Etiology of community-acquired pneumonia in a population-based study: Link between etiology and patients characteristics, process-of-care, clinical evolution and outcomes

Background The etiologic profile of community-acquired pneumonia (CAP) for each age group could be similar among inpatients and outpatients. This fact brings up the link between etiology of CAP and its clinical evolution and outcome. Furthermore, the majority of pneumonia etiologic studies are based on hospitalized patients, whereas there have been no recent population-based studies encompassing both inpatients and outpatients. Methods To evaluate the etiology of CAP, and the relationship among the different pathogens of CAP to patients characteristics, process-of-care, clinical evolution and outcomes, a prospective population-based study was conducted in Spain from April 1, 2006, to June 30, 2007. Patients (age >18) with CAP were identified through the family physicians and the hospital area. Results A total of 700 patients with etiologic evaluation were included: 276 hospitalized and 424 ambulatory patients. We were able to define the aetiology of pneumonia in 55.7% (390/700). The most frequently isolated organism was S. pneumoniae (170/390, 43.6%), followed by C. burnetti (72/390, 18.5%), M. pneumoniae (62/390, 15.9%), virus as a group (56/390, 14.4%), Chlamydia species (39/390, 106%), and L. pneumophila (17/390, 4.4%). The atypical pathogens and the S. pneumoniae are present in pneumonias of a wide spectrum of severity and age. Patients infected by conventional bacteria were elderly, had a greater hospitalization rate, and higher mortality within 30 days. Conclusions Our study provides information about the etiology of CAP in the general population. The microbiology of CAP remains stable: infections by conventional bacteria result in higher severity, and the S. pneumoniae remains the most important pathogen. However, atypical pathogens could also infect patients in a wide spectrum of severity and age.


Background
Community-acquired pneumonia (CAP) is a common clinical disorder with an estimated incidence range from 1.6/1,000 to 16/1,000 per year [1][2][3][4]. The majority of pneumonia-related morbidity, mortality, and health care expenditures occur among persons who are hospitalized, however, 50 percent to 80 percent of adult patients with CAP are treated on an ambulatory basis [1,2,5,6].
Initial antibiotic selection for treatment of CAP is empirical, though a thorough understanding of the likely pathogens raises important questions regarding the current care of patients with CAP. Although the microbiology of CAP has probably remained relatively stable over the last decade, indeed could have changed over time due to the success of vaccinations and a population which is getting older, and sicker. Until now the majority of pneumonia etiologic studies are based on selected series of hospitalized patients, and there have been no recent, comprehensive population-based studies including hospitalized patients and outpatients treated by family physicians.
Streptococcus pneumoniae is the most common etiologic agent in hospitalized patients with CAP, particularly those with severe disease [7]. Some studies have found atypical bacteria to be the predominant microorganism for cases of ambulatory CAP [6] and for patients hospitalized with milder episodes of pneumonia [8]. However, population-based studies [1,9] have showed that S. pneumoniae was also the organism most frequently isolated in outpatients, and for each age group, the etiologic profile could be similar among inpatients and outpatients [9]. A better understanding of pathogens causer of CAP could lead to effective efforts to define their natural history and optimize the therapy.
The aim of the present study was to provide an assessment of the etiology of CAP that occurred in the adult population of a defined geographical area. Our second objective was to evaluate the relation that may exist between the different pathogens of CAP to patient characteristics, process-of-care, clinical evolution and outcomes.

Setting, design and study population
The Comarca Interior region, situated in the Basque Country (northern Spain), has a mixed urban, suburban, and rural population of 300,299 (254,523 aged ≥18). The age distribution, educational level, sources of employment, socioeconomic status, and health care services of the population are representative of the overall Basque Country [10]. Health care in this region is provided by the public network, which provides free unrestricted care to nearly 100% of the population.
We conducted a population-based study of all cases of adults (age ≥18 years) CAP within our area during a 15 months period. From April 1, 2006, to June 30, 2007, patients with CAP were recruited from a teaching hospital (Galdakao Hospital), the only hospital in the area, and 150 Family Practice (FP) physicians working in the Comarca Interior region. The adult cases of confirmed pneumonia that occurred during the study period were prospectively and consecutively enrolled in the study.
All eligible participating patients were informed of the study goals and gave informed consent to participate in the study. The project was approved by the ethics committee of Galdakao Hospital Definition of pneumonia CAP was defined as pulmonary infiltrate on chest radiograph not known to be old and symptoms consistent with pneumonia, not acquired in a hospital or a nursing home residence. Patients were excluded if they were known to be positive for human immunodeficiency virus, were chronically immunosuppresed (defined as immunosuppression for solid organ transplantation, postsplenectomy, receiving ≥10 mg/day of prednisone or the equivalent for more than 30 days, treatment with other immunosuppressive agents, or neutropenia, i.e., <1.0 × 10 9 /L neutrophils), or who had been discharged from an acute care hospital or an on-site subacute care unit or from palliative care within the previous 14 days.
Cases were confirmed if there was a radiological finding suggestive of pneumonic infiltrate as reviewed by two members of the research team (PPE and AC).

