| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2005;129:518-524
© 2005 The American Association for Thoracic Surgery
Surgery for Acquired Cardiovascular Disease |
a Service de Réanimation Médicale, Hôpital La Pitié–Salpêtrière Assistance Publique–Hôpitaux de Paris, Paris, France
b Service de Chirurgie Thoracique et Cardiovasculaire, Hôpital La Pitié–Salpêtrière Assistance Publique–Hôpitaux de Paris, Paris, France
Received for publication December 8, 2003; revisions received July 2, 2004; accepted for publication July 13, 2004. * Address for reprints: Jean-Louis Trouillet, MD, Service de Réanimation Médicale, Pr Gibert, Institut de Cardiologie, Hôpital Pitié–Salpêtrière 47-83 boulevard de l'Hôpital, 75651 Paris Cedex 13, France (E-mail: jean-louis.trouillet{at}psl.ap-hop-paris.fr).
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
METHODS: Over a 10-year period, 316 consecutive patients with mediastinitis occurring less than 30 days after sternotomy were treated in a single unit. First-line therapy was closed-drainage aspiration with Redon catheters. Variables recorded, including patient demographics, underlying disease classification, clinical and biologic data available at intensive care unit admission and day 3, and their association with intensive care unit mortality, were subjected to multivariate analyses.
RESULTS: Intensive care unit mortality (20.3%) was significantly associated with 5 variables available at admission: age greater than 70 years (odds ratio, 2.70), operation other than coronary artery bypass grafting alone (odds ratio, 2.59), McCabe class 2/3 (odds ratio, 2.47), APACHE II score (odds ratio, 1.12 per point), and organ failure (odds ratio, 2.07). After introducing day 3 variables into the logistic regression model, independent risk factors for intensive care unit death were as follows: age greater than 70 years, operations other than coronary artery bypass grafting alone, McCabe class 2/3, APACHE II score, mechanical ventilation still required on day 3, and persistently positive bacteremia. For patients receiving mechanical ventilation for less than 3 days, mortality was very low (2.4%). In contrast, for patients receiving mechanical ventilation for 3 days or longer, mortality reached 52.8% and was associated with non–coronary artery bypass grafting cardiac surgery, persistently positive bacteremia, and underlying disease.
CONCLUSIONS: In patients requiring intensive care for acute poststernotomy mediastinitis, age, type of cardiac surgery, underlying disease, and severity of illness at the time of intensive care unit admission were associated with intensive care unit death. Two additional factors (mechanical ventilation dependence and persistently positive bacteremia) were identified when the analyses were repeated with inclusion of day 3 patient characteristics.
|
Mediastinitis is a severe complication of median sternotomy responsible for significant morbidity and mortality in adults. Thirty years ago, these wounds were treated with open packing or continuous mediastinal irrigation.1,2 The poor results obtained in earlier studies, with mortality approaching 50%, encouraged surgeons to develop other techniques, such as complex plastic surgery procedures with muscle flaps.3 In 1989, Durandy and colleagues4 proposed a simple, primary closed-drainage technique with small multiperforated rigid catheters (Redon catheters) to drain all infected areas after meticulous wound debridement. The results of our previous experience suggested that this first-line treatment was effective.5
Compared with other surgical techniques, closed drainage with Redon catheters has several advantages: it is simple and rapidly put in place, it is less work intensive for nurses, and it leaves a discreet scar. Thus first-line surgical treatment of mediastinitis has been standardized with this technique for over 10 years in all cardiothoracic centers in Paris, France, and, more recently, in other European centers.6
However, although technical procedures have been described in detail and the epidemiology of deep sternal wound infections has been extensively studied,7-10 few reports deal with prognostic factors in patients with mediastinitis.11,12 Most of the factors considered in those previous studies were related to the initial cardiac surgery. The heterogeneity in antibiotic regimens and in surgical techniques, which are used according to each team's experience in the absence of consensual guidelines,2-5,13-16 contributes to the difficulty of performing such studies.
