Content uploaded by David V Pilcher
Author content
All content in this area was uploaded by David V Pilcher on Dec 19, 2013
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
ORIGINAL ARTICLES
Critical Care and Resuscitation Volume 14 Number 1 March 201214
Crit Care Resusc ISSN: 1441-2772 1 March
2012 14 1 14-19
©Crit Care Resusc 2012
www.jficm.anzca.edu.au/aaccm/journal/publi-
cations.htm
Original articles
Stroke is the second most common cause of mortality and a
major cause of disability.1 Apart from thrombolysis in highly
selected patients,2 the only general interventions that influence
outcome are aspirin3 and general supportive measures pro-
vided by dedicated stroke units.4 Thus, it is desirable to identify
therapeutic interventions that are effective and safe, and that
can be routinely administered to all patients with stroke.
Strokes can be divided into ischaemic and haemorrhagic
subtypes, with the former accounting for more than 80% of
all strokes.5 During the development of an ischaemic stroke,
nerve cells lose the ability to maintain ionic homoeostasis as
free radicals accumulate and degrade cell membranes.6 These
changes eventually lead to nerve cell death. This occurs
rapidly in some patients and more gradually in others, over a
matter of hours or days.7 This phenomenon of gradual
progression is due to the existence of areas of marginally
viable brain (the “ischaemic penumbra”) and has led to
interest in various forms of oxygen therapy that may protect
reversibly injured brain cells.8 These potential benefits are
weighed against a risk of free radical-mediated damage.9
The effect of oxygen administration soon after intensive
care unit admission on outcome in critically ill patients ventil-
ated after ischaemic stroke has not been previously reported.
The appropriate oxygen therapy target for these patients is
not clear. To address this important issue, we examined the
relationship between PaO2 in the first 24 hours in ICU and
outcome in ventilated stroke patients in the Australian and
New Zealand Intensive Care Society Adult Patient Database
(ANZICS APD). We hypothesised that early hypoxia would be
associated with increased mortality, and that early hyperoxia
may be associated with either benefit or harm.
Methods
Data were extracted from the ANZICS APD. This database is
an established binational voluntary database, which con-
tains data from more than one million ICU admissions.10
Ventilated adult patients (> 17 years of age) who were
admitted to the ICU with a stroke at one of 129 particip-
ating centres between 1 January 2000 and 31 December
2009 were included. The primary Acute Physiology and
Chronic Health Evaluation (APACHE) III diagnostic code 403
(stroke) was used to identify suitable patients. An altern-
ative code (402) exists for intracerebral haemorrhage, so it
is likely that our dataset exclusively comprises patients with
ischaemic strokes. Readmissions and patients whose
records did not contain arterial blood gas analysis, APACHE
III risk of death, or vital status at discharge were excluded.
Access to the data was granted by the ANZICS Centre for
Outcome and Resource Evaluation (CORE) Management
Committee in accordance with standing protocols. Data are
collected under the quality assurance legislation of Part VC
ABSTRACT
Background: There are conflicting data that suggest that
hyperoxia may be associated with either worse or better
outcomes in patients suffering a stroke.
Objectives: To investigate the association between PaO2 in
the first 24 hours in the intensive care unit and mortality
among ventilated patients with acute ischaemic stroke.
Design: Retrospective cohort study.
Setting: Data were extracted from the Australian and New
Zealand Intensive Care Society Adult Patient Database.
Participants: Adults ventilated for ischaemic stroke in 129
ICUs in Australia and New Zealand, 2000–2009.
Main outcome measures: The primary outcome was the
odds ratio for inhospital mortality associated with “worst”
PaO2 considered as a categorical variable, with data divided
into deciles and compared with the mortality of the 10th
decile. For patients on an FiO2 of ⭓50% at any time in the first
24 hours, “worst” PaO2 was defined as the PaO2 associated
with the highest alveolar–arterial (A–a) gradient. For patients
on an FiO2 of < 50%, it was defined as the lowest PaO2.
Secondary outcomes were ICU and hospital length of stay and
the proportion of patients in each decile discharged home.
Results: Of the 2643 patients eligible for study inclusion, 1507
(57%) died in hospital. The median “worst” PaO2 was
117 mmHg (interquartile range, 87–196 mmHg). There was no
association between worst PaO2 and mortality, length of stay
or likelihood of discharge home.
