Skip Navigation

A Prediction Rule for Estimating the Risk of Bacteremia in Patients with Community-Acquired Pneumonia

  1. Miquel Falguera1,
  2. Javier Trujillano2,
  3. Sílvia Caro1,
  4. Rosario Menéndez5,
  5. Jordi Carratalà6,
  6. Agustín Ruiz-González1,
  7. Manuel Vilà3,
  8. Mercè García4,
  9. José Manuel Porcel1,
  10. Antoni Torres7, and
  11. NAC-CALIDAD (Proyecto Integrado de Investigación de la Sociedad Española de Patología del Aparato Respiratorio sobre Infecciones Respiratorias de Vías Bajas) Study Groupa
  1. 1Services of Internal Medicine, Microbiology, Hospital Universitari Arnau de Vilanova, Universitat de Lleida, Institut de Recerca Biomèdia de Lleida, Ciber de Enfermedades Respiratorias (Ciberes 06/06/0028), Lleida, Barcelona, Spain
  2. 2Services of Critical Care, Microbiology, Hospital Universitari Arnau de Vilanova, Universitat de Lleida, Institut de Recerca Biomèdia de Lleida, Ciber de Enfermedades Respiratorias (Ciberes 06/06/0028), Lleida, Barcelona, Spain
  3. 3Services of Pneumology, and Microbiology, Hospital Universitari Arnau de Vilanova, Universitat de Lleida, Institut de Recerca Biomèdia de Lleida, Ciber de Enfermedades Respiratorias (Ciberes 06/06/0028), Lleida, Barcelona, Spain
  4. 4Services of Microbiology, Hospital Universitari Arnau de Vilanova, Universitat de Lleida, Institut de Recerca Biomèdia de Lleida, Ciber de Enfermedades Respiratorias (Ciberes 06/06/0028), Lleida, Barcelona, Spain
  5. 5Services of Pneumology Service, Hospital Universitari La Fe, Ciber de Enfermedades Respiratorias (Ciberes 06/06/0028), Valencia, Barcelona, Spain
  6. 6Services of Infectious Diseases Service, Hospital Universitari de Bellvitge, Universitat de Barcelona, Institut d'Investigació Biomèdica de Bellvitge, Barcelona, Spain
  7. 7Services of Pneumology Department, Clinic Institute of Thorax, Hospital Clinic of Barcelona, Universitat de Barcelona, Institut d'Investigacions Biomèdiques August Pi I Sunyer, Ciber de Enfermedades Respiratorias (Ciberes 06/06/0028), Barcelona, Spain
  1. Reprints or correspondence: Dr Miquel Falguera, Service of Internal Medicine, Hospital Universitari Arnau de Vilanova, Rovira Roure 80, 25198 Lleida, Spain (mfalguera{at}comll.cat).

  • ↵a Study group members are listed at the end of the text.

 
Next Section

Abstract

Background. We endeavored to construct a simple score based entirely on epidemiological and clinical variables that would stratify patients who require hospital admission because of community-acquired pneumonia into groups with a low or high risk of developing bacteremia.

Methods. Derivation and internal validation cohorts were obtained by retrospective analysis of a database that included 3116 consecutive patients with community-acquired pneumonia from 2 university hospitals. Potential predictive factors were determined by means of a multivariate logistic regression equation applied to a cohort consisting of 60% of the patients. Points were assigned to significant parameters to generate the score. It was then internally validated with the remaining 40% of patients and was externally validated using an independent multicenter cohort of 1369 patients.

Results. The overall rates of bacteremia were 12%-16% in the cohorts. The clinical probability estimate of developing bacteremia was based on 6 variables: liver disease, pleuritic pain, tachycardia, tachypnea, systolic hypotension, and absence of prior antibiotic treatment. For the score, 1 point was assigned to each predictive factor. In the derivation cohort, a cutoff score of 2 best identified the risk of bacteremia. In the validation cohorts, rates of bacteremia were <8% for patients with a score ⩽1 (43%-49% of patients), whereas blood culture results were positive in 14%-63% of cases for patients with a score ⩾2.

Conclusions. This clinical score, based on readily available and objective variables, provides a useful tool to predict bacteremia. The score has been internally and externally validated and may be useful to guide diagnostic decisions for community-acquired pneumonia.

