Intraoperative Fraction of Inspired Oxygen and
Postoperative Outcomes
(MPOG PCRC 60)
David McIlroy, Frederic T. (Josh) Billings IV, (Vanderbilt)
Jennifer Morse, Matthew Shotwell (Vanderbilt)
Joanna Olsen (Oregon Health Science University)
Sachin Kheterpal, (U. of Michigan)
Shelley Vaughn, Michelle Romanowski, (U. of Michigan)
• Disclosures
No disclosures to report.
• Grant support
IARS IMPACT award
R01GM112871
R01AG048915
R01HL135144
P
e
r
i
o
p
e
r
a
t
i
v
e
M
e
d
i
c
i
n
e
P
a
i
n
M
a
n
a
g
e
m
e
n
t
V
a
n
d
e
r
b
i
l
t
A
n
e
s
t
h
e
s
i
o
l
o
g
y
E
s
t
a
b
l
i
s
h
e
d
1
9
4
6
Disclosures and Acknowledgements
P
e
r
i
o
p
e
r
a
t
i
v
e
M
e
d
i
c
i
n
e
P
a
i
n
M
a
n
a
g
e
m
e
n
t
V
a
n
d
e
r
b
i
l
t
A
n
e
s
t
h
e
s
i
o
l
o
g
y
E
s
t
a
b
l
i
s
h
e
d
1
9
4
6
Outline
• Background and rationale for study
• Study design
• Execution (and challenges)
• Preliminary results
• Future plans
Background and rationale: Oxygen
Increased tissue pO
2
Bactericidal killing
Protection from Hypoxia during Surgery?
O
2
-
H
2
O
2
OH
ONOO
-
HIF1a
ï‚Ÿ
ï‚Ÿ
Oxidative
Damage
Altered gene
expression
Oxygen Toxicity during Surgery?
Broad objective
To generate evidence that informs the debate on best practice
for intraoperative oxygen administration
Multicenter Perioperative Outcomes Group (MPOG)
To improve patient
care, using electronic
healthcare data.
Multicenter Perioperative Outcomes Group (MPOG)
Specific Aims
1. Explore the association between intraoperative O
2
delivery
and adverse perioperative outcomes
2. Generate preliminary data to support funding for a clinical
trial testing the impact of strategies for varied intraoperative
F
i
O
2
3. Vet MPOG centers for quality/completeness of current data
delivery
Primary hypothesis
Excess (modifiable) intraoperative oxygenation is associated with
postoperative organ injury
Methods
• Multicenter observational cohort study of patients at
participating MPOG centers
• Inclusion criteria
– Adults, duration of surgery >120 minutes, GA with ETT
• Exclusion criteria
– Outpatient surgery or repeat surg within 90d
– Pregnant, jet ventilation, airway surgery, one-lung ventilation
– Intraop desaturation (SpO
2
<90% for ≥3 min)
– Infrequent documentation of FiO2 or SpO2 (≥ 5min of missing data)
Exposure variable (Oxygen)
Oxygen exposure quantified as the AUC of FiO
2
above 0.21 (air)
calculated for minutes during surgery when the SpO
2
is >92%.
The FiO
2
for minutes when the oxygen saturation ≤92% is NOT included
in the AUC calculation as this may be considered necessary oxygen
delivery rather than excess (and therefore modifiable) oxygen delivery
Outcomes variables
• Co-primary outcomes
1. AKI defined by creatinine-based KDIGO criteria
2. Myocardial injury in patients undergoing non-cardiac surgery (MINS)
defined using concentrations of troponin I or troponin T above ULN
within 72 hours of surgery
Outcomes variables
• Secondary outcomes
30-d mortality
Acute lung injury (AHRQ criteria)
Surgical site infection (NSQIP defined)
Peak postoperative lactate within 24 hours of surgery
In-hospital MI (ICD codes)
In-hospital cardiac arrest, stroke
Missing data
Oxygen exposure: Imputed when FiO
2
gap <5 mins
Case excluded if longer gap in FiO
2
Outcomes:
Only centers providing a NON-ZERO event count for any given
outcome will be included in the analysis of that outcome.
Once a center was included, patients without measurement
data (eg troponin or creatinine) considered NON-EVENTS
Statistical Analysis Plan
For co-primary endpoints:
Multivariable mixed-effects logistic regression
Including modeling of heterogeneity among centers
Adjusted for pre-specified covariates (potential confounders)
age, BMI, ASA status, Elixhauser comorbidity index, emergency surgery,
preop Hb, S
Cr
, N
2
O exposure, PEEP, intraoperative fluid/blood
administration and intraoperative hypotension
Statistical Analysis Plan – sensitivity analyses
Recalculate O
2
exposure as:
AUC of FIO
2
above 0.21 (air) during surgery – excluding those
minutes where SpO2 <96%
While variations in FiO2 during these periods may reflect hyperoxia
within the alveoli (PAO2) it is unlikely that hyperoxemia of the blood
reaching other organs was simultaneously present.
?Excess P
A
O
2
vs P
a
O
2
Statistical Analysis Plan – sensitivity analyses
Restricting the cohort for analysis:
Complete case analysis (excl patients with no outcome)
Excluding centers reporting low frequency of outcome
measurement (eg troponin data)
Subgroup Analyses:
Age, high-risk, sex, preoperative Hb, duration of surgery
Limitations
• Unmeasured/unrecognized confounding by indication
– The same factors that placed patients at increased risk for
adverse outcomes may have led clinicians to administer a
higher FiO2
• Ascertainment bias
– The same factors that led to a higher F
i
O
2
administration
may also lead to greater surveillance for outcomes
Challenges
• Identifying study cohort within existing MPOG variable
structure
• Inconsistent methods by centers of providing data to MPOG (eg
OI vs F
i
O
2
)
• Understanding and handling missing data
• Identifying and handling data that is in inconsistent with
dictionary definition (eg in-hosp mortality)
• Using outcomes such as troponin (varied ULN, changing assays)
Challenges
16 months … and ongoing
1000’s of emails
countless hours
Results
Study Flow Chart
Study Flow Chart
O
2
administration – Median F
i
O
2
per case
O
2
administration – Median F
i
O
2
per case –
by center
Oxygen exposure – calculated AUC
AUC
FiO2
(when SpO
2
%>92.minutes)
Cases (n)
Outcomes - AKI
AKI status
N (%)
No AKI
172,393 (48.7%)
Stage 1 AKI
16,345 (4.6%)
Stage 2 AKI
2344 (0.7%)
Stage 3 AKI
806 (0.2%)
Insufficient data
156,003 (44.1%)
Preoperative renal failure
6119 (1.7%)
Outcomes – AKI
AUC
FiO2
(when SpO
2
%>92.minutes)
log odds of AKI
Outcomes – AKI: subgroup analysis by duration of
surgery
Future plans
• Complete data cleaning and analysis
• Manuscript preparation (Anesthesiology)
• Pragmatic clinical trial informed by current data
– Characterize current variation usual practice
– Primary endpoint and plausible effect size
– Logistics and feasibility
Supplemental Oxygen Therapy during Surgery
• Reduce tissue hypoxia?
• Increase reperfusion injury?
• Increase the production of ROS?
• Increase the production of ATP?
• Increase atelectasis?
• Suppress HIF signaling
• Decrease surgical site infection?
2016
Increased tissue pO
2
Bactericidal killing
Protection from Hypoxia during Surgery?
Risk of Oxygen during Cardiac Surgery (ROCS) Trial
AKI
Delirium
Myocardial Injury
Safety Markers
of Ischemia
NCT02361944
R01GM112871
Lopez et al, Trials, 2017.
