Factors Predisposing to Median Sternotomy Complications*

Transcription

1 Factors Predisposing to Median Sternotomy Complications* Deep vs Superficial Infection Anoar Zacharias, MD; and Robert H. Habib, PhD Study objectives: Median sternotomy infections are a serious complication of cardiac surgery. The purpose of this study was to determine the patient characteristics and operative variables that predict incidence of sternal infection, and possibly its severity. Design: Univariate and multivariate retrospective analysis comparing patient, operative, and postoperative data in patients with and without sternal infections. Setting: Cardiac surgery program of a 580-bed private hospital in Toledo, Ohio. Patients: We studied 2,317 consecutive (June 1991 to December 1994) patients undergoing cardiac surgery. Results: Forty-one sternal infections were documented. Of these, 21 (0.91%) were deep infections with mediastinal involvement and 20 (0.86%) were superficial. Two patients with deep infections died (2141, 5%). Ten variables were associated with infection by univariate analysis (p<0.05), and of these, five were independent predictors by multivariate logistic regression. These predictors were obesity (p<0.001), insulin-dependent diabetes (p<0.001), use of internal mammary artery grafts (p=0.02), surgical reexploration of the mediastinum (p=0.003), and postoperative transfusions (p=0.01). Predictors of deep and superficial sternal infection did not differ. Length of hospitalization was substantially longer for patients with deep (32:±::21 days) vs superficial infection (13:±::10 days). Conclusions: The present study confirms previous findings that obesity, insulin-dependent diabetes, and internal mammary artery grafting (especially bilaterally) increase the risk of sternal infection. In addition, chest surgical reexploration and blood transfusions were postoperative factors that predisposed patients with median sternotomy to infection. Unlike their associated morbidity and mortality, predictors of deep and superficial sternal infections are similar. (CHEST 1996; 110: ) Key words: mediastinitis; multivariate analysis; risk factors; univariate analysis Abbreviations: CABG=coronary artery bypass grafting; IABP=intra-aortic balloon pump; IMA=intemal mammary artery; LOS=length of stay Median sternotomy infections are an infrequent, yet potentially devastating, complication following cardiac surgery. Although antibiotic prophylaxis and evolving surgical practices have led to an appreciable decline in their incidence, sternal wound infections remain an important problem because of the associated high mortality and morbidity.1 2 Several studies have investigated patient and operative risk factors that are associated with an increased incidence of sternal infections.3-15 Yet, there remains a lack of consensus on what combination of variables or risk factors best predict the incidence of this serious *From the Department of Cardiothoracic Surgery, St. Vincent Medical Center, Toledo, Ohio. Manuscript received May 8, 1996; accepted May 31. Rlprint requests: Dr. Zacharias, St. Vincent Medical Center, 2213 CTie!""J St, ACC 309, Toledo, OH Surgicor primenet.com complication. Milano et al 16 recently explored the patient characteristics and operative variables that independently predisposed to mediastinitis in a large series of patients undergoing myocardial revascularization, and compared their findings with findings from other published studies. However, they excluded postoperative variables, arguing that these are complications and hence may not be true independent predictors of mediastinitis. 16 We contend that identifying postoperative complications that increase the likelihood of developing sternal infections is important. Herein we report a multivariate analysis for the incidence of sternal infections that includes postoperative variables. Also, because treatment strategy and patient outcome differ substantially for superficial and deep sternal infections, we investigated the potential association of risk factors with severity of infection. CHEST I 110 I 5 I NOVEMBER,

2 Table!