American Journal of Respiratory and Critical Care Medicine

Microbiologic studies (MBSs) fail to identify a specific pathogen in more than 50% of patients with community-acquired pneumonia (CAP). The 1993 American Thoracic Society guideline (ATS-GL) for the management of CAP advised selecting initial antibiotic regimens based on severity of illness and comorbidities. Our study evaluated the role of initial MBS in adult patients hospitalized with CAP and treated according to the ATS-GL. In 184 patients hospitalized at our facility for CAP in 1996, and treated according to the ATS-GL, 25 (14%) failed to respond to initial antibiotic regimens. In these nonresponders, there was no difference in mortality between those in whom antibiotics were changed empirically, and those with MBS-guided changes. We conclude that initial MBS may not be warranted in many adult patients admitted for CAP. Exceptions include patients with conditions that predispose to less common, more resistant pathogens.

Community-acquired pneumonia (CAP) continues to be a major cause of morbidity and mortality in the United States despite the availability of potent antimicrobial therapies. There are 800,000 hospital admissions for CAP each year in this country (1), and pneumonia is the leading cause of death due to infectious disease (1, 2). The utility of initial microbiologic studies (MBSs) in the management of CAP remains unclear (3-5). Extensive diagnostic testing yields a specific microbiologic etiology in less than 50% of cases (6, 7).

An empirical approach to initial antibiotic therapy in CAP, based on severity and comorbidity, was developed by an expert panel of the American Thoracic Society (ATS) and published as a guideline (ATS-GL) in 1993 (8). The ATS-GL recommends blood cultures for all patients admitted with CAP, and sputum smear and culture in selected cases. In general, initial MBSs continue to be ordered routinely by practitioners caring for patients admitted with CAP. The purpose of our study was to examine the value of initial MBSs in adult patients who were hospitalized for CAP and managed according to the ATS-GL.

The Long Island College Hospital (Brooklyn, NY) is a 450-bed acute care facility. In 1995, a clinical guideline based on the ATS-GL (8) was introduced for the management of CAP. The medical records of all adult patients with CAP discharged in 1996 were reviewed. Patients with the conditions listed in Table 1 were excluded.


2. Malignancies
3. Sickle cell anemia
4. Neutropenia or other immunosuppressed states
5. Splenectomy
6. Intravenous drug abuse
7. Bronchiectasis
8. Discharge from acute care hospital within the preceding 12 wk
9. Not treated according to the American Thoracic Society Guideline, 1993.

The diagnosis of pneumonia was based on the presence of an acute lower respiratory infection with parenchymal densities on chest radiograph that were neither preexisting nor of any known cause. Severe CAP (SCAP) was defined according to the ATS-GL (8). The remaining patients were classified as having nonsevere CAP (NSCAP). Nonresponders were defined as patients with persistent or increasing fever and/or leukocytosis, or those showing marked clinical deterioration after the first 24 h, but within 5 d of hospitalization. Sputum specimens were considered adequate for analysis if they had more than 25 polymorphonuclear leukocytes and fewer than 10 epithelial cells per low-power field. These smears were considered positive when more than 10 of 20 oil immersion fields showed the same morphologic type of organism. Sputum cultures were considered positive if growth was moderate or large. Sputum samples that were received more than 24 h after admission were excluded from the study. Blood cultures were done using the Bactec 9240 system (Becton Dickinson, Sparks, MD). Blood culture data were analyzed only if at least two sets of cultures had been obtained before initiation of antibiotics. The following isolates were considered contaminants and were excluded from analysis: coagulase-negative staphylococci, Micrococcus spp., Streptococcus bovis, Propionibacterium spp., and diphtheroids. Statistical analyses were done using the Student t test, χ2 test, the Fisher exact test, and the binomial test as appropriate. The α level for rejecting the null hypothesis was equal to 0.05.