Conclusions: We found no association between worst arterial
oxygen tension in the first 24 hours in ICU and outcome in
Crit Care Resusc 2012; 14: 14–19
ventilated patients with ischaemic stroke.
The association between early arterial oxygenation and
mortality in ventilated patients with acute ischaemic stroke
Paul Young, Richard Beasley, Michael Bailey, Rinaldo Bellomo, Glenn M Eastwood, Alistair Nichol, David V Pilcher,
Nor’azim M Yunos, Moritoki Egi, Graeme K Hart, Michael C Reade and D James Cooper
on behalf of the Study of Oxygen in Critical Care (SOCC) Group
ORIGINAL ARTICLES
Critical Care and Resuscitation Volume 14 Number 1 March 2012 15
of the Health Insurance Act 1973
(Cwlth). In New Zealand, use of
anonymous collected quality data
for research is classified as low-risk
audit activity and is exempt from
requirements for formal ethics
approval.
Data for oxygen values
All arterial blood gas measurements
taken during the first 24 hours of
ICU admission are collected and
entered into a standard data collec-
tion system. In accordance with the
APACHE III scoring system, the
most abnormal set of arterial blood
gas measurements by analysis of
simultaneous recordings of FiO2
and PaO2 are entered in the data-
base. If the FiO2 is ⭓0.5, the PaO2
associated with the highest alveo-
lar–arterial (A–a) gradient is
selected, and if the FiO2 is < 0.5,
the measurement with the lowest
PaO2 is selected. If arterial blood gases are taken on both an
FiO2 < 0.5 and an FiO2 ⭓0.5 during the first 24 hours, the
PaO2 derived from measurements taken on ⭓0.5 is used. In
our study, this PaO2 value was defined as the “worst” PaO2.
To explore the relationship between the worst PaO2
recorded in the adult patient database and the peak, median
and mean PaO2 measured during the first 24 hours and over
the duration of the ICU stay in patients with stroke, we
examined details of all recorded arterial blood gas measure-
ments (906 measurements) for a convenience sample of 49
stroke patients admitted to five tertiary ICUs in Australia with
a diagnosis of ischaemic stroke. The measurements were
collected between 2000 and 2009, and, of these, 311 were
collected from the first 24 hours of the ICU stay.
Data extraction
Data of the size, type and location of the hospital were
collected. At a patient level, the following variables were
extracted: demographics, APACHE III chronic comorbidi-
ties, hospital and ICU admission source, intubation, treat-
ment limitation, year of admission, physiological and
arterial blood gas parameters over the first 24 hours in the
ICU, vital status at hospital discharge (alive or dead),
discharge destination, and an APACHE III risk of death
score.11 To apply a marker for severity of illness that was
independent of arterial oxygenation, an adjusted APACHE
risk of death (AP3-no-ox) was calculated for each patient,
whereby the oxygen component of the APACHE III scoring
system was removed and an
adjusted score independent of
oxygen was recalculated.
Outcomes
The primary outcome was the odds
ratio for the risk of inhospital mor-
tality associated with the worst
PaO2 in the first 24 hours in ICU
considered as a categorical variable
with the data divided into deciles,
and compared with the mortality of
the 10th decile. We compared
between deciles the proportion of
patients who were discharged
home, the ICU length of stay and
the hospital length of stay as sec-
ondary outcome variables.
Subgroup analyses
We compared patients who were
admitted to the ICU from the emer-
gency department with those who
were admitted to the ICU from the
ward. We also compared patients who lived at home before
admission with patients who were in hospitals or residential
care facilities.
Statistical analyses
All analyses were performed using SAS, version 9.2 (SAS
Institute Inc, Cary, NC, USA). Continuous data are pre-
sented as mean (SD) or median (interquartile range [IQR]),
depending on the underlying distribution of the data.
Categorical data are reported as number (%).
