VAP diagnosis: a true quality
indicator or game of chance?
Tim Walsh
Professor of Critical Care
Edinburgh University
Disclosure
• In past received unrestricted educational
sponsorship from Astra Zeneca and Wyeth to
support epidemiology and quality
improvement research relating to ICUacquired infection
Ventilator Associated Pneumonia
(VAP)
• Commonest ICU acquired infection
• Associated with higher mortality, duration of
MV, healthcare costs
– Uncertain attributable mortality
• Melson et al. CCM 2009; 37: 2709
• Bekaert et al. JAMA 2011; 184: 1133
• Delays in diagnosis and appropriate therapy
associated with adverse outcomes
• Modifiable risks:
– Sedation and weaning practice
– Ventilator circuit management
– Hand-washing/infection control
– Antibiotic use
• Preventative interventions:
– Selective oral decontamination
Pileggi et al. Critical Care 2011, 15:R155
• Antibiotics
(RR 0.27 (95% CI = 0.16 to 0.37))
• Antiseptic
(RR 0.36 (95% CI = 0.18 to 0.50))
– Patient Position
Drakulovic MB et al. Lancet 1999; 354:1851–1858
van Nieuwenhoven CA et al. Crit Care Med 2006; 34:396–402
– Sub-Glottic suction (RR 0.61, 95% CI: 0.46–0.79)
Wang et al. J Trauma. 2012;72: 1276–1285
– Reduced biofilm formation and colonisation
Kollef et al JAMA 2008; 300:805–813.
– Probiotics
Quality measure
• Associated with relevant, measurable adverse outcomes
– Clinicians
– Patients
– Administrators
• Modifiable risk factors at system level
• Relatively cheap interventions amenable to system-wide
implementation
• Focus of National Quality Programmes
• Potential link to reimbursement
• Potential driver to explore system “failure”
Ventilator-associated pneumonia rates at major trauma centers compared
with a national benchmark: a multi-institutional study of the AAST.
Michetti CP; et al. AAST Ventilator-Associated Pneumonia Investigators
The Journal of Trauma and Acute Care Surgery. 72(5):1165-73, 2012
Range: 1.8 to 57.6 per 1,000 ventilator days
2
ICU admission
Mechanical ventilation
Population
Risk period
Surveillance method
• Personnel
• Methodology
Diagnostic approach
• Clinical definitions
• Microbiological approach
Extubation
Decannulation
ICU admission
Mechanical ventilation
Population
Risk period
Surveillance method
• Personnel
• Methodology
Diagnostic approach
• Clinical definitions
• Microbiological approach
Extubation
Decannulation
Population
• Exclusions
• Risk factors
– Chronic lung disease; smoking
– Acute illness-related factors (sepsis; ARDS;
trauma; neurological disease)
– Prior antibiotic exposure
• “Pre-morbid illness” and “Acute illness” –
related bias
ICU admission
Mechanical ventilation
Population
Risk period
Surveillance method
• Personnel
• Methodology
Diagnostic approach
• Clinical definitions
• Microbiological approach
Extubation
Decannulation
Risk Period
• “Sampling” bias
– 105 trauma/surgery patients; BAL within 48 hours
– 58% >104 CFU; many organisms subsequently implicated in
VAP
Wahl WL et al Surgery 2011; 150:665–672.
• “Exposure” bias
– Case Mix
• Individual patient cumulative risk varies according to duration of
MV
• Peak risk early first week?
– Cook DJ et al Ann Intern Med 1998; 129:433–440.
• Sundar KM et al. J Crit Care 2011.
– Community versus Academic Institution (0 versus 2.4 per 1000
ventilator days)
– Same organisation, staffing pattern, medical staff; VAP bundles
ICU admission
Mechanical ventilation
Population
Risk period
Surveillance method
• Personnel
• Methodology
Diagnostic approach
• Clinical definitions
• Microbiological approach
Extubation
Decannulation
Surveillance method
Bedside based
Clinician
•
•
•
•
Infection control
Office based
Other
Prospective vs Retrospective
Diagnostic criteria
Independent vs non-independent
Validated vs non-validated
Clinical
•
•
• Audience (manager vs clinician)
• Feedback mechanisms
Rate per 1000 ventilator days
Non-clinical
Data sources
Validation
Diagnosis
• Definition used
• Inter-observer variability
– Within groups
Surveillance staff: rates 20-40%; low consensus
Am J Infect Control 2010; 38:237–239.
