The Association Between Oxygenation Thresholds and ... · The Association Between Oxygenation...
Transcript of The Association Between Oxygenation Thresholds and ... · The Association Between Oxygenation...
Laveena Munshi, MD, MSc
November 1, 2016
Critical Care Canada Forum
Interdepartmental Division of Critical Care MedicineMount Sinai Hospital/University Health NetworkUniversity of Toronto
Toronto, Canada
The Association Between Oxygenation
Thresholds and Mortality During
Extracorporeal Life Support
Disclosures:
No Financial Disclosures
Extracorporeal Life Support Organization Research Grant
Is It Possible to
O.D. on O2??
Hyperoxia Not Uncommon During ECLS?
ObjectivesTo evaluate the association between oxygen thresholds and hospital mortality in patients undergoing ECLS in 3 cohorts:
1.Venovenous ECMO
for respiratory failure
2.Venoarterial ECMO
for cardiogenic shock
3. eCPR for cardiac
arrest20
30
40
50
60
70
80
90
100
0 80 160 240 320 400 480
HYPOTHESIS
paO2
Perc
ent
Mo
rtal
ity
Methods
Retrospective Cohort Study
Adult Patients Undergoing ECLS
2010-2015
Methods
COHORT 1: VV ECMO RESP Venous outflow/inflow or bicaval dual lumenRespiratory Indication (eg. ARDS)
COHORT 2: VA ECMO CARDIAC Venous outflow/arterial inflow or venous/arterial inflowCardiac Indication (eg. post MI cardiogenic shock)
COHORT 3: eCPR Veno-arterial cannulation during cardiac arrest
EXPOSURE Oxygenation determined by ABG 24 hours after ECLS initiationHypoxemia: PaO2 <60mmHg
Normoxia: PaO2 60-100mmHg (ref)
Moderate Hyperoxia:PaO2101-300 mmHg
Extreme Hyperoxia: PaO2 >301mmHg
OUTCOME In-hospital Mortality
STATISTICAL ANALYSIS Multivariable logistic regression analysis controlling for all clinically relevant confounders (Patient demographic, pre ECLS clinical data, post-ECLSparameters)
Results
Cohort Creation
VV ECMO VA ECMO VA ECMO
Configuration, Indication and ABG Datan=7,337
VV ECMO RESP VA ECMO CARDIAC VA ECMO eCPR
Cohort Characteristics
Mean (± Standard Deviation)
Median (Interquartile Range)VV RESP VA CARDIAC eCPR
Age 44 ± 16 52 ± 15 53 ± 16
Sex (% Male) 64% 67% 71%
Weight 87 ± 28 82 ± 23 81 ± 22
Duration MV pre ECMO
55 hours (18-144) 15 hours (6-40) 2 hours (0-17)
ABG pre ECMOph/pCO2/O2/HCO3
7.20/64/67/26 7.30/40/124/20 7.20/47/120/18
ECMO Mode VV 61%BCDL 38%
VA 99%VVA 1%
VA 99%VVA 1%
ECMO Flow 4 ± 1 4 ± 1 3.5 ± 1
ABG post ECMOph/pCO2/O2/HCO3
7.23/41/94/26 7.40/36/180/22 7.40/34/195/23
Duration of ECMO (days)
8 (4-15) 5 (3-8) 4 (3-8)
Discharged Alive 59% 41% 39%
0
10
20
30
40
50
60
70
80
90
100
VV ECMO RESP VA ECMOCARDIAC
eCPR
Hypoxemia <60 mmHg
Normoxia 61-100 mmHg
Mod Hyperoxia 101-300 mmHg
Severe Hyperoxia >300 mmHg
Distribution of Oxygenation
21%
52%
25%
2%8%
19%
58%
15%
23%
8%
46%
22%
Perc
ent Extreme Hyperoxia >301 mmHg
Crude Association Between O2 and
Mortality
paO2
Perc
ent
---- VV ECMO RESP---- VA ECMO CARDIO---- eCPR
VV ECMO Respiratory Failure:
Moderate Hyperoxia and Hypoxemia
Associated with Increased Mortality
OR 95% CI
Hypoxemia
paO2<60 mm Hg
(161 patients, 21%)
1.68 (1.09-2.57)
Normoxia
paO2 60-100 mm Hg
(394 patients, 52%)
1 reference
Moderate Hyperoxia
paO2 101-300 mm Hg
(194 patients, 25%)
1.66 (1.11-2.50)
Extreme Hyperoxia
paO2 >301 mm Hg
(15 patients, 2%*)
0.75 (0.21-2.64)
*Older age, liver failure, higher peak inspiratory pressure at 24 hours, higher HCO3,
lower pH at 24 hours also statistically significantly associated with increased mortality
101-120 mmHg
121-140 mmHg
141-160 mmHg
161-180 mmHg
181-200 mmHg
201-220 mmHg
221-240 mmHg
241-260 mmHg
VA ECMO Cardiogenic Shock:
No Association with Mortality
OR 95% CI
Hypoxemia
paO2<60 mm Hg
(62 patients, 8%)
1.67 (0.74-3.75)
Normoxia
paO2 60-100 mm Hg
(145 patients, 19%)
