Pathogens in Early‐Onset and Late‐Onset Intensive Care Unit–Acquired Pneumonia
Objectives. To compare the type of pathogens isolated from patients with early‐onset intensive care unit (ICU)‐acquired pneumonia with those isolated from patients with late‐onset ICU‐acquired pneumonia and to study risk factors for the isolation of pathogens that are potentially resistant to multiple drugs.
Design. Prospective cohort study.
Setting. Patients admitted to the ICU of a 677‐bed, university‐affiliated teaching hospital in Belgium during 1997‐2002.
Methods. ICU‐acquired pneumonia was defined as a case of pneumonia that occurred 2 days or more after admission to the ICU in combination with a positive results of radiologic analysis, clinical signs and symptoms, and a positive culture result. All cases of pneumonia were categorized as either early onset (within 7 days after admission) and late onset (7 days or more after admission), with or without previous antibiotic treatment, and the corresponding pathogens were analyzed. Risk factors for the isolation of pathogens potentially resistant to multiple drugs (ie, Pseudomonas aeruginosa, Serratia marcescens, Enterobacter species, Morganella morganii, methicillin‐resistant Staphylococcus aureus, Citrobacter species, Acinetobacter species, Burkholderia species, extended‐spectrum β‐lactamase–producing pathogens, and Stenotrophomonas maltophilia) were analyzed using logistic regression analysis.
Results. A total of 4,200 patients stayed at the ICU for 2 or more days, 298 of whom developed ICU‐acquired pneumonia, for an overall incidence of 13 cases (95% confidence interval [CI], 11‐14 cases) per 1,000 ICU‐days. Pathogens potentially resistant to multiple drugs were isolated from 52% of patients with early‐onset pneumonia. Risk factors for the isolation of these pathogens were greater age and previous receipt of antibiotic prophylaxis (adjusted odds ratio [aOR], 4.6 [95% CI, 1.6‐13.0]) or antibiotic therapy (aOR, 8.2 [95% CI, 2.8‐23.8]). The length of ICU admission and hospital stay were weaker risk factors for the isolation of these pathogens.
Conclusions. Pathogens potentially resistant to multiple drugs were isolated in 52% of cases of early‐onset ICU‐acquired pneumonia. Previous antibiotic use (both prophylactic and therapeutic) is the main risk factor for the isolation of these pathogens.
Received November 28, 2005; accepted April 3, 2006; electronically published February 28, 2007.
Hospital‐acquired pneumonia (HAP) is a major complication in patients hospitalized in intensive care units and has a negative influence on both patient morbidity and mortality.1 An international prevalence study involving more than 1,000 intensive care units (ICUs) indicated that pneumonia is responsible for almost half of the infections in critically ill patients in Europe.2 Risk factors for HAP are receipt of mechanical ventilation, admission for trauma, and presence of comorbidities.3 The incidence of HAP in Belgium is approximately 9 cases per 1,000 ICU‐days.4 According to international studies, the incidence of HAP varies between 9 and 20 cases per 1,000 ICU‐days, depending on the definitions used, and is much higher among patients receiving mechanical ventilation.5
HAP in Onze Lieve Vrouw Ziekenhuis (Aalst, Belgium) is treated according to the guidelines of the Belgian Infectious Disease Advisory Board (IDAB), which are based on the article by Trouillet et al.6,7 In these guidelines, HAP is classified into the following 4 categories: early‐onset HAP (pneumonia occurring within 7 days after admission or initiation of mechanical ventilation) without previous receipt of antibiotics, early‐onset HAP with previous receipt of antibiotics, late‐onset HAP (ie, pneumonia occurring 7 or more days after admission or initiation of mechanical ventilation) without previous receipt of antibiotics, and late‐onset HAP with previous receipt of antibiotics. Early‐onset HAP in patients without previous receipt of antibiotics is treated with amoxicillin–clavulanic acid or a second‐generation cephalosporin, because it is believed to be comparable to community‐acquired pneumonia in terms of causal microorganisms. Because the potential for antibiotic resistance is higher in pathogens from patients with early‐onset HAP with a previous history of antibiotic use and patients with late‐onset HAP without previous history of antibiotic use, monotherapy with a fourth‐generation cephalosporin is the preferred treatment for these patients. Microorganisms most likely to be recovered from patients with late‐onset HAP who were previously treated with antibiotics are Pseudomonas aeruginosa and drug‐resistant Enterobacteriaceae.7 In these patients, empirical treatment with a fourth‐generation cephalosporin in combination with an aminoglycoside is started.
