Experimental study for evaluation of the efficacy of a biofilm-embedded bacteria-based vaccine against Staphylococcus chromogenes-associated mastitis in sheep.

Although coagulase-negative staphylococci are the primary aetiological agents of subclinical mastitis in ewes, there is little information regarding vaccination against that infection. The objective of this study was to evaluate the efficacy of a vaccine against staphylococcal mastitis in ewes under experimental conditions. The antigen in the vaccine is based on a bacterin of Staphylococcus aureus strain, expressing the exopolysaccharide poly-N-acetylglucosamine (PNAG), which is involved in biofilm formation by these bacteria. Ewes in groups A (n = 17) or B (n = 6) were given an initial vaccination 5 weeks before expected lambing, followed by a repeat administration 21 days later. Ewes in groups C (n = 8) or D (n = 6) were unvaccinated controls. Ewes in group A (n = 17) or C (n = 8) were challenged with a biofilm-forming S. chromogenes; animals in subgroups A1 or C1 were challenged on the 10th and those in A2 or C2 on the 50th day after lambing. Ewes in groups B or D were uninoculated controls. Clinical examinations of ewes, ultrasonographic examinations of udder, milk yield measurements, blood sampling for detection of anti-PNAG specific antibodies and milk sample collection for bacteriological and cytological examinations were performed up to 52nd day post-challenge. Finally, biopsies were performed for mammary tissue collection for histopathological examination. Among group A ewes, 29% developed systemic signs and 59% signs in the inoculated gland; the respective figures for group C were 50% and 100% (P =  0.040 for mammary signs). The median total clinical score was 2.0 for A and 5.5 for C ewes (P =  0.025). For A, but not for C, clinical scores decreased progressively during the study (P =  0.018 and P =  0.47, respectively). The duration of mastitis was shorter in A (4 days) than in C (17.5 days) ewes (P =  0.022). Bacterial counts were lower in milk samples from A than from C ewes, for samples collected from the inoculated and the uninoculated (P <  0.01) mammary glands of these ewes. Somatic cell counts in samples from inoculated and uninoculated mammary glands of A ewes were higher than in samples of C ewes (P <  0.02). There were differences for gray-scale evaluations during ultrasonographic examination and for milk yield measurements between groups (P <  0.01). Median bacterial counts in tissue samples from A ewes (0 cfu g-1) were lower than in ones from C (6.5 cfu g-1) ewes (P =  0.041). The median score for histopathological findings in tissue samples from inoculated glands of A was lower than that for C ewes: 1 versus 2 (P =  0.014). It is concluded that mastitis was less severe in vaccinated animals, as indicated by a wide array of measures.

Evaluation of efficacy of a biofilm-embedded bacteria-based vaccine against staphylococcal mastitis in sheep-A randomized, placebo-controlled field study.

Our objective was to evaluate the efficacy of a vaccine against staphylococcal mastitis in 5 dairy sheep farms, with 316 ewes in the vaccinated (V) group and 307 in the control (C) group studied throughout a lactation period. Two administrations of the vaccine were performed during the last stage of gestation of ewes. Starting 15 d after lambing and at monthly intervals thereafter, up to 9 milk samplings were performed for bacteriological and cytological examinations. Staphylococcal isolates recovered were examined for biofilm formation. Blood samples were collected for measurement of IgG poly-N-acetylglucosamine-specific antibodies. The most frequently isolated bacteria were staphylococci: 56.4 and 76.1%, respectively, of total isolates recovered from ewes of group V and C, respectively; staphylococci as causal agents of mastitis were isolated less frequently from V (5.3%) than in ewes in C (10.3%). Among mastitis-associated staphylococcal isolates recovered from V ewes, a smaller proportion was biofilm-forming than among ones from C: 53.2% versus 74.9% of isolates; biofilm-forming staphylococci as causal agents of mastitis were isolated less frequently from ewes in group V (2.3%) than in ewes in group C (6.0%). Anti-poly-N-acetylglucosamine-specific antibody values increased in V ewes and were higher than in C; a greater proportion of ewes with low antibody titers developed staphylococcal mastitis (41.4%) than of V ewes with high antibody titers (17.0%). Incidence risk of mastitis, staphylococcal mastitis, and biofilm-associated staphylococcal mastitis was smaller in V than in C: 36.7, 17.1, and 8.0% versus 44.3, 30.9, and 18.9%, respectively. The first case of staphylococcal mastitis occurred later in V than in C: third versus second sampling point. Overall, efficacy of the vaccine was 44.6% for staphylococcal mastitis, 57.7% for biofilm-associated staphylococcal mastitis, 33.1% for staphylococcal intramammary infection, and 51.5% for biofilm-associated staphylococcal intramammary infection. Nevertheless, vaccination should not be the only means for controlling mastitis; other udder health management measures should be included therein to improve control of the infection.

