American Journal of Respiratory and Critical Care Medicine

The aim of the study was to determine the causes and prognostic implications of antimicrobial treatment failures in patients with nonresponding and progressive life-threatening, community-acquired pneumonia. Forty-nine patients hospitalized with a presumptive diagnosis of community-acquired pneumonia during a 16-mo period, failure to respond to antimicrobial treatment, and documented repeated microbial investigation ⩾ 72 h after initiation of in-hospital antimicrobial treatment were recorded. A definite etiology of treatment failure could be established in 32 of 49 (65%) patients, and nine additional patients (18%) had a probable etiology. Treatment failures were mainly infectious in origin and included primary, persistent, and nosocomial infections (n = 10 [19%], 13 [24%], and 11 [20%] of causes, respectively). Definite but not probable persistent infections were mostly due to microbial resistance to the administered initial empiric antimicrobial treatment. Nosocomial infections were particularly frequent in patients with progressive pneumonia. Definite persistent infections and nosocomial infections had the highest associated mortality rates (75 and 88%, respectively). Nosocomial pneumonia was the only cause of treatment failure independently associated with death in multivariate analysis (RR, 16.7; 95% CI, 1.4 to 194.9; p = 0.03). We conclude that the detection of microbial resistance and the diagnosis of nosocomial pneumonia are the two major challenges in hospitalized patients with community-acquired pneumonia who do not respond to initial antimicrobial treatment. In order to establish these potentially life-threatening etiologies, a regular microbial reinvestigation seems mandatory for all patients presenting with antimicrobial treatment failures.

The majority of patients hospitalized with community-acquired pneumonia (CAP) respond satisfactorily to antimicrobial treatment. However, it has been estimated that approximately 10 to 25% of patients with CAP do not resolve within the anticipated time (1), and an additional 10% may experience progressive life-threatening pneumonia (2, 3). In fact, despite effective antimicrobial regimens, mortality rates have remained virtually unchanged in the last decades, ranging from 4 to 14% in the general hospitalized population (2-7).

Fein and Feinsilver (8) have provided useful practical definitions for different types of treatment failures. They suggest that delayed clinical and/or radiographic improvement should be addressed as “slowly resolving pneumonia,” persistent infiltrates in chest radiograph as “non-resolving pneumonia,” and clinical deterioration as “progressive pneumonia.” In this view, the first two types of treatment failures would represent nonresponse in the clinically stable patient, and the latter would represent nonresponse as a clinical emergency. Most but not all patients with progressive pneumonia experience primary treatment failures within 48 to 72 h after in-hospital antimicrobial treatment (9). Potential causes for the first type of treatment failure have been described in detail (1, 10). Important causes comprise pulmonary infectious complications, unusual pathogens, and a variety of noninfectious mimics of CAP. Much less attention has been paid to the description of causes of progressive pneumonia. Overall, surprisingly little information is available about the role of microbial susceptibility and nosocomial pneumonia in patients with treatment failures.

We therefore studied the causes of antimicrobial treatment failures in hospitalized patients with CAP. We were especially interested in determining the role of microbial susceptibility and nosocomial pneumonia. Finally, we also wanted to determine the potential prognostic implications of different causes of treatment failures.

Study Population

The database was derived from a prospective study on community- acquired pneumonia conducted at our 1,000-bed, tertiary-care, university teaching hospital. All adult patients hospitalized with a presumptive diagnosis of community-acquired pneumonia, failure to respond to antimicrobial treatment, and documented repeated microbial investigation during a 16-mo period, including two winter seasons, were recorded.

Community-acquired pneumonia was assumed in the presence of a new infiltrate on chest radiograph on admission together with symptoms suggestive of a lower respiratory tract infection. Patients with severe immunosuppression (HIV infection, solid organ transplantation, or neutropenia < 1.0 × 109/L, steroid treatment > 20 mg/d or any other immunosuppressive regimen) were excluded.


Failure to respond to antimicrobial treatment was classified as nonresponding or progressive pneumonia. Nonresponding pneumonia was defined as persisting fever > 38° C and/or clinical symptoms (malaise, cough, expectoration, dyspnea) after at least 72 h of antimicrobial treatment. Progressive pneumonia was defined as clinical deterioration in terms of the development of acute respiratory failure requiring ventilatory support and/or septic shock after at least 72 h of antimicrobial treatment. The definitions of septic shock were adopted from Bone and coworkers (11). According to this definition, septic shock is present in cases of severe sepsis with hypotension despite adequate fluid resuscitation along with the presence of perfusion abnormalities.

The causes of treatment failures were further classified into six categories.

Primary infections. Demonstration of a pathogen not detected in initial investigations (“atypical” or unusual pathogens or pathogens associated with the development of empyema).

Definite persistent infections. Demonstration of the same pathogen in initial and repeated investigations.