1 reference
Moderate Hyperoxia
paO2 101-300 mm Hg
(450 patients, 58%)
0.89 (0.59-1.34)
Extreme Hyperoxia
paO2 >301 mm Hg
(117 patients, 15%)
1.43 (0.80-1.00)
*Older age, lower mean arterial pressure, post ECMO MV FiO2, lower pH & pCO2,
statistically significantly associated with a higher mortality
VA ECMO eCPR:
Moderate Hyperoxia Associated with
Increased Mortality
OR 95% CI
Hypoxemia
paO2<60 mm Hg
(96 patients, 23%)
1.33 (0.48-3.69)
Normoxia
paO2 60-100 mm Hg
(34 patients, 8%)
1 reference
Moderate Hyperoxia
paO2 101-300 mm Hg
(191 patients, 46%)
1.77 (1.03-3.03)
Extreme Hyperoxia
paO2 >301 mm Hg
(91 patients, 22%)
1.92 (0.9-3.69)
*post ECMO MV FiO2 & lower pH statistically significantly associated with a higher
mortality
MECHANISM of HARM ASSOCIATED with HYPEROXIA:
• Direct Lung Toxicity
• Interstitial fibrosis, atelectasis, tracheobronchitis, ocular toxicity
• Systemic Hyperoxia Effect
• pro-inflammatory response, increased reactive O2 species
• vasoconstriction leading to end organ dysfunction
ECLS Types VV ECMO
RESPIRATORY
FAILURE
VA ECMO
CARDIOGENIC
SHOCK
VA ECMO
eCPR
Association
w/ Mortality
Moderate Hyperoxia
Hypoxemia
No Association Moderate Hyperoxia
Mechanism for Harm – VV ECMO
• Oxygen free radicals exacerbated on ECMO
• Vasoconstriction induced by hyperoxia leading to MSOF
• Mechanisms to achieve higher arterial paO2 may be harmful
• Higher ECMO flow
• Higher ventilation intensity
• Higher FiO2
No Harm Seen - VA ECMO
• Predominant mechanism of death may be due to underlying disease
• Death due to underlying disease may precede harm potentially attributable to hyperoxia
• VA ECMO deaths occurred earlier than VV ECMO
Death on VA ECMO Occurred
Earlier than VV ECMO
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
24 HOURS 48 HOURS 72 HOURS 96 HOURS 120 HOURS 230 HOURS
VA ECMO
VV ECMO
VA – 90% deaths occurred by day 12
VV – 90% 0f deaths occurred by day 30
Mechanism for Harm – eCPR
• Consistent with reports of increased harm in literature with for anoxic brain injury population
• Neurologic injury and brain edema following cardiac arrest particularly susceptible to O2 free radicals and vasocontriction
Limitations:
• Exploratory analysis
• Adequate capture of exposure? (24 hour time point ABG)
• Bias towards the null
• Clustering at hospital level
• Would impact standard errors but not point estimates
• Selection Bias
• Sufficient Control for Confounding
O2: Too Much of a Good Thing??
• Recent growing body of literature over the past 6 years suggesting harm associated with hyperoxia
• Recent growing body of literature surrounding lower O2
thresholds at which potential harm may be seen
RCT, 434 Patients, MSICU
MSICU, anticipated admission >72 hours
CONSERVATIVE CONVENTIONAL
paO2 70-100 mmHg paO2 up to 150 mmHg
Saturation 94-98% Saturation 97-100%
Exclude ARDS with PaO2/FiO2 <150mmHgChronically hypercapnic COPD
Potential Overestimation
of Treatment Effect
• Stopped Early
• Baseline imbalances
• Small number of Outcomes
• Underpowered
Conclusions and Future Directions• Moderate hyperoxia (paO2 101-300 mmHg) associated with
increased mortality in select cohorts of ECLS (VV ECMO Respiratory Failure and VA ECMO for eCPR)
• Possible mechanisms: vasoconstriction and oxygen free radicals or harm associated with increased ECLS or MV intensity to achieve higher O2
• Lack of association between hyperoxia and mortality in VA ECMO may be attributable to death due to underlying disease process
• Future research needed to further confirm results, explore thresholds, understand duration and dose-response as well as mechanisms for harm
A word of caution for the visitors to Toronto regarding your Tuesday night plans…..
