A retrospective observational study was performed in the medical ICU of the Seoul National University Hospital between January 2011 and August 2011. We included patients with pneumonia treated in the medical ICU. Patients were enrolled if they had pneumonia on admission or developed pneumonia during their ICU stay. A total of 242 patients were admitted and treated in the medical ICU over the study period; in addition, data from 102 patients with pneumonia were analyzed retrospectively.

Pneumonia was clinically classified based on the American Thoracic Society/Infectious Disease Society of America guidelines [11,12]. According to these guidelines, healthcare-associated pneumonia (HCAP) was defined as pneumonia in any patient admitted to an acute care hospital for ≥ 2 days within 90 days of the infection; who resided in a nursing home or long-term care facility; who received recent intravenous antibiotic therapy, chemotherapy, or wound care within 30 days of onset of the current infection; or who attended a hospital or hemodialysis clinic within 30 days. HAP was defined as pneumonia that developed ≥ 48 hours after admission. Community-acquired pneumonia (CAP) was defined as pneumonia that did not meet any of the HCAP and HAP criteria.

The causative organisms were considered to be micro-organisms that had been isolated from specimens, including blood, bronchoalveolar lavage fluid, bronchial wash, and pleural effusion, and which grew to greater than a threshold concentration in quantitative cultures. Specimens obtained by endotracheal aspiration or in sputum expectorant were evaluated as appropriate using counts of white blood cells and epithelial cells, and the micro-organisms identified were considered the causative pathogen. Growth below the threshold was considered to be caused by colonization or contamination. Legionella and Streptococcus urinary antigen tests, viral polymerase chain reaction, and antigen tests were also used to identify the pathogens. Pathogens may have been confirmed in additional samples after a patient developed pneumonia, but a secondary infection due to a hospital- or ICU-acquired pathogen was ruled out. Specimens sampled within 3 days after a patient developed pneumonia were considered significant.

Empirical antibiotic regimens to treat pneumonia were reviewed, and their response and relevance to subsequent changes in the antibiotic regimen were analyzed. The initial empirical antibiotic regimen was administered according to the American Thoracic Society/Infectious Disease Society of America guidelines [11,12]. Initial non-responders to empirical antibiotics were defined as cases in which the initial antibiotic was changed due to expansion or a switch in the antibiotic spectrum due to persistence or worsening of the clinical course within 1 week after developing pneumonia. This did not include cases in which empirical antibiotics were changed to the targeted regimen based on drug susceptibility in clinically improving patients.

We analyzed ICU mortality, duration of ICU stay, duration of mechanical ventilation, frequency of re-intubation, and frequency of tracheostomy as clinical outcomes of pneumonia. These outcomes were compared between patients with CAP, HCAP, and HAP.

This retrospective data collection was approved by the Institutional Review Board of Seoul National University (IRB no: H-1208-150-424) and was performed in accordance with the Declaration of Helsinki.

Data are presented as means and standard deviations for continuous variables and as numbers (percentages) for categorical variables. We analyzed baseline characteristics, clinical outcomes, and prognoses according to specific pathogens using the Kruskal-Wallis test. We used the Mann-Whitney test to compare data for prognosis according to whether a pathogen or MDR pathogen was identified. Associations between prognostic factors and survival rate were investigated using the Cox proportional-hazard analysis after adjusting for age, sex, Acute Physiology and Chronic Health Evaluation (APACHE) II score [13], Sequential Organ Failure Assessment (SOFA) score [14], Charlson comorbidity index [15], pneumonia severity index [16], the identification of the pathogen or MDR pathogen, and the pneumonia classification. Likelihood ratio tests were used to examine the goodness of fit of the model, and no lack of fit was found. Adjusted odd ratios (aOR) are presented with 95% confidence intervals (CIs). A p < 0.05 was considered significant. Data were analyzed using SPSS version 18 (SPSS Inc., Chicago, IL, USA).

During the study period, 242 patients were admitted to the ICU; 102 had pneumonia, and all were eligible for this study. The study population was comprised of 20 patients (19.6%) with CAP, 26 (25.5%) with HCAP, and 56 (54.9%) with HAP.

