| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2007;133:397-403
© 2007 The American Association for Thoracic Surgery
Surgery for Acquired Cardiovascular Disease |
a Department of Internal Medicine E, Rabin Medical Center, Beilinson Hospital and Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel
b Infection Control Unit, Rabin Medical Center, Beilinson Hospital, Tel-Aviv, Israel
c Department of Anesthesia, Rabin Medical Center, Beilinson Hospital, Tel-Aviv, Israel.
Received for publication July 30, 2006; revisions received September 30, 2006; accepted for publication October 9, 2006. * Address for reprints: Mical Paul, MD, Internal Medicine E and Infectious Diseases Unit, Rabin Medical Center, Beilinson Hospital, Petah Tikva, 49100, Israel. (Email: pil1pel{at}zahav.net.il).
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
METHODS: We included 809 consecutive patients undergoing coronary artery bypass surgery. Data were collected prospectively. Infections were defined as deep sternal wound infection or mediastinitis by using established criteria and evaluated 60 days after surgical intervention. All-cause mortality was assessed at 30 days and 6 months. We assessed the ability of the National Nosocomial Infections Surveillance risk index, the EuroSCORE, and the Society of Thoracic Surgeons risk score to predict infection and mortality. Discrimination was assessed using the area under the receiver operating curve.
RESULTS: The rate of surgical site infection was 3.6% (29/809 patients). The National Nosocomial Infections Surveillance risk index showed moderate discrimination for infection (area under the receiver operating curve of 0.64) and poor ability to stratify patients into infection risk groups. The EuroSCORE predicted infection and 30-day and 6-month mortality with good discrimination (area under the receiver operating curve of 0.72, 0.78, and 0.77, respectively). Ranking patients by the EuroSCORE and dividing the cohort into 3 roughly equal risk groups yielded an ascending risk for infection of 0.7%, 3.0%, and 7.2%. The preoperative and intraoperative Society of Thoracic Surgeons risk scores showed good discrimination for surgical site infection (area under the receiver operating curve of 0.72 and 0.76, respectively) and excellent discrimination for early and late mortality (area under the receiver operating curve of >0.80). Risk grouping based on the Society of Thoracic Surgeons score yielded an ascending risk for infection of 0.7%, 3.6%, and 6.4%.
CONCLUSIONS: The EuroSCORE and the Society of Thoracic Surgeons risk score can be used for joint risk stratification for surgical site infection and mortality. Both scores performed better than the National Nosocomial Infections Surveillance risk index.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Surgical site infection (SSI) is an infrequent but devastating complication after cardiac surgery. Deep sternal wound infections and mediastinitis are associated with increased mortality, repeated surgical procedures, prolonged hospital stay, and increased costs.1-3
Several risk scores for SSI have been developed. The National Nosocomial Infections Surveillance (NNIS) risk index is well accepted for general surgery.4 However, it was not adapted specifically for coronary artery bypass graft (CABG) operations. This score is based on the American Society of Anesthesiologists (ASA) score, wound class, and operation duration. It does not take into account particular risk factors among patients undergoing CABG and permits little variability because CABG operations are always clean and patients’ ASA scores are greater than 2 per definition. Fowler and colleagues2 developed a simple bedside risk score using the Society of Thoracic Surgeons (STS) National Cardiac Database, including more than 300,000 patients undergoing CABG surgery. The STS score predicts the risk for major infections, including SSIs or septicemia, after CABG. A preoperative score, including variables available before surgical intervention and an intraoperative score, including preoperative and intraoperative variables, performed well in the validation sample of the STS database. Finally, the EuroSCORE is a risk score for 30-day mortality after cardiac surgery that has been well validated worldwide.5-8 It has been proposed that the EuroSCORE can predict other adverse outcomes, including infections, after surgical intervention.9 The similarity between risk factors for SSI and mortality might permit joint risk stratification.
