What is the most common cause of community-acquired acute pneumonia?

• PDF
• Split View
• • Article contents
  • Figures & tables
  • Video
  • Audio
  • Supplementary Data

Community-acquired pneumonia (CAP) is commonly diagnosed and is a frequent reason for hospital admission. Despite advances in medical care and antimicrobial therapy, the mortality rate for admitted patients remains at ∼12% [1]. Because of delays in receiving results for etiological investigations, initial antibiotic choices are usually empirical. In general, recommendations for therapy are based on both knowledge of local etiology results and also the severity of the episode. Australian guidelines are distinctive in that suggestions for empirical antibiotic treatment are stratified according to the patient's Pneumonia Severity Index (PSI) classification (table 1) [2, 3].

In Australia, published data on the etiology of CAP are limited [4–7]. The most rigorous of these studies was published in 1989 and consisted of results for 106 patients at 1 hospital. The most common causes of CAP were Streptococcus pneumoniae (in 42% of cases), respiratory viruses (in 18%), Haemophilus influenzae (in 9%), Mycoplasma pneumoniae (in 8%), and gram-negative bacteria (in 8%) [4]. Some of the diagnostic tests used in this study are no longer in common use, and some criteria used for making diagnoses would no longer be accepted. The other studies also had major limitations [5–7]. This paucity of local data has led clinicians to rely on international studies, which may not necessarily be appropriate.

The Australian Community-Acquired Pneumonia Study (ACAPS) was a prospective, multicenter, observational study of immunocompetent adults with CAP. Its aim was to provide Australian data on CAP etiology, usefulness of severity assessment tools, and treatment outcomes. The etiology data from ACAPS forms the largest ever study of CAP etiology and is one of the few studies worldwide to assess CAP etiology using the full array of diagnostic methods, including bacterial cultures, urinary antigens, viral PCR, and acute- and convalescent-phase serologic tests [8–11].

Methods

Study design and setting. Patients were recruited from the emergency departments of 6 hospitals in 3 states in Australia. The hospitals involved were Austin Health (840 beds; Melbourne), The Alfred Hospital (350 beds; Melbourne), Monash Medical Centre (MMC; 637 beds; Melbourne), Princess Alexandra Hospital (780 beds; Brisbane), and Royal Perth Hospital (RPH; 750 beds; Perth), all of which are adult tertiary teaching hospitals; and West Gippsland Hospital (83 beds; Worragul), a regional hospital in rural Victoria. Recruitment occurred during a 28-month period for 4 sites (1 June 2004 to 30 September 2006—that is, 3 southern hemisphere winters) and during a 12-month period for 2 sites (MMC and RPH; 1 June 2004 to 1 June 2005).

Inclusion criteria for the CAP study were as follows: age, >18 years; chest radiograph within 24 h after hospital admission demonstrating features consistent with acute pneumonia; and at least 2 symptoms consistent with pneumonia (e.g., fever or hypothermia, rigors, sweats, new cough [with or without sputum], chest discomfort, or new-onset of dyspnea) [12]. Patients were excluded if they had been hospitalized within the preceding 14 days, were significantly immunosuppressed (because of HIV infection, use of >10 mg of prednisolone per day in the preceding month, use of other immunosuppressive agents, active treatment for cancer, or prior organ transplantation), had received parenteral antibiotics prior to obtainment of blood samples for culture, had aspiration pneumonitis, or had active treatment withdrawn within 12 h because of a poor prognosis. Patients were recruited within 48 h after hospital presentation. Consent was obtained at the time of recruitment. Human research ethics committee approval was obtained at all hospitals.

Blood specimens for culture—and, when possible, sputum specimens for Gram stain and culture—were obtained before receipt of parenteral antibiotics. Nose and throat swabs for respiratory virus PCR testing, standard nonconcentrated urine samples for pneumococcal and Legionella antigen testing, and acute-phase serum samples were collected at the time of recruitment within 48 h after hospital admission. At the time of patient follow-up (4–8 weeks after hospital discharge), convalescent-phase serum samples were obtained. Antibiotic therapy was given at the discretion of the treating clinician, although the hospitals involved were encouraging staff to follow the Australian guidelines (table 1) [2]. If patients did not attend follow-up appointments, they were telephoned by a study coordinator to assess 30-day mortality and cure and to arrange convalescent-phase serologic testing.

