JSTOR

Methods

This study was conducted at the University of North Carolina (UNC) Hospitals. During the period analyzed, 2000 through 2003, infection control surveillance was conducted by 5 infection control professionals supervised by 2 full‐time faculty. Comprehensive hospitalwide surveillance (ie, of all patients and hospital locations) was performed using criteria developed by the Centers for Disease Control and Prevention (CDC),¹² with the exception that the diagnosis of pneumonia required demonstration of an infiltrate on a chest radiograph. This modified definition is similar to the most recent definition used by hospitals participating in the National Nosocomial Infections Surveillance (NNIS) System.¹³ Hospital‐acquired pneumonia (HAP) was defined as pneumonia that developed 48 hours or more after admission that was not incubating at the time of admission. Ventilator‐associated pneumonia (VAP) was defined as pneumonia that developed more than 48 hours after endotracheal intubation. Our definitions for HAP and VAP were consistent with those used for these diagnoses by the American Thoracic Society (ATS) and the Infectious Diseases Society of America (IDSA).⁸ Our database did not include patients with a diagnosis of community‐acquired pneumonia (CAP) or healthcare‐associated pneumonia (HCAP). Cases were entered into a computerized database after having been reviewed by a nurse supervisor and a physician who specialized in infectious diseases and critical care medicine. Specimens were considered adequate for processing by the UNC Microbiology Laboratory if they were of 1 of the following 3 types: (1) bronchoscopically obtained specimens (cultures of bronchoalveolar lavage fluid samples were performed quantitatively, and isolates reported if results showed 10⁴ or more colony forming units (cfu) per mL; cultures of bronchial brush samples were performed semiquantitatively); (2) expectorated sputum in which the number of polymorphonuclear cells exceeded the number of epithelial cells and organisms (other than just yeast) seen during oil immersion examination; and (3) tracheal aspirates in which there were 25 epithelial cells or fewer per low‐power field and organisms (other than yeast) seen during oil immersion examination. Pathogens isolated from the respiratory tract were identified using standard techniques.

Data available from this prospectively maintained database included the patient’s name, age, sex, unit number, and admission date; the date of onset of infection; the site of infection; indwelling devices present (ie, endotracheal tube and/or central venous catheters), and the infecting pathogen(s). The antimicrobial susceptibilities of the infecting flora were recorded on patient case forms but not entered into the computerized database, with the exception of oxacillin resistance in Staphylococcus aureus and vancomycin resistance in Enterococcus species. P values were calculated using the 2‐tailed Fisher exact test.

Results

Epidemiology

During the period of this study, 2000 through 2003, our hospital had 158,519 patients admitted. Overall, 5,000 (3.15%) of these patients developed a total of 5,997 healthcare‐associated infections (3.78 infections per 100 patients admitted). The number of nosocomial pneumonia cases by year was as follows: in 2000, there were 141 infections; in 2001, there were 145 infections; in 2002, there were 133 infections; and in 2003, there were 169 infections. A total of 588 lower respiratory tract infections were reported in 556 patients (Table 1). Thus, the overall incidence of nosocomial pneumonia was 0.37%.

Table 1.  Epidemiology of Ventilator‐Associated Pneumonia (VAP) and Hospital‐Acquired Pneumonia in Nonventilated Patients (HAP)

Open New Window

Approximately 85% of the VAP cases occurred in patients from the surgical service. Although HAP occurred most commonly in patients from the surgical service (56%), patients from the medical service accounted for approximately one‐third of all cases. More than 90% of the VAP cases occurred in patients housed in ICUs, whereas approximately 67% of the HAP cases occurred in patients not housed in ICUs.

For patients with VAP, the time from admission to the onset of infection was as follows: mean, 22.4 days; median, 12.0 days; and mode, 4 days. For patients with HAP, the time from admission to infection was as follows: mean, 25.0 days; median, 11.5 days; and mode, 4 days. Overall, 14.4% of the VAP cases and 20.9% of the HAP cases occurred within the first 4 days of hospitalization.

