1. Ram E. Rajagopalan, MBBS
AB (Int Med) AB (Crit Care)
Department of Critical Care Medicine
SUNDARAM MEDICAL FOUNDATION
Chennai, India
Debunking (some) ICU Myths
2. So… how do Myths develop?
Reification of an opinion
The best way to convert fiction
into fact is to “publish” it
-Anon.
The finality of the printed word
Text-Books
Peer-reviewed journals
and…now…
The darned “ ’Net”
3. “Nonsensus Consensus”*
*Lancet 1990; 335: 1446-7.
*P. Skrabanek
Consensus enforces the tyranny of the majority
Never a part of the scientific method
Ideal for resolving political discordance
If “popularity” was a
necessary component
of scientific acceptance,
we would still be living
on a flat earth in the
centre of the Universe!!
4. Myth: Restricting ICU visitation
is beneficial to everyone
Unrestricted visitation of patients is
considered to be undesirable:
Interferes with patient care
Provokes anxiety
(patient, family & caregiver)
Promotes infection
Disrespects “privacy”
Minerva Anestesiol 2007; 73: 299-306
5. RCT of ICU Visit Restriction
Circulation 2006; 113: 946-52
RCT of unrestricted (UVP) vs. restricted (RVP)
Single 6-bed Coronary Care Unit
Random assignment of UVP/RVP for
2-month periods
n= 226
30% J in numbers of visitors
2.5X J in duration of visits
7. Caution!
Circulation 2006; 113: 946-52
J in bacterial
colonization;
without changes in
infection
Could severity of
illness / invasiveness
affect infection
acquisition?
8. Myth: Shoe Covers reduce
Infection in the ICU?
Another ritual in most Indian ICUs
Logical?
We walk through dirty streets so,
like we do in the OT, shouldn’t we
change footwear or at least cover
them?
9. Shoe Covers in the Indian ICU
MJAFI 2007; 63 : 334-6
ICU
environmental
contamination
studied in two
2-week phases
with & without
covers
No K in colonization; but basal extent of
colonization was very high
No clinical data
10. Paradoxical Pakistani Data
1151 patients in
medical & surgical
ICUs
Observed in two 3-
month periods with &
without covers
“Infection”
ambiguously defined
Higher “infection” during shoe-cover phase
explained away as hand contamination while
donning the shoe covers
Pak J Med Sci 2014; 30: 272-5
11. Wash Hands or Cover Shoes?
We have no idea of user compliance in
these studies
Covers may not reduce colonization of
highly contaminated environments?
Do we ever provide hand disinfectants
at sites where shoe-covers are
dispensed?: Pay attention to the
basics of hand hygiene
12. Myth: Protective value of N95 Mask
Limited availability of N 95
Costs are high
Fit testing is essential
13. Can We Use Surgical Masks?
RCT of surgical vs. fit tested N95 masks
8 Canadian Hospitals; 2008-9 ’Flu season
Nurses in medical / ED/ Paediatric units
Block randomization 4-nurses
n=212 n=210
To wear assigned mask taking care of all febrile
respiratory illness
JAMA 2009; 302: 1865-71
16. Dealing with conflicting priorities
Get things right!
Consider the N95
if high infection risk
Bronchoscopy
Intubation
Open suctioning
MDR TB
17. Myth: Physical Restraints prevent
unplanned extubation
Soft wrist restraints are commonly used
in the ICU to prevent ‘self-harm’
Typical frequency 30-50% of intubated
patients have restraints
Unclear if restraint will minimize
self-extubation
18. Study on Unplanned Extubation
AJCC 2008; 17: 408-16
Case-control study from a base
cohort of 126 self-extubations
in 1455 ventilated patients (8.7%)
100 patients matched with 200
controls who did not self-extubate
(Age/ sex/ diagnosis/ date of hospital stay)
Assessed for factors associated with
unplanned extubation
19. Multivariate Predictors
Use of restraints
Higher GCS &
Nosocomial Infection
Independent
predictors of
unplanned extubation
AJCC 2008; 17: 408-16
20. Physical Restraints Do Not Help
Restraints increase the RR
of self extubation ~3x
May not be the provocation,
but may be a marker of
unrecognized delirium*
Wrist restraints ineffective
especially with HOB J 30o
* Crit Care Med 2005; 33: 1260-5
21. Myth: The CVP is an excellent
guide of adequate volume
resuscitation
Even today:
Many “guidelines” consider it an
acceptable option
Every candidate in our
Echo workshop wanted
CVP as a guide
(or the IVC size on USG)
22. Central Venous Pressure
We assume
RV preload = RV EDV = RV EDP = RAP = CVP
Responsiveness of output to ventricular stretch
Volume as marker of stretch
Linear P-V relationship
No AV valve pathology
24. Preload D is not Preload
Responsiveness
Preload
Stroke
Volume
(Crit Care 2000, 4: 282-9)
