Abstract

INTRODUCTION

Pneumonia is the leading infectious cause of mortality in the United States with 53,667 attributed deaths in 2011 (1). The etiology of pneumonia varies depending on the patient population being observed. Traditionally, pneumonia has been divided into community-onset pneumonia, which is most commonly due to Streptococcus pneumoniae, atypical bacteria, and respiratory viruses, and hospital-acquired pneumonia (HAP), which more frequently involves resistant pathogens such as Pseudomonas aeruginosa and methicillin-resistant Staphylococcus aureus (MRSA) (2, 3).

Due to the increase in complicated and chronically ill patients treated in the outpatient setting, the 2005 American Thoracic Society (ATS)/Infectious Disease Society of America (IDSA) guidelines introduced the health care-associated pneumonia (HCAP) classification as a way to identify patients from the community who may be at risk for infection with antimicrobial-resistant pathogens (3). This recommendation greatly expanded the number of patients treated with broad-spectrum antimicrobial agents. For example, over 30% of 58,118 community-onset pneumonia patients in 128 Veterans Affairs (VA) hospitals received empirical treatment with vancomycin (4). The HCAP classification has been controversial, because its ability to predict multidrug-resistant organisms (MDROs) has not been consistently replicated, and the frequency of MDROs in patients with HCAP has varied significantly (5, 6). A recent systematic review and meta-analysis of 24 studies worldwide demonstrated this and found that the prevalence of MRSA and P. aeruginosa in HCAP varied from 0.7% to 30% and 0.7% to 23%, respectively (7). In addition, viruses are a frequent primary cause of community-onset pneumonia and are often not described in these studies (8,–10).

Due to the varying rates of MDRO isolation and overall lack of U.S. data, determining the regional etiology of community-onset pneumonia (CAP and HCAP) is essential for optimizing patient outcomes while minimizing unnecessary broad-spectrum antimicrobial use. The ATS/IDSA guidelines state that one “recognize the variability of bacteriology from one hospital to another, specific sites within the hospital, and from one time period to another, and use this information to alter the selection of an appropriate antibiotic treatment regimen” (3). More data are required from U.S. centers characterizing the etiology of CAP and HCAP. The aim of this study was to characterize the etiology, MDRO risk factors, and clinical outcomes for patients with CAP and HCAP at a large U.S. academic medical center.

MATERIALS AND METHODS

This was a retrospective observational cohort study of all consecutive patients admitted with CAP or HCAP to The Nebraska Medical Center (a 627-bed academic medical center in Omaha, NE, USA) between 1 January 2010 and 31 December 2011. This protocol was approved by the University of Nebraska Medical Center Institutional Review Board.

RESULTS

A total of 521 patients (50.5% CAP and 49.5% HCAP) were included in the study with characteristics summarized in Table 1. The median age of the cohort was 65 years, and 55% were female.

TABLE 1

Baseline characteristics of patients with community-acquired pneumonia

Characteristica	No. of CAP patients (%) unless otherwise specified			P valueb
	All patients (n = 521)	Patients with MDRO pneumonia (n = 20)	Patients with no MDRO isolated (n = 501)
Age, yr [median (IQR)]	65 (52, 79)	73 (62, 82)	65 (52, 78)	0.059
Female	289 (55.5)	9 (45)	280 (55.9)	0.337
CURB-65 score
Median (IQR)	1 (0, 2)	2 (1, 3)	1 (0, 2)	0.005
Score of >1	217 (41.7)	15 (75)	202 (40.3)	0.002
Score of >2	85 (16.3)	7 (35)	78 (15.6)	0.021
ICU admission	40 (7.7)	4 (20)	36 (7.2)	0.035
Ventilator required	43 (8.3)	3 (15)	40 (8)	0.264
Comorbidities
CVA history	68 (13.1)	6 (30)	62 (12.4)	0.022
CHF	86 (16.5)	7 (35)	79 (15.8)	0.023
COPD	144 (27.6)	6 (30)	138 (27.5)	0.81
Liver disease	30 (5.8)	1 (5)	29 (5.8)	0.877
BUN level of >19 mg/dl	206 (39.5)	15 (75)	191 (38.1)	0.001
Diabetes	148 (28.4)	9 (45)	139 (27.7)	0.093
Immunocompromised	59 (11.3)	1(5)	58 (11.6)	0.363
Aspiration risk	107 (20.5)	7 (35)	100 (20)	0.103
CAP	263 (50.5)	5 (25)	258 (51.5)	0.02
HCAP	258 (49.5)	15 (75)	243 (48.5)	0.02
Admission from a nursing home	127 (24.4)	13 (65)	114 (22.8)	<0.001
Hospitalization for 48 h in the last 90 days	167 (32.1)	9 (45)	158 (31.5)	0.206
Home infusion therapy	4 (0.8)	0	4 (0.8)	0.688
Home wound care	21 (4)	2 (10)	19 (3.8)	0.166
Hemodialysis clinic	28 (5.4)	1 (5)	27 (5.4)	0.94
No. of days hospitalized in the last 90 days [median (IQR)]	0 (0, 3)	0 (0, 10)	0 (0, 3)	0.092
No. of days hospitalized in the last 180 days [median (IQR)]	0 (0, 7)	6 (0, 22)	0 (0, 6)	0.03
Antibiotic use in the last 90 days	166 (31.9)	12 (60)	154 (30.7)	0.006
History of colonization with:
Any MDRO	57 (10.9)	7 (35)	50 (10)	<0.001
MRSA	28 (5.4)	3 (15)	25 (5)	0.052
Pseudomonas	21 (4)	5 (25)	16 (3.2)	<0.001
CRE	2 (0.4)	0	2 (0.4)	0.777
ESBL-producing bacteria	2 (0.4)	0	2 (0.4)	0.777
VRE	14 (2.7)	1 (5)	13 (2.6)	0.514