Knowledge of both the identity of the causative pathogen and the susceptibility of the pathogen to antimicrobials is essential in the management of patients with infectious diseases. Unfortunately, with currently available diagnostic techniques this information is not available for many patients with community-acquired pneumonia (CAP). Most subjects do not produce a sputum sample of good quality for Gram staining, and the results obtained for sputum culture lack specificity. Additionally, samples of pleural fluid usually cannot be obtained, and the usefulness of urine antigen-detection tests and serum serological methods is restricted to the identification of specific agents, leaving out the most aggressive and drug-resistant pathogens [1].

In this context, blood cultures could be an important tool in the search for the etiology of CAP; in fact, they have become one of the most used diagnostic methods [1–3]. Blood culture is a polyvalent diagnostic test that is simple to obtain and relatively inexpensive. Moreover, the specificity of the test is high if skin contaminants are excluded, and it allows for the recognition of antimicrobial resistance. However, the yield of the procedure is low, with sensitivity levels of ∼10%-15%. Consequently, many investigators have questioned the cost-effectiveness of the technique [4, 5].

Previous reports have suggested that the risk of bacteremia can be influenced by some clinical factors. As expected, it has been recognized that prior antibiotic therapy reduces the sensitivity of culture [1]. There is also evidence suggesting a relationship between bacteremia and some epidemiological or clinical factors, such as immunosuppression, septic shock, and a severe clinical course [6, 7]. Finally, a previous retrospective study demonstrated an association between bacteremia and liver disease, 3 vital signs, and 3 laboratory abnormalities [8].

The purpose of the present study was to construct and validate a scoring method based on simple epidemiological and clinical data available at the bedside for estimating the risk of bacteremia in an unselected population of immunocompetent adult patients with CAP.

Previous SectionNext Section

Methods

Derivation and Internal Validation Step

Selection of patients. Patients included in this study were identified from a common database available at 2 acute-care university hospitals in Catalonia, Spain—the Hospital de Bellvitge in Barcelona and the Hospital Arnau de Vilanova in Lleida. The study was approved by the scientific and ethics committees of these institutions.

At both centers, data on all consecutive immunocompetent patients who were ⩾18 years old and hospitalized with CAP were prospectively collected over a 10-year period (January 1996 through December 2005). CAP was defined as an acute illness with features of a lower respiratory tract infection, the presence of an infiltrate on the chest radiograph consistent with acute infection, and 2 or more of the following clinical parameters: fever, chills, cough, sputum production, pleuritic pain, and signs of lung consolidation.

Exclusion criteria were tuberculosis or infection caused by fungi or opportunistic microorganisms; immunosuppression, including that caused by human immunodeficiency virus infection, hematologic neoplasms, solid-organ and bone-marrow transplantation, neutropenia, and treatment with immunosuppressive drugs or corticosteroids at a daily dose of ⩾20 mg of prednisolone or the equivalent; absence of blood cultures performed before the start of treatment in the hospital; and nosocomial or health care-associated acquisition of pneumonia.

Data collection. A unified protocol was established for the collection of data and the definition of the variables. The database contained >80 parameters related to the baseline characteristics of patients, microbiological results, therapeutic regimens, and outcomes. However, to derive and validate the present clinical rule we used only epidemiological and clinical parameters at entry. The results of the blood analysis were used to categorize the severity of illness at admission, according to the pneumonia severity index (PSI) [9]. The following variables were evaluated as potential predictive factors: epidemiologic data (age, sex, smoking habit, alcohol abuse, prior antibiotic therapy, chronic obstructive pulmonary disease, diabetes mellitus, chronic heart failure, neoplasm, chronic renal insufficiency, chronic liver disease, and cerebrovascular disease with significant sequelae); clinical findings (acute onset of disease, duration of clinical picture, chills, cough, expectoration, pleuritic pain, signs of pulmonary consolidation, and acute mental confusion); vital signs measured at the time of blood collection (body temperature, respiratory rate, heart rate, arterial systolic and diastolic blood pressure, and oxygen saturation measured by pulse oximetry with the patient not receiving supplemental oxygen); radiological features (pleural effusion and extent of pulmonary infiltrate [lobar vs multilobar]); and PSI.

Blood collection and culture. Blood was obtained and cultured using routine techniques. Briefly, after the diagnosis of CAP was established and before treatment, 10 mL of blood was obtained and inoculated into 2 culture bottles (5 mL was inoculated into an aerobic bottle and 5 mL into an anaerobic bottle). Thirty minutes later, an additional 10 mL of blood was obtained from a different site, and the procedure was repeated. Blood culture bottles were conventionally processed in an automated blood culture system. The following isolates were always considered to be contaminants: coagulase-negative staphylococci, Micrococcus species, Propionibacterium species, and diphteroids.