First detection of an optrA-positive, linezolid-resistant ST16 Enterococcus faecalis from human in Greece.

Until now, in Greece, the resistance of enterococci to linezolid was associated with mutations of domain V of 23S ribosomal RNA (G2576T). Here we report the first linezolid-resistant optrA-positive Enterococcus faecalis sequence type (ST) 16 isolated from a patient with a urinary tract infection (UTI). No travels overseas, contact with food-producing animals or previous treatment with linezolid were reported. Plasmid analysis suggested the chromosomal location of optrA gene. Additionally, whole genome sequencing data revealed the association of optrA with transposon Tn554 and the coexistence with fexA, spc and ermA-like resistance genes. A similar genetic structure has been previously identified in an ST767 E. faecalis from Taiwan.

Study of potential environmental factors predisposing ewes to subclinical mastitis in Greece.

Objectives of the work presented herewith were to investigate association of prevalence of subclinical mastitis with environmental (climatic and topographic) factors and to identify factors potentially predisposing ewes to the disease. Milk samples were collected from 2198 sheep in 111 farms, in all 13 administrative regions of Greece, for bacteriological and cytological examination. Data on farm location were collected in the field using hand-held Global Positioning System Garmin units. The geo-references were resolved to specific farm level. Prevalence of subclinical mastitis was 0.260. Main aetiological agents were staphylococci (Staphylococcus aureus and coagulase-negative species), which accounted for 0.699 of all isolates recovered. In a multivariable mixed-effects analysis, the two environmental variables found to be associated with increased prevalence of subclinical mastitis were the minimum temperature of coldest month (coefficient: -0.084 ± 0.033, P = 0.014) and the mean temperature for 30 days prior to sampling date (coefficient: 0.031±0.014, P = 0.029).

Combination of vial culture and broad-range PCR for the diagnosis of spontaneous bacterial peritonitis: experience in a Greek tertiary care hospital.

Spontaneous bacterial peritonitis (SBP) is often difficult to diagnose because bacteria in ascites cannot be detected accurately by conventional culture. In this study, we evaluated the use of broad-range 16S rRNA PCR, applied either directly to a total of 32 ascitic fluids (AFs) or to the AF vial cultures, after a long incubation of 14 days; the results were compared with those of AF vial cultures. Escherichia coli was isolated in four of 32 AF vial cultures (12.5%). The application of 16S rRNA PCR directly to AF detected only one of the four positive samples (sensitivity 25%, specificity 100%, positive predictive value (PPV) 100%, negative predictive value (NPV) 90.32%). However, the application of 16S rRNA PCR to AF vial cultures after 14 days of incubation correctly identified all the positive samples, including one more that was positive for Brucella mellitensis (sensitivity 100%, specificity 80%, PPV 80%, NPV 100%). The elongation of the incubation period of the AF vial cultures, combined with the use of 16S rRNA in negative vials, increases the possibility of identifying the causative agents of SBP and could be applied in the clinical laboratory.

Antimicrobial susceptibility and mechanisms of resistance of Greek Clostridium difficile clinical isolates.

