Key words: antimicrobial treatment; diagnosis; Gram stain; outcome; pneumonia; sputum

Applying Sputum as a Diagnostic Tool in Pneumonia* Limited Yield, Minimal Impact on Treatment Decisions Santiago Ewig, MD; Matthias Schlochtermeier, MD; Norbert Göke, MD; Michael S. Niederman, MD, FCCP Study objectives: We evaluated the role of sputum examination in the management of patients with community-acquired pneumonia (CAP) in a primary-care hospital without microbiologic laboratory facilities. Design and interventions: A diagnostic strategy using regular collection of sputum samples, Gram staining in a local laboratory, and mailing of samples to a commercial laboratory for culture analysis. Setting: A 200-bed primary-care hospital without subspeciality physicians. Patients: One hundred sixteen consecutive patients with a diagnosis of CAP were prospectively evaluated during a 12-month period. Results: Of 116 patients, 42 patients (36%) were capable of producing a sputum sample. Age > 75 years (odds ratio [OR], 0.4; 95% confidence interval [CI], 0.18 to 0.93) and prior ambulatory antimicrobial treatment (OR, 3.2; 95% CI, 1.2 to 8.4) were independent predictors of sputum production. A delay in collection and processing of sputum samples of > 24 h was present in 31% and 39%, respectively. A delay in collection yielded an increased number of Gram-negative enteric bacilli and nonfermenters (44% vs 7%, p 0.056). A delay in processing was associated with an increased number of Candida spp isolates (33% vs 9%, p 0.16). The overall diagnostic yield was low (10 of 116 patients, 9%) due to a limited number of valid samples (n 23 of 42 patients, 55%) and a limited number of definitely or probably positive samples on Gram s stain and culture (n 10 of 42 patients, 24%). Prior ambulatory antimicrobial treatment was associated with a reduction in diagnostic yield (14% vs 56%, p 0.09). The impact of diagnostic results on antimicrobial treatment decisions was minimal, with antimicrobial treatment directed to diagnostic results in only one patient. Conclusions: We conclude that in this setting representative of primary-care hospitals in Germany, sputum had a low diagnostic yield and did not contribute significantly to patient management. (CHEST 2002; 121:1486 1492) Key words: antimicrobial treatment; diagnosis; Gram stain; outcome; pneumonia; sputum Abbreviations: CAP community-acquired pneumonia; CI confidence interval; ERS European Respiratory Society; GNEB Gram-negative enteric bacilli; OR odds ratio *From the Medizinische Universitätsklinik und Poliklinik Bonn (Dr. Ewig), Bonn, Germany; Dreifaltigkeits-Krankenhaus Wesseling (Drs. Schlochtermeier and Göke), Wesseling, Germany; and Winthrop University Hospital (Dr. Niederman), Mineola, NY. Manuscript received February 22, 2001; revision accepted November 1, 2001. Correspondence to: Santiago Ewig, MD, Medizinische Universitätsklinik und Poliklinik Bonn, Innere Medizin/Kardiologie und Pneumologie, Sigmund-Freud-Strasse 25, 53105 Bonn, Germany; e-mail: santiago.ewig@ukb.uni-bonn.de Current guidelines for the management of adult patients with community-acquired pneumonia (CAP) generally agree in recommending an initial empiric antimicrobial treatment based on both general and local microbiology and susceptibility patterns. 1 7 However, the role of sputum as a rapid diagnostic tool that could direct antimicrobial treatment is a matter of controversy. While some authorities 1,5,8 11 have outlined important limitations of this tool in terms of sensitivity, reliability, and impact on treatment decisions, others 3,4,12 15 have insisted on placing sputum at the top of any management algorithm. The recent update of the American Thoracic Society guidelines 5 recommends obtaining sputum samples only if a drug-resistant pathogen or an organism not covered by usual empiric treatment is suspected. 1486 Clinical Investigations

