• Antibiotherapy;
  • Gram-negative bacteria;
  • mediastinitis;
  • mortality;
  • outcome


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Contributors
  8. Transparency Declaration
  9. References

The aim of this study was to describe the features of a large cohort of patients with postoperative mediastinitis, with particular regard to Gram-negative bacteria (GNB), and assess their outcome. This bicentric retrospective cohort included all patients who were hospitalized in the Intensive Care Unit with mediastinitis after cardiac surgery during a 9-year period. Three hundred and nine patients developed a mediastinitis with a mean age of 65 years and a mean standard Euroscore of six points. Ninety-one patients (29.4%) developed a GNB mediastinitis (GNBm). Of the 364 pathogens involved, 103 GNB were identified. GNBm were more frequently polymicrobial (44% versus 3.2%; p <0.001). Being female was the sole independent risk factor of GNBm in multivariate analysis. Initial antimicrobial therapy was significantly more frequently inappropriate with GNBm compared with other microorganisms (24.6% versus 1.9%; p <0.001). Independent risk factors for inappropriateness of initial antimicrobial treatment were GNBm (OR = 8.58, 95%CI 2.53–29.02, p 0.0006), and polymicrobial mediastinitis (OR = 4.52, 95%CI 1.68–12.12, p 0.0028). GNBm were associated with more drainage failure, secondary infection, need for prolonged mechanical ventilation and/or use of vasopressors. Thirty-day hospital mortality was significantly higher with GNBm (31.9 % versus 17.0%; p 0.004). GNBm was identified as an independent risk factor of hospital mortality (OR = 2.31, 95%CI 1.16–4.61, p 0.0179).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Contributors
  8. Transparency Declaration
  9. References

Mediastinitis is a severe complication of cardiac surgery, occurring in 0.25–2.9% of patients with sternotomy [1-5]. Although aggressive treatments including early surgical debridement and closed aspirative catheter drainage have dramatically improved the outcome, the mortality rate remains high (7–40%) [3, 6-9]. Epidemiological data of mediastinitis have been reported in previous studies [1, 2, 10, 11], but only a few studies have investigated prognostic factors, mainly related to intraoperative events [3, 12]. Furthermore, the limited number of cases in some studies and the heterogeneity of surgical management make the assessment of the prognostic factors difficult.

Even though previous studies have investigated the management of anti-infectious therapies, such as the use of antibiotic prophylaxis or the appropriateness of empirical antibiomicrobial therapy, microbiological data and particularly Gram-negative bacteria (GNB) mediastinitis have rarely been studied as prognostic factors. GNB could be a frequent cause of inappropriate treatment with adverse prognostic potential because these pathogens are underestimated in mediastinitis [2, 3, 5, 13]. Thus far, no studies have systematically studied the impact of GNB infections on the outcome of mediastinitis. The aim of this study was to describe the demographic, clinical and microbiological features of a large cohort of patients with postoperative mediastinitis and more particularly to assess the outcome of GNB mediastinitis (GNBm).

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Contributors
  8. Transparency Declaration
  9. References

Study population

This retrospective cohort analysis was conducted in the intensive care units (ICU) of two hospitals: Européen Georges Pompidou, Paris, France and Centre Cardiologique du Nord, Saint-Denis, France. Patients admitted to the ICU for mediastinitis after cardiac surgery were included over a period of 9 years (January 2000 to December 2008). The diagnosis of mediastinitis was based on the Centers for Disease Control definitions [14] and defined as the presence of a purulent organ space infection of the mediastinum that was confirmed by sternal debridement. In both institutions, a specific committee composed of surgeons, microbiologists, intensivists and hygienists validated mediastinitis cases twice a year. We retrospectively analysed the data of these prospectively validated cases. GNBm was defined by the identification of at least one GNB in the culture samples.

