Ventilator Associated Pneumonia - Angelfire

15 Pages·1999·0.34 MB·English

Anaesthesia,1999,54,pages1183–1197
................................................................................................................................................................................................................................................
REVIEW ARTICLE
Ventilator-associated pneumonia
Diagnosis, pathogenesis and prevention
P. J.Young1and S. A.Ridley2
1 SpecialistRegistrar,Anaesthesiaand2 DirectorofIntensiveCareandConsultantAnaesthetist,CriticalCareComplex,
NorfolkandNorwichAcuteNHSTrust,BrunswickRoad,NorwichNR13SR,UK
Summary
Ventilator-associatedpneumoniais common,difficultto diagnose, affectsthe mostvulnerableof
patientsand carriesa highmortality.During prolongedmechanical ventilation theoropharynx,
sinuses,dentitionand stomachof criticallyillpatientsbecomecolonised with pathogenicbacteria.
Colonised secretionspoolintheoropharynxand subglottic space. These secretionsrepeatedlygain
access tothelower airwaysby leakage past thetracheal tube cuff. Ifhost defence mechanismsare
overwhelmed,multiplicationoccursinthe lower respiratorytract producing aninflammatory
responseinthe bronchiolesand alveoli.Theinflammatoryresponseischaracterised bycapillary
congestion,leucocyteand macrophage infiltration and fibrinous exudationintothe alveolar spaces.
Ifthisinflammatory responseoccursmorethan 48hafter intubation,it is called ventilator-
associated pneumonia. Prevention dependsonreducingupper airwayand gastrointestinal reservoirs
ofbacteria, reducingorabolishingaspirationof these bacteriapast the trachealtube cuff and
enhancing bacterialclearancefrom the lowerairways.
Keywords Ventilation,mechanical; ventilator-associatedpneumonia. Intubation;tracheal.
......................................................................................
Correspondenceto:DrP.J.Young
Accepted:29April1999
Despiteanincreasedunderstandingofthepathogenesisand pneumonia as the most prevalent infection found in
diagnosis of ventilator-associated pneumonia (VAP), pre- European ICUs [2]. The incidence of VAP in adult ICU
vention has remained problematic. There is now good patientsrangesfrom3to 52%[3].Suchlargevariationin
evidenceforcontinuousleakageofinfectedsecretionspast incidenceisduetocase-mixdifferences,differencesinthe
thetrachealtubecuffbeingresponsibleforVAP.Asaresult, diagnostic criteria used, and the variable sensitivity and
VAPstrikescriticallyillpatientsandincreasestheirmorbid- specificityoftheavailablediagnostictests.Theincremental
ityand mortality. The aim of this review is to discuss the riskofVAPinamedicaladultICUpopulationis,1%per
diagnosisandpathogenesisofVAP,includingabriefreview dayofventilation(Fig.1);forexample,thecumulativerisk
ofnaturaldefencemechanisms.Thereviewalsocoversthe at 10 days of mechanical ventilation was 6.5% and at
prevention of VAP with special reference to new devel- 20 days this had risen to 19% [4]. Tracheal intubation
opments aimedat reducingaspirationpast the cuff. andmechanicalventilationincreasetheriskofpneumonia
in hospitalised patients by 7–21 times [5, 6]. As well as
Definition increasing length of stay by up to threefold [7, 8] and
increasing cost [9], mortality rates are higher for patients
VAP is usually a bacterial nosocomial pneumonia that
withVAP(71%)than for thosewithout(29%)[4]. How-
developsinpatientswithacuterespiratoryfailuresupported
ever,itislikelythatmanypatientsdiewith,ratherthanof,
by mechanical ventilation. The pneumonia was neither
VAP.Crudemortalityratesfor patientswithVAPdepend
presentnor incubatingat thetime ofintubation [1].
on the case mix and range from 13 to 71% [10], and the
mortality directly attributable to the VAP is estimated at
Incidence,morbidityandmortality
27%althoughitishigherwithPseudomonasaeruginosaand
The European Prevalence of Infection in Intensive Care Acinetobacter infections [8]. The bacteria responsible for
(EPIC)studyidentifiedintensivecareunit(ICU)-acquired VAP vary with case mix, institution and antibiotic usage.
Q1999BlackwellScienceLtd 1183
P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia Anaesthesia,1999,54,pages1183–1197
................................................................................................................................................................................................................................................
Figure1 CumulativehazardofVAPin567
patients.Datapointsaremeanvalues(1SD)
(adaptedfromFagonetal.[4]with
permission).