Score derivation. Data were analyzed using a commercially available software package (SPSS for Windows, version 12.0; SPSS). Patients were randomized into 2 groups to create a derivation cohort with the 60% of the samples and an internal validation cohort with the remaining 40%. We evaluated all the selected epidemiological and clinical variables by performing univariate analysis to determine the significance and strength of the association between each candidate predictor and bacteremia. We assessed significance by the χ2or Fisher exact test for categorical variables and the Student t test for continuous variables. Two-tailed P<.05 was considered to indicate statistical significance.

Multivariate analysis was performed to determine factors that were independently associated with bacteremia. For these analyses, we selected all potential predictive factors, including all variables found to be significant in the univariate analysis, and entered them into a final model using stepwise multiple logistic regression analysis. To simplify the application of the rule, continuous variables were recategorized into binary factors using the most discriminant cutoff point. Results of multivariate analysis are reported as odds ratios with 95% confidence intervals (CIs) and P values.

The logistic regression model coefficients for statistically significant predictors of bacteremia were used to assign the value (risk score ) of each variable. We then computed the score for each patient reflecting the probability of bacteremia and chose the cutoff value that discriminated between the low and the high probability groups. Finally, we performed a receiver-operating characteristic (ROC) curve analysis and computed the area under the ROC curve (AUC) and its corresponding 95% CI.

Internal validation. We applied the resultant prediction score to the cohort including the 40% of patients reserved for the internal validation, and rates of bacteremia for each prediction level were calculated. We also performed ROC curve analysis in this population.

External Validation Step

We validated the score in an independent cohort of patients included in a prospective study promoted by the Group of Respiratory Infections from the Sociedad Española de Patología del Aparato Respiratorio. A total of 13 Spanish general hospitals entered into the study, which included all consecutive patients admitted to the hospital for CAP in a 12-month period (January through December 2007). The study was approved by the scientific and ethics committees of participating institutions. Definitions of CAP and variables, inclusion and exclusion criteria, and methods for blood cultures were similar to those used for the derivation and internal validation groups.

Similar to the internal validation, we assessed the predictive accuracy of the score in the external validation cohort by calculating the prevalence of bacteremia in the clinical probability categories according to the point assignment. We also assessed the discrimination ability of the score by means of an ROC curve analysis.

Previous SectionNext Section

Results

Derivation and Internal Validation

Study population. A total of 3116 patients with recorded cases satisfied the inclusion and exclusion criteria and were enrolled in the retrospective study for derivation and internal validation. Patients with missing values were excluded from the analysis, and the remaining 2286 patients were analyzed for the study.

True pathogens were detected in 322 episodes (15%), and contaminants were detected in 148 (6%). Among the true pathogens, Streptococcus pneumoniae was the most frequently isolated bacterium, found in 277 (82%) patients. This was followed by Haemophilus influenzae, found in 14 (4%); viridans streptococci, found in 11 (3%); Staphylococcus aureus, found in 6 (2%); Pseudomonas aeruginosa, found in 5 (2%); Escherichia coli, found in 5 (2%); and Klebsiella pneumoniae, found in 4 (1%). Other pathogens were found in 13 (4%). Polymicrobial infections were detected in 3 patients.

Of these 2286 patients, 1386 (60%) were randomly assigned to the derivation cohort, and 900 (40%) were assigned to the internal validation cohort. The baseline characteristics and microbiological findings for both cohorts of patients are shown in tables 1and 2.

Table 1
View larger version:
Table 1

Demographic, Clinical, and Radiological Characteristics of and Rates of Bacteremia among Patients in the Derivation, Internal Validation, and External Validation Cohorts

Table 2
View larger version:
Table 2

Microorganisms Isolated from Blood in the Derivation, Internal Validation, and External Validation Cohorts

Score derivation. The univariate analysis identified 14 variables that were significantly associated with bacteremia, including epidemiological factors (chronic liver disease and absence of prior antibiotic use), clinical findings (acute onset of disease, length of disease, chills, cough, expectoration, and pleuritic pain), vital signs (heart rate, respiratory rate, systolic blood pressure, and oxygen saturation), and radiological features (pleural effusion and multilobular infiltration) (table 3). The distribution of patients in the PSI categories was also related to bacteremia, with the risk increasing progressively (8%, 9%, 12%, 15%, and 29% for classes 1–5, respectively).

