All SCAP patients hospitalized in our 16-bed ICU were prospectively studied from January 2005 to January 2009 in a 420-bed university and teaching hospital that serves a population of 725,000 inhabitants. According to the SCAP definition by the Infectious Diseases Society of America (IDSA)/American Thoracic Society (ATS), the presence of symptoms of lower respiratory tract infection was required, along with a new infiltrate on chest radiography and no other alternative diagnosis during follow-up, in non-hospitalized patients for at least the previous 72h. The patients were included by a study protocol and the final decision on ICU patient admission was made according to clinical judgement by the physician in charge. The institutional review board and the local ethical committee for clinical research approved the study protocol. The written consent of the patient or the patient's legally authorized representative was obtained before inclusion in the study.

The data collected for each patient were divided into three groups: (1) demographic and clinical data prior to admission; (2) clinical data at the time of admission; and (3) clinical data obtained during the clinical process. The following data were recorded upon admission: age, gender, smoking and alcohol habits, co-morbid illness, antimicrobial treatment prior to hospital admission, clinical symptoms and clinical presentation (body temperature, confusion, respiratory rate, heart rate, arterial systolic pressure, saturation O2, pleural effusion), analytical data (haemoglobin, BUN, glucose, sodium, leukocytes, arterial pH and PaO2/FiO2). During the clinical process the following variables were collected: results of microbial investigations, presence of septic shock, need for mechanical ventilation and non-invasive ventilation, days of mechanical ventilation, renal failure development, radiographic progression – increased involvement in the chest radiography with an additional lung lobe respect to emergency department admission – emphysema, bacteraemia, shock, acute respiratory distress syndrome (ARDS), empiric antibiotic treatment and length of ICU stay. Mortality was evaluated at 28 days.

Co-morbidities were defined as follows: cardiac illness: congestive heart failure with ventricular dysfunction documented by clinical, radiology and echocardiography or presence of valve heart disease or previous coronary artery disease; pulmonary: treatment for asthma or chronic obstructive pulmonary disease (COPD); central nervous system disorders: acute or chronic vascular or nonvascular encephalopathy; diabetes mellitus: diagnosis of intolerance to glucose and treatment with oral antidiabetics or insulin; hepatic: pre-existing viral or toxic hepatopathy or liver cirrhosis; renal: pre-existing renal disease with documented abnormal serum creatinine level outside the period of the pneumonia episode; neoplasic illness: any solid tumour; immunosupression: for severe immunosupression such as manifested by neutropenia (<1.0×109cells/l), or from human immunodeficiency virus (HIV) infection with counting of CD4<350cell/mm3, solid-organ or bone-marrow transplantation, or steroid treatment within the previous 30 days. Alcohol abuse was defined as the ingestion of ≥40g in males and >24g in females per day for at least one year before presentation. Smokers were defined as current smokers of ≥10 cigarettes/day during at least the preceding year.

Admission pneumonia scores were calculated with data of 24h of ICU admission (PSI-Pneumonia Severity Index-, CURB and CURB-65), according to the original publications,5–7 but not used to admit or not the patient in ICU. High-risk patients were defined as those with PSI risk class IV–V, a CURB risk class 3–4 or CURB-65 risk class 3–5 upon admission.

All patients were classified into IDSA/ATS criteria, according to the 2007 IDSA/ATS guidelines,1 and SCAP was defined as the presence of two major criteria (receipt of invasive mechanical ventilation and septic shock with the need for vasopressors), or presence of three minor criteria (respiratory rate≥30breaths/min, PaO2/FiO2≤250mmHg, multilobar infiltrates, confusion and/or disorientation), uraemia (BUN level≥20mg/dl), leukopenia (WBC count<4×109cells/l), thrombocytopenia (platelet count<100×109platelets/l), hypothermia (core temperature≤36°C), hypotension (SBP≤90mmHg; requiring aggressive fluid resuscitation).

Septic shock was defined as sepsis induced hypotension persisting despite adequate fluid replacement. Sepsis induced tissue hypoperfusion is defined as septic shock, an elevated lactate or oliguria.10,11

ARDS was defined as acute onset of bilateral radiographic infiltrates, PaO2/FiO2≤250mmHg, pulmonary artery wedge pressure<18mmHg if this information is available or lack of clinical evidence of left ventricular failure.12

Acute renal failure (ARF) was defined using de RIFLE criteria as tripling of basal creatinine or creatinine≥4mg/dl with an increase of 0.5mg/dl, 75% reduction in glomerular filtration rate, urine output≤0.3ml/kg/h for 24h, or anuria for 12h.13

As a comparative baseline of severity, SAPS II (Simplified Acute Physiology Score) was also calculated.14

In non-intubated patients, the routine sampling for microbiological examination consisted of a sputum, two blood cultures, and a sample of urine for culture and for the determination of Legionella sp. and Streptococcus pneumoniae urinary antigen assays, and pleural fluid culture in case of large pleural effusions. All intubated patients with mechanically ventilation, underwent at least one tracheobronchial aspirate (TBA) or protected bronchoalveolar mini-lavage (mini-BAL).15 Patients with any level of immunosupression or/and bilateral pulmonary infiltrates underwent bronchoscopic sampling techniques (protected specimen brush [PSB] or bronchoalveolar lavage [BAL]). During the period of study, urine, TBA, and blood were processed for polymerase chain reaction (PCR) analysis of Legionella pneumophila.

