1. Ventilator-Associated Events A Patient Safety Opportunity
CUSP for Ventilator-Associated Pneumonia
December 4, 2014
Michael Klompas MD, MPH, FRCPC, FIDSA, FSHEA
Harvard Medical School, Harvard Pilgrim Health Care Institute, and
Brigham and Women’s Hospital, Boston, USA

2. Disclosures
Grant funding from the US Centers for Disease Control and Prevention Honoraria from Premier Healthcare Alliance for lectures on VAP surveillance

3. Ventilator-associated pneumonia
Affects ~5-10% of ventilated patients
Increases ICU length of stay by ~4-7 days
Increases hospital length of stay by ~14 days
Crude mortality rate 30-50%
Attributable mortality 8-12%
Adds ~$10-50,000 to cost of hospital stay
VAP is a serious problem
CMS 1533-P, 2007 Safdar et al, Crit Care Med 2005; 33:2184 Tejerina et al, J Crit Care 2006; 21:56 Muscedere et al, J Crit Care 2008;23:5-10 Eber et al, Arch Intern Med 2010;170: Nguile-Makao et al, Intensive Care Med 2010;36:781-9 Beyersmann et al, Infect Control Hosp Epidemiol 2006;27:493 3

4. VAP
However, as we start to think about VAP surveillance, we have to appreciate that VAP surveillance in this country does not occur within a vacuum in the hospitial but is very much in the public spotlight

5. States with mandatory reporting legislation for healthcare-associated infections
Mandatory reporting enacted
Study bill
There is mandatory reporting legislation in most U.S.
Association for Professionals in Infection Control and Epidemiology 2012 5

6. Prevent ventilator-associated pneumonia
2016 National Patient Safety Goal
(proposed)
Prevent ventilator-associated pneumonia
The Joint Commission has repeatedly proposed making VAP prevention a National Patient Safety Goal

7. Medicare has repeatedly noted it’s wish to make VAP a non-reimbursable complication of care
“Centers for Medicare and Medicaid Services (CMS) announced its decision to cease paying hospitals for some of the care made necessary by ‘preventable complications’” 7

8. …but VAP is a difficult diagnosis.
These initiatives all presume we can accurately identify and track who does and does not have VAP….
…but VAP is a difficult diagnosis.
These policies all presume we can accurately identify which patients do and don’t have VAP … 8

9. Diagnostic Criteria for VAP
High Temp
Low Temp
High WBC
Low WBC
Low P:F Ratio
Increased vent settings
Purulent secretions
Gram stain neutrophils
New Antibiotic
Start
Infiltrate
CDC Old Definition ✓
CDC New Definition
HELICS Criteria
ACCP Criteria
Clinical Pulmonary Infection Score
Johanson’s Criteria
First of all, there’s no consensus on how to define VAP. Here are 6 different published definitions and as you can see they all vary in their criteria.
Ego et al. Chest 2014;ePub ahead of print

10. Impact of Diagnostic Criteria on VAP Prevalence
Prospective surveillance, 1824 patients, Tertiary Med-Surg Unit, Belgium
And not surprisingly, they vary in both the number and identity of patients they identify
Ego et al. Chest 2014;ePub ahead of print

11. CDC’s old surveillance definition for VAP
Patient must fulfill each of the three categories below:
Chest
Radiograph
Any one of the following:
1. New, progressive, or persistent infiltrate
2. Consolidation
3. Cavitation
Systemic
Signs
1. Temperature >38°C
2. WBC <4,000 or >12,000 WBC/mm3
3. For adults 70 years old, altered mental status with no other recognized cause
Pulmonary
Any two of the following:
1. New onset of purulent sputum, or change in character of sputum, or increased respiratory secretions, or increased suctioning requirements
2. New onset or worsening cough, or dyspnea, or tachypnea
3. Rales or bronchial breath sounds
4. Worsening gas exchange, increased oxygen requirements, or increased ventilation demand
Let’s look more closely at CDC’s old VAP definition before they introduced VAE. The old definition required patients to fulfill radiographic, systemic, and pulmonary criteria. As you look at the list of signs on this list, however, we note that none of these is sensitive or specific (all these signs have a broad differential diagnosis). More concerningly, the signs are very subjective. What constitutes “purulent sputum, a change in the character of the sputum, increased respiratory secretions, increased suctioning requirements, worsening gas exchange, etc.? There’s nothing in the definition to define these signs. Who makes the call (MD, RN, RT, IP)? How many observations are required? How long do signs need to continue for? How many ml’s constitutes increased secretions? All is left to the discretion of the observer. And the worry then, is that if the observer knows that the VAP rates he she reports are not just going to be reviewed by clinicians but will be used to determine the hospital’s reputation, accreditation, and / or compensation, one cannot help but worry that observers will interpret these subjective signs more strictly. 11

12. Complicated Labor Intensive Subjective Non-Specific
In essence, the problem with the old definition was that it was very complicated and labor intensive yet subjective and non-specific. Lots of work to generate uncertain results.