Microbiological studies. Aerobic and anaerobic blood cultures, as well as Gram stain and culture of sputum samples, were performed by conventional methods. PCR testing of the combined nose and throat swab samples was as described elsewhere for influenza A and B viruses, parainfluenza virus types 1–3, respiratory syncytial virus, adenoviruses, and picornaviruses [13]. Urine testing with the BinaxNOW pneumococcal urinary antigen test (Binax) and the BinaxNOW Legionella urinary antigen test (Binax) was performed in accordance with the manufacturer's instructions. Standard serologic test methods were used to detect antibodies to the following organisms: Legionella species (immunofluorescence [14]), Mycoplasma pneumoniae (total antibody by particle agglutination [Serodia, Fujirebio] and IgM by EIA [Savyon Diagnostics]), Chlamydophila (SeroELISA Chlamydia IgG; Savyon Diagnostics) to determine the genus, with or without determination of the species (Chlamydia SeroFIA IgG; Savyon Diagnostics), and influenza virus (complement fixation test [15]).

All relevant bacterial isolates, viral swabs, viral extracted DNA, serum specimens, and urine samples were stored. Viral and serologic tests that were performed in Brisbane and Perth were retested in Melbourne to ensure consistent results.

Classification of etiology. The etiology of CAP was determined to be definite if any of the following criteria were met: the organism was cultured from blood samples; urinary antigen tests were positive for S. pneumoniae or L. pneumophila; PCR of nose and throat swab samples yielded positive results; purulent sputum sample (presence of >25 polymorphonuclear leukocytes and < 10 squamous cells per low-power field), with a predominant organism cultured and compatible results from Gram stain; presence of IgM antibodies for M. pneumoniae; or 4-fold increases in IgG antibody titers for M. pneumoniae, L. pneumophila, C. pneumoniae, C. psittaci, or influenza virus. Diagnosis was defined as “presumptive” if purulent sputum sample cultured a predominant organism but without compatible Gram stain or if serologic test titers did not meet the aforementioned criteria but were suggestive of recent infection [11].

Outcome measures. The measured outcomes were etiology, antibiotic treatment received, time to “step down” from intravenous to oral antibiotics, time to clinical stability (defined as a temperature ⩽37.2°C and a respiratory rate ⩽24 breaths/min for 24 h) [16], the need to “step up” to broader-spectrum antibiotics, length of stay, the need for intensive respiratory or vasopressor support (IRVS; defined as the use of mechanical ventilation via endotracheal intubation, noninvasive ventilation, or receipt of infusion of vasoactive agents) in the intensive care unit (ICU), and in-hospital and 30-day mortality rates. The PSI and CURB-65 scores were calculated for all patients, as described elsewhere [3, 17]. On the basis of CURB-65 scores, patients were classified in the following groups: group 1, scores of 0–1; group 2, score of 2; and group 3, scores of 3–5 [17].

Statistical analysis. To detect differences between specified groups, either the χ2 test, Fisher's exact test, or the Mann-Whitney U test were used, as appropriate, with P values <.05 considered to be statistically significant. Statistical calculations were performed using Stata software (Intercooled Stata, version 9 for Windows; Stata Corporation).

Results

Patient characteristics. More than 2500 persons were assessed for possible study recruitment, with the major reason for exclusion being a chest radiograph finding that was either normal or inconsistent with CAP. Other reasons for exclusion were receipt of parenteral antibiotics prior to blood cultures, hospitalization within the preceding 2 weeks, a diagnosis of aspiration, receipt of palliative treatment, or lack of provided consent. Eight hundred sixty-five patients with a total of 885 episodes of CAP were recruited during the 28 months of the study, the majority of whom presented to 3 of the urban teaching hospitals. The demographic characteristics of patients are presented in table 2. The overall 30-day mortality rate was 5.6% (50 of 885 episodes). The mean age (±SD) was 65.1±19.9 years (range, 18–100 years).