We conducted a retrospective study to investigate whether the microorganisms associated with the 4 categories of HAP correspond to the assumptions specified in the Belgian IDAB guidelines.6 We focused our research on patients admitted to the ICU, because of the homogeneity of this population and because we had access to a database containing data on all patients admitted to the ICU.
Methods
Setting
In 1992, the study hospital, Onze Lieve Vrouw Ziekenhuis, began participating in the Belgian National Program for the Surveillance of Hospital Infections project,4 and since 2000 this program has participated in the European program Hospitals in Europe Link for Infection Control Through Surveillance (HELICS).8 Information on all patients admitted to the ICU is collected on a daily basis and entered into a database. Onze Lieve Vrouw Ziekenhuis is a 677‐bed, university‐affiliated teaching hospital with a 24‐bed ICU. Approximately 2,000 patients are admitted to the ICU each year, of whom 60% undergo cardiac surgery. All patients admitted to the ICU are monitored for the presence of nosocomial pneumonia or septicemia during their ICU stay.
The database contains information on patient demographic characteristics, admission and discharge dates, and simplified acute physiology score (SAPS) II at the time of admission. In addition, information on comorbidities, reason for ICU admission, and previous use of antibiotics is available. On a day‐by‐day basis, information on the use of mechanical ventilation, central venous catheters, antibiotics, and nasointestinal or orointestinal tubes and receipt of nasointestinal or orointestinal or parenteral nutrition is gathered. For patients with nosocomial pneumonia or septicemia, dates of infection onset and clinical, radiographic, and microbiological data are recorded.
Study Population
The study population comprised all patients admitted to the ICU between January 1, 1997, and December 31, 2002. Patients were observed from the time of ICU admission until the occurrence of ICU‐acquired pneumonia (IAP), death, or discharge from ICU, whichever occurred first.
Case Identification and Validation
IAP was defined as a case of pneumonia that began after an ICU stay of more than 48 hours (as specified in the HELICS protocol). The primary criteria for diagnosis were abnormal chest radiograph findings in combination with fever (temperature greater than 38.3°C) and/or leukocytosis (leukocyte count greater than 10,000 leukocytes/mL). Additional criteria were clinical signs of pneumonia and/or either new onset of purulent sputum or change in sputum characteristics. Only patients with a positive result of culture were included in the study. Specimens for culture were obtained by tracheal or bronchial aspiration, and a semiquantitative culture technique was used. For patients not undergoing ventilation, samples for culture were obtained via bronchoscopy with bronchoalveolar lavage.
The following microorganisms were classified as pathogens potentially resistant to multiple drugs (hereafter, “the study pathogens”): Pseudomonas aeruginosa, Serratia marcescens, Enterobacter species, Morganella morganii, methicillin‐resistant Staphylococcus aureus (MRSA), Citrobacter species, Acinetobacter species, Burkholderia species, extended‐spectrum β‐lactamase (ESBL)–producing pathogens, and Stenotrophomonas maltophilia. Results of susceptibility tests were only used to distinguish between MRSA and methicillin‐susceptible S. aureus (MSSA) and between ESBL‐producing pathogens and non–ESBL‐producing pathogens.
Cases of IAP were stratified into 2 categories: early‐onset pneumonia, defined as pneumonia that occurred less than 7 days after ICU admission, without (group 1) or with (group 2) receipt of antibiotic therapy in the preceding 30 days; and late‐onset pneumonia, defined as pneumonia that occurred 7 or more days after ICU admission, without (group 3) or with (group 4) receipt of antibiotic therapy in the preceding 30 days. Receipt of antibiotic prophylaxis during the preceding 30 days was not taken into account in the stratification into 4 groups, because it was not known to be a risk factor for the isolation of the study pathogens. We used an ICU stay of less than 7 days as the cutoff value to distinguish between early‐onset and late‐onset pneumonia, because the median time to onset of pneumonia during the period under study was 6 days.
