Ventilator-associated pneumonia in patients undergoing major heart surgery: an incidence study in Europe

Introduction Patients undergoing major heart surgery (MHS) represent a special subpopulation at risk for nosocomial infections. Postoperative infection is the main non-cardiac complication after MHS and has been clearly related to increased morbidity, use of hospital resources and mortality. Our aim was to determine the incidence, aetiology, risk factors and outcome of ventilator-associated pneumonia (VAP) in patients who have undergone MHS in Europe. Methods Our study was a prospective study of patients undergoing MHS in Europe who developed suspicion of VAP. During a one-month period, participating units submitted a protocol of all patients admitted to their units who had undergone MHS. Results Overall, 25 hospitals in eight different European countries participated in the study. The number of patients intervened for MHS was 986. Fifteen patients were excluded because of protocol violations. One or more nosocomial infections were detected in 43 (4.4%) patients. VAP was the most frequent nosocomial infection (2.1%; 13.9 episodes per 1000 days of mechanical ventilation). The microorganisms responsible for VAP in this study were: Enterobacteriaceae (45%), Pseudomonas aeruginosa (20%), methicillin-resistant Staphylococcus aureus (10%) and a range of other microorganisms. We identified the following significant independent risk factors for VAP: ascending aorta surgery (odds ratio (OR) = 6.22; 95% confidence interval (CI) = 1.69 to 22.89), number of blood units transfused (OR = 1.08 per unit transfused; 95% CI = 1.04 to 1.13) and need for re-intervention (OR = 6.65; 95% CI = 2.10 to 21.01). The median length of stay in the intensive care unit was significantly longer (P < 0.001) in patients with VAP than in patients without VAP (23 days versus 2 days). Death was significantly more frequent (P < 0.001) in patients with VAP (35% versus 2.3%). Conclusions Patients undergoing aortic surgery and those with complicated post-intervention courses, requiring multiple transfusions or re-intervention, constitute a high-risk group probably requiring more active preventive measures.


Results
Overall, 25 hospitals in eight different European countries participated in the study. The number of patients intervened for MHS was 986. Fifteen patients were excluded because of protocol violations. One or more nosocomial infections were detected in 43 (4.4%) patients. VAP was the most frequent nosocomial infection (2.1%; 13.9 episodes per 1000 days of mechanical ventilation). The microorganisms responsible for VAP in this study were: Enterobacteriaceae (45%), Pseudomonas aeruginosa (20%), methicillin-resistant Staphylococcus aureus (10%) and a range of other microorganisms. We identified the following significant independent risk factors for VAP: ascending aorta surgery (odds ratio (OR) = 6.22; 95% confidence interval (CI) = 1.69 to 22.89), number of blood units transfused (OR = 1.08 per unit transfused; 95% CI = 1.04 to 1.13) and need for re-intervention (OR = 6.65; 95% CI = 2.10 to 21.01). The median length of stay in the intensive care unit was significantly longer (P < 0.001) in patients with VAP than in patients without VAP (23 days versus 2 days). Death was significantly more frequent (P < 0.001) in patients with VAP (35% versus 2.3%).
Conclusions Patients undergoing aortic surgery and those with complicated post-intervention courses, requiring multiple transfusions or re-intervention, constitute a high-risk group probably requiring more active preventive measures.

Introduction
Patients undergoing major heart surgery (MHS) represent a special subpopulation at risk for nosocomial infections. Postoperative infection is the main non-cardiac complication after MHS and has been clearly related to increased morbidity, use of hospital resources and mortality [1,2].
Ventilator-associated pneumonia (VAP) is the most common infection in patients admitted to intensive care units (ICUs) [3,4] and is a leading cause of morbidity and mortality [5,6]. The situation of VAP in patients undergoing MHS has been assessed only from the perspective of single institutions with the bias of the case-mix at a particular centre [1,[7][8][9][10]. Our group led the collection of retrospective data of VAP in MHS from several European institutions [11], but prospective data from a large group of European centres were lacking.
Our study aims were to determine the incidence, aetiology, risk factors and outcome of VAP in a large sample of patients who have undergone MHS in Europe.