Any standardized approach to patients with CAP must be adapted to local needs and limitations. Since the vast majority of patients with CAP in Germany are treated in primary-care hospitals, this is an important setting to validate any diagnostic approach. The value of collecting sputum samples is limited by the absence of a microbiology laboratory in most German primary-care hospitals. Therefore, in order to evaluate the role of sputum in the management of patients with CAP in this setting, we implemented a program of regular collection of sputum samples and Gram staining at the local, on-site laboratory as a diagnostic strategy. At the same time, we relied on the European Respiratory Society (ERS) guidelines adapted to current German susceptibility data for the selection of empiric antimicrobial treatment in the absence of a conclusive diagnostic result. 6 Study Hypothesis Materials and Methods The study was part of a follow-up interventional evaluation of the management of patients with CAP at the Dreifaltigkeits- Krankenhaus in Wesseling, a primary-care hospital with about 200 beds. 16 One part of the intervention consisted of evaluation of sputum samples as a diagnostic tool. The study hypothesis was that the sampling of sputum in this setting would result in a low diagnostic yield with a minimal impact on antimicrobial treatment decisions. The other part of the study included the implementation of the ERS guidelines for the selection of initial empiric antimicrobial treatment. 6 Setting The main characteristics of this hospital are the absence of any subspecialists within the Department of Internal Medicine, and the absence of a microbiology laboratory, resulting in the practice of mailing all diagnostic samples to a commercial laboratory for analysis. Patients Between June 1999 and May 2000, a total of 116 consecutive patients with a diagnosis of CAP were prospectively evaluated. The diagnosis of CAP was based on finding a new infiltrate on chest radiography, symptoms suggestive of a lower respiratory tract infection, and no alternative diagnosis made during followup. All patients were nonimmunosuppressed, and had not been hospitalized during the preceding month. Exclusion criteria were as follows: (1) the presence of severe immunosuppression (HIV infection, neutropenia 1 10 9 /L, and organ transplantation); (2) patients referred from another hospital after initiation of antimicrobial treatment; (3) patients with witnessed gross aspiration; and (4) patients with pulmonary tuberculosis. The severity of pneumonia was classified according to the pneumonia severity index of Fine et al. 17 All patients were intended to be treated according to the ERS guidelines with modifications made in accordance with current German susceptibility patterns. Interventions An attempt to collect a sputum sample was made in all patients as soon as possible. The procedure of expectoration was explained in detail by the attending nurse, and the sputum sample was collected in a sterile tube. The nurse also documented the reasons for a failure to produce sputum. Immediately after obtaining the sample, it was processed in the local hospital laboratory. The macroscopic appearance was graded as mucoid or purulent. Then it underwent Gram staining and read at low ( 100) magnification by the local hospital laboratory technicians. Thereafter, samples meeting microscopic validity criteria were mailed to a commercial laboratory for culture. Otherwise, samples were discarded. The regular mailing time was 24 h. Antimicrobial treatment was never delayed in order to obtain a sample from any patient. Definitons A sample was judged macroscopically valid when the appearance was purulent. Following the criteria of Lentino and Lucks, 9 stained specimens were then scored 0, 1, or 2 according to the number of leukocytes, and 0, 1, and 2 according to the number of squamous epithelial cells seen per low power field. Specimen with scores of 1 were considered microscopically valid. 