Microbiology

A pathogen was isolated from 92.4% of patients with VAP but from only 76.6% of patients with HAP. Overall, 400 pathogens were isolated from 327 cases of VAP, and 270 pathogens were isolated from 261 cases of HAP (Table 2).

Table 2.  Relative Frequency of Isolation of Selected Pathogens from Patients With Ventilator‐Associated Pneumonia (VAP) and Nonventilated Patients With Hospital‐Acquired Pneumonia (HAP)

Open New Window

Among patients with VAP, gram‐positive cocci were isolated from 32.0% of infections, gram‐negative bacilli from 59.0%, and the remainder of isolates were miscellaneous pathogens. The most common gram‐positive pathogen isolated was Staphylococcus aureus. Approximately two‐thirds of S. aureus isolates were oxacillin resistant. Approximately two‐thirds of the gram‐negative isolates were non‐Enterobacteriaceae bacilli, principally Pseudomonas aeruginosa, Acinetobacter species, Stenotrophomonas maltophilia, and Haemophilus species.

Among patients with HAP, gram‐positive cocci were isolated from 42.59% of infections, gram‐negative bacilli from 39.63%, and the remainder of isolates were miscellaneous pathogens. The most common gram‐positive pathogen isolated was S. aureus. Sixty percent of S. aureus isolates were oxacillin resistant. Slightly less than half of the gram‐negative pathogens isolated were non‐Enterobacteriaceae bacilli, principally P. aeruginosa, Acinetobacter species, Haemophilus species, and Moraxella catarrhalis.

Compared with pneumonia in ventilated patients, pneumonia in nonventilated patients was more likely to be caused by gram‐positive cocci ( ), especially Streptococcus pneumoniae ( ). Pneumonia in nonventilated patients was less likely to be caused by non‐Enterobacteriaceae bacilli ( ), including P. aeruginosa ( ), Acinetobacter species ( ), and S. maltophilia ( ). The percentage of infections due to Enterobacteriaceae was similar in both groups ( ).

When patients with VAP and patients with HAP were stratified by the location where they received care (ICU vs non‐ICU ward), there were no dramatic differences apparent with regard to etiologic agents (Table 3). Patients with VAP had a significantly lower rate of infection with Escherichia coli while residing in the ICU, whereas patients with HAP had a significantly lower rate of infection with oxacillin‐resistant S. aureus while residing in the ICU.

Table 3.  Relative Frequency of Isolation of Selected Pathogens From Patients With Ventilator‐Associated Pneumonia (VAP) and Nonventilated Patients With Hospital‐Acquired Pneumonia (HAP), as a Function of Hospital Location of Care

Open New Window

Microbiology As a Function of Time of Infection

In patients with VAP, the pathogens statistically associated with early onset of infection (ie, 0‐4 days after endotracheal intubation) included oxacillin‐susceptible S. aureus, S. pneumoniae, and Hemophilus species (Table 4). Pathogens occurring more frequently after 5 or more days of hospitalization included Acinetobacter species and S. maltophilia. Enteric gram‐negative pathogens were not statistically associated with either early or late onset of infection.

Table 4.  Frequency of Isolation of Selected Pathogens from Patients With Ventilator‐Associated Pneumonia (VAP), as a Function of Duration of Hospitalization

Open New Window

In patients with HAP, the pathogens statistically associated with early onset of infection (ie, 0‐4 days after hospitalization) included only S. pneumoniae, although oxacillin‐susceptible S. aureus infections tended to occur early (Table 5). Pathogens occurring more frequently after 5 or more days of hospitalization included oxacillin‐resistant S. aureus and P. aeruginosa. Enteric gram‐negative pathogens were not statistically associated with either early or late onset of infection.