Measurements and management evaluation
Management for patients who initially visited an FP was left to the FP's discretion. Patients attending the emergency department of Galdakao Hospital during the study period were managed according to a clinical guideline [11]. Clinical and demographic characteristics (age and sex) of each patient were recorded, along with all of the variables for the CURB-65 (Confusion, Urea nitrogen, Respiratory rate, Blood pressure, age ≥65 years) score [12] within the first 24 hours after diagnosis.
Process-of-care variables included the therapy employed (initial choice of antibiotic treatment, whether it was consistent with recommendations of the Spanish Society of Pneumology (SEPAR) [13] or the Infectious Diseases Society of America/American Thoracic Society (IDSA/ATS) guidelines [14], and antibiotics taken prior to admission), and the duration of antibiotic therapy.
Clinical evolution measures included treatment failure (defined as the development of clinical deterioration with hemodynamic instability, demonstrated respiratory failure or the appearance of it, required mechanical ventilation, demonstrated radiographic progression of pneumonia or the appearance of a new infectious foci, or had persistent fever or the reappearance of fever if change in treatment was needed) [15]; severe sepsis (defined as sepsis associated with organ dysfunction [16]); septic shock (defined as systolic arterial tension <90 mmHg and requirement of vasopressors for a minimum of 4 hours) [17]; hospitalization; admission to the intensive care unit (ICU); need for mechanical ventilation.
Outcome measures included vital status at 30 days after diagnosis; in-hospital mortality; hospital readmission within 30 days; length of hospital stay; length of time needed to return to normal daily activities; and subsequent hospitalization for patients initially treated as outpatients.
Vital status and readmission information for all patients were initially determined by telephone interviews up to 90 days after discharge. All reported deaths and dates of death were confirmed by a review of medical reports, public death registries, or both.

Bacteriologic studies
The strategy for bacteriologic diagnosis included two blood cultures, a urinary antigen during the acute phase of the infection, and serological tests for atypical bacteria during the acute and remittance phases. A mixed infection was defined as pneumonia due to more than one pathogen.
In the first three days after diagnosis and 4 to 6 weeks thereafter, sera were collected, and tested for the presence of IgG and IgM to Mycoplasma pneumoniae, Chlamydia pneumoniae, Chlamydophila psittaci, Coxiella burnetii, and Legionella pneumophila. Antibodies to M. pneumoniae, C. Burnetti, and L. pneumophila were tested by immunofluorescent antibody assay and to Chlamydophila spp by microimmunofluorescence assay. Tests to detect influenza virus types A and B, parainfluenza virus types 1 to 3, adenovirus, and respiratory syncytial virus were performed with monoclonal antibodies for direct immunofluorescence assay. The collected urine samples were stored at -70°C or −20°C until testing. The detection of L. pneumophila was performed by means of an enzyme immunoassay system and Streptococcus pneumoniae by means of an immunochromatographic membrane assay (Binax Inc; Scarborough, ME) in a nonconcentrated urine sample.
An etiologic diagnosis was considered to be definitive if one of the following criteria were met: 1) isolation of respiratory pathogen in a sterile specimen (blood and pleural fluid); 2) four-fold rise in IgG titers for M. pneumoniae, C. pneumoniae, C. psittaci, C. burnetii, and L. pneumophila; 3) a single increased IgM titer for M. pneumoniae (≥1:16) or C. pneumoniae (≥1:10) or C. psittaci (≥1:32) or C. burnetii (≥1:64); 4) positive urinary antigen for L. pneumophila type 1 or S. pneumoniae; and 5) a positive result for one respiratory virus. An etiologic diagnosis was considered to be presumptive if a single increased IgG titer for M. pneumoniae (≥1:180), for C. pneumoniae (≥1:512), for C. psittaci (≥1:128), for C. burnetii (≥1:512), for L.pneumophila (≥1:256). In the present study the bacteriological examination of sputum was not compulsory, however, an etiologic diagnosis was considered to be presumptive if a validated sputum sample yielded one or more bacterial strains.
Since not all patients without etiologic diagnostic were evaluated by all diagnostic methods, the following minimal set of diagnostic tools was required for inclusion in the analysis: 1) urinary antigen and 2) paired serology or only 2°serology.

