Invasive and Non-Invasive Diagnostic Approaches for Microbiological Diagnosis of Hospital-Acquired Pneumonia

Otavio T. Ranzani; Tarek Senussi; Francesco Idone; Adrian Ceccato; Gianluigi Li Bassi; Miquel Ferrer; Antoni Torres


Crit Care. 2019;23(51) 

In This Article


Of the 488 patients enrolled during the cohort period, we excluded 288 (59%) patients who were diagnosed with pneumonia while receiving mechanical ventilation (i.e., VAP). Therefore, we analyzed 200 (41%) patients with HAP.

Patient Characteristics

The main clinical characteristics upon ICU admission and at onset of HAP are shown in Table 1. Mean age was 66 years, and there was a high proportion of males. Approximately one third had a chronic comorbidity. The main cause of ICU admission was acute respiratory failure followed by shock and postoperative status. One hundred twenty-two patients (61%) required iMV after the onset of HAP (Figure 1), and 72 (59%) intubations occurred within 24 h of diagnosis. The median ICU length of stay was 13[7–26] days, and 85 (43%) patients died in the hospital. Patients who needed iMV after HAP diagnosis presented higher hospital mortality than those who did not [62 (51%) vs. 23 (30%), p = 0.003].

Figure 1.

Time flow-chart for the microbiological assessments performed in 200 patients with hospital-acquired pneumonia. BAL bronchoalveolar lavage, EAT endotracheal aspirate, FBS fiberoptic-bronchoscopy, FBAS fiberoptic-bronchoscopy aspirate, HAP hospital-acquired pneumonia, iMV invasive mechanical ventilation

Diagnostic Approach

In the 200 patients with HAP, 89% underwent at least two methods for microbiological assessment (median 3 [2–4] methods). Patients who required iMV had a higher number of microbiological assessments than those who did not (3 [2–4] vs. 2 [2,3], p < 0.001, respectively). Respiratory samples were obtained in 186 (93%) patients, and at least two respiratory methods were applied in 40%. Blood cultures (79%), urinary antigen (48%), and FBAS (47%) were the methods most commonly applied to microbiological assessment (Figure 2, Table 2, and Additional file 1: Table S1). Sputum and BAL were performed in almost one third of patients, while 18% had pleural liquid cultures. Sputum, EAT, FBAS, and BAL were the methods that obtained the highest proportions of positivity (Figure 2, Table 2, and Additional file 1: Table S1), followed by pleural liquid, blood culture, and urinary antigen testing.

Figure 2.

Sampling methods and corresponding positivity in a whole cohort, b patients not requiring invasive mechanical ventilation, and c patients requiring invasive mechanical ventilation. BAL bronchoalveolar lavage, EAT endotracheal aspirate, FBAS fiberoptic-bronchoscopy aspirate, iMV invasive mechanical ventilation. *Percentage among those in whom the method was performed

Microbiological diagnosis was possible in 99 (50%) patients. Patients who required iMV had a higher proportion of microbiological diagnosis than those who did not (56 vs. 40%, P = 0.027, Table 2). Thirty-eight (19%) patients received a new antibiotic before sample collection and had a lower proportion of microbiological diagnosis than those who did not (34 vs. 53%, p = 0.036). Overall, the most common pathogens identified were Gram-negative non-fermenting bacteria (39/99, 39%), followed by Staphylococcus aureus (24/99, 24%) and Gram-negative enteric bacteria (24/99, 24%). The prevalence of polymicrobial HAP was 17% (17/99), while 40% had a MDR pathogen. The distribution of causative pathogens was similar in those who required iMV and those who did not (Table 2). The cross-tabulation of different methods for microbiological assessment and their agreement on the same pathogen, when positive, are shown in Figure 3. The average overall agreement was 80% (40/50). Indeed, there was 85% agreement for sputum with other respiratory samples (11/13), 80% for EAT (8/10), 81% for FBAS (13/16), and 91% for BAL (10/11).

Figure 3.

Distribution and agreement of different sampling methods (a cross-tabulation of different methods; b agreement on the same pathogen when both methods were positive). Square colors divided as dark blue for agreement ≥ 75%, blue for agreement between ≥ 50 and < 75%, light blue for agreement between ≥ 25 and < 50%, and grey for agreement < 25%. BAL bronchoalveolar lavage, EAT endotracheal aspirate, FBAS fiberoptic-bronchoscopy aspirate

The majority of microbiological diagnoses were determined by only one method (69/99, 70%), with differences among those who required iMV and those who did not (p = 0.015). FBAS was the only method responsible for the diagnosis of 42% (29/69) patients, followed by EAT (23%), sputum (15%), BAL (9%), and blood culture (7%).

One hundred twenty-five (63%) patients underwent invasive sampling, of whom 78 (39%) were applied fiberoptic-bronchoscopy while not receiving iMV (Figure 1). Patients who required iMV after invasive sampling were more severe at HAP diagnosis (Additional file 2: Table S2). There was no significant difference in the proportion of final microbiological diagnoses when stratifying by fiberoptic-bronchoscopy when receiving or not receiving iMV (p = 0.112); however, among the patients who did not require iMV, the rate of microbiological diagnosis was 10% higher (95% CI, − 12 to 32%) in those who underwent fiberoptic-bronchoscopy. When stratifying patients according to non-invasive (sputum and EAT) or invasive (FBAS and BAL) respiratory methods, we observed higher proportions of microbiological diagnoses in those who underwent at least one invasive method (56 vs. 39%, risk difference 17%, 95% CI, 3–31%, p = 0.018), mainly due to those who required iMV.

Antibiotic Management and Duration

The majority of patients received the initial antibiotic regimen in accordance with the 2005 ATS/IDSA guidelines; empiric antibiotic treatment was adequate in 71% (70/99 patients) (Table 3). Patients who had a microbiological diagnosis more frequently changed their empirical antibiotic regimen (P = 0.006), driven by de-escalation (30 vs. 8%). However, patients who had a microbiological diagnosis also received longer total antibiotics duration than patients without microbiological diagnosis, although similar duration when considered only the empiric antibiotic scheme.