Patient management

For initial surgery, antimicrobial prophylaxis measures were the same in both institutions; cefamandole was infused within 1 h before the incision, and reinfusion was performed every 2 h if the surgery lasted for more than 4 h. Vancomycin was used in patients allergic to β-lactams or for patients hospitalized in the ICU for more than 7 days before the surgery. For mediastinitis, surgical and medical management were similar in both ICU. Surgical care consisted of a mediastinal exploration via sternotomy. Several swabs and collections of samples were performed for bacteriological analysis. Subsequently, any fibrin deposits were removed and necrotized areas of the sternum were resected. As previously described [7, 8], three to ten Redon catheters for aspirating drainage were inserted in infected areas, including the mediastinal cavity and dissected subcutaneous chest wall area. Each catheter was connected to a bottle in which a negative pressure was maintained (Drainobag® 150; B. Braun Melsungen AG, Melsungen, Germany). The effluents collected in each bottle were bacteriologically cultured three times a week. The catheters were progressively removed after two consecutive negative culture results if the daily effluent volume was less than 5 mL. In addition, the usual ICU treatments (mechanical ventilation, vasopressors or renal replacement therapy) were administered as needed; medical care consisted of an intravenous infusion of antibiotics.

Microbiological methods and antibiotherapy

For each patient, direct examination with Gram-staining and cultures of mediastinal needle aspirates and/or surgical material were performed. The initial antibiotic therapy was determined based on a direct examination of Gram-staining. Vancomycin was administered when Gram-positive cocci were identified, β-lactams were administered when GNB were identified. In both case, gentamicin was associated. When Gram-staining was not conclusive, β-lactams, vancomycin and gentamicin were administered. The choice of β-lactams depended on host factors, including the risk for infection with antibiotic-resistant pathogens, with guidance from local susceptibility patterns. This therapy was subsequently modified, if necessary, according to the antibiogram of the responsible microorganisms. Antibiotic therapy was prolonged 6 weeks after the last surgical debridement.

The initial antimicrobial therapy was considered ‘appropriate’ if the treatment regimen included at least one antibiotic active in vitro against identified pathogens, except aminoglycosides alone, and if the dosage and route of administration conformed to current medical standards. We considered antimicrobial therapy ‘inappropriate’ if the drugs used did not have in vitro activity against the isolated strain, if the patient did not receive any antibiotics initially.

Data collection

Baseline clinical and surgical data including the following were collected: age, gender, body mass index, comorbidities (chronic pulmonary disease, extracardiac arteriopathy, diabetes, preoperative renal insufficiency), clinical risk factors of mediastinitis (type of surgery, bypass time and postoperative emergency surgery for bleeding). The perioperative mortality risk status was assessed with the standard European System for Cardiac Operative Risk Evaluation (EuroSCORE) [15].

Collected data related to mediastinitis were time to onset (defined as the time between first surgical procedure and mediastinal debridement), isolated pathogens, mediastinitis-associated bacteraemia, and appropriate or inappropriate characteristics of empirical antimicrobial therapy. Time to onset was classified in two groups: before and after 14 days [16]. Isolated pathogens were independently analysed. The resistant bacteria included methicillin-resistant Staphylococcus aureus (MRSA) and GNB resistant to one or more of extended-spectrum cephalosporins, one of aminoglycosides (tobramycin, amikacin or gentamicin) and ciprofloxacin.

Outcome measures

Outcome data included the in-hospital survival rates, the length of catheter drainage, the ICU length of stay, the incidence of a complicated ICU course (defined as the need for mechanical ventilation for more than 3 days and/or the use of vasopressors), failure of drainage (defined as secondarily infected Redon catheters and/or failure of aspirating Redon catheters), and mediastinitis-associated endocarditis.

This study complied with the Declaration of Helsinki. The data base concerned had previously been approved by the French national commission for computerized files and liberty (CNIL, commission Nationale de l'Informatique et des Libertés).