Statistical analysis
Pathogens detected were compared among inpatients and outpatients, and according to age groups (18-44, 45-64, 65-74 and >74 years). Characteristics of patients, and process-of care and outcomes were compared in relation Clinical evolution and outcomes were also compared according to etiologic categories adjusting by age, sex, and comorbidities. For continuous dependent variables (length of hospital stay and length of time needed to return to normal daily activities), the general linear model was used. Due to the non-normal distribution of these dependent variables, a log transformation was performed. Parameter estimates and standard errors were estimated after exponentiation. For all other dependent variables, all dichotomous, multivariate logistic regression models were performed. Odds ratios (OR) and 95% confidence intervals (95% CI) are presented.
All effects were considered significant at P < 0.05. All statistical analyses were performed using SAS for Windows statistical software, version 9.1 (SAS Institute, Inc., Carey).
Microbial etiology by age group is shown in Table 2. A trend toward a higher proportion at older ages was found for Pneumococcal infection (trend P < 0.0001). In contrast, for atypical infection a trend toward a low proportion at older ages was found (trend P < 0.0001). Virus infection as a group was significantly more frequent among patients aged ≥65 years than among patients aged <65 years (22.2% vs. 7%; P < 0.0001).
Characteristics of patients according to etiologic categories are provided in Table 3. In contrast with patients infected by atypical agents, those infected by conventional bacteria were elderly, had more comorbidities, or had a higher severity of CAP at diagnostic. The rate of patients immunized against pneumococcal infection was very low for all groups.
Process-of-care is shown in Table 4. First-line antibiotic therapy was a monotherapy in 92.8% patients. Antibiotics initially prescribed were overcoat quinolones (60%) for all etiologic categories. Almost a quarter of CAP by atypical agents were treated with betalactam antibiotics.
Clinical evolution and outcomes are provided in Table 5. In contrast with patients infected by atypical agents, those infected by conventional bacteria had a higher hospitalization rate, showed worse evolution and higher length of hospital stay . However, among patients infected by atypical agents 25.8% needed hospitalization with an inhospital mortality rate of 5.1%. There was no difference according to etiologic categories for inhospital mortality, readmission within 30 days, and the mean duration of return to daily activity. No deaths were found for patients with mixed, virus or L. pneumophila infections, and none of them were admitted to an ICU. After adjusting by sex, age, and comorbidities, the previous analyses yielded similar results ( Table 6). All the patients who passed away, both in hospital as well as at 30-days had a definitive etiologic diagnostic. Out of all patients admitted at the ICU [15], for 14 (Table 5) the diagnostic was definitive, while 1 patient was not diagnosed etiologically: blood culture, negative; urinary antigen, negative; sputum culture, negative; serologic test 1°a nd 2°, negative. Out of all patients with septic shock [10], for 9 (Table 5) the diagnostic was definitive, while 1 patient was not diagnosed etiologically: blood culture, negative; urinary antigen, negative; sputum culture, negative; serologic test 1°and 2°, negative. Out of all patients with severe sepsis (131), for 105 (Table 5) the diagnostic was definitive, while 26 patients were not diagnosed etiologically: blood culture, negative for 17 (9 were not done); urinary antigen, negative for all; sputum culture, negative for 5 (21 not done); serologic test 2nd sample, negative for all, serologic test 1st sample, negative in 16 (10 not done).