9 Gram stain was graded positive in the presence of a predominant organism, negative in the absence of any microorganism, and mixed flora in the absence of a predominant microorganism. A sputum sample was considered as definitely diagnostic for a cultured bacterial isolate when the following criteria were met: (1) the sample was macroscopically purulent and microscopically valid, (2) the Gram stain result was positive, (3) the same microorganism grew in culture as was seen on Gram stain, and (4) there were moderate-to-high amounts of growth according to semiquantitative evaluation criteria used by the laboratory. The sample was considered probably diagnostic in the presence of mixed flora on Gram stain, while meeting all other criteria. Candida spp were not considered an etiologic pathogen. Delay in collection or processing of sputum was referred to as a delay in obtaining or processing a sample for 24 h. Nonresponse to antimicrobial treatment was defined as persistent temperature 38 C and/or persistent clinical symptoms (malaise, cough, expectoration, or dyspnea) or clinical deterioration (development of acute respiratory failure requiring ventilatory support and/or septic shock) after 72 h of in-hospital antimicrobial treatment. Measurements The following end points were recorded: the number of patients capable of producing sputum, reasons for failure to produce sputum, macroscopic and microscopic validity of sputum samples and the association of both criteria, and the number of definite and probably diagnostic samples. In addition, the influence of the presence of COPD, prior ambulatory antimicrobial treatment, and delays in sampling and laboratory processing of samples on diagnostic results was analyzed. Finally, the impact of diagnostic results on antimicrobial treatment decisions was investigated, ie, the rate of using a targeted antimicrobial treatment according to diagnostic results, instead of using an empiric regimen. This assessment was done retrospectively, correlating treatment decisions to diagnostic results. Statistics Results are expressed as mean SD. Continuous variables were compared by Student s t test, and categorical variables were www.chestjournal.org CHEST / 121 / 5/ MAY, 2002 1487

compared by 2 test or by Fisher Exact Test where appropriate. Predictors of productive cough were evaluated by a multivariate stepwise forward logistic regression analysis. The level of significance was set at p 0.05. Results General Patient Characteristics and Outcome The patient population included 72 men and 44 women (62% and 34%, respectively). Mean age was 68 16 years (range, 24 to 96 years), including 82 elderly patients (71%; age 65 years) of whom 50 patients (43%) were very elderly (age 75 years). Twenty-four patients (21%) were admitted to the hospital from nursing homes. At least one comorbidity (cardiac, pulmonary, hepatic, renal, CNS, diabetes mellitus, or neoplastic as previously defined 18 ) was present in 103 patients (89%). Using the classification according to the pneumonia severity score, there were 13 patients (11%) in risk class I, 7 patients in risk class II (6%), 11 patients in risk class III (10%), 48 patients in risk class IV (41%), and 37 patients in risk class V (32%). Initial antimicrobial treatment corresponded to the ERS guidelines in 85 cases (83%) and was inadequate treatment (according to guideline standards) in 17 cases (17%). Fourteen patients were not rated for adequacy of therapy because of decisions to limit treatment. Twenty-seven patients received monotherapy (aminopenicillin/ -lactamase-inhibitor [n 10], second-generation cephalosporin [n 5], macrolide [n 5], and levofloxacin [n 7]), and 89 patients received combination treatment (aminopenicillin/ -lactamase-inhibitor plus macrolide [n 71], aminopenicillin/ -lactamase-inhibitor plus levofloxacin [n 1], second-generation cephalosporin plus macrolide [n 9], and third-generation cephalosporin plus macrolide [n 8]). Twenty-two patients (19%) did not respond to initial antimicrobial treatment. Twenty patients (17%) were admitted at the ICU. Mortality rate was 15% (17 patients; 5 patients [10%] from risk class IV and 12 patients [32%] from risk class V). Sampling of Sputum Overall, 42 of 116 patients (36%) produced a sputum sample. Of these, 13 samples (31%) were not collected at hospital admission but 1 day after (n 10), 2 days after (n 2), and 3 days after (n 1). Reasons for failure to produce a sputum sample included unproductive cough efforts (n 31, 27%), weakness (n 23, 20%), and noncompliance (n 20, 17%). Age 75 years (odds ratio [OR], 0.45; 95% confidence interval [CI], 0.2 to 0.99; p 0.046), admission to the hospital from a nursing