Table1showsthecommonorganismsisolated,but40%of Post-mortem studies demonstrated that following pro-
cases ofVAParepolymicrobial [4]. longed mechanical ventilation, there is widespread poly-
microbial colonisation and infection of the lungs of
critically ill patients [12, 13]. Importantly, the bacterial
Diagnosis
counts from lung biopsy cultures appeared to be poorly
Goldstandard relatedtothepresence,absenceorhistopathologicalgrade
The gold standard for the diagnosis of VAP is combined ofpneumonia.LackofadefinitivemeasureofVAPisone
histopathologicalandmicrobiologicalexaminationoflung ofthemajorconfoundinginfluencesinstudyingstrategies
tissueshowingbothaninflammatoryresponseandmicro- for reducingVAP.
organisms (Fig. 2). Clearly, this is rarely possible in
Clinical diagnosis
critically ill patients when an ante-mortem diagnosis is
TheclinicaldiagnosisofVAPisdifficultanddependsupon
required. However, even this gold standard has recently
been called into question when the rate of histological a clinical suspicion and laboratory tests (see below). The
recognitionofVAPvariedbetweenpathologists[11],and laboratorytestsarerequiredbecauseofthepoorspecificity
themicrobiologialinfectionoflungtissueandhistological (i.e. many false positives) when clinical suspicion alone
is used.
changes of pneumonia were found to be poorly related.
Classification ofprobabilityofVAP
Table1 Frequencyoforganismsrecoveredfromprotectedspeci- Thefollowingdiagnosticcriteriahavebeenadaptedfrom
men brush specimens (>103cfu.ml(cid:255)1) in 83 episodes of VAP. the recommendations of the first international consensus
AdaptedfromFagonetal.[4].
conferenceonthe clinicalinvestigationofVAP[14].
ClinicalsuspicionofVAP dependsupon:
Organism Frequency(%) 1 fever >38.38C, leucocytosis and deterioration in gas
exchange;
Gram-negativebacteria
Pseudomonasaeruginosa 31 2 radiographicappearanceofnewandpersistentinfiltrates;
Acinetobacterspp. 15 3 grosslypurulent tracheobronchialsecretions.
Proteusspp. 15
Unfortunatelyclinicalcriteriaalonehavepoorspecificity
Haemophilusspp. 10
Branhamellacatarrhalis 10 butwithfurther investigation,patientscanbecategorised
Escherichiacoli 8 accordingtothelikelihoodofVAP(Table2).Clearlymost
Klebsiellaspp. 4
diagnosesofVAPfallintotheprobableVAPgroup,andin
Enterobactercloacae 2
Gram-positivebacteria centres relying on endotracheal aspirate (ETA) sampling
Staphylococcusaureus 33 thediagnosis falls outside thisclassification.
Streptococcuspneumoniae 6
Otherstreptococci 15
Thediagnosticvalue of thechestX-ray
Corynebacteriumspp. 8
Anaerobes 2 The consensus diagnostic criteria [14] and the Centers
of Disease Control (CDC) criteria [15] imply that the
1184 Q1999BlackwellScienceLtd
Anaesthesia,1999,54,pages1183–1197 P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia
................................................................................................................................................................................................................................................
Figure2 HistologicalsampleoflungtissueinVAP.
radiographic appearance of new and persistent infiltrates Protectedspecimenbrushing (PSB)
must alwaysbe presentinVAP(i.e.100% sensitivity;Fig. In order to obtain a reliable lower respiratory tract speci-
3). However, many patients without new and persistent men that is not contaminated by tracheal organisms,
infiltrates at post-mortem had histological or microbio- protected specimen brushings can be employed. With
logicalpneumonia[13].Only44%ofpulmonarydensities this technique, the brush is enclosed in a concentric dual
onICUchestradiographsareinfactinfectious[16].Thus, catheterdevicewithitsdistalendoccludedbyadissolvable
while pulmonary infiltrates are essential for the diagnosis plug. When the catheter has been bronchoscopically
of VAP, the cause of such infiltrates is frequently non- placed in the desired distal airway the brush is advanced
infective and VAP mayexistwithout infiltrates. expelling the plug. A brushing is taken and the brush is
Table2 First InternationalConsensus Conferenceon the Clinical Investigationof Ventilator-AssociatedPneumoniacriteria for the
definitionofpneumonia.AdaptedfromPingletonetal.[14].
DefiniteVAP
Clinicalsuspicionplusoneofthefollowing:
1.Positiveneedleaspirateculturefromapulmonaryabscess.
2.Histopathologicalevidenceofpneumoniaonopenlungbiopsyorpost-mortemexamination,andapositivequantitativecultureoflung
parenchymaat>104micro-organismspergramoftissue.
ProbableVAP
Clinicalsuspicionplusoneofthefollowing:
1. The presence of positive quantitative culture by reliable lower respiratory specimen (BAL or protected specimen brush at >104 and
>103cfu.ml(cid:255)1,respectively).
2.Thepresenceofapositivebloodcultureunrelatedtoanothersourceof,andidenticalorganismto,thatrecoveredfromthelowerrespiratory
tract.
3.Positivepleuralfluidcultureofanidenticalmicro-organismtothatrecoveredfromthelowerrespiratorytract.
ProbableabsenceofVAP
Lackofsignificantgrowthinareliablerespiratoryspecimenwithoneof:
1.ResolutionofclinicalsuspicionofVAPwithoutantibiotics.
