original article
The
new england journal
of
medicine
n engl j med 360;2 nejm.org january 8, 2009
140
Arterial Blood Gases and Oxygen Content
in Climbers on Mount Everest
Michael P.W. Grocott, M.B., B.S., Daniel S. Martin, M.B., Ch.B.,
Denny Z.H. Levett, B.M., B.Ch., Roger McMorrow, M.B., B.Ch.,
Jeremy Windsor, M.B., Ch.B., and Hugh E. Montgomery, M.B., B.S., M.D.,
for the Caudwell Xtreme Everest Research Group*
From the Centre for Altitude, Space, and
Extreme Environment Medicine, Univer-
sity College London Institute of Human
Health and Performance, London. Address
reprint requests to Dr. Grocott at the
Centre for Altitude, Space, and Extreme
Environment Medicine, University College
London Institute of Human Health and
Performance, 1st Fl., Charterhouse Bldg.,
Archway Campus, Highgate Hill, London
N19 5LW, United Kingdom, or at mike.
grocott@ucl.ac.uk.
Drs. Grocott and Martin contributed equal-
ly to this article.
*The members of the Caudwell Xtreme
Everest Research Group are listed in the
Appendix.
N Engl J Med 2009;360:140-9.
Copyright © 2009 Massachusetts Medical Society.
Abstract
Background
The level of environmental hypobaric hypoxia that affects climbers at the summit
of Mount Everest (8848 m [29,029 ft]) is close to the limit of tolerance by humans.
We performed direct f ield measurements of arterial blood gases in climbers breath-
ing ambient air on Mount Everest.
Methods
We obtained samples of arterial blood from 10 climbers during their ascent to and
descent from the summit of Mount Everest. The partial pressures of arterial oxygen
(PaO
2
) and carbon dioxide (PaCO
2
), pH, and hemoglobin and lact ate concentrations
were measured. The arterial oxygen saturation (SaO
2
), bicarbonate concentration,
base excess, and alveolar–arterial oxygen difference were calculated.
Result s
PaO
2
fell wit h increasing altitude, whereas SaO
2
was relatively st able. The hemoglobin
concentration increased such that the oxygen content of arterial blood was main-
tained at or above sea-level values until the climbers reached an elevation of 7100 m
(23,294 ft). In four samples taken at 8400 m (27,559 ft) — at which altitude the
barometric pressure was 272 mm Hg (36.3 kPa) — the mean PaO
2
in subjects breath-
ing ambient air was 24.6 mm Hg (3.28 kPa), with a range of 19.1 to 29.5 mm Hg
(2.55 to 3.93 kPa). The mean PaCO
2
was 13.3 mm Hg (1.77 kPa), with a range of
10.3 to 15.7 mm Hg (1.37 to 2.09 kPa). At 8400 m, the mean arterial oxygen content
was 26% lower than it was at 7100 m (145.8 ml per liter as compared with 197.1 ml
per liter). The mean calculated alveolar–arterial oxygen difference was 5.4 mm Hg
(0.72 kPa).
Conclusions
The elevated alveolar–arterial oxygen difference that is seen in subjects who are in
conditions of extreme hypoxia may represent a degree of subclinical high-altitude
pulmonary edema or a functional limitation in pulmonary diffusion.
arterial blood gases and oxygen content in climbers on mount everest
n engl j med 360;2 nejm.org january 8, 2009
141
T
he partial pressure of atmospheric
oxygen falls progressively as barometric
pressure decreases with increasing altitude.
Correspondingly, the ability to perform work
(e.g., walking or climbing) diminishes with the
decreased availability of atmospheric oxygen for
aerobic respiration.
1,2
At the summit of Mount
Everest (8848 m [29,029 ft]), the highest point on
the earth’s surface, the partial pressure of in-
spired oxygen (P
i
O
2
) is believed to be very close
to the limit that acclimatized humans can toler-
ate while maintaining functions such as ambula-
tion and cognition.
3
Hillary and Tenzing used
supplemental oxygen to achieve the first ascent
of Everest in 1953. It was not until 25 years after
their ascent that the first ascent of Everest with-
out supplemental oxygen was made by Messner
and Habeler.
4
Currently, less than 4% of persons
who climb Everest do so without the use of sup-
plemental oxygen (Salisbury R., Himalayan data-
base: personal communication).
The only published measurements of the par-
tial pressure of oxygen in arterial blood (PaO
2
)
at such a low barometric pressure were reported
in two studies — Operation Everest II and Opera-
tion Everest III (Comex ’97) — that were designed
to simulate an ascent of Mount Everest by plac-
ing subjects in a hypobaric chamber.