Covariates
For all pathogens recovered from patients with IAP, microbiological characteristics and, eventually, antibiotic resistance patterns were determined. In addition, data were available on SAPS II at admission, patient's sex and ward or institution of origin, medical history of respiratory disease, antibiotic use prior to ICU stay, previous hospital admission in the 30 days prior to ICU admission, reason for ICU admission (medical procedure, scheduled surgical procedure, or unscheduled surgical procedure), infection at the time of ICU admission, and type of prior surgery.
Antibiotic use prior to ICU admission was verified by reviewing prescription records from the hospital pharmacy database. Antibiotic prophylaxis involved use of a first‐generation cephalosporin (cefazolin) within 24 hours before surgery. Antibiotic therapy involved use of β‐lactam agents other than a third‐ or fourth‐generation cephalosporin or imipenem, a third‐ or fourth‐generation cephalosporin, imipenem, quinolones, aminoglycosides, and other agents.
Statistical Analysis
The incidence rate of IAP was calculated by dividing the number of cases of IAP by the accumulated number of hospital‐days. The Poisson distribution was used to calculate 95% confidence intervals (CIs). Distribution of the variables was determined by means of the Kolmogorov‐Smirnov test. Differences between patients with and patients without nosocomial pneumonia were studied by means of the Pearson χ2 test (for categorical variables) and the Student t test or the Mann‐Whitney U test (for continuous variables). Descriptive statistical methods were used to study the distribution of the pathogens over the different groups. A multivariate logistic regression analysis was used to study risk factors for the isolation of the study pathogens. In the final model, we included all factors that were associated with the outcome in the univariate analysis. A P value of less than .05 was considered to be statistically significant. All statistical analysis was conducted with the statistical software package SPSS/PC, version 11.5 (SPSS).
Results
Patient Characteristics and Incidence of IAP
During the study period, 12,162 patients were admitted to the ICU, 4,200 of whom stayed in the ICU for more than 2 days. The mean age (± SD) of these 4,200 patients was 
years, and they were hospitalized for 23,791 days. A total of 498 patients (12%) received a diagnosis of pneumonia, of whom 298 were classified as having IAP. Of the 200 patients without IAP, 82 had onset of pneumonia less than 48 hours after ICU admission, and 98 had no clinical symptoms of pneumonia. Fifty‐six of the 298 patients with IAP developed more than 1 episode of pneumonia; 32 episodes were classified as new episodes of IAP, for a total of 330 episodes of IAP. A total of 289 (90%) of these episodes were ventilator associated. When only considering the first episode of pneumonia, the overall incidence of nosocomial pneumonia was 13 cases per 1,000 ICU‐days (95% CI, 11‐14 cases per 1,000 ICU‐days).
Differences between patients with and patients without nosocomial pneumonia are given in Table 1. A greater percentage of patients with IAP were male, received ventilation (before the diagnosis of nosocomial pneumonia), and were admitted because of trauma, and they had a higher mean SAPS II. Importantly, patients with IAP had a longer mean ICU stay and a higher risk of dying during their ICU stay, compared with patients without IAP (31.5% vs 13.5%; 
). However, we do not know whether death during ICU stay was attributable to pneumonia, because information on the cause of death is missing in our database.
Distribution of Microorganisms
Analysis of the distribution of the different microorganisms for the 330 episodes of IAP revealed that the most common isolated pathogens were P. aeruginosa (isolated in 16% of episodes), Haemophilus influenzae (16%), MSSA (15%), Escherichia coli (15%), S. marcescens (15%), Enterobacter species (14%), Klebsiella pneumoniae (13%), and Proteus species (10%). Of the 54 episodes in which P. aeruginosa was isolated, 14 occurred in patients who had a medical history of chronic obstructive pulmonary disease, and 1 occurred in a patient with a medical history of bronchiectasis. Three of 45 Enterobacter isolates and 2 of 42 K. pneumoniae isolates were ESBL producers. A total of 103 episodes (31%) were polymicrobial.
Of the 141 episodes in group 1, P. aeruginosa was detected in 16 (11%), Enterobacter species were detected in 14 (10%), and S. marcescens was detected in 21 (15%) (Table 2). Demographic and clinical characteristics of the 16 patients in group 1 from whom P. aeruginosa was isolated are given in Table 3.
