Health care-associated Staphylococcus aureus pneumonia

181 1Department of Medical Microbiology, Dalhousie University, Queen Elizabeth II Health Sciences Centre, Halifax, Nova Scotia; 2Provincial Laboratory for Public Health, Department of Laboratory Medicine & Pathology; 3Division of Infectious Diseases, Department of Internal Medicine, University of Alberta, Edmonton, Alberta Correspondence and reprints: Dr Duncan Webster, Department of Medical Microbiology, Dalhousie University, Queen Elizabeth II Health Sciences Centre, 5788 University Avenue, Halifax, Nova Scotia B3H 1V8. Telephone 902-473-7716, fax 902-473-2234, e-mail dvwebste@dal.ca Received for publication August 28, 2006. Accepted December 21, 2006

H ealth care-associated pneumonia (HCAP) is being increasingly recognized as a clinical entity with significant morbidity and higher mortality than traditional communityacquired pneumonia (CAP) (1).With unique epidemiological, clinical and bacteriological characteristics, HCAP should be considered in the differential diagnosis of individuals with recent health care contact who present to hospitals with signs and symptoms of lower respiratory tract infection.Although presenting from the community, relevant health care contact is considered to include long-term care facility (LTCF) residence, ambulatory clinic hemodialysis, intravenous chemotherapy or home intravenous therapy, specialized home nursing care and recent hospitalization in an acute care hospital.When an individual presents with health care contact of this nature, empirical therapy may be tailored to the probable causative pathogens of HCAP, including Staphylococcus aureus.The American Thoracic Society and the Infectious Diseases Society of America have recently published guidelines that address the management of HCAP (2).
Although generally uncommon as a cause of CAP, S aureus has been implicated as a predominant pathogen in HCAP (1).S aureus pneumonia has classically been described as a secondary bacterial infection in the setting of primary influenza virus upper respiratory tract infection (3)(4)(5)(6)(7)(8).However, S aureus has a long history as a cause of nosocomial pneumonia (9,10), and has been known for years to play a significant role in nursing home-acquired pneumonia (11).Thus, as health care expands further into the community with shorter inpatient stays and more expansive outpatient management, the predominance of HCAP, and thus S aureus pneumonia, can be expected to increase.
In the present observational study, a cohort of S aureus pneumonia cases presenting from the community were reviewed to determine the role of HCAP.Cases were identified from a larger prospective study of CAP cases in Edmonton, Alberta, over a two-year period and reviewed retrospectively.Cases were determined to be community acquired or health care associated.Epidemiological, clinical and microbiological data were analyzed.

PATIENTS AND METHODS
Patients older than 17 years of age who presented to hospitals in Edmonton, Alberta, between November 2000 and November 2002 were eligible for the study.Four teaching hospitals, two community hospitals and one freestanding emergency centre in Edmonton participated.Patients presenting with at least two of the following were managed according to a critical pathway for treatment of CAP: fever (body temperature higher than than 38°C), productive cough, chest pain, shortness of breath and crackles on auscultation, in addition to a chest radiograph interpreted as pneumonia (12).Pregnant or nursing women, neutropenic patients (white blood cell count less than 1.0×10 9 /L), critical care patients, cystic fibrosis patients and patients hospitalized within 48 h before presentation were ineligible for the treatment pathway.
As per the critical pathway protocol, patients who were admitted to hospital had blood cultures drawn, and those who were treated for pneumonia on an ambulatory basis had blood cultures performed at the discretion of the attending physician.Data collected on all patients included demographic variables, signs and symptoms at presentation, comorbidities and predisposing factors, laboratory values, microbiology results and diagnostic imaging, as well as course of treatment and outcome.A pneumonia severity index (PSI) score was calculated using a validated prognostic scoring model (13).Given the low sensitivity and specificity of sputum cultures, the causative organism was determined to be the definite source of infection only when isolated from blood or pleural fluid (14).To provide a strict analysis of patients with S aureus pneumonia, only patients with blood cultures positive for S aureus were included in the study cohort and underwent further chart and radiological review.

