, and adult congenital procedures were performed in 1,675 (13.7%) patients. The incidence of DSWI was 0.83% in patients undergoing CABG, 0.63% in patients undergoing valvular procedures, and 0.54% for other procedures (p 0.325). Patients who acquired DSWI had longer intensive care unit stays ( versus days, p 0.001) and hospital stays (32 21 versus 10 9 days, p 0.001) than patients who did not have DSWI. Perioperative mortality was also higher in patients with DSWI (9.8% versus 3.7%, p 0.002). A univariate analysis of potential risk factors for development of DSWI was performed using the variables listed in Table 1, and a definition of these variables is supplied in Appendix 1. Significant preoperative predictors of DSWI were male sex, body mass index, diabetes, smoking status, and chronic obstructive pulmonary disease. Intraoperative predictors were cardiopulmonary bypass time, operation time, use of left ITA graft, and use of bilateral ITA grafts. Postoperative predictors were low cardiac output syndrome, ventilation time, and postoperative resternotomy. (Postoperative resternotomy refers to patients returning to the operating room after their cardiac surgical procedure for bleeding, dehiscence, or delayed primary closure of the sternum. It does not refer to patients with a previous history of cardiac operation performed through a median sternotomy.) All significant variables (p 0.25) from the univariate analysis were entered into a stepwise multiple logistic

3 1052 BORGER ET AL Ann Thorac Surg DEEP STERNAL WOUND INFECTION 1998;65: Table 1. Variables Assessed as Univariate Predictors of Deep Sternal Wound Infection Variable Incidence in Patients Without DSWI (%) (n 12,175) Incidence in Patients With DSWI (%) (n 92) p Value a Age (y) Male sex Diabetes Hypertension Peripheral vascular disease Smoking status Never Former Active COPD Renal failure CHF LV ejection fraction Reoperation Preoperative endocarditis Procedure Isolated CABG Valve CABG Other Timing Elective Urgent Emergent Preoperative length of stay (mean Body mass index (mean CPB time (mean Cross-clamp time (mean Operation time (mean LITA LITA RITA Perioperative MI Postoperative low output syndrome IABP Ventilation time (mean Postoperative resternotomy a Continuous data analyzed by Student s t test; categoric data analyzed by 2 test or Fisher s exact test. CABG coronary artery bypass grafting; CHF congestive heart failure; COPD chronic obstructive pulmonary disease; CPB cardiopulmonary bypass; DSWI deep sternal wound infection; IABP intraaortic balloon pump; LITA left internal thoracic artery; LV left ventricle; MI myocardial infarction;rita right internal thoracic artery; SD standard deviation. regression analysis, the results of which are displayed in the upper panel of Figure 1. Because operative procedure was not a univariate predictor of DSWI (see Table 1), we originally performed logistic regression on all patients regardless of procedure. Independent predictors of DSWI in all patients were diabetes mellitus (odds ratio [OR], 2.6; 95% confidence interval [CI], 1.7 to 4.0) and male sex (OR, 2.2; CI, 1.3 to 3.9). The incidence of DSWI was 1.5% in diabetic patients and 0.9% in male patients (versus 0.75% in all patients). The Hosmer-Lemeshow goodness of fit statistic for this model had a probability value of (a p value 0.05 claims there are no significant differences between the predicted outcomes and the observed data), and the area under the receiver operator characteristic curve was Bilateral ITA grafting was a significant univariate predictor of DSWI in all patients (see Table 1). At our institution, however, bilateral ITA grafts are performed exclusively in patients undergoing isolated CABG (ie, this procedure was never used in patients undergoing valve and CABG or other combined procedures). We therefore performed a second multivariable logistic regression analysis after dividing our patient population into two groups those undergoing isolated CABG, and all other patients. The results of this analysis are displayed in the middle and bottom panels of Figure 1. Independent risk factors for DSWI in CABG patients Fig 1. Multivariable predictors of deep sternal wound infection in all cardiac surgical patients (top panel), in isolated coronary artery bypass grafting (CABG) patients (middle panel), and in non-cabg patients (bottom panel). (BITA bilateral internal thoracic artery grafts; C.I. confidence interval; DM diabetes mellitus.)