We validated these 3 risk scores in a sample of 809 patients undergoing CABG during 1 year in a single center in Israel. Our objective was to compare the performance of these risk scores and to validate the STS scores of patients outside the STS database.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Preoperative data collected included patient demographics and detailed data regarding hospital stay and invasive procedures before surgical intervention, preoperative risk factors using EuroSCORE definitions,10 hemoglobin A1c (HBA1c) when available, previous antibiotic use, and preoperative patient preparation. Intraoperative data included the type of operation and procedures performed; operation duration; bypass use and duration; hourly arterial glucose level, continuous temperature for on-pump surgery, and mean intraoperative temperature for off-pump surgery; timing, dosing, and type of antibiotic prophylaxis; administration of blood products; and mean fraction of inspired oxygen and PaO2. Postoperative data included temperature and oxygenation on arrival to the intensive care unit; duration of hypothermia, mechanical ventilation, and intensive care unit stay; complications during the first 3 postoperative days; and 30-day and 6-month all-cause mortality. The EuroSCORE and NNIS risk index values were collected prospectively. SSIs were diagnosed and classified according to Centers for Disease Control and Prevention criteria.11
Data were obtained from patients’ hospital charts and electronic databases. The electronic databases available for routine work at our hospital present data from all hospitals, primary care clinics, and laboratories belonging to the health maintenance organization. Thus data on background conditions, previous hospitalizations, HBA1c, and postdischarge follow-up with hospital or clinic notes, microbiologic studies, and surgical procedures were available. Data for this study were collected as part of the routine surveillance conducted by the infection control unit at our hospital. The study was conducted in accordance with the requirements of the local ethics committee.
The primary outcome assessed was SSI 60 days after surgical intervention, including deep sternal wound infection or mediastinitis. Because most continuous variables did not have a normal distribution, the Mann–Whitney U test (nonparametric) was used for comparisons. For comparisons of dichotomous data, we used the 
2 test. A multiple logistic regression model was constructed to assess independent risk factors for SSI. Variables associated with SSI at a P value of less than .1 on univariate analysis were entered by using stepwise forward logistic regression. We did not include composite risk scores in the model because we assessed their individual components.
We assessed the ability of each of the risk scores to predict occurrence of the primary outcome and mortality. We used the area under the receiver operating characteristic (ROC) curve with 95% confidence intervals to assess the discriminatory ability of each of the risk scores. Calibration was assessed by using the Hosmer–Lemeshow goodness-of-fit statistic. For each score, we grouped patients into low-, intermediate-, and high-risk patients to assess the ability of the score to identify a clinically meaningful percentage of the study population at high risk for infection. All analyses were conducted with SPSS 13.0 statistical software (SPSS, Inc).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
Nearly all patients were classified as risk index 1 or 2, as expected. The respective SSI rates for these risk groups were 2.0% and 5.8% (Table 3). The discrimination of the score for SSI and mortality was moderate (Table 4).
|
|
Ranking patients by the EuroSCORE and dividing the cohort into 3 roughly equal groups yielded a high-risk group consisting of 33% of patients with an observed rate of SSI of 7.2% (Table 3). The discriminative power of the EuroSCORE with regard to prediction of SSI was good (areas under the ROC curve of 0.72 for the additive model and 0.73 for the logistic score). Originally modeled to predict mortality, the EuroSCORE performed well in our cohort, with an area under the ROC curve of 0.78 for 30-day mortality and 0.77 for 6-month mortality (Table 4).
STS Risk Score
Several variables in our database were defined differently from the original variable definitions used in the STS database. We defined chronic lung disease and diabetes as conditions necessitating chronic treatment; myocardial infarction was recorded within 3 months before the operation. We did not collect data on insertion of an intra-aortic balloon pump.