Materials and methods
Our study is a joint venture between the European Study Group of Nosocomial Infection (ESGNI), the European Society of Clinical Microbiology and Infectious Diseases (ESC-MID) and the European Working Party of Cardiothoracic Intensivists (EWCI). The Ethics Committee of Hospital General Universitario Gregorio Marañón (Madrid, Spain) approved the study and indicated that individual informed consent was not necessary in this study because no intervention was performed and confidentiality was respected.
Our study (ESGNI 09 study) was a prospective (one-month enrolment) analysis of patients undergoing MHS in Europe who developed suspicion of VAP. During a one-month period participating units submitted a protocol of all patients admitted to their units who had undergone MHS. Specific variables on VAP diagnosis and evolution were included.
Units and investigators willing to participate sent data regarding the type of hospital, (public or private, teaching or nonteaching, total population surveyed, number of beds, and the number of hospital admissions for 24 hours or longer during the month of the study) and data regarding the ICU used for postoperative care of MHS patients in each institution (ICU specific for MHS or mixed with other types of patients and number of beds available).
Individual data for patients admitted to European postsurgical ICUs included: hospital admission date, sex, age, prior illnesses, clinical characteristics of the patient and New York Heart Association (NYHA) functional class. Patients' underlying diseases were classified according to the criteria of McCabe and Jackson [12] as rapidly fatal, ultimately fatal and non-fatal; their morbidity scores were based on the Charlson co-morbidity index [13]. The American Society of Anesthesiologists physical status grading system [14] and EuroSCORE [15] were used to value surgical risk.
Data regarding the surgical procedure included type of indication (elective, urgent or emergent), type of surgical procedure, duration (from the skin incision until closure), time on cardiopulmonary bypass, aortic cross-clamp time, surgical antimicrobial prophylaxis, transfusion needs, overall period with chest drainages, number of reinterventions and need for inotropic support, intra-aortic balloon or circulatory assistance. Surgical prophylaxis was performed according to each centre's protocol.
Recorded postsurgical events included ICU admission and discharge date, days spent on mechanical ventilation, preventive methods for VAP, type of nosocomial infection and patient evolution. If the patient had VAP, a specific part of the questionnaire was completed including criteria for diagnosis, Clinical Pulmonary Infection Score (CPIS) [16], microbiological data (microorganisms causing pneumonia) and outcome.

Definitions
VAP was diagnosed upon the presence of new and/or progressive pulmonary infiltrates on chest radiograph plus two or more of the following criteria: fever (≥ 38.5°C) or hypothermia (< 36°C), leucocytosis (≥ 12 × 10 9 /L), purulent tracheobronchial secretions or a reduction of partial pressure of arterial oxygen (PaO 2 )/fraction of inspired oxygen (FiO 2 ) of 15% or higher in the past 48 hours according to the definitions of the Centers for Disease Control and Prevention [17]. Also, as pneumonia cases we included those patients with a CPIS higher than six [16].
Tracheobronchitis was defined as the presence of purulent tracheobronchial secretions plus two or more of the following criteria: fever (≥ 38.5°C) or hypothermia (< 36°C), leucocytosis (≥ 12 × 10 9 /L), or significant bacteriological counts in respiratory secretions in patients without pulmonary infiltrates suggesting pneumonia on chest radiograph [17]. Cases with either VAP or tracheobronchitis had to be microbiologically confirmed.
The ICUs were classified as specific if more than 95% of their beds were addressed to patients undergoing MHS or as mixed if this criterion was not met.

Data analysis
Reports from individual centres were sent to the coordinating centre either by regular mail or via the internet. Individual reports were reviewed by one of the authors before being entered into the database and analysed using SPSS Version 12 (SPSS Inc., Chicago, IL, USA).
We expressed continuous variables as the median and interquartile range (IQR) if their distribution was skewed, and discrete variables as percentages. Measures of significance were assessed by univariate and stratified analysis. Continuous variables were analysed by the Mann-Whitney U test, and categorical variables were analysed with Fisher's exact test or the chi-squared test. All statistical tests were two-tailed. The independent contribution of predictor variables for the development of VAP and mortality after MHS was assessed by stepwise logistic regression analysis, and associations between variables expressed as odds ratios (OR) and respective 95% confidence intervals (CI). As candidate variables we included in the model all those which showed univariate significance less than P < 0.1. The aim of the study was to find pre-operative, operative and immediate postoperative risk factors of VAP. For this reason risk factors were analysed in two models, with and without the inclusion of the number of days of mechanical ventilation. The models were validated by means of the jack-knifing technique [18]. Variables which did not yield the same results in at least 90% of the 20 jack-knifing runs were discarded. No significant first-order interactions were found in the models.

Participating institution characteristics
Overall, 25 hospitals in eight different European countries participated in the study ( Table 1). The participating institutions were either teaching (88%) or non-teaching hospitals (12%) Overall, 477 patients (49.1%) were transfused and the median number of units was three (IQR = two to six). The patients needed inotropic support (59.1%), intra-aortic balloon (6.1%) or circulatory assistance (0.5%) because of different degrees of ventricular dysfunction. The median length of stay in the ICU was two days (IQR = one to three).

Risk factors
We analysed preoperative, operative and immediate postoperative risk factors for the development of VAP. In the univariate analysis preoperative factors associated with VAP were (Table  3): mixed ICU (relative risk [RR] = 2.8), peripheral vascular disease (RR = 3), renal disease (RR = 7.9) and American Society of Anesthesiologists score more than 3 (RR = 3.5). For surgical risk factors, the following were associated with VAP: need for inotropic support (RR = 15), need for intra-aortic balloon (RR = 5.5), ascending aorta surgery (RR = 9.7) and median duration of surgery. For postoperative risk factors, the following were associated with VAP: mean number of blood units transfused, need for re-intervention (RR = 12.3) and days of mechanical ventilation until onset of VAP.
Regarding multivariate analysis, two different models were performed, not including or including the days on mechanical ventilation in the model. As for the logistic regression model