13. Nothing brings home the challenge of surveillance using traditional VAP definitions than the x-ray. We can all see that this x-ray is abnormal but none of us can say with certainty exactly what’s going on. Is it underlying chronic lung disease, atalectasis, edema, effusion, contusion, hemorrhage, infarction, or some combination of these? 13

14. “Diffuse patchy airspace disease right greater than left with obliteration of both hemi-diaphragms. Opacities possibly slightly increased since yesterday accounting for changes in patient position and inspiration. This could represent atelectasis, pneumonia, or effusion.” 14

15. Sources of fever and infiltrates
ARDS
Diffuse alveolar damage
Thromboembolic disease
Hemorrhage
Infarction
Fibrosis
Carcinoma
Lymphoma
Contusion
Tracheobronchitis
CLABSI
UTI
Drug fever
PLUS
Pulmonary edema
Atelectasis
Contusion
Fibrosis
Meduri, Chest 1994; 106: Petersen, Scand J Infect Dis 1999; 31:

16. Accuracy of clinical diagnosis of VAP Relative to 253 autopsies
100%
Loose definition:
Infiltrate and 2 of
temp / wbc / purulence
80%
Sensitivity / Positive Predictive Value
60%
Strict definition:
Infiltrate and 3 of
temp / wbc / purulence
40%
20% 0%
Sensitivity
Positive Predictive Value
Tejerina et al., J Critical Care 2010;25:62

17. Accuracy of quantitative BAL cultures Relative to histology
100%
80%
Sensitivity / Positive Predictive Value
60%
40%
20% 0%
Sensitivity
Positive Predictive Value
Kirtland, Chest 1997;112:445 Fabregas, Thorax 1999;54:867 Chastre, Am Rev Respir Dis 1984;130:924 Torres, Am J Resp Crit Care Med 1994;149:324 Marquette, Am J Resp Crit Care Med 1995;151:1878 Papazian, Am J Resp Crit Care Med 1995;152:1982

18. Implications for surveillance

19. Interobserver agreement in VAP surveillance
50 ventilated patients with respiratory deterioration
IP 1 (11 VAPs)
IP 2 (20 VAPs) 3 7 1 7 3
Not surprisingly, thoughtful, decent people tend to disagree. This is an analysis of 50 patients with respiratory deterioration evaluated by 3 different infection preventionists. They called between 11 and 20 VAPs and only agreed on 7 patients all together. 5
IP 3 (15 VAPs)
Kappa = 0.40
Klompas, AJIC 2010:38:237

20. 6 Case Vignettes Presented to 43 Surveyors
Similarly, here’s a study in which 43 infection preventionists from across the country were given 6 different vignettes of ICU patients and asked whether each had VAP or not according to CDC criteria. You can see that some thought there was just 1 VAP, some thought 2, some thought 3, some thought 4, some thought 5, and one person thought there were 6. Agreement here was essentially random.
Crit Care Med 2014;42:497

21. Ways to lower VAP rates Without meaningfully changing patient care
Narrowly interpret subjective clinical signs
Narrowly interpret radiographs
Seek consensus between multiple surveyors
Allow clinicians to veto surveillance determinations
Increase use of quantitative BAL for diagnosis
Taking this all together, one can put on a cynical hat and exploit the data we’ve reviewed together to lower your hospital’s VAP rates without doing anything to change patient care. All 5 initiatives on this list seem like good things at first blush to increase the rigor of your surveillance program but based on the data we’ve reviewed, we can predict that they will all lead to lower VAP rates even though they have nothing to do with patient care.
Klompas, Clin Infect Dis 2010:51: Klompas, Am J Infect Control 2012;40:408-10

22. U.S. National VAP rates United States, 2004-2012
And not surprisingly when we look at U.S. national VAP rates, we find that they’ve plummeted over the last decade. The problem is that we don’t know if this is due to better care, surveillance bias, or both. And when you can’t tell if rates are lower because of better care or because of surveillance bias then arguably the surveillance definition has failed.
SICUs
MICUs
Source: CDC NNIS and NHSN

23. International VAP Rates
And there are some clear hints that bias does play a role in lower VAP rates. Here are VAP rates from a number of countries with first world medical systems. U.S. VAP rates are a full order of magnitude lower than our European peers and yet I don’t think anyone would dare argue that our care is 10x better than theirs.
Source: CDC Europe and CDC USA

24. Increasing gap between clinical and surveillance VAP rates
15% of ICU pts on
VAP Rx
on cross-sectional surveys
No. of Patients
Thomas et al. Am Surgeon 2011;77:998 Skrupky et al. Crit Care Med 2012;40:281 Koulenti et al. Crit Care Med 2009;37:2360 Vincent et al. JAMA 2009;302:2323

25. Where does this leave hospitals?
We need to publicly report VAP rates to catalyze improved quality of care and save lives!
But the definition of VAP is ambiguous, hard to implement, and open to be gamed!
Where does this leave hospitals?