Oxygen bars are places people can go to inhale
high purity Oxygen for recreation and relaxation. It is a fun and profitable business that succeeds in many different settings.
The first "Oxygen Bar" in North America was the "Oxygen Spa Bar" opened by <name omitted> in Toronto Canada in 1995.
$5.00/3 minutes
Acknowledgements
Dr. Eddy Fan
Dr. Alex Kiss
Dr. Niall Ferguson
Dr. Shaf Keshavgee
Dr. Marcelo Cypel
ELSO Research Grant
VV ECMO RESP
MORTALITY RespAcidosis<7.35, >45
RespAlkalosis>7.45, <35
Full VV RespCohort
Hypoxemia 48% (56) 39% (77) 36%
Normoxia 43% (86) 41% (34) 39%
Hyperoxia 27% (11) 42% (52) 45%
Hyperoxia 100% (3) 33% (6) 46%
Interaction (exploratory)VA ECMO CARDIAC
MORTALITY RespAcidosis<7.35, >45
RespAlkalosis>7.45, <35
Full VA Cardiac Cohort
Hypoxemia <60
92% 61% 64%
Normoxia 61-100
76% 53% 57%
Hyperoxia101-300
56% 52% 56%
Hyperoxia>301
79% 60% 67%
VV ECMO RESP VA ECMO CARDIAC eCPR
Sex
Race
Age
Weight
Year
Pre ECMO ABG data
Pre ECMO Duration of MV
Pre ECMO Vent (FiO2, PIP,
PEEP)
Pre ECMO MAP
Pre ECMO Vasopressors
Pre ECMO CVVHD
Pre ECMO Narcotics
Pre ECMO NMBA
Pre ECMO iNO
Pre ECMO Steroids
Acute Comorbid Conditions Pre
ECMO (cardiac, AKI, acute liver
failure)
Post ECMO flow
Post ECMO vent parameters
Post ECMO ABG data
Sex
Race
Age
Weight
Year
Pre ECMO ABG data
Pre ECMO Duration of MV
Pre ECMO Vent (FiO2, PIP, PEEP)
Pre ECMO MAP
Pre ECMO Vasopressors
Pre ECMO CVVHD
Pre ECMO Narcotics
Pre ECMO CPB, iNO
Pre ECMO paced
Pre ECMO Steroids
Acute Comorbid Conditions Pre
ECMO (resp failure, AKI, acute
liver failure)
Post ECMO flow
Post ECMO vent parameters
Post ECMO ABG data
Sex
Race
Age
Weight
Year
Post ECMO flow
Post ECMO vent parameters
Post ECMO ABG data
MAP
*did not include pre-ECMO ABG
and MV data as the majority are
not intubated pre-arrest
Appendix Table 1: Variables Incorporated Into Each Model (*determined by clinical relevance)
ECMO paO2 mmHg OR 95% CI
100-120 1.97 (0.84-4.65)
120-140 1.17 (0.29 – 4.66)
141-160 0.717 (0.10-5.34)
161-180 3.56 (0.26-49.38)
181-200 3.16 (0.14-71.34)
201-220 3.99 (0.10-153.99)
221-240 3.62 (0.04-361.24)
241-260 1.49 (0.01-201.86)
280-300 0.96 (0.002-415.50)
VV ECMO paO2 Detailed Overview
Moderate Hyperoxia
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
24 HOURS 48 HOURS 72 HOURS 96 HOURS 120 HOURS 230 HOURS
VA ECMO
VV ECMO
(VA) 90% of deaths occurred by day 12
(VV) 90% of deaths occurred by day 30
Timing of Death Across those who Died on ECMO
0
10
20
30
40
50
60
70
80
90
100
0 2 4 6 8 10 12
VV ECMO
VA ECMO
eCPR
Days
0
10
20
30
40
50
60
70
80
90
100
Hypoxia Normoxia ModerateHyperoxia
SevereHyperoxia
VV ECMO RESP
VA ECMO CARDIAC
eCPR
Distribution of Oxygenation
<60mmHg 61-100mmHg 101-300mmHg >300mmHg
Perc
ent