Endogenous source of bacteria in tracheal tube and proximal ventilator breathing system in intensive care patients

British Journal of Anaesthesia 1998; 80: 41–45

Endogenous source of bacteria in tracheal tube and proximal ventilator breathing system in intensive care patients T. J. J. INGLIS, E.-W. LIM, G. S. H. LEE, K.-F. CHEONG AND K.-S. NG

Summary Although bacteria from both the ventilator breathing system and the gastrointestinal tract have been implicated in the pathogenesis of ventilator-associated pneumonia, an endogenous source of bacteria in the proximal respiratory breathing system has yet to be demonstrated conclusively. We investigated a potential route of bacterial colonization from the stomach contents to the efferent limb of the ventilator breathing system by bacterial culture of daily specimens from six sites in 20 surgical intensive care patients. Gram-negative bacilli were isolated in a progressively increasing proportion of samples at successive sampling points, consistent with an endogenous-to-external route of spread (patients, chi-square:14.12, P
0.02; samples, chi-square:106.15, P
0.001). Identical strains of gram-negative bacilli, confirmed by REPS typing, were found at two or more sites in seven patients. In all seven, gram-negative bacilli were first isolated from a site in the patient. In none of the 20 patients was there evidence of a sequence of colonization from the ventilator tubing or Y-piece connector towards the patient. Probable colonization sequences plotted from the time of first isolation supported the proposed sequence in six patients, and in five began with the stomach contents. Isolation sequences contrary to the proposed direction of colonization involved four bacterial species and two patients, and did not extend beyond two sample sites. These findings imply that the retrograde route of bacterial colonization of the ventilated lung extends into the proximal respiratory breathing system and may help to identify additional targets for preventive intervention. (Br. J. Anaesth. 1998; 80: 41–45)

series of bacteriological events would thus be expected to have major implications for the preventive management of the critically ill. We proposed previously that bacteria from the biofilm on the luminal surface of the tracheal tube might be disseminated into the lungs during the inspiratory phase of mechanical ventilation.14 Further studies on the static and dynamic properties of tracheal tube biofilm in tubes from intensive care patients support the occurrence of fluid dynamic events in the tracheal tube and other parts of the proximal respiratory breathing system,15 16 but bacteria in the luminal biofilm and their subsequent dissemination within the proximal breathing system have yet to be linked to a gastrointestinal source. In this study, we have examined a possible retrograde route of bacterial transfer from the stomach via the tracheal tube to the efferent tubing in intensive care patients undergoing mechanical ventilation.

Patients and methods Entry to the study was by consecutive admission to the adult surgical intensive care unit (ICU) of a tertiary referral hospital. Selection criteria were a requirement for mechanical ventilation at the time of ICU admission and access to stomach contents by nasogastric tube. Admission and clinical progress data were collected without reference to laboratory results. Tracheal suction was applied by nursing staff using single use suction catheters, and a gloved-hand technique. In-line, enclosed suction devices were not used during this study. Ventilator filters were not placed between the Y-piece connector and the ventilator tubing during the study. SPECIMENS

Keywords: infection, nosocomial; intensive care; ventilation, mechanical; infection, bacterial; infection, breathing systems; equipment, breathing systems

A complete specimen set was obtained early every morning, 7 days a week, until the patient died or no longer required mechanical ventilation. All specimens were collected by the same investigator (T. J. J. I.) after preparing specimen containers, undergoing a

Despite a substantial body of evidence implicating bacteria in the gastrointestinal tract as the cause of ventilator-associated pneumonia,1–9 the role of bacteria in the stomach contents remains controversial.10 The route followed by these bacteria during their initial entry into and colonization of the lower respiratory tract is an important consideration in preventive strategies designed to prevent ventilatorassociated pneumonia.11–13 Any new insight into this

TIMOTHY J. J. INGLIS, DM, MRCPATH, DTM&H, FRCPA, Division of Microbiology and Infectious Diseases, Western Australian Centre for Pathology and Medical Research, Locked Bag 2009, Nedlands, WA 6009, Australia. EK-WANG LIM, MIBIOL, GINA S. H. LEE, BSC(HONS), (Department of Microbiology); KENG-FATT CHEONG, MB, BS, (Department of Anaesthesia); National University Hospital, Singapore. KIM-SWEE NG, MB, BS, MMED(ANAES), Tao Payoh Hospital, Singapore. Accepted for publication: September 2, 1997. Correspondence to T. J. J. I.