The study population consisted of 184 patients. Patient characteristics are shown in Table 2. Ages ranged from 18 to 99 yr, with a mean of 61.7 ± 21 yr. Patients with SCAP were significantly older than those with NSCAP, were more likely to have come from long-term care facilities (LTCFs), had a longer length of stay (LOS), and experienced a higher mortality.


NSCAPSCAPTotalp Value*
Study population13054184
Age, yr56.3 ± 20.774.7 ± 15.761.7 ± 21<0.001
LTCF before admission10 (8%)33 (61%)43 (23%)<0.01
Antibiotics before admission 19 (15%) 7 (13%)26 (14%)<0.9
Mean LOS, d5.9 ± 3.211.2 ± 8.27.5 ± 5.7<0.001
Nonresponders6 (4.6%)19 (35%)25 (13.6%)<0.0001
Mortality1 (0.8%)14 (26%)15 (8%)<0.0001

Definition of abbreviations: LOS = length of stay; LTCF = long-term care facility; NSCAP = Nonsevere community-acquired pneumonia; SCAP = severe community- acquired pneumonia.

*NSCAP versus SCAP.

The frequencies of positive microbiologic studies are shown in Table 3. Pathogens isolated from sputum and blood cultures are shown in Tables 4 and 5, respectively. Sputum specimens were collected from 127 patients and were considered adequate for evaluation in 116 patients (91%). Blood cultures were done in 174 patients (94.6%). Sputum cultures were positive in 40 patients (23% in NSCAP versus 67% in SCAP). Streptococcus pneumoniae was isolated in sputum cultures from five patients; these isolates were susceptible to penicillin. Nineteen patients had positive blood cultures (11% in both NSCAP and SCAP). Patients with bacteremia had greater mortality than nonbacteremic patients (21 versus 6.5%, p < 0.05). Streptococcus pneumoniae was by far the most common pathogen isolated from blood cultures (11 NSCAP, 3 SCAP). In all instances, the isolates were susceptible to penicillin.


NSCAPSCAPTotalp Value*
Sputum smear 38/86 (44%)22/30 (73%)60/116 (52%) <0.01
Sputum culture 20/86 (23%)20/30 (67%)40/116 (34%)<0.005
Blood culture13/121 (11%) 6/53 (11%)19/174 (11%) >0.9

Definition of abbreviations: NSCAP = nonsevere community-acquired pneumonia; SCAP = severe community-acquired pneumonia.

*NSCAP versus SCAP.


Streptococcus pneumoniae 23
Haemophilus influenzae 91
Staphylococcus aureus 44
Escherichia coli 13
Klebsiella pneumoniae 23
Pseudomonas aeruginosa 22
Proteus mirabilis 01
Group A streptococci01

Definition of abbreviations: MRSA = methicillin-resistant S. aureus; NSCAP = nonsevere community-acquired pneumonia; SCAP = severe community-acquired pneumonia.


Streptococcus pneumoniae 113
Haemophilus influenzae  20
Escherichia coli  01
Klebsiella pneumoniae  01

Definition of abbreviations: MRSA = methicillin-resistant S. aureus; NSCAP = nonsevere community-acquired pneumonia; SCAP = severe community-acquired pneumonia.

All patients with SCAP received erythromycin plus ceftazidime, or ticarcillin/clavulanate. If they were penicillin allergic they received aztreonam and erythromycin. In the NSCAP group, 99 (76%) received cefuroxime and erythromycin, 23 (18%) received only cefuroxime, 8 (6%) who had penicillin allergy received alternative regimens including erythromycin and trimethoprim–sulfamethoxazole (TMP + SMX). In those patients with NSCAP who were treated with cefuroxime alone there were two (8.6%) nonresponders both of whom had empiric antibiotic changes and no mortality. These frequencies were four (4%) and one (1%), respectively, in the group that received both a macrolide and β-lactam. The frequencies of nonresponse and mortality were not significantly different between patients with NSCAP who were receiving a macrolide and β-lactam, compared with patients receiving a macrolide alone.