home (OR, 0.39; 95% CI, 0.13 to 1.1; p 0.078), prior ambulatory antimicrobial treatment (OR, 2.9; 95% CI, 1.1 to 7.3; p 0.024), and pulmonary comorbidity (OR, 2.0; 95% CI, 0.9 to 4.4; p 0.07) were the only variables predictive of sputum production at a level of significance of p 0.1. In particular, the pneumonia severity score was not associated with the probability of producing sputum (32% in pneumonia severity score classes 1 to 3 as compared to 38% in classes 4 and 5). In a multivariate analysis including the four variables with p 0.1, only age 75 years (OR, 0.4; 95% CI, 0.18 to 0.93; p 0.034) and prior ambulatory antimicrobial treatment (OR, 3.2; 95% CI, 1.2 to 8.4; p 0.019) remained as independent predictors of sputum production. Validity of Sputum Samples Of 42 samples, 35 samples (83%) were purulent according to macroscopic judgment. All 12 bacterial isolates cultured in moderate or high amounts in 10 patients were found in purulent sputum samples. Overall, macroscopic appearance was 100% (10 of 10 samples) sensitive and 22% (7 of 32 samples) specific for the prediction of growth in culture. Only 23 of 42 samples (55%) were considered microscopically valid as defined above. Macroscopic judgement predicted microscopic validity with a sensitivity of 91% (21 of 23 samples) and a specificity of 26% (5 of 19 samples). Gram Stains of Microscopically Valid Samples Gram stains were positive in 5 samples (22%) [with Gram-positive cocci in all cases] and negative in 1 sample (4%); the remaining 17 samples (74%) revealed a mixed flora. Sputum Cultures of Microbiologically Valid Samples Sputum samples were processed in the laboratory within the first 24 h in 11 cases (48%), within 48 h in another 10 cases (44%), and within 72 h and 96 h in 1 case each (4%). Overall, 16 pathogens (excluding 5 Candida spp) were cultured from 14 patients, including Streptococcus pneumoniae (n 4), Staphylococcus aureus (n 2), Haemophilus influenzae (n 4), Gram-negative enteric bacilli (GNEB; n 3), Pseudomonas aeruginosa (n 2), and Stenotrophomonas maltophilia (n 1). Twelve pathogens were present in moderate-to-high amounts. All four pathogens isolated in low amounts were either GNEB (n 2), P aeruginosa (n 1) or S maltophilia (n 1). However, of 23 samples, only 5 samples had a predominant organism on Gram stain; of these 5 samples, only 3 samples had concomitant growth in culture (2 1488 Clinical Investigations

samples with S pneumoniae and 1 sample with S aureus). The remaining two cases may have been false-negative results because of prior ambulatory antimicrobial treatment. Thus, applying strict criteria for the validity of a given isolate, overall 10 of 42 patients (24%) had a definite or probable diagnostic result in sputum samples (10 of 23 patients [44%] with valid sputum samples), with 12 pathogens isolated (Table 1). The number of valid specimen related to the diagnostic yield and the consequences for antimicrobial treatment decisions are summarized in Figure 1. Sputum investigation did not lead to reduction in the number of empiric antimicrobial agents in any case. Impact of Delays in Collecting and Processing of Sputum Samples Since empiric IV antimicrobial treatment was initiated as soon as possible after hospital admission, a delay in collecting a sputum sample always resulted in patients receiving therapy prior to sample collection (antimicrobial pretreatment). Thus, the number of isolates yielding GNEB and nonfermenters increased when pretreatment was done (4 of 9 isolates [44%] vs 1 of 14 isolates [7%]; OR, 10.4; 95% CI, 0.9 to 117.2; p 0.056). Three GNEB isolates were present in low quantities and were not diagnostic by definition. One significant isolate (Klebsiella oxytoca) was successfully treated with the empirical antimicrobial regimen (amoxicillin/clavulanic acid). A delay in processing was associated with an increase in the number of Candida spp isolates (4 of 12 isolates [33%] vs 1 of 11 isolates [9%]; OR, 5.0; 95% CI, 0.4 to 54.0; p 0.16). Isolates of GNEB and nonfermenters as well as Candida spp had both an 80% positive predictive value for delays in collecting and processing, respectively. Impact of COPD on Diagnostic Results In