Statistical analysis

Quantitative variables were described using the mean (± SD) for normally distributed data, the median (interquartile range) for non-normally distributed data and frequency and percentage for qualitative variables. Comparisons of qualitative variables were assessed using the Student's t-test, analysis of variance and Kruskal–Wallis's test as appropriate. Qualitative variables were compared with the chi-square test or Fisher's exact test as appropriate. Parameters from centres A and B were mixed after having verified absence of difference between the both centres (data not shown). Two logistic regression analyses were used to identify independent risk factors associated with the probability of GNBm and with the probability of mortality. All relevant variables achieving a p value <0.05 in the univariate analysis were proposed in a backward stepwise model. Postoperative survival times were assessed from the date of mediastinal debridement to the date of death or the date of departure of the hospital for censored observation. Survival curves were computed according to Kaplan–Meier and compared with the log-rank test. Values of p <0.05 were considered to be significant. The R.2.14.1 statistical software was used for all analyses [17].


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Contributors
  8. Transparency Declaration
  9. References

Microbiological results

During this study, 15 237 cardiac surgical procedures were performed. There were 309 patients (2.03%) that experienced postoperative mediastinitis (centre A, n = 127/7098; centre B, n = 182/8139). Three hundred and sixty-four pathogens were identified in 305 patients (98.7%). In four patients (1.3%) no pathogen was recovered (Table 1). The 364 pathogens identified were split as followed: methicillin-susceptible Staphylococcus aureus (MSSA) (n = 116, 31.9%), GNB (n = 103, 28.3%), coagulase-negative Staphylococcus (n = 68, 18.7%), Streptococcus (n = 31, 8.5%), MRSA (n = 25, 6.9%), and other pathogens (n = 21, 5.8%). In the GNB group, 103 pathogens were isolated. Enterobacteriaceae were the most frequent GNB involved: Escherichia coli (n = 23, 22.3%), Klebsiella pneumoniae (n = 19, 18.5%), Enterobacter cloacae (n = 19, 18.5%), Proteus mirabilis (n = 8, 7.8%), Serratia marcescens (n = 6, 5.8%), Citrobacter koseri (n = 6, 5.8%), Enterobacter aerogenes (n = 5, 4.8%), Morganella morganii (n = 3, 2.9%), Proteus vulgaris (n = 1, 1.0%) and Hafnia alvei (n = 1, 1.0%). Other GNB associated with mediastinitis were Pseudomonas aeruginosa (n = 6, 5.8%), Haemophilius influenzae (n = 3, 2.9%), Bacteroides fragilis (n = 2, 1.9%) and Stenotrophomonas maltophilia (n = 1, 1.0%).

Table 1. Microbiological analysis of the 309 cases of mediastinitis
 All mediastinitis with 364 pathogens
  1. CNS, coagulase-negative staphylococcus; GNB, Gram-negative bacteria; MSSA, methicillin susceptible Staphylococcus aureus; MRSA, methicillin-resistant Staphylococcus aureus.

  2. a

    Including MRSA.

GNB, n (%)103 (28.3)
MSSA, n (%)116 (31.9)
MRSA, n (%)25 (6.9)
CNS, n (%)68 (18.7)
Streptococcus, n (%)31 (8.5)
Other pathogens, n (%)21 (5.8)
Polymicrobial, n (%)47 (15.2)
No. of pathogens isolated n (%)4 (1.3)
Multi-resistant bacteriaa, n (%)68 (22.0)

Among these 309 cases of mediastinitis, 262 were monomicrobial and 47 were polymicrobial. Polymicrobial mediastinitis was documented with four microorganisms in two cases (4.3%), three microorganisms in eight cases (17.0%) and two microorganisms in 37 cases (78.7%). GNB were involved in 40 (85%) of these 47 cases of polymicrobial mediastinitis: multiple GNB (n = 10, 21.3%) or GNB associated with Gram-positive cocci (n = 30, 63.8%). Enterobacteriaceae were the most frequent pathogens associated with polymicrobial mediastinitis (n = 36, 76.6%). Bacteraemia occurred in 173 (56.0%) of the 309 cases of mediastinitis.