Table 5.  Frequency of Isolation of Selected Pathogens From Nonventilated Patients With Hospital‐Acquired Pneumonia (HAP), as a Function of Duration of Hospitalization

Open New Window

Discussion

Nosocomial pneumonia is an important cause of morbidity and mortality for hospitalized patients. The epidemiology, diagnosis, treatment, prevention, and outcome of nosocomial pneumonia have been the subject of several recent reviews.^9,14‐17 In the United States, pneumonia accounts for approximately 15% of all healthcare‐associated infections and 25% of nosocomial infections in ICUs. The primary risk factor for the development of nosocomial pneumonia is receipt of mechanical ventilation. The microbiology of VAP has been well described. Chastre and Fagon⁹ recently summarized the etiology of VAP as determined by bronchoscopic techniques in 24 studies. On this analysis, the etiologic agents of VAP included gram‐positive cocci (34% of isolates [S. aureus, 20%]), Enterobacteriaceae (14%), P. aeruginosa (24%), Acinetobacter species (8%), S. maltophilia (2%), Haemophilus species (10%), and miscellaneous pathogens (8%). Only limited data are available regarding the epidemiology of HAP, including the causative pathogens.^10,11 Because current treatment guidelines recommend the same empirical antibiotic therapy for VAP and HAP, we reviewed our prospectively obtained data on nosocomial pneumonia to determine whether the pathogens isolated from patients with VAP and those isolated from patients with HAP were similar.

Between 2000 and 2003, there were 588 lower respiratory tract infections reported in 556 patients cared for in our hospital. Our data revealed that the bacterial etiology of VAP and HAP differed quantitatively, but were qualitatively similar. Patients with VAP, compared with those with HAP, were more likely to be from the surgical service and to be cared for in an ICU. The etiologic agents of VAP at our hospital were similar to those reported by Chastre and Fagon⁹ in their review of the literature. While the overall frequency of infection with gram‐positive cocci was virtually identical, our patients had a higher frequency of infections with S. aureus. The frequency of infection with enteric bacilli, approximately 15%, was similar. The frequency of infection with gram‐negative pathogens that are more likely to exhibit resistance to multiple antibiotics (ie, P. aeruginosa, Acinetobacter species, and S. maltophilia)¹⁸ was also similar, although we had fewer patients infected with P. aeruginosa and more patients infected with S. maltophilia. Compared with patients with VAP, our patients with HAP had a higher frequency of infection with gram‐positive cocci and a lower frequency of infection with nonenteric gram‐negative bacilli. However, more than 13% of patients with HAP had pneumonia due to P. aeruginosa, Acinetobacter species, or S. maltophilia. Recently, Kollef and colleagues¹¹ noted that Pseudomonas species or Acinetobacter species were associated with more than 20% of cases of HAP. From a clinical perspective, this frequency is clinically high enough to warrant the use of antibiotics with activity against these pathogens when choosing an empirical regiment for the treatment of nosocomial pneumonia. Hence, the recommendations of the current ATS/IDSA guideline⁸ are appropriate for the treatment of patients with either VAP or HAP. The guideline also recommends similar therapy for patients with risk factors for infection with more‐resistant pathogens (eg, methicillin‐resistant Staphylococcus aureus [MRSA] and P. aeruginosa), such as admission from an extended care facility, recent hospital admission, or recent antibiotic therapy (such cases of pneumonia are now termed healthcare‐associated pneumonia). However, patients with healthcare‐associated pneumonia are not ascertained using Centers for Disease Control and Prevention nosocomial pneumonia criteria and therefore were not included in our database.