patients with a valid sputum sample, the presence of COPD was associated with a diagnostic yield twice as high as patients without COPD (7 of 12 Table 1 Overview of the Results of Microscopically Validated Sputum Samples in Relation to the Presence of COPD, Prior Ambulatory Antimicrobial Treatment, and Delays in Sampling and Processing* IDNR Macroscopic Validity Gram Stain Pathogens in Culture Bacterial Quantity COPD Prior ATB Delay in Sampling Delay in Processing 1 Purulent Mixed flora S pneumoniae No No No No H influenzae 2 Mucoid Mixed flora No Cef No No 3 Purulent Mixed flora P aeruginosa Yes No Yes No 4 Purulent Positive S pneumoniae Yes No No No 5 Purulent Mixed flora Yes Cef Yes Yes 6 Purulent Mixed flora No No No Yes 7 Purulent Positive S pneumoniae Yes No No Yes Candida spp 8 Purulent Mixed flora S pneumoniae No No Yes Yes K oxytoca 9 Purulent Mixed flora No No No Yes 10 Purulent Mixed flora Serratia spp Yes No No No 11 Purulent Mixed flora H influenzae Yes No No Yes 12 Purulent Positive No Pen Yes No 13 Purulent Mixed flora S maltophilia, No Tetra Yes No Candida spp 14 Purulent Mixed flora S aureus No No Yes No 15 Purulent Positive No Quino No No 16 Purulent Mixed flora H influenzae, Yes No No Yes Candida spp 17 Purulent Mixed flora H influenzae Yes No Yes Yes 18 Purulent Mixed flora Candida spp Yes Cef No Yes 19 Purulent Positive S aureus Yes Tetra No No 20 Mucoid Mixed flora Candida spp Yes No Yes Yes 21 Purulent Mixed flora Escherichia coli No No Yes No 22 Purulent Negative P aeruginosa Yes No No Yes 23 Purulent Mixed flora No No No Yes *IDNR patient identification number; ATB antimicrobial treatment; Cef oral cephalosporins; Pen oral penicillins; Tetra doxycycline; Quino oral quinolone. moderate-to-high growth on culture plates; low growth on culture plates. For definitions, see Methods section. www.chestjournal.org CHEST / 121 / 5/ MAY, 2002 1489

Figure 1. Diagnostic yield of sputum cultures in 116 patients with CAP (36% of the population studied). patients [58%] with COPD vs 3 of 11 patients [27%] without COPD; OR, 3.7; 95% CI, 0.65 to 21.6; p 0.13), but this difference did not reach significance. Impact of Ambulatory Antimicrobial Treatment on Diagnostic Results Prior ambulatory antimicrobial treatment was associated with a fourfold reduction in diagnostic yield, although this difference did not achieve significance (1 of 7 patients [14%] vs 9 of 16 patients [56%]; OR, 0.13; 95% CI, 0.13 to 1.3; p 0.06). Impact of Diagnostic Results on Antimicrobial Treatment Decisions Only one patient with a positive result in a valid sample (S aureus) who did not respond to initial empiric antimicrobial treatment received a targeted treatment according to culture results that differed from the ERS guideline-recommended empiric treatment. This directed treatment was successful. Discussion The present study demonstrates the limited value of sputum as a diagnostic tool in the initial evaluation of patients with CAP admitted to a primary-care hospital. The limitations include failure to obtain a sputum sample from most patients, delay in collection and laboratory processing of samples, low diagnostic yield, and minimal impact on therapeutic decisions. Of the 42 sputum samples collected from this population of 116 patients, only 23 samples (55%) were valid when standard microscopic criteria were applied, thereby substantially reducing the number of evaluable patients. Using these criteria, a macroscopic purulent appearance had a high false-positive rate of predicting validity. The same was true for predictions of subsequent growth in culture based on macroscopic appearance. These observations confirm and extend previous data. 8 Thus, it is hardly possible to replace microscopic validation by the use of more readily available macroscopic criteria. Only a minority of patients (36%) were able to produce a sputum sample. Failures were due to nonproductive cough, weakness, and noncompliance. An age 75 years was the most important predictor of a failure to produce sputum. The positive association between prior ambulatory antimicrobial treatment and ability to produce a sample may simply reflect the fact that general practitioners believed that patients with productive cough were more likely to have bacterial pneumonia and need antibiotic therapy. Our findings show a generally lower frequency of collecting sputum samples than in previous studies, which reported failure in 20 to 70% of patients, 19 22 with particularly low rates in the elderly. 