Etiology. Etiology was identified in 404 (45.6%) of 885 episodes, and >1 pathogen was isolated in 75 episodes (8.5%). Results are shown in table 3. The most common organisms were S. pneumoniae (14% of episodes) and M. pneumoniae (9%), but together, the respiratory viruses (influenza virus, picornaviruses, respiratory syncytial virus, parainfluenza virus, and adenovirus) were identified in 15% of episodes. In only 48 (5.4%) of 885 episodes was a pathogen found that would not be adequately treated with the combination of a penicillin plus either doxycycline or a macrolide, and many of the affected patients had significant risk factors to alert the clinician to their unusual pathogen (see “Associations between specific etiologies and comorbidities or age” below).

For the 123 episodes of pneumococcal CAP, the diagnosis was made by urinary antigen data alone in 58 episodes, sputum data alone in 20, both blood culture and urinary antigen data in 17, sputum and urinary antigen data in 12, blood culture data alone in 8, and all 3 tests in 8. Of the 33 episodes of bacteremic pneumococcal infection, 2 (6.1%) involved bacteria with intermediate susceptibility to penicillin (MIC, 1 µg/mL). There were no penicillin-resistant isolates in blood cultures. Of the 40 patients for whom sputum cultures yielded pneumococcus, 8 (20.0%; 6 [17.1%] of 35 definite cases) had cases with reduced penicillin susceptibility. Among the 8 penicillin-nonsusceptible isolates recovered from sputum specimens, 7 had intermediate susceptibility to penicillin (MIC, 0.25–1 µg/mL), and 1 was resistant to penicillin (MIC, 2 µg/mL).

The causes of CAP, according to severity, are shown in table 4. Mycoplasma species was the most frequent pathogen in the less severe cases (i.e., those with PSI classes I and II), with pneumococcus being the most common in all other classes, as well as in those that needed IRVS. Viral etiology was found frequently in all classes.

Open in new tabDownload slide

Etiology by severity of community-acquired pneumonia (CAP) and in the subgroup that had not received prior antibiotics.

Of the 19 patients with legionellosis diagnosed by urinary antigen testing, 15 had both acute- and convalescent-phase serum samples obtained, and only 5 (33.3%) experienced seroconversion. Eight of the cases were associated with a point-source outbreak of infection at a shopping center close to one of the study sites. Overall, serologic testing demonstrated elevated but stable titers of antibodies (⩾1:128) against Legionella species in 118 (13.3%) of 885 episodes.

Copathogens. More than 1 pathogen was identified in 75 (8.5%) of 885 episodes, with 67 involving 2 pathogens and 8 involving 3 pathogens. The 30-day mortality rate for these episodes was not significantly higher than that for the remaining episodes (6 [8.0%] of 75 vs. 44 [5.4%] of 810; P=.36). Respiratory viruses and a bacterial copathogen were detected in 47 episodes (5.3%), whereas a respiratory virus alone was noted almost twice as frequently (91 episodes [10.3%]).

Associations between specific etiologies and comorbidities or age. Chronic obstructive pulmonary disease was the major comorbidity (12 of 16 episodes) for episodes in which aerobic, gram-negative bacteria (e.g., Pseudomonas species) were isolated. The majority of patients with pseudomonal infection were known to have been previously colonized.

Cultures yielded gram-negative enteric bacilli (GNEB; e.g., Escherichia coli, Klebsiella pneumoniae, and Proteus mirabilis) for patients from long-term care facilities or for those with extensive comorbidities. Eight patients had GNEB bacteremia, and for 4 of these 8 patients, urine cultures also yielded the same pathogen.

There was an association between legionellosis and a history of renal impairment (P=.04). Mycoplasma infection was associated with an age <50 years (P<.001).

For the 55 episodes involving patients from nursing homes, the identified pathogens were respiratory viruses in 9 (16.4%), pneumococcus in 8 (14.5%; 1 isolate had intermediate susceptibility to penicillin, and all others were susceptible to penicillin), GNEB in 4 (7.3%), and H. influenzae in 2 (3.6%). This cohort had a high 30-day mortality rate (16 [29.1%] of 55 episodes) but a low rate of IRVS requirement (3 episodes [5.5%]).