4 Ann Thorac Surg BORGER ET AL 1998;65: DEEP STERNAL WOUND INFECTION 1053 Fig 2. Risk of deep sternal wound infection (DSWI) in coronary artery bypass grafting (CABG) patients receiving bilateral internal thoracic artery grafts (BITA) with and without other risk factors. All patient subgroups exhibited a significantly increased risk of DSWI with BITA grafts. were bilateral ITA grafts (OR 3.2; CI 1.1 to 8.9), diabetes (OR, 2.7; CI, 1.6 to 4.3), and male sex (OR, 1.8; CI, 0.9 to 3.7). The Hosmer-Lemeshow goodness of fit probability value for this model was and the area under the receiver-operator characteristic curve was Independent risk factors for DSWI in non-cabg patients were age more than 74 years (OR, 3.3; CI, 1.1 to 10.1), male sex (OR, 3.0; CI, 1.1 to 8.1), and diabetes (OR, 2.3; CI, 0.9 to 5.8). The Hosmer-Lemeshow probability value was and the area under the receiver-operator characteristic curve was 0.58 for this model. Of the three multivariable predictors of DSWI in CABG patients, only the use of bilateral ITA grafts is under the surgeon s discretion. The other risk factors, male sex and diabetes, are known at the time a decision is made whether or not to perform bilateral ITA grafts. We therefore examined the risk of development of DSWI in male and diabetic CABG patients, with and without bilateral ITA grafts (see Fig 2). The incidence of DSWI was 0.7% in CABG patients not receiving bilateral ITA grafts versus 2.2% in those receiving bilateral ITA grafts (p 0.029). In male CABG patients, bilateral ITA grafts raised the risk of DSWI from 0.8% to 2.6% (p 0.020). Similarly, diabetic patients had an increased risk of DSWI with bilateral ITA grafts from 1.3% to 14.3% (p 0.001). Male diabetic patients undergoing CABG went from a baseline incidence of DSWI of 1.6% to 20.0% with the addition of bilateral ITA grafts (p 0.001). Table 2 shows the results of cultures obtained at the time of operative therapy for the 92 patients with DSWI. Staphylococcus aureus was cultured from the wounds of 38 patients (42%) and coagulase-negative staphylococci from 22 patients (24%). Multiple organisms were cultured from 20 patients (22%). Five patients (6%) had no growth from all cultures (ie, they had clinical signs and operative findings consistent with DSWI, but negative cultures). All patients with negative-culture DSWI received broadspectrum intravenous antibiotics before their operative procedure, which may have contributed to their culturenegative status. It should be mentioned, however, that intraoperative wound cultures at our institution are not routinely cultured for mycoplasma or fungi, and therefore some of the culture-negative infections may have been caused by these less common organisms. Twentyfive patients had infections at distant sites before being diagnosed with DSWI. Twelve of 20 patients (60%) with multiple organisms cultured from their DSWI had previous infections: 6 with leg wound infections, 5 with pneumonia, and 1 with prosthetic valve endocarditis. Nine of 60 patients (15%) with staphylococcal DSWI had preexisting infections: 7 with leg wound infections and 2 with pneumonia. Clinical signs of DSWI included purulent discharge in 80.4% of patients, fever in 39.1%, and sternal click or gross instability in 30.4%. Patients were divided into two groups according to the operative procedure for their wound infection sternal debridement with primary closure in 45 patients and sternectomy with muscle flap reconstruction in 46 patients and follow-up data were collected to examine outcomes for each of these groups of patients. One patient had a cardiac arrest and died on the way to the operating room and is excluded from any of the follow-up data. Of the 46 patients who underwent sternectomy with muscle flap reconstruction, 41 received bilateral pectoralis flaps and 5 received combined pectoralis and rectus abdominis flaps. As stated previously, the decision as to which operative procedure the patients would receive was at the discretion of the attending surgeon, with patients tending to receive debridement alone if the infection was thought to be localized, and flap reconstruction if there was gross purulent drainage or evidence of mediastinitis. The median length of time from date of initial cardiac operation to diagnosis of DSWI was 12 days in those patients receiving flap reconstruction (range, 3 to 71 days) and 11 days in those receiving debridement (range, 3 to 27 days). The median length of time from diagnosis of DSWI to date of operative therapy for DSWI was 3 days in the flap group (range, 0 to 33 days) and 1 day in the debridement group (range, 0 to 36 days). Follow-up was obtained on 100% of patients, with the median length of time to follow-up being 11.4 months after the operative procedure for infection (range, 7.3 to Table 2. Bacteriology Organism No. of Patients (%) Staphylococcus aureus 38 (42) Coagulase-negative staphylococci 22 (24) Enterobacter species 2 (2) Enterococcus 1 (1) Escherichia coli 1 (1) Pseudomonas aeruginosa 1 (1) Propionibacterium acnes 1 (1) Multiple organisms 20 (22) Negative culture 5 (6)