Discussion
In our population-based study we assessed the etiology for CAP, in search of evidence and insight into the different roles and pathlines concerning diverse pathogens. Our results confirm many already well documented features, i.e. -S. pneumoniae was the most common pathogen, causing a greater degree of severity. We bring to light as a new finding the fact that, viruses, mixed and L. pneumophila infections do not imply greater severity. What is more, atypical pathogens and the S. pneumoniae are found in a wide spectrum of severity and age levels. Our findings emphasize the part that atypical organisms play. The study contains distinctive high points. It is a prospective population-based survey that includes all cases of CAP occurrence in the overall total population of a defined geographical area, and consequently avoids any selection bias. We surveyed a large number of hospitalized and ambulatory unselected adult patients for whom we had gathered highly detailed data. Furthermore, our study is original in that it links the etiology to the clinical evolution and the consequent outcomes.
In agreement with other studies S. pneumoniae is still the predominant infection [1][2][3]18,19]. It was the most common agent in hospitalized patients and the second most common cause of illness in outpatients. C. burnetti had unusual frequencies in our study due to regional epidemiological factors which are markedly different from those in other areas of the western world [20,21]. In our region the sheep are the usual reservoir and the infection source for other domestic and wildlife species and the human population [22]. M. pneumoniae was the third most common pathogen isolated, although it was less common than in some earlier studies [23,24] probably due to cyclicity of Mycoplasma epidemics. Consistent with previous studies [1,19,25], the frequency of L. pneumophila infection was very low (<2%) and was not implicated in greater severity. The mixed pneumonia group is a heterogeneous group in which proportions varied among studies [9,26]. We found mixed infections in 9% of patients, and, like others [19], nearly all of our cases were non-severe CAP, contrary to the view that mixed infections can be severe [27,28]. Viral infections were attributable to old age patients with chronic obstructive pulmonary disease. In contrast with previous studies [29], viral etiology was associated with low severity of illness, possibly related to the high rate of influenza immunization in this group. In previous population-based studies about CAP in adult population [1][2][3] H. influenzae infection has been diagnosed in 1%-10%. The widely varying results are more likely to reflect differences in the bacteriologic methods than true differences in the etiology. In the present study the bacteriological examination of sputum was not compulsory. This approach implicated a failure to detect the true incidence of H. influenza, S. aureus, and P. aeruginosa. Moreover, the infection by S. aureus is always very infrequent in CAP, especially among outpatients.
According to etiologic categories the profile of patients is similar to the combined information from previous studies of CAP [1][2][3][4]8]: infections by atypical agents were characterized by milder episodes, while more severe infections were caused by conventional bacteria. In fact, patients with conventional bacteria have higher hospitalization rate, higher CURB65 scores at admission, more bilateral radiographic involvement and more adverse clinical evolution and outcomes such as severe sepsis, septic shock, ICU admission and longer length of stay. However, our findings confirm that atypical pathogens and the S. pneumoniae are present in pneumonias of a wide spectrum of severity and age, as had been stated in a previous study [30,31]. Almost 26% of patients infected with any atypical agents needed to be hospitalized and, like others [32], atypical pathogens were present in a considerable proportion of inpatients with CAP (30.6%). Conversely, among patients with conventional bacteria almost 25% were treated at home with excellent evolution. In the present study the mortality can be compared between conventional bacteria and those atypicals who require a convalescent serum, because all the patients who passed away, both in hospital as well as at 30-days, has a definitive etiologic diagnostic. Therefore, the assessment of this outcome is not biased towards a better outcome for atypical organisms. In our study antibiotic treatment was consistent with the Spanish guideline strategy [13], which recommends targeting pneumococcal and atypical infection for inpatients and outpatients. Likewise, the IDSA/ATS [14] guideline strategy recommends empirical antibiotic treatment to cover typical and atypical pathogens. Our data show that S. pneumoniae as well as atypical agents were causing pneumonia in hospitalized and ambulatory patients of all ages, questioning the empirical use of an antibiotic to cover atypical and not atypical agents in ambulatory pneumonia. There is concern about this question because evidence is lacking that clinical outcomes are improved by using antibiotics against atypical pathogens in all-cause non-severe CAP [33], and a greater dependence on quinolones as initial therapy raises concerns regarding the potential for the development of antibiotic resistance.
Our study has some limitations. We made a great effort to identify all patients with CAP in the catchment area: all local health care centers and the referral hospital participated in the reporting of patients; all primary health care centers were closely supervised by a local FP, and all information was coordinated by a member of the research team. However, full participation of all 150 FP was not obtained and, therefore, we must assume as a limitation the fact that not all real cases have been included. As in most pneumonia studies [1,6,25,29,34], specific etiologic diagnoses could not be assigned for

Conclusions
Our study updates the information concerning the epidemiology and management of CAP. S. pneumoniae remains the most important pathogen, and almost 50% of the patients were infected with C. burnetti, M. pneumoniae, Chlamydia sp, or L. pneumophila. Although all identified agents infect patients of a wide spectrum of severity and age, infection by conventional bacteria showed higher severity, and there is a high incidence of S. pneumoniae in elderly patients. Adjusted analyses were performed comparing the clinical evolution and outcomes between Conventional Bacteria vs. the rest of groups, and between Conventional Bacteria vs. Atypical Agents, adjusting for age, sex and comorbidities. ║ For these two continuous dependent variables, the general linear model was used, and due to the non-normal distribution of these variables, a log transformation was performed. Parameter and standard errors were estimated after exponentiation. Therefore, the parameter estimated is the ratio between the length of hospital stay or the length to return to daily activities of Conventional Bacteria vs. the reference group.