Clinical analysis
All cases were initially considered to be community acquired based on traditional definitions.However, on further review, a case was reclassified as health care associated if one of the following criteria was fulfilled: the patient received home intravenous therapy or specialized home nursing care within the 30 days before the infection; attended a hospital or hemodialysis clinic, or received intravenous chemotherapy within the 30 days before the infection; was hospitalized in an acute care hospital for a minimum of two days in the 30 days before infection; or resided in an LTCF.To determine the appropriate classification, information gathered by research nurses at the time of entry into the study was reviewed, and cases were retrospectively reviewed by one of the authors (DW) using inpatient hospital charts in every case and outpatient clinic charts as available.
On retrospective review of each case, the pathogenesis of the bacteremic pneumonia was also assessed.If S aureus infection was determined to have originated in the respiratory tract, it was defined as a primary S aureus pneumonia.This designation was based on clinical signs and symptoms of respiratory tract infection preceding or coinciding with the isolation of S aureus from blood, with conditions predisposing to primary pulmonary infection and without a further extrapulmonary source that may have predisposed to hematogenous seeding of the lungs.Infection determined to have originated outside of the respiratory tract with subsequent seeding of the lungs was defined as a secondary S aureus pneumonia.This classification was based on clinical evidence of an extrapulmonary source of S aureus infection, such as endocarditis, that appeared to precede the development of pulmonary infection, or radiographic evidence, such as multiple nodular cavitary lesions, that was suggestive of hematogenous spread to the lungs.
Demographic and clinical data of the CAP and HCAP groups were compared.Statistical analysis of means was performed using the unpaired t test with Welch correction.Nominal data were analyzed using Fisher's exact test.A two-tailed P value was calculated, and P<0.05 was considered significant.

Microbiological analysis
S aureus isolates were identified by routine laboratory procedures.Susceptibility assays were performed using VITEK instrumentation (bioMerieux Inc, USA).VITEK cards (GPS-105) for susceptibility assays were inoculated and incubated according to the manufacturer's recommendations (bioMerieux Inc).Methicillin-resistant S aureus (MRSA) was determined as per the Clinical and Laboratory Standards Institute guidelines (15) using an oxacillin screen plate assay.Any isolates exhibiting growth on the screen plate were further characterized by detection of the penicillin-binding protein 2' (PBP2') using the PBP2' latex agglutination test (Oxoid Ltd, United Kingdom) according to the manufacturer's instructions.All S aureus isolates exhibiting growth on the oxacillin screen plate and that were PBP2'-positive were considered to be MRSA.All isolates were  (16).A positive control was run concurrently with the test samples.
Molecular typing with pulsed-field gel electrophoresis (PFGE) was used to further characterize all the S aureus strains.The protocol was carried out as described by Mulvey et al (17).Analysis was performed using the Bio-Rad Gel Documentation System (Bio-Rad Laboratories, USA) and BioNumerics Software (Applied Maths, USA).A dendrogram was generated using the Dice coefficient with 1% tolerance.The molecular relatedness of strains was interpreted by PFGE analysis according to criteria described by Bannerman et al (18).Identical PFGE patterns were considered to be the same strain.Banding patterns with three or less band differences were considered to be subtypes.Banding patterns with more than three band differences were interpreted to be different strains.Epidemiological data were used to guide the interpretation of the molecular data.

Clinical analysis
Of 3043 patients who met the inclusion criteria for entry into the larger prospective CAP study between November 2000 and November 2002, 2008 patients (66%) had blood cultures performed, and 28 patients with bacteremic S aureus pneumonia were identified.Of these 28 patients, 23 were admitted on initial presentation, while five patients had blood cultures drawn in conjunction with initial ambulatory management.Twelve of the 28 patients (43%) remained classified as having CAP while 16 patients (57%) met criteria for HCAP.
The demographic features are presented in Table 1.Two distinct populations presented from the community with S aureus pneumonia.The CAP cohort was significantly younger than the HCAP cohort (mean age 49.0±23.7 years versus 67.8±18.6 years; P=0.035), with higher rates of smoking (67% versus 12%; P=0.005) and intravenous drug use (IVDU) (50% versus 0%; P=0.002).In contrast, the HCAP cohort was generally older, with higher rates of noninfectious chronic disease, such as diabetes mellitus (38% versus 0%; P=0.024).As expected with systemic S aureus infection, both cohorts tended to present with severe illness.Overall, 75% of the entire cohort presented with a risk class IV or V PSI score; however, the HCAP group presented with more severe illness overall and a significantly higher mean PSI score (143.1±41.1 versus 98.2±54.6;P=0.028) There was no trend in seasonal distribution and none in the two cohorts were reported to have had a preceding influenzalike illness; none were tested for influenza.The most common presenting symptoms overall (data not shown) included dyspnea (79%), fever (64%) and cough (61%).Common laboratory findings included anemia (85% had a hemoglobin level lower than 135 g/L), neutrophilic leukocytosis (64% had a white blood cell count higher than 11.0×10 9 /L, 78% had a neutrophil count higher than 7.5×10 9 /L) and renal dysfunction (50% had a creatinine level higher than 130 μmol/L).Chest radiographs revealed effusion in 54% of individuals and cavitation in 18% of inidividuals.
Complications and outcomes are shown in Table 2, and distinct differences between the CAP and HCAP cohorts are again noted, although none reached statistical significance.Those with community-acquired S aureus pneumonia were noted to have higher rates of complications generally associated with S aureus bacteremia.Complications included endocarditis (50% versus 19%; P=0.114), cerebral emboli (8% versus 0%; P=0.429), epidural abscesses (17% versus 0%; P=0.175) and septic arthritis (33% versus 6%; P=0.133).However, despite fewer embolic complications, the HCAP cohort suffered a higher mortality rate (31% versus 0%; P=0.052), with a trend toward statistical significance in this latter category.A shorter mean  length of stay among the HCAP cohort was due, in part, to this higher mortality rate, because death occurred within 48 h of admission in two HCAP patients.An assessment of the pathogenesis and summary of predisposing factors is provided in Table 3.In 11 of 12 cases (92%) of S aureus CAP, the pneumonia was determined to be secondary to hematogenous seeding of the lung parenchyma in the setting of a primary bacteremia.A break in skin integrity was the most common source of the primary bacteremia, with IVDU being the most frequent cause of the skin portal.A primary bloodstream infection with a secondary pneumonia was also common, although less so, among the HCAP cohort; this was observed in eight of 16 cases (50%).A break in skin integrity -frequently due to S aureus culturepositive decubitus ulcers, skin abrasions or infected intravascular catheters -was also the common source of the bacteremia among the HCAP cohort.LTCF residence was a prominent potential predisposing factor among the HCAP cohort, observed in seven of 16 (44%) cases.Readmission within 30 days of an acute care hospitalization was observed in four cases of HCAP (25%), although no single hospital predominated, because all four cases had been previously admitted to four different Edmonton hospitals.