Outcomes associated with specific pathogens. A higher risk of IRVS requirement was associated with picornavirus infection (P=.01), but there were weaker associations with influenza (P=.05), pneumococcus infection (P=.06), and legionellosis (P=.09). The mortality rate was higher for CAP due to Pseudomonas species than for other causes of CAP (P<.001), although this likely reflects the very poor premorbid conditions of patients for whom CAP was due to this organism.

Utility of sputum examination and blood culture. A sputum sample was obtained in 524 (59.2%) of 885 episodes, and the CAP diagnosis was established using sputum analysis in 103 (19.7%) of these 524 episodes (i.e., 103 [11.6%] of all 885 study episodes). Blood cultures identified a pathogen in 54 (6.2%) of 868 episodes, with S. pneumoniae being the most frequent isolate in 33 (61.1%). There was a likely contaminant in the blood cultures for an additional 36 (4.1%) of 868 episodes, the most common contaminant being coagulase-negative staphylococcal species. Blood culture results were more likely to be positive in patients who had not received prior oral antibiotics (44 of 604 vs. 10 of 264; P=.05).

Treatment and outcome. Choice of antibiotic therapy was consistent with the recommendations of the Australian antibiotic guidelines in 729 episodes (82.4%). Simple β-lactam therapy (e.g., benzylpenicillin, amoxycillin, or ampicillin), in combination with either a macrolide or doxycycline, was given in 494 (55.8%) of 885 episodes (mean PSI score ± SD, 86±56). Ceftriaxone-based therapy was used in 326 (36.8%) of 885 episodes (mean PSI score ± SD, 112±41). Penicillin allergy was documented in 102 (31.3%) of 326 episodes treated with ceftriaxone. Moxifloxacin was administered to 7 patients, all of whom had penicillin allergy. Of the 10 patients who had pneumococcal infection with reduced susceptibility, 5 were successfully treated with penicillin or amoxycillin; this included 1 patient whose sputum data indicated that the infection was resistant to penicillin and 1 patient whose blood culture indicated intermediate susceptibility to penicillin. The other 5 patients received ceftriaxone.

The 30-day mortality rate among patients who received penicillin-based therapy was lower than that among patients who received ceftriaxone-based therapy (3.8% vs. 8.6%), although patients with higher mortality risk (i.e., pneumonia severity index class V or treatment in an ICU) were more likely to receive ceftriaxone. Unadjusted outcomes of therapy are shown in table 5. A comparison of antibiotics used and outcomes based on whether an etiologic agent was identified is presented in table 6.

Open in new tabDownload slide

Antibiotics used and outcomes among episodes with a known etiology versus those for which no etiology was found.

Discussion

This study is one of the largest prospective studies of the etiology of CAP ever performed. As expected, S. pneumoniae was the most common single organism, although it was identified at a rate lower than the 35%–75% rate that has been quoted in some previous studies [4, 18–20]. The decrease in relative frequency of identified pneumococcal cases of CAP has been described elsewhere, although the reasons for this decrease are not clear [21]; possibilities include pneumococcal vaccination programs, increased preadmission treatment with oral antibiotics, and a decrease in the quality of sputum testing by laboratory staff. However, the pneumococcal urinary antigen test may help detect patients with false-negative sputum or blood culture results [22].

An important finding was the relatively high frequency of PCR-detected respiratory viruses, both as sole pathogens and among all categories of CAP severity, including among patients who required IRVS. Also notable was the low number of cases (4% overall) caused by GNEB, Pseudomonas species, or S. aureus, even among patients with more severe CAP. Many of these episodes were potentially predictable on the basis of prior colonization or presence of extensive comorbidities.

Certain organisms were associated with more severe CAP: S. pneumoniae, influenza virus, and picornavirus were found more frequently in patients who required IRVS. Cases of legionellosis, although not common, generally involved L. pneumophila (83%); only 3 were cases due to Legionella longbeachae. Outcomes were not significantly worse for patients with legionellosis than for those with CAP due to other pathogens.