We ranked our cohort by the STS risk score and divided it into 3 risk groups (Table 3). The number of patients within each risk group was similar to the distribution observed in the original STS cohort (low risk, 35% vs 31% of patients; intermediate risk, 30% vs 30%; high risk, 35% vs 38% for our cohort versus the STS database; data supplied by the authors). The observed rate of infections in the high-risk group was 6.4% (mean predicted value for the group, 5.3%). Both the preoperative and intraoperative (omitting intra-aortic balloon pump) scores discriminated well between patients with and without SSIs (Table 4). The area under the ROC curve was 0.72 for the preoperative score and 0.76 for the intraoperative score, indicating good discrimination. The respective values in the original STS validation cohort were 0.70 and 0.71. Calibration was assessed against the predicted probability of infection, as reported in the original model derivation. The Hosmer–Lemeshow 2 value was 9.86 (P > .25) for the preoperative and 6.55 (P > .5) for the intraoperative scores, indicating adequate calibration. The score we used was designed to predict major infections but predicted mortality in our cohort with excellent discrimination (areas under the ROC curves of 
0.80 for both the preoperative and intraoperative models, Table 4).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The STS risk score was developed by using a national database assessing mainly cardiovascular process and outcome measures.2 There were several differences between our methods and the STS database. Risk factors were collected by using the EuroSCORE definitions.10 The primary outcome of interest in our study included deep sternal wound infection and mediastinitis, whereas Fowler and colleagues2 included superficial site infection and septicemia in their primary end point of major infection. Our surveillance included postdischarge follow-up for SSI, and indeed, more than 30% of infections were identified after discharge. Thus the absolute number of infections in our cohorts is not comparable. However, despite these differences, the model performed very well in our cohort. In the highest risk group, consisting of one third of the cohort, the rate of SSI was 6.4% compared with 0.7% in the low-risk group. An important risk factor for SSI emerging in our cohort was intraoperative glucose control (Tables 1 and 2), which was not assessed in the STS database. It should be investigated whether inclusion of perioperative glucose values might improve the performance of the model. Interestingly, the score predicted mortality with excellent discrimination.
The EuroSCORE is a well-validated score for prediction of mortality after cardiac surgery.6-8 All variables included in the EuroSCORE were prospectively collected by using their original definitions. The score predicted both short-term (30-day) and long-term (6-month) mortality rates with very good discrimination in our cohort (area under the ROC curve of 0.78 and 0.77, respectively), somewhat better than described in an earlier cohort at our center.5 Performance of the EuroSCORE with regard to prediction of SSI has been assessed in a single study previously.9 In our dataset the discriminative power of the EuroSCORE for SSI was better (area under the ROC curve of 0.72 for SSI in our cohort vs 0.57 for deep sternal wound infection9).
The NNIS risk index is widely used for benchmarking within and between centers. We show that this score is less well adapted for SSI after cardiac surgery than the scores specifically designed for these operations.
There are limitations and strengths to our study. We used a single-center, small cohort of patients. However, the small cohort permitted close and detailed surveillance. The electronic records used in the hospital permitted preadmission and postdischarge data collection. Thus the presence or absence of an SSI was confirmed at 60 days postoperatively in all but 2 patients, in whom no SSIs were recorded in our hospital, but postdischarge follow-up was not available. All-cause mortality was ascertained in all patients. We included only deep and organ space sternum infections in our outcome because the diagnosis of these infections is unequivocal, and these infections are associated with significant morbidity and mortality. The small number of outcomes in our study precluded the assessment of specific pathogens. Optimally, a risk score should be able to predict infections that can be prevented, such as those caused by Staphylococcus aureus. Finally, we did not assess the effect of risk stratification prospectively. Ultimately, risk stratification should be used to improve patient outcomes.