Probable persistent infections. Demonstration of a pathogen in initial but not in repeated investigations.

Nosocomial infections. Demonstration of a pathogen not present in the initial evaluation (sputum or tracheobronchial aspirate [TBAS]) usually associated with early- or late-onset ventilator-associated pneumonia.

Noninfectious etiology. Definite alternative diagnosis.

Nondiagnostic. Neither definite infectious nor noninfectious etiology.

Comorbidities were defined as follows.

Cardiac comorbid illness. Treatment for coronary artery disease or congestive heart failure or presence of valvular heart disease.

Pulmonary. Treatment for asthma or chronic obstructive pulmonary disease (COPD) or presence of interstitial lung disorders.

Renal. Preexisting renal disease with documented abnormal serum-creatinine outside the pneumonia episode.

Hepatic. Preexisting viral or toxic hepatopathy.

CNS disorders. Presence of symptomatic acute or chronic vascular or nonvascular encephalopathy, with or without dementia.

Diabetes mellitus. Diagnosis of intolerance to glucose and treatment with oral antidiabetics or insulin.

Neoplastic illness. Any solid tumor active at the time of presentation or requiring antineoplastic treatment within the previous year.

Alcohol abuse was defined as the ingestion of an estimated amount of 80 g alcohol per day during at least the previous year. Smokers were defined as current cigarette smokers of > 10 cigarettes/d during at least the previous year.

Study Protocol

Regular initial microbial investigation after hospital admission included sampling of sputum, two sets of blood cultures, and serology. Initial empiric antimicrobial treatment was administered according to general guidelines in the hospital. It mainly included a combination regimen of an intravenously administered third-generation cephalosporin with an orally administered macrolide. Clinical, radiographic, and microbial reinvestigation after 72 h of in-hospital antimicrobial treatment was ordered at the discretion of the attending physician. Radiographic reinvestigation consisted of repeated chest radiographs and/or CT scan. Microbial reinvestigation included a repeated sampling of sputum and two sets of blood cultures as well as additional techniques as judged appropriate by the attending physician. These included serology, pleural puncture, transthoracic needle aspiration (TTA), flexible bronchoscopy with protected specimen brush (PSB), bronchoalveolar fluid (BALF), and/or transbronchial biopsy, and TBAS in intubated patients. Autopsy results were additionally retrieved.

Data Collection

Clinical, radiographic, and laboratory data were recorded in a data sheet and entered in a computer database. The following demographic data were recorded: age, sex, smoking and alcohol habits, comorbid illnesses, antimicrobial treatment prior to hospital admission, and duration of symptoms prior to the diagnosis of pneumonia. In addition, the following parameters were recorded on admission and reassessment: clinical symptoms (body temperature, pleuritic chest pain, cough, expectoration, dyspnea, confusion [i.e., disorientation with regard to person, place, or time that is not known to be chronic, stupor, or coma]), clinical presentation (presence of rales, respiratory rate, heart rate, arterial systolic and diastolic blood pressure, APACHE II score), blood gas analysis (PaO2 , PaCO2 , PaO2 /Fi O2 ), radiographic parameters (extension and evolution of infiltrates in chest radiograph, presence of pleural effusion), laboratory parameters (leukocyte count, LDH, serum-creatinine, and blood urea nitrogen), the requirement for mechanical ventilation and the presence of septic shock, results of microbial investigations, and in-hospital outcome.

Microbiologic Evaluation

Sputum was Gram-stained. Representative sputum originating from the lower respiratory tract was validated by the criteria > 25 granulocytes and < 10 epithelial cells per low power field (total magnification: ×100) (12). Validated sputum, blood culture samples, pleural fluid, transthoracic needle aspiration samples, and undiluted and serially diluted TBAS, PSB, and BAL fluid samples were plated on the following media: blood-sheep agar, CDC agar, and chocolate agar, as well as Sabouraud agar. Undiluted PSB and BAL fluid samples were also cultured on charcoal-yeast extract agar. Urine was tested for the presence of Legionella spp. antigen. Identification of microorganisms and susceptibility testing were performed according to standard methods (13). Results of quantitative cultures were expressed as colony-forming units per milliliter (CFU/ml).