Demographic, clinical and microbiological features of GNBm

We compared demographic, clinical and microbiological data of GNBm with mediastinitis due to other pathogens (Table 2). GNBm occurred more often in women (p 0.02), and were more frequently polymicrobial (44.0% (n = 40) versus 3.2% (n = 7); p <0.0001). No other difference between the two groups was identified. The sole independent risk factor of GNBm identified by multivariate analysis was female sex (OR = 2.02, 95% CI 1.15–3.55; p 0.01).

Table 2. Demographic and surgical characteristics of mediastinitis caused by Gram-negative bacteria (GNB)
 All (= 309)GNB (= 91)Other pathogens (= 218)p
  1. Data are number (%) of patients, unless otherwise indicated. BMI, body mass index; GNB, Gram-negative bacteria; IQR, interquartile range; CABG, coronary artery bypass graft.

  2. a

    Renal Insufficiency defined by creatinine >200 μM.

Demographic characteristics
Mean age, years (IQR)67 (57–75)67 (56–74)67 (58–75)0.69
Female sex74 (23.9)30 (33.0)44 (20.2)0.02
Obesity (BMI >30 kg/m2)81 (32.4)31 (38.3)50 (29.6)0.37
Chronic pulmonary disease41 (13.3)12 (13.2)29 (13.3)0.98
Diabetes126 (41.0)44 (48.9)82 (37.8)0.07
Extracardiac arteriopathy70 (22.7)21 (23.1)49 (22.5)0.91
Preoperative renal insufficiencya29 (9.4)8 (8.9)21 (9.7)0.83
Median EuroSCORE, standard (+1 point) (IQR)6 (3–8)6 (3.5–8)5 (3–7)0.11
Immunosuppressive agents9 (2.9)3 (3.3)6 (2.8)0.73
Median left ventricle Function,% (IQR)60 (45–60)60 (40–60)60 (49–60)0.45
Previous cardiac surgery34 (11)12 (13.2)22 (10.1)0.44
Surgical characteristics
Isolated CABG174 (56.3)49 (53.8)125 (57.3)0.57
CABG + valve procedure26 (8.4)9 (9.9)17 (7.8)0.55
Use of internal mammary arteries192 (62.1)56 (61.5)136 (62.4)0.89
Isolated valve replacement85 (27.5)25 (27.5)60 (27.5)0.99
Emergency surgery24 (7.8)8 (8.8)16 (7.4)0.67
Median bypass time, min (IQR)85 (65–114)83 (65–106)87 (65–115)0.74
Median duration of ischaemia, min (IQR)57 (45–73)58 (44–71)57 (46–73)0.79
Reoperation for bleeding39 (12.6)12 (13.2)27 (12.4)0.85
Time of onset to mediastinitis,0.12
<14 days175 (57.9)57 (64.8)118 (55.1) 
≥14 days127 (42.1)31 (35.2)96 (44.9) 

Appropriateness of initial antibiotherapy and GNBm

Appropriateness of antimicrobial therapy could be evaluated among 304 of the 309 patients (98.4%). Initial antimicrobial therapy was inappropriate in 24.6% (n = 22) of the GNB group and in 1.9% (n = 4) of the non-GNB group (p <0.0001). Regarding polymicrobial mediastinitis, antimicrobial therapy was more often inappropriate compared with monomicrobial mediastinitis (34.7% (n = 16/46) versus 3.9% (n = 10/258), p <0.0001). Multivariate analysis showed that GNBm (OR = 8.58, 95% CI 2.53–29.02; p 0.0006) and polymicrobial mediastinitis (OR = 4.52, 95% CI 1.68–12.12; p 0.0028) were independent risk factors of inappropriateness of initial antimicrobial therapy.

GNBm outcome

Regarding postoperative outcome, the failure of drainage, the occurrence of a complicated ICU course, secondary infection and mortality were significantly more frequent with GNBm. However, duration of drainage and hospital length of stay were comparable (Table 3).