2.Alternativediagnosisestablishedforfeverandinfiltrates.
DefiniteabsenceofVAP
1.Post-mortemshowsnohistologicalsignsoflunginfection.
2.Definitealternativeaetiology,andnegativereliablelowerrespiratoryspecimen.
Q1999BlackwellScienceLtd 1185
P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia Anaesthesia,1999,54,pages1183–1197
................................................................................................................................................................................................................................................
Figure3 ChestX-rayofapatientwithVAP.
then retracted and isolated within the catheter, before untilresistanceisfelt.Aspecimenisobtainedbyexpressing
beingwithdrawnthroughthebronchoscope.Thebrushis andretractingtheinnercannulaandbrushaswithbroncho-
placed in 1ml of transport medium and a quantitative scopic PSB. Nonbronchoscopic BAL is performed by
culture obtained. A growth of 103 colony forming units guidingadirectionaltipcathetertothemainstembronchus
(cfu)permlindicatesapositiveresult.Thisequatesto105– ofthelungwithsuspectedpneumonia.Theinnercannula
106cfu.ml(cid:255)1 in the bronchial secretions. In 16 reported is then advanced until resistance is felt and the catheter is
studies,PSBquantitativecultureshadameansensitivityof wedged. Up to 100ml of saline is instilled and suctioned
82%and a meanspecificityof92% [17]. until at least 25ml of specimen is obtained. This blind
techniqueisquickerandcheapertoperformandhasgood
Bronchioalveolarlavage(BAL) concordance with bronchoscopic techniques [19]. The
Bronchioalveolar lavage is performed by wedging the goodagreementbetweenbronchoscopicandnonbroncho-
bronchoscope in a distal airway and instilling 120ml of scopic techniques emphasises the disseminated nature of
normalsaline[18].Whenlavagevolumesof10–20mlare lower respiratorytract colonisation and infectionin intu-
used this is called mini-BAL and if a balloon-tipped bated critically ill patients. In nine reported studies of
catheterisusedtoisolatethebronchopulmonarysegment, nonbronchoscopic techniques (PSB and BAL), the mean
this is known a protected BAL. As much of the saline sensitivitywas 80%and amean specificity was 82%[17].
as possible is aspirated and quantitatively cultured. This
methodsamplesalargerportionoflungparenchymathan Endotracheal aspirates
PSB.Agrowthof104cfu.ml(cid:255)1indicatesaVAP.Thisalso An ETA is collected in a sputum trap by passing a sterile
equates to 105–106cfu.ml(cid:255)1 in the bronchial secretions. suction catheter blindly down the tracheal or tracheost-
AlsoBALfluidenablesearlierdiagnosisasthemicroscopic omytubeintothedistaltracheaorproximalbronchialtree
identification ofintracellularorganismsin>2%ofphago- and applying suction as the catheter is withdrawn. The
cyticcellsis >95%specific for pneumonia. bacterial origin may therefore be from bronchial or
In seven reported studies, BAL quantitative cultures trachealsecretions,tubebiofilmorcontaminant.Specificity
had a mean sensitivity of 86% and a mean specificity of is therefore unfortunately low, particularly if nonquanti-
87%[17]. tative cultures are used. In five reported studies of ETA
nonquantitative cultures, the mean sensitivity was 78%
NonbronchoscopicPSBandBAL withameanspecificityof19%.Inninereportedstudiesof
Nonbronchoscopic PSB is performed by inserting the ETA quantitative cultures, the mean sensitivity was 69%
brush directly through the tracheal tube to <35cm or with a mean specificity of 80% [17]. The significant
1186 Q1999BlackwellScienceLtd
Anaesthesia,1999,54,pages1183–1197 P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia
................................................................................................................................................................................................................................................
growth level of 106cfu.ml(cid:255)1 is often, but not invariably, relationship between bacterial counts from lung biopsy
used.Iflower valuesareusedthenthetestbecomesmore and the presence of pneumonia brings into question the
sensitiveand lessspecific. validity of bronchoscopically guided BAL and PSB ana-
lyses for the diagnosis of pneumonia. Kirtland et al. [13]
Complications ofBAL,PSB and ETA found that qualitative ETAs had a sensitivity of 87% in
The most common complication of these procedures is recognising bacteria simultaneously present in lung par-
hypoxaemia, particularly associated with BAL. A median enchyma, and suggested that empirical antibiotic therapy
fallinarterialoxygenpressureof1kPawasseeninastudy canbeaccuratelyguidedbyETAs.QuantitativeETAsmay
of 35 patients with unchanged inspiratory oxygen before beasusefulasBALandPSBforthediagnosisofVAPand
and after BAL [20]. PSB can cause minor bleeding and as a guide to appropriate antibiotic treatment [25, 26]; in
threepneumothoracesoccurredin25patientsinonestudy fact, bronchoscopic techniques may lead to greater anti-
[21]. However, ETAs are safe and, apart from minor bioticusewithnoreductioninmortality[27].Advantages
haemorrhage, serious complications are uncommon. ofETAcomparedwithbronchoscopictechniquesinclude
simplicity (as it can be performed by nonmedical staff),
Choicebetween PSB,BALorETA fewer delays in sampling and therefore more expedient
Thechoiceofdiagnostictestdependsuponthefollowing: antibiotictreatment,fewercomplicationsandreducedcosts.