In the total study group of 184 patients, 25 (14%) did not respond to initial regimens. Of these nonresponders, 6 had NSCAP and 19 had SCAP. The events defining nonresponse are shown in Table 6. None of the six nonresponding patients with NSCAP had a positive MBS. Empiric antibiotic changes were made in this group and one patient died of septic shock and respiratory failure. Of the 19 nonresponding patients with SCAP, 4 were admitted with shock and respiratory failure and died within 72 h. There were no antibiotic changes in these four patients. Three of the four patients had positive MBSs with pathogens that were sensitive to their antibiotic regimens. MBSs were negative in the fourth patient. Of the remaining 15 nonresponding patients with SCAP, 1 was continued on the initial treatment and eventually improved. The other 14 patients in this group had changes in their antibiotic regimens. MBSs were positive in 13 of these 14 patients. In 11 patients, the initial antibiotics were appropriate for the pathogens cultured, but changes were made empirically because of clinical deterioration. The remaining three patients had MBS-guided antibiotic changes. In two patients, gentamicin was added because sputum cultures showed Pseudomonas aeruginosa. In the other patient, sputum and blood cultures grew methicillin-resistant Staphylococcus aureus, and vancomycin was added to the regimen. These three patients with SCAP with MBS-guided antibiotic changes were from LTCFs. There was no difference in mortality between nonresponding SCAP patients with empiric antibiotic changes and those with MBS-guided changes (Table 7). There were no patients in whom antibiotic changes were based on the results of sputum smears.


Persistent fever and/or leukocytosis4/6 (67%) 9/19 (47%)
Respiratory failure1/6 (17%)14/19 (74%)
Sepsis syndrome2/6 (33%)15/19 (79%)
Empyema3/6 (50%)1/19 (5%)
Lung abscess01/19 (5%)

Definition of abbreviations: NSCAP = nonsevere community-acquired pneumonia; SCAP = severe community-acquired pneumonia.

* Note: Some patients had multiple events.


Number of PatientsMortality
Antibiotics changed per MBS 32 (67%)
Antibiotics changed empirically117 (64%)

Definition of abbreviation: MBS = microbiological studies.

Our retrospective study attempts to assess the value of an initial MBS for patients hospitalized with CAP and treated according to the ATS-GL. While sputum and blood culture positivity rates were comparable to those in previous reports (5, 9-11), we found that initial MBSs had little influence on therapeutic decisions and outcomes. Overall, failure to respond to initial antibiotic therapy occurred in only 14% of the entire study group, and changes in initial antibiotic regimens were made in 11% (four patients with SCAP died within 72 h without changes in antibiotic treatment). Sputum smears never influenced antibiotic choices, and culture results rarely did so. When antibiotics were changed for nonresponse, the changes were more often made empirically (85%) than on the basis of the results of MBS (15%). However, the mortality between these groups did not differ. It should be noted that because our study was retrospective, all antibiotic choices were made by the treating physicians. For patients included in the study, the choices for initial therapy were guided by our hospital protocol, which is based on the ATS-GL. However, changes in antibiotics for nonresponse were not protocol driven and were at the discretion of the physicians caring for those patients.

There have been several prior reports that addressed the utility of an MBS in CAP (5, 7, 8). Although their conclusions were in accord with ours, these studies did not specifically focus on patients managed according to the ATS-GL. Chalasani and co-workers (5), in a retrospective study of 517 patients with CAP, found that blood cultures did not have a significant impact in directing antibiotic changes. Similarly, a study of 93 patients admitted with severe CAP, of whom 69 had failed initial empiric antibiotics, showed little benefit from microbiologic studies (9). A prospective study by Pachon and colleagues (10), evaluating the impact of MBS on survival in 67 cases of severe CAP, also showed that the identification of the causative agents failed to improve mortality. Moine and co-workers (11), in another prospective study involving 132 patients with severe CAP, found mortality was the same in patients with and without a specific microbiologic diagnosis.