These etiology results suggest that broad-spectrum antibiotics are not necessary for the vast majority of Australian patients with CAP. Only 5.4% of episodes involved isolation of a pathogen with in vitro resistance to benzylpenicillin plus either doxycycline or a macrolide. The likely etiology of CAP was pneumococcal, viral, or “atypical,” even for the 94 patients who required IRVS, with only 7 cases (7.4%) due to more resistant organisms (e.g., Pseudomonas species, GNEB, or S. aureus). Our 30-day mortality data and rate of IRVS are comparable with those in international studies of hospitalized patients, even though the majority of patients in ACAPS received a narrow-spectrum penicillin combined with either doxycycline or a macrolide [3, 17, 23]. One-half of the patients who were infected with pneumococcal isolates with reduced penicillin susceptibility were cured with benzylpenicillin or amoxycillin, as has been reported by others [24]. The antibiotics used and outcomes were similar regardless of whether patients had a pathogen isolated.

Our finding that antibiotic-resistant pathogens were infrequent causes of CAP contrasts starkly with laboratory-based studies that have assessed resistance rates among consecutive bacterial isolates, as opposed to consecutive patients with CAP. Although our data suggest a low prevalence of pneumococcal resistance in Australia, a previous laboratory-based study found that >25% of consecutive pneumococcal isolates were classified as penicillin nonsusceptible [25]. In the absence of data from clinical studies similar to ours, it is possible that other nations may have also overestimated the likely rates of resistance among patients with CAP.

Numerous authors and international CAP management guidelines—including the most recent guidelines from the United States [26]—have promoted the use of “respiratory” fluoroquinolones (e.g., moxifloxacin). However, prospective studies comparing these agents with combination therapy containing a β-lactam and a macrolide have shown equivalence [27, 28]. Similarly, other studies that compared fluoroquinolones with a β-lactam plus the discretionary use of a macrolide have also shown equivalence [29–33], with only 1 exception [34]. The increased rate of resistance observed with the widespread use of these drugs is cause for concern, and unlike for benzylpenicillin, treatment failures among patients with pneumococcal CAP or bacteremia have been reported [35–41]. The use of fluoroquinolones has also been associated with infections due to clinically aggressive strains of Clostridium difficile [42] and extended-spectrum β-lactamase–producing gram-negative bacilli [43]. Thus, considering the availability of many other antibiotics with equivalent efficacy for treatment of CAP, it would seem to be prudent to avoid the widespread use of fluoroquinolones for this indication. Given the paucity of new antibiotics in development, any measures to prolong the useful lifespan of antibiotic classes, such as the fluoroquinolones, should be encouraged [44].

Our study has some limitations, including the fact that the majority of patients were inpatients recruited from 3 large, urban tertiary care referral centers. Therefore, we cannot be sure that our data can be generalized to other settings (e.g., primary care settings), to significantly immunosuppressed patients, or to patients with hospital-acquired pneumonia (who were excluded from ACAPS). Similar to other studies, we noted that, among patients who had recently received oral antibiotics, sputum specimens were of limited diagnostic utility—a finding that may explain why no etiology was identified in more than one-half of our patients [45]. Paired serum samples were not obtained from 24% of patients, and nose and throat swab samples for viral testing were not obtained from 13%. With regard to treatment outcomes, this was an observational study, with antibiotic choices made by the treating clinician; thus, there is no control group for comparison.

Despite these limitations, ours is the most significant study of CAP etiology in Australia, with the study population consisting of patients with CAP rather than sequential bacterial isolates. Strong supportive evidence for the use of narrow-spectrum β-lactam antibiotics, combined with either doxycycline or a macrolide, for treatment of CAP is provided by the low frequency of antibiotic-resistant pathogens and the good 30-day mortality rates and rate of IRVS requirement. Because CAP is such a frequent indication for antibiotic prescribing, greater use of such agents has the potential to have a significant impact on antibiotic selection pressure exerted by often more expensive and broader-spectrum drugs, such as fluoroquinolones.

The ACAPS Collaboration

Patrick Charles, Lindsay Grayson, Robert Pierce, Barrie Mayall, Paul Johnson, Michael Whitby, John Armstrong, Graeme Nimmo, Wendy Munckhof, David Looke, Luke Garske, Geoffrey Playford, Andrew Fuller, Robert Stirling, Denis Spelman, Tom Kotsimbos, Peter Holmes, Tony Korman, Philip Bardin, Grant Waterer, Keryn Christiansen, Christopher Heath, Alistair Wright, Michael Catton, Christopher Birch, Julian Druce, Norbert Ryan, Lou Irving, and David Hart.