OBJECTIVES: This study examined the antimicrobial susceptibility and resistance mechanisms of Clostridium difficile recovered in Greek hospitals during 2012-2015. METHODS: C. difficile isolates (n=88) were collected from clinically-confirmed C. difficile infection from symptomatic patients in 10 Greek hospitals. Minimum inhibitory concentrations (MICs) of various antimicrobial agents were determined by Etest. Isolates were typed by multilocus sequence typing (MLST). Toxin and resistance genes were detected by PCR. Chromosomal mutations in gyrA, gyrB and rpoB were identified by PCR and sequencing. The genetic environment of resistance genes was characterised by Illumina sequencing. RESULTS: The 88 C. difficile isolates comprised 27 sequence types (STs), with ST37 (n=26) and ST11 (n=21) being the most prevalent. All isolates were susceptible to vancomycin and metronidazole, with variable resistance rates to other antimicrobials. Of the 88 isolates, 45.5% were multidrug-resistant and the majority belonged to ST11 and ST37. The presence of chromosomal mutations in gyrA, gyrB and rpoB was mainly observed in high-risk clones such as ST11 and ST37. The antimicrobial resistance genes ermB, mefA, msrA and tetM were identified at different prevalences and combinations. Additionally, cfrB and cfrC were identified for the first time in Greece and were carried by a Tn6218 transposon and a novel plasmid, respectively. CONCLUSIONS: To our knowledge, this is the first study examining the resistance profiles and respective mechanisms of C. difficile recovered in Greek hospitals. Gut commensals such as C. difficile may serve as hubs for further transfer of antimicrobial resistance genes.

Use of geographical information system and ecological niche modelling for predicting potential space distribution of subclinical mastitis in ewes.

Objective was to recognise areas potentially of high risk for increased frequency of subclinical mastitis in ewes. Milk samples were collected, for bacteriological and cytological examination, from 2198 clinically healthy ewes in 111 farms in all administrative regions of Greece. Data on farms were located in the field using hand-held Global Positioning System Garmin units. Collected data were analysed by an Ecological Niche Model under the framework of a geographic information system. Two separate analyses were performed: one for subclinical mastitis independently of causal agent (prevalence in population sampled: 0.260) and one for subclinical mastitis caused specifically by slime-producing staphylococci (prevalence in population sampled: 0.153). A model was constructed in which sheep farms were divided into two clusters, according to prevalence of subclinical mastitis: farms in the upper three quartiles of prevalence were used as occurrence points for the Ecological niche modelling procedure ('infected farms'); farms in the lower quartile of prevalence within each category were (pseudo)negative points. Significant differences in environmental parametres prevailing in locations of farms into the study, were identified for up to 13 parametres between locations of farms according to management system applied in farms. When farms in each management system were considered separately, differences became evident between farms in each management system, as well as between the two infections. The factor with the highest relative contribution in the analyses was the distance from other sheep farms; other factors also of importance in the predictive models were the altitude, the maximum temperature of warmest month and the total precipitation of driest month. Verification of the model revealed that ≥ 0.760 of infected farms' were located in areas predicted as high risk for prevalence of subclinical mastitis or slime staphylococcal subclinical mastitis. The paper describes for the first time potential association of mastitis with environmental factors and presents predictive models for mastitis in ewes taking into account environmental parametres.

Slime-producing staphylococci as causal agents of subclinical mastitis in sheep.

Hitherto, research work in slime production from staphylococcal strains of mastitis origin has focused in laboratory properties of these organisms. Objective of present work was to study subclinical mastitis in sheep, caused specifically by slime-producing staphylococci: to investigate its frequency and to identify potential factors playing a role therein. Slime production was evaluated in 708 staphylococcal isolates recovered from cases of subclinical mastitis in a field study in 2198 ewes performed in an extensive countrywide field investigation across Greece. Isolates were studied by means of microbiological and molecular methods. Of these strains, 262 were characterised as slime-producing, 227 as weak slime-producing and 219 as non slime-producing. Most frequently detected genes were eno and icaB; Staphylococcus aureus possessed more genes than coagulase-negative strains; greater number of genes was detected in slime-producing than in weak slime-producing or non-slime-producing strains. Subclinical mastitis caused specifically by slime-producing staphylococci was detected in 337 ewes: prevalence in population sampled was 0.153. A multivariable mixed-effects model revealed that milking mode (highest prevalence in hand-milked flocks) and flock management system (highest prevalence in semi-intensive flocks) were the two factors associated with increased prevalence of mastitis in flocks. The results confirmed the significance of slime producing staphylococcal strains of importance in the aetiology of subclinical mastitis of sheep. Hand-milking was identified as the most important factor predisposing to that infection.

Extensive countrywide field investigation of subclinical mastitis in sheep in Greece.