Table 3
View larger version:
Table 3

Univariate Analysis of Epidemiological, Clinical, and Radiological Findings in Patients With or Without Bacteremia in the Derivation Cohort

To perform the multivariate analysis, continuous variables were categorized into binary factors using the following threshold values: age >65 years, length of illness <72 h, temperature <35°C or >40°C, heart rate >125 beats/min, respiratory rate >30 cycles/min, systolic blood pressure <90 mm Hg, diastolic blood pressure <60 mm Hg, and oxygen saturation <92%. In the multivariate analysis, only 6 of these potential predictive factors remained significantly associated with bacteremia: absence of prior antibiotic therapy, chronic liver disease, pleuritic pain, tachycardia, tachypnea, and systolic hypotension (table 4).

Table 4
View larger version:
Table 4

Multivariate Analysis of Predictive Factors Associated with Bacteremia in the Derivation Cohort and Assignment of Points for the Clinical Score

Results from the multivariate analysis were used to develop the clinical prediction rule. According to the regression coefficients, we assigned 1 point to each independent parameter. The application of the score to patients enabled us to categorize subjects into 5 different classes on the basis of the predicted probability of bacteremia: 122 (9%) had 0 points, 482 (35%) had 1 point, 496 (36%) had 2 points, 244 (18%) had 3 points, and 42 (3%) had 4 or more points (only 2 patients had >4 points). The incidence of bacteremia in these groups was 3%, 8%, 15%, 24%, and 38%, respectively (table 5). When measured by the AUC, the prediction rule showed an accuracy of 0.70 (95% CI, 0.66–0.74). For patients admitted to the intensive care unit, bacteremia rates were 0% (0 of 6), 12% (3 of 25), 37% (19 of 51), 58% (7 of 12), and 50% (1 of 2), respectively, showing an AUC of 0.67 (95% CI, 0.59–0.76).

Table 5
View larger version:
Table 5

Rates of Bacteremia by Clinical Score in the Derivation, Internal Validation, and External Validation Cohorts

Internal validation. The score predicted the incidence of bacteremia well when it was applied to the internal validation cohort across all risk classes. For patients included in the internal validation group, we obtained incidences of bacteremia that began at 3% for patients who had no risk factors and that progressively increased to 63% for patients with 4 or more risk factors (table 5). This is an accuracy of 0.69 (95% CI, 0.65–0.74), as measured by the AUC.

External Validation

A total of 1369 patients who were admitted to 13 Spanish hospitals and were prospectively enrolled in the study formed the external validation cohort. However, 242 were excluded because they had variables with missing values; thus, 1127 patients were analyzed. Table 1shows the baseline demographic and clinical characteristics of these patients, and table 2shows microbiological results of blood cultures.

The incidences of bacteremia ranged from 3% for patients without risk factors to 29% for subjects with 4 or more risk factors (table 5), with an AUC of 0.67 (95% CI, 0.62–0.71). A comparison of the results obtained with the application of the score to the different cohorts of patients and categories of risk is shown in table 5. In the 3 cohorts, between 43% and 49% of patients were categorized as having 1 or 0 points, with a corresponding bacteremia rate <8%.

Previous SectionNext Section

Discussion

In the present study, a very simple score was generated that estimates the risk of bacteremia in patients with CAP, and the model was internally and externally validated. The presence of chronic liver disease, pleuritic pain, tachycardia, tachypnea, and systolic hypotension and the absence of prior antibiotic therapy were identified through multivariate analysis as predictors of bacteremia. The coincidence of 2 or more of these predictive factors carried a substantial risk of bacteremia, thereby justifying culture of blood from patients with these risk factors.

