Mediastinitis after cardiac valve operations

J

THoRAc CARDIOVASC SURG

90:517-522, 1985

Mediastinitis after cardiac valve operations Impact upon survival Mediastinitis after cardiac valve replacement is a dreaded complication with consequent mortafity estimated as high as 70%. We have reviewed 2,491 patients with cardiac valve operatiom to assess the impactof mediastinitis uponmortafityin our institutionin the past 10 years. Mediastinitisdeveloped after valve replacement in 36 patients (1.4%). All patients required operative intervention for mediastinal infection with positive bacterial cultures. Twelve of these patients bad other perioperative problems associated with a high mortafity independent of mediastinitis: bacterial endocarditis not cured by valve replacement (three), recent preoperative myocardial infarction(four), triple valve disease with biventricular failure (one), and severe perioperative cerebraldamage (four~ Ten of these bigh-riskpatients died(83.3%~ The impactof mediastinitis upon survival is best evaluated in the remaining 24 patients withoutbigh-risk perioperative problems. Eight of these patients were managed before 1980 with debridement and irrigationas the primary treatment, with two hospital deaths (25% ~ Pectoral or reens meseleflaps were frequently used after 1980 (flaps in 11 of 16 patients), leading to a significantly shorter time between diagnosis of infection and hospitaldischargefree of infection (62 versus 385 days, p < 0.05). Only one of these 16 patients died. Valve re-replacement for endocarditis was performed in three of these 24 patients although 13 of 24 had positive blood cultures.Mediastinitisafter valve operatiom in the absenceof other bigh-risk perioperative problems can be successfully managed. Early debridement and mltiCle flap closure bas led to a 94 % survival rate in 16 patients during the past 4 years.

Edson H. Cheung, M.D. (by invitation), Joseph M. Craver, M.D., Ellis L. Jones, M.D., Douglas A. Murphy, M.D. (by invitation), Charles R. Hatcher, Jr., M.D., and Robert A. Guyton, M.D. (by invitation), Atlanta, Ga.

Rrulent drainage from a sternal wound after an otherwise uncomplicated cardiac operation signals the beginning of a period of major morbidity and perhaps death. The incidence of these infections is less than 5%, but the associated mortality ranges from 10% to 71%.1-6 This complication is even more fearful if, beneath the infection, a prosthetic valve raises the specter of endocarditis. The occurrence of mediastinitis after prosthetic valve implantation raises difficult clinical problems. Can the treatment of mediastinitis be altered to minimize the possibility of subsequent endocarditis? Does the presFrom the Carlyle Fraser Heart Center, Crawford W. Long Memorial Hospital and Emory University Hospital, Department of Surgery, Emory University School of Medicine, Atlanta, Ga. Read at the Sixty-fifth Annual Meeting of The American Association for Thoracic Surgery, New Orleans, La., April 29-May I, 1985. Address for reprints: Robert A. Guyton, M.D., Carlyle Fraser Heart Center, Crawford W. Long Memorial Hospital, Atlanta, Ga.

30365.

ence of bacteremia necessarily mean that the valve must be replaced? What is the impact of mediastinitis upon survival after prosthetic valve implantation? Our struggle with these questions in the past few years has led to this review of our experience in the treatment of this potentially disastrous complication.

Patients and methods From May, 1974, through May, 1984, 14,531 patients undergoing cardiac operations at Emory University Hospital and Crawford W. Long Hospital were recorded by the Emory University Cardiac Data Bank. Within this group, 2,491 patients (17%) had cardiac valve operations. Fifty-five percent of these patients were male and 45% were female. Their mean age was 55. Fifty-two percent of these patients had aortic valve replacement, 40% had mitral valve operations, and 8% had various combinations of valves or valve plus other cardiac operations. Ninety-four percent of the diseased valves were replaced with porcine xenografts and the remainder were replaced with mechanical valves. 517

.The Journ-al of Thoracic and Cardiovascular Surgery

5 1 8 Cheung et al.

