Content uploaded by Denis Barclay
Author content
All content in this area was uploaded by Denis Barclay on Dec 02, 2014
Content may be subject to copyright.
Q
Macmillan Press Ltd. 1994
Europeun
Journul
of
Clinicul
Nutrition
(1994)
48.
625-632
Receiwai
14
September
1993: accepted
I?
April
1994
Altitude correction
for
hemoglobin
H.
Dirren',
M.
H. G.
M.
Logman',
D.
V.
Barclay'
and
W.
B.
Freire'
'Nestlé
Research
Centre, Nestec
Ud,
RO.
Box
44.
Ch-1000
Lausanne
26,
Switzerland:
und
'the
Consejo
Nucional
de
Desurrollo,
RO.
Box
537-C, Quito,
Ecitador
Objective:
To
propose a correction
for
the hemoglobin (Hb) increase induced by
altitude-associated hypoxia.
Design:
Part of a national study
of
nutrition and health of preschool children (0-59
months), based on a stratified, probabilistic, cluster sample.
Sefting:
Coastal and Andean regions of Ecuador, comprising about
97%
of the
population, living at altitudes ranging from sea level
to
3400m.
Subjects:
Subsample of 469 girls and boys, 6-59 months old, with normal iron (Fe)
status parameters, ¡.e serum ferritin
al
O
pg/l, transferrin saturation
2
12%,
zinc
protoporphyrin
S2.8
pg/g Hb.
Results:
Exponential regression curves are adapted through the Hb values of the
children, grouped by altitude ranges. and through the data reported by Hurtado in
1945
for
male adults. From these exponential curves, correction factors for Hb are
derived for altitudes ranging from sea level
to
34".
Conclusion:
The striking parallelism between the hypoxia-induced hemoglobin
increase with altitude
in
young children (girls and boys) and that in male adults
strongly suggests that the proposed correction factors for Hb are applicable for all
ages and possibly both genders, at least
in
the Andes.
Descrintors:
altitude. Andes. anemia. children. hemoelobin. hvooxia
I
Introduction has long been recognized to have an effect on the
red blood cell count and the hemoglobin level
Iron deficiency anemia is a widespread nutri- (Hurtado, Merino
&
Delgado, 1945). Investiga-
tional problem in developing countries, partic- tions into the mechanisms enabling residents to
ularly
in
infants and young children (DeMaeyer live successfully under the low atmospheric
p02
&
Adiels-Tegman,
1985),
where it is related to
of
high altitude often focused on hematological
depletion of iron stores in the first few months parameters because of the essential role of blood
after birth, to the use of weaning foods that lack
in
oxygen delivery. The decline
in
oxygen partial
sufficient iron. and
to
the rapid growth rate pressure with altitude is accompanied by a
(Dallman, Siimes
&
Stekel,
1980).
Later on, decline in the oxygen saturation
of
arterial blood
during the preschool period, anemia
is
likely to and
by
an
increased concentration of hemoglobin
occur because of low iron content of the food, (Hurtado
et
al.,
1945).
Andean
1
the presence
of
substances inhibiting iron chronically exposed to high altitude
are
charac-
absorption, thc prefercntial intake
of
heme iron terized by high concentrations
of
hemoglobin
by
adults, and hookworm infestation (Layrisse relative to those
of
sea level residents (e.g.
&
Roche, 1964; Dallman
et
al.,
1980).
18.2-19.0g/dl for healthy adult males at 3400m)
Beside nutritional aspects, other factors may as found by Hurtado
et
al.
(1945). This has
been
influence the indicators of iron status. Altitude interpreted as an adaptive response which main-
Correspondence t<>:
Dr
H.
Dirren. Nsstle Rcbearch Centre, Nestec Ltd,
PO
Boa
44,
CH-1000 Lausanne
26,
Switzerland.
'
626
H.
Dirren
et
al.
tains oxygen supply under conditions of arterial
hypoxia.
This finding is
so
consistent that altitude-
based hemoglobin corrections, using the
Andean findings of Hurtado
et
al.,
have been
published and utilized
in
population surveys
(Cook
et
al., 1971).
A simplified correction factor of 4% of
hemoglobin increase per 1OOOm has also been
used to adjust for the incrfease in hemoglobin at
high altitudes (Dallman
et
al.,
1980; Estrella
et
al.,
1987). However, Hurtado's hemoglobin
curve
as
a
function of altitude behaves exponen-
tially and does
not
support this correction factor.
Compared with an exponential curve, such
a
4%
correction would lead to an underestimation of
the prevalence of anemia at high altitudes, and to
a slight overestimation at lower altitudes.
Hemoglobin corrections for altitude are
important to estimate anemia in community
nutrition surveys or to diagnose anemia in
individuals living at high altitudes, and to make
it possible to compare hemoglobin levels
between studies conducted at different altitudes.
To estimate these hemoglobin corrections for
children, an exponential model was fitted
through the hemoglobin values
of
the Ecua-
dorian children who participated in the National
Nutrition and Health Survey of
1986
(Freire
et
al.,
1988). Their applicability
to
other age
groups was evaluated by comparing them with
those of Hurtado for adult males.