To predict bacteremia in patients with CAP, prior experiences have been reported without fully resolving the issue [8, 10, 11]. The most substantive study was published by Metersky et al [8] in 2004. They developed a prediction tool using the following variables as predictors: recent antibiotic treatment, liver disease, 3 vital signs (systolic blood pressure <90 mm Hg, temperature <35°C or ⩾40°C, and pulse ⩾125 beats/min), and 3 laboratory measurements (blood urea nitrogen level ⩾30 mg/dL, sodium level <130 mmol/L, and white blood cell count <5000 or >20000 cells/mm3). Patients were categorized into 3 groups with a low (2%), moderate (5%), and high (11%) risk of bacteremia. However, the authors recognized that the study had some limitations. Specifically, data were retrospectively collected from Medicare patients (only 10% of patients were <65 years old), and subjects had been discharged with a diagnosis of pneumonia, septicemia, or respiratory failure. In addition, the spectrum of isolated bacteria was not concordant with the expected flora, showing a relatively low incidence of S. pneumoniae and a high frequency of relatively uncommon agents, such as S. aureus and gram-negative bacilli [12]. Furthermore, this model was not prospectively validated.

Because of this poor scientific evidence, the current guidelines are imprecise and recommend that blood cultures be performed for all patients with severe CAP; the culture of blood from the remaining subjects requiring hospital admission is optional [13]. Undoubtedly, severity of illness, usually estimated using the PSI or CURB-65 prognostic rules, cannot be used as the exclusive criterion for selecting patients at risk for bacteremia. In these rules, parameters such as age and, in the PSI, some underlying diseases (conditions unrelated to bacteremia, as specific studies have found) play a very important role [14, 15]. Previous reports had already demonstrated a weak correlation between the PSI classification and bacteremia [5, 6], and our results are consistent with this observation. Thus, 45% of patients with bacteremia in our study were classified in low-severity categories (PSI class I, II, or III).

Several aspects of the immune system—including the activity of the reticuloendothelial system, polymorphonuclear leukocyte function (eg, leukocyte adherence, chemotaxis, and phagocytosis), and bactericidal activity of serum—are depressed in patients with chronic liver diseases, who have a high predisposition to bacterial infections [16]. Thus, chronic liver diseases have been identified as risk factors for bacteremia in patients with sepsis [17]. Similarly, many epidemiological studies have demonstrated that prior use of antibiotics reduces the rate of positive results from diagnostic tests based on the isolation of microorganisms from culture [1, 18–20]. In the study by Metersky et al [8], both chronic liver disease and the lack of prior antibiotic therapy were also associated with bacteremia.

On the other hand, a greater presence of clinical symptoms and signs of pneumonia—particularly pleuritic pain—has been traditionally related to the bacterial etiology, not to atypical agents [21, 22]. In the present study, most of these clinical manifestations were associated with bacteremia in the univariate analysis, and pleuritic pain was identified as an independent factor in the multivariate analysis. Pleuritic pain probably summarizes features that define the typical clinical picture that is more frequently caused by pathogens that can be recovered from the blood. The remaining parameters selected in this model are 3 vital signs that are related to a more severe clinical presentation.

In patients with severe CAP, antibiotic treatment should be initiated promptly after the diagnosis of pneumonia is established on the basis of a chest radiograph [13, 23]. Similarly, specimens for culture should be obtained before the start of treatment in order to increase the likelihood of detecting bacteremia [19]. Therefore, the collection of blood for culture needs to be done early, usually when laboratory data are still not available. Even more, it would be advisable to simultaneously collect blood for laboratory analyses and culture. To apply our scoring system, only 6 epidemiological and clinical variables need to be known, and they are available at presentation for almost all patients, are very easy to obtain, and are not subject to the ability of the physician.

Our predictive model clearly discriminates between 2 situations. A subgroup of patients have a low risk of bacteremia (score, ⩽1), with a rate of positive blood culture results <8%. In these low-risk cases, many clinicians can rightfully consider this diagnostic method unjustified. Conversely, the bacteremia rate can reach between 14% and 63% for the remaining patients (score, ⩾2). For these patients, blood cultures can be considered useful diagnostic methods. If this prediction rule is applied, the number of tests performed for patients with pneumonia will be substantially reduced. Although this practice may not detect ∼10% of cases, it is more desirable than the elimination of blood cultures from the diagnostic approach to CAP, as some authors have reasonably suggested [24].

The accuracy of our method is good but not excellent, as the statistical analysis demonstrates (AUC, 0.70). We can find positive blood culture results for patients who had no risk factors. In such patients, an underlying immunodeficiency—particularly human immunodeficiency virus infection—must be excluded [7]. Additionally, even among patients with many predictive factors, only 1 positive result will be obtained for every 2 or 3 patients evaluated.