Table I. Bacterial infections in 36 patients with mediastinitis after valve operations Non-high risk In = 24) High risk In = 12) Bacteria Gram-negative

COCCI

Wound

I

Blood

Wound

3

3

2

S. aureus

3

S. epidermidis

o

Gram-negative rods

3

Mixed culture

7

Total patients with positive cultures Percent

Group A In = 8)

I

Group B In = 16)

Blood

Wound

3

7

2

2

5

3

a

2

I

Blood 3

4

3

3

o 3

5

6

12

9

8

4

16

9

lOa

75

100

50

100

56.3

General management of patients undergoing valve operations. All patients received povidone-iodine (Betadine) scrubs and were shaved from chin to knee or ankle on the night prior to operation or on the morning of operation. The operative field was scrubbed and painted with povidone-iodine solution, and adhesive disposable drapes were applied to all surfaces of the operative field prior to draping. Prophylactic antistaphylococcal or broad-spectrum antibiotics were used in all Patients prior to operation and continued for 48 hours postoperatively or until chest tubes were removed. Oxacillin and cephalothin were the most frequently used drugs in the early period of the study; in the past few years, other cephalosporins were administered routinely. For patients with penicillin allergy, vancomycin was the antibiotic of choice. Surgical technique. In nearly all patients, the operative approach was through a median sternotomy with cardiopulmonary bypass and systemic hypothermia (25 0 to 28 0 C). Hypothermic hyperkalemic cardioplegia has been used routinely in our institution for myocardial preservation since 1976 instead of coronary perfusion. The technique of valve implantation has been previously reported.?" Follow-up of patients treated for mediastinitis was obtained by patient contact, report from the family physician, cardiologist, or surgeon, or from chart review. Statistical analysis was performed by standard methods. The Student's t test was employed for the comparison of the mean and variance. Fisher's exact test and chi square analysis with Yates' correction were used in the evaluation of survival rates.

Results In the past 10 years, the overall rate of sternal or mediastinal wound infection was 1.1 % (153/14,531). Following cardiac valve operations, 48 patients (1.9%) developed sternal or mediastinal wound infections. The difference in wound infection rates in valve versus non-valve patients was significant (1.9% versus 0.9%, p < 0.01). For the purpose of this study, mediastinitis was defined as clinically significant infection involving the mediastinum below the sternum with purulent drainage and positive bacterial cultures. Twelve of the aforementioned 48 patients were found to have wound problems or superficial infections that did not penetrate below the sternum. The remaining 36 patients had mediastinitis by our definition. The retrospective analysis of these 36 patients with mediastinal infection is the focus of this report. Mediastinitis. All 36 patients with mediastinitis had positive wound cultures and 23 (64%) had positive blood or valve cultures. Twenty-two patients had positive blood cultures with the majority of these patients having multiple positive samples. Three patients had positive valve cultures, including one patient who had both positive blood and valve cultures (Table I). There were 13 deaths among these 36 patients (36.1%). However, 12 of these patients had other serious perioperative problems associated with a high mortality independent of mediastinitis. These patients merit special consideration. High-risk patients. The 12 patients with high-risk perioperative problems can be divided into four groups: three patients with persistent postoperative bacterial

Volume 90 Number 4

Mediastinitis 5 I 9

October. t 985

Table Il, Profile of 24 patients with mediastinitis No. of patients Age (yr) Mean Range Male: female No. of valves replaced per patient No. of "redo" patients Diagnosis of infection Mean postop, days Range (days) Duration of antibiotic treatment (days) Mean Range Time from diagnosis to hospital discharge free of infection (days) Mean Range

Group A

Group B

8

16

51.3 33-75 7:1 1.13

57.1 28-77 10:6 1.31

2

6

11.3 5-35

11.4 6-24

41.4 14-88

46.5 14-125

385 23-1,710

62 15-150

endocarditis, four patients with preoperative myocardial infarction, four patients with major perioperative cerebral damage, and one patient with profound biventricular failure preoperatively. Three patients had persistent bacterial endocarditis not cured by antibiotics and valve replacements. One patient, who was a drug addict, had recurrent Staphylococcus aureus endocarditis despite appropriate antibiotics and three aortic valve replacements in 8 months. He finally died of multi-organ failure with S. aureus mediastinitis. At autopsy, bacterial endocarditis was again confirmed. The second patient had prosthetic valve endocarditis (S. aureus) 4 years after his original aortic valve replacement. Mediastinitis (S. epidermidis) developed after the second valve operation. He had persistent bacteremia (S. aureus) and developed aortic insufficiency that necessitated a third valve operation. This patient could not be weaned from cardiopulmonary bypass and subsequently died. The last patient was transferred to our institution in septic shock with renal and respiratory failure. She developed endocarditis (Hemophilus parainjluenzae) on a prolapsed mitral valve while receiving steroid treatment for ulcerative colitisand she died 2 weeks after valve replacement with continuing sepsis, Pseudomonas and Candida mediastinitis, and multi-organ failure. Four patients had preoperative myocardial infarction with hemodynamic deterioration associated with their valve lesions, which forced urgent operations. All four patients were over 70 years of age. Two had aortic stenosis with persistent angina and heart failure 2 to 3