5 1054 BORGER ET AL Ann Thorac Surg DEEP STERNAL WOUND INFECTION 1998;65: months). To capture only those events related directly to the sternal wound infection, data are displayed only for the first 6 months of follow-up. Outcome data is supplied in an intention-to-treat format for the two groups of patients (ie, if patients underwent a different operative procedure for a second operation for infection, they were still classified by the original operative procedure for infection). Figure 3 displays the 6-month eventfree survival for patients undergoing sternal debridement, which was 75.6% 2.9%. Three of the 45 patients (6.7%) in the debridement group had persistent sepsis postoperatively leading to multisystem organ failure and death, despite repeat surgical interventions in 2 of the 3 patients. Seven patients (15.6%) required a reoperation for control of infection. These 7 patients all received a pectoralis flap for the second operative procedure, and 2 suffered ongoing sepsis and eventually died. Two debridement patients (4.4%) required readmission to the hospital as a direct result of the infection; however, both of these patients received conservative treatment with good recovery. Figure 4 displays the outcomes for patients receiving sternectomy with flap reconstruction for their DSWI. The 6-month event-free survival was 76.1% 2.9% for this group of patients. Five flap patients (10.9%) had persistent sepsis postoperatively with resultant multisystem organ failure and death, despite repeat surgical interventions in 3 of these patients. Any reoperation was required in 6 of the 46 patients (13.0%) undergoing flap reconstruction as the initial operative procedure. Four of the 6 patients requiring reoperation underwent repeat flap procedures (of which 2 died of ongoing sepsis), 1 underwent simple drainage of a hematoma with no adverse sequelae, and 1 underwent multiple debridements of the flap before dying of uncontrolled sepsis. Readmission for complications related to sternal infection was required in 3 patients (6.5%), with all 3 patients being treated conservatively with good recovery. Three of our 92 patients underwent early reoperation for a sternal dehiscence and were found to have evidence of infection at reoperation. During the period of this Fig 3. Six-month event-free survival in patients treated with sternal debridement and primary closure. Events included death, reoperation, or readmission to hospital. Fig 4. Six-month event-free survival in patients treated with sternectomy and flap reconstruction. Events included death, reoperation, or readmission to hospital. study an additional 38 patients underwent sternal rewiring for dehiscence at our institution, and 4 acquired DSWI later in their hospital admission. Therefore 7 of 41 patients who required reoperation for sternal dehiscence (17.1%) had DSWI. Comment Deep sternal wound infection is a well-described complication of operations performed through a median sternotomy [1]. This study identifies risk factors for DSWI in a large cohort of consecutive cardiac surgical patients. All procedures were considered together because we did not find a difference in DSWI incidence for the different surgical procedures (see Table 1). The overall incidence of DSWI in our study was 0.75%, which is in the lower end of the range of values commonly quoted in the literature [1 4, 14, 15]. Patients with DSWI had longer intensive care unit and hospital stays and higher mortality, emphasizing the seriousness of this complication. Our multiple logistic regression analysis revealed two independent predictors of DSWI in the entire cohort of patients: diabetes and male sex (Fig 1, top panel). Diabetes is known to impair wound healing and cellular and humoral immunity, which may increase the risk of infection [16]. Male sex is cited as a risk factor for DSWI in at least two other studies [2, 15], including a prospective, multicenter study performed by the Parisian Mediastinitis Study Group [2]. The reason for male sex as an independent predictor is unclear, particularly in view of the fact that female sex is traditionally associated with increased morbidity and mortality in cardiac surgical patients. It may be that males tend to have increased tension on their sternal incision, which in turn leads to increased risk of sternal instability. Sternal rewiring for dehiscence was performed in 41 patients during the time period of this study, of which a large proportion (17.1%) went on to have DSWI. We divided patients into two groups isolated CABG

6 Ann Thorac Surg BORGER ET AL 1998;65: DEEP STERNAL WOUND INFECTION 1055 and all other patients to assess the risk of bilateral ITA grafting on the development of DSWI. Only isolated CABG patients received bilateral ITA grafts at our institution. Bilateral ITA grafting was a strong univariate predictor of DSWI in our overall patient population. The middle and bottom panels of Figure 1 display the factoradjusted odds ratios for the predictors of DSWI in these two patient subpopulations. The use of bilateral ITA grafts was the strongest predictor of DSWI in CABG patients, followed by diabetes and male sex. In all other cardiac surgical patients independent predictors of DSWI were age more than 74 years, male sex, and diabetes, in decreasing order. Elderly age has been associated with sternal wound infection in other studies [13, 20]. Left ITA grafting to the left anterior descending coronary artery increases bypass graft patency rates and decreases late cardiac events. Extraarterial grafts may result in further improvements