Microbiological analysis
Susceptibilities were performed on 27 of 28 S aureus isolates because one of the isolates -isolate 12 shown in Figure 1 failed to grow on the VITEK.Of the 27 isolates for which results were available, all were susceptible to vancomycin and rifampin.Eighty-five per cent of the isolates were betalactamase positive and, thus, penicillin resistant.One isolate was resistant to trimethoprim-sulfamethoxazole.Two isolates were found to be methicillin resistant, one of which was also resistant to clindamycin.Of these two MRSA isolates, both were cultured from individuals with a history of IVDU.
Polymerase chain reaction revealed that all isolates were negative for the Panton-Valentine leukocidin gene.The PFGE patterns of the 28 study isolates are illustrated in Figure 1.Three isolates, numbers 4, 5 and 6, were interpreted as having indistinguishable band patterns.Dendrogram analysis with epidemiological information showed that identical isolates 5 and 6 along with subtype 8 were recovered from three elderly residents of three different LTCFs in Edmonton in 2002.The LTCFs were located in separate areas of the city, and no further epidemiological link was clearly identified in these cases.Isolate 4 was cultured from an individual living in the community and was classified as a CAP isolate.Isolates 3 and 7 were also subtypes within this group.Both were classified as CAP isolates, with the latter cultured in the setting of IVDU.
Isolates 20 and 21 were subtypes by PFGE pattern and were cultured in the setting of health care-associated infections less than one month apart in 2001.Both individuals were elderly with multiple comorbidities and were admitted to the same acute care hospital.Isolate 20 was cultured from the blood of an 89-year-old woman who was an LTCF resident with a left base infiltrate and endocarditis involving a porcine mitral valve.Isolate 21 was cultured from a 66-year-old man with Hodgkin's lymphoma and bilateral patchy infiltrates.He was a palliative patient and died within hours of his admission.
Isolates 26 and 27 were also subtypes cultured from individuals with health care-associated infections.These isolates were both cultured in 2002.The former isolate was cultured from the blood and sputum of a 71-year-old man with metastatic small cell lung cancer and recurrent admissions for pneumonia.The latter strain was cultured from the blood of a 59-year-old quadriplegic man with chronic pressure ulcers and a left upper lobe infiltrate.
The two MRSA isolates were numbers 13 and 23.Both were CAP isolates cultured in the setting of IVDU; however, PFGE analysis demonstrated no relatedness.Further assessment of the band patterns of these two isolates confirmed that the strains did not fall within a Canadian MRSA clonal prototype, or within the USA300 or USA400 prototypes.