23,24 Since the low rate of collecting a sample is unlikely to be increased by any intervention, sputum sampling is unable to be part of the initial microbial investigation in a considerable proportion of hospitalized patients with CAP. The delay in collection and laboratory processing of sputum samples was considerable, with 31% collected and 52% processed 24 h after hospital admission. These delays are mostly explained by weekends and nonworking days, and may represent a problem even in prospective studies. 25 Delays in collection resulted in antimicrobial pretreatment, thereby affecting the specificity of culture results, which demonstrated an increased recovery of GNEB and nonfermenters. Although we are not able to define exactly the impact of processing delays on the sensitivity of sputum cultures, it is known that common respiratory pathogens such as S pneumoniae and H influenzae are easily missed when samples are processed after 4 h. In addition, it was evident that delays in processing were associated with an increase in the isolation of Candida spp. Thus, the isolation of Candida spp in sputum samples of nonimmunosuppressed patients may be regarded as marker of overgrowth with colonizing organisms and not true pathogens. Gram stains had a low diagnostic yield, and a low 1490 Clinical Investigations

number of positive samples had a corresponding growth in culture. One explanation may be the lack of investigators who were familiar with the evaluation of Gram stains. In fact, the yield of Gram stains has proven to be highly dependent on a skilled investigator applying strict criteria. 26,27 In one study, 15% of the samples prepared by a house officer were technically inadequate, and the interpretations were often incorrect, with 50% of the pneumococcal readings being false-positive, and many of the H influenzae samples being false-negative. 27 However, if experienced laboratory technicians are not able to examine specimens, it is highly improbable that Gram stains can be accurately used in primarycare hospitals. In a study correlating Gram stain with the culture recovery of S pneumoniae from culture, it could be shown that applying liberal criteria such as any Gram-positive diplococcus resulted in a sensitivity of 83%, but with a specificity of only 38%. If the criterion for positive was a preponderance of Grampositive cocci, then the sensitivity dropped to 48% while specificity rose to 100%. 7 The updated American Thoracic Society guidelines advise that if a Gram stain is used to guide initial antimicrobial treatment, it should be with highly sensitive criteria, with the primary purpose being to visualize a microorganism that was not anticipated by the clinical scenario. 5 In our study, the yield of sputum was limited, even when liberal criteria were applied. It provided diagnostic results in only 10 of 116 patients (9%), 10 of 42 patients (24%) who produced sputum, and 10 of 23 patients (44%) with a microscopically valid sputum sample, respectively. The yield was somewhat higher in COPD patients, but since colonization with high bacterial loads may be present in up to 25%, it may be difficult to account for the effect of colonizing pathogens in the interpretation of sputum culture results. 28,29 Clearly, prior ambulatory antimicrobial treatment decreased the diagnostic yield, but in view of data 30,31 suggesting that any delay in treatment may be harmful in terms of outcome, it is not appropriate to discourage ambulatory treatment in order to increase diagnostic yield. The impact of sputum sample data on therapeutic decisions was negligible. Only one patient had a benefit from the identification of the underlying pathogen in sputum. This patient might have been diagnosed equally if he was reinvestigated after the failure of the initial antimicrobial treatment. Similar results have also been reported when evaluating the value of routine microbial investigation in different hospital settings. 