Discussion
Our multicentre European study confirms that VAP was the main cause of postoperative infection in patients undergoing MHS in several European centres. Our results showed incidence data between that reported from institutions with very different case mixed.
Figures of incidence of nosocomial infections in general ICUs vary from 9 to 37%, mostly depending on the type and severity of illness of that population and the definitions used [19,20]. In patients undergoing MHS, figures of postoperative nosocomial infections range from 9 to 45%, depending also on the type of heart surgery performed [1,21].
According to the National Nosocomial Infections Surveillance report of 2004, the median rate of VAP in 47 cardiothoracic surgery ICUs was 6.3 (IQR = 2.9 to 12.6) per 1000 ventilation days [26]; however, the proportion of patients undergoing MHS in that population is not clear. In a previous study, several European MHS units retrospectively estimated their incidence of VAP which occurred in 3.8% of all patients undergoing MHS [11]. The present study provides a prospectively collected incidence rate of 2.1% (incidence density of 13.9 of 1000 ventilation days) which is lower than previous results found in similar patients [1,7,8], although it is comparable with those reported in other studies [9,10].
Microorganisms causing VAP vary considerably according to the characteristics of the patients in the different ICU types, the length of hospital stay and intubation. Common pathogens include P. aeruginosa, S. aureus and Enterobacteriaceae [27]. There is no evidence that the microorganisms causing VAP after MHS are substantially different [1,7,9] to those in other types of patients in ICUs. In a paper from Kollef and colleagues [1], 59 of 605 MHS patients developed VAP. Enterobacteriaceae (15 cases) and P. aeruginosa (9 cases) predominated, as happened in our series. Our series also  showed the potential presence of S. aureus and particularly the risk of methicillin-resistant isolates. The proportion of polymicrobial VAP ranged from 13 to 55% in different studies [28][29][30]. In our series 25% of the VAP episodes had more than one microorganism present.
Various risk factors have been associated with the development of VAP in patients undergoing MHS, including the duration of mechanical ventilation, need for reintubation, transfusion needs, empirical administration of broadspectrum antibiotics, type of surgery, age over 60 years, supine position during the first 24 hours, history of chronic obstructive pulmonary disease, NYHA score of 3 or higher and need for mechanical intravascular support [1,7,8,21,23,24,31]. Some of these factors were confirmed in our study, in particular, transfusion needs and type of surgery. At the same time, our study underscores other risk factors such as the need for re-intervention with haemorrhage or cardiac tamponade in the immediate postoperative period. Our study was oriented to find preoperative, intraoperative and immediate postoperative factors amenable to intervention in the population undergoing MHS. Due to this, we decided not to include the variable 'days of mechanical ventilation' in the model because it completely overshadowed the importance of the other variables we specifically wanted to address.
Because of this, we decided to include the variable 'days of mechanical ventilation' in a separate model. After analysing this new model, transfusion needs lost statistical significance.
Most unfortunately, the majority of the variables that significantly predict VAP are not amenable to intervention. In our opinion the use of anticipative or pre-emptive antimicrobial therapy should be explored as one of the few potential interventions to avoid VAP in the high-risk population. It is known that inadequate empirical therapy is associated with an increase in VAP-related mortality, even if it is corrected in the following hours. Singh and colleagues demonstrated that the administration of three days of ciprofloxacin to patients with suspicion of VAP had a very favourable impact on the cost and length of antimicrobial use, and reduced the rate of superinfections and the emergence of resistance [32]. Also, the use of oral decontamination, along with three days of cefotaxime or ceftriaxone, has been demonstrated to have the potential benefit of antimicrobial pre-emptive therapy in patients at high risk of VAP [33,34]. Other potential preventive measures include continuous aspiration of subglottic secretions [35] or the use of polyurethane cuffed tubes [36].
The overall mortality rate for VAP in patients undergoing MHS may be as high as 16 to 57% [1,7,9], but many critically ill patients with VAP die because of their underlying disease rather than of pneumonia. Crude mortality rate of patients with VAP was found to be 35% in our study and attending physicians attributed 13.8% of excess deaths to VAP. However, it should be stated that because multiple comorbidities in these patients, the attribution of mortality to VAP should always be interpreted with caution.
Some limitations of this investigation should be mentioned. Countries and institutions were not randomly selected among the whole continent and the relative weight of the European countries is not equilibrated. However, this study includes 25 centres from eight European countries and constitutes, to our knowledge, the best data available to date to estimate the dimension of this problem. On the other hand, the number of patients with VAP is relatively low. However, our study population includes almost 1000 cases and during a whole month all patients undergoing operations were systematically included.

Conclusions
These data, representing several European institutions, suggest that VAP is still the main cause of nosocomial infection during the postoperative period following MHS. Due to the scarcity of variables for intervention, anticipative or pre-emptive antimicrobial therapy should be explored as one of the few potential interventions to avoid VAP in the population remaining under mechanical ventilation for more than 48 hours.