A
B
25. AJRCCM 2000; 162: 134-8.
Failure of Static Measures
PAoP
RAP
Sensitivity
(True
+)
1-Specificity (False +)
RAP = R Atrial Pressure
PAoP = PA occ Pr
0
1
1
Neither RAP/ CVP
nor PAoP
discriminates
patient response to
fluid loading
26. Myth 1: Lectures at conferences are
meant to be educational & informative
How can they be?
12-15 minutes to deliver a
lecture is certainly not a
generous amount of time!
Q.E.D.
27. Myth: The cuff leak test perfectly
identifies post-extubation obstruction
Upper airway obstruction after extubation
occurs in ~15% of adult patients
Significant stridor may require re-
intubation (~5%)
Knowledge of airway patency prior to
extubation will help
May be indicated by air-leak around a
deflated ET cuff
28. Cuff Leak Test
No standardized Method
Administer a Vt of 10 cc/kg
With deflated cuff:
“Normal” Leak > 110 – 140 ml
Or
>10-15%
E.g.: Vt = 600 ml
Exhalation Vt = 450 ml (D= 150 ml)
10-15 % Loss = 60-90 ml
Leak influenced by
Crs (Compliance)
& Vt
29. Best Data
Intens Care Med 2009; 35:1171-9
Meta analysis
11 studies
(n=2303)
9 studies Up Aw Obs
3 studies re-intubation
Significant heterogeneity &
publication bias
30. Making Sense of it all
Assumes that in
Indian women
Stridor ~30%*
Re-intubation ~10%*
(* personal data)
31. The Cuff Leak as I see it
Not perfect outcome data
Cumbersome to perform;
potentially decompensating
Standardization of
methodology needed
Multi-dose steroids & NIV reduce need for
re-intubation significantly
32. Myth: Early empirical anti-
biotic Rx saves lives
DOI: 10.1097/CCM.
0b013e31827e83af
1. Administration of effective intravenous
antimicrobials within the first hour of
recognition of septic shock (grade 1B) and
severe sepsis without septic shock (grade 1C)
as the goal of therapy.
“ ”
33. The Importance of Timing
CCM 2006; 34: 1589-96
Each hour delay of appropriate abx.:
~12% K survival
Anand Kumar, et al
34. Is Mortality related to Delay?
After adjustment to SSS*, department &
Geographic location the association
becomes miraculously immaculate!!
CCM 2014; 42: 1749-55
36. Current Empirical Strategies
The Big Picture
Suspected NI
Rx
NI No NI
MDRO
Colonization
Adverse outcome
Cx Appropriate
De-escalated
Good outcome
Cx Inappropriate
Adverse outcome
? Diagnostic certainty
?Rates of resistance
? Effect on outcome
37. Diagnostic Certainty?
Diagnosis of
most infection is
‘retrospective’
Culture results
take days
In a prospective ICU cohort; 8 US/EU ICUs
Empirical Abx started on suspicion of NI
Infection retrospectively adjudicated by a panel
Suspected NI
195
No NI Confirm NI
156 (80%) 39 (20%)
117 (¾) 26 ( 2/3)
Empirical Rx Empirical Rx
Int Care Med 2007; 33: 1369-78 WBC, APACHE & “Hospital” independent predictors of empirical Rx
38. Remember the Asian
Resistance Patterns?
