Patient demographics and ICU characteristics
Out of the 250 patients enrolled, 241 were mechanically ventilated for > 2 days; one additional patient was excluded due to missing clinical information crucial for analysis (Additional file 3: Fig. S1, Additional file 7). The average age among the remaining 240 patients was 64 years, and a slight male predominance was seen. The mean LoS and MV duration were 20 and 9 days, respectively. We addressed the potential differences in patient characteristics between those admitted to SICU and MICU, respectively. Significant differences were observed: patients in the SICU were hospitalized and ventilated significantly longer, but displayed significantly lower mortality rates compared to those in the MICU. Additionally, patients with respiratory failure were significantly more likely to be admitted to the MICU, whereas subjects requiring an emergency surgery or having a history of recent trauma – to the SICU, as expected by the ICU profiles. Other characteristics, including most frequently observed comorbidities, history of smoking, alcohol abuse, use of antibiotics and of immunosuppressive medications, did not differ significantly among the groups. VAP was detected in approximately one-third of the patients irrespective of the ICU type (Table 1).
Few clinical indicators such as clinical pulmonary infection score (CPIS) and disease severity scores could not be analyzed as some of the clinical parameters required for their calculation were not available, as they were not recorded for every patient, or were recorded at different time points during the ICU stay.
Bacterial pathogens isolated from the ETA samples
Upon study commencement the most recent, yearly assessed global rates of local bacterial susceptibility to the antibiotics were provided, summarized and adapted for the purposes of the study to the organisms most commonly isolated from the ETA in Additional file 1: Table S1. ETA sampling frequency was approximately 2 out of every 3 MV days in the overall ICU population studied. The ETA microbiological culture results for patients were grouped into three main categories: yielding S. aureus (n = 71), Gram-negative bacteria (18 species; n = 75), or displaying no bacterial growth or only NRF (n = 115). 35.0% of the study subjects (84/240) yielded a single pathogen throughout the study, whereas 12.9% had two, and 4.2% yielded three or more pathogens. The most common bacterial species in ETAs was S. aureus, isolated from 29.6% of the study patients (71/240), and 56.8% (71/125) of patients with pathogen-positive ETAs (Additional file 9). 40.8% (29/71) of S. aureus positive patients carried MRSA. This was similar to the global MRSA rate (38%) reported for Lahey Hospital and Medical Center in the year preceding the study (Additional file 1: Table S1). The three most commonly detected Gram-negative pathogens, K. pneumoniae, P. aeruginosa and E. coli, were present in ETA samples of 18.3% (44/240), and 15 other Gram-negative species were detected in 22.1% (53/240) of patients (Table 2). 47.8, 43.8, and 42.7% of S. aureus, K. pneumoniae, and P. aeruginosa respectively were detected in the ETA by day 2 of MV. E. coli, S. maltophilia, E. cloacae, and E. aerogenes appeared later in the course of MV, with only 17–33% of these species isolated within the first two MV days (Fig. 1).
ETA microbiology during VAP events
Eighty-five of the 240 study patients experienced VAP during MV (Additional file 8). In 41 (48.2%) of these patients, potential pathogenic bacteria were recovered from the ETA samples at the time of VAP diagnosis, ±2 days (Fig. 2A). S. aureus was isolated from 18 patients; however, in 7 of these cases Gram-negative pathogens, and in one case S. pneumoniae were also isolated (Fig. 2B, Additional file 2: Table S2, Additional file 4: Fig. S2, Additional file 5: Fig. S3). Twenty-seven patients (65.8%) had at least one Gram-negative pathogen, the most common ones being P. aeruginosa, E. coli, and S. maltophilia (Fig. 3 and Additional file 2: Table S2).
Both medical and surgical ICU had similar proportions of S. aureus and Gram-negative VAP cases (Table 1). In 51.8% of the VAP patients, no culturable bacteria (other than NRF) were detected in the ETA (Fig. 2A).
Association of airway colonization with subsequent pneumonia
Patients yielding ETA samples with S. aureus and Gram-negative (pooled) organism were divided into heavily or lightly colonized. Seventy-one of the 240 patients had at least one ETA sample positive for S. aureus: 37 were heavily colonized and 34 were lightly colonized. 12 of the 37 heavily colonized patients (32.4%) developed VAP associated with S. aureus during the VAP-relevant period, compared to only 6 out of 34 lightly colonized patients (Fig. 4). The difference between progression to VAP from heavily colonized (32.4%) compared with lightly colonized (17.6%) did not reach statistical significance, possibly due to low sample size (p = 0.1808). Notably, out of the 18 total VAP cases with S. aureus present 15 yielded at least one ETA with only S. aureus at any time during the VAP-relevant period.
Of the 71 patients yielding S. aureus positive ETAs, 41 had MSSA, 27 MRSA, 2 had both, and for 1 patient the isolate was not received and its methicillin resistance could not be assessed. 48.6% (18/37) of the S. aureus from ETAs with heavy colonization were MSSA, and 67.6% (23/34) with light colonization were MSSA. Patients heavily colonized only with MSSA progressed to VAP in 44.4% (8/18) of cases, while only 23.5% of patients heavily colonized with MRSA did so (4/17, p = 0.289). In total, MSSA was implicated in 11 out of 18 VAP cases with S. aureus present, and 7 out of 10 where only S. aureus was isolated during the whole VAP-relevant period. Notably, in all S. aureus VAP patients colonizing and VAP strains had the same methicillin resistance status.
The first isolation of S. aureus in ETA occurred early during MV (median: 2 days) (Fig. 1), and no temporal difference was observed between MRSA and MSSA carrying patients. In patients who later developed S. aureus-only VAP (9 patients: 10 total, and 1 excluded due to insufficient ETA coverage prior to VAP event), detection of S. aureus in the ETA preceded VAP clinical symptoms by a median value of 4 days. Of these 9 patients, only 1 did not have an ETA positive for S. aureus prior to the S. aureus-only VAP event (Fig. 5, Additional file 4: Fig. S2).
Out of the 75 patients with ETA samples colonized with any Gram-negative pathogen, 60 were heavily colonized and 15 were lightly colonized. Eighteen of the 60 Gram-negative heavily colonized progressed to Gram-negative VAP (30.0%), while none (0.0%) of the 15 lightly colonized did so (p = 0.0156) (Fig. 3). In 94.4% (17/18) of Gram-negative VAP cases patients were colonized with at least one of the species causing the infection. On average, Gram-negative organisms were first detected in the ETA 2.5 days (median value) before the onset of VAP.
Impact of VAP on duration of ventilation, hospital stay and mortality
Regardless of presence of ETA colonization, patients with at least one VAP episode required MV support and hospital care for twice as long (mean 13.0 and 16.6 days, respectively) as those who did not develop VAP (mean 6.9 and 25.1 days) (Fig. 6A). This trend was seen for all VAP patients regardless of the type of ICU (medical or surgical) (Fig. 6B-C), pathogen presence and causative bacterial species (Fig. 7A-D), with the exception of the Gram-negative VAP group, where no significant difference in hospital LoS was seen. Although statistical significance was not reached, notable observations regarding all-cause mortality were made. Whereas overall mortality rate between VAP and non-VAP patients were similar (Fig. 6A), an increase of approximately 10% was seen in VAP patients whenever a pathogen was implicated (Fig. 7B-D). This was not true for the bacteria-negative/NRF VAP group, where VAP was paradoxically associated with a 10% decrease in mortality.
Interestingly, a temporal difference was observed between bacterial pathogen-positive VAP peaking in November–December, and bacterial pathogen-negative ones in the January–February period (Additional file 6: Fig. S4).