Acknowledgments

We would like to acknowledge the research nurses (Barbara Johnson, Michelle Hooy, Kathy Bailey, Sue Colby, Claire Forsdyke, and Bernadette Dunlop), medical staff, and laboratory workers who assisted with the performance of ACAPS.

Financial support. 20ICC Research Fund and the Victorian Department of Human Services.

Potential conflicts of interest. All authors: no conflicts.

References

1

,  ,  , et al. 

Guidelines for the management of adults with community-acquired pneumonia: diagnosis, assessment of severity, antimicrobial therapy, and prevention

Am J Respir Crit Care Med

7

,  ,  , et al. 

Community-acquired pneumonia at the Alfred Hospital, Melbourne: a prospective study with particular reference to Chlamydia pneumoniae [abstract]

8

,  ,  ,  ,  ,  . 

Improved diagnosis of the etiology of community-acquired pneumonia with real-time polymerase chain reaction

10

,  ,  , et al. 

Prospective multicenter study of the causative organisms of community-acquired pneumonia in adults in Japan

11

,  ,  , et al. 

Etiology of community-acquired pneumonia in hospitalized patients in Chile: the increasing prevalence of respiratory viruses among classic pathogens

12

,  ,  ,  ,  ,  . 

Practice guidelines for the management of community-acquired pneumonia in adults

16

,  ,  , et al. 

Time to clinical stability in patients hospitalized with community-acquired pneumonia: implications for practice guidelines

17

,  ,  , et al. 

Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study

18

Research Committee of the British Thoracic Society and the Public Health Laboratory Service

Community-acquired pneumonia in adults in British hospitals in 1982–1983: a survey of aetiology, mortality, prognostic factors and outcome. The British Thoracic Society and the Public Health Laboratory Service

22

,  ,  , et al. 

Rapid diagnosis of bacteremic pneumococcal infections in adults by using the Binax NOW Streptococcus pneumoniae urinary antigen test: a prospective, controlled clinical evaluation

26

,  ,  , et al. 

Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults

27

,  ,  , et al. 

A multicenter, open-label, randomized comparison of levofloxacin and azithromycin plus ceftriaxone in hospitalized adults with moderate to severe community-acquired pneumonia

28

,  ,  ,  ,  ,  . 

Moxifloxacin monotherapy compared to amoxicillin-clavulanate plus roxithromycin for nonsevere community-acquired pneumonia in adults with risk factors

Eur J Clin Microbiol Infect Dis

29

,  ,  . 

Cost-effectiveness of gatifloxacin vs ceftriaxone with a macrolide for the treatment of community-acquired pneumonia

30

,  ,  ,  ,  ,  . 

Full-course oral levofloxacin for treatment of hospitalized patients with community-acquired pneumonia

Eur J Clin Microbiol Infect Dis

31

,  ,  ,  ,  ,  . 

Oral gemifloxacin versus sequential therapy with intravenous ceftriaxone/oral cefuroxime with or without a macrolide in the treatment of patients hospitalized with community-acquired pneumonia: a randomized, open-label, multicenter study of clinical efficacy and tolerability

33

,  ,  ,  ,  . 

Treatment with sequential intravenous or oral moxifloxacin was associated with faster clinical improvement than was standard therapy for hospitalized patients with community-acquired pneumonia who received initial parenteral therapy

34

,  ,  , et al. 

Randomized controlled trial of sequential intravenous (i.v.) and oral moxifloxacin compared with sequential i.v. and oral co-amoxiclav with or without clarithromycin in patients with community-acquired pneumonia requiring initial parenteral treatment

Antimicrob Agents Chemother

35

,  ,  ,  ,  ,  . 

Emergence of levofloxacin-resistant pneumococci in immunocompromised adults after therapy for community-acquired pneumonia

38

,  ,  ,  . 

Levofloxacin treatment failure in a patient with fluoroquinolone-resistant Streptococcus pneumoniae pneumonia

42

,  ,  , et al. 

A predominantly clonal multi-institutional outbreak of Clostridium difficile-associated diarrhea with high morbidity and mortality

45

,  ,  . 

Diagnostic value of microscopic examination of gram-stained sputum and sputum cultures in patients with bacteremic pneumococcal pneumonia

Major Articles