Open in a separate window

^aIQR, interquartile range; ICU, intensive care unit; CVA, cerebrovascular accident; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; BUN, blood urea nitrogen; CAP, community-acquired pneumonia; HCAP, health care-associated pneumonia; MDRO, multidrug-resistant organism; MRSA, methicillin-resistant Staphylococcus aureus; CRE, carbapenem-resistant Enterobacteriaceae; ESBL, extended-spectrum beta-lactamase; VRE, vancomycin-resistant Enterococcus.

^bP values reported for comparisons between patients with MDRO pneumonia versus patients with no MDRO isolated.

Pneumonia etiology.

Microbiologic tests obtained for patients in the cohort included the following: blood cultures (88.7%), respiratory culture (42.6%), S. pneumoniae urine antigen (18%), Legionella urine antigen (15.7%), respiratory virus panel (15%), and influenza virus rapid antigen (16.3%). Bacterial investigations occurred in the same proportion among patients with HCAP and CAP. Patients with CAP more frequently had positive respiratory viral panels (19%) and influenza virus rapid antigen tests (19%) obtained than HCAP patients did (11% and 13%, respectively). The etiology of pneumonia was determined in 19% of the cohort and is listed in Table 2. The most common identified etiology was primary viral pneumonia followed by S. pneumoniae. At least one MDRO was isolated in 20 (3.8%) patients in the overall cohort with MDRO occurring in 15/258 (5.9%) HCAP patients and 5/263 (1.9%) CAP patients. MRSA was isolated in 11/521 patients (2.1%), and P. aeruginosa was isolated in 9/521 patients (1.7%). No ESBL-producing Enterobacteriaceae or CIRE were isolated.

TABLE 2

Identified etiology of pneumonia

Etiology of pneumoniaa	No. of patients (%)b
	All patients (n = 99)	HCAP patients (n = 46)	CAP patients (n = 53)
Gram-positive bacteria	48 (48)	21 (46)	27 (51)
Streptococcus pneumoniae	24 (24)	5 (12)	19 (36)
Staphylococcus aureus	17 (17)	11 (24)	6 (11)
MRSA	11 (11)	8 (17)	3 (5.7)
MSSA	6 (6)	3 (6.5)	3 (5.7)
Streptococcus spp. other than pneumococcus	5 (5)	3 (6.5)	2 (3.8)
Other	2 (2)	2 (4.3)	0
Gram-negative bacteria	33 (33)	21 (46)	12 (23)
Enterobacteriaceae	13 (13)	8 (17)	5 (9.4)
Pseudomonas aeruginosa	9 (9)	7 (15.2)	2 (3.8)
Haemophilus influenzae	5 (5)	1 (2.2)	4 (7.5)
Moraxella catarrhalis	4 (4)	3 (6.5)	1 (1.9)
Other	2 (2)	2 (4.3)	0
Atypical	2 (2)	0	2 (3.8)
Polybacterial culture	9 (9)	6 (13)	3 (5.7)
Virus	32 (32)	13 (28)	19 (36)
Rhinovirus	12 (12)	3 (6.5)	9 (17)
Influenza virus	9 (9)	3 (6.5)	6 (11.3)
Influenza A virus	6 (6)	2 (4.3)	4 (7.5)
Influenza B virus	3 (3)	1 (2.2)	2 (3.8)
Parainfluenza virus	5 (5)	2 (4.3)	3 (5.7)
Metapneumovirus	4 (4)	2 (4.3)	2 (3.8)
RSV	2 (2)	2 (4.3)	0
Adenovirus	1 (1)	1 (2.2)	0
Primary viral pneumonia	26 (26.3)	11 (24)	15 (28)
Mixed bacterial/viral pneumonia	6 (6)	1 (2.2)	5 (9.4)

