Microorganisms and clinical outcomes of early- and late-onset ventilator-associated pneumonia at Srinagarind Hospital, a tertiary center in Northeastern Thailand

Back ground: Ventilator-associated pneumonia (VAP) is a common nocosomial infection in intensive care unit (ICU). Local microbiological surveillance of pathogens and resistance patterns for early-onset VAP (EOVAP) and late-onset VAP (LOVAP) will help to choose appropriate empiric antibiotics. Objective: To compare the multi-drug resistant (MDR) pathogens, treatment outcomes, and factors associated with hospital mortality of VAP. Method: A cross-sectional study between 1 January 2015 and 31 December 2017 at Srinagarind hospital, Khon Kaen University was conducted. The demographic data, causative pathogens, hospital length of stay (LOS), ICU LOS, mechanical ventilator (MV) days, and hospital mortality were retrospectively reviewed. Results: One hundred and ninety patients were enrolled; 42 (22%) were EOVAP and 148 (78%) were LOVAP. Acinetobacter baummanii was the most common pathogen in both groups (50 % EOVAP vs 52.7% LOVAP). MDR pathogens were signicant greater in LOVAP (81.8 %) than EOVAP (61.9%) (p = 0.007). The EOVAP had a signicantly better ICU LOS (median 20.0 (11.0, 30.0) vs. 26.5 (17.0, 43.0) days), hospital LOS (median 26.5 (15.0, 44.0) vs. 35.5 (24.0, 56.0) days) shorter MV days (14.0 (10.0, 29.0) vs. 23.0 (14.0, 35.5) days) and lower hospital mortality (11.9 % VS 27.7%) than LOVAP ( p < 0.05). The factor associated with hospital mortality was having simplied acute physiology score (SAP) ≥ 40 with an adjusted odds ratio (aOR) of 2.22 (95%CI, 1.08-4.54, p = 0.02). Conclusion: LOVAP had signicantly higher MDR pathogens, MV days, ICU LOS, hospital LOS and hospital mortality


Background
Pneumonia is the most common hospital-acquired infection with a prevalence of approximately 22% [1,2]. Ventilator-associated pneumonia (VAP) is pneumonia developing after 48-72 hours of endotracheal intubation [3][4][5]. VAP is the most common nosocomial infection, developed in about 5-40 % of mechanically ventilated patients [5][6][7]. Data from the International Nosocomial Infection Control Consortium (INICC) collected summary data from 50 countries including Southeast Asia during 2010-2015 indicated the VAP rate was 13.1 per 1000 mechanical ventilator-days in the medical and surgical intensive care unit (ICU) [8]. Similar results of Reechaipichitkul et al who determined that VAP rates in Srinagarind Hospital, Khon Kaen University, a tertiary-care hospital in northeastern Thailand were 13.6 and 12.6 per 1000 mechanical ventilator-days in 2008 and 2009. This study also demonstrated that more than half of the costs of nocosomial treatment in 2008 and 2009 were the costs for hospital acquired pneumonia (HAP) and VAP, 16.8 and 17.5 million Baht [9]. Melsen WG et al performed a meta-analysis and suggested that overall attributable mortality in mechanical ventilator patients from VAP was 13% [10].
VAP was categorized into early-onset VAP (EOVAP) and late-onset VAP (LOVAP) depending upon when it occurred on which days after hospitalization. The cutoff point of a range 4-7 days onset varied across the studies [11][12][13][14][15][16]. Recent guideline for HAP and VAP management from The Infectious Disease Society of America (IDSA)/American Thoracic Society(ATS) and the International ERS/ESICM/ESCMID/ALAT use the cutoff point of 5 days after hospitalization [2,17,18]. It is believed that in EOVAP, the causative pathogen was not drug-resistant bacteria such as Streptococcus pneumoniae, Haemophilus in uenzae, antibiotic-sensitive enteric gram-negative bacilli or methicillinsensitive Staphylococcus aureus (MSSA). There is a greater risk that the causative organisms in LOVAP are multidrug-resistant (MDR) such as Acinetobacter baumannii, Pseudomonas aeruginosa, methicillinresistant S. aureus (MRSA), extended-spectrum beta-lactamase-producing bacteria (ESBL) and other gram-negative bacilli [5,17,19,20]. The prevalence of MDR pathogens between EOVAP and LOVAP in several studies remained a controversy. Several studies demonstrated that EOVAP had a signi cantly lower prevalence of MDR pathogens [21][22][23]. Subsequent studies, however, did not show a signi cant difference in MDR pathogens between EOVAP and LOVAP groups [11,12,14,24]. Therefore, the study was conducted and aimed to compare the pathogens, clinical characteristics, treatment outcomes between EOVAP and LOVAP groups, and factors associated with hospital mortality.