in patency rates and cardiac events. The use of bilateral ITA grafts, however, may lead to increased risk of DSWI by decreasing sternal blood flow [11, 12]. The risk of DSWI in patients receiving bilateral ITA grafts is controversial, with some studies suggesting no increased risk in selected patients [1, 13] and others finding a consistent increase in risk in all patient groups [2, 14]. Figure 2 demonstrates that in patients undergoing CABG at our institution, bilateral ITA grafts significantly increased the risk of DSWI in all subgroups of patients. In particular, diabetic patients who received bilateral ITA grafts had a much higher incidence of DSWI compared with those who did not (14.3% versus 1.3%, p 0.001). There were 14 diabetic patients who received bilateral ITA grafts in this study population, of which 3 were insulin-dependent and 11 were diet-controlled or on oral hypoglycemic agents. Two diabetic patients had DSWI: 1 insulin-dependent patient and 1 with diet-controlled diabetes. We would therefore conclude that any type of diabetes is a significant risk factor for development of DSWI for patients receiving bilateral ITA grafts. Risk factor identification permits the assessment of factors that may be modifiable. Of the risk factors for DSWI identified in this study, only the use of bilateral ITA grafts is at the surgeon s discretion. Bilateral ITA grafts demonstrate excellent patency rates and may decrease long-term mortality [16, 17]. However, our findings would suggest that diabetes is a contraindication to the use of bilateral ITA grafting. If extraarterial grafts are being contemplated for a diabetic patient, we would suggest the use of other arterial conduits, such as the radial, inferior epigastric, or gastroepiploic artery. The results of wound cultures are presented in Table 2. As in several previous studies, the most common organisms were S aureus and coagulase-negative staphylococci [1, 14, 19], followed by multiple-organism infections. Patients with multiple organisms cultured from the sternal wound had a high incidence of preexisting infections at distant sites, in contrast to those with staphylococcal infections (60% versus 15%, p 0.002). Staphylococcal infections may be caused by wound contamination from local skin flora, whereas multiple-organism infections may be caused by hematogenous spread or by contamination from distant sites. Patients with DSWI were divided into two groups according to the operative therapy chosen for their wound infection: sternal debridement with primary closure or sternectomy with flap reconstruction. The choice of therapy depended on the clinical status of the patient, and therefore outcomes for the two groups cannot be directly compared. However, we can state that patients receiving debridement underwent operative therapy earlier than those undergoing flap reconstruction (median length of time from diagnosis of DSWI to operative therapy 1 versus 3 days, respectively.) This was probably related to the fact that flap reconstructions were performed in conjunction with plastic surgeons at our institution, and therefore required coordination of operative schedules. Follow-up was obtained on 100% of patients and results are displayed for the 6-month period after the operative intervention for DSWI for both groups of patients in Figures 3 and 4. Both treatment strategies resulted in a 6-month event-free survival of approximately 75%. Mortality was high in both groups, occurring in 6.7% of debridement patients and 10.9% of flap patients. We did not identify any complications attributable to the infection that occurred later than 6 months after operative intervention for DSWI. In conclusion, DSWI is a devastating complication of operations performed through a median sternotomy. Although the incidence of this complication is low (0.75% in our study population), high-risk patients can be identified. Male sex and diabetes are predictors of DSWI in all cardiac surgical patients. Bilateral ITA grafting increased the risk of DSWI in all subgroups of CABG patients. Diabetic patients receiving bilateral ITA grafts have a 14% incidence of DSWI, and we therefore suggest against the use of bilateral ITA grafts in this population. Longterm follow-up reveals that a good result can be achieved in 75% of patients treated with debridement or sternectomy with flap reconstruction. Supported in part by the Heart and Stroke Foundation of Ontario. Doctors Borger, Rao, and Cohen are Research Fellows of the HSFO. Doctor Weisel is a Career Investigator of the HSFO. References 1. Loop FD, Lytle BW, Cosgrove DM, et al. Sternal wound complications after isolated coronary artery bypass grafting: early and late mortality, morbidity, and cost of care. Ann Thorac Surg 1990;49: The Parisian Mediastinitis Study Group. Risk factors for deep sternal wound infection after sternotomy: a prospective, multicenter study. J Thorac Cardiovasc Surg 1996;111: Blanchard A, Hurni M, Ruchat P, et al. Incidence of deep and superficial sternal infection after open heart surgery: a ten years retrospective study from 1981 to Eur J Cardiothorac Surg 1995;9: Ottino G, De Paulis R, Pansini S, et al. Major sternal wound infection after open-heart surgery: a multivariate analysis of risk factors in 2,579 consecutive operative procedures. Ann Thorac Surg 1987;44:173 9.