In summary, the risk of bacteremia in CAP can be estimated by evaluating simple epidemiological and clinical factors. Our prediction model allows for the stratification of patients according to the risk of bacteremia. This identifies a subgroup of low-risk patients for whom blood cultures would have little diagnostic value. Conversely, blood collection for culture appears to be fully justified for the remaining subjects.

Previous SectionNext Section

NAC-CALIDAD (Proyecto Integrado de Investigación de la Sociedad Española de Patología del Aparato Respiratorio sobre Infecciones Respiratorias de Vías Bajas) Study Group Members

R. Menéndez (Hospital Universitario La Fe, Valencia, Spain), J. Aspa (Hospital La Princesa, Madrid, Spain), A. Capelastegui (Hospital de Galdakano, Vizcaya, Spain), I. Alfageme (Hospital de Valme, Sevilla, Spain), S. Bello (Hospital Miguel Servet, Zaragoza, Spain), L. Borderias (Hospital San Jorge, Huesca, Spain), J. J. Martín (Hospital Carlos Haya, Málaga, Spain), L. Molinos (Hospital Central de Asturias, Oviedo, Spain), J. Rello (Hospital Joan XXIII, Tarragona, Spain), F. Rodríguez de Castro (Hospital Doctor Negrín, Las Palmas de Gran Canarias, Spain), J. Ruiz Manzano (Hospital Germans Trías i Pujol, Badalona, Spain), R. Zalacaín (Hospital de Cruces, Vizcaya, Spain), and A. Torres (Hospital Clínic, Barcelona, Spain).

Previous SectionNext Section

Acknowledgments

Financial support. Ciber de Enfermedades Respiratorias from Instituto de Salud Carlos III, Madrid, Spain.

Potential conflicts of interest. All authors: no conflicts.