1 The sensitivity and specificity of the three techniques Overall, while bronchoscopic techniques are probably
(Table 3). The choice of technique depends upon how themostscientificallyrobustatpresent,theeaseofuseand
important the clinician feels the diagnosis is. If no cases comparable performance of quantitative ETAs makes this
shouldbemissedthenthetestwiththehighestsensitivity thetechnique ofchoice inmostICUs.
should be chosen (e.g. bronchoscopic BAL) or lower
thresholds used for the quantitative culture (reduced from
104to 103cfu.ml(cid:255)1). Pathogenesis
2 Prior antibiotic therapy. If given prior to the test, anti- Colonisationimpliesthepresenceofbacteriaonthemucosa
biotics alter the investigation’s receiver–operator charac- without a host response. In health the oropharynx is
teristic curve and so alter the combination of sensitivity colonised by nonpathogenic bacteria and the lower
and specificity [22]. Ideally specimens need to be taken respiratory tract is sterile. In the ventilated patient, the
priortotheintroductionofnewantibiotics,butinclinical colonisationofthecontiguousstructuresofthesubgingi-
practicethis may be difficult. valplaque,peridontalareas,oropharynx,sinuses,stomach
3 Complicationrate,easeofsamplingandspeedofresult and trachea with pathogenic bacteria have been termed
alsoinfluence choice. transcolonisation [28]. This transcolonisation normally
BALand ETA allowearlier identification of organisms occurs prior to the development of VAP [29]. Colonisa-
than PSB as the fluid can be examined microscopically. tionpatternsvaryandalthoughtheentericGram-negative
PSBandBALaremoreinvasive,expensive,labour-inten- bacteria (GNB) usually colonise the oropharynx before
sive and associated with more complications than ETA. thetrachea,theoppositemaybetrueofPseudomonasspp.
The need for specialised equipment and staff to obtain a [30, 31]. This is caused by the increased binding affinity
BALorPSBspecimencoulddelayantibioticintroduction of Pseudomonas spp. to ciliated tracheal epithelial cells
compared with collecting an ETA. The introduction of comparedwithbuccalcells.Thereforeconcomitantexpo-
earlyappropriateantibiotictreatmentshouldbeapriority sure of the oral and tracheal surfaces to an inoculum of
as this improves the outcome of VAP [23, 24]. The poor Pseudomonas mayonlyleadto tracheal colonisation [28].
Table3 Sensitivityandspecificitydatafromclinicaltrialsforquantitativeculturesofprotectedspecimenbrushes,BAL,nonbroncho-
scopicPSBorBALandendotrachealaspirates;andnonquantitativeendotrachealaspirates.MeanvaluesadaptedfromGriffin&Meduri
[17].
Bronchoscopic BronchoscopicBAL Nonbronchoscopic Endotracheal Endotracheal
PSB BAL BALorPSB aspirate aspirate
Technique (Quantitative) (Quantitative) (Quantitative) (Quantitative) (Nonquantitative)
Numberofstudies 16 7 9 9 5
Meansensitivity[range];% 82[62–100] 86[72–100] 80[60–100] 69[38–91] 78[57–88
Meanspecificity[range];% 92[60–100] 87[69–100] 82[66–100] 80[59–92] 19[0–33]
Q1999BlackwellScienceLtd 1187
P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia Anaesthesia,1999,54,pages1183–1197
................................................................................................................................................................................................................................................
Normaladultsaspirateoropharyngealfloraduringsleep Sourcesof colonising bacteria – upper airwayand
[32] yet rarely develop pneumonia. There is a balance gastrointestinalbacterialreservoirs
betweenthesizeandnatureoftheinoculumandthehost’s
defence capabilities as to the susceptibility to VAP. For Dentition
example, healthy hamsters withstand infection with an During critical illness effective oral hygiene, despite best
intratracheal inoculation of 6· 103 organisms of Ps.aeru- nursingcare,willdeclineandthereisarapidshiftofdental
ginosa;however,whenincreasedto5· 106organisms,foci plaque colonisation from nonrespiratory pathogens to
ofbronchopneumoniaareobservedwhichresolvewithout aerobic GNB and Staphylococcus aureus [39]. Dental
treatment [33]. Bolus inoculation or aspiration of a nidus plaque increases with ICU stay, and the bacterial flora
ofcontaminatedmaterial,asopposedtoaerosolisation,are frequentlymirrorsthat oftracheal aspirates [40].
both more efficient in producing pneumonia in animal
models [28].