Diagnostic Criteria

Infectious etiology of pneumonia was classified as presumptive if a valid sputum sample yielded one or more predominant bacterial strains. It was considered definite if one of the following criteria were met: (1) blood cultures yielding a bacterial or fungal pathogen (in the absence of an apparent extrapulmonary focus); (2) pleural fluid and transthoracic needle aspiration cultures yielding a bacterial pathogen; (3) seroconversion, i.e., a fourfold rise in IgG titers for Chlamydia pneumoniae (IgG ⩾ 1:512), Chlamydia psittaci (IgG ⩾ 1:64), Legionella pneumophila ⩾ 1:128, Coxiella burnetii ⩾ 1:80, and respiratory viruses, i.e., influenza virus A and B, parainfluenza virus 1 to 3, respiratory syncytial virus, adenovirus; (4) single elevated IgM titer for Chlamydia pneumoniae ⩾ 1:32, Coxiella burnetii ⩾ 1:80, and Mycoplasma pneumoniae (any titer); (5) a single titer ⩾ 1:128 or a positive urinary antigen for Legionella pneumophila; (6) bacterial growth in cultures of TBAS ⩾ 105 CFU/ml, in PSB ⩾ 103 CFU/ml, and in BALF ⩾ 104 CFU/ml. Coagulase-negative staphylococci were only accepted as diagnostic when concomitantly isolated from blood and transthoracic needle aspiration. Aspergillus spp. were accepted as definite in the presence of concomitant lung abscess and/or histologic confirmation.

Independent of microbiologic results, a diagnosis of probable aspiration was made in cases of witnessed aspiration or in the presence of risk factors for aspiration (severely altered consciousness, abnormal gag reflex, or abnormal swallowing mechanism) (14). Empyema was defined as the presence of gross pus at thoracocentesis with or without demonstration of a pathogen in culture.


Results are expressed as means ± SD. Continuous variables were compared using the t-test for independent or paired samples, categorical variables were compared using the chi-square test or Fisher's exact test when appropriate. Multiple comparisons of continuous variables were performed by ANOVA with post-hoc Bonferroni correction. Multivariate analysis was performed by stepwise forward logistic regression. All reported p values are two-tailed. The level of significance was set at 5%.

Patient Population

Overall, 444 patients were hospitalized during the study period because of suspected community-acquired pneumonia. Of these, 49 patients (11%) had a repeated investigation because of antimicrobial treatment failure and were included in the study. The main demographic characteristics are summarized in Table 1. Thirty-five patients (71%) had at least one comorbidity, including 13 (21%) who had more than one. Pulmonary comorbidity was by far the most common (45%), followed by diabetes mellitus (16%), CNS disorders (14%), and cardiac comorbidities (12%).


Smokers, n (%)32 (65)
Alcoholism, n (%)15 (31)
Comorbidity present, n (%)35 (71)
No. of comorbidities, n (%)
 122 (45)
 2 9 (18)
 34 (8)
Type of comorbidities, n (%)
 Cardiac 6 (12)
 Pulmonary22 (45)
 Hepatic4 (8)
 Renal3 (6)
 CNS 7 (14)
 Diabetes mellitus 8 (16)
 Malignancy2 (4)
Oral ambulatory antimicrobial pretreatment, n (%)20 (41)
Duration of symptoms until hospital admission, h
 Mean ± SD158 ± 142
Time to reinvestigation, h
 Mean ± SD98 ± 49

Clinical characteristics of the study population on admission are summarized in Table 2. Sixteen of 49 patients (33%) had severe CAP on admission (within the first 4 h of evaluation) and were referred to the intensive care unit. Of these, nine (56%) had acute respiratory failure requiring ventilatory support, including two (13%) with septic shock.


VariablesOn AdmissionOn Repeated Investigationp Value
Respiratory rate, breaths/min*  30 ± 9 28 ± 80.16
Heart rate, beats/min*  99 ± 17 99 ± 180.9
Systolic arterial blood pressure, mm Hg* 127 ± 20115 ± 330.006
Diastolic arterial blood pressure, mm Hg*  73 ± 13 61 ± 16< 0.0001
Confusion present, n (%)15 (31)21 (43)0.21
Probable or witnessed gross aspiration, n (%) 7 (14)
APACHE II, mean ± SD 14 ± 6 16 ± 70.28
Leukocytes × 109/L* 13.6 ± 7.314.2 ± 6.40.56
PaO2 /Fi O2 250 ± 101176 ± 111< 0.0001
Type of pulmonary infiltrates, n (%)
 Consolidating41 (84)42 (86)0.78
 Bronchopneumonic 4 (8) 5 (10)1.0
 Interstitial 4 (8) 2 (4)0.68
Multilobar infiltrates, n (%)17 (35)32 (65)0.002
Pleural effusion, n (%) 9 (18)21 (43)0.009

*Values are mean ± SD.

Including mechanically ventilated patients.

Initial Microbial Investigation

Sputum was obtained in 25 patients, blood cultures in 41, tracheobronchial aspirates in eight, pleural fluid in three, and Legionella antigen in one.

An etiologic diagnosis could be made within 48 h in 15 patients (31%). The pathogens involved are listed in Table 3. Streptococcus pneumoniae was the most frequently isolated pathogen, followed by Pseudomonas aeruginosa.