-Summary of Selected Patient Characteristics and Operative Data* Sternal Wound Infection Categorical Variables Deep Superficial All, No. (%) Control Subjects, No. (%) All patients ,276 CABG only (88) 1,877 (82) Valve only 0 1 (2.4) 206 (9.1) Other (9.8) 193 (85) Female/male 10/11 ll/9 21/20 (51/49) 700/1,576 (31/69) Smoking (56) 1,341 (59) Coronary artery disease (98) 2,018 (89) COPD (20) 371 (16) Insulin-dependent diabetes (46) 197 (8.7) Obesity (51) 324 (14) Emergency (12) 143 (6.3) Redo operation (10) 161 (7.1) Vein grafts only (15) 366 (16) Left IMA only (73) 1,583 (70) Right IMA only 2t 0 2 (4.9) 23 (1.0) Bilateral IMA (4.9) 46 (2.0) Blood transfusion (49) 622 (27) Prolonged ventilation (22) 115 (5.1) Surgical reexploration (15) 77 (3.4) Postoperative bleeding (12) 47 (21) Age 63:+::7 62:+::10 63:+::9 62±11 Perfusion time 90:+::41 104:+::.37 97:+::39 89±37 *Numbers in parentheses indicate percent incidence relative to the total number of patients in each group; quantities in brackets are [minimum-maximum, median].!both right IMA-only patients were redo CABC with prior left IMA grafting. MATERIALS A;\ID METHODS From January 1991 through December 1994, 2,317 patients underwent cardiac surgery at St. Vincent Medical Center. Overall operative mortality in this patient series was 3.06% (71/2,317). All patients underwent a median sternotomy incision and cardiopulmonary bypass. Normothermic perfusion was used in 96% of patients. Cold 02 crystalloid and cold blood cardioplegia were used in 68% and 32% of patients, respectively. The sternum was closed using interrupted stainless steel wires and "delayed" absorbable sutures were used for the subcutaneous tissues. The skin was closed with interrupted nylon or subcuticular absorbable sutures. Prophylaxis to infection included a 2-g dose of a first-generation cephalosporin (cefazolin sodium [Ancef]) administered 2 h prior to surgery, with a similar dose given at the end of bypass. Six or eight additional doses were administered postoperatively at 8-h intervals in patients who had undergone coronary bypass and valve surgeries, respectively. Diagnosis of sternal infections was based on wound tenderness, drainage, cellulitis, fever, and sternal instability. Deep infections were defined as those involving the mediastinum, bone, and/or cartilage, and typically with extensive necrosis of tissues. Infections limited to subcutaneous and soft tissues without mediastinal involvement were considered to be superficial. Patient and operative data were collected retrospectively from the cardiothoracic surgery database (Society of Thoracic Surgery). Univariate analysis was performed using Fisher's Exact Test for categorical variables, the "nonparametric" 'Wilcoxon's two-tailed test for count variables with limited range (eg, number of bypass grafts), or Student's t tests for continuous variables. Risk factors were considered to be univariately significant when their p value was dl.05. Multivmiate analysis was performed with backward elimination logistic regressions. Herein we used univariate significance levels of p<0.2 for covariate inclusion and p<0.05 was chosen for covariate retention. A list of all analyzed variables is provided in Appendix I. Statistical analyses were performed with a statistical software package (SAS; SAS Institute Inc; Cary, NC). Incidence RESULTS A total of 41 sternal wound infections were documented (1.77%). Of these, 21 were deep infections (0.91 %) and 20 (0.86%) were superficial. Patient characteristics and operative data are summarized and compared with the general cardiac surgery population in Table 1. Thirty-six of 41 patients with sternal infections underwent coronary artery bypass grafting (CABG) only, 4 underwent combined CABG and valve procedures, and 1 had valve-only surgery. The average age of patients with sternal infections was 63 years and was similar to the rest of the cardiac surgery patients. Bacteriology Infectious microorganisms were isolated in 95% of patients. Infections were mostly unimicrobial (76%) with no fungal infections documented. Gram-positive bacteria were prevalent (31/48). Staphylococci (29/48), particularly Staphylococcus aureus (20 patients), were the most frequent organisms. Pseudomonas, Serratia, and Klebseilla were the prevalent Gram-negative bacteria at four each Clinical Investigations

3 Table 2-Morbidity and Mortality Associated With Sternal Infections* Sternal Infection Superfical (%) Deep,(%) All,(%) Time to diagnosis 19±9 15±5 16.7±7.3 [8-35, 15) [5-24, 14) [5-35, 14) Mortality 0 of20 (0) 2121 (9.5) 2141 (4.9) Morbidity Recurrence 0 of20 (O) 1/19 (5.2) 1/41 (2.5) LOS 13±10 32± ±18.7 [0-35, 12) [6-87, 27) [0-87, 18) *Quantities in brackets are [minimum-maximum, median). 1Denotes statistical significance (p<0.001) compared with superficial infection group. Diagnosis, Treatment, and Outcome Initial diagnosis of infection varied considerably among patients (Table 2). Time to diagnosis was greater for deep (19±9 days) compared with superficial (15±5) infections, but the difference was not statistically significant. Method of treatment has been described elsewherep Briefly, treatment with IV antibiotics was started immediately following diagnosis, and was followed by prompt wound debridement, and open packing with diluted povidone-iodine (Betadine) with frequent dressing changes (three to four times per day). U pan resolution ofinfection, the chest was closed by delayed primary closure in most patients (37/41). Flap transposition was indicated in four gravely ill patients who had excessively large