Table 3. Postoperative evolution after debridement
 All (= 309)GNB (= 91)Others pathogens (= 218)p
  1. Data are number (%) of patients, unless otherwise indicated. ICU, Intensive care unit; IQR, interquartile range. Failure of drainage was defined as secondarily infected Redon catheters and/or failure of aspirating Redon catheters; Complicated ICU course was defined as the need for mechanical ventilation for more than 3 days and/or the use of vasopressors.

Median length of stay, days (IQR)40 (0–170)38 (1–170)41 (0–136)0.89
Median duration of drainage, days (IQR)21 (0–70)19 (0–70)21 (0–62)0.114
Failure of drainage19 (6.2)10 (11.0)9 (4.1)0.023
Complicated ICU course102 (33.0)38 (41.8)64 (29.4)0.035
Secondary infection42 (13.6)20 (22.0)22 (10.1)0.005
Endocarditis14 (4.5)3 (3.3)11 (5.0)0.71
Death66 (21.4)29 (31.9)37 (17.0)0.004

Thirty days after mediastinitis debridement, death occurred in 29 (31.9%) of 91 patients in the GNB group and in 37 (17.0%) of 218 patients in the non-GNB group (p 0.004). The overall mortality was 21% (n = 66).

Survival using Kaplan–Meier method showed a significant statistical difference between GNB and other mediastinitis (Fig. 1). Death was directly related to the mediastinitis (including mediastinal septic shock or major mediastinal bleeding) in 33 patients (50%), the other causes of death were other causes of septic shock (13 patients, 19.7%), acute haemorrhage (three patients, 4.5%), neurological accidents (four patients, 6.1%), acute heart failure (three patients, 4.5%), and ten cases of others causes (15.2%).


Figure 1. Kaplan–Meier estimates of the probability of survival. GNB, Gram-negative bacteria.

Download figure to PowerPoint

The median time from the initial surgery to mediastinitis debridement was 14 (10–21) days for the survivors and 11 (8–22) days for the non-survivors (p 0.066).

To identify risk factors of mortality, univariate analysis of preoperative, procedure-related, postoperative variables was performed by comparing different subsets of patients. All the variables significantly associated with death in the univariate analysis were integrated into the multivariate analysis (age, sex, preoperative renal insufficiency, previous cardiac surgery, isolated valve replacement, coronary artery bypass graft, use of one or both internal mammary arteries, median bypass time, GNBm, failure of drainage and complicated ICU course). Backward logistic regression found six independent risk factors of mortality: age, baseline renal insufficiency, isolated valve replacement surgery, failure of drainage, a complicated ICU course and GNBm (Table 4). We performed two other univariate analyses where the appropriateness of initial antibiotherapy was adjusted and where patients with inappropriate initial antibiotherapy were excluded. In both cases, the multivariate analysis results did not change.

Table 4. Multivariate analysis of the risk factors for death after mediastinitis in the intensive care unit
 OR (95% CI)p
  1. Results of multivariate analysis after logistic regression: n = 309; 66 non-survivors and 243 survivors.

Age1.07 (1.04–1.11)0.0001
Preoperative renal insufficiency3.58 (1.34–9.51)0.0107
Isolated valve replacement2.25 (1.11–4.49)0.0253
Complicated ICU course6.33 (3.17–12.63)<0.0001
Failure of drainage3.94 (1.19–13.04)0.0245
Presence of Gram-negative bacteria2.31 (1.16–4.61)0.0179


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Contributors
  8. Transparency Declaration
  9. References

In this large cohort of patients, GNBm were frequent, often polymicrobial and frequently associated with an inappropriate initial antimicrobial therapy. GNBm were associated with a high rate of complications such as drainage failure, secondary infection, prolonged mechanical ventilation and/or use of vasopressor and increase of mortality rate.

The microbiological characteristics of mediastinitis in this study are similar to those previously reported [2, 11, 13, 18, 19]. The most common pathogen involved was MSSA followed by GNB. MRSA rate was slightly lower than reported rates in some recent studies [13, 16, 20]. However, the relative frequency of pathogens could be affected by several factors, such as local bacterial ecology, institutional antibiotic policy or both. These variations could partially explain the discrepancy between pathogens reported in different studies. Data about identified GNB are scarce. In our study, Enterobacteriaceae were the most frequent GNB identified. Chen et al. [19] found similar results, but these authors do not distinguish mediastinitis from other postoperative infections, such as bloodstream infections and superficial wound infections.