  • Received December 10, 2008.
  • Accepted March 10, 2009.
Previous Section
 

References

  1. ↵
    1. Skerrett SJ
    . Diagnostic testing to establish a microbial cause is helpful in the management of community-acquired pneumonia. Semin Respir Infect 1997;12:308-21.
    Medline
    1. Luna CM,
    2. Famiglietti A,
    3. Absi R,
    4. et al
    . Community-acquired pneumonia: etiology, epidemiology, and outcome at a teaching hospital in Argentina. Chest 2000;118:1344-54.
    Abstract/FREE Full Text
  3. ↵
    1. Wattanathum A,
    2. Chaoprasong C,
    3. Nunthapisud P,
    4. et al
    . Community-acquired pneumonia in Southeast Asia: the microbial differences between ambulatory and hospitalized patients. Chest 2003;123:1512-9.
    Abstract/FREE Full Text
  4. ↵
    1. Sanyal S,
    2. Smith PR,
    3. Saha AC,
    4. Gupta S,
    5. Berkowitz L,
    6. Homel P
    . Initial microbiologic studies did not affect outcome in adults hospitalized with community-acquired pneumonia. Am J Respir Crit Care Med 1999;160:346-8.
    Abstract/FREE Full Text
  5. ↵
    1. Campbell SG,
    2. Marrie TJ,
    3. Anstey R,
    4. Dickinson G,
    5. Ackroyd-Stolarz S
    . The contribution of blood cultures to the clinical management of adult patients admitted to the hospital with community-acquired pneumonia. Chest 2003;123:1142-50.
    Abstract/FREE Full Text
  6. ↵
    1. Waterer GW,
    2. Wunderink RG
    . The influence of the severity of community-acquired pneumonia on the usefulness of blood cultures. Re spir Med 2001;95:78-82.
  7. ↵
    1. Falguera M,
    2. Martín M,
    3. Ruiz-González A,
    4. Pifarré R,
    5. García M
    . Community-acquired pneumonia as the initial manifestation of serious underlying diseases. Am J Med 2005;118:378-83.
    CrossRefMedlineWeb of Science
  8. ↵
    1. Metersky ML,
    2. Ma A,
    3. Bratzler DW,
    4. Houck PM
    . Predicting bacteremia in patients with community-acquired pneumonia. Am J Respir Crit Care Med 2004;169:342-7.
    Abstract/FREE Full Text
  9. ↵
    1. Fine MJ,
    2. Auble TE,
    3. Yealy DM,
    4. et al
    . A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997;336:243-50.
    CrossRefMedlineWeb of Science
  10. ↵
    1. Marrie TJ
    . Blood cultures in ambulatory patients who are discharged from emergency with community-acquired pneumonia. Can J Infect Dis 2004;15:21-4.
    Medline
  11. ↵
    1. Müller B,
    2. Harbarth S,
    3. Stolz D,
    4. et al
    . Diagnostic and prognostic accuracy of clinical and laboratory parameters in community-acquired pneumonia. BMC Infect Dis 2007;7:10.
    CrossRefMedline
  12. ↵
    1. Barlett JG,
    2. Mundy LM
    . Community-acquired pneumonia. N Engl J Med 1995;333:1618-24.
    CrossRefMedlineWeb of Science
  13. ↵
    1. Mandell LA,
    2. Wunderink RG,
    3. Anzueto A,
    4. et al
    . Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults. Clin Infect Dis 2007;44:27-72.
    CrossRef
  14. ↵
    1. Fernández-Sabe N,
    2. Carratalà J,
    3. Rosón B,
    4. et al
    . Community-acquired pneumonia in very elderly patients. Causative organisms, clinical manifestations, and outcomes. Medicine 2003;82:159-69.
    CrossRefMedline
  15. ↵
    1. Falguera M,
    2. Pifarré R,
    3. Martin M,
    4. Sheick A,
    5. Moreno A
    . Etiology and outcome of community-acquired pneumonia in patients with diabetes mellitus. Chest 2005;128:3233-9.
    Abstract/FREE Full Text
  16. ↵
    1. Navasa M,
    2. Rodés J
    . Bacterial infections in cirrhosis. Liver Int 2004;24:277-80.
    CrossRefMedlineWeb of Science
  17. ↵
    1. Bates DW,
    2. Sands K,
    3. Miller E,
    4. et al
    . Predicting bacteremia in patients with sepsis syndrome. J Infect Dis 1997;176:1538-51.
    Abstract/FREE Full Text
  18. ↵
    1. Ewig S,
    2. Schlochtermeier M,
    3. Göke N,
    4. Niederman MS
    . Applying sputum as a diagnostic tool in pneumonia: limited yield, minimal impact on treatment decisions. Chest 2002;121:1486-92.
    Abstract/FREE Full Text
  19. ↵
    1. Marston BJ,
    2. Plouffe JF,
    3. File TM,
    4. et al
    . Incidence of community-acquired pneumonia requiring hospitalization: results of a population-based active surveillance study in Ohio. Arch Intern Med 1997;157:1709-18.
    Abstract/FREE Full Text
  20. ↵
    1. Dorca J,
    2. Manresa F,
    3. Esteban L,
    4. et al
    . Efficacy, safety, and therapeutic relevance of transthoracic aspiration with ultrathin needle in nonventilated nosocomial pneumonia. Am J Respir Crit Care Med 1995;151:1491-6.
    Abstract/FREE Full Text
  21. ↵
    1. Ruiz-González A,
    2. Falguera M,
    3. Vives M,
    4. Nogués A,
    5. Porcel JM,
    6. Rubio-Caballero M
    . Community-acquired pneumonia: development of a bedside predictive model and scoring system to identify the etiology. Respir Med 2000;94:505-10.
    CrossRefMedlineWeb of Science
  22. ↵
    1. Kauppinen MT,
    2. Saikku P,
    3. Kujala P,
    4. Herva E,
    5. Syrjälä H
    . Clinical picture of community-acquired Chlamydia pneumoniae pneumonia requiring hospital treatment: a comparison between chlamydial and pneumococcal pneumonia. Thorax 1996;51:185-9.
    Abstract/FREE Full Text
  23. ↵
    1. Houck PM,
    2. Bratzler DW,
    3. Nsa W,
    4. Ma A,
    5. Barlett JG
    . Timing of antibiotic administration and outcomes for Medicare patients hospitalized with community-acquired pneumonia. Arch Intern Med 2004;164:637-44.
    Abstract/FREE Full Text
  24. ↵
    1. Luna CM
    . Blood cultures in community-acquired pneumonia: are we ready to quit? Chest 2003;123:977-8.
    FREE Full Text
| Table of Contents

This Article

  1. Clin Infect Dis. (2009) 49 (3): 409-416. doi: 10.1086/600291
  1. AbstractFree
  2. » Full Text (HTML)Free
  3. Full Text (PDF)Free

Classifications

Share

  1. Email this article

Navigate This Article

Current Issue

  1. March 1, 2012 54 (5)
  1. Clinical Infectious Diseases
  1. Alert me to new issues

Most