Subjects
and
methods
Population
This study was part of the National Nutrition
and Health Survey of Ecuador, implemented in
1986. It involved the examination of 8100
children aged 0-59 months, representing
about
1.2 million in that age group in Ecuador, living
at altitudes ranging from
sea
level to 3400m
(Freire
e?
al.,
1988). From the total sample,
a
subsample
of
1620 children was selected to
investigate specific nutritional deficiencies by
measuring biochemical parameters and dietary
intakes. Blood sampling was carried out in 1570
children giving 1552 blood samples suitable for
the measurement of hemoglobin
(Hb)
levels;
most samples were collected by venipuncture,
the rest by heel- or finger-prick. Of these
1552
samples, 1413 were from children aged
6-59
months, and for
951
of the latter samples
sufficient blood was available for the measure-
ment of several iron status parameters
as
described below.
For the determination
of
the change of
hemoglobin, and thus the hemoglobin correction
as
a
function
of
altitude, only data for those 469
children, aged 6-59 months, who had normal
iron status parameters, were used, i.e. those with
serum ferritin
10
pgA, transferrin saturation
3
1296,
zinc protoporphyrin ~2.8 pg/g hemoglo-
hin (Cook
et
al..
1985). Children aged e6 months
were excluded: because of rapid changes of
hemoglobin concentrations following birth
(Dallman
ef
al.,
1980).
Biochemical analyses
Fasting venous blood samples were collected in
most children 26 months
of
age. Blood was
collected on ethylene diamine tetra-acetic acid
(EDTA) for hematological measurements.
In
order to prepare serum, whole blood was left to
clot in tubes without anticoagulant, in the dark,
for
1
h in portable refrigerators at about
10-15"C, after which serum was separated,
frozen and stored at -80°C until analyzed.
All biochemical measurements were carried
out in duplicate. Hemoglobin concentration was
measured
in
the field by the cyanomethemoglo-
hin method (ICSH, 1978), using
a
Boehringer
Mannheim kit (Boehringer Mannheim GmhH,
Mannheim, Germany) with
a
battery-operated
Compur
M
1000 D2 photometer (Compur
Electronic GmbH, München, Germany), and
hematocrit was determined by centrifugation
with
a
battery-operated Compur MllOO micro-
hematocrit centrifuge (Compur Electronic
GmhH, München, Germany). Zinc protopor-
phyrin was measured by front-face fluorometry
(Blumherg
er
al.,
1977)
with
an
AVIV 206
hematofluorometer (AVIV Biomedical Inc.,
Lakewood, New Jersey. USA). Serum ferritin
was measured by immunoradiometry (Miles
et
al.,
1974) with Ramco kits (Ramco Laho-
ratories Inc., Houston, Texas, USA). Because
of
the lack
of
intemational standardization of the
serum ferritin assay (VanOost
et
al.:
1982)
prevailing
at
the time of the survey, two vials of
the intemational ferritin standard
No.
80/602
(National Biological Standards Board, London,
UK)
were measured and the ferritin values were
adapted according to the linear regression
calculated through the given and the measured
values (Ramco kit ferritin value
=
0.68
*
I
Altitude
correction
for
hemoglobin
621
International ferritin standard
-
0.54).
Serum .In order to compare the hemoglobin-altitude
iron and total iron-binding capacity (TIBC) relationship of our study with that of Hurtado
et
were measured with the ferrozine colorimetric
al.
(1945),
a
similar exponential model was
method (Williams, Johnson
&
Haut,
1977)
calculated for their hemoglobin data. The latter
using Roche reagents (Roche Diagnostica, include their own data at sea level,
3730
m and
Basel.
Switzerland) and transferrin saturation,
4540111,
as well as data reported by Andresen
expressed
as
a percentage, was calculated by and Mugrage (1936) at 15” in Denver, and
dividing the serum iron concentration by the data reported by Talhott and Dill (1936) at
total iron-binding capacity. The acute phase 5335
m
in Chile. This gave the set of constants:
proteins C-reactive protein (CRP) and alpha-
a
=
6.83,
b
=
0.000445
and
c
=
153.2.
1
-acid glycoprotein (AGP) were determined by Finally, an exponential function parallel to
,
turbidimetry (Müller-Matthesius
&
Opper. that of Hurtado
et
al.
(it. with the same
a
and
)
with a COBAS FARA centrifugal ana-
b
parameters) was fitted through the hemoglo-
lyzer (Roche Diagnostica, Basel. Switzerland), bin data of the Ecuadorian children (referred to
using Behring reagents, antibodies and stan- later on as parallel exponential regression).
dards (Behring Werke. Marburg, Germany). For estimating the goodness of fit of the
Infection and/or inflammation were considered parallel exponential regression curve, the sum
to be present when elevated levels of AGP of squares
of
the pure error was calculated by
(>1.4g/l) or CRP
(>12mg/l)
were detected one-way ANOVA. and the residual sum of
(Ingenbleek
&
Carpentier,
1985).
squares (total error) was obtained from the
Quality control was carried out using com- exponential regression. The goodness
of
fit
mercial controls with known target values and esimation was calculated by subtracting the
with identified and anonymous serum
pools.
pure error from the total error.