26. CDC recognized this fundamental tension and therefore assembled leaders from the key stakeholder societies to do something about this.

27. Critical Care Medicine 2013;41:2467-2475
The group came up with VAE definitions.
Critical Care Medicine 2013;41:

28. An alternative approach to surveillance
Broaden the focus from pneumonia alone to the syndrome of ventilator complications in general
More accurate description of what can be reliably determined using surveillance definitions
Emphasizes the importance of preventing all complications of mechanical ventilation, not just pneumonia
Streamline the definition using quantitative criteria
Reduce ambiguity
Improve reproducibility
Enable electronic collection of all variables
this is not just a new definition for VAP but rather a brand new approach to surveillance. The group reasoned that if the core lesion is that we cannot say with any certainty who does or does not have VAP then why are we pretending to do so? Let’s rather focus surveillance on what we can be reasonably sure about which is whether or not a patient developed new onset respiratory deterioration. This carries many advantages. First of all, it’s a truer description of what we can and cannot tell using surveillance definitions. Secondly, it emphasizes the importance of preventing all complications of mechanical ventilation, not just VAP. Thirdly, it allows us to make the surveillance definition much simpler. We just need markers of respiratory deterioration. We do not need to try to work out if it’s pneumonia or not. Allows us to create an objective, reproducible, mathematical definition that’s even amenable to electronic implementation.

29. Ventilator-associated conditions (VAC)
Sustained rise in daily minimum PEEP ≥3cm or FiO2 ≥20 points after a period of stable or improving daily minimum PEEP or FiO2
Date
PEEP (min)
FiO2
(min)
Jan 1 10
100
Jan 2 5 50
Jan 3 40
Jan 4
Jan 5 8 60
Jan 6
Jan 7
Jan 8
Jan 9
VAC
The core condition for VAE surveillance is a VAC, or ventilator-associated condition. Conceptually, it’s simply a patient who was stable or improving on a vent and then suffered a period of sustained deterioration. This is measured by looking at the vent settings. We look for increases in daily minimum PEEP and FiO2. We use the daily minimum in order to capture the patient’s best value of the day.

30. VAC IVAC Possible Pneumonia Probable Pneumonia
Ventilator-Associated Condition
IVAC
Infection-related
Ventilator-Associated Complication
There are then subcriteria to try to work out which of the VACs may actually be due to infections and which may be pneumonias. If the patient had abnormal temp or wbc and a new antibiotic start continued for at least 4 days at the time of VAC then it’s an Infection-related ventilator-associated complication (IVAC). If at the time of IVAC the patient has a positive culture and/or purulent secretions (as measured by looking at the quantity of polys on a Gram stain) then it’s a Possible or Probable Pneumonia. These subcriteria (IVAC and Pneumonia) are really just to give a hospital some insight into what MIGHT be causing their VAEs. Make no mistake, there’s no reason to believe that the VAE Pneumonia definition is any more (or less) accurate than all the other imperfect definitions for pneumonia. It’s strength is simply that it’s more objective and amenable to electronic ascertainment.
Possible
Pneumonia
Probable
Pneumonia

31. Infection-related ventilator-associated complications (IVAC)
VAC with concurrent abnormal temp or WBC count
AND ≥4 days of new antibiotics
Date
PEEP (min)
FiO2
(min)
T min
T max
WBC min
WBC max
Antibiotic
Jan 1 10
100
Jan 2 5 50
Jan 3 40
99.1
99.9
8.4
10.1
Jan 4
101.9
9.9
11.2
Linezolid
Cefepime
Jan 5 8 60
98.6
102.2
12.1
15.3
Jan 6
98.8
100.3
14.1
17.4
Jan 7
96.8
15.0
16.1
Jan 8
Jan 9
IVAC