DISCUSSION
In terms of epidemiology and pathogenesis, the clinical entity of S aureus pneumonia has evolved over time.A literature review, shown in Table 4, suggests some trends.Although a very heterogenous set of studies, the review suggested that the role of influenza has decreased in proportion to other risk factors over the past century, and has cycled in and out of the equation.Influenza remains an important predisposing factor; however, the contribution of other factors has grown.Table 4 shows the emergence of IVDU as an important risk factor for secondary S aureus pneumonia as highlighted by Julander et al (19), and MRSA as a major threat was prominent in a number of studies (20)(21)(22).S aureus pneumonia, as a significant nosocomial entity, is noted in Rebhan and Edwards' ( 23) 1960 review of 329 cases of staphylococcal pneumonia at The Hospital for Sick Children (Toronto, Ontario).In this 1950s cohort, 32% of the cases involved recent hospital contact and 3% of all cases occured in hospitalized premature infants.
With the significant role of the health care system and its extension into the community, S aureus has become a more frequent pathogen among cases of pneumonia presenting from the community.The American Thoracic Society and the Infectious Diseases Society of America have acknowledged HCAP as a unique and important clinical entity, with recognition of S aureus as a common causal pathogen (2).This designation is supported by our present study, which also found a significant number of S aureus pneumonia cases within a traditional CAP cohort that were found to be more precisely defined as HCAP.
This new designation also finds support in a recently published retrospective cohort study (1), which analyzed a large multi-institutional database of American acute care hospitals.In the study, Kollef et al (1) hypothesized that HCAP would involve pathogens more commonly associated with nosocomial pneumonia.This large study included 2221 patients with CAP, 988 patients with HCAP, 835 patients with hospital-acquired pneumonia and 499 patients with ventilator-associated pneumonia.The distribution of pathogens was somewhat unusual, in that S aureus was the most common pathogen in all four groups.It was noted in the related editorial that this may have been an artifact of the inclusion criteria (24).Nonetheless, the frequency of occurrence of S aureus in the HCAP group (46.7%) was comparable with the hospital-acquired pneumonia (47.1%) and the ventilator-associated pneumonia (42.5%) groups, and was significantly higher than the CAP group (25.5%).
Other investigators have also noted the relevance of distinguishing health care-associated infection from communityacquired infection.In a study of 504 bloodstream infections in North Carolina, USA, Friedman et al (25) found that 37% of bacteremic episodes that would normally have been classified as community-acquired, were more accurately identified as health care-associated.Similar to nosocomial infections, S aureus was prominent and in fact the most common pathogen.
A significant predisposing factor among the small cohort in our present study was the acquisition of health care in the community setting.Among the cohort, it was found that HCAP was more common than CAP, with the former identified in 16 cases (57%) and the latter in 12 cases (43%).
While one-half of the CAPs occurred in the setting of active IVDU, the most prominent predisposing factor in the older and sicker HCAP group was a break in skin integrity providing a portal of entry for bacteremia, which occured in 50% of cases; LTCF residence was a factor in 44% of cases.All five patients who died were from the HCAP group and three of the five were LTCF residents.Although the number of cases was small, a distinction in terms of predisposing factors, and perhaps outcomes, appeared to emerge, demonstrating the importance of distinguishing between these differing cohorts.
With epidemiological data to guide the interpretation of PFGE analysis, we were able to examine the relatedness of S aureus isolates in the present study.Within the community setting of the study, three identical isolates with related subtypes and two further subtype pairs were identified from the 28 isolates.Among these, seven were HCAP isolates.There is evidence of relatedness among 44% of the HCAP isolates and, therefore, modest evidence to suggest the spread of common strains.
Furthermore, additional evidence of relatedness may be observed through the application of the less stringent Tenover et al (26) rules.Although isolates 9 and 10 did not meet the predefined study criteria as subtypes, by the Tenover et al rules, they would be classified as 'possibly related' based on a four-to six-band difference, consistent with two genetic events.These isolates were cultured one month apart in 2001.Both involved individuals with chronic disease closely linked to the health care system.Isolate 9 was isolated from a 41-year-old woman with primary pulmonary hypertension, who developed bacteremic S aureus pneumonia in the setting of an infected intravascular Broviac catheter.Isolate 10 was cultured from a 45-year-old splenectomized man with hepatitis C and alcohol-related cirrhosis, who had been undergoing routine paracenteses when he developed a large left lower lobe pneumonia and S aureus bacteremia.Both had been receiving regular outpatient care at medical clinics within the same hospital.
The present study also provides some insight into the demographics of S aureus HCAP and is relevant in light of recently published guidelines for the management of HCAP (2).These guidelines recommend that HCAP be empirically treated as a multidrug-resistant infection with knowledge of local resistance patterns with further guidance by the principles of antibiotic stewardship and the targeting of modifiable risk factors.In our small study, MRSA accounted for 7% of cases.The need for empirical MRSA coverage with vancomycin, linezolid or an alternative may be best determined at a local level with periodic review.
However, given the exclusion criteria of the study, the findings may not be applicable to all populations, specifically pregnant women, critical care patients, individuals with cystic fibrosis or neutropenia, and those recently hospitalized.Prior studies and knowledge of these groups may provide some insight into the epidemiology of pneumonia in these specific settings.For example, observational studies of pneumonia in pregnancy with routine microbiological investigations have suggested that the range of causal pathogens is similar to those of the nonpregnant adult (27).It may be assumed that should a pregnant woman have exposures that would put her at risk for health care-associated infection, she would also be at increased risk for S aureus pneumonia.Although, admittedly, clear evidence from focused studies is lacking.
In terms of the critically ill, studies have found that S aureus is a more prominent etiological agent in the setting of severe CAP requiring admission to an intensive care unit compared with less severe CAP requiring hospitalization or ambulatory management (14).S aureus may also be more prominent as a cause of pneumonia presenting from the community among the other groups excluded from the study.Certainly, the airways of cystic fibrosis patients are commonly colonized by S aureus within the first few years of life, and these individuals are prone to developing respiratory tract infection due to this pathogen at rates higher than the general population (28).
The risk factors for the neutropenic group excluded from the study depend on the cause and duration of the neutropenia, although in general, this would certainly be a group predisposed to receiving frequent and specialized medical care.In this respect, selected cases would also be at increased risk for S aureus infection, although, akin to the cystic fibrosis patients, there would frequently be other opportunistic pathogens that need to be considered.Finally, patients hospitalized within 48 h of presentation were excluded from the current study, because the cohort was derived from a larger prospective CAP study.However, these cases would meet the definition of HCAP, and would be considered to be at risk for developing S aureus infections, including pneumonia.With regard to the role of influenza, which has been classically described as a risk factor for the development of S aureus pneumonia, it is difficult to comment on the role that this predisposing viral infection may have played in the present study.It was noted that there was no trend in seasonal distribution, and none in the cohorts were reported to have had a preceding influenza-like illness.Furthermore, preceding influenza infection was unlikely to have contributed to the development of S aureus pneumonia in the present study, because influenza A was not prominent in the region over the two-year study period, with only 314 confirmed cases in the Capital Health region (Edmonton, Alberta), which consists of over one million people.Yet, none of the cohort were tested for influenza, and thus, the role of this viral pathogen as a predisposing factor cannot be absolutely delineated in the present study.Note also that influenza would predispose a patient to a primary S aureus pneumonia rather than a secondary pneumonia via hematogenous spread.Primary pneumonia accounted for only a fraction of the total cases in the study.