Oropharynx
Host susceptibility is also an important determinant of
In health, the oropharynx is colonised predominantly by
whethertissueinvasionandVAPwilldevelop.Pre-existing
nonpathogenic bacteria, but in hospitalised patients, this
lung injury in the hamster model increased the bacterial
flora is replaced with pathogenic bacteria, predominantly
concentrationsinlungtissueandtheseverityofpneumonia
aerobicGNBandstaphylococci,andthisincreasestherisk
at a given inoculum [33]. In humans, VAP occurs more
ofVAP[41].Prolongedexposureoftheepithelialsurface
often in patients with acute respiratory distress syndrome
tothebacterialadhesinmoleculesandthechangesinhost
(ARDS)[34,35].Thereisalsoachangeinbacterialflora
epithelial cell receptors in critical illness promote this
with illness. At post-mortem the pulmonary parenchyma
abnormal colonisation. Fibronectin is normally secreted
of previously healthy individuals invariably contained
into the oropharynx with the saliva and coats the buccal
oropharyngealflorawithinfrequentaerobicGNB.How-
epithelium. The fibronectin creates an adhesive surface
ever, hospitalised patients’ lungs have a 62% incidence of
favouring the binding of commensal oral strepococci.
aerobic GNB [36]. Aerobic GNB colonisation increases
Decreases in serum fibronectin associated with sepsis
with increasing severity of illness, as indicated by the
maycausedecreasesinsalivaryfibronectin.Alsoorophar-
presence of coma, respiratory tract disease, hypotension,
yngeal inflammation and the release of polymorphonuc-
tracheal intubation, acidosis, uraemia and either leuco-
lear elastase increase proteolytic breakdown of buccal
cytosisorleucopenia[37].Comorbiditiessuchasdiabetes
fibronectin and many respiratory pathogens also release
mellitus, malnutrition, renal failure, neoplastic and liver
proteases capable of degrading fibronectin. Fibronectin
diseaseallincreasetherateofGNBcolonisationandimpair
depletion reduces streptococcal binding sites and favours
leucocyte function. Pulmonary oedema may impair
aerobic GNB colonisation [28]. Aerobic GNB colonisa-
respiratory bacterial clearance; resultant surfactant loss is
tion rates increase with severity of illness and systemic
associated with impaired alveolar macrophage function
antibiotictherapy[37]byreducingthenormalinhibitory
andalveolarfluidcanserveasabacterialgrowthmedium
flora. In the critically ill, supine, ventilated patient oro-
[38]. The combination of comorbidities, and especially
pharyngeal secretions pool in the oropharynx and sub-
lungpathology,weakensthehostdefencesagainstbacterial
glotticspaceabovethetrachealtubecuff[42],soforminga
colonisationandhenceVAPincriticallyillpatients(Fig.4).
reservoirofinfectedsecretions.
Sinuses
The risk of nosocomial sinusitis is increased with naso-
tracheal and nasogastric intubation [43]. The incidence
(1.4–100%) of sinus infection is often underestimated
because of a lack of clinical signs. The predominant
organisms involved are aerobic GNB, there is a clear
association between sinusitis and VAP, and sinusitis
increasesthe riskofVAPby3.8 [44].
Stomach
The stomach is a potential large reservoir for aerobic
GNB. Reductions in gastric acidity increase the rate of
gastric colonisation. H antagonists and antacids increase
2
Figure4 ThepathogenesisofVAP. pHandgastriccolonisationmorethan thecytoprotective
1188 Q1999BlackwellScienceLtd
Anaesthesia,1999,54,pages1183–1197 P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia
................................................................................................................................................................................................................................................
agentsucralfate[45,46].Riskfactorsforgastriccolonisation
includeduodenal[47]andgastric[48]enteralfeeding.The
importanceofthegastropulmonaryrouteforVAPpatho-
genesis remains controversial because studies of colonisa-
tion sequences prior to the development of VAP have
failed to show consistently the stomach as an initial or
precedingsiteofcolonisation[48–50].However,manyof
theorganismscausingVAPdoarisefromthegastrointesti-
naltractandthisisthebasisforselectivedecontamination
ofthedigestive tract (SDDT).
Faeco-oral orfaeco-trachealroute
Cross-colonisationortransferofrectalbacteriabynursing
andmedicalstaffmayplayaroleincolonisation[50],and
good infection control measures are essential, but this
route should only play a minor role in the pathogenesis
ofVAP inhospitalised patients.
Microbialentry mechanisms
Aspirationpastthe trachealtube cuff
Thehigh-volumelow-pressure(HVLP)cuffof atracheal
tube does not prevent the leakage of potentially infected
fluid from the subglottis to the trachea. Dye placed in
the subglottic space will pass along folds within the cuff
wall to the trachea in 87–100% of cases [51–53]. In a
ventilatory model [54] and in the cadaveric trachea [55],
therateofleakagealongthecufffoldswasfoundtobeof Figure 5 An HVLP cuff inflated to an intracuff pressure of
the order of magnitude of millilitres per minute and 30cmH O in a 2-cm-diameter cylinder demonstrating the
2
leakage occurred despite increasing the cuff pressure mechanismofleakagealongfoldswithinthecuffwall.
above normal(Fig.5).