Although some patients in group 1 had study pathogens isolated, most isolates of study pathogens were recovered from patients in groups 2 and 4. Among patients in groups 1 and 2, isolation of study pathogens occurred in 101 (52%) of 193 episodes.
We performed a second analysis, taking into account the duration of hospitalization instead of the duration of ICU stay. Again, we found that the pathogens isolated from patients in group 1 were P. aeruginosa (in 8 [9%] of 90 episodes), Enterobacter species (in 12 [13%] of 90 episodes), and S. marcescens (in 11 [12%] of 90 episodes). Similar results were obtained in a third analysis, which involved the duration of mechanical ventilation (data not shown).
Changing the cutoff value for early‐onset IAP from onset less than 7 days after admission to onset less than 6, less than 5, or less than 4 days did not result in dramatically different results. The percentage of episodes in which study pathogens were recovered in group 1 changed from 45% (63 of 141 episodes) for a cutoff of less than 7 ICU‐days to 43% (47 of 110 episodes) for a cutoff of less than 6 ICU‐days, 48% (35 of 73 episodes) for a cutoff of less than 5 ICU‐days, and 45% (17 of 38 episodes) for a cutoff of less than 4 ICU‐days. The percentage of episodes decreased somewhat when the cutoff was applied to the duration of hospitalization: study pathogens were recovered in 42% (38 of 90) of the episodes that occurred less than 7 days after hospital admission, compared with 41% (27 of 66 episodes) for a cutoff of less than 6 days, 39% (14 of 36 episodes) for a cutoff of less than 5 days, and 36% (5 of 14 episodes) for a cutoff of less than 3 days.
The mortality rate among patients with IAP was 32% (94 of 298). This rate was slightly higher among patients from whom study pathogens were isolated (35% [52 of 148]), compared with patients from whom such pathogens were not isolated (28% [42 of 150]). The mortality rate was the highest (25 of 63; 39%) for patients in group 1 from whom study pathogens were isolated.
Risk Factors for the Recovery of Study Pathogens
Factors that increased the risk for the recovery of study pathogens were age, SAPS II, duration of hospitalization, ward or institution of origin, and previous history of antibiotic use (both prophylactic and therapeutic) (Table 4). After adjusting for age, SAPS II, ward or institution of origin, medical history of respiratory diseases, and duration of hospital stay, only age and previous history of antibiotic use remained risk factors (adjusted odds ratio for therapeutic use, 8.2 [95% CI, 2.8‐23.8]; adjusted odds ratio for prophylactic use, 4.6 [95% CI, 1.6‐13.0]).
Additional analysis identified age and previous antibiotic use (both prophylactic and therapeutic) as risk factors for the isolation of P. aeruginosa and ESBL‐producing pathogens (Table 5). Isolation of study pathogens was most strongly associated with previous use of cephalosporins (third‐ and fourth‐generation agents), aminoglycosides, and imipenem.
Discussion
In this cohort study, we showed that one or more of the study pathogens was isolated from more than 50% of the patients in groups 1 and 2 and that the main risk factor for isolation of these pathogens was the previous use of antibiotics, both as therapy and as prophylaxis.
Known determinants of the etiology of nosocomial pneumonia are the interval between admission and onset of pneumonia, previous history of antibiotic therapy, and unit‐specific epidemiological factors.7,9 Numerous reports have described the relationship between isolation of the study pathogens and late‐onset pneumonia.7,11‐19 Trouillet et al.7 studied risk factors for the isolation of such pathogens from patients with ventilator‐associated pneumonia and found that the duration of mechanical ventilation and the history of antibiotic use, especially use of broad‐spectrum agents, were strongly related to the isolation of these pathogens. In our study, we confirmed the impact of prior therapeutic use of antibiotics on the etiology of pneumonia, but one of the most important findings is the fact that even prophylactic use of antibiotics was associated with an increased risk for the isolation of study pathogens. This was nicely illustrated by the results from group 1, in which 14 of the 16 patients from whom P. aeruginosa was isolated had received antibiotic prophylaxis before surgery. We know from the literature that receipt of broad‐spectrum antibiotic treatment before onset of pneumonia is linked to an increased risk of pneumonia caused by highly virulent organisms, such as P. aeruginosa and Acinetobacter species.9,10,19 On the basis of our findings, it seems that prophylactic use of antibiotics also alters the microbial flora of hospitalized and critically ill patients by favoring colonization with the study pathogens.