Table m. Postoperative complications Pulmonary infection necessitating tracheostomy Urinary tract infection Acute renal failure necessitating dialysis Postop. bleeding, reexploration Arrhythmias Endocarditis Seizures Gastrointestinal bleeding Cholecystitis

Group A

Group B

4 1

o

8 3 6

2

2

o 1

I

o 1

I

o

2 3 2 3 1

o 1

weeks following infarction. The other two patients developed acute mitral regurgitation after their myocardial infarction. All four patients had protracted postoperative courses and developed mediastinitis. Their wounds were successfully treated with antibiotics and muscle flaps. Two survived and were discharged after 3 months of hospitalization. The other 2 patients died suddenly, presumably of ventricular tachyarrhythmias, I and 3 months after valve replacement. Four other patients had' major perioperative cerebral damage and were neurologically unresponsive throughout their postoperative courses. One of these patients also had a preoperative myocardial infarction with mitral insufficiency that prompted his valve operation. All four patients had tracheostomies and required long-term ventilatory support and eventually died of multi-organ failure. The remaining patient had a history of rheumatic fever with triple valve disease, hepatic dysfunction, and severe biventricular failure preoperatively. He underwent aortic, mitral, and tricupsid valve replacement. Postoperatively, the patient bled massively with a severe coagulopathy and required mediastinal exploration twice in 48 hours for bleeding. He remained in low cardiac output with progressive hepatic and renal failure and died 22 days after operation. The impact of mediastinitis upon survival and the effectiveness of its management is difficult to evaluate when mortality is probable because of other factors. The remaining patients without high-risk problems present a more reasonable test of the consequences of mediastinitis after valve replacement. Patients without high-risk problems. There were 24 patients without serious high-risk perioperative problems in the 10 year period between 1974 and 1984 who developed mediastinitis after otherwise successful valve operations (Tables II and III). Eight patients (Group A) were treated before 1980. The principal treatment of

The Journal of Thoracic and Cardiovascular Surgery

5 2 0 Cheung et al.

ca

64% in 36 patients from 1974 to 1984

Q

88% in 24 patients without other high risk perioperative problem

G

94% in 16 of 24 patients in whom muscle flap closure was aggressively used

Fig. 1. Survival rate in patients with mediastinitis after valve operations.

mediastinitis in these patients was appropriate intravenous antibiotics based on wound culture, thorough wound debridement and sternal rewiring (with or without skin closure), and continuous antibiotic irrigation of the wound with 0.25% neomycin solutionfor at least 48 hours after debridement. In the late 19708, pectoralis major or rectus abdominis muscle flap closure methods were developed in our institution." These procedures were used as secondary therapy for persistent wound infections when other conventional treatment modalities proved unsuccessful. Of the eight patients in Group A, four eventuallyrequired muscleflap woundclosureafter three or more unsuccessful surgical attempts to control the infection. Seven flap procedures were performed in these four patients, with an infection control rate of 57%. Two patients having valveoperationsin the early period, who had persistent wound infections that were finally controlled with muscle flap or conventional methods, developed mediastinitis with the same organisms after valve re-replacement 2 to 3 years later. Both patients subsequently required more than one muscle flap operation before the infections were finally cured. Sixteen patients (Group B) were treated after 1980 with a different approach to mediastinitis. Since 1980, the muscle flap operation has been the preferred primary therapy in our institution for patients with serious mediastinitis after cardiac operations. The procedure is usually performed in two stages. In the first operation, the mediastinal wound is debrided and the wound is left open to be treated thrice daily with dressing changes. The wound is then closed with muscle flaps by our plastic surgery colleagues a few days later when there has been complete open drainage of the wound with no evidence of loculatedareas of infection. In four Group B patients, the wounds were judged to be clean enough to be closed primarily without muscle flaps. One patient died after debridement and drainage, but before wound closure, Eleven of the 16 Group B patients had their