32 35 In these studies, sputum smears and cultures added very little to the management and outcome of patients who received an appropriate initial antimicrobial regimen. In patients with treatment failures, mortality was not different, regardless of whether treatment was changed empirically or changed on the basis of diagnostic tests. We are unsure to what extent the results of the present study can be applied to other hospital settings. In fact, it has been repeatedly demonstrated that higher diagnostic yields can be achieved in different hospital settings and patient populations. 18,20 22,25 However, even in the most favorable report, originating from an experienced, university tertiary-care hospital, valid sputum samples could only be obtained in 39% of patients, with a predominant microorganism on Gram stain in 31% of patients. Moreover, the final difference in broad-spectrum vs pathogen-directed antimicrobial treatment, in the groups with and without diagnostic Gram stain, was only 14%. 36 Thus, the impact of diagnostic testing on therapeutic decisions also seems limited in settings interested in the study of pneumonia. However, data 37 highlighting an improved outcome for patients treated with a cephalosporin/macrolide combination or a quinolone, compared to a cephalosporin monotherapy regimen, raise serious doubts about the value of using focused antimicrobial treatment according to sputum findings. We believe that our results, reflecting implementation of a diagnostic strategy in a primary-care hospital not particularly interested in the field of respiratory infections, are representative of comparable hospitals in Germany. Moreover, our findings demonstrate that the yield and impact of sputum samples should be evaluated in each particular setting in order to establish a management approach for CAP that is designed according to local capabilities and facilities. As a result, we conclude that sputum collection cannot be recommended as part of a management algorithm in German patients with CAP treated in a primary-care hospital that does not have on-site microbiology facilities. References 1 American Thoracic Society. Guidelines for the initial management of adults with community-acquired pneumonia: diagnosis, assessment of severity, and initial antimicrobial therapy. Am Rev Respir Dis 1993; 148:1418 1426 2 British Thoracic Society. Guidelines for the management of community-acquired pneumonia in adults admitted to the hospital. Br J Hosp Med 1993; 49:346 350 3 Bartlett JG, Dowell SF, Mandell LA, et al. Practice guidelines for the management of community-acquired pneumonia in adults. Clin Infect Dis 2000; 31:347 382 4 Mandell LA, Marrie TJ, Grossman RF, et al. Canadian guidelines for the initial management of community-acquired pneumonia: an evidence-based update by the Canadian Infectious Diseases Society and the Canadian Thoracic Society. Clin Infect Dis 2000; 31:385 421 5 American Thoracic Society. Guidelines for the management www.chestjournal.org CHEST / 121 / 5/ MAY, 2002 1491

of adults with community-acquired pneumonia: diagnosis, assessment of severity, antimicrobial therapy, and prevention. Am J Respir Crit Care Med 2001; 163:1730 1754 6 European Study on Community-Acquired Pneumonia (ESO- CAP) Committee: guidelines for management of adult community-acquired lower respiratory tract infections. Eur Respir J 1998; 11:986 991 7 Deutsche Gesellschaft für Pneumologie: Empfehlungen zur Therapie der ambulant erworbenen Pneumonie. Pneumologie 1998; 52:450 462 8 Rein MF, Gwaltney JM Jr, O Brien WM, et al. Accuracy of Gram s stain in identifying pneumococci in sputum. JAMA 1978; 239:2671 2673 9 Lentino JR, Lucks DA. Nonvalue of sputum culture in the management of lower respiratory tract infections. J Clin Microbiol 1987; 25:758 762 10 MacDonald KS, Scriver SR, Low DE. Community-acquired pneumonia: the future of the microbiology laboratory; focused diagnosis or syndromic management? Semin Respir Infect 1994; 9:180 188 11 San Pedro GS, Campbell GD Jr. Limitations of diagnostic testing in the initial management of patients with communityacquired pneumonia. Semin