Data from a prospective cohort study at a university‐affiliated urban teaching hospital in the United States showed that the pathogens recovered from patients with early‐onset IAP were similar to pathogens recovered from patients with late‐onset IAP.15 This was recently confirmed by Giantsou et al.18 in a study of pathogens recovered from patients with ventilator‐associated pneumonia. Rello et al.14 repeated the study of Trouillet et al.7 in 3 different Spanish ICUs and found that the isolated pathogens differed from hospital to hospital and that the pathogens targeted in our study were recovered from patients with early‐onset pneumonia and no history of antibiotic use.
We could not confirm the relationship between the isolation of study pathogens and the length of ICU stay. We used an ICU stay of less than 7 days as the cutoff to distinguish between early‐onset and late‐onset pneumonia, because the median time to onset of pneumonia during the study period was 6 days. Small differences in the percentage of episodes of early‐onset pneumonia in which study pathogens were isolated were observed with the cutoff set at less than 6, less than 5, and less than 4 days. According to the American Thoracic Society guidelines on the treatment of adults with HAP, early‐onset HAP is defined as a case of pneumonia that occurs within the first 4 days after hospital admission.9 The percentage of episodes in which study pathogens were isolated decreased somewhat when the cutoff was applied to the duration of hospitalization (from 42% for a cutoff of less than 7 days to 36% for less than 4 days) but was still substantial.
The fact that many patients with ventilator‐associated pneumonia or HAP are infected with the study pathogens we investigated impairs the adequacy of initial, empirical treatment with antibiotics. Clearly, appropriate selection of antibiotics for empirical treatment is critical because inadequate antimicrobial coverage is associated with increased mortality.20 Although the pathogens isolated differed from those we expected to recover, patients in groups 2, 3, and 4 were adequately covered by initial empirical therapy. Only patients in group 1 had inadequate antibiotic coverage in case of colonization with study pathogens, because they received empirical treatment with amoxicillin–clavulanic acid or monotherapy with a second‐generation cephalosporin. Mortality in this group was the highest (39%). However, because information on the cause of death is missing from our database, it would be presumptuous to conclude that the high mortality rate was solely related to inappropriate antibiotic coverage.
As with all observational research, this study has a number of limitations. First, we performed semiquantitative culture of endotracheal or bronchial aspirates to assess the etiology of pneumonia episodes. Although we are well aware that semiquantitative culture of tracheal aspirates is not as reliable as more‐invasive techniques to confirm the presence of pneumonia and the need for antibiotic therapy, recent guidelines recognize that, although this method is not as reliable as quantitative culture, it is useful for guiding decisions about antibiotic therapy.9,21,22 Second, information about previous antibiotic use might have been missing for patients transferred from another hospital or admitted from the community. Although a rigorous attempt was made by ICU staff to record all relevant data on patients in the study, the number of patients with a past history of antibiotic use was probably underestimated.
We believe that our results confirm that hospital guidelines on empirical antibiotic treatment should be based on institutional surveillance data and not solely on international or national guidelines. We also want to stress the importance of a good anamnesis concerning antibiotic use (both prophylactic and therapeutic) during the 30 days prior to admission, because of the high prevalence of colonization by the study pathogens we investigated among these patients. An incomplete medical history leads to misclassification of these patients, administration of inadequate empirical treatment, and potentially death. On the basis of our findings, we will amend our current antibiotic guidelines by including prophylactic antibiotic use as risk factor for the isolation of the study pathogens, and despite the weaker correlation, we will change our definition of early‐onset pneumonia as onset of pneumonia less than 5 days after hospital admission.
References
- 1. Bowton DL. Nosocomial pneumonia in the ICU—year 2000 and beyond. Chest 1999; 115:28S–33S.