mediastinal wounds treated in this manner. Two patients required repeat muscle flap operations, using a different muscle, when the initial flaps failed to control the infection. All 11 patients survived, with an infection control rate of the muscle flap procedure of 85%. The bacteriologic data in both groups of patients are presented in Table I. All patients with bacteremia received intravenous antibiotics for at least 4 weeks. Of particular interest were three patients in Group B who developed S. aureus prostheticvalve endocarditis 3 to 4 weeks after the diagnosis of mediastinitis was made. All three were cured with muscle flap coverage, valve replacement, and antibiotics. The diagnosis of mediastinitis was usually made 1 to 3 weeks postoperatively in all patients. There was no significant difference between the two groups in the duration of antibiotic treatment. The mean time period from initial diagnosis of mediastinitis to fmal hospital discharge free of wound infection was significantly shorter in Group B (62 days in Group B versus385 days in Group A, p < 0.05). Mortality. In the entire group of patients with mediastinitis, the mortality rate was 36.1% (13/36). Excluding those patients with serious perioperative high-risk problems, there were three deaths in 24 patients (12.5%). Two were in Group A (before 1980) (25%) and one was in Group B (after 1980) (6.25%) (Fig. 1). Becauseof the small sample size, there was no significantstatistical differencein mortality between the two groups. FoUow-up. Follow-up records were obtained in 20 of the 21 surviving "non-high risk" patients. One patient was lost to follow-up after the first postoperative visit. There was one late death of a noncardiac cause that occurred 4 months after hospital discharge. The mean duration of follow-up in the 19 long-term survivors was 28.1 months with a range of 9 to 85 months after fmal hospital discharge cured of infection. There have been no recurrent infections, no flap-related hernias, and no other late wound complications. Tow patients required valve re-replacement 5 and 10 years after their initial valve operations. Discussion Sternal and mediastinal infections after cardiac operations are dreaded complications that cause significant morbidity and mortality. Various reports have discussed the management of sternal wound complications following cardiac operations.":" However, in most of these studies, patients who underwent various cardiac operations and had sternal or mediastinalwoundconditions of differing severity were included. This heterogeneous

Volume 90 Number 4 October, 1985

patient population makes interpretation and comparison of results difficult. In reviewing our 10 year experience in over 2,000 patients who underwent cardiac valve operations at Emory University hospitals, we found an overall incidence of 1.9% for sternal infection and/or mediastinitis. This infection rate is comparable to that reported in other series.l-" 13 We found a significantly higher rate of sternal wound problems in patients undergoing valve operations than in patients undergoing operations not involving the cardiac valves. This result is not unexpected. Valve operations are more likely to include lengthy procedures related to extensive disease or reoperation after previous valve repair or replacement. Furthermore, valve operations are often performed on patients with endocarditis, many of whom have positive bacterial cultures from the operative field. In our series, sternal infections presented in three different forms. Twelve patients had wound breakdown or superficial infections that did not penetrate beneath the sternum. These infections were generally benign. Twelve other patients were gravely ill in the perioperative period and had major problems such as persistent endocarditis, severe neurological damage, or multiorgan failure from low cardiac output. Mediastinitis presented in these patients often as a preterminal event, as multi-organ failure extended to include wound breakdown and failure of defense against bacterial invasion. In these situations, mediastinitis must be viewed more as an accomplice than as a perpetrator of the death. The fmal group of 24 patients are those of most interest to this discussion-patients with otherwise successful operations who develop mediastinitis. It is in these patients that the most knowledge can be gained about the impact of mediastinitis and its treatment upon morbidity and mortality. Overall mortality in the 24 patients who had undergone otherwise successful valve operations and developed mediastinitis was 12.5% (3/24). In the 10 year study period, our philosophy of treatment has changed. Prior to 1980, treatment consisted of drainage, debridement, and sternal closure with antibiotic irrigation. This method of treatment was successful in only two of eight patients treated prior to 1980. Four patients required subsequent treatment with muscle flap procedures, and two patients died. The mean time from diagnosis of infection to hospital discharge free of infection was terribly long (385 days). One patient required seven operations before the infection was controlled. Well-vascularized muscle flaps have been shown to be an effective treatment for persistent and chronic wound infections.v'v'r" We 9 have previously demonstrated

Mediastinitis

521

that the use of pectoralis major and rectus abdominis muscle flaps in wound coverage might reduce the mortality and morbidity associated with mediastinitis after cardiac operation. Since 1980, the expeditious use of muscle flap closure has been the cornerstone of our therapy for mediastinitis. Of 16 "non-high risk" patients treated since 1980 with this technique, only one has died. Bacteremia was documented in 58% of the 24 "non-high risk" patients. It is unclear whether this complication was bacteremia secondary to the wound infection or prosthetic valve endocarditis. We had no direct evidence to show that mediastinitis caused endocarditis. However, the incidence of endocarditis in these patient was fifteen times higher than the overall rate of endocarditis in our institution among patients after valve operations (12.5% versus 0.8%, p < 0.01). Three patients had muscle flap closure and valve replacement done in the same setting. One patient had the muscle flap procedure performed I week prior to the second valve operation. At the time of the second valve operation, the muscle flap was elevated and the heart was exposed. A small amount of additional muscle was mobilized to help close the wound after valve rereplacement. All three patients with endocarditis received intravenous antibiotics for at least 4 weeks, and all three patients survived. The fact that none of our surviving patients developed endocarditis after the mediastinal wound healed during the long-term follow-up period suggests that early control of mediastinitis in patients after valve replacement may reduce the incidence of consequent valve infection. Postoperative infection with S. aureus presented special problems in the 24 "non-high risk" patients. Six patients had S. aureus wound infections. Five of the six patients also had S. aureus bacteremia. Of these patients with S. aureus bacteremia, two developed endocarditis and two died. The sixth patient with S. aureus wound infection but negative blood cultures developed persistent fever and hemodynamic instability. The valve was replaced and valve cultures grew S. aureus. All of these S. aureus infections were methicillin resistant. The bacteremia with organisms other than S. aureus occurred in seven other patients in this "nonhigh risk" subgroup, and none of these patients required valve replacement for prosthetic valve endocarditis. We conclude that early use of muscle flaps for mediastinitis may decrease the incidence of consequent endocarditis, promote faster eradication of bacteria from the mediastinum and sternal tissues leading to faster wound closure, and reduce chronic wound problems. With its use, we have achieved a 94% survival rate