42 thorough hand wash with chlorhexidine solution, and drying and donning of sterile gloves. The ventilator was then inactivated under the guidance of a member of the intensive care staff for the short time required to obtain bacteriological swabs. The sequence of specimen collection was in the opposite direction to the putative colonization sequence (i.e. from expiratory trap to stomach contents) in order to avoid bias caused by accidental carryover. The ventilator trap in the expiratory limb of the ventilator tubing was disconnected briefly and a sterile dry cotton swab run around the inner circumference of the afferent limb of the trap. The Y-piece connector was then sampled, after disconnecting from the angle piece, by running a sterile dry cotton swab around its inner circumference. Next, the angle piece was removed briefly from the top of the tracheal tube and a dry, sterile cotton swab inserted without touching any surface other than the inside of the access port plug. A sterile swab was then run around the luminal circumference of the upper end of the tracheal tube. The connections and positioning of tubing were checked, and the ventilator cycle restarted. A specimen of gastric contents was obtained with a sterile 50-ml syringe via the nasogastric tube, after first discarding a smaller 10-ml syringe containing 5 ml or more of nasogastric tube contents. The previously used (early morning) tracheal suction catheter was kept for analysis by the attending intensive care nurse, who wrapped the catheter in her glove by turning the glove inside out during final removal. This specimen was placed in a self-seal bag, labelled and stored in the ICU specimen refrigerator if the other specimens had yet to be collected.

BACTERIOLOGICAL METHODS

Specimens were obtained for processing the same morning, other than those collected during weekends which were stored at 0–4 ⬚C until the following Monday. Specimen recording, processing and further analytical work were performed without reference to clinical data. All results were withheld until after completion of the specimen collection phase of the study. Each specimen was inoculated onto a single pre-prepared 5% horse blood agar 8-cm plate and an 8-cm MacConkey agar plate (Becton Dickinson, Heidelberg, Australia). Cotton swabs were applied and the inoculae spread following the conventional pattern. The lowermost 5 cm was cut from the tip of the tracheal suction catheter onto the blood agar plate using heat sterilized scissors, rolled to make a square impression. This was repeated on a MacConkey plate. The gastric contents specimen was mixed by repeated inversion and a 0.1-ml aliquot transferred by sterile pipette to the surface of the agar plate and spread with a sterile wire loop in the conventional manner. All plates were incubated for 18 h at 37 ⬚C in air before examination. Any growth of gram-negative bacteria was noted and the colony isolated was identified to species level using standard methods.17 Antibiotic susceptibility testing was performed on all isolates using the Kirby–Bauer method and NCCLS interpretation criteria.18 Each isolate