Sputum was stained for Gram and only cultured with the presence of more than 25 granulocytes and less than 10 epithelial cells. PSB and BAL fluid samples were cultured for aerobic and anaerobic bacteria pathogens, mycobacteria and fungi. Negative bacterial cultures were discarded after 7 days. Micro-organism identification was completed according to standard microbiological methods. Results of quantitative cultures were expressed as colony-forming units per millilitre (cfu/ml), according to published standards.16

Pneumonia aetiology was considered definite whenever one of the following criteria was met: (1) blood cultures yielding a bacterial pathogen; (2) pleural fluid cultures yielding a bacterial pathogen; (3) a positive urinary antigen for Legionella or S. pneumoniae; (4) bacterial growth in cultures of TBA≥105cfu/ml, in PSB≥103cfu/ml, and in mini-BAL and BAL≥104cfu/ml; or (5) specific PCR amplification of L. pneumophila.

Descriptive statistics of demographic and clinical variables included means and standard deviations for quantitative variables, and percentages for qualitative variables. For unadjusted comparisons between or among groups, continuous variables were compared by using Student's t test, if data are normally distributed, or Mann–Whitney U test for non-normally distributed data. Categorical variables were compared by using the chi-square test or Fisher's exact test where appropriate. Analysis of risk factors for mortality at 28 days was performed initially by using a bivariate analysis, whereas a multivariate analysis was performed by logistic regression. A multiple logistic regression model was used to assess the relationships between predictors of mortality and death at 28 day by all causes. Our primary predictors of interest are the general severity score (SAPS II), scores of pneumonia (PSI, CURB and CURB 65) upon admission, and other predictors such as septic shock, ARDS, and acute renal failure within the first 24h of ICU admission.

The list of other candidate predictors was narrowed to include those with a bivariate significance at p<0.2 level, as well as clinically significant variables. A forward selection model, using an entry criterion of p≤0.05, was carried out to limit co-linearity problems. Because the number of deaths was limited (N=56), the model was constrained to a total number of 5–6 degrees of freedom. No confounding was found, so this term was not included in the final model. Model discrimination and predictive power were assessed by the area under the receiver operator characteristic curve (AUC-ROC). Possible over-fitting was evaluated with the Hosmer–Lemeshow goodness-of-fit test. Data are presented as odds ratios (OR) with 95% confidence intervals (CI). p value<0.05 was considered statistically significant for all comparisons. Statistical analyses were performed using SPSS (version 12; SPSS®, Inc., Chicago, IL).

During the period from 2005 to 2009, 242 SCAP patients were prospectively collected. One hundred and sixty-seven patients (69%) were male, and 75 (31%) female, while average age was 56.8±16. Considering co-morbidity, 99 patients (40.9%) had previous COPD diagnosis, 49 patients (20.2%) had congestive heart failure, and 43 patients (17.8%) had diabetes mellitus. Only 12 patients (5%) had chronic renal disease. Upon ICU admission, patient severity by SAPS II was 37.2±15.5 points.

During the first 24h, 150 patients (62%) showed some level of respiratory failure that required ventilation support, and 141 patients (58.3%) displayed involvement of multiple lobes and radiographic progression. Average ICU stay was 11.0±11.8 days, and 28-day mortality rate was 23.1%. Demographic data, risk factors, co-morbidities, clinical data and outcome are shown in Table 1.

Microbiological aetiology could be determined in 173 out of 242 (71.5%) patients. The most frequent pathogen was S. pneumoniae (33.5% of all patients). The second and third most common pathogens were L. pneumophila and Haemophilus influenzae. Pseudomonas aeruginosa was isolated in 8 patients (3.3%). No difference in microbiological aetiology was found between survivor and non-survivor patients. See Table 2.

Demographics and comorbidities associated with death at 28 days are shown in Table 3. Non-survivor patients were older than survivors (64.7±13.4 vs. 54.7±16.1; p=0.0001). In addition, neoplasic disease, immunosupression, chronic renal disease, invasive mechanical ventilation, ADRS, acute renal failure, bacteraemia and septic shock were risk factors of mortality at 28 days. On the other hand, the support treatment with non-invasive ventilation was associated with better survival (Table 4).

After applying severity scores (SAPS II, PSI, CURB and CURB 65) to our population, both SAPS II and higher PSI were associated with significantly higher mortality. Most survivor patients were classified into CURB 1–2 and CURB 65 1–3 groups (Table 5).

A predictive model of mortality applicable during the first 24h of ICU admission was built. This model shows that age in years (OR: 1.04; 95%CI: 1.02–1.06), CURB score 2 and 3 (OR: 3.36; 95%CI: 1.58–7.15), septic shock (OR: 4.03; 95%CI: 1.74–9.74), ARDS (OR: 2.89; 95%CI: 1.25–6.69), and acute renal failure (OR: 2.41; 95%CI: 1.07–5.38) were independent predictive factors of mortality at 28 days in SCAP patients (Table 6).