The mechanisms of mediastinal inoculation by GNB are unclear. The use of a vein graft harvested from a contaminated donor site in coronary artery bypass surgery has been proposed as an explanation [18]. However, this mechanism could not take in account all GNBm. GNBm has been shown to be associated with concomitant infections, especially pneumonia or bacteraemia [21]. Chen et al. [19] demonstrated that patients with major postoperative infections due to GNB have higher rates of renal failure, vascular disease, myocardial infarction, repeat surgery and immunosuppressive treatment, with longer postoperative length of stay. However, in our study, we identified only female sex as an independent risk factor of GNBm. This unexpected result could not be easily explained. We could hypothesize that women have higher prevalence rates of GNB urinary tract infection or colonization, but we did not have any data supporting that assumption. GNBm is not associated with time to onset of mediastinitis in this work, as in the study by Mekonto Dessap et al. [16].

In this study, GNBm is an independent risk factor of mortality. This result was not previously reported. Chen et al. [19] have found that GNB infections are independently associated with postoperative mortality in all patients who underwent cardiac surgery, but not when compared with other pathogens in patients with postoperative infections. Inappropriateness of initial antibiotherapy may explain, in part, the worse outcome of GNBm. We identified GNBm as an independent risk factor of inappropriate initial antimicrobial therapy. In patients with septic shock, initial inappropriate antibiotherapy is associated with a fivefold decrease in survival [22]. Zahar et al. [23] showed that, in patients with severe sepsis, the characteristics of the infectious process (site and causative organism) were not associated with mortality, but early appropriate antimicrobial therapy was consistently associated with better survival.

Our study had some limitations. This study might have yielded different results if we had included additional data such as the delay between the first clinical signs of infection and the initiation of antibiotics or surgical debridement. Unfortunately, these data cannot be reliably provided regarding the retrospective design of this study. Moreover, despite a large cohort of patients, it was not possible to calculate a mortality rate for each type of GNB involved.

In conclusion, we showed that GNBm was associated with more severe outcome leading to decrease in survival. No alterable risk factor for GNBm was observed. However, we supported that this severe outcome was partly related to an inappropriateness of initial antimicrobial therapy. This result emphasizes the importance of knowledge of the local ecology in the choice of empirical antibiotherapy and the necessity of the earliest identification of the involved pathogens to allow determination of their susceptibility to antibiotics. A more frequent use of antibiotics directed against resistant GNB should be considered according to local microbial environment.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Contributors
  8. Transparency Declaration
  9. References