Contaminationthrough the respiratoryapparatus Ventilator circuit
Theusualsequenceofcontaminationofthetrachealtube A recentsystematic reviewof randomised trials of airway
andventilatorcircuithasbeendemonstratedrecently[56]. managementintheICU[58]showedthatthefrequencyof
Normallythebacteriaareacquiredendogenouslyfromthe ventilatorcircuitchangeshadlittleinfluenceontherateof
oropharynxorstomach,seedthetracheapastthetubecuff VAP, although infrequent changes were associated with a
andsubsequentlycontaminatethetrachealtubeandventi- modest decrease in VAP rate. The increased rate of VAP
lator tubing. However, Ps.aeruginosa may colonise the withcircuitchangesmightbeduetomoremanipulations
tracheawithoutfirstcolonisingtheoropharynxasdiscussed ofthepatient,thetrachealtubeandtheventilatortubing.
previously[30,31].Rarelydirectinoculationthroughthe Thiscanresultintheinadvertentflushingofcontaminated
respiratoryapparatusmayberesponsible,insomecasesdue tubing condensate into the patient’s trachea or increased
to cross-contamination or inadequate cleaning and steril- leakage of secretions past the cuff of the tracheal tube. A
isation procedures. recent study of 637 patients showed that 7- and 30-day
circuit change intervals have lower risks for VAP than
Biofilms 2-day changeintervals [59].
The inner lumen of a tracheal tube rapidly develops Closedsuctionsystemshavebeendevelopedformultiple
viscous adhesive layers of accretions containing bacteria, usetoremovetherequirementtobreaktheclosedcircuit
including aerobic GNB and Staph.aureus, at concentra- inordertopassasuctioncatheter.Thisdevicereducesthe
tions of up to 106cfu.cm(cid:255)1 of tube. Particles of this incidenceofarterialdesaturationduringsuctioningdueto
biofilm can be propelled into the trachea and distal lung themaintenanceofventilationandpositiveend-expiratory
by shear forces of gas flow and passing tracheal suction pressure (PEEP) [60]. However, there is no difference in
catheters[57]. VAP rates with closed, as opposed to open, tracheal
Q1999BlackwellScienceLtd 1189
P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia Anaesthesia,1999,54,pages1183–1197
................................................................................................................................................................................................................................................
suction systems (increased tracheal colonisation occurred suchasantimicrobiallysozyme,secretoryimmunoglobulin
with the closed system in one study [61]), despite their (Ig)Aandantiproteases,areusuallyresponsibleforbacter-
11-foldcost[62].Furtherevidenceofthebeneficialeffect ial degradation and clearance. In health a cuffed tracheal
ofclosed suction systemsisawaited. tubecausesmucosalinjuryatthevocalcords,atthelevelof
thecuffandatthetubetip[68].Bileandgastricaspiration
Humidifier may exacerbate these injuries and delay healing [69].
Heat and moisture exchange (HME) humidification, as Tracheal toilet with suction catheters causes tracheal and
opposed to heated humidifiers, may have a protective brochialmucosalinjury[70].Mucosalinjuryandexposure
effectthroughminimisingtubingcondensateandcolonis- of basement membrane predispose to bacterial adhesion
ation. HME humidifiers have a relative risk of VAP [28].
occurrenceof0.34–0.86[58].Ventilatortubecondensate
frequentlyhashighbacterialcountswithamedianlevelof
colonisationat24hof7· 104organisms.ml(cid:255)1;condensate Alveoli
cancollectincircuitsatameanrateof30ml.h(cid:255)1[63].This Bacteria entering the alveolus are exposed to opsonins
(e.g. IgA, IgG, complement 3b, surfactant phospholipids)
infected condensate may wash back from the ventilator
inthefluidliningtheepithelium.Coatingofthebacteria
tubing intothe trachea.
withopsoninsenhancesalveolar macrophagephagocytosis.
Phagocytosedbacteriaareexposedtoahighlybacteriocidal
Nebulisers
mixture of proteases, lysozyme, acid and free radicals.
Asmanyas68%ofin-circuitnebulisers,ifusedrepeatedly
Alveolar macrophagesreleaseneutrophilchemotacticfac-
inthesamepatient,becomecontaminatedwithhighlevels
of organisms (>103.ml(cid:255)1). These bacteria are aerosolised tors and early inflammatory cytokines such as tumour
necrosis factor (TNF)-a and interleukin (IL)-1b. TNF-a
to small particles capable of reaching the peripheral lung
and IL-1b stimulate lung endothelial cells to express sur-
[64]. Thorough cleaning of the nebulisers between treat-
faceadhesionmolecules,intercellullaradhesionmolecule-1
mentsreducedthis contaminationto 20%,and theuse of
and extracellular leucocyte adhesion molecule-1. These
moderndisposablesystemsmayreducethis further.
bindtocirculatingneutrophilsfacilitatingtheir migration
intotheinterstitiumandultimatelytothealveolus.Thein
Hostdefence mechanisms – microbialclearance
vitro clearance of bacteria by neutrophil phagocytosis is
Microbial clearance falls into two broad categories,
dose dependent, the maximum bacteriocidal capacity
mechanical mechanisms and immunological mechanisms.
being a ratio of 10 bacteria to one neutrophil [71]. Even
the defences of healthy lung tissue will be overwhelmed
Cough
whenexposed to anexcessive bacterial load.