5,6
The sub-
jects in the two studies had a mean (±SD) resting
PaO
2
of 30.3±2.1 mm Hg (4.04±0.28 kPa)
5
and
30.6±1.4 mm Hg (4.08±0.19 kPa),
6
respectively,
at a barometric pressure equivalent to the summit
of Mount Everest (253.0 mm Hg, or 33.73 kPa).
Such profound hypoxemia was tolerable because
the subjects had been gradually acclimatized to
the simulated altitude over a period of 37 to 40
days. In 1981, the partial pressures of oxygen
and carbon dioxide (PaCO
2
) at end expiration were
measured in a single person on Everest’s summit
after the person had been breathing without
supplemental oxygen for approximately 10 min-
utes.
7
With the use of a classic Bohr integration,
the PaO
2
for this climber was estimated to be
28 mm Hg (3.73 kPa).
We made direct field measurements of PaO
2
and arterial oxygen content (CaO
2
) in climbers
breathing ambient air at these extreme altitudes.
Methods
Study Participants
We obtained approval for this study from the
University College London Committee on the Eth-
ics of Non-NHS Human Research. All part icipants
gave written informed consent. The subjects in
this study were 10 healthy climbers (9 men and
1 woman, ranging in age from 22 to 48 years),
who were ascendi ng Everest by it s sout he ast ridge
a s pa r t of a m e d i c a l r es e a rch e x p ed it ion (C au dw e l l
Xtreme Everest).
8,9
All subjects had ascended higher than 6800 m
(22,310 ft) without incident on previous expedi-
tions, and all were well acclimatized, with no evi-
dence of ill effects from high altitude or of other
illnesses. Subjects who were ascending higher
than 7950 m (26,083 ft) had all previously ascend-
ed higher than that altitude without incident.
Collection of Blood Samples
Arterial blood samples were obtained in London,
at an altitude of 75 m (246 ft); at the Everest base
camp, at an altitude of 5300 m (17,388 ft); in
Camp 2, at an altitude of 6400 m (20,997 ft);
in Camp 3, at an altitude of 7100 m (23,294 ft);
and during the descent from the summit at a fea-
ture known as the Balcony, at an altitude of 8400 m
(27,559 ft) (Fig. 1). The samples that were obtained
in London and at the Everest base camp were
obtained with the subject at rest, with the use
of indwelling radial arterial cannulae that were
placed as part of other study protocols; these
samples were analyzed immediately. Samples ob-
tained at an altitude higher than the Everest base
camp were obt ained from the right femoral arter y,
identified by digital palpation. Intraarterial place-
ment of the needle (21-gauge) was confirmed by
pulsatile f illing of a heparinized 2-ml oiled glass
syringe (Fisher Scientific). Syringes were imme-
diately sealed with an airtight cap and placed in
a plastic bag, which in turn was placed in an ice-
water slurry inside an insulated vacuum flask.
The flask was rapidly transported to a laboratory
at Camp 2 in the Western Cwm; the length of t ime
for this transfer to be completed was recorded.
Barometric pressure was measured at t he alt itude
at which the blood samples were taken, with the
use of a handheld digital barometer (GPB 2300,
Greisinger Electronic). Arterial samples were ob-
tained by two investigators, both of whom had
extensive experience with cannulation of the fem-
oral artery and blood sampling.
Supplemental Oxygen
Supplemental oxygen was used only at or above
Camp 3 (7100 m), with the following flow rates:
2 to 3 liters per minute while the subject was
The
new england journal
of
medicine
n engl j med 360;2 nejm.org january 8, 2009
142
climbing and 0.5 liter per minute while the sub-
ject was sleeping. Supplemental oxygen was infre-
quently used while the subjects were resting at
Camp 3 and Camp 4 (7950 m). At Camp 3, arte-
rial samples were obtained after the subjects had
been breathing ambient air for at least 4 hours.
At the Balcony, samples were obtained after the
subjects had been breathing ambient air for 20
minutes in order to achieve an adequate washout
of supplemental oxygen.
Analysis of Blood Samples
Arterial blood samples were analyzed with the
use of the RapidLab 348 (Siemens Medical Solu-
tions Diagnostics) blood gas analyzer, which does
not contain a co-oximeter. The PaO
2
, the PaCO
2
,
and the pH were measured. Values for the bicar-
bonate concentration, blood base excess, and oxy-
gen saturation (SaO
2
) were calculated with the
use of formulas currently approved by the Clini-
cal Laboratory Standards Institute
10
(
Table 1
). The
blood lactate concentration was measured with a
separate device (Lactate Scout, EKF Diagnostic).
Barometric pressure was measured at the site of
analysis wit h the use of the same model of barom-
eter as that used at the sampling site.