32. Ventilator-associated pneumonia
IVAC with concurrent purulent sputum (Gram stain neutrophils)
and / or positive pulmonary cultures
Date
PEEP (min)
FiO2
(min)
Gram Stain
Polys
Epis
Culture
Jan 1 10
100
Jan 2 5 50
Jan 3 40
Jan 4 3+
Klebsiella pneumoniae
Jan 5 8 60
Jan 6
Jan 7
Jan 8
Jan 9
PVAP

33. VAE Definition Changes - 2015
The third tier of the VAE algorithm will be consolidated
Possible VAP and Probable VAP will be combined: PVAP
PVAP defined as IVAC + one of the following:
Positive quantitative or semi-quantitative culture above threshold
Culture (any quantity) with purulent secretions
Positive pleural fluid culture, lung histology, diagnostic test for Legionella, or diagnostic test for a respiratory virus

34. VAE Definition Changes - 2015
Community-acquired fungal pathogens will be excluded
Cryptococcus
Histoplasma
Coccidioides
Paracoccidioides
Blastomyces
Pneumocystis

35. VAE Definition Changes - 2015
Option to collect and report a new denominator
Episodes of mechanical ventilation
Rationale
Many best practices in critical care are designed to decrease duration of mechanical ventilation
Decreasing average duration of mechanical ventilation means fewer total ventilator days and a paradoxical increase in the VAE rate per 1000 vent-days
Compare:
3 VAEs / 1000 vent days  3 VAEs per 1000 vent-days
3 VAEs / 900 vent days  3.3 VAEs per 1000 vent-days

36.
Proof that these definitions can be automated comes from CDC’s online calculator that hospitals can use to enter raw data about a patient and get an instant determination on whether the patient meets criteria for VAC and IVAC. There are also a couple of papers describing automation of hospital wide surveillance.

37. Clinical Significance & Interpretation

38. Canadian Critical Care Trials Group ABATE Study 11 ICUs, 1330 patients, VAE vs VAP Surveillance
9.9 events
per 1000 vent days
VAP
10.6 events
per 1000 vent days VS
100 39
109
What do we know then about VAEs? Well there are now almost a dozen studies of VAE epidemiology. This one is from the Canadian Critical Care Trials Group. They compared VAE vs VAP surveillance in 11 ICUs, more than 1300 patients. They found that VAE and VAP rates were very similar but they captured very different populations of patients. There was very little overlap between the two groups. It’s worth thinking for a moment about the 109 patients that were VAP positive but VAE negative. By definition, these were VAPs that did not require a sustained increase in ventilator support (if they had, they would have been VAEs). One wonders about the clinical significance of these physiologically more benign events.
Muscedere et al. Chest 2013;144:1453

39. VAE ≠ VAP
So, the key message here is that VAE and VAP are not the same thing.
Image from

40. Qualitative analysis of 153 VAEs Royal Brisbane & Women’s Hospital, Queensland, Australia
Abx + Furosemide 6%
Pneumonia
38%
Edema
26%
There are also now at least 5 studies assessing the clinical correlates of VAE. What sorts of clinical events tend to trigger VAEs? The results are consistent across all 5 studies. Almost all VAEs are triggered by 1 of 4 conditions: pneumonia, volume overload, atelectasis, or ARDS.
Atelectasis
15%
Other 8%
ARDS 6%
Hayashi et al. Clin Infect Dis 2013;56:

41. VAE = VAP + CHF + ARDS + Atelectasis +
Others
Rather, VAEs are a mix of pneumonias, volume, ARDS, and atelectasis.

42. Attributable Mortality of VAC versus VAP
USA – 3 centers
PLoS ONE 2011;6:e18062
USA – 8 centers
Crit Care Med 2012;40:3154
Canada – 11 centers
Chest 2013;144:1453
Netherlands – 2 centers
Am J Resp Crit Care Med 2014;189:947
USA – 2 centers
Crit Care Med 2014;ePub
USA – 1 center
Infect Control Hosp Epidemiol 2014;5:502
What is clear, though, is that VAEs are bad for patients. Here are findings from 6 studies estimating the attributable mortality of VAEs. The results are very consistent. VAEs appear to double the risk of dying compared to similar patients without VAEs. Four of the six studies also estimated the attributable mortality of VAP and we can see that in 3 out of 4 studies, the attributable mortality for VAE was higher.
0.5 1 2 5 10
Odds Ratio or Hazard Ratio 42

43. VAE Prevention

44. Strategies for preventing VAEs
Decrease duration of mechanical ventilation
Target the primary conditions associated with VAEs
So, this brings us to the key question of how do we prevent VAEs? There are two key strategies to do so: decrease exposure to the ventilator by speeding extubation and prevent the primary conditions associated with VAEs. In practice, these are often one and the same interventions

45. Strategies for preventing VAEs
Minimize sedation
Paired SATs and SBTs
Early mobility
Low tidal volume ventilation
Conservative fluid
management
Minimize blood
transfusions
Decrease duration of mechanical ventilation
Target the primary conditions associated with VAEs
Minimize sedation, practice paired daily SATs and SBTs, early mobility, low tidal volume ventilation, conservative fluid management, and minimize blood transfusions. Now remarkably, none of these interventions should come as a surprise to you. These are all emerging best practices in critical care. What we’re seeing here is a coming together of emerging best practices in critical care and CDC’s new surveillance metric. It is very likely that VAE will turn out to be a very efficient means to monitor the fidelity and impact of the adoption of best practices for ventilated patients.