To ensure that the nature of the relationship between
PaO2 and mortality was not masked by confounding varia-
bles, multivariate analysis was conducted using logistic
regression for mortality adjusting PaO2 levels for FiO2 levels,
illness severity (AP3-no-ox) and year of admission. All first-
order interactions were tested for statistical significance,
with none being significant. A two-sided P of 0.05 was
considered statistically significant. Data are reported in
accordance with the Strengthening the Reporting of Obser-
vational Studies in Epidemiology (STROBE) guidelines.12
Results
Overall, 3148 patients met the inclusion criteria of mechan-
ical ventilation, age older than 17 years and admission to ICU
with an ischaemic stroke. There were 505 patients who did
not have available information for hospital mortality (163),
arterial blood gas measurements (50), APACHE III risk of
Table 1. Baseline characteristics
Characteristic
Mean age (SD) 65.6 (14.6)
Male, no. (%) 1584 (60%)
Mean APACHE III score (SD) 75.8 (27.9)
Treatment limitation or palliative, no. (%) 92 (3%)
Chronic conditions, no. (%)
Cardiovascular disease 306 (12%)
Liver disease 18 (1%)
Renal disease 54 (2%)
Respiratory disease 89 (3%)
ICU admission source, no. (%)
Emergency 1420 (54%)
Theatre 46 (2%)
Other hospital 586 (22%)
Ward 586 (22%)
Vital signs, mean (SD)
Glasgow Coma Scale score 7.07 (4.1)
Heart rate 88.8 (38.0)
MAP 99.5 (32.7)
APACHE= Acute Physiology and Chronic Health Evaluation.
ICU = intensive care unit. MAP= mean arterial pressure.
ORIGINAL ARTICLES
Critical Care and Resuscitation Volume 14 Number 1 March 201216
death (277) or were readmissions (15). The remaining 2643
patients were drawn from ICUs of 129 contributing hospitals
(33 rural, 33 metropolitan, 34 tertiary referral centre and 29
private hospitals). Most hospitals (75) were small to medium
(< 300 beds), 33 hospitals were large (300–500 beds) and 21
hospitals were extra-large (>500 beds). The median number
of acute ischaemic stroke patients per hospital over the study
period was 8 (IQR, 3–21).
The mean patient age was 66 years (SD, 15 years) and
1584 (60%) were men. A total of 1674 were living at home
before admission (63%). The ICU admission source was the
emergency department for 1420 patients (54%), the ward
for 586 patients (22%), other hospitals 586 (22%) and the
operating theatre for 46 (2%). Admission source data were
missing for five patients (0.2%). Eighteen per cent of patients
(476) had documented pre-existing APACHE III chronic
comorbidities. The median APACHE III risk of death was 45%
(IQR, 21%–69%) and 1507 patients died in hospital (57%).
Baseline characteristics are shown in Table 1.
There was no apparent relationship between mortality
and PaO2 levels in the first 24 hours in ICU, with mortality
levels across the 10 deciles of PaO2 ranging between 50%
(5th decile, PaO2 range 103–117 mmHg) and 63% (2nd
decile, PaO2 range 69–83 mmHg). (Figure 1). After adjust-
ment for FiO2 levels (odds ratio [OR], 1.44 [95% CI, 0.97–
2.14]) AP3-no-ox (OR, 1.03 [95% CI, 1.03–1.04]) and year
of admission (OR, 1.02 [95% CI, 1.00–1.04]), there was no
relationship between PaO2 and mortality (Figure 2), as no
decile was significantly different from the reference cate-
gory (10th decile, PaO2 range 341–611 mmHg). There was
also no apparent relationship between PaO2 and length of
ICU stay, length of hospital stay or likelihood of being
discharged home. Outcome data for each of the 10 deciles
of worst PaO2 are shown in Table 2.
After adjustment for confounding variables, there were
no differences in inhospital mortality between the PaO2
deciles for any of the predefined subgroups (Table 3).