Atrachealtubebypassestheepiglottistherebyreducingan
effectivecough.Sedation,opioidanalgesia,neuromuscular
alsoall depressthe coughreflex.
Prevention
Mucociliary clearance Reduction in pathogenic bacterialreservoirs
Cilialbeatingonviscousmucuspropelsdebristowardsthe Reductions in pathogenic bacterial reservoirs can be
larynx for expectoration. As well as providing a physical achieved bythe methodsdescribedbelow.
barriertomucusclearance,trachealintubationcanreduce
tracheal mucus velocity [65]. Increasing oxygen concen-
Nursingcare
trationsareassociatedwithtracheobronchialinflammation
Strictinfection-controlproceduresshouldbeinplace,for
and epithelial sloughing and a progressive reduction in
example,handwashingandthesterilisationofequipment.
tracheal mucus velocity[66].
Routineoralhygieneandtheremovalofsecretionsinthe
Inadequate humidification causes dehydration of the
mouth and subglottis are important [42, 72]. The use of
periciliarymucuslayeranddamagetociliatherebyimpair-
oral chlorhexidine rinse in cardiac surgical patients
ingthefunctionofthemucociliaryelevator;thedegreeof
reduced postoperative respiratory infections by 69%, and
damageisproportionaltothedurationofventilationwith
subsequently reduced antibiotic usage. Antibiotic resis-
drygases[67].Thus,intubationitselfpredisposedtoVAP
tancewasunaffected[73].
throughthe breakdownofhost defences.
Tracheobronchialtree Oralvs.nasal intubation
Immunologically active cells, such as neutrophils and The use of oral rather than nasal tracheal intubation has
macrophages, and substances within the mucus layers, beenassociatedwithlower ratesofVAP,andisduetothe
1190 Q1999BlackwellScienceLtd
Anaesthesia,1999,54,pages1183–1197 P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia
................................................................................................................................................................................................................................................
reducedincidenceofsinusitiswithoralasopposedtonasal appear to reduce gastro-oesophageal reflux in normal
tracheal and gastrictubes (95.5%vs. 22.5%)[43, 44]. subjects[82].
Selectivedecontamination of the digestivetract Enteral feeding
SDDT involves infection control by general hygiene, Both continuous and intermittent (with a 6-h break)
topical nonabsorbable bacteriocidal antimicrobials admi- enteral feeding increase gastric pH, and are associated
nistered to the oropharynx and gut often with parenteral with Gram-negative colonisation of the stomach (80%
antibiotics. SDDT reduces colonisation rates and lower gastriccolonisationinbothgroups)[83].Adequatenutri-
respiratorytractinfectionwithoutdemonstrablechangein tional status prevents VAP [30] and feeding enterally is
mortality[74],althoughsubgroupanalysishasshownsome undoubtedly the route of choice, so the clinician must
benefitintraumapatients[75].Itsuseremainscontrover- decideuponthebalanceofrisks.Avoidingstomachover-
sialbecauseofconcernsregardingincreasesincolonisation distension by monitoring residual gastric volume during
and infection rates with multiresistant pathogens [76]. feedingprotocols,andtheplacingofjejunalfeedingtubes
SDDT reinforces the point that patients may die with maybe appropriate.
rather than of VAP. Respiratory failure is an important
organ failure in the critically ill and possible mechanisms Ventilator circuit
for reducingVAParetobe encouraged. The ventilator circuit should be actively managed so that
condensate fluid does not wash back into the patient but
can be removed from water traps. HME filters should be
Gastric alkalinisation
considered. Tubing should be changed only if heavy
TheuseofantacidstoincreasegastricpHisassociatedwith
soilingoccurs.
the appearance of more new pathogens in the gastric
contents than if a surface-active agent such as sucralfate
Reduction in aspiration pastthe cuff
isused(anincidenceof35%intheantacidgroupvs.17%
inthesucralfategrouponthethirdpostoperativeday)[46].