Over a 10-year period, we prospectively collected specific data that might have an effect on the outcomes of patients with acute poststernotomy mediastinitis. Therefore the aim of this study is to describe the experience of a single intensive care unit (ICU) in the treatment of poststernotomy mediastinitis with closed-drainage aspiration with Redon catheters and to identify independent factors associated with ICU death.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Finally, 17 (4.8%) of the 353 patients were initially treated with the open technique. Among them, 14 had been referred to our ICU during the first year (1991) of the study period, when surgical teams transferring patients to our unit were using indiscriminately open and closed techniques. In the 3 other patients, the thorax was left open because of extensive necrotic lesions. Eventually, 8 of the 17 patients died. Among 336 patients first treated with closed-drainage aspiration with Redon catheters, 20 were transferred to our ICU 30 days or more after the initial cardiac surgery; none of them died. Therefore, 316 patients hospitalized within 30 days after initial cardiac surgery were included in the present study. Initial surgical procedures were distributed as follows: 185 (58.5%) coronary artery bypass grafting (CABG) operations, 68 (21.5%) valve operations, 43 (13.6%) CABG plus valve replacements, 6 (1.9%) heart transplantations, and 14 (4.4%) other operations (eg, aortic dissection and pericardial surgery). The mean ± SD incubation time was 13 ± 5.2 days (median, 13 days). The microorganisms responsible for mediastinitis are listed in Table 1.
|
In addition, the patients were reassessed on day 3 after ICU admission. This time point was chosen because it takes at least 48 hours to get valuable bacteriologic data and because mechanical ventilation (MV) duration of greater than 48 hours is considered prolonged respiratory support and associated with high rates of adverse events after cardiac surgery.20 Four variables were therefore recorded on day 3: microorganism responsible for mediastinal infection classified as S aureus or other microorganism; presence of methicillin-resistant S aureus; persistently positive bacteremia, defined as 2 or more positive blood cultures drawn more than 12 hours apart with the microorganism responsible for infection; and requirement for MV. In the case of polymicrobial infection, the microorganism found in all mediastinal samples and blood cultures was considered to be the main causative microorganism. When a Staphylococcus organism was present, it was considered the main causative microorganism.
Outcome measures included the following: in-ICU mortality; duration of MV; duration of ICU stay; time required to sterilize mediastinal fluid, calculated as the interval between the day of debridement and the first day at which the collected mediastinal fluid was sterile; duration of Redon drainage; and failure of treatment to eradicate local infection, defined as the need for redebridement. Life-threatening complications, defined as definite secondary endocarditis of an operated valve and massive mediastinal hemorrhage occurring after initial mediastinal debridement, were also noted. Mediastinitis-related death was defined as death occurring as a consequence of massive mediastinal hemorrhage, associated prosthetic valve endocarditis, or uncontrolled sepsis caused by mediastinitis and leading to multiorgan failure.
Statistical analyses
Univariate analysis was performed to assess the effects on ICU survival of clinical characteristics known within 24 hours of ICU admission or available 3 days after ICU admission. Continuous variables, expressed as means ± SD, were compared with the Student t test or the Mann-Whitney U test as appropriate. Categorical variables, expressed as percentages, were compared with the 
2 test or the Fischer exact test, as appropriate.
Second, 3 separate multivariate logistic regression analyses of risk factors for ICU mortality were performed: for the entire population of patients, analyzing variables known within 24 hours or 3 days after ICU admission, and on day 3 for the subgroup of patients still mechanically ventilated because ICU mortality was very low when patients were weaned off the ventilator within 3 days after debridement. Variables with P values of .05 or less according to the univariate analysis were included in the multivariate models. Thereafter, the analyses were repeated, including all candidate risk factors in the multivariate models. Variables were considered to be independent risk factors for ICU death when the final multivariate P value was less than .05. The discriminating ability of each model was assessed by using the area under the receiver operating characteristic curve. Calibration of each model was assessed with the Hosmer-Lemeshow goodness-of-fit statistic.
Additionally, these models were tested when the population was randomly divided into a derivation cohort, including 65% of the patients, and a validation cohort, including the rest of the patients. Finally, stability of the obtained models was evaluated by using the bootstrap procedure, which generated a random series of 1000 samples, and thereafter, mean parameter estimates as discriminant values were calculated.
All analyses were conducted with the SAS 6.12 software (SAS Institute).
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
|
|
On day 3 after ICU admission, all 4 newly available variables were significantly associated with a poor prognosis (Table 2). Concerning MV and prognosis, 5 (2.4%) of 206 patients weaned off MV before day 3 died versus 57 (52.8%) of 108 patients requiring MV for 3 or more days after debridement.