Figure 1. Hospital mortality by PaO2 decile
PaO
2
decile
Mortality (%)
0
-
69
>
69
-
83
>
83
-
93
>
93
-
103
>
103
-
117
>
117
-
140
>
140
-
174
>
174
-
226
>
226
-
341
>341
-
611
50
10
0
30
20
40
60
70
Table 2. Outcomes associated with deciles of “worst” PaO2
Worst PaO2,
mmHg
Median ICU LOS
(IQR), hours
Median hospital
LOS (IQR), hours
Inhospital
mortality, no. (%)
Adjusted* OR for
inhospital mortality
(95% CI)
OR for failure to
discharge to home
(95% CI)
Adjusted* OR for
failure to discharge
to home (95% CI)
0–69 60.5 (31.5–111. 1) 162.4 (65.0–476. 8) 163/ 264 (62%) 1.14 (0.76–1.72) 0.84 (0.55–1.30) 0.93 (0.57–1.51)
> 69–83 65.1 (318.0–121.7) 162.3 (67.0–428. 9) 165/ 264 (63 %) 1.15 (0.76–1.74) 0.91 (0.59–1.40) 0.98 (0.60–1.6)
> 83–93 70.8 (39.7–142. 7) 178.0 (75.3–563. 1) 145/ 265 (55%) 0.99 (0.65–1.51 ) 0.66 (0.43–1.00) 0.82 (0.51–1.33)
> 93–103 63.8 (32.8–118.5) 180.0 (73.8–435.3) 159/264 (60%) 1.48 (0.97–2.26) 0.84 (0.55–1.30) 1. 20 (0.73–1.97)
> 103–117 75.9 (41.9–120.9) 233.6 (93.2–594.3) 133/264 (50%) 0.94 (0.62–1.43) 0.71 (0.47–1.08) 0.97 (0.60–1.58)
> 117–140 57.3 (32.1–116.0) 210.8 (66.3–518.5) 141/265 (53%) 1.01 (0.67–1.53) 0.78 (0.51–1.19) 1.05 (0.64–1.71)
> 140–174 58.1 (32.0–115.2) 187.5 (68.0–394.6) 156/264 (59%) 1.20 (0.80–1.81) 0.83 (0.54–1.27) 1.03 (0.64–1.66)
> 174–226 55.5 (32.0–124.4) 190.8 (64.8–569.3) 141/265 (53%) 0.82 (0.55–1.22) 0.78 (0.51–1.19) 0.89 (0.56–1.42)
> 226–341 63.8 (29.9–138.4) 188.8 (72.3–515.0) 148/264 (56%) 1.01 (0.69–1.47) 0.91 (0.59–1.4) 1.08 (0.68–1.71)
> 341–611 57.2 (27.0–98.0) 165.6 (63.3–491. 5) 163/264 (62 %) 1.00 1.00 1.00
AP3-no-ox= adjusted Acute and Chronic Health Evaluation (APACHE) III risk of death, whereby the oxygen component of the APACHE III scoring system was
removed and an adjusted score independent of oxygen was recalculated. ICU =intensive care unit. IQR =interquartile range. LOS = length of stay.
OR =od ds ratio. * Odds ratio is adjusted for FiO2 levels, AP3-no-ox and year of admission. All odds ratios are relative to th e 10th worst PaO2 decile.
Figure 2. Odds ratios for PaO2 deciles adjusted for
FiO2 level, AP3-no-ox and year of admission
AP3-no-ox =adjusted Acute and Chronic Health Evaluation (APACHE) III
risk of death, whereby the oxygen component of the APACHE III scoring
system was removed and an adjusted score independent of oxygen was
recalculated.
PaO2 decile comparison
Odds ratio
1
v
10 2
v
10 3
v
10 4
v
10 5
v
10 6
v
10 7
v
10 8
v
10 9
v
10
Adjusted odds ratio
95% CI
0.5
0
1.0
1.5
2.0
2.5
ORIGINAL ARTICLES
Critical Care and Resuscitation Volume 14 Number 1 March 2012 17
The median worst PaO2 value was 117mmHg (IQR, 87–
196 mmHg). Using data from 906 arterial blood gas meas-
urements derived from 49 ventilated stroke patients in the
ICU, we showed that the worst PaO2 defined in the database
correlated well with the peak PaO2 measured in the first 24
hours (r=0.79), and the mean PaO2 measured in the first 24
hours (r=0.68), although there was a weaker correlation
with the median PaO2 measured in the first 24 hours (r=
0.49). The correlation between the worst PaO2 and the
mean and median PaO2 for the entire ICU stay was r=0.46
and r=0.30, respectively. For 86% of these patients the
worst PaO2 value that would have been entered in the
ANZICS CORE database was derived from an arterial blood
gas measurement taken when the patient had an FiO2 ⭓0.5.
Discussion
We found no evidence that in mechanically ventilated
patients with ischaemic stroke, differing levels of the worst
PaO2 in the first 24 hours in ICU influenced mortality, length
of ICU or hospital stay or the likelihood of being discharged
home. The relationship between worst PaO2 and mortality
was similar for patients admitted to hospital from their own
home compared with patients admitted from other hospi-
tals and residential care facilities. It was also similar for
patients admitted to ICU from the ward compared with
patients admitted to ICU from the emergency department.