Subglotticsecretiondrainage
Although there are conflicting data, the gastric colonisa-
Subglotticsecretiondrainageisperformedusingatracheal
tionassociatedwithH antagonistscomparedwithsucral-
2 tube incorporating a separate dorsal lumen ending in the
fateshowsatrendtowardsanincreaseinVAPrates[77].A
subglottic space just above a HVLP cuff (HI-LO EVAC
meta-analysis of trials comparing ulcer prophylaxis in the
tube, Mallinckrodt). Fluid can be drained along the
ICU showed lower rates of lower respiratory tract infec-
channelwithsuction.Thiscanreduce,butnoteliminate,
tionswithsucralfate(a45%riskreductioncomparedwith
the volume of fluid aspirated into the lungs. Such a tube
pH-alteringdrugs)[78].However,inarecentmulticentre,
reduced the incidence of VAP from 29.1% to 13% with
randomisedstudyof1200patients,nosignificantdifference
intermittentdrainage[42],andfrom32.5%to18.4%with
inVAPrates,ICUstayormortalitycouldbedemonstrated,
continuousdrainage[84] ofthe subglotticspace.
although therewas a trendtowardslower VAP rateswith
sucralfate. The ranitidine group had lower rates of clini-
Cufftype
callysignificantgastrointestinalbleedingthanthesucralfate
HVLP tracheal tube cuffs do not provide the barrier to
group (1.7% vs. 3.8%) [79]. These conflicting data make
fluid entering the tracheobronchial tree afforded by low-
recommendations difficult. However, ulcer prophylaxis
volume high-pressure (LVHP) cuffs [53]. Unfortunately
with sucralfate alone maybe inadequate for patients at
LVHPcuffsareassociatedwithtrachealwallinjuryandare
highest riskofpeptic ulceration.
unsuitable for use in long-term ventilation. Two new
tracheal tubes have beendescribedthatappear to prevent
Gastro-oesophageal reflux aspiration with acceptable tracheal wall pressures. The
There is a fourfold reduction in pulmonary aspiration of pressure-limited cuff (PLC) is a LVHP cuff with inflation
radioactive-labelled gastric contentswith the semirecom- characteristicsthatenabletrachealwallpressurecontrolat
bent(458angle)ratherthansupineposition.Furthermore, 30cmH Owhilesimultaneouslyprovidingcompletepro-
2
the isolation of the same organisms from the stomach, tection against aspiration to the lungs [55] (Fig. 6). The
pharynx and endobronchial samples occurred in 32% of cuff is used in combination with a constant-pressure
semirecombent patients compared with 68% of supine inflation device and has been shown to reduce the
patients [80]. Gastro-oesophageal reflux,however, occurs aspiration of dye placed in the subglottis from 87% with
irrespective of body position in mechanically ventilated aconventionalHVLPcuffto0%withthePLC[52].The
patients with nasogastric tubes [81]. The use of fine bore ‘no pressure’ laryngeal seal design by Kolobow and
(8F)asopposedtolargebore(14F)feedingtubesdoesnot co-workers has been evaluated in sheep. This tube has
Q1999BlackwellScienceLtd 1191
P.J.YoungandS.A.Ridley • Ventilator-associatedpneumonia Anaesthesia,1999,54,pages1183–1197
................................................................................................................................................................................................................................................
Figure 6 The PLC inflated to a tracheal wall pressure of Figure7 AnHVLPcufflubricatedwithKYjellyandinflatedto
anintracuffpressureof30cmH Oina2-cm-diametercylinder.
30cmH Opreventsleakageina2-cm-diametercylinder. 2
2
KYjellypreventsleakage.
12–20 thin polyurethane film gills that are positioned at
Enhancement ofbacterial clearance fromairways
theglottistherebypreventingairleaksandfluidaspiration
andlung tissue
[85,86].InvitrostudiessuggestthatPEEPandlubrication
Aswellastrachealsuctioning,threemechanismsmayhelp
(Fig. 7) of HVLP cuffs may, at least temporarily, prevent
bacterialclearance.
leakageof fluid past thecuff [54].
Kinetic beds
Noninvasiveventilation In five studies (only one study showing a significant
Arecentstudyshowedthatextubationandweaningwith difference),kineticbedshavebeenassociatedwithareduc-
noninvasive (by facemask), as opposed to invasive (by tioninVAP,althoughnodifferenceinICUstayormortality
tracheal tube), ventilation can reduce VAP (28% vs. 0%) was reported [58]. Possible advantages of rotational bed
and mortality in suitable patients with respiratory failure therapyincludeintrathoracicposturaldrainage,thelimita-
secondary to chronic obstructive airways disease [87]. tionofpooledsecretionsintheupperairwaywithrotation
Survival rates at 60 days were significantly higher in the andimprovementsinpulmonarygasexchange.Thesebeds
noninvasivegroup(92%vs.72%);theexcessdeathsrelated are expensive; there are inconsistent data demonstrating
to the complications of conventional mechanical ventila- theireffectiveness,androutineusecannotberecommended
tion, inparticular pneumonia. at present.
Unplanned extubations Physiotherapy
Appropriate securing of tracheal tubes and the preven- Chest physiotherapy, postural drainage, percussion and
tion of the need for reintubations may reduce the risk of vibration in association with tracheal suctioning, are used
VAP [88]. topromotesecretiondrainageandimprovelungexpansion
1192 Q1999BlackwellScienceLtd