British Journal of Anaesthesia was stored by suspension in proprietary broth containing sterile beads (Microbank, USA) and frozen at 970 ⬚C. Molecular typing of isolates was performed on isolates resuscitated from the frozen culture collection by REP (repetitive extragenic palindromic) sequence analysis, as described previously,9 19 but having adapted the method for use in capillary tubes with the rapid air thermal cycler.20 Briefly, bacterial isolates were cultivated on nutrient agar at 37 ⬚C for 48 h. A single colony of each isolate was picked off with a sterile toothpick and emulsified in 50 ␮l of sterile distilled water. The suspension was boiled for 5 min and then cooled on ice. A PCR mastermix was prepared in a laminar flow cabinet from 18 ␮l 10 PCR reagent buffer (Promega, Madison, WI, USA), 18 ␮l of dNTPs 2 mmol litre91 (Promega), 18 ␮l of 10% Triton X100 (Sigma Chemical Co., St Louis, MO, USA), 18 ␮l of sterile bovine serum albumin (Idaho Technology, Idaho Falls, ID, USA), 18 ␮l of dimethyl sulphoxide (E Merck, Darmstadt, Germany), 18 ␮l of MgCl2 10 mmol litre91 (Idaho Technology), 18 ␮l of sterile distilled water, 25.2 ␮l of REP1-RI primer (5’IIIICGICGICATCIGGC-3’, 500 mmol litre91 stock solution) (Biosynthesis, Lewisville, KY, USA), 25.2 ␮l of REP2-I primer (5’ ICGICTTATCIGGCCTAC-3’, 500 mmol litre91 stock solution) (Biosynthesis) and 18 ml of Taq DNA polymerase (8 ␮l of 5 u. ml91 stock and 12 ␮l of enzyme diluent, Promega). Mastermix (10 ␮l) was dispensed into each pre-labelled sterile 1.5-ml Eppendorf tube, to which was added 1.2 ␮l of template DNA. After mixing, the reagents were drawn into a capillary reaction tube, the ends of which were heat sealed. The air thermal cycler (Idaho Technology) was programmed to run as follows: initial denaturation for 3 min at 95 ⬚C, then 35 cycles of denaturation 0 min at 90 ⬚C, annealing 1 min at 45 ⬚C and elongation 8 min at 65 ⬚C, with a final elongation at 65 ⬚C for 16 min. Capillaries were then scored with a sapphire cutter and the contents dispensed into pre-labelled 1.5-ml Eppendorf tubes: 2-␮l 5running buffer and dye were added to each tube and pipette mixed, and 12 ␮l of molecular marker (DNA 100 base pair ladder, Promega) and buffer were dispensed into the first of 15 wells in a 1% agarose (Sigma Chemical Co., St Louis, MO, USA) submarine gel, under Tris EDTA buffer. The amplified PCR product and buffer were dispensed into the remaining 14 wells and run at 100 V for 1.5 h. Staining was performed with ethidium bromide and visualization was with an ultraviolet transilluminator. The gel was photographed using Polaroid 667 film.

ANALYSIS

Isolates from different sites in the same patient, or the same site on different days, were regarded as possibly identical if the biochemical profile identified the same species, and antibiotic susceptibility differed by no more than one result. Isolates were referred to as identical only when amplified REP sequences produced an identical pattern on agarose gel electrophoresis. Identical first isolates from a given sample site were then plotted against the day the respective sample was collected.

Bacteria in ventilator breathing system

43

Table 1 Patient characteristics (mean (range) or number)

Discussion

n Sex (M/F) Admission category Neurosurgical—trauma Neurosurgical—other Abdominal Thoracic ENT Orthopaedic—trauma Other Age (yr) APACHE score Days in study

In this study of bacterial colonization in patients undergoing mechanical ventilation, the source of gram-negative species contaminating the proximal respiratory breathing system was the patient in every case where identical gram-negative bacilli were isolated from two or more specimen sites. There was no evidence to support progression of bacteria from the ventilator, or even, carryover during the sampling procedure. On the contrary, the significant increase in gram-negative bacilli at successive sampling points implied that a colonization sequence that started in the ventilator tubing was unlikely, and was consistent with a progressive colonization of luminal surfaces in the opposite direction. In the majority of bacterial isolate series, the time of first isolation at a given site was consistent with the proposed luminal route of bacterial transfer, that is from the stomach contents via the oesophagus, into the trachea, down the outside of the tracheal tube and, on the suction catheter up into the tracheal tube, from where spread to the angle piece connector, the bifurcation of the Y-piece and finally the expiratory tubing could occur. We did not attempt to determine if the oropharynx acted as a portal of entry, a secondary source of bacteria or a staging post in retrograde colonization, as resolution of the role of the oropharynx would have required sampling at multiple intra-oral sites. Instead, the study aimed to give equal prominence to the six locations sampled and to use rigorous bacteriological criteria to exclude similar but non-identical isolates. Inclusion of a genetic typing method in the assessment criteria necessitated exclusion of several possible isolate series, permitting a greater degree of confidence in the postulated colonization sequence.21 In five (25%) of the patients studied the gramnegative species concerned were observed first in the stomach contents. However, in two other patients the bacteria appeared first on the suction catheter tip, several days before the stomach contents. We have shown previously that gastric bacterial overgrowth may be intermittent in intensive care patients22 but it remains possible that the gastric contents were contaminated by bacteria from the oropharynx, prior

20 13/7 5 3 6 2 1 1 2 54.9 (21–91) 15.6 (3–32) 5.25 (2–18)

STATISTICAL METHODS

The chi-square test was calculated using epidemiological software (C-stat, version 1.0, Cherwell Scientific, Oxford, UK) running under DOS 5.0. A threshold P value of 0.05 was used in assessments of statistical significance.