– Between groups
Intensivists 28%; Surveillance staff 8%
Am Surg 2011; 77:998–1002.
• “Ascertainment” bias
– Staff group
– Implications of “the positive”
Ventilator-associated pneumonia rates at major trauma centers compared with
a national benchmark: a multi-institutional study of the AAST.
Michetti CP; et al. AAST Ventilator-Associated Pneumonia Investigators
The Journal of Trauma and Acute Care Surgery. 72(5):1165-73, 2012
• Range: 1.8 to 57.6 per 1,000 ventilator days
• No correlation with ISS, diagnostic methodology used
hospital size
• Service capturing data (P = 0.001):
Trauma service 26.4% Mixed 18.9% Surveillance 11.3%
2
American College of Chest Physicians
New or progressive consolidation on chest radiographs and
At least two of the following:
fever, abnormal white blood cell count, and/or purulent secretions
Chest 1992; 102:553S–556S
CDC National Healthcare Safety Network
Agreement between intensivists
poor ( 0.47)
Tejerina J Crit Care 2010; 5:62
• NHSNetwork (CDC diagnosis): 1.2 per 1000 ventilator days
– Surveillance led
• ACCP diagnosis: 8.5 per 1000 ventilator days
– Clinician led
• Similar proportion positive microbiology (88% vs 92%)
•  statistic 0.26
Definitions
Clinical Definition
Microbiological
Confirmation
•
•
•
•
Clinician judgement
CDC criteria
ACCP criteria
HELICS criteria
•
•
•
•
•
•
Culture negative
ETA (qualitative)
ETA (quantitative)
BAL (qualitative)
BAL (quantitative)
Other
ICU admission
Mechanical ventilation
Population
Risk period
Surveillance method
• Personnel
• Methodology
Diagnostic approach
• Clinical definitions
• Microbiological approach
Extubation
Decannulation
53 cases with clinical VAP
Paired ETA and BAL samples
BAL as “Reference standard”
Simulation based on 12
months of prospective
surveillance data for clinical
VAP
Statistical process chart showing incidence of confirmed VAP in the preintervention and post-intervention periods.
Rate of diagnosis with BAL
37%
Incidence of clinically suspected VAP 29
Incidence of microbiologically
confirmed VAP
18
(per 1000 ventilator days)
58%
25
P = 0.16
9
P = 0.0012
Morris A C et al. Thorax 2009;64:516-522
An “external” (regional; national; international)
quality indicator?
• Not fit for purpose
• Require significant investment to ensure
robust and meaningful
• Potentially subject to multiple systematic
biases in reporting
• “Zero” rating potentially encourages apathy
An “Internal” quality indicator?
Chlorhexidine 1% >95%
>30% head-up
>95%
“Wake and wean” 70%
Pre-VAP bundle
Post-VAP bundle
Patients >48 hours
1460
501
Clinical VAP /1000
ventilation days
32
12
<0.001
Micro confirmed VAP/1000
ventilation days
14
6
<0.001
Mortality (%)
25
20
0.03
ICU LOS (median)
6
6
0.5
Antibiotic days (median)
4
4
0.2
Pre-VAP bundle
Post-VAP bundle
Patients ≥ 14 days
335
113
Clinical VAP /1000
ventilation days
37
18
<0.001
Micro confirmed VAP/1000
ventilation days
13
8
<0.001
Mortality (%)
25
16
0.07
ICU LOS (median)
22
20
0.01
Antibiotic days (median)
16
13
<0.001
VAP diagnosis: a true quality indicator or game
of chance?
• For comparison between ICUs VAP rates subject to a
myriad of potential confounders and biases
• As a quality indicator within a service:
– Awareness of factors influencing data
– Presentation of data to drive “Plan-Do-Study-Act” cycles
– Measurement of outcomes relevant to clinicians and
patients (as well as administrators)