CONCLUSION
In the present series of 28 cases of bacteremic S aureus pneumonia presenting to hospital from the community, health careassociated infection accounted for the majority of cases that, using traditional definitions, would have been classified as CAP.The HCAP cohort was significantly older, presented with a higher mean PSI score and showed a trend toward a higher mortality rate.LTCF residence and skin portals including decubitus ulcers and intravascular lines were important predisposing factors for the development of S aureus HCAP.The cohort with community-acquired infection was younger in age, and had higher rates of smoking and IVDU.
MRSA accounted for a small but significant percentage of these S aureus HCAP cases.There is evidence of relatedness among 44% of the HCAP isolates in the present study and modest evidence to suggest spread of common strains.S aureus is a predominant HCAP pathogen, and the present study supports distinguishing between HCAP and CAP, as well as the empirical coverage of S aureus among certain high-risk groups.
Webster et al Can J Infect Dis Med Microbiol Vol 18 No 3 May/June 2007 tested for the presence of the Panton-Valentine leukocidin gene by polymerase chain reaction using primers established by Lina et al Health care-associated Staphylococcus aureus pneumonia Can J Infect Dis Med Microbiol Vol 18 No 3 May/June 2007 183

TABLE 3
Potential predisposing factors and pathogenesis among 28 patients with Staphylococcus aureus pneumonia Can J Infect Dis Med Microbiol Vol 18 No 3 May/June 2007 184 *Immunosuppression is defined as hypogammaglobulinemia, CD4 lymphocyte count less than 200×10 6 /L, treatment with conventional immunosuppressives and corticosteroid use of oral prednisone ≥10 mg dailyFigure 1) Pulsed-field gel electrophoresis dendrogram analysis of Staphylococcus aureus isolates.CAP Community-acquired pneumonia; HCAP Health care-associated pneumonia; IVDU Intravenous drug use; LTCF Long-term care facility; MRSA Methicillin-resistant S aureus