Acinetobacter: 34%
Ps. Aeruginosa: 24%
K. Pneumoniae: 15%
Staph Aureus: 14%
ANSRP
65% Carbepenem
Resistant;
~85% resistance
in India, Malaysia &
Thailand
30%
Carbepenem
Resistant
AJRCCM 2011; 184: 1409-17
39. The Risk of Continued
Empirical Antibiotic
¾ of patients
without NI
adjudicated get
empirical Rx
started
Int Care Med 2007; 33: 1369-78
59% of patients without NI still on Rx day 5
28-day mortality: Rx <4d = 7.7% >4d = 32%
Continuing Abx >4 days was an independent
predictor of death (AOR: 3.75 (0.91-15.49) p-0.07)
40. Current Empirical Strategies
Quantifying the Big Picture
Suspected NI
Rx
Cx Appropriate
De-escalated
Good outcome
NI No NI
20% 80%
MDRO Colonization
Adverse outcome
No de-escalation
Cx Inappropriate
4x J odds of
death with
>4 day Rx
41. VAP: A Decision Analysis
A decision analysis based on available
data suggests:
Empirical Rx (using clinical diagnosis) :
66% mortality in the untreated patients
68% mortality in treated patients
Chest 1996; 110:1025-34
Reasons for unexpected death with Rx:
Diagnostic certainty is low (0.23)
Colonization c resistant strains
Limitations: Old data; incomplete info
43. Guidelines don’t care
about “Evidence”….
ACC/ AHA
53 Guidelines
1984-2008
7196 recommendations
…one recommendation
every 1.2 days!!!
11%
41%
48%
18%
82% of “Class I”
recommendations
were based on low-
grade evidence
JAMA 2009; 301: 831-41
44. … to Survive Sepsis….
Strong: 1 Weak: 2
RECOMMENDATIONS
Evidence strength
A: High
B: Moderate
C: Low
D: Very low
Crit Care Med 2008; 36: 296-327 Only 8/73 (11%) recommendations were Grade A
45. “Before-After” Studies
Crit Care Med 2006; 34: 943-9
Influence of a “Standard Operating
Procedure” on sepsis outcome
SOP:
EGDT
Glucose <150
Steroid use
Drotrecogin a
46. “Before-After” Studies
Crit Care Med 2006; 34: 943-9
SOP:
EGDT
Glucose <150
Steroid use
Drotrecogin a
53%
27%
Intuitively &
emotionally
appealing, but….
2008 2013
Wow!
47. ..the Science is Appalling!
EGDT: The value of EGDT is
being questioned
Glucose <150: Is only a compromise
position in glycaemic
control; no demonstrable
benefit
Steroid use: Challenged by the
CORTICUS data
Drotrecogin a: Withdrawn for inefficacy
Can therapies
of questionable
efficacy, reduce
mortality by ½ ?
The biological implausibility of outcome
raises concerns re. “before-after” reports
48. Cohorts, Confounders &
the Hawthorne Effect
Problem of historical controls:
Time-dependent improvement in care
Being in the guidelines cohort may J better
delivery of other therapies as well
Hawthorne effect:
Observation affecting the observed;
Patient care improves in the “Guidelines
phase” because caregivers are being watched
49. The Need for RCTs
“The effects of implementing
(a) guideline using a robust,
multifaceted practice change
strategy were assessed in a
cluster RCT.”
27 ICUs in ANZ; 20-week evaluation
1118 patients recruited
11 Evidence-based recommendations
JAMA 2008; 300: 2731-41
50. Successful Implementation..
Complex process:
Identifying local
“opinion leaders”
Initial educational
outreach
Academic detailing
Active reminders
Audit & Feedback
Passive reminders
In-servicing
JAMA 2008; 300: 2731-41
Earlier feeding
0.75 vs. 1.37 days
Calorie goals met
in 6.1 of 10 fed days
vs. 5 of 10
51. …Unsuccessful Result
Only 2 / 11 of the EBRs were supported
by high levels of evidence
Hospital mortality:
28.9 vs. 27.4%
Hospital LOS:
24.2 vs. 24.3 days
ICU LOS:
9.1 vs. 9.9 days
JAMA 2008; 300: 2731-41
Under-recruitment,
but power remained
adequate to detect
8% D mortality
52. Can’t Hurt! Can they?
they
Lancet Inf Dis. 2011;11: 181-9
Prospective, observational study of Rx of VAP
Compliance with ATS/ IDSA guidelines assessed
(IMPACT-HAP Investigators)
303 patients
129 compliant 174 non-compliant
28-day survival:65% vs. 79%
(p 0.0042; Kaplan-Meier)
D persists after adjusting for severity of illness
No D in median LOS / median ventilator days