Etiology determined15 (31)
Streptococcus pneumoniae 6SPU (2), BC (5), TBAS (1)
β-hemolytic streptoccci1BC
Streptococcus milleri 1PF
Coagulase-negative staphylococci1BC
Hemophilus influenzae 2SPU (1), TBAS (1)
Pseudomonas aeruginosa 4SPU (4)
Pseudomonas stutzeri 1BC
Legionella spp.1LEGAG

Definition of abbreviations: SPU = sputum; BC = blood culture; TBAS = tracheobronchial aspirate; PF = pleural fluid culture; LEGAG = Legionella antigen.

*Results fo initial evaluation (within 48 h) and medium of diagnosis.

Two patients had mixed (double) infections.

Initial Antimicrobial Treatment

Overall, 16 patients (33%) received oral ambulatory antimicrobial pretreatment, and four patients had received multiple antimicrobial agents. The regimen included amoxycillin (n = 3), amoxycillin plus β-lactamase inhibitor (n = 6), oral cephalosporins (n = 3), macrolides (n = 5), ciprofloxacin (n = 3). One additional patient was receiving aerosolized gentamicin treatment.

Initial in-hospital empiric antimicrobial treatment consisted of monotherapy: aminopenicillin plus β-lactamase inhibitor (n = 2), third-generation cephalosporin (n = 1); dual combination therapy: third-generation cephalosporin plus macrolide (n = 24), fourth-generation cephalosporin plus macrolide (n = 3), ciprofloxacin plus macrolide (n = 1); triple combination therapy: third-generation cephalosporin plus macrolide plus rifampicin (n = 7), or plus aminoglycoside (n = 3), or plus cotrimoxazole (n = 2), or plus ciprofloxacin (n = 1), fourth-generation cephalosporin plus macrolide plus aminoglycoside (n = 3); quadruple combination therapy: third-generation cephalosporin plus macrolide plus vancomycin plus aminoglycoside (n = 1), or plus ciprofloxacin (n = 1).

Characteristics of Treatment Failures

Thirty of 49 patients (61%) had nonresponding pneumonia and 19 of 49 (39%) had progressive pneumonia according to the given definition, including six of 16 (38%) with initial intensive care unit admission. All patients with nonresponding pneumonia presented with persistent fever, and 15 (50%) presented with both persistent fever and symptoms. Progressive pneumonia included acute respiratory failure in nine patients (47%), septic shock in four (21%), and both in six (32%).

Accordingly, clinical characteristics at reassessment changed significantly in terms of systolic and diastolic blood pressure, PaO2 /Fi O2 , multilobar infiltrates, and pleural effusion (Table 2). A spread of radiographic infiltrates in a repeated chest radiograph was present in 24 of 49 patients (49%), including eight (16%) ⩾ 50%.

Time Course of Different Types of Treatment Failures

The time course of different types of treatment failures was not significantly different, ranging from a mean of 81 ± 18 (range, 72 to 120) for nosocomial infections to 135 ± 95 h after initiation of in-hospital antimicrobial treatment (range, 72 to 312) for primary infections (see Table 4).


Type of Treatment FailureMean Hours after Initiation of in-hospital Antimicrobial TreatmentSDRange (h)
Primary infection1359572–312
Definite persistant infection 903672–144
Probable persistant infection1002872–144
Nosocomial infection 811872–120
Noninfectious etiology1054872–192
Nondiagnostic 852972–144

*Four patients with diverse double etiologies were excluded. Differences were not statistically significant (p = 0.30).

Etiology of Treatment Failures

A definite etiology of treatment failure could be established in 32 of 49 patients (65%). Etiologies comprised primary infection (n = 8), definite persistent infection (n = 4), nosocomial infection (n = 8, including one patient with two pathogens), noninfectious causes (n = 8), and dual causes in four patients (primary infection plus probable persistent infection [n = 1], primary infection plus nosocomial infection [n = 1], definite persistent infection plus nosocomial infection and noninfectious cause, respectively [n = 1 each]).

Nine additional patients (9 of 49, 18%) had a probable etiology, including probable persistent infection (n = 6), empyema without pathogen (n = 1), and gross aspiration without pathogen (n = 2).

Causes summed up to 46 (36 definite and 10 probable). Definite infectious causes were the most frequent (27 of 46, 59%). Noninfectious causes were present in nine of 46 (20%). The types of treatment failures and pathogens involved are listed in Table 5.