Results Twenty consecutive patients admitted to the surgical intensive care unit were investigated (patient and clinical data are shown in tables 1 and 2). The proportion of patients and specimens from which gram-negative bacilli were isolated increased at successive sampling points in the specimen sequence followed (table 3). In seven patients and 12 series of isolates, gram-negative bacilli from more than one specimen site were identical by REP sequence analysis (table 4). In all seven patients and 12 bacterial series, the first isolate was obtained from a site within the patient, and in all but five instances the temporal sequence of isolation of a given REP sequence pattern was consistent with the proposed luminal route of bacterial colonization. However, in four series of identical bacterial isolates (three from one patient), isolates from the suction catheter tip or tracheal tube lumen antedated isolation of the same strains from the stomach contents. In one other patient, first isolation was at the angle piece and was followed by isolation from the suction catheter tip. No confounding colonization sequence involving more than two adjacent sites was observed.

Table 2 Patients studied; clinical data Patient No.

Reason for ICU admission Concurrent disease

Antibiotics at ICU admission

Duration of intubation

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Shock Subtotal hepatectomy Septic shock Shock Perforated gastric ulcer Head injury, multiple fractures Head injury Total gastrectomy Craniotomy Craniotomy Craniotomy Laparotomy Cardiac and renal failure Craniotomy Obstructed airway Septic shock Septic shock

Vancomycin Ceftriaxone, metronidazole Gentamicin, co-amoxyclav Ciprofloxacin, vancomycin Amikacin, metronidazole Cloxacillin, metronidazole Cefazolin Ceftriaxone Nil Cefazolin Cefazolin Ceftriaxone Nil Ceftriaxone Ciprofloxacin Imipenem, fluconazole

8 days 36 h 19 days 15 days 2 days 4 days 7 days 4 days 36 h 36 h 14 days 24 h 2 days 2 days 3 days 2 days

18

Head injury

2 days 7 days

19 20

Oesophageal surgery Head injury

Ceftriaxone, metronidazole Gentamicin, metronidazole, ampicillin Cefuroxime Cephazolin

Myocardial infarction, intracranial haemorrhage Hepatocellular carcinoma Intestinal perforation Gastric carcinoma, leaking anastomosis Cardiorespiratory failure

Perforated DU Extradural haematoma Brainstem haemangioma Haematoma evacuation Peritonitis, ruptured appendix Ischaemic heart disease Subarachnoid haemorrhage Ludwig’s angina Diabetes, renal failure Mesenteric embolus, bronchial asthma, ischaemic heart disease Aspiration Hepatic cirrhosis, portal hypertension

2 days 2 days

44

British Journal of Anaesthesia

Table 3 Patients and specimens from which gram negative bacilli (GNB) were isolated. ET:Expiratory trap, YP:Y-piece, AP:angle piece, TTL:tracheal tube lumen, SCT:suction catheter up, GA:gastric aspirate. *Chi-square:14.12, df:5, P<0.02; †chi-square:106.15, df:5, P<0.001 Patients*

First isolated (day)

Specimens†

Site No GNB GNB (%) No GNB

GNB (%) Range Median

ET* 15 YP 15 AP 14 TTL 9 SCT 7 7 GA

8 (9) 10 (11) 14 (15) 30 (32) 53 (60) 44 (52)

5 (25) 5 (25) 6 (30) 11 (55) 12 (63) 12 (63)

85 83 78 63 35 40

1–9 4–9 1–9 1–8 1–4 1–6

5.0 7.0 3.5 2.0 1.0 1.0

Table 4 Series of identical bacterial isolates from intensive care patients and their corresponding respiratory breathing system. GA:Gastric aspirate; SCT:suction catheter tip; TTL:tracheal tube lumen; AP:angle piece connector; YP:Y-piece; ET:expiratory trap. Numbers represent day of specimen collection