Eubaric hyperoxia has been shown to increase oxygen
delivery to brain tissue in animal stroke models13 and in
patients with traumatic brain injury.14 Hyperoxia also pre-
vents degradation of the blood–brain barrier during focal
cerebral ischaemia.15 It has been proposed that hyperoxia
shunts blood from regions of normal brain to ischaemic
brain.16 It does this by selectively vasoconstricting cerebral
arteries that perfuse normal brain without affecting arteries
in areas of ischaemic brain, thereby potentially protecting
the ischaemic penumbra. Hyperoxia during ischaemia and
reperfusion in rats subjected to middle cerebral artery
occlusion leads to a reduction in infarct size and neurologi-
cal scores.17 In animals subjected to focal cerebral ischae-
mia, eubaric hyperoxia causes upregulation of antioxidant
enzymes18 and glutamate transporters19 and alters expres-
sion of inflammatory cytokines.20
Conversely, oxygen can reduce cerebral blood flow21 and,
when resulting in hyperoxia, can increase oxidative stress
through the production of oxygen free radicals22 that may
be important in the pathogenesis of ischaemic stroke.9 The
potential harms of oxygen therapy in brain injured patients
are suggested by recent evidence that, in patients with
global hypoxic brain injury after cardiac arrest, hyperoxia
increases mortality23 and, more generally, by the demon-
stration that in critically ill patients mortality increases with
increasing levels of hyperoxia.24
There is evidence that administration of high concentra-
tions of oxygen under eubaric conditions may reduce the
neurological deficit caused by an acute stroke in animal
models.17-21,25 We were unable to assess for such an effect in
critically ill patients with ischaemic stroke and could only use
surrogate measures, such as discharge home, to assess
neurological outcome.
However, experimental administration of oxygen in animal
models differs from use of oxygen in ICU patients with stroke
in two ways. Firstly, in many cases, animal models of stroke
typically involve brief transient arterial occlusion rather than
prolonged arterial occlusion as typically occurs in stroke
patients. Secondly, administration of oxygen in models of
stroke typically occurs at or soon after the onset of brain
ischaemia, whereas, oxygen administration to ICU patients
Table 3. Adjusted odds ratios* (95% CI) for inhospital mortality across deciles of PaO2 for predefined subgroups
“Worst” PaO2,
mmHg
Admitted to hospital
from home
Admitted to hospital from residential care
or transferred from another hospital
Admitted to ICU
from ward
Admitted to ICU
from ED
0–69 1.33 (0.80–2.21) 0.87 (0.42–1.79) 1.72 (0.78–3.79) 1.04 (0.58–1.87)
> 69–83 0.99 (0.59–1.66) 1.49 (0.73–3.03) 1.66 (0.72–3.86) 0.77 (0.43–1.37)
> 83–93 1.12 (0.66–1.88) 0.83 (0.40–1.72) 1.82 (0.78–4.25) 0.78 (0.44–1.39)
> 93–103 1.52 (0.91–2.54) 1.34 (0.63–2.82) 1.74 (0.73–4.13) 1.34 (0.76–2.36)
> 103–117 0.94 (0.56–1.56) 0.92 (0.44–1.93) 1.13 (0.47–2.70) 0.82 (0.46–1.45)
> 117–140 1.00 (0.60–1.69) 1.04 (0.51–2.12) 1.71 (0.69–4.24) 0.82 (0.47–1.44)
> 140–174 1.30 (0.78–2.15) 1.05 (0.52–2.13) 1.5 (0.65–3.44) 0.91 (0.52–1.59)
> 174–226 0.89 (0.54–1.47) 0.76 (0.39–1.48) 1.06 (0.47–2.37) 0.76 (0.44–1.32)
> 226–341 1.05 (0.66–1.66) 0.91 (0.46–1.79) 1.51 (0.71–3.2) 0.77 (0.45–1.30)
> 341–611 1.00 1.00 1.00 1.00
AP3-no-ox= adjusted Acute and Chronic Health Evaluation (APACHE) III risk of death, whereby the oxygen component of the APACHE III scoring system was
removed and an adjusted score independent of oxyge n was recalculated. ED = emergency departmen t. ICU = intensive care unit. AOR = adjusted od ds ratio.
* Adjusted for FiO2 levels, AP3-no -ox and year of admission. All odds ratios are relative to the 10th worst PaO2 decile.