chest wall defects following debridement of their deep sternal infections. Omentum only was used in two patients, omentum and left pectoralis major in one, and bilateral pectoralis in one. Patient outcome and morbidity differed significantly between the superficial and deep infection groups (Table 3). The overall mortality encountered in patients with wound complications was relatively low at 4.9%. Both deaths occurred in patients with deep infections (9.5%). Recurrence of superficial infection was documented in one patient with a previous deep infection, and was successfully treated with local wound care. Hospital length of stay (LOS) was highly variable among patients (Table 2) and was prolonged by other concomitant complications in some. Overall median LOS was 18 days and was substantially longer for deep (27 days) vs superficial (12 days) infections. Risk Factors Ten of 30 variables were associated with increased risk of sternal infections by univariate analysis (Table 3). In order of importance, these were as follows: insulin-dependent diabetes, obesity, postoperative bleeding, prolonged ventilation, surgical reexploration, use of one or two internal mammary artery (IMA) grafts, number of diseased vessels, postoperative intra-aortic balloon pump (IABP) use, blood transfusions, and female gender. Coronary artery disease (p=0.07), number of grafts (p=0.08), and perfusion time (p=o.ll) approached significance. No variables were significantly different for patients with deep vs superficial infections. The above 13 variables were then included in the multivariate analysis from which 5 independent predictors of sternal infections were identified (Table 4). These were insulin-dependent diabetes, obesity, use of I MAs, surgical reexploration, and transfusion of blood products. DISCUSSION Development of sternal infections following cardiac surgery is multifactorial. This study was aimed at identifying the patient characteristics, operative variables, and postoperative conditions that may predispose these patients to this serious complication. Also, because the severity of infection varies, we attempted to determine whether factors associated with superficial vs deep infections are different. Superficial (0.86%) and deep (0.91%) infections occurred with nearly equal frequency in this patient Table 3-Results of Univariate Risk Factor Analysis* Odds Factor Ratio 95% cr 1 p Value Diabetes overall p<0.001 Insulin dependent 9.5 ( ) p<0.001 Noninsulin dependent 1.3 ( ) p=0.8 Obesity 6.3 ( ) p<0.001 Postoperative bleeding 6.2 ( ) p=0.002 Prolonged ventilation 5.3 ( ) p<0.001 Chest surgical reexploration 4.6 ( ) p=0.004 IMA grafts overall p=0.03 Bilateral vs none 3.9 ( ) p=0.2 Left IMA vs none 1.7 ( ) p=0.3 Right IMA vs none 7.8 ( ) p=0.04 V esse! disease overall p= vs none 3.1 ( ) p= vs none 1.4 ( ) p= vs none 5.2 ( ) p=0.08 IABP 2.9 ( ) p=o.ol Blood transfusion 2.5 ( ) p=0.004 Female 2.5 ( ) p=0.01 *Obesity defined as more than 150% ideal body weight. 1Cl=confidence interval. CHEST I 110 I 5 I NOVEMBER,

4 Table 4-Multivariate Risk Factor Model for Incidence of Sternal Infections Odds Factor Ratio 95% CI* p Value Obesity 4.3 ( ) p<0.001 Diabetes overall p<0.001 Insulin dependent 5.9 ( ) p<0.001 Noninsulin dependent 1.2 ( ) p=0.7 IMA grafts overall p=0.02 Bilateral vs none 9.2 ( ) p=o.ol Left IMA vs none 2.2 ( ) p=0.09 Right IMA vs none 13.3 ( ) p=0.004 Chest surgical reexploration 4.6 ( ) p=0.003 Blood transfusion 2.48 ( ) p=o.ol *CI=confidence interval. series. While a large majority (96%) of the patients undeiwent normothermic cardiopulmonary bypass, incidence of sternal complications was similar for patients with hypothermia and normothermia. The incidence of sternal infections in this predominantly "normothermia" patient series is comparable to other studies in which hypothermia was used. 13.l 4,17 Such similarity in incidence suggests that body core temperature during surgery is not an important factor. A total of ten variables were linked univariately to sternal infections, and these included five postoperative complications or treatments (Table 3). Only five of the ten variables were independent predictors by multivariate analysis: obesity, insulin-dependent diabetes, IMA grafting, blood transfusions, and surgical reexploration. Although previous authors have associated these factors with sternal infection, to our knowledge, this patient series is the first to have all five factors implicated. Also, in contrast with Milano et al, 16 our results highlight the importance of postoperative data in models predicting the incidence of sternal infections. Our results concur with others ,16 that obesity is a major risk factor for sternal infection. In a prospective study, Wilson et al9 found obesity to be an important risk factor for developing median sternostomy wound complications as did two