Patient A Escherichia coli Citrobacter diversus Acinetobacter baumannii Patient B Citrobacter diversus Patient C Enterobacter aerogenes Klebsiella pneumoniae Acinetobacter baumannii Enterobacter cloacae Patient D Acinetobacter baumannii Patient E Acinetobacter baumannii Patient F Enterobacter cloacae Patient G Klebsiella pneumoniae

GA

SCT

TTL AP

6 7 8

1 1 2

4

3

1

2

8 2 6 6

1

2

1

1

1

1

1

1

4

YP

7

ET

10

9 2

9 9

2

2

4 4

gram-negative colonization at this site being common in hospitalized patients.23 In two patients more than one gram-negative species followed the proposed colonization pathway. As the same phenomena appear to occur with different bacterial genera, much of the process of bacterial transfer from site to site could be explained by movement of fluids within the lumina of anatomical structures or respiratory devices. Our observations are consistent with previous findings that the patient’s end of the ventilator breathing system is more likely to be contaminated by gram-negative bacilli.24 Moreover, these findings would suggest that the bacteria observed on the luminal surface of the tracheal tube originated in the patient’s endogenous flora; either in the stomach contents, or possibly the oropharynx. If this is indeed the case, the suction catheter must be considered a means of transferring these bacteria and the other components of tracheal tube biofilm from the tracheal mucosa to the inside of the tracheal tube.15 The subsequent dissemination of bacteria-laden particles from the luminal biofilm has to be inferred,14 16 as direct observation is not possible in vivo. The first point in the expiratory gas pathway that could act as an inertial impaction filter to trap aerosolized bacteria is where expired gases negotiate a 90⬚ change in direction at the angle piece connector. The inner surface of the angle piece connector, where angle piece swabs were obtained, was therefore chosen

as the most likely location for bacterial deposition after aerosolization of tracheal tube biofilm during the expiratory phase of ventilation. The presence of gramnegative bacteria on the inside of the angle piece in two isolate series after only 2 days, without identical isolates on the luminal surface of the adjacent tracheal tube, would therefore be consistent with aerosolization of bacteria during the expiratory phase of ventilation. Bacterial contamination of additional locations in the proximal respiratory breathing system prone to high velocity turbulent flow during the inspiratory phase of mechanical ventilation would be expected to increase the risk of dissemination of a bacteria-laden aerosol into the patient’s lungs. The angle piece may have been a point of entry for bacteria subsequently isolated from the trachea in one case. If so, this may represent an isolated occurrence of exogenous colonization. We have identified a possible route of bacterial colonization from the stomach via the tracheal tube and onwards into the ventilator breathing system. The results presented here prompt several important questions, particularly regarding the interaction between the suction catheter, tracheal secretions and the luminal surface of the tracheal tube. The extension of the so-called “retrograde” route of bacterial colonization of the ventilated lung proposed here should lead to additional target sites for preventive interventions, if our provisional conclusions are confirmed in further studies using other cohorts of intensive care patients.

Acknowledgements We thank Steven Chan and Timothy Lee, and also Goh Kah-Wei for assistance with the specimens. This study was supported by a grant (RP358038) from the National University of Singapore academic research fund.