The pathogens associated with HAP have been reported to differ depending on the time of onset of infection.^19‐24 Early‐onset pneumonia has been associated with higher frequencies of infection with S. pneumoniae, H. influenzae, and oxacillin‐susceptible S. aureus. In contrast, late‐onset pneumonia has been associated with higher frequencies of infection with oxacillin‐resistant S. aureus, P. aeruginosa, Acinetobacter species, and S. maltophilia. Comparisons among studies are limited by differing definitions of early‐onset infection (ie, ranging from less than 4 days to less than 7 days) and of late‐onset infection (ie, ranging from 4 days or more to 7 days or more), baseline time (ie, admission to hospital, admission to ICU, or initiation of mechanical ventilation), and initial exclusionary period for nosocomial pneumonia (ie, 48‐72 hours). Our data, which were based on a 4‐day cutoff between early‐ and late‐onset pneumonia, as in the ATS/IDSA guideline,⁸ demonstrated that early‐onset infections were more likely to be caused by S. pneumoniae, oxacillin‐susceptible S. aureus, and H. influenzae, whereas late‐onset infections were more likely to be caused by oxacillin‐resistant S. aureus, Acinetobacter species, S. maltophilia, and P. aeruginosa. This pattern was found for both VAP and HAP. However, more than 26% of cases of early‐onset VAP and more than 19% of early‐onset cases of HAP were due to pathogens requiring treatment with broad‐spectrum agents (ie, P. aeruginosa, S. maltophilia, or Acinetobacter species) or therapy active against oxacillin‐resistant S. aureus. Thus, our data are similar to those of Ibrahim and colleagues,²² who reported that oxacillin‐resistant S. aureus and P. aeruginosa can be important pathogens in early‐onset pneumonia. Using a 7‐day cutoff, Giantsou and colleagues²⁴ reported that similar proportions of early‐ and late‐onset cases of VAP were caused by multiresistant pathogens, P. aeruginosa, and MRSA. The explanation for our findings, and those of others, is that factors other than time of onset (such as recent hospitalization and/or prior receipt of antibiotics) affect the frequency of infection with specific drug‐resistant pathogens.^18,25,26 For example, the duration of mechanical ventilation and prior antibiotic use has been shown to independently predict infection with resistant pathogens.¹⁸ Unfortunately, we could not further stratify our patients by factors such as receipt of prior antibiotic therapy because this information is not included in our database. Such stratification of patients eligible for narrow‐spectrum antibiotic therapy (ie, those developing hospital‐acquired pneumonia within 4 days of hospitalization who have received no prior antimicrobial therapy and have no risk factors for multidrug‐resistant pathogens) is likely to result in only a small percentage of patients being eligible for such therapy.

The strengths of our study include the consistent and strict use of CDC criteria for ascertainment of nosocomial pneumonia, the prospective acquisition of the data, and the use of comprehensive hospitalwide surveillance. There are several limitations that affect our data. First, no data were available on the use of antibiotics prior to the onset of pneumonia. As previously noted, the prior use of antibiotics is a well‐described and independent risk factor for pneumonia due to drug‐resistant pathogens.¹⁸ Second, infecting flora was most often isolated from tracheal aspiration samples, for patients with VAP, and from expectorated sputum samples, for patients with HAP. Although sputum samples obtained by tracheal aspiration are known to be a less accurate reflection of the true infecting flora associated with VAP, the close match between our etiologic data and those obtained in studies that have used bronchoscopic techniques argues that our data are appropriate for comparing the pathogens associated with VAP and HAP. Finally, our analysis was based on the time from admission to the onset of infection, similar to the method for calculating time intervals used in the ATS/IDSA guideline.⁸ We were unable to analyze our data for patients with VAP according to duration of mechanical ventilation.

In conclusion, our data demonstrated that patients with HAP, compared with those with VAP, less commonly had infections with nonenteric bacteria, including P. aeruginosa, Acinetobacter species, and S. maltophilia. However, the overall frequency of infection with these pathogens was sufficiently high to warrant the use of empirical therapy likely be active against these pathogens in hospitalized, nonventilated patients with suspected pneumonia. In addition, similar to patients with VAP, patients with HAP can have the nature of their infecting flora influenced by duration of hospitalization.