Open in a separate window

^aMRSA, methicillin-resistant Staphylococcus aureus; MSSA, methicillin-susceptible Staphylococcus aureus; RSV, respiratory syncytial virus.

^bThe percentages may equal >100%, given some patients had polybacterial or mixed bacterial/viral pneumonia. HCAP, health care-associated pneumonia; CAP, community-acquired pneumonia.

DISCUSSION

In this 2-year observational cohort study of 521 patients with CAP and HCAP, there was a 3.8% prevalence of MDRO. This low prevalence of MDRO community-onset pneumonia contrasts with previous studies in the United States (12,–14). Kollef et al. evaluated the etiology of pneumonia in a large database of hospitalized patients from 59 U.S. centers and found a high rate of MDRO HCAP (26.5% with MRSA, 25.3% with P. aeruginosa) (12). However, only culture-positive patients were included, viral etiologies of pneumonia were not evaluated, and patients were not assessed for study inclusion by their clinical, laboratory, and radiological presentation; these limitations likely contributed to an overestimation of the MDRO prevalence. Two subsequent retrospective studies, each conducted at the same U.S. institution, found a similarly high rate of MDRO, including MRSA and P. aeruginosa in patients with culture-positive HCAP (13, 14). In contrast, a prospective study of all patients admitted to a United Kingdom medical center with community-onset pneumonia over a 4.5-year period found a low rate of MDRO in HCAP with 2.2% of patients having MRSA and 2.2% having P. aeruginosa (15). Similarly, Aliberti et al. conducted a prospective study of consecutive hospitalized patients with community-onset pneumonia at a single center in Italy and found an overall MDRO rate of 3.3% (16).

There are multiple potential reasons for the different MDRO isolation rates. One explanation may be the method of analysis. The previously mentioned United States-based studies included only patients with a bacterial pathogen isolated (12,–14). These studies do not reflect the true spectrum of CAP and HCAP, since prospective studies have defined a specific microbiologic etiology only in 16% to 63.8% of CAP patients and 21% to 67.5% of those with HCAP (15,–19). Moreover, pathogens such as S. aureus and P. aeruginosa may be more easily grown from sputum specimens compared to S. pneumoniae (2). This may result in overdetection of MDRO pathogens relative to more typical respiratory pathogens. Similarly, it is less likely that patients with community-onset pneumonia and an unrevealing respiratory tract culture have pneumonia due to an MDRO; thus, there is clinical value to a negative culture when deciding whether broad-spectrum antibiotics are warranted.

Furthermore, the variability of MDRO prevalence may be based on geographic location and patient population. The National Healthcare Safety Network reports that there is “great variation in the degree of spread of antimicrobial-resistant infections” (20). As an example, 29% of Klebsiella pneumoniae from 14 New York City hospitals harbored the carbapenemase KPC (21). In contrast, no CIRE or ESBL-producing Gram-negative bacilli were found in our study, and only 10 patients have had KPC-positive Gram-negative bacilli at our institution over the last 4 years (unpublished data). Based upon our 2011 institutional antibiogram, third-generation cephalosporin resistance and carbapenem resistance was identified in only 4.9% and 0.3% of institutional cultures yielding Enterobacteriaceae. Our microbiologic findings in community-onset pneumonia are more consistent with studies conducted outside the United States and may be more relatable to areas with low rates of MDRO isolation (15,–17, 22, 23). These findings strongly suggest that treatment choices should be guided by local microbial epidemiology.

Multiple independent risk factors for MDRO were identified in our study. A novel finding was the association between the duration of previous hospitalization in the last 90 or 180 days and the isolation of MDRO. Previous studies have found hospitalization for >48 h in the last 90 days to be a predictor of MDRO isolation, but to our knowledge, none have attempted to quantitate the risk based on previous durations of stay (16, 19, 24). The 48-h cutoff used in the HCAP definition was likely based upon the traditional definition of a nosocomial infection (3). However, the risk of acquiring an MDRO is likely increased with increased duration of prior hospitalization, previous antimicrobial exposure, and the rate of MDRO in the local environment. Epidemiologic studies of MRSA and ESBL Gram-negative colonization support this by demonstrating that colonization is associated with a prolonged hospital stay (25, 26).