The blood gas analyzer was altered from its
original specification so that it would function at
high altitude. The analyzer’s internal barometer
was bypassed with a fixed resistor so that the
analyzer always read as if the barometric pressure
was a constant 450 mm Hg (60.0 kPa), regard-
less of altitude. This modification was necessary
in order to circumvent an inbuilt mechanism that
prevented the analysis of samples at a baromet-
ric pressure lower than 400 mm Hg (53.3 kPa).
To replicate the barometric-pressure correction
that the machine would normally apply in its un-
modified form, true gas partial-pressure values
were obtained by inserting the machine-derived
values into Equation 1, shown in
Table 1
. This
calculation is identical to that performed inter-
nally by the arterial blood gas analyzer during
normal function at lower altitudes.
The subjects’ temperatures at the time of sam-
pling were assumed to be the same as the tem-
perature of the blood gas analyzer — namely,
37.0°C (98.6°F). The analyzer was validated in a
hypobaric chamber at the equivalent of 4000 m
(13,123 ft) and then revalidated in the field, at
5300 m and 6400 m, the altitudes at which mea-
surements of arterial blood gas were performed
Figure 1. Barometric Pressure (P
b
) and Partial Pressure of Inspired Oxygen
(P
i
O
2
) in Blood Samples Obtained from Subjects Breathing Ambient Air
at Various Altitudes between London and the Summit of Mount Everest.
In Panel A, the measurements at the summit are reported from West et al.
7
The other measurements were performed by the investigators.
n engl j med 360;2 nejm.org january 8, 2009
143
arterial blood gases and oxygen content in climbers on mount everest
in this study. Validation involved the analysis of
aqueous trilevel quality-control solutions (RapidQC
Plus, Bayer HealthCare) with known values of pH,
PaO
2
, and PaCO
2
. Two-point calibration of the
RapidLab 348 gas sensors and electrodes was
performed automatically according to the manu-
facturer’s specifications with the use of standard
gases and electroly te solutions, respectively. Each
arterial blood sample was analyzed three times,
and the mean of these values is reported. Because
the pulse oximeters available to us were not cali-
brated below 70% SaO
2
, we chose to calculate
SaO
2
using Equation 2, shown in
Table 1
. All
reported values for SaO
2
are calculated values,
except for the values for four subjects at an alti-
tude of 5300 m; for these subjects, values ob-
tained by peripheral-pulse oximetry (Onyx 9500,
Nonin) are reported owing to an isolated failure
of the pH electrode on the blood gas machine,
an electrode that was subsequently replaced. All
measured and calculated values for SaO
2
at an
altitude of 5300 m fell within the calibrated range
of the pulse oximeter.
The hemoglobin concentration was measured
in venous blood collected from subjects in London
before the expedition, at the Everest base camp
(at 2-week intervals during the expedition), and
at Camp 2, with the use of a handheld photo-
Table 1. Equations Used for the Calculation of Arterial Blood Gas Values and Arterial Oxygen Content.*
1. Actual partial pressure of oxygen in arterial blood samples (PaO
2
)
measured P
b
at site of ABG machine – SVP
resistor set P
b
(i.e., 450 mm Hg) – SVP
2. Arterial oxygen saturation (SaO
2
)†
N
4
– (15 × N
3
) + (2045 × N
2
) + (2000 × N)
N
4
– (15 × N
3
) + (2400 × N
2
) – (31,100 × N) + (2.4 × 10
6
)
N = PO
2
× 1 0
[0.48 × (
p
H – 7.4) – 0.0013 × BE]
3. Bicarbonate concentration (HCO
3
−
)†
HCO
3
−
(mmol/liter) = 0.0307 × PaCO
2
× 1 0
(
p
H – 6.105)
4. Blood base excess (BE)†
BE = (1 − 0.014 × Hb) × (HCO
3
−
− 24.8) + (7.7 + 1.43 × Hb) × (pH −7.4)
5. Partial pressure of inspired oxygen (P
i
O
2
)
P
i
O
2
= (P
b
– SVP) × F
i
O
2
6. Partial pressure of alveolar oxygen (P
a
O
2
) — The “alveolar gas equation”
P
a
CO
2
+
[
P
a
CO
2
× F
i
O
2
×
1 – RER
]
RER
RER
P
a
CO
2
was assumed to be equal to PaCO
2
7. Arterial oxygen content (CaO
2
)
CaO
2
= SaO
2
× Hb × 1.39 + (PaO
2
× 0.03)
* ABG denotes arterial blood gas, BE base excess, F
i
O
2
fraction of inspired oxygen, Hb hemoglobin concentration, P
a
CO
2
partial pressure of alveolar carbon dioxide, PaCO
2
partial pressure of arterial carbon dioxide, P
b
barometric pressure,
PO
2
partial pressure of oxygen, RER respiratory exchange ratio, and SVP saturated vapor pressure at body temperature
(47 mm Hg).