Primary infections10
 Gram-positive Streptococcus milleri 1*,
 Atypical Mycoplasma pneumoniae  2
Coxiella burnetii  1
Mycobacterium tuberculosis  3
 Fungal Histoplasma capsulatum  2
 Other Nocardia asteroides  1
Definite persistent infection 6
 Gram-positive Streptococcus pneumoniae 2 (1*)
Coagulase-negative staphylococci1
 Gram-negative Pseudomonas aeruginosa  3
Probable persistent infection 7
 Gram-positive Streptococcus pneumoniae 3 (1*, 1)
β-hemolytic streptococci1*
Streptococcus milleri 1*,
 Gram-negative Pseudomonas aeruginosa  1
Pseudomonas stutzeri  1
Nosocomial infection11
 Gram-positive Streptococcus pneumoniae  1
 Gram-negativeGNEB 2
Pseudomonas aeruginosa  2
Stenotrophomonas maltophilia  1
 Fungal Aspergillus fumigatus  2
Noninfectious 9
 Malignancy 3
 Interstitial lung diseaseBOOP 2
Histiocytosis X 1
 Cardiopathy 2
 Foreign body1
 Empyema without pathogen1*
 Aspiration without pathogen2
 No definite or probable etiology 8

Definition of abbreviations: GNEB = Gram-negative enteric bacilli; MRSA = methicillin resistant S. aureus; BOOP = bronchiolitis obliterans with organizing pneumonia.

*Pathogen associated with empyema.

Pathogen associated with aspiration.

An empyema was present in six of 49 patients (12%) (Streptococcus pneumoniae [n = 2], Streptococcus milleri [n = 2], β-hemolytic streptococci [n = 1], no pathogen [n = 1]).

Seven patients with gross aspiration had Streptococcus pneumoniae (n = 1), empyema with Streptococcus milleri (n = 2), coagulase-negative staphylococci (n = 1), foreign-body aspiration (n = 1), and no pathogen (n = 2).

Patients with nonresponding pneumonia had significantly more often primary, definite, and probable persistent infections (n = 14), as compared with nosocomial infections (n = 2), than did patients with progressive pneumonia (n = 4 and 6, respectively, excluding patients with dual infectious causes; p = 0.03). Seven of eight (88%) noninfectious causes were detected in patients with nonresponding pneumonia, whereas patients with nondiagnostic investigations were equally distributed in both groups (five of 11 [46%] versus six of 11 [54%]).

Association of Persistent Pathogens with Microbial Susceptibility

In five of six patients with definite persistent pathogens, microbial resistance to initially administered antimicrobial treatment could be documented. The remaining patient had empyema. Conversely, six of seven patients with probable persistent infection initially isolated pathogens were susceptible to initially administered antimicrobial treatment, and of these, three had empyema (Table 6).


PathogenOral Ambulatory PretreatmentIn-hospital Initial Antimicrobial TreatmentSusceptibility
Definite persistent infections
Streptococcus pneumoniae NoneCefotaxime/erythromycinIntermediately resistant to both
Streptococcus pneumoniae (empyema and malignancy) NoneCeftriaxone/erythromycin/ciprofloxacinSusceptible to cefotaxime/erythromycin
Coagulase-negative staphylococciNoneCefotaxime/erythromycinResistant to both
Pseudomonas aeruginosa CiprofloxacinCiprofloxacin/erythromycinResistant to both
Pseudomonas aeruginosa Amoxicillin/Clavulanic AcidCefotaxime/erythromycinResistant to both
Pseudomonas aeruginosa NoneCefepime/amikacin/erythromycinResistant to amikacin and erythromycin Cefepime not tested
Probable persistent infections
Streptococcus pneumoniae NoneCefotaxime/erythromycin/rifampicinSusceptible to cefotaxime and erythromycin Rifampicin not tested
Streptococcus pneumoniae NoneCeftriaxone/azithromycinSusceptible to both
Streptococcus pneumoniae(empyema)NoneCeftriaxone/erythromycinSusceptible to both
β-hemolytic streptococci (empyema)NoneCeftriaxone/azithromycinSusceptible to both
Streptococcus milleri(empyema)NoneCefotaximeSusceptible
Pseudomonas aeruginosa NoneCefepime/amikacin/azithromycinSusceptible to amikacin, cefepime  not tested
Pseudomonas stutzeri NoneCeftriaxone/erythromycinResistant to both

Media of Diagnosis in Treatment Failures

Noninvasive techniques had a diagnostic yield ranging from 6 to 41%, with blood cultures having the lowest and TBAS the highest. Paired serology had a higher yield than did single testing (20% versus 4%). Bronchoscopic techniques of lower respiratory secretion sampling (PSB and BAL) had a yield of 40 and 42%, respectively, and transbronchial biopsy of 57% (Table 7).


Diagnostic TechniqueSamples Obtained (n)Positive Results n (%)
Sputum183 (17)
Blood cultures332 (6)
Single281 (4)
Paired164 (20)
Total445 (11)
PSB125 (42)
BAL104 (40)
Transbronchial biopsy 74 (57)
TBAS229 (41)
Pleural effusion172 (12)
Transthoracic puncture 31 (33)
Legionella antigen370

Definition of abbreviations: PSB = protected specimen brush; BAL = bronchoalveolar lavage; TBAS = tracheobronchial aspirate.