Certain categories of patients were excluded from our study and our conclusions cannot be extrapolated to those situations. Indeed, MBSs may have the greatest impact in these circumstances. Patients in these groups, such as those with human imunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS), bronchiectasis, and recent discharge from acute care hospitals, are more likely to be infected with antibiotic-resistant organisms. Patients from LTCFs may be an additional group that should have initial MBSs. Our data suggest that some of these patients may harbor resistant organisms. The three patients in our study who had MBS-guided antibiotic changes for nonresponse were from LTCFs.

In our institution the costs for initial MBSs in 1996–1997 were $180 for two sets of blood cultures, $30 for sputum Gram's stain, and $28 for sputum culture. Significant cost savings on a nation-wide basis might be realized if initial MBSs were ordered only for patients more likely to be infected with unusual or resistant organisms.

In conclusion, our results suggest that the ATS-GL provides a reasonable approach to initial antibiotic therapy for adult patients admitted to the hospital with CAP. Furthermore, MBSs may not be warranted at the outset for most patients. Exceptions include patients in the categories excluded from our study and certain other groups such as patients from LTCFs, who have a greater likelihood of harboring resistant pathogens.

1. Garibaldi, R. A. 1985. Epidemiology of community-acquired respiratory tract infections in adults: incidence, etiology and impact. Am. J. Med. 78(Suppl. 6B):32S–37S.
2. U.S. Department of Commerce. 1984. Bureau of the Census Statistical Abstract of the United States (104th Edition). U.S. Government Printing Office, Washington, DC.
3. Reed W. W., Byrd G. S., Gates R. H., Howard R. S., Weaver M. J.Sputum Gram's stain in community-acquired pneumococcal pneumonia—a meta-analysis. West. J. Med.16541996197204
4. Mandell L. A.Community acquired pneumonia: etiology, epidemiology and treatment. Chest108199535S42S
5. Chalasani N. P., Valdecans M. I., Gopal A. K., McGowan J. E., Jurado R. L.Clinical utility of blood cultures in adult patients with community-acquired pneumonia without underlying risks. Chest1081995932936
6. Bates J. H., Campbell G. D., Barron A. L., McCravken G. A., Morgan P. N., Moses E. B., Davis C. M.Microbial etiology of acute pneumonia in hospitalized patients. Chest101199210051012
7. Woodhead M. A., Arrowsmith J., Chamberlain-Webber R., Wooding S., Williams I.The value of routine microbial investigation in community-acquired pneumonia. Respir. Med.851991313317
8. Niederman M. S., Bass J. B., Campbell G. D., Fein A. M., Grossman R. F., Mandell L. A., Marne T. J., Sarosi G. A., Torres A., Yu V. L.Guidelines for the initial management of adults with community-acquired pneumonia, diagnosis, assessment of severity, and initial antimicrobial therapy. Am. Rev. Respir. Dis.148199314181423
9. Ewig S., Bauer T., Hasper E., Marklein G., Kubin I. R., Luderitz B.Value of routine microbial investigation in community-acquired pneumonia treated in a tertiary care center. Respiration631996164169
10. Pachon J., Prados M. D., Capote F., Cuello J. A., Garnacho J., Verono A.Severe community-acquired pneumonia, etiology, prognosis and treatment. Am. Rev. Respir. Dis.1421990369373
11. Moine P., Vercken J.-B., Chevret S., Chastang C., Gajdos P.Severe community-acquired pneumonia: etiology, epidemiology and prognosis factors. Chest105199414871495
Correspondence and requests for reprints should be addressed to Peter R. Smith, M.D., Division of Pulmonary Medicine, The Long Island College Hospital, 339 Hicks Street, Brooklyn, NY 11201.


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record