recent large retrospective studies We found a strong role for insulin-dependent diabetes in the development of sternal infections, while other types of diabetes mellitus did not. Multivariate analyses reported elsewhere have diverged as to the role of diabetes.8.lo,ll.l3-15 For example, Rutledge et al,8 He et al,15 and Ottino et al10 did not find a significant role for diabetes in sternal infections following cardiac surgery, contrary to the experience of Loop et al. 13 Grossi and coworkers14 implicated diabetes as a risk factor, but did not find an increased incidence with insulin dependence. Hazelrigg et al11 reported, similar to us, that only insulin-dependent diabetes increased the risk of sternal infections. Use of IMA grafts was linked to a higher incidence of sternal infections in this patient series, especially when dissected bilaterally. Although similar findings have been reported by others,11.l2.l4 the literature is not unanimous with regards to the role ofima use. He et al15 reported that bilateral IMA dissection did not increase the risk of infection compared to using saphenous veins exclusively, or when only one IMA is used. Loop et al13 found that only diabetics who receive bilateral IMA grafts were at increased risk. Our results corroborate those of Loop et al13 that the coincidence of diabetes (especially insulin dependence) and bilateral IMA use increases the risk of developing the complication. But, we also found that both of these variables were independent predictors of sternal infection as well. Decreased sternal perfusion has been hypothesized as a cause for increased incidence of sternal infections following IMA use. However, the impact of IMA harvesting on sternal perfusion remains unresolved. Arnold18 provided evidence that sternal blood flow is severely limited after IMA mobilization, whereas Gromljez and Barner19 contradicted this finding, claiming that the intercostal collateral blood flow from the periosteal plexus compensated for that loss. More recently, Seyfer et al20 showed in primates that IMA mobilization reduced blood flow by 90% within the target sternal half. Similar to us, several other studies3.4.lo.l4 have correlated chest surgical reexploration with increased incidence of sternal complications. Intuitively, one may suspect that the increased time of exposure of internal tissues to airborne pathogens during the reoperation could itself increase the likelihood of infection. Our data also confirmed previous findings that postoperative transfusion of blood products, presumably due to its associated suppression of leukocyte immune function, is correlated with an increased incidence of sternal infections.10.l3 Multivariate statistical models are advantageous because they avoid redundancy in the identified predictors. However, multivariate analysis may exclude true predictors due to characteristics of the patient series under study (eg, sample size). Thus, comparing univariate and multivariate results may further our understanding of causes of sternal infection. For example, we and others found a higher incidence of sternal infections in female patients, yet gender in itself was not a predictor. This is explained by the prevalence of insulin-dependent diabetes (13.3% vs 7.5%), obesity (27.6% vs 9.1 %), and postoperative transfusions ( 43.8% vs 20.4%) in female patients. It is not obvious why female patients received more blood 1176 Clinical Investigations

5 products postoperatively, but we found that reoperation for bleeding was 30% more frequent in women. Postoperative bleeding, prolonged ventilation, and IABP use were not predictors of sternal infections despite their significantly increased frequency in patients with the complication. Postoperative bleeding was redundant with other predictors (surgical reexploration and blood transfusions) because all patients with sternal infection with this complication underwent surgical reexploration and transfusion. Why prolonged mechanical ventilation and postoperative IABP use were not predictors is not clear, but may be related to sample size. Other studies have associated prolonged ventilatory requirement with development of sternal infection.7 22 Hazelrigg et allllinked the use ofiabp to a higher incidence of sternal infection, while others have related prolonged low cardiac output syndrome with the complication. Both forms of treatment indicate a sicker state that may prolong hospitalization, which increases exposure to infection. Ideally, models capable of predicting sternal complications are useful for treatment of patients undergoing cardiac surgery. For example, the studies of Hazelrigg et alll and Loop et al13 provided the basis for avoiding bilateral IMA grafting in obese and insulindependent diabetic patients. From a similar perspective, we believe that significant postoperative bleeding