References 1. Atherton ST, White DJ. Stomach as a source of bacteria colonising respiratory tract during artificial ventilation. Lancet 1978; ii: 968–969. 2. du Moulin GC, Paterson DG, Hedley-Whyte J. Aspiration of gastric bacteria in antacid treated patients: a frequent cause of post-operative colonisation of the airway. Lancet 1982; i: 242– 245. 3. Hillman KM, Riodan T, O’Farrell SM. Colonisation of the gastric contents in critically ill patients. Critical Care Medicine 1982; 109: 444–447. 4. Craven DE, Kunches LM, Kilinsky V. Risk factors for pneumonia and fatality in patients receiving continuous mechanical ventilation. American Review of Respiratory Disease 1986; 133: 792–796. 5. Reusser P, Scheidegger D, Marbet GA. Role of gastric colonisation in nosocomial infections and endotoxinaemia: a prospective study in neurosurgical patients on mechanical ventilation. Journal of Infectious Diseases 1989; 160: 414–420. 6. Aerdts SJA, Clasener HAL, van Dalen R. Prevention of bacterial colonisation of the respiratory tract and stomach of mechanically ventilated patients by a novel regimen of selective decontamination in combination with initial systemic cefotaxime. Journal of Antimicrobial Chemotherapy 1990; (Suppl. A): 59–76. 7. Jacobs S, Chang RWS, Leer B. Continuous enteral feeding: a major source of pneumonia among ventilated intensive care unit patients. Journal of Parenteral and Enteral Nutrition 1990; 14: 352–357. 8. Torres A, Serra-Batilles J, Ros E. Pulmonary aspiration of gastric contents in patients receiving mechanical ventilation: the effect of body position. Annals of Internal Medicine 1992; 116: 540–543. 9. Inglis TJJ, Sherratt MJ, Sproat LJ. Gastroduodenal dysfunction and bacterial colonisation of the ventilated lung. Lancet 1993; 341: 911–913.

Bacteria in ventilator breathing system 10. Bonten MJM, Gaillard CA, van Tiel FH. The stomach is not a source for colonisation of the upper respiratory tract and pneumonia in ICU patients. Chest 1994; 105: 878–884. 11. Ledingham IMcA, Alcock SR, Eastaway AT. Triple regimen of selective decontamination of the digestive tract, systemic cefotaxime, and microbiological surveillance for prevention of acquired infection in intensive care. Lancet 1988; i: 785–790. 12. Tryba M. Sucralfate versus antacids or H2 antagonists for stress ulcer prophylaxis-a meta-analysis on the efficacy and pneumonia rate. Critical Care Medicine 1991; 19: 942–949. 13. Mahul PH, Auboyer C, Jospe R. Prevention of nosocomial pneumonia in intubated patients: respective role of mechanical subglottic drainage and stress ulcer prophylaxis. Intensive Care Medicine 1992; 18: 20–25. 14. Inglis TJJ, Millar MR, Jones JG. Tracheal tube biofilm as a source of bacterial colonisation of the lung. Journal of Clinical Microbiology 1989; 27: 2014–2018. 15. Inglis TJJ. Evidence for dynamic phenomena in residual tracheal tube biofilm. British Journal of Anaesthesia 1993; 70: 22–24. 16. Inglis TJJ, Jones JG, Paxton S. Penetration of an aerosol, produced by film atomization, through the carinal bifurcation. British Journal of Anaesthesia 1993; 70: 527–523. 17. Murray PR, Barron EJ, Pfaller MA, eds. Manual of Clinical Microbiology, 6th Edn. Washington DC: ASM Press, 1995.

45 18. National Council for Clinical Laboratory Standards. Zone diameter interpretive standards and equivalent inhibitory concentration (MIC) breakpoints. National Council for Clinical Laboratory Standards 1993; 13: no. 24, M2-A5 (disk diffusion), table 2. 19. Versalovic JT, Koeuth T, Lupskin JR. Distribution of repetitive DNA sequences in eubacteria and application to fingerprinting of bacterial genomes. Nucleic Acid Research 1991; 19: 6823–6831. 20. Swerdlow H, Dew-Jager K, Gesteland RF. Rapid cycle sequencing in an air thermal cycler. Biotechniques 1993; 15: 512–519. 21. Inglis TJJ. The stomach is not a source for infection in ICU patients? Chest 1995; 107: 883. 22. Inglis TJJ, Sproat LJ, Sherratt MJ. Gastroduodenal dysfunction as a cause of gastric bacterial overgrowth in patients undergoing mechanical ventilation of the lungs. British Journal of Anaesthesia 1991; 68: 499–502. 23. Johanson WG, Pierce AK, Sanford JP. Changing pharyngeal bacterial flora of hospitalised patients. Emergence of Gram negative bacilli. New England Journal of Medicine 1969; 281: 1137–1140. 24. Craven DE, Goularte TA, Make BJ. Contaminated condensate in mechanical ventilator circuits: a risk factor for nosocomial pneumonia? American Review of Respiratory Disease 1984; 129: 625–628.