Multivariate analyses
When logistic regression analysis was applied to the factors available within 24 hours of admission, the 5 independently associated with in-ICU mortality (Table 4), with the exception of the type of operation, were linked to the severity of the underlying disease and the severity of the acute illness. The multivariate analysis, including variables available as of day 3, retained 5 factors as being independently and significantly associated with mortality: 3 known within the first 24 hours in the ICU and identified in the prior analysis (age >70 years, operation other than CABG alone, and McCabe and Jackson class 2/3 [rapidly or ultimately fatal]) plus persistently positive bacteremia and MV for 3 or more days (Table 4).
|
|
The discriminant values, as determined by the areas under the receiver operating characteristic curves, were 0.91 (95% confidence interval [CI], 0.85-0.96) for prediction on the basis of variables available within 24 hours of ICU admission, 0.95 (95% CI, 0.90-0.99) for those available on day 3, and 0.80 (95% CI, 0.71-0.88) for the subpopulation of patients still receiving MV for 3 or more days.
The same independent variables were identified when statistical analyses were performed with the data of the derivation cohort. In addition, the odds ratios and 95% CIs were very close to those obtained for the entire population (data not shown). Discriminant values for the 3 previously described models applied to the validation cohort were 0.90 (95% CI, 0.80-0.99), 0.92 (95% CI, 0.83-1.00), and 0.77 (95% CI, 0.61-0.93), respectively.
Results obtained with the bootstrap procedure were very close to those reported above, confirming the validity of factors identified by using the logistic regression analyses (data not shown).
Finally, risk factors for mediastinitis-related death (n = 33 patients) retained by means of multivariate analyses were similar to those associated with ICU death (except for age and persistently positive bacteremia for the subgroup of patients still mechanically ventilated on day 3; data not shown).
It should be noted that the same independent risk factors were retained by using multivariate analyses when all candidate risk factors (ie, age, sex, McCabe class, CABG, incubation time, APACHE II, number of organ failures, persistently positive bacteremia, and MV 
day 3) were included in the models (data not shown).
Clinical evolution in ICU survivors and nonsurvivors
MV duration (mean ± SD, median, interquartile range) was shorter for survivors than for nonsurvivors: 5.8 ± 17 days, 1 day, and 0 to 2 days versus 25.6 ± 15 days, 1 day, and 6 to 38 days (P < .0001). The length of ICU stay (mean ± SD, median, interquartile range) was 24.9 ± 18.8 days, 21 days, and 17 to 26 days versus 31.3 ± 29.7 days, 19 days, and 9 to 44 days, respectively (P = .03). The time to obtain mediastinal sterilization and duration of Redon catheter drainage did not differ significantly between survivors and nonsurvivors (mean ± SD): 8.7 ± 6.8 days versus 7.8 ± 5.1 days and 19.0 ± 5.5 days versus 18.9 ± 5.6 days, respectively. Redebridement for local failure was required by 34 (10.6%) patients: 21 (8.3%) patients among survivors and 13 (20.3%) among nonsurvivors (P = .006). For 22 patients, new Redon catheters were placed in the mediastinum, and the thorax was closed. The thorax was left open in 12 patients, and thereafter, the mediastinum was rapidly covered with the omental flap in 6 patients, was closed by means of rigid fixation and soft tissue reconstruction with pectoralis muscle advancement flaps in 1 patient, and was surgically closed at a later time in another patient. In the last 4 patients, no surgical closure was attempted. Two (0.8%) survivors and 3 (4.7%) nonsurvivors experienced life-threatening mediastinal hemorrhage (P = .06). Endocarditis was associated with mediastinitis in 21 of the 111 patients who had undergone valve replacement: 12 (18.8%) in the nonsurvivor group and 9 (3.6%) in the survivor group.
In-ICU mortality according to duration of MV, underlying disease, cardiac surgery, and bacteremia
In contrast to the low mortality rate observed for patients weaned from the ventilator before day 3, among patients requiring MV for 3 days or longer after surgical debridement, the mortality rate was 20-fold higher. However, big differences in mortality rates existed in this subgroup. Thus the patients without fatal underlying disease who had undergone CABG alone and who did not experience persistently positive bacteremia had a mortality rate of 21.4%. On the other hand, among the patients who had undergone surgical intervention other than CABG alone and who experienced sustained bacteremia, this percentage reached 89% for patients also experiencing an ultimately or rapidly fatal underlying disease (McCabe and Jackson class 2/3).
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Mediastinal infection is still a dreadful complication with a high mortality rate, ranging between 5% and 47%.6,11-13,21-24 Mortality rates reported in large series of patients were around 20%.13 Other authors described lower mortality rates.3,4,9,15,25 However, these series included patients with sterile sternal dehiscence4,25 and patients without signs of severity. For example, in one of these recent reports, 30.6% of the patients did not require ICU transfer.12 It should also be noted that other authors included patients with very long incubation times and with local treatment problems but without systemic complications.