Methods
A cross-sectional study was conducted at Srinagarind Hospital, Faculty of Medicine, Khon Kaen University which is a 1466-bed tertiary care center in Northeast Thailand. The study was approved by the Human Research Ethics Committee, Khon Kaen University (approval number HE611281). All VAP patients recorded by the infectious control (IC) unit from January 1, 2015, to December 31, 2017, were enrolled.

Study subjects
VAP was diagnosed by the following criteria: 1) a pulmonary infection occurring 48 hours after mechanical ventilation 2) new pulmonary in ltration on chest radiograph 3) at least two of the three following characteristics: temperatures > 38.3 °C or < 36.5 °C, purulent tracheal secretions, and leukocytosis (white blood cell > 12,000 cells/mm 3 ) or leukopenia (white blood cell < 4,000 cells/mm 3 ) [4,25].{, 2005 #41} The exclusion criteria were as following: 1) patients who had previous abnormal chest imaging including pulmonary edema, adult respiratory distress syndrome, pulmonary embolism, alveolar hemorrhage, pulmonary tuberculosis, and recent pneumonia 2) Immunocompromised patients who received any immunosuppressive agents, chemotherapy, or prednisolone equivalence ≥ 15 mg/day

Data collection
The medical records of demographic data, hospital department, laboratory results, chest radiological ndings, microbiological pro les, tracheostomy tube placement, hospital length of stay (LOS), intensive care unit (ICU) LOS, mechanical ventilator (MV) days and hospital mortality were reviewed.
De nition and outcome EOVAP de ned as VAP developed before 5 calendar days of hospitalization while LOVAP was VAP occurred at least 5 calendar days of hospitalization. Multi-drug resistant (MDR) bacteria were de ned as organisms that resisted al least 3 classes of antibiotics [26]. MDR pathogens included extended-spectrum beta-lactamase-producing (ESBL) bacteria, carbapenem-resistant enterobacteriaceae (CRE), MRSA, and other MDR bacteria that were reported from the microbiological laboratory. The causative organisms were de ned as one or more of the following: 1) an isolated organism from hemoculture 2) an isolated organism from pleural effusion 3) an isolated numerous growth organism on a semiquantitative method or isolated organism on the quantitative method i.e. endotracheal aspirate > 10 5 colony-forming unit (CFU)/ml, bronchoalveolar lavage > 10 4 CFU/ml or protected specimen brush ≥ 10 3 CFU/ml. Hospital mortality was death occurring during the same admission of VAP diagnosis.
The primary outcome was to compare the MDR pathogen between EOVAP and LOVAP. The secondary outcome was to compare causative pathogens, hospital length of stay (LOS), ICU LOS, mechanical ventilator (MV) days, and hospital mortality between EOVAP and LOVAP. Factors associated with hospital mortality of VAP were identi ed.

Statistical analysis
The categorical data were shown as numbers and percentages. The normal distributed continuous data were presented as mean and standard deviation (SD) while the non-normal distributed data were presented as the median and interquartile range (IQR). A comparison of category data used the Chisquare test and Fisher's exact test depending on data. The nonparametric data used the Mann-Whitney U test for comparison. The factors associated with hospital mortality in VAP subjects were evaluated by univariate logistic regression analysis. The stepwise backward multiple logistic regression analysis of factors with a p-value <0.2 on univariate analysis or factors with previous reports of clinical signi cance was performed. Crude odds ratio (cOR) and adjusted odds ratio (aOR) with their 95% con dence intervals (95% CI) were demonstrated. A p-value of less than 0.05 was considered statistically signi cant The statistical analysis was performed by Stata version 10.1(StataCorp, Texas, USA).

Patients
During the study period, 190 patients were diagnosed as VAP. Forty-two patients were EOVAP and 148 patients were LOVAP. The mean (SD) age of these was 64.3 (16.2) years. Males were 127 (66.8%) and females were 63 (33.2%). One hundred and seven subjects were admitted to the Medicine Department (96 medical ICU and 11 general medicine ward). Eighty-three subjects were admitted to the Surgical Department (73 surgical ICU and 10 general surgery ward). One hundred and forty-eight patients had an underlying disease. The common underlying diseases were hypertension (41.6%), diabetes mellitus (27.4 %), cardiovascular disease (26.8%). The mean (SD) of the simpli ed acute physiology score (SAP) II score was 43.7 (13.3). Lobar pneumonia was the most common nding on chest radiography (75.8%). Pleural effusion developed in 28.4% of all subjects. The demographic data of EOVAP and LOVAP patients were shown in table 1. LOVAP patients had a higher mean age and more comorbidities than EOVAP patients while the chest radiographic ndings were similar between groups.
The proper empiric antibiotics were used to treat 130 (68.4%) study subjects; 61.9% of EOVAP and 70.3% of LOVAP. The percentage of proper empiric treatment was similar between groups (p = 0.30). (Table 2