Respir Infect 1997; 12:300 307 12 Gleckman R, DeVita J, Hibert D, et al. Sputum Gram stain assessment in community-acquired bacteremic pneumonia. J Clin Microbiol 1988; 26:846 849 13 Bartlett JG, Mundy MA. Community-acquired pneumonia. N Engl J Med 1995; 333:1618 1624 14 Skerrett SJ. Diagnostic testing to establish a microbial cause is helpful in the management of community-acquired pneumonia. Semin Respir Infect 1997; 12:308 321 15 Dorca J, Manresa F. Community-acquired pneumonia: initial management and empirical treatment. Eur Respir Mon 1997; 2:36 55 16 Ewig S, Seifert K, Kleinfeld T, et al. Management of patients with community-acquired pneumonia in a primary care hospital: a critical evaluation. Respir Med 2000; 94:556 563 17 Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997; 336:243 250 18 Ruiz M, Ewig S, Marcos MA, et al. Etiology of communityacquired pneumonia: impact of age, comorbidity, and severity. Am J Respir Crit Care Med 1999; 160:397 405 19 Marrie TJ, Durant H, Yates L. Community acquired pneumonia requiring hospitalisation: 5-year prospective study. Rev Infect Dis 1989; 11:586 599 20 Karalus NC, Cursons RT, Leng RA, et al. Communityacquired pneumonia: aetiology and prognostic index evaluation. Thorax 1991; 46:413 418 21 Blanquer J, Blanquer R, Borras R, et al. Aetiology of community acquired pneumonia in Valencia, Spain: a multicentre prospective study. Thorax 1991; 46:508 511 22 Neill AM, Martin IR, Weir R, et al. Community-acquired pneumonia: aetiology and usefulness of severity criteria on admission. Thorax 51:1010 1016 23 Marrie TJ, Haldane EV, Faulkner R, et al. Community acquired pneumonia requiring hospitalization: is it different in the elderly? J Am Geriatr Soc 1985; 33:671 680 24 Riquelme R, Torres A, El-Ebiary M, et al. Communityacquired pneumonia in the elderly: a multivariate analysis of risk and prognosis. Am J Respir Crit Care Med 1996; 154:1450 1455 25 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. QJM 1987; 239:195 220 26 Fine MJ, Orloff JJ, Rihs JD, et al. Evaluation of housestaff physicians preparation and interpretation of sputum Gram stains for community-acquired pneumonia. J Gen Intern Med 1991; 6:189 198 27 Reed WW, Byrd GS, Gates RH Jr, et al. Sputum Gram s stain in community-acquired pneumococcal pneumonia: a metaanalysis. West J Med 1996; 165:197 204 28 Cabello H, Torres A, Celis R, et al. Distal airway bacterial colonisation in healthy subjects and chronic lung diseases: a bronchoscopic study. Eur Respir J 1997; 10:1137 1144 29 Monsó E, Ruiz J, Rosell A, et al. Bacterial infection in chronic obstructive pulmonary disease: a study of stable and exacerbated outpatients using the protected specimen brush. Am J Respir Crit Care Med 1995; 152:1316 1320 30 Meehan TP, Fine MJ, Krumholz HM, et al. Quality of care, process, and outcomes in elderly patients with pneumonia. JAMA 1997; 278:2080 2084 31 Ruiz M, Ewig S, Torres A, et al. Severe community-acquired pneumonia: risk factors and follow-up epidemiology. Am J Respir Crit Care Med 1999; 160:923 929 32 Woodhead MA, Arrowsmith J, Chamberlain-Webber R, et al. The value of routine microbial investigation in communityacquired pneumonia. Respir Med 1991; 85:313 317 33 Ewig S, Bauer T, Hasper E, et al. Value of routine microbial investigation in community acquired pneumonia treated in a tertiary care center. Respiration 1996; 63:164 169 34 Sanyal S, Smith PR, Saha AC, et al. Initial microbiologic studies did not affect outcome in adults hospitalized with community-acquired pneumonia. Am J Respir Crit Care Med 1999; 160:346 348 35 Theerthakarai R, El-Halees W, Ismail M, et al. Nonvalue of the initial microbiological studies in the management of nonsevere community-acquired pneumonia. Chest 2001; 119: 181 184 36 Roson B, Carratala J, Verdaguer R, et al. Prospective study of the usefulness of sputum Gram stain in an approach to community-acquired pneumonia requiring hospitalization. Clin Infect Dis 2000; 31:869 874 37 Gleason PP, Meehan TP, Fine JM, et al. Associations between initial antimicrobial therapy and medical outcomes for hospitalized elderly patients with pneumonia. Arch Intern Med 1999; 159:2562 2572 1492 Clinical Investigations