ORIGINAL ARTICLES
Critical Care and Resuscitation Volume 14 Number 1 March 201218
takes place many hours after the onset of ischaemia due to
the time it takes for stabilisation and transfer to the ICU. We
are unable to ascertain whether our measurements are
reflective of oxygen measurements taken earlier on in the
patient’s prehospital (ambulance) or hospital course.
Existing human data from stroke patients are limited and
conflicting.16,26,27 The largest controlled trial of eubaric
oxygen therapy was performed in a single centre in Norway
and involved 550 patients with a stroke (of which 87.6%
were ischaemic) who were allocated by a quasi-randomised
design to 24 hours of treatment with either 3 L of oxygen or
room air.26 In this study, there was no significant difference
in 1-year survival between the oxygen and the room air
groups.26 However, in a subgroup analysis of patients with
minor or moderate strokes, survival was lower in the oxygen
group than the control group (82% v 91%; odds ratio, 0.45
[95% CI, 0.23–0.90]; P= 0.02). For patients with the most
severe strokes, treatment with oxygen did not have an
effect on 1-year mortality (53% v 48%; odds ratio, 1.26
[95% CI, 0.76–2.09]; P= 0.54).26
Chiu and colleagues investigated the feasibility of eubaric
hyperoxia therapy among a group of 46 patients with severe
ischaemic stroke involving more than one-third of the middle
cerebral artery territory.
27
In a non-randomised trial, they com-
pared 40% oxygen administered via a Venturi mask with 2L
oxygen administered via nasal prongs. No significant differ-
ences in mortality or other outcome measures were demon-
strated, although the analyses were limited by low power.
Similar limitations applied in a pilot randomised control-
led trial that investigated the effects of high-flow oxygen in
16 acute ischaemic stroke inpatients with perfusion–diffu-
sion mismatch on magnetic resonance imaging (MRI) scan
(an abnormality thought to correspond to the presence of
ischaemic but potentially salvageable brain tissue).16 It
demonstrated that during hyperoxia there were transient
MRI and clinical improvements within the first 4 and 24
hours respectively; however, these improvements were not
evident by the time of 1-week or 3-month follow-up.16
Given the correlation between worst PaO2 and peak PaO2
in the first 24 hours, our findings suggest that peak arterial
oxygen tension in the first 24 hours was not associated with
a change in the risk of mortality, length of ICU or hospital
stay, or likelihood of being discharged home among venti-
lated critically ill patients with ischaemic stroke. However,
the retrospective nature of our study means that detailed
clinical conclusions cannot be drawn. Furthermore, we
cannot exclude the possibility of benefit or harm among
particular subsets of patients such as those with less severe
strokes, as suggested by the Norwegian study.26
The major strength of our study is its power to detect an
effect, with more than 2600 patients studied. Our findings
are generalisable to ICU practice in that the data were
contributed by 129 ICUs in Australia and New Zealand.
They also included a multifaceted assessment of the inde-
pendent relationship between hyperoxia and outcome with
adjustment for illness severity. However, like other studies of
association using a large database, it is limited by the nature
of the data available. Additionally, 16% of eligible records
were not included in the analysis because of missing data.
The assessment of oxygenation status in the first 24
hours was based on the worst possible arterial blood gas
result in accordance with the PaO2 criteria used for this
component of the APACHE III risk of death score. It would
have been preferable to use the highest (or lowest) PaO2,
regardless of FiO2; however, these data were not available
in the ANZICS APD. However, in a validation study of
arterial blood gas results from 49 patients with ischaemic
stroke admitted to ICU, we determined that the “worst”
PaO2 was moderately well correlated with the peak and
mean PaO2 in the first 24 hours, and was usually taken from
an early blood gas measurement taken on an FiO2 of ⭓0.5.
As a result, we consider that this measure is an acceptable
surrogate for the PaO2 levels in the first 24 hours of ICU
care. An additional weakness of our data is that we did not
adjust for carbon dioxide levels, which are known to
influence cerebral perfusion.28
Our data do not provide any information about the
potential benefits or harms of eubaric hyperoxia in the early
period after acute stroke or exclude a potential effect of
such therapy in particular subgroups of patients. We only
studied patients admitted to ICU. The mortality rate of over
50% seen in our cohort of patients may reflect factors such
as the severity of the stroke and underlying comorbidities or
functional limitations that might drive clinicians to with-
draw active therapy and may confound the detection of an
effect of hyperoxia on outcome. Our results do not provide
information about the usefulness or otherwise of hyperoxia
in stroke patients in non-ICU settings.