Because our unique referral pattern was mediastinitis serious enough to require ICU treatment, comparison with other studies dealing with prognostic factors of patients with poststernotomy mediastinitis is not easy.3,11,12 Nevertheless, some similar factors associated with bad outcome emerged from those series. Septicemia was associated with mortality of patients treated with muscle-flap coverage.3 Older age was also associated with treatment failure in the study of Kirsch and associates.12 Conversely, S aureus methicillin resistance was not associated with fatal outcome in our multivariate analysis. The absence of a statistical link between methicillin resistance and mortality in various clinical situations has been reported by others.26-28
Nonetheless, our study design provides several advantages compared with previous observational investigations. First, our cohort was large. Second, these were consecutive patients treated in a single center with the same management protocol concerning both the first-line surgical approach and antibiotic regimen. Moreover, the mortality rate was stable from one year to another. Third, clinical and biologic parameters were routinely recorded prospectively for these patients. Fourth, our study used a time-honored comorbidity index to analyze the effect of underlying disease on the risk of death,17 a validated acute disease severity score,18 and clinical and biologic characteristics explicitly defined and easily identifiable by the physician at the bedside. Finally, the same independent factors were identified when we randomly selected 65% of the population to constitute a derivation data set, and the concordance indexes were good when we tested the models in a distinct validation cohort or when we used a bootstrapping method.
The present study also confirms that a first-line approach with closed-drainage aspiration with Redon catheters achieves excellent results, with low morbidity and low mortality rates in the absence of severe systemic factors. In our current practice, 99% of patients with mediastinitis within 30 days after cardiac surgery are treated with the Redon catheter technique, which is not applicable only in the case of massive sternal necrosis, skin necrosis, or both. In the case of patent first-line treatment failure, surgical redebridement is performed, and new Redon catheters are repositioned in the mediastinum. When tissue necrosis does not permit wound closure, open wound packing is applied temporarily, and one of the plastic surgery techniques is used soon afterward to close the mediastinum.
However, our study has some limitations. First, we cannot precisely determine the number of cases of cardiac surgery and the frequency of deep sternal wound infections during this 10-year period because patients were addressed from 7 different centers. Second, this was a retrospective study, even though data were recorded prospectively. Third, we did not classify mediastinal infections according to the system proposed by El Oakley and Wright7 because many of our patients were treated in our ICU before its publication. Fourth, our study design differs markedly from those of other studies dealing with prognostic factors. In particular, our very simple database did not include factors associated with the occurrence of mediastinitis, such as preoperative (eg, diabetes mellitus) and perioperative (eg, use of both internal thoracic arteries, use of an intra-aortic balloon pump, perioperative myocardial infarction, and surgical duration) variables. We cannot exclude that these variables might have been associated with death in the ICU if tested in our multivariate models. However, it should be noted that none of these factors were associated with fatal outcome in previous studies and that the most relevant factors of poststernotomy mediastinitis outcomes might be those evaluated at the onset of the infection. Finally, the absence of systematic follow-up with respect to further infection or mortality is another limitation of the present study.
In conclusion, this study provides a set of simple variables strongly associated with ICU death for patients with acute poststernotomy mediastinitis requiring initial ICU care and managed by means of debridement and closed-drainage aspiration with Redon catheters. These variables are linked to patients' underlying disease, severity of the infection at ICU admission, and type of initial cardiac operation. Reassessment of patient characteristics on day 3 permits an update of the probability of ICU death. Importantly, the ICU mortality rate was very high for patients with rapidly or ultimately fatal underlying disease or diseases who had undergone cardiac surgery other than CABG alone, who experienced persistently positive bacteremia, and who were still mechanically ventilated on day 3.
Conversely, patients weaned off the ventilator within 3 days after debridement had a low mortality rate (around 2%), regardless of the association with other ICU mortality risk factors.
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. E. Molina, E. C. Nelson, and R. R.A. Smith Treatment of postoperative sternal dehiscence with mediastinitis: Twenty-four year use of a single method J. Thorac. Cardiovasc. Surg., October 1, 2006; 132(4): 782 - 787. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Karra, L. McDermott, S. Connelly, P. Smith, D. J. Sexton, and K. S. Kaye Risk factors for 1-year mortality after postoperative mediastinitis. J. Thorac. Cardiovasc. Surg., September 1, 2006; 132(3): 537 - 543. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