The Journal of Thoracic and Cardiovascular Surgery

5 2 2 Cheung et al.

since 1980 in patients who have developed mediastinitis after otherwise successful valve operations. We wish to thank Ms. Marianne Falls for preparation of the manuscript and the continued support of the Emory University Cardiac Data Bank.

2

3

4

5

6

7

8

REFERENCES Culliford AT, Cunningham IN Jr, Zeff RH, Isom OW, Teiko P, Spencer FC: Sternal and costochondral infections following open-heart surgery. A review of 2,594 cases. J THORAC CARDlOVASC SURG 72:714-726, 1976 Engelman RH, Williams D, Gouge TH, Chase RM, Falk EA, Boyd AD, Reed GE: Mediastinitis following open heart surgery. Review of two years' experience. Arch Surg 107:772-778, 1973 Grmoljez PF, Barner HH, Willman VL, Kaiser GC: Major complications of median sternotomy. Am J Surg 130:679-681, 1975 Jimenez-Martinez M, Arguero-Sanchez R, Perez-Alvarez J, Mena-Castenada P: Anterior mediastinitis as a complication of median sternotomy incisions. Diagnostic and surgical considerations. Surgery 67:929-934, 1970 Serry C, Bleck PC, Javid H, Hunter JA, Goldin MD, DeLaria GA, Najafi H: Sternal wound complications. J THORAC CARDlOVASC SURG 80:861-867, 1980 Ochsner JL, Mills NL, Woolverton WC: Disruption and infection of the median sternotomy incision. J Cardiovasc Surg 13:394-399, 1972 Craver JM, Jones EL, McKeown P, Bone DK, Hatcher CR Jr, Kandrach M: Porcine cardiac xenograft valves. Analysis of survival, valve failure and explanation. Ann Thorac Surg 34:16-21, 1982 Jones EL, Craver JM, Morris DC, King SB III, Douglas JS, Franch RH, Hatcher CR Jr, Morgan EA: Hemody-

9

10

11

12

13

14

15

16

namic and clinical evaluation of the Hancock xenograft bioprosthesis for aortic valve replacement (with emphasis on management of the small aortic root): J THORAC CARDlOVASC SURG 75:300-308, 1978 Jurkiewicz MJ, Bostwick J, III, Hester TR, Bishop JB, Craver JM: Infected median sternotomy wound. Successful treatment by muscle flaps. Ann Surg 191:738-744, 1980 Shumaker HB, Mandelbaum L: Continuous antibiotic irrigation in the treatment of infection. Arch Surg 86:384387, 1963 Bryant LR, Spencer FC, Trinkle JK: Treatment of median sternotomy infection by mediastinal irrigation with an antibiotic solution. Ann Surg 169:914-920, 1969 Thurer RJ, Bognolo D, Vargas A, Isch JH, Kaiser GA: The management of mediastinal infection following cardiac surgery. J THORAC CARDIOVASC SURG 68:962-968, 1974 Sarr MG, Gott VL, Townsend TR: Mediastinal infection after cardiac surgery. Ann Thorac Surg 38:415-423, 1984 Nahai F, Morales L, Bone DK, Bostwick J, III: Pectoralis major muscle turnover flaps for closure of the infected sternotomy wound with preservation of form and function. Plast Reconstr Surg 70:471-474, 1982 Herrera HR, Ginsburg ME: The pectoralis major myocutaneous flap and omental transposition for closure of infected median sternotomy wounds. Plast Reconstr Surg 70:465-470, 1982 Pairolero PC, Arnold PG: Management of recalcitrant median sternotomy wounds. J THoRAe CARDlOVASC SURG 88:357-364, 1984

[For Discussion see page 528.}