The objectives of this work were (1) to investigate prevalence of subclinical mastitis, (2) to identify etiological agents involved, and (3) to study factors potentially predisposing ewes to subclinical mastitis. Milk samples were collected from 2,198 ewes in 111 farms with a total population of 35,925 ewes, in all 13 administrative regions of Greece, for bacteriological and cytological examination. Prevalence of subclinical mastitis was 0.260. Main etiological agents were staphylococci (Staphylococcus aureus and coagulase-negative species), which accounted for 0.699 of all isolates recovered; prevalence of staphylococcal mastitis was 0.191. In a multivariable mixed-effects analysis, the primary factor found to be associated with increased prevalence of subclinical mastitis was the management system practiced in flocks (flocks under a semi-intensive system had the highest prevalence). Other factors that were included in the multivariable model were the stage of lactation period (ewes in the 2nd month postpartum showed the highest prevalence) and application of postmilking teat dipping. In contrast, measures taken at the end of a lactation period (e.g., intramammary administration of antimicrobial agents) were not found to have an effect on prevalence of subclinical mastitis. The results confirmed the significance of subclinical mastitis as a frequent problem of ewes, with staphylococci as the primary etiological agent. The findings confirm the multifactorial nature of subclinical mastitis and indicate that its control should rely on many approaches.

First description in Europe of the emergence of Enterococcus faecium ST117 carrying both vanA and vanB genes, isolated in Greece.

OBJECTIVES: An Enterococcus faecium isolate (Efa-125) carrying both the vanA and vanB genes was recovered from a patient with bacteraemia treated in a Greek hospital. Since this is the first description in Europe of E. faecium carrying both vanA and vanB genes, the isolate was further studied. METHODS: Susceptibility to several antibiotics was determined using the VITEK®2 automated system. The isolate was typed by multilocus sequence typing (MLST). To define the genetic units of the vanA and vanB genes, the plasmid content of Efa-125 was analysed by pulsed-field gel electrophoresis (PFGE) of total DNA digested with S1 nuclease followed by hybridisation with digoxigenin-labelled vanA and vanB probes. In addition, plasmids and chromosomes were sequenced using the Illumina MiSeq platform. RESULTS: E. faecium Efa-125 belonged to ST117 and expressed resistance both to vancomycin and teicoplanin, with minimum inhibitory concentrations (MICs) for both of 256mg/L. The vanA gene was carried on a 29 320-bp plasmid exhibiting high similarity to pA6981 previously characterised from Enterococcus gallinarum A6981, whereas vanB was part of a Tn1549-like transposon integrated into the chromosome. Expression of the VanA phenotype was correlated with the presence of intact vanZ and vanS genes. CONCLUSIONS: This is the first detection in Greece of vanA-vanB genotype/VanA phenotype E. faecium and indicates an evolving epidemiology of vancomycin-resistant enterococci.

Viruses and endogenous retroviruses in multiple sclerosis: From correlation to causation.

Multiple sclerosis is an immune-mediated disease with an environmental component. According to a long-standing but unproven hypothesis dating to initial descriptions of multiple sclerosis (MS) at the end of the 19th century, viruses are either directly or indirectly implicated in MS pathogenesis. Whether viruses in MS are principally causal or simply contributory remains to be proven, but many viruses or viral elements-predominantly Epstein-Barr virus, human endogenous retroviruses (HERVs) and human herpesvirus 6 (HHV-6) but also less common viruses such as Saffold and measles viruses-are associated with MS. Here, we present an up-to-date and comprehensive review of the main candidate viruses implicated in MS pathogenesis and summarize how these viruses might cause or lead to the hallmark demyelinating and inflammatory lesions of MS. We review data from epidemiological, animal and in vitro studies and in doing so offer a transdisciplinary approach to the topic. We argue that it is crucially important not to interpret "absence of evidence" as "evidence of absence" and that future studies need to focus on distinguishing correlative from causative associations. Progress in the MS-virus field is expected to arise from an increasing body of knowledge on the interplay between viruses and HERVs in MS. Such interactions suggest common HERV-mediated pathways downstream of viral infection that cause both neuroinflammation and neurodegeneration. We also comment on the limitations of existing studies and provide future research directions for the field.

Emergence of sequence type 11 Klebsiella pneumoniae coproducing NDM-1 and VIM-1 metallo-ß-lactamases in a Greek hospital.