Discussion
The study revealed that the most common pathogens were a gram-negative organisms. A. baumannii, K. pneumoniae, P. aeruginosa were common pathogens in both groups while S. maltophilia was increased in late-onset VAP. The pathogens from this study did not differ between EOVAP and LOVAP. The results of this study were similar to other tertiary centers in Thailand [27,28]. Of these, A. baumannii, K. pneumoniae, P. aeruginosa were the common pathogens of VAP. These studies, however, did not address the causative organisms into early-onset VAP and late-onset VAP. Three studies from different tertiarycare centers of India had results similar to the present study [14,15,29]. A. baumannii, K. pneumonia and P. aeruginosa were common pathogens in both EOVAP and LOVAP. The pathogens of EOVAP from this study differed from pathogens mentioned in the recent guideline [17]. The results supported that empiric treatments should be guided by a local distribution of pathogens that recognized and treatments are recommended by the Management of Adults with Hospital-acquired and Ventilator-associated Pneumonia in 2016 by IDSA/ATS guideline [2]. Papazian et al suggested that microbiological con rmation is strongly recommended when considering a diagnosis of VAP and pathogens may vary depending on many factors including the duration of MV, hospital LOS, ICU LOS, previous antibiotics exposure, the occurrence of epidemic phenomena in a given ICU and local distribution of organisms [5]. 2008-2009. The study indicated MRSA was responsible for 6-7 % of the total causative organisms [9]. The majority of S. aureus colonization in the respiratory tract is in the nares and throat. Chlorhexidine is a topical antiseptic, which is most active against gram-positive bacteria [31]. Our center has applied selective oral decontamination (SOD) with chlorhexidine since 2011. This might have reduced the incidence of VAP due to MRSA.
The purpose of differentiation of VAP into EOVAP and LOVAP was to guide empiric antibiotic treatment to cover MDR bacteria. Inappropriate and delayed empirical therapy is associated with higher mortality in VAP patients [32][33][34]. The study found that LOVAP had a signi cantly higher proportion of MDR pathogens than EOVAP (p= 0.007). The results endorsed the Management of Adults with Hospital-acquired and Ventilator-associated Pneumonia in 2016 by IDSA/ATS suggested that VAP developed after 5 days of hospitalization had a greater risk of MDR pathogen presence than VAP developed earlier [2]. Therefore empiric broad-spectrum antibiotics against MDR pathogens were recommended for LOVAP.
Furthermore, this current study demonstrated that LOVAP had signi cantly longer MV days, ICU LOS, and hospital LOS than EOVAP. The hospital mortality was signi cantly greater in LOVAP (35.1% VS 16.7%, p=0.02). These worse outcomes of LOVAP were also observed by Khan et al [24]. The implementation of VAP prevention might reduce the cost of hospitalization and unnecessary mortality, especially in LOVAP [35].
A meta-analysis from Melsen et al suggested that overall attributable mortality from VAP was 13% and the higher mortality were found in surgical patients, acute physiology and chronic health evaluation (APACHE) score of 20-29 and SAPS II score of 35-58 [10]. Bekaert et al revealed the SAPS II score of 28-40 was signi cantly greatest associated with ICU death per additional day since the onset VAP [36]. Similar to our study, on stepwise backward multivariate analysis, a SAPII score ≥ 40 was signi cantly associated with hospital mortality of VAP patients.
The strengths of this study were that the recorded data were complete because VAP was under regular surveillance of our institute by infection control ward nurses (ICWNs) and con rmed by the infection control unit.
This study had some limitations. First, the sample size is small, especially in EOVAP. This affected the statistical power. Second, this was a retrospective study, some data might be di cult to determine such as previous antibiotic exposure within 90 days, prior hospitalization preceding 90 days. These factors are associated with the infection of MDR pathogen [2,37]. Third, the results of this study were unable to be applied to VAP in immunocompromised patients. Fourth, this study was from a single tertiary center, which had some limitations for considering empiric antibiotics treatment to general hospitals. Pathogens and resistance patterns could vary between hospitals, regions and countries [2]. The local pathogens and pattern resistance of each hospital were the crucial factors for the selection of empiric antibiotics.

Conclusion
In conclusion, LOVAP was signi cantly higher MDR pathogen, MV days, ICU LOS, hospital LOS and hospital mortality than EOVAP. A broad-spectrum antibiotic to cover MDR pathogens should be considered in LOVAP. The factor associated with hospital mortality of VAP was a SAPII score ≥ 40.

Declarations
Ethics approval and consent to participate This study was approved by the Human Research Ethics Committee, Khon Kaen University (approval number HE611281). Because the study was descriptive without intervention, the need of consent from the participants had also been waived.

Consent for publication
According to no individual patient data is presented in our study, consent for publication is not applicable.

Availability of data and materials
The datasets used and/or analyzed in this study are available from the corresponding author on