46. Conservative fluid management
About a third of VACs are due to pulmonary edema Elevated central venous pressures associated with increased mortality rates Randomized controlled trial showing conservative fluid management associated with more ventilator-free days compared to liberal fluid management
Boyd et al., Crit Care Med 2011;39:259
ARDSnet, NEJM 2006;354:2564

47. BNP Driven Fluid Management
Randomized controlled trial of ventilator weaning
304 patients randomized to daily BNP levels versus usual care
Patients randomized to daily BNP levels
More diuretics
More negative fluid balance
Less time to extubation
50% fewer VAEs
P=.02
The second study looked at conservative fluid management. In this study, patients were randomized to daily BNP checks versus routine care. Patients with high BNPs were managed with a fluid depletive strategy – less volume infused and more diuretics. This not only lead to a significant decrease in ventilator days but remarkably was also associated with a 50% decrease in VAEs. This study affirms our earlier observation that a substantial fraction of VAEs are caused by excess volume.
Usual Care
Daily BNP
Mekontso Dessap et al. Chest 2014; ePub ahead of print

48. Canadian Critical Care Trials Group ABATE Study Enhanced care for vented patients, 11 ICUs, 1330 patients
Let me now show you results from the 3 studies that have been done thus far assessing the preventability of VAEs. The first is again from the Canadian Critical Care Trials Group. They worked with 11 ICUs over 2 years to increase adoption of what they considered to be best practices for ventilated patients. You can see that they did realize some improvements although truth be told, many of the improvements were quite modest.
Sinuff et al. Crit Care Med 2013;41:15-23

49. Canadian Critical Care Trials Group ABATE Study Enhanced care for vented patients, 11 ICUs, 1330 patients
Nonetheless, they did see a significant decrease in VAE rates
Muscedere et al. Chest 2013;144:

50. CDC Prevention Epicenters’ Wake Up and Breathe Collaborative
Prospective quality improvement collaborative Goal: prevent VAEs through less sedation and earlier liberation from mechanical ventilation Mechanism: increase performance of paired daily spontaneous awakening trials and breathing trials (SATs and SBTs) 12 ICUs affiliated with 7 hospitals
Finally, the CDC Prevention Epicenters just recently presented the results of their VAE prevention collaborative. 12 ICUs affiliated with 7 hospitals sought to decrease VAEs by decreasing duration of mechanical ventilation. They focused on enhancing the performance of paired daily SATs and SBTs.
Klompas et al., Am J Resp Crit Care Med 2014;ePub

51. CDC Prevention Epicenters’ Wake Up and Breathe Collaborative
SATs / SBTs
VAEs
63% in SATs 16% in SBTs 81% in SBTs done with sedatives off
37% in VACs 65% in IVACs
Over a 19 month period they observed a substantial increase in SATs and SBTs which in turn was mirrored by substantial decreases in VACs and IVACs.
Klompas et al., Am J Resp Crit Care Med 2014;ePub

52. Ventilator-associated events A patient safety opportunity
Broaden Awareness
VAE surveillance provides hospitals with a fuller picture of serious complications in mechanically ventilated patients
Catalyze Prevention
A significant portion of VAEs are likely preventable
Reflect and Inform Progress
VAE surveillance provides an efficient and objective yardstick to track one’s progress relative to oneself and to peers
I’d like to close by suggesting to you that CDC’s new surveillance metric has the potential to be more than just a rate measure but a real opportunity to improve patient safety and outcomes. This new surveillance metric purposefully broadens the scope of surveillance and hence our awareness to more than just pneumonia alone. There are other critical complications of mechanical ventilation associated with high morbidity and mortality. If we now know about these additional complications our patients are suffering than that should prompt us to enhance our prevention programs. And finally we have an objective metric that we can use to measure our progress without the worry that lower rates are due to surveillance bias rather than true improvements.
NEJM 2013;368:1472

53. Michael Klompas (mklompas@partners.org)
Thank You!
Michael Klompas