*Diagnostic yield of different techniques.

Autopsy Results

Eight patients underwent autopsy. Bilateral pneumonia was confirmed in all. Of these, two had extensive pulmonary hemorrhage, one multiple lung abscesses, and one diffuse alveolar damage. In one patient, miliary tuberculosis could be confirmed, and Aspergillus pneumonia could be confirmed in another.


On repeated investigation, 15 patients required secondary treatment at the ICU. One additional patient was not admitted to the ICU because of multimorbidity. Thus, overall, 31 patients (63%) were treated at the ICU at any time during the course of their disease.

In-hospital mortality was recorded in 21 of 49 patients (43%). Mortality rates were three of eight (38%) for primary infections, three of four (75%) for definite persistent infections, one of six (17%) for probable persistent infections, seven of eight (88%) for nosocomial pneumonia, two of eight (25%) for noninfectious etiologies, three of 11 (27%) for nondiagnostic patients, and two of four (50%) for mixed etiologies.

APACHE II ⩾ 14 (RR, 3.6; 95% CI, 1.2 to 10.5; p = 0.003) as well as nosocomial pneumonia (RR, 2.6; 95% CI, 1.6 to 4.2; p = 0.01), but no other etiologies were significantly associated with death. When nosocomial pneumonia was added to APACHE II ⩾ 14 in a multivariate model, both remained independent predictors of death (nosocomial pneumonia: RR, 16.7; 95% CI, 1.4 to 194.9; p = 0.03, and APACHE II: RR, 9.0; 95% CI, 1.7 to 47.7; p = 0.01, respectively).

The present study provides three important results: (1) treatment failures of hospitalized patients with CAP were mainly infectious in origin and included primary, persistent, and nosocomial infections; (2) definite but not probable persistent infections were mostly due to microbial resistance to the administered initial empiric antimicrobial treatment; (3) nosocomial infections were particularly frequent in patients with progressive pneumonia and were the only cause of treatment failure independently associated with death.

The population studied consisted mainly of early treatment failures. Moreover, a major proportion of patients experienced life-threatening progressive pneumonia. This is important to consider when comparing our results with those of others. One of the major strengths of our study was the repeated investigation design, thereby allowing to differentiate primary, persistent, and nosocomial infections. We argue that this classification is of paramount importance in order to interpret the results of microbial investigation in treatment failures correctly. On the other hand, we thought that from a clinical point of view it would be important to include all types of treatment failures in our analysis. This notion was further supported by the finding that there were no apparent differences in the time course of the diverse treatment failures, which would have allowed us to differentiate patients with distinct etiologies simply with respect to the time course of treatment failure.

In accordance with previous reports, we found a variety of unusual pathogens as primary infections. These included “atypical” bacterial pathogens detected by serology, tuberculosis, and histoplasmosis, as well as nocardiosis detected by special media. These pathogens form part of the recognized microbial patterns of CAP, but they represent rare causes, and, therefore, they are typically associated with a delay in diagnosis. Because this delay may be responsible for an unfavorable prognosis, new molecular diagnostic tools may be useful in order to reduce the frequency of this type of treatment failure. However, in contrast to the findings of others, we found even more frequently persistent and nosocomial infections.

Definite persistent pathogens were nearly exclusively observed in the presence of microbial resistance to the administered initial antimicrobial treatment. The only exception was a patient who developed empyema. The most frequent persistent pathogen was Pseudomonas aeruginosa. Only recently, persistent P. aeruginosa pneumonia has been found to occur in 35% of patients mechanically ventilated (15). These data stress the need to appropriately assess and to treat according to susceptibility patterns of each pseudomonal isolate. Of concern, one patient had persistent Streptococcus pneumoniae in the presence of intermediate resistance to the third-generation cephalosporin and macrolide administered. Although we recently could show that the outcome of pneumococcal pneumonia is not adversely affected by drug resistance in the presence of appropriate antimicrobial treatment (16), this case illustrates the need for continuous surveillance of patterns of pneumococcal resistance in order to prevent disease consequences of resistance. Overall, mortality from definite persistent infections was very high (75%).

The main reason for persistent disease without isolation of the same pathogen in the repeated investigation (probable persistent infection) was empyema. Resistance to the antimicrobial regimen administered was only exceptionally present in one patient with bacteriemic pneumonia caused by Pseudomonas stutzeri. Two patients with Streptococcus pneumoniae and another patient with P. aeruginosa experienced a protracted clinical course despite susceptibility to antimicrobial agents. Because the choice, route, and dosing of antimicrobial treatment was correct in each case, the most probable explanation is an ongoing inflammatory response despite a significant reduction of the bacterial load.