To
calculate the
The coefficients
of
variation of the hemoglobin ANOVA, an altitude grouping was used in
measurements, determined with commercial which ‘sea level’ (0-28 m) and ‘Quito’
controls (Merz and Dade AG, Diidingen. Swit- (2501-2800m) formed two groups, and the
zerland) for the levels of 85, 110 and 160 g/l other altitude levels of the children formed eight
were 0.8,
1.1
and
I.
1%
respectively
(n
=
73),
groups with approximately equal numbers of
over the 8 months
of
the survey. cases (about 20-30 cases per group).
.
;tical
analyses
Results
“ata were analyzed with the BMDP statis-
tical package (BMDP Statistical Software. The median of altitude. the range, the mean and
1990). Statistical analyses reported in this paper the standard error of hemoglobin and hemato-
comprised one-way ANOVA, exponential crit. and the number of children for each altitude
regression (with hemoglobin
as
dependent and
group
used to perform the exponential regres-
altitude
as
independent variables),
a
test of sions and the one-way ANOVA are show1
goodness of
fit,
the Pearson X2-test. and the Table
1.
Kruskal-Wallis test. The average
age
of the 469 childí-en included
To
investigate whether the data fitted an in the analyses was 40.2 months (SD
=
12.6).
exponential model in order
to
obtain an altitude
No
differences in age distributions between the
correction for hemoglobin,
an
exponential different altitude groups were found (Kruskal-
regression of the following form was performed Wallis test statistic
=
6.53.
P
=
0.70).
through the 469 hemoglobin values: The free exponential curve fitted through the
data of the Ecuadorian children is of the form:
Hh
=
a
*
exp’
*
ALT
+
c
Hh ~
3.44
*
exp0.000633*ALT
+
116.9 (free fit)
where Hb is the hemoglobin concentration
(gfl).
ALT is the altitude (m). and
a.
b and
c’
are The goodness of fit test (Ho: the exponential
constants calculated in the regression (referred model fits with the actual data) shows that this
to later on
as
free exponential regression). To be curve has
a
good fit with
our
data (Le.
F7,459
=
ahle to calculate the goodness
of
fit, the data 1.33,
P
=
0.24). The exponential curve through
were regrouped by altitude ranges
as
discussed the reported Ecuadorian data parallel to that
below. fitted through Hurtado’s data has an equally
.
630
H.
Dirren
et
al.
exponential correction is 29.1
%
compared to
only 8.8% when using the
4%
correction.
The exponential curves fitted through the
data for healthy male adults reported by Hur-
tado
et al.
and through the data of 469
Ecuadorian children 6-59 months of age show-
ing no biochemical signs of iron store depletion
(serum ferritin in normal range) or of iron
deficiency (transferrin saturation and proto-
porphyrin in normal range) are virtually paral-
lel. Since, however, the latter showed
a
sharp,
non-physiological increase at altitudes above
3400 m, an exponential function parallel to that
of Hurtado was
also
fitted through the child-
ren’s data and subsequently used for calculating
the hemoglobin correction. The goodness of fit
was equally good for the free and the parallel
exponential regressions and the hemoglobin
differences between them were physiologically
insignificant (<2g/l) at any altitude up to
3400111, the highest inhabited altitude in the
present survey.
Several factors could distort the hemoglobin-
altitude relationship induced by the pure effect
of hypoxia. They include factors influencing
hemoglobin and which could show an altitude-
dependent distribution. such
as
iron status,
infection and age. Therefore, cases with iron
store depletion or iron deficiency were excluded
from the exponential regression, thereby reduc-
ing the possibility that altitude-related regional
differences in iron status (Freire, Dirren
&
Barclay. 1990a) would influence the hemoglo-
bin relationship with altitude. Infection and
inflammation (defined here as AGP
>
1.4g/l
and/or CRP
>
12
mgfl), are known to affect the
biochemical parameters of iron status (Lip-
schitz, Cook
&
Finch, 1974; Dallman
et
al.,
1980;
Kushner, 1982) and this has been con-
firmed
in
the present survey (Freire, Dirren
&
Barclay, 1990b). However, the exclusion
of
cases with abnormal values of zinc proto-
porphyrin and transferrin saturation eliminated
most of the children with infection
or
inflamma-
tion. The few remaining cases with elevated
AGP or CRP showed no difference in mean
hemoglobin compared to the non-infected
groups (one-way ANOVA.
F,,447
=
0.03,
P
=
0.86) and were therefore kept
in
the calculation.
Finally. the age dependence of hemoglobin
during childhood could distort the hemoglobin-
altitude relationship if the age distribution were
not similar in the different altitude groups.
Hemoglobin increases particularly rapidly with
age between 3 and 12 months in industrialized
countries (Dallman
et al.,
1980). In the present
study, the increase was much slower and began
to level