- 2. Vincent JL, Bihari DJ, Suter PM, et al. The prevalence of nosocomial infection in intensive care units in Europe. Results of the European Prevalence of Infection in Intensive Care (EPIC) Study. EPIC International Advisory Committee. JAMA 1995; 274:639‐644.
- 3. Leroy O, Soubrier S. Hospital‐acquired pneumonia: risk factors, clinical features, management, and antibiotic resistance. Curr Opin Pulm Med 2004; 10:171‐175.
- 4. National Surveillance of Infection in Hospitals Web site. Available at: http://www.nsih.be. Accessed February 16, 2007.
- 5. Hoffken G, Niederman MS. Nosocomial pneumonia: the importance of a de‐escalating strategy for antibiotic treatment of pneumonia in the ICU. Chest 2002; 122:2183‐2196.
- 6. Infectious Diseases Advisory Board. Diagnosis and Antibacterial Treatment of VAP. Belgium; 2002.
- 7. Trouillet JL, Chastre J, Vuagnat A, et al. Ventilator‐associated pneumonia caused by potentially drug‐resistant bacteria. Am J Respir Crit Care Med 1998; 157:531‐539.
- 8. Hospitals in Europe Link for Infection Control Through Surveillance Web site. Available at: http://helics.univ‐lyon1.fr/home.htm. Accessed February 16, 2007.
- 9. Guidelines for the management of adults with hospital‐acquired, ventilator‐associated, and healthcare‐associated pneumonia. Am J Respir Crit Care Med 2005; 171:388‐416.
- 10. Park DR. The microbiology of ventilator‐associated pneumonia. Respir Care 2005; 50:742‐765.
- 11. Bonten MJ, Bergmans DC, Ambergen AW, et al. Risk factors for pneumonia, and colonization of respiratory tract and stomach in mechanically ventilated ICU patients. Am J Respir Crit Care Med 1996; 154:1339‐1346.
- 12. Chevret S, Hemmer M, Carlet J, Langer M. Incidence and risk factors of pneumonia acquired in intensive care units: results from a multicenter prospective study on 996 patients. European Cooperative Group on Nosocomial Pneumonia. Intensive Care Med 1993; 19:256‐264.
- 13. George DL, Falk PS, Wunderink RG, et al. Epidemiology of ventilator‐acquired pneumonia based on protected bronchoscopic sampling. Am J Respir Crit Care Med 1998; 158:1839‐1847.
- 14. Rello J, Sa‐Borges M, Correa H, Leal SR, Baraibar J. Variations in etiology of ventilator‐associated pneumonia across four treatment sites: implications for antimicrobial prescribing practices. Am J Respir Crit Care Med 1999; 160:608‐613.
- 15. Ibrahim EH, Ward S, Sherman G, Kollef MH. A comparative analysis of patients with early‐onset vs late‐onset nosocomial pneumonia in the ICU setting. Chest 2000; 117:1434‐1442.
- 16. Akca O, Koltka K, Uzel S, Cakar N, et al. Risk factors for early‐onset, ventilator‐associated pneumonia in critical care patients: selected multiresistant versus nonresistant bacteria. Anesthesiology 2000; 93:638‐645.
- 17. Leroy O, Meybeck A, d’Escrivan T, Devos P, Kipnis E, Georges H. Impact of adequacy of initial antimicrobial therapy on the prognosis of patients with ventilator‐associated pneumonia. Intensive Care Med 2003; 29:2170‐2173.
- 18. Giantsou E, Liratzopoulos N, Efraimidou E, et al. Both early‐onset and late‐onset ventilator‐associated pneumonia are caused mainly by potentially multiresistant bacteria. Intensive Care Med 2005; 31:1488‐1494
- 19. Chastre J, Fagon JY. Ventilator‐associated pneumonia. Am J Respir Crit Care Med 2002; 165:867‐903
- 20. Fagon JY, Chastre J. Antimicrobial treatment of hospital‐acquired pneumonia. Clin Chest Med 2005; 26:97‐104.
- 21. Niederman MS. The clinical diagnosis of ventilator‐associated pneumonia. Respir Care 2005; 50:788‐796.
- 22. Baughman RP. Microbiologic diagnosis of ventilator‐associated pneumonia. Clin Chest Med 2005; 26:81‐86.