One of the most important findings in our study was the high frequency of nosocomial pneumonia. Overall, 11 pathogens were involved, and most belonged to the pattern usually found in the presence of risk factors or in late-onset pneumonia (17). In addition, our findings confirm that Aspergillus spp. must be taken into account as a nosocomial pathogen also in the absence of severe systemic immunosuppression (18). High respiratory tract colonization rates of 34 to 59% with subsequent pneumonia in 5 and 12% in patients with CAP receiving antimicrobial treatment were reported in studies originating from the 1960s (19, 20). More recently, colonization of the respiratory tract during antimicrobial treatment for CAP could be demonstrated in 38% (21). In this study, colonization was an adverse prognostic factor associated with twice the length of hospital stay, slower recovery, and a 14-fold higher mortality. However, only one of 93 colonized patients had definite nosocomial pneumonia. In a subsequent study, the same investigators could confirm respiratory tract colonization as an independent adverse prognostic factor in multivariate analysis (5). The high frequency of severe nosocomial pneumonia in our study probably reflects a more severely compromised population, with a high incidence of pulmonary comorbidity and severe CAP. In fact, nosocomial pneumonia independently predicted mortality in addition to APACHE II score. These data provide a strong additional argument to perform a repeated microbial investigation in all patients with antimicrobial treatment failures in order to detect possible nosocomial pneumonia, which would require a specific secondary antimicrobial treatment approach.

The most frequently detected “syndromes” of treatment failures were empyema and aspiration, or both. In one patient, even foreign body aspiration could be demonstrated. Therefore, the presence of pleural effusion should always be checked, thoracocentesis of pleural effusion should be rapidly performed, and the possibility of aspiration should be raised in patients with antimicrobial treatment failures.

Noninfectious causes of treatment failures included known mimics of CAP (10). Consistent with other reports, malignancy was quite rare, accounting for only three cases. The same was true for interstitial lung disease. Bronchiolitis obliterans with organizing pneumonia represents a typical mimic of CAP, and it is very important to detect it because of its good response to corticosteroid medication. The use of high resolution CT may obviate the need for transbronchial biopsy in selected cases.

Although this study was not designed to compare the diagnostic yield of diagnostic techniques, some clarifications should be made. TBAS generally had a higher yield than sputum, which probably is due to the fact that it was applied in ventilated patients with a higher probability of nosocomial pneumonia. As expected, serology had a low yield when only a single sample was obtained but an acceptable yield if paired serology was available. The low yield of Legionella antigen may simply reflect a low prevalence of Legionellosis (3). TBAS and bronchoscopically retrieved PSB and BAL had comparable yields, a trend compatible with observations in ventilator- associated pneumonia (22, 23). Bronchoscopy was particularly useful in determining noninfectious causes of treatment failures.

In conclusion, the main challenge in patients with antimicrobial treatment failures, which might have been currently underestimated, is to detect microbial resistance to the initially administered antimicrobial regimen and nosocomial pneumonia. In order to accomplish with these potentially life-threatening etiologies, a regular microbial reinvestigation seems mandatory in all patients presenting with antimicrobial treatment failures.

Supported by Commisionat per a Universitats i Recerca de la Generalitat de Catalunya 1997 SGR 00086, IDIBAPS Hospital Clinic Barcelona, and FISS 98/0138.