The 2005 ATS/IDSA guidelines did not specifically include patients with a known colonization or previous infection with an MDRO in the HCAP definition. However, we found that patients with a clinical culture positive for P. aeruginosa in the previous 12 months was independently associated with MDRO isolation in patients with pneumonia. Furthermore, a history of P. aeruginosa in the previous 12 months was associated with P. aeruginosa pneumonia. Previous MRSA colonization/infection in the previous 12 months did not reach statistical significance for predicting MRSA pneumonia (P = 0.11). Shindo et al. recently found that a previous MRSA history in the last 90 days was associated with MRSA pneumonia (19). These findings suggest that previous colonization/infection with MDRO pathogens should be included in empirical antimicrobial decisions.

While the HCAP classification was not independently associated with MDRO in our propensity score-based multivariate analysis, one component of the definition, admission from a nursing home, was. Many previous studies have attempted to validate the utility of the HCAP definition for identification of risk factors for MDRO isolation with mixed results (7, 19). A recent meta-analysis of 24 studies suggested that the HCAP classification poorly discriminated between patients at a high or low risk for MDRO (7). Importantly, the authors suggest that utilizing the current HCAP classification will yield unnecessary MDRO therapy in areas with a low prevalence of MDRO and undertreatment in areas with high rates of MDRO. A new definition that incorporates MDRO epidemiology in the patient population of interest and patient-specific risk factors is needed. Reflexive prescribing of combinations of broad-spectrum antibiotics directed at MDROs for all patients meeting the current definition of HCAP is inappropriate and results in increased costs of care and increased risk of drug side effects and toxicity and contributes to selective pressure favoring the emergence of antibiotic resistance.

Primary viral pneumonia was identified in 5% of the overall cohort (5.7% of patients with CAP, 4.3% of patients with HCAP) and was the most commonly identified pneumonia etiology. Consistent with our findings, previous studies have shown that primary viral infections are a common cause of CAP (8, 10). Although primary viral HCAP has not typically been a focus of investigation, a recent study suggests patients admitted with nursing home-acquired pneumonia are at a high risk for primary viral pneumonia (9). The recognition of viruses as primary causes of CAP and HCAP may increase over time with the increased utilization of more-sensitive molecular assays (27). The identification of a primary viral pneumonia has implications for antimicrobial stewardship, given that patients with primary viral pneumonia could potentially have antibacterial treatment withheld and antiviral therapy initiated (e.g., influenza).

Mortality at 30 days occurred infrequently in our cohort, which is consistent with the low median CURB-65 score of 1. Only 16.3% of the cohort had a CURB-65 score of >2, a cutoff that would be associated with an anticipated mortality of greater than 3% (11). Previous studies have suggested an increased mortality rate with HCAP compared to CAP in univariate analysis (12, 15, 17, 23). In our study, the high mortality with HCAP was likely due to more comorbidities and higher admission disease severity. Chalmers et al. found that HCAP was not independently associated with mortality after adjustments for differences in baseline severity of illness, comorbidities, and particularly withholding aggressive therapy, which was frequent in the HCAP group (15).

Our study has some limitations. It is a retrospective study and is subject to misclassification bias. While this study was conducted at a single center in the United States and the epidemiologic findings may not apply elsewhere, they are useful in defining the “variability of bacteriology from one hospital to another” and accounting for local resistance patterns as suggested in the ATS/IDSA guidelines (3). Identified independent predictors for MDRO were based on a limited number of patients given the infrequency of isolating MDRO pathogens in our patient population. Respiratory cultures were obtained in 43% of patients, and a specific etiology of pneumonia was identified in only 19% of our cohort, so other potential etiologies of pneumonia may not have been identified.

Our study makes substantial contribution to the existing body of epidemiologic data on pneumonia in the United States and suggests that MDROs may be infrequent causes of community-onset pneumonia at some centers. Based upon our findings and those of others, it is unlikely that all patients with HCAP require empirical treatment for MRSA and P. aeruginosa as originally suggested (3, 12). More data are needed to better describe the etiology of community-onset pneumonia in the United States, specifically prospective, multicenter studies which include characterization of primary viral pneumonia and culture-negative pneumonia. In conclusion, MDRO prevalence in the local patient population and specific MDRO risk factors should be considered when selecting empirical therapy for community-onset pneumonia.

ACKNOWLEDGMENT

Footnotes

REFERENCES