Interventions that have been shown to reduce SSIs include antibiotic prophylaxis,13 surveillance for and decontamination of Staphylococcus aureus,14,15 tight perioperative glucose control,16-20 and optimal skin preparation.21 In our study vancomycin was used mainly in combined operations, and the recommended dose of cefazolin for on-pump operations was 1 g. Both combined and on-pump operations were associated with an increased rate of SSI. Previous studies have suggested that ß-lactams might be superior to glycopeptides for the prevention of SSIs,22 even in locations with a high prevalence of methicillin-resistant Staphylococcus aureus,23 and that the use of 1 g of cefazolin dosing might be insufficient for operations with cardiopulmonary bypass.24 Other interventions, such as management of chronic obstructive pulmonary disease before admission, should be explored. The benefit of surgical intervention versus percutaneous coronary intervention, when possible, can be reconsidered in light of preoperative assessment,25-27 as well as the necessity of concomitant surgical intervention. Although general measures for prevention of infections acquired during surgical intervention, such as antibiotic prophylaxis, apply to all patients, special interventions can be considered for patients at high risk for SSIs. The 2 scores we validated can be used to define patients at high risk, candidates for special interventions to reduce SSIs. The threshold of predicted SSI used to implement an intervention should be assessed. Intraoperative and postoperative risk stratification will allow for close monitoring and early diagnosis of infections among high-risk patients.
In summary, we show that both the STS risk score and the EuroSCORE can be used to identify SSI risk among patients undergoing CABG. The similarity of risk factors allows for joint risk stratification for SSI and mortality. Future studies should assess the effect of implementing risk assessment, with or without special interventions targeting high-risk patients.
| Earn CME credits at http://cme.ctsnetjournals.org
|
|
Acknowledgments |
|---|
for supplying additional data for their study. |
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. C. Berg, K. E. Kjorstad, P. E. Akselsen, B. E. Seim, H. L. Lower, M. N. Stenvik, N. K. Sorknes, and H.-M. Eriksen National surveillance of surgical site infections after coronary artery bypass grafting in Norway: incidence and risk factors Eur J Cardiothorac Surg, December 1, 2011; 40(6): 1291 - 1297. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Mannien, J. C. Wille, J. J. Kloek, and B. H. B. van Benthem Surveillance and epidemiology of surgical site infections after cardiothoracic surgery in The Netherlands, 2002-2007 J. Thorac. Cardiovasc. Surg., April 1, 2011; 141(4): 899 - 904. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Ariyaratnam, M. Bland, and M. Loubani Risk factors and mortality associated with deep sternal wound infections following coronary bypass surgery with or without concomitant procedures in a UK population: a basis for a new risk model? Interact CardioVasc Thorac Surg, November 1, 2010; 11(5): 543 - 546. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Wilson Further reading OSH Cardiac Anaesthesia, January 1, 2010; 1(1): med-9780199209101-div1-90 - med-9780199209101-div1-90. [Full Text]
|
|
|
|
 |
|
 I. S. Modrau, T. Ejlertsen, and B. S. Rasmussen Emerging Role of Candida in Deep Sternal Wound Infection Ann. Thorac. Surg., December 1, 2009; 88(6): 1905 - 1909. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Z. Atkins, M. K. Wooten, J. Kistler, K. Hurley, G. C. Hughes, and W. G. Wolfe Does Negative Pressure Wound Therapy Have a Role in Preventing Poststernotomy Wound Complications? Surgical Innovation, June 1, 2009; 16(2): 140 - 146. [Abstract] [PDF]
|
|
|
|
|
|
M. Hartrumpf, T. Claus, M. Erb, and J. M. Albes Surgeon performance index: tool for assessment of individual surgical quality in total quality management Eur J Cardiothorac Surg, May 1, 2009; 35(5): 751 - 758. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. A. M. Nashef Predicting morbidity after coronary surgery Eur J Cardiothorac Surg, May 1, 2009; 35(5): 760 - 768. [Full Text] [PDF]
|
|
|
|
|
|
J. Nakano, H. Okabayashi, M. Hanyu, Y. Soga, T. Nomoto, Y. Arai, T. Matsuo, M. Kai, and M. Kawatou Risk factors for wound infection after off-pump coronary artery bypass grafting: Should bilateral internal thoracic arteries be harvested in patients with diabetes? J. Thorac. Cardiovasc. Surg., March 1, 2008; 135(3): 540 - 545. [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 |