The baseline characteristics and clinical outcomes of the patients treated for pneumonia in the ICU are presented in Table 1. Mean age of the entire population was 64.1 ± 14.3 years, and 75.5% were men. No significant differences were found between the CAP, HCAP, and HAP groups for age, sex distribution, or severity as evaluated by the APACHE II score and the SOFA score. However, the Charlson comorbidity index and pneumonia severity index scores were significantly higher in the HCAP and HAP groups than in the CAP group; the scores did not differ between the HCAP and HAP groups.

The causative pneumonia pathogens were identified in 50 patients (49.0%). No significant differences in baseline characteristics were observed according to the pathogens identified except for a history of solid tumor within 5 years and the presence of a MDR pathogen.

Among 50 cases in which a pathogen was identified, sputum (24, 48%), blood (17, 34%), endotracheal aspirate (12, 24%), bronchoalveolar lavage (4, 8%), serology (4, 8%), and bronchial wash (2, 4%) were the specimens in which pathogens were confirmed. The pathogens causing pneumonia were identified in nine patients (45%) with CAP, in 11 (42.3%) with HCAP, and in 30 (53.6%) with HAP. No significant difference in the incidence of pathogens was observed among the pneumonia types after adjusting for culture frequency. The incidence of pneumonia with a MDR pathogen was 22 patients (21.6%) in the entire population. The incidence was significantly higher in the HAP group than in the HCAP and CAP groups (HAP, 32.1%; HCAP, 11.5%; CAP, 5.0%; p = 0.015), but was not different between the CAP and HCAP groups. The MDR micro-organisms were: carbapenem-resistant A. baumannii, MRSA, coagulase-negative Staphylococcus, vancomycin-resistant enterococci, extended-spectrum b-lactamase-producing Escherichia coli, Klebsiella, and MDR Pseudomonas species.

Table 2 shows the distribution of specific pathogens and the respective prognoses. S. aureus was the most frequently isolated pathogen among all patients in all groups (10, 20%). Pseudomonas species were detected in eight patients (16%), Klebsiella pneumoniae in seven (14%), and A. baumannii in six (12%). No significant difference was observed in the clinical outcomes between specific pathogens.

The empirical antibiotics and their responses are shown in Table 3. The most commonly used antibiotics as an empirical regimen, including combination therapy, were β-lactam/β-lactamase inhibitors (62, 60.8%), followed by quinolones (57, 55.9%). About 58% of the patients responded initially to the empirical antibiotic regimen. Appropriate antibiotics were initially used in 60% of cases when a retrospective review was performed considering pathogen drug susceptibility and the antibiotic spectrum in cases in which the pathogen was identified. In these cases, a subsequent change to the targeted antibiotic occurred in 56% of cases, and most (89.3%) were appropriate considering the drug susceptibility of the pathogen.

The prognoses of pneumonia based on identifying the pathogens are shown in Table 4. The ICU mortality rate was 47.1% in the entire study population, and did not differ according to whether the pathogen was identified (unidentified pathogen 46.2% vs. identified pathogen 48.0%, p = 0.853). Other clinical outcomes such as duration of ICU stay, need for and duration of mechanical ventilation, and frequencies of re-intubation and tracheostomy were different based on whether the pathogen was identified. These results were consistent with cases in which MDR pathogens were identified.

A subgroup analysis was performed according to pneumonia type, and the clinical outcomes did not differ significantly between the groups.

A Cox proportional-hazard analysis was performed to evaluate the relative risk of ICU mortality caused by pneumonia, and the data are presented in Table 5. The ICU mortality rate was significantly higher in the group with higher APACHE II and SOFA scores in both univariate and multivariate analyses ([APACHE II score: aOR, 1.07; 95% CI, 1.03 to 1.11]; [SOFA score: aOR, 1.26; 95% CI, 1.14 to 1.38]). Age was also a significant prognostic factor for ICU mortality in the multivariate analysis (aOR, 1.04; 95% CI, 1.01 to 1.07). The pneumonia severity index was significantly associated with ICU mortality in the univariate analysis, but significance was lost after adjustment. Age, sex, the Charlson comorbidity index, identification of the pathogen or MDR pathogen, initial responder to empirical antibiotic therapy, and pneumonia classification were not significant prognostic indicators in the ICU mortality evaluation.