1. Feinsilver S. H., Fein A. M., Niederman M. S., Schultz D. E., Faegenburg D. H.Utility of fiberoptic bronchoscopy in non-resolving pneumonia. Chest98199013221326
2. Marrie T. J., Durant H., Yates L.Community-acquired pneumonia requiring hospitalization: 5-year prospective study. Rev. Infect. Dis.111989586599
3. Ruiz M., Ewig S., Marcos M. A., Martinez J. A., Gonzalez J., Arancibia F., Mensa J., Torres A.Etiology of community-acquired pneumonia: impact of age, comorbidity, and severity. Am. J. Respir. Crit. Care Med.1601999397405
4. British Thoracic SocietyCommunity-acquired pneumonia in adults in British hospitals in 1982–1983: a survey of aetiology, mortality, prognostic factors and outcome. Q.J.M.2391987195220
5. Örtquist A., Hedlund J., Grillner L., Jalonen E., Kallings I., Leionen M., Kalin M.Aetiology, outcome and prognostic factors in community-acquired pneumonia requiring hospitalization. Eur. Respir. J.3199011051113
6. Fang G. D., Fine M., Orloff J., Arisumi D., Yu V. L., Kapoor W., Grayston J. T., Wang S. P., Kohler R., Muder R. R., Yee Y. C., Rihs J. D., Vickers R. M.New and emerging etiologies for community-acquired pneumonia with implications for therapy. A prospective multicenter study of 359 cases. Medicine (Baltimore)691990307316
7. Neill A. M., Martin I. R., Weir R., Anderson R., Chereshsky A., Epton M. J., Jackson R., Schousboe M., Frampton C., Hutton S., Chambers S. T., Townl G. I.Community acquired pneumonia: aetiology and usefulness of severity criteria on admission. Thorax51199610101016
8. Fein, A. M., and S. H. Feinsilver. 1995. Nonresolving and slowly resolving pneumonia. In S. H. Feinsilver and A. M. Fein, editors. Textbook of Bronchoscopy. Williams & Wilkins, Philadelphia, 286–301.
9. Örtquist A., Kalin M., Leideborn L., Lundberg B.Diagnostic fiberoptic bronchoscopy and protected brush culture in patients with community-acquired pneumonia. Chest971990576582
10. Lynch P., Sitrin R. G.Non-infectious mimics of community-acquired pneumonia. Semin. Respir. Infect.71992282293
11. Bone R. C., Balk R. A., Cerra F. B., Dellinger R. P., Fein A. M., Knaus W. A., Schein R. M. H., Sibbald W. J.Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest101199216441655
12. Murray T. J., Washington J. A.Microscopic and bacteriologic analysis of expectorated sputum. Mayo Clin. Proc.501975339344
13. Balows, A., W. J. Haussler, Jr., K. L. Herrmann, H. D. Isenberg, and H. J. Shadomy, editors. 1991. Manual of Clinical Microbiology, 5th ed. American Society of Microbiology, Washington, DC. 147–150.
14. Leroy O., Vandenbusche C., Coffinier C., Bosquet C., Georges H., Guery B., Thevenin D., Beaucaire G.Community-acquired aspiration pneumonia in intensive care units: epidemiological and prognosis data. Am. J. Respir. Crit. Care Med.156199719221929
15. Crouch Brewer, S., R. G. Wunderink, C. B. Jones, and K. V. Leeper, JrVentilator-associated pneumonia due to Pseudomonas aeruginosa. Chest109199610191029
16. Ewig S., Ruiz M., Torres A., Marco F., Martinez J. A., Sanchez M., Mensa J.Pneumonia acquired in the community through drug-resistant Streptococcus pneumoniae. Am. J. Respir. Crit. Care Med.159199918351842
17. American Thoracic SocietyHospital-acquired pneumonia in adults: diagnosis, assessment of severity, initial antimicrobial therapy, and preventive strategies. Am. J. Respir. Crit. Care Med.153199517111725
18. Rodrigues J., Niederman M. S., Fein A. M., Pai R. B.Non-resolving pneumonia in steroid-treated patients with obstructive lung disease. Am. J. Med.9319922934
19. Louria D. B., Kaminski T.The effects of four antimicrobial drug regimens on sputum superinfections in hospitalized patients. Am. Rev. Respir. Dis.851962649665
20. Tillotson J. R., Finland M.Bacterial colonization and clinical superinfections of the respiratory tract complicating antibiotic treatment of pneumonia. J. Infect. Dis.1191969597624
21. Örtquist A., Hammers-Berggren S., Kalin M.Respiratory tract colonization and incidence of secondary infection during hospital treatment of community-acquired pneumonia. Eur. J. Clin. Microbiol. Infect. Dis.91990725731
22. Marquette C. H., Copin M. C., Wallet F., Neviere R., Saulnier F., Mathieu D., Durocher A., Ramon P., Tonnel A. B.Diagnostic tests for pneumonia in ventilated patients: prospective evaluation of diagnostic accuracy using histology as a diagnostic gold standard. Am. J. Respir. Crit. Care Med.151199518781888
23. Fàbregas N., Ewig S., Torres A., El-Ebiary M., Ramirez J., Puig de la Bellacasa J., Bauer T. T., Cabello H.Clinical diagnosis of ventilator associated pneumonia revisited: comparative validation using immediate postmortem lung biopsies. Thorax541999867873
Correspondence and requests for reprints should be addressed to Dr. Antoni Torres, Servei de Pneumologia i Al.lergia Respiratoria, Villarroel, 170, E-08036 Barcelona, España. E-mail:

Dr. Arancibia was a 1997 Research Fellow from the Insituto Nacional del Tórax, Santiago de Chile, Chile.

Dr. Ewig was a Research Fellow from the Medizinische Universitätsklinik and Poliklinik Bonn, Bonn, Germany.

Dr. Ruiz was a 1997 European Respiratory Society Research Fellow from the Hospital Clı́nico de la Universidad de Chile, Santiago de Chile, Chile.

Dr. Bauer was a Research Fellow from the Abteilung für Pneumologie, Allergologie und Schlafmedizin, Medizinische Klinik, Bergmannsheil-Universitätsklinik, Bochum, Germany supported in 1999 by IDIBAPS, Hospital Clı́nic, Barcelona, Spain.


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