Sequence type 11 Klebsiella pneumoniae, coproducing NDM-1 and VIM-1 metallo-ß-lactamases, were isolated in a Greek hospital. blaNDM-1 was part of a Tn125 derivative, located on an ~90-kb plasmid similar to the NDM-1-encoding plasmid pB-3002cz. blaVIM-1 was located in an In-e541-like integron, carried on a multireplicon (IncA/C and IncR) plasmid of ~180kb.

Pitfalls in the identification of Enterococcus species and the detection of vanA and vanB genes.

UNLABELLED: The aims were to assess the performance of Vitek 2 in identifying enterococcal species and the implementation of GeneXpert(®) vanA/vanB PCR for the detection of vancomycin-resistant enterococci (VRE). Gram-positive cocci from clinical and environmental specimens (n = 431) suspicious of being enterococci by conventional methods were evaluated by Vitek 2. This system identified 296 Enterococcus faecium, 87 Enterococcus faecalis, 10 Enterococcus villorum, 9 Enterococcus gallinarum, 9 Enterococcus durans, 5 Enterococcus casseliflavus, 1 Enterococcus spp. and 14 isolates as Non-Enterococcus. All strains were submitted to pulsed field gel electrophoresis (PFGE) analysis showing 64 banding patterns. Representative strains from each banding pattern were further characterized to species level by 16S rDNA sequencing. The misidentification rate by Vitek 2 to species level among 429 molecularly identified enterococci was 6% (26 isolates). Additionally, 372 rectal swabs were obtained from critically ill patients. They were evaluated for the presence of VRE by ChromID VRE combined with in-house PCR vs GeneXpert(®) . GeneXpert(®) showed high (>92%) sensitivity, specificity, accuracy for vanA-positive Enterococcus detection, as well as, sensitivity and specificity for vanB-positive strains. Positive predictive value for detection of vanB-positive enterococci by GeneXpert(®) vanA/vanB was low (30%). GeneXpert(®) showed the same efficacy as ChromID VRE in detecting vanA-positive enterococci, but lower for vanB-gene detection. SIGNIFICANCE AND IMPACT OF THE STUDY: The study shows that even though the performance of Vitek 2 Advanced Expert System was good in identifying enterococci to species level, it is important to verify results by a molecular method when phenotypic findings are discordant with epidemiologic patterns. Furthermore, GeneXpert(®) vanA/vanB PCR and ChromID VRE combined with in-house PCR were applied in rectal samples for the detection of VRE colonization among critically ill patients. GeneXpert(®) showed an excellent performance in detecting vanA-positive enterococci, but false-positive results for vanB-gene detection render its application problematic in departments with high incidence of vanB-positive enterococci.

Rapid dissemination of colistin and carbapenem resistant Acinetobacter baumannii in Central Greece: mechanisms of resistance, molecular identification and epidemiological data.

BACKGROUND: Colistin-resistant/carbapenem-resistant Acinetobacter baumannii is a significant challenge for antibiotic treatment and infection control policies. Since 2012, in Central Greece an increase of colistin/pan- resistant A. baumannii has occurred, indicating the need for further analysis. METHODS: A total of 86 colistin-resistant/carbapenem-resistant out of 1228 A. baumannii clinical isolates, consecutively collected between 2012 and 2014 in a tertiary Greek hospital of Central Greece, as well as one environmental isolate from surveillance cultures were studied. Molecular typing and mechanisms of resistance to colistin and to carbapenems were assessed, whereas, epidemiological and clinical data of the patients were reviewed. RESULTS: During the study period, the rate of colistin resistance gradually increased and reached 21.1 % in 2014. All colistin-resistant/carbapenem-resistant A. baumannii belonged to 3LST ST101 clone that corresponds to the international clonal lineage II. Carbapenem resistance was associated with the presence of bla oxa-23-like, while resistance to colistin probably correlated with G54E and R109H amino acid substitutions in PmrA and PmrC, respectively. CONCLUSIONS: Epidemiological data of the patients indicated that the first detection of colistin-resistant/carbapenem-resistant ST101 clone in the University Hospital of Larissa (UHL) was associated with a patient who previously had received colistin, while, the movement of the infected patients into the hospital probably resulted to its spread.

Human Staphylococcus aureus lineages among Zoological Park residents in Greece.