7 1056 BORGER ET AL Ann Thorac Surg DEEP STERNAL WOUND INFECTION 1998;65: Scully HE, Leclerc Y, Martin RD, et al. Comparison between antibiotic irrigation and mobilization of pectoral muscle flaps in treatment of deep sternal infections. J Thorac Cardiovasc Surg 1985;90: Johnson JA, Gall WE, Gundersen E, Cogbill TH. Delayed primary closure after sternal wound infection. Ann Thorac Surg 1989;47: Calvat S, Trouillet J, Nataf P, et al. Closed drainage using Redon catheters for local treatment of poststernotomy mediastinitis. Ann Thorac Surg 1996;61: Ringelman PR, Vander Kolk CA, Cameron D, et al. Longterm results of flap reconstruction in median sternotomy wound infections. Plast Reconstr Surg 1994;93: Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections Am J Infect Control 1988;16: Rao V, Ivanov J, Weisel RD, et al. Predictors of low cardiac output syndrome after coronary artery bypass. J Thorac Cardiovasc Surg 1996;112: Lust RM, Sun YS, Chitwood R. Internal mammary artery use: sternal revascularization and experimental infection patterns. Circulation 1991;84(Suppl 3): Parish MA, Asai T, Grossi EA, et al. The effects of different techniques of internal mammary artery harvesting on sternal blood flow. J Thorac Cardiovasc Surg 1994;104: He GW, Ryan WH, Acuff TE, et al. Risk factors for operative mortality and sternal wound infection in bilateral internal mammary artery grafting. J Thorac Cardiovasc Surg 1994; 107: Grossi EA, Esposito R, Harris LJ, et al. Sternal wound infections and use of internal mammary artery grafts. J Thorac Cardiovasc Surg 1991;102: Demmy TL, Park SB, Liebler GA, et al. Recent experience with major sternal wound complications. Ann Thorac Surg 1990;49: Rayfield EJ, Ault MJ, Keusch GT, et al. Infection and diabetes: the case for glucose control [Review]. Am J Med 1982;72: Fiore AC, Naunheim KS, Dean P, et al. Results of internal thoracic artery grafting over 15 years: single versus double grafts. Ann Thorac Surg 1990;49: Cameron AC, Green GE, Brongno DA, Thornton J. Internal thoracic artery grafts: 20-year clinical follow-up. J Am Coll Cardiol 1995;25: Mossad SB, Serkey JM, Longworth DL, et al. Coagulasenegative staphylococcal sternal wound infections after open heart operations. Ann Thorac Surg 1997;63: Horneffer PJ, Gardner TT, Manolio TA, et al. The effects of age on outcomes after coronary bypass surgery. Circulation 1987;76(Suppl 5):6 12. Appendix 1. Variables assessed in predictive models include the following: Age Sex Body mass index: Patient s weight (in kilograms) divided by square of height (in meters) Diabetes mellitus Hypertension: Patient on antihypertensive medication preoperatively Peripheral vascular disease: History of peripheral or carotid vascular disease Smoking status: Past or current smoker Chronic obstructive pulmonary disease: Chronic use of inhalational or oral bronchodilators or steroids Renal failure: Serum creatinine 200 mmol/l or history of renal failure Congestive heart failure: History of hospital admission for congestive heart failure Left ventricular ejection fraction: Left ventricular ejection fraction as assessed by angiography or two-dimensional echocardiography Reoperation Preoperative endocarditis Operative procedure: CABG, valvular procedure with or without CABG, or other Timing: Elective: Elective admission Urgent: Operation during same hospitalization Emergent: Cardiac event within 12 hours of operation Preoperative length of stay: Length of stay in hospital (in days) before operation Cardiopulmonary bypass time Cross-clamp time Left ITA: Use of left ITA graft Left ITA right ITA: Use of left ITA and right ITA grafts (in isolated CABG patients only) Perioperative myocardial infarction: Development of new Q waves with rise in creatine kinase-mb Postoperative low output syndrome: Use of inotropic agents (excluding dopamine 3 g kg 1 min 1 ) for more than 90 minutes postoperatively for a cardiac index less than 2.0 L min 1 m 2, or use of intraaortic balloon pump Intraaortic balloon pump: Use of intraaortic balloon pump Ventilation time Postoperative resternotomy: Repeat sternotomy after initial surgical procedure for any reason except infection