All authors contributed to this study design draft by EG. HC, JMM, OM, EP and PLB contributed to data recovery. MF performed the statistical analysis and verified its accuracy. All authors helped draft this report or critically revise the draft. All authors reviewed and approved the final version of the report.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Contributors
  8. Transparency Declaration
  9. References
  • 1
    The Parisian Mediastinitis Study Group. Risk factors for deep sternal wound infection after sternotomy: a prospective, multicenter study. J Thorac Cardiovasc Surg 1996; 111: 12001207.
  • 2
    Borger MA, Rao V, Weisel RD et al. Deep sternal wound infection: risk factors and outcomes. Ann Thorac Surg 1998; 65: 10501056.
  • 3
    De Feo M, Renzulli A, Ismeno G et al. Variables predicting adverse outcome in patients with deep sternal wound infection. Ann Thorac Surg 2001; 71: 324331.
  • 4
    Eklund AM, Lyytikainen O, Klemets P et al. Mediastinitis after more than 10,000 cardiac surgical procedures. Ann Thorac Surg 2006; 82: 17841789.
  • 5
    Lepelletier D, Perron S, Bizouarn P et al. Surgical-site infection after cardiac surgery: incidence, microbiology, and risk factors. Infect Control Hosp Epidemiol 2005; 26: 466472.
  • 6
    Braxton JH, Marrin CA, McGrath PD et al. 10-year follow-up of patients with and without mediastinitis. Semin Thorac Cardiovasc Surg 2004; 16: 7076.
  • 7
    Calvat S, Trouillet JL, Nataf P, Vuagnat A, Chastre J, Gibert C. Closed drainage using Redon catheters for local treatment of poststernotomy mediastinitis. Ann Thorac Surg 1996; 61: 195201.
  • 8
    Durandy Y, Batisse A, Bourel P, Dibie A, Lemoine G, Lecompte Y. Mediastinal infection after cardiac operation. A simple closed technique. J Thorac Cardiovasc Surg 1989; 97: 282285.
  • 9
    Milano CA, Kesler K, Archibald N, Sexton DJ, Jones RH. Mediastinitis after coronary artery bypass graft surgery. Risk factors and long-term survival. Circulation 1995; 92: 22452251.
  • 10
    El Oakley RM, Wright JE. Postoperative mediastinitis: classification and management. Ann Thorac Surg 1996; 61: 10301036.
  • 11
    L'Ecuyer PB, Murphy D, Little JR, Fraser VJ. The epidemiology of chest and leg wound infections following cardiothoracic surgery. Clin Infect Dis 1996; 22: 424429.
  • 12
    Kirsch M, Mekontso-Dessap A, Houel R, Giroud E, Hillion ML, Loisance DY. Closed drainage using Redon catheters for poststernotomy mediastinitis: results and risk factors for adverse outcome. Ann Thorac Surg 2001; 71: 15801586.
  • 13
    Trouillet JL, Vuagnat A, Combes A et al. Acute poststernotomy mediastinitis managed with debridement and closed-drainage aspiration: factors associated with death in the intensive care unit. J Thorac Cardiovasc Surg 2005; 129: 518524.
  • 14
    Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC definitions of nosocomial surgical site infections, 1992: a modification of CDC definitions of surgical wound infections. Infect Control Hosp Epidemiol 1992; 13: 606608.
  • 15
    Roques F, Nashef SA, Michel P et al. Risk factors and outcome in European cardiac surgery: analysis of the EuroSCORE multinational database of 19030 patients. Eur J Cardiothorac Surg 1999; 15: 816822.
  • 16
    Mekontso DA, Vivier E, Girou E, Brun-Buisson C, Kirsch M. Effect of time to onset on clinical features and prognosis of post-sternotomy mediastinitis. Clin Microbiol Infect 2011; 17: 292299.
  • 17
    R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, 2012. Available at:
  • 18
    Gardlund B, Bitkover CY, Vaage J. Postoperative mediastinitis in cardiac surgery – microbiology and pathogenesis. Eur J Cardiothorac Surg 2002; 21: 825830.
  • 19
    Chen LF, Arduino JM, Sheng S et al. Epidemiology and outcome of major postoperative infections following cardiac surgery: risk factors and impact of pathogen type. Am J Infect Control 2012; 40: 963968.
  • 20
    Jarlier V, Trystram D, Brun-Buisson C et al. Curbing methicillin-resistant Staphylococcus aureus in 38 French hospitals through a 15-year institutional control program. Arch Intern Med 2010; 170: 552559.
  • 21
    Bor DH, Rose RM, Modlin JF, Weintraub R, Friedland GH. Mediastinitis after cardiovascular surgery. Rev Infect Dis 1983; 5: 885897.
  • 22
    Kumar A, Ellis P, Arabi Y et al. Initiation of inappropriate antimicrobial therapy results in a fivefold reduction of survival in human septic shock. Chest 2009; 136: 12371248.
  • 23
    Zahar J-R, Timsit J-F, Garrouste-Orgeas M et al. Outcomes in severe sepsis and patients with septic shock: pathogen species and infection sites are not associated with mortality. Crit Care Med 2011; 39: 18861895.