Staphylococcus aureus is a part of the microbiota flora in many animal species. The clonal spread of S. aureus among animals and personnel in a Zoological Park was investigated. Samples were collected from colonized and infected sites among 32 mammals, 11 birds and eight humans. The genes mecA, mecC, lukF/lukS-PV (encoding Panton-Valentine leukocidin, PVL) and tst (toxic shock syndrome toxin-1) were investigated by PCR. Clones were defined by Multilocus Sequence Typing (MLST), spa type and Pulsed-Field Gel Electrophoresis (PFGE). Seven S. aureus isolates were recovered from four animals and one from an employee. All were mecA, mecC and tst-negative, whereas, one carried the PVL genes and was isolated from an infected Squirrel monkey. Clonal analysis revealed the occurrence of seven STs, eight PFGE and five spa types including ones of human origin. Even though a variety of genotypes were identified among S. aureus strains colonizing zoo park residents, our results indicate that colonization with human lineages has indeed occurred.

Mycoplasma pneumoniae and Chlamydophila pneumoniae: a comparative study in patients with nasal polyposis and healthy controls.

INTRODUCTION: The role played by Mycoplasma pneumoniae and Chlamydophila pneumoniae in the pathogenesis of chronic rhinosinusitis with nasal polyps has been the object of ongoing debate. We used real-time polymerase chain reaction to investigate the prevalence of both microorganisms in the nasal tissue samples of patients and controls. METHODS: We extracted DNA from nasal polyp samples obtained during functional endoscopic sinus surgery and the inferior turbinate samples of controls undergoing septoplasty. We used the highly sensitive real-time polymerase chain reaction to detect the presence of M pneumoniae and C pneumoniae DNA. RESULTS: Patients with chronic rhinosinusitis with nasal polyps consisted of 62 individuals (39 men; mean age 51 years); the control group consisted of 24 individuals (13 men; mean age 45 years). All samples from both groups were negative for M pneumoniae and C pneumoniae DNA. CONCLUSION: We have demonstrated that the likelihood of M pneumoniae and C pneumoniae acting as an ongoing inflammatory stimulus in chronic rhinosinusitis with nasal polyps is slim.

Disseminated Bacillus Calmette-Guerin infections after intravesical therapy.

Intravesical instillation of Bacillus Calmette-Guerin (BCG) is the treatment of choice for superficial bladder carcinoma. Disseminated BCG infection presenting as granulomatous hepatitis or pneumonitis is a very rare complication of this treatment. Here we report a case series of seven patients previously treated with BCG presenting with pneumonitis. In two of the cases, identification of Mycobacterium bovis was achieved with molecular methods.

Risk factors for enterococcal infection and colonization by vancomycin-resistant enterococci in critically ill patients.

PURPOSE: Vancomycin-Resistant Enterococci (VRE) are important causes of Intensive Care Unit (ICU) infections. Our goal was to identify the prevalence and risk factors for VRE colonization upon ICU admission and during ICU stay, as well as, their impact in enterococcal infection including vancomycin-susceptible cases (VSE). METHODS: A prospective study regarding patients admitted in ICU (n = 497) was conducted during a 24-month period. Rectal swabs were collected upon admission and during hospitalization and inoculated onto selective medium. Enterococci were phenotypically characterized. van genes were investigated by PCR and clones were identified by Pulsed-Field Gel Electrophoresis and Multilocus Sequence Typing. Epidemiologic data were collected from the ICU database. RESULTS: Risk factors for VRE carriage upon ICU admission (71/497) were: duration of previous hospitalization, glycopeptide administration, chronic heart failure, malignancy, insulin-dependent diabetes mellitus, and previous enterococcal infection (VRE and/or VSE). Risk factors for VRE colonization during ICU stay (36/250) were: quinolone administration, chronic obstructive pulmonary disease, chronic renal failure, and number of VRE-positive patients in nearby beds. Risk factors for enterococcal infection during ICU stay (15/284), including VRE and VSE cases, were: administration of third- or fourth-generation cephalosporins, cortisone use before ICU admission and VRE colonization, whereas, enteral nutrition was a protective factor. CONCLUSIONS: Previous VRE colonization and antibiotic usage are essential parameters for enterococcal infection (by VRE or VSE) during ICU stay. Previous enterococcal infection, co-morbidities and antibiotic usage are associated with VRE colonization upon ICU admission, whereas, patient to patient transmission, co-morbidities and antibiotic usage constitute risk factors for VRE colonization during ICU hospitalization.