Finally, we are unable to comment on the cause of death
or consider other potential confounding variables that might
have affected the relationship between oxygenation and
mortality but were not collected as part of the ANZICS APD.
Summary
In a large multicentre cohort study of patients admitted to
the ICU and ventilated after an ischaemic stroke, we found
no significant association between worst arterial oxygen
pressure in the first 24 hours of ICU admission and in
hospital mortality, length of stay or likelihood of being
discharged home.
Competing interests
None declared.
ORIGINAL ARTICLES
Critical Care and Resuscitation Volume 14 Number 1 March 2012 19
Author details
Paul Young, Intensivist,1 and Honorary Senior Research Fellow2
Richard Beasley, Physician,1 and Director2
Michael Bailey, Chief Biostatistician3
Rinaldo Bellomo, Co-director,3 and Director of Research4
Glenn M Eastwood, Research Manager4
Alistair Nichol, Associate Professor,3 and Intensivist5
David V Pilcher, Intensivist,5 and Director, Adult Patient Database6
Nor’azim M Yunos, Consultant Intensivist3
Moritoki Egi, Consultant Intensivist7
Graeme K Hart, Director4
Michael C Reade, Consultant Intensivist,4 currently, Professor of
Military Medicine and Surgery, Australian Defence Force and Burns
Trauma and Critical Care Research Centre, University of Queensland,
Brisbane, QLD, Australia
D James Cooper, Professor of Intensive Care Medicine and Director,3
and Director of Research5
on behalf of the Study of Oxygen in Critical Care (SOCC) Group
1 Intensive Care Unit, Wellington Regional Hospital, Capital and Coast
District Health Board, Wellington, New Zealand.
2 Medical Research Institute of New Zealand, Wellington, New
Zealand.
3 Australian and New Zealand Intensive Care Research Centre, School
of Public Health an d Preventive Medicine, Monash University,
Melbourne, VIC, Australia.
4 Department of I ntensive Care, Austin Hospital, Melbourne, VIC,
Australia.
5 Department of Intensive Care, Alfred Hospital, Melbourne, VIC,
Australia.
6 Australian and New Zealand Intensive Care Society Centre for
Outcome and Resource Evaluation, Melbourne, VIC , Australia.
7 Department of Anaesthesiology and Resuscitology, Okayama
University Medical School, Okayama, Japan.
Correspondence: Paul.Young@ccdhb.org.nz
References
1 Lopez AD, Mathers CD, Ezzati M, et al. Global and regional burden of
disease and risk factors, 2001: systematic analysis of population
health data. Lancet 2006; 367: 1747-57.
2 Tissue plasminogen activator for acute ischaemic stroke. The National
Institute of Neurological Disorders and Stroke rt-PA Stroke Study
Group. N Engl J Med 1995; 333: 1581-7.
3 Chen ZM, Sandercock P, Pan HC, et al. Indications for early aspirin use
in acute ischaemic stroke: a combined analysis of 40 000 randomized
patients from the chinese acute stroke trial and the international
stroke trial. On behalf of the CAST and IST collaborative groups.
Stroke 2000; 31: 1240-9.
4 Govan L, Weir CJ, Langhorne P; for the Stroke Unit Trialists’ Collabora-
tion. Organized Inpatient (Stroke Unit) Care for Stroke. Stroke 2008;
Jun 12. [Epub ahead of print]
5 Feigin VL, Lawes CM, Bennett DA, Anderson CS. Stroke epidemiol-
ogy: a review of population-based studies of incidence, prevalence,
and case-fatality in the late 20th century. Lancet Neurol 2003; 2:
43-53.
6 Dirnagl U, Iadecola C, Moskowitz MA. Pathobiology of ischaemic
stroke: an integrated view. Trends Neurosci 1999; 22: 391-7.
7 Fisher M, Takano K. The penumbra, therapeutic time window and
acute ischaemic stroke. Baillieres Clin Neurol 1995; 4: 279-95.
8 Singhal AB. Oxygen therapy in stroke: past, present, and future. Int J
Stroke 2006; 1: 191-200.
9 Chan PH. Reactive oxygen radicals in signaling and damage in the
ischaemic brain. J Cereb Blood Flow Metab 2001; 21: 2-14.