leading to blood transfusions and! or surgical reexploration may be a signal for an increased likelihood of sternal wound complications. Clearly, the chance of developing such infections will be further increased if additional associated factors are present. In those instances, one may argue that additional measures such as wound irrigation/drainage and extending the period of antibiotic coverage are warranted. The usefulness of such measures, however, is only speculative and systematic investigation is necessary. In summary, this study simultaneously linked obesity, insulin-dependent diabetes, surgical reexploration of the mediastinum, use of IMA grafts, and transfusion of blood products to the increased incidence of sternal infections. Moreover, bilateral IMA use increases the risk of sternal infection independent of the presence of diabetes. Finally, risk factors associated with deep vs superficial infections were similar. ACKNOWLEDGMENTS: We thank Nancy Buderer, MS (biostatistician) for performing the statistical analyses, Marsha Shoemaker for database assistance, and Jean Yarberry, RN/PA, for help in compilation of data. REFERENCES 1 Hehrlein FW, Herrmann H, Kraus J. Complications of median sternotomy in cardiovascular surgery. J Cardiovasc Surg 1972; 13: Oschner JL, Mills NL, Woolverton WC. Disruption and infection of median sternotomy incision. J Cardiovasc Surg 1972; 13: Engleman RM, Williams CD, Gouge TH, et al. Mediastinitis APPENDIX I-CLINICAL AND OPERATIVE VARIABLES UsED IN UNNARIATE AND MuLTIVARIATE ANALYSES Preoperative Age Weight Sex Obesity Diabetes mellitus Smoking COPD Dialysis Renal failure Coronary artery disease Vessel disease (none, 1, 2, or 3) Intraoperative Emergency surgery Redo surgery Type of operation (CABG only, valve only, or other) Operative time Perlusion time Cross-clamp time Lowest core temperature Normothermic/hypothermic perfusion Number of grafts (0,1,... ) Vein grafts IMA grafts (none, left, right, or both) Postoperative IABP Reoperation bleeding Reoperation other Surgical reexploration Transfusion of bank blood products Prolonged ventilation Sepsis (prior to infection) Time to diagnosis following open heart surgery: review of 2 years' experience. Arch Surg 1973; 107: Culliford AT, Cunningham JN Jr, Zeff RH, et al. Sternal and costochondral infections following open heart surgery. J Thorac Cardiovasc Surg 1976; 72: Simchen E, Shapiro M, Marin G, et al. Risk factors for postoperative wound infection in cardiac surgery patients. Infect Control 1983; 4: Sarr MG, Gott VL, Townsend TR. Mediastinal infections after cardiac surgery. Ann Thorac Surg 1984; 38: Grossi EA, Culliford AT, Krieger KH, et al. A survey of 77 major infectious complications of median sternotomy: a review of 7,949 consecutive operative procedures. Ann Thorac Surg 1985; 40: Rutledge R, Applebaum RE, Kim BJ. Mediastinal infection after open heart surgery. Surgery 1985; 97: Wilson APR, Livesey SA, Treasure T, et al. Factors predisposing to wound infection in cardiac surgery: a prospective study of 517 patients. Eur J Cardiothorac Surg 1987; 1: Ottino G, DePaulis R, Pansini S, et al. Major sternal wound infections after open heart surgery: a multivariate analysis of factors in 2,579 consecutive operative procedures. Ann Thorac Surg 1987; 44: ll Hazelrigg SR, Wellon HA Jr, Schneider JA, et al. Wound complications after median sternotomy: relationship to internal mammary grafting. J Thorac Cardiovasc Surg 1989; 98: Kouchokos NT, Waering TH, MurphyS, et al. Risks of bilateral CHEST I 110 I 5 I NOVEMBER, n

6 mammary artery bypass grafting. Ann Thorac Surg 1990; 49: Loop F, 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: Grossi, EA, Esposito R, Harris LJ, et al. Sternal wound infections and use of internal mammary artery grafts. Ann Thorac Surg 1991; 102: He G, 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: Milano CA, Kesler K, Archibald N, et al. Mediastinitis after coronary artery bypass graft surgery: risk factors and long term survival. Circulation 1995; 92: Zacharias A, Habib RH. Delayed primary closure of deep sternal wound infections. Texas Heart Inst J 1996 (in press) 18 Arnold M. The surgical anatomy of sternal blood supply. J Thorae Cardiovasc Surg 1972; 64: Gromljez PF, Barner HB. Bilateral internal mammary artery mobilization and sternal healing. Angiology 1978; 29: Seyfer AE, Shriver CD, Miller TR, et al. Sternal blood flow after median sternotomy and mobilization of the internal mammary arteries. Surgery 1988; 104: Breyer RH, Mills SA, Hudspeth AS, et al. A prospective study of sternal wound complications. Ann Thorac Surg 1984; 38: Scully HE, Leclerc Y, Martin RD, et al. Comparison between antibiotic irrigation and mobilization of pectoral muscle flaps in treatment of deep sternal wound infections. J Thorac Cardiovasc Surg 1985; 90: Clinical investigations