† This equation is currently approved by the Clinical Laboratory Standards Institute (http://www.clsi.org).
10
× value of PO
2
given by ABG machine
× 100
PaO
2
=
SaO
2
(%)
=
P
a
O
2
= P
i
O
2
–
The
new england journal
of
medicine
n engl j med 360;2 nejm.org january 8, 2009
144
metric device (HemoCue Whole Blood Hemoglo-
bin System, HemoCue). Venous samples were
obtained at the same time as arterial samples in
London, at the Everest base camp, and at Camp 2.
For the hemoglobin concentration at Camp 3 and
the Balcony, we used the mean of the hemoglo-
bin values obtained at the Everest base camp
9 days before and 8 days after the arterial sample
at the Balcony was obtained (Fig. 1); these values
were used in the calculation of CaO
2
, bicarbonate
concentration, and blood base excess.
The partial pressure of alveolar oxygen (P
a
O
2
)
at the time of sampling was estimated by apply-
ing the alveolar gas equation to the calculated
P
i
O
2
(Equations 5 and 6 in
Table 1
). The resting
respiratory exchange ratios necessary for these
calculations were obtained for three of the sub-
jects at the South Col of Everest on the day be-
fore summiting, with the use of breath-by-
breath analysis equipment (MetaMax 3B, Cortex
Biophysik).
Results
Collection of Samples
The climbers reached the summit of Mount Ever-
est on the morning of May 23, 2007, after having
spent 60 days at an elevation higher than 2500 m
(8202 ft). The location, altitude, barometric pres-
sure, and P
i
O
2
for each sampling site are shown
in Figure 1. All femoral arterial blood samples
were obtained without complications on the first
attempt. Ten samples were obtained in London,
nine at the Everest base camp, nine at Camp 2,
six at Camp 3, and four at 8400 m. The reasons
for not obtaining samples were as follows: at the
Everest base camp, one subject was unwell; at
Camp 2, one subject was unwell; at Camp 3, four
subjects were not present when the Sherpa was
available to transport the sample; and at the Bal-
cony, two subjects did not reach this altitude, and
four subjects were not present when the Sherpa
was available to transport the sample. One sam-
ple at Camp 2 repeatedly clotted in the arterial
blood gas machine, so no data are available for
that sample. In all cases, the interval bet ween sam-
pling and analysis was less than 2 hours.
Arterial Blood Gases
Measured PaO
2
and hemoglobin values, along
with calculated SaO
2
and CaO
2
values, are shown
in Figure 2. The CaO
2
value at sea level was main-
tained up to an altitude of 7100 m and fell below
baseline only at 8400 m; at this altitude, the mean
CaO
2
for the four subjects was calculated to be
145.8 ml per liter. Mean PaCO
2
values fell with
increasing altitude, from 36.6 mm Hg (4.88 kPa)
at sea level to 20.4 mm Hg (2.72 kPa) at 5300 m,
18.2 mm Hg (2.43 kPa) at 6400 m, and 16.7 mm Hg
(2.23 kPa) at 7100 m; corresponding pH values
were 7.40, 7.46, 7.51, and 7.53.
The results of the arterial blood gas analysis
and the hemoglobin and lactate concentrations in
four subjects at 8400 m are shown in
Table 2
. The
mean PaO
2
and PaCO
2
values were 24.6 mm Hg
(3.28 kPa) and 13.3 mm Hg (1.77 kPa), respec-
tively. The mean P
i
O
2
value at 8400 m was calcu-
lated to be 47.0 mm Hg (6.27 kPa) at the time of
arterial sampling. Calculated values for P
a
O
2
,
resting respiratory exchange ratios, and the alveo-
lar–arterial oxygen difference in four subjects at
8400 m are shown in Table 2. The mean P
a
O
2
and
22p3
250
Mean Value
150
200
100
50
0
75 5300 6400 7100 8400
Altitude (m)
No. of Subjects
10 9964
AUTHOR:
FIGURE:
JOB:
4-C
H/T
RETAKE
SIZE
ICM
CASE
EMail
Line
H/T
Combo
Revised
AUTHOR, PLEASE NOTE:
Figure has been redrawn and type has been reset.
Please check carefully.
REG F
Enon
1st
2nd
3rd
Grocott
2 of 2
01-08-09
ARTIST: ts
36002 ISSUE:
Partial pressure
of arterial oxygen
(mm Hg)
Arterial oxygen
saturation (%)
Hemoglobin
concentration
(g/liter)
Arterial oxygen
content (ml/liter)
Figure 2. Changes in the Arterial Mean Partial Pressure of Oxygen, Oxygen
Saturation, Hemoglobin Concentration, and Oxygen Content in Climbers
on Mount Everest.
I bars denote standard deviations.