10 Stow PJ, Hart GK, Higlett T, et al. Development and implementation of
a high-quality clinical database: the Australian and New Zealand
Intensive Care Society Adult Patient Database. J Crit Care 2006; 21:
133-41.
11 Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic
system. Risk prediction of hospital mortality for critically ill hospitalized
adults. Chest 1991; 100: 1619-36.
12 von Elm E, Altman DG, Egger M, et al; STROBE Initiative. The
Strengthening the Reporting of Observational Studies in Epidemiology
(STROBE) statement: guidelines for reporting observational studies.
Lancet 2007; 370: 1453-7.
13 Shin HK, Dunn AK, Jones PB, et al. Normobaric hyperoxia improves
cerebral blood flow and oxygenation, and inhibits peri-infarct depo-
larizations in experimental focal ischaemia. Brain 2007; 130: 1631-42.
14 Menzel M, Doppenberg EM, Zauner A, et al. Increased inspired
oxygen concentration as a factor in improved brain tissue oxygenation
and tissue lactate levels after severe human head injury. J Neurosurg
1999; 91: 1-10.
15 Liu W, Hendren J, Qin XJ, Shen J, Liu KJ. Normobaric hyperoxia
attenuates early blood-brain barrier disruption by inhibiting MMP-9-
mediated occludin degradation in focal cerebral ischemia. J Neuro-
chem 2009; 108: 811-20.
16 Singhal AB, Benner T, Roccatagliata L, et al. A pilot study of
normobaric oxygen therapy in acute ischaemic stroke. Stroke 2005;
36: 797-802.
17 Henninger N, Bouley J, Nelligan JM, et al. Normobaric hyperoxia
delays perfusion/diffusion mismatch evolution, reduces infarct vol-
ume, and differentially affects neuronal cell death pathways after
suture middle cerebral artery occ lusion in rats. J Cereb Blood Flow
Metab 2007; 27: 1632-42.
18 Bigdeli MR. Preconditioning with prolonged normobaric hyperoxia
induces ischaemic tolerance partly by upregulation of antioxidant
enzymes in rat brain tissue. Brain Res 2009; 1260: 47-54.
19 Bigdeli MR, Hajizadeh S, Froozandeh M, et al. Normobaric hyperoxia
induces ischaemic tolerance and upregulation of glutamate transport-
ers in the rat brain and serum TNF-alpha level. Exp Neurol 2008; 212:
298-306.
20 Bigdeli MR, Khoshbaten A. In vivo preconditioning with normobaric
hyperoxia induces ischaemic tolerance partly by triggering tumor
necrosis factor-alpha converting enzyme/tumor necrosis factor-alpha/
nuclear factor-kappaB. Neuroscience 2008; 153: 671-8.
21 Floyd TF, Clark JM, Gelfand R, et al. Independent cerebral vasocon-
strictive effects of hyperoxia and accompanying arterial hypocapnia at
1 ATA. J Appl Physiol 2003; 95: 2453-61.
22 Yusa T, Beckman JS, Crapo JD, Freeman BA. Hyperoxia increases
H2O2 production by brain in vivo. J Appl Physiol 1987; 63: 353-8.
23 Kilgannon JH, Jones AE, Shapiro NI, et al. Association between arterial
hyperoxia following resuscitation from cardiac arrest and in-hospital
mortality. JAMA 2010; 303: 2165-71.
24 de Jonge E, Peelen L, Keijzers PJ, et al. Association between adminis-
tered oxygen, arterial partial oxygen pressure and mortality in
mechanically ventilated intensive care unit patients. Crit Care 2008;
12: R156.
25 Flynn EP, Auer RN. Eubaric hyperoxemia and experimental cerebral
infarction. Ann Neurol 2002; 52: 566-72.
26 Rønning OM, Guldvog B. Should stroke victims routinely receive
supplemental oxygen? A quasi-randomized controlled trial. Stroke
1999; 30: 2033-7.
27 Chiu EH, Liu CS, Tan TY, Chang KC. Venturi mask adjuvant oxygen
therapy in severe acute ischaemic stroke. Arch Neurol 2006; 63: 741-4.
28 Bulte DP, Chiarelli PA, Wise RG, Jezzard P. Cerebral perfusion response
to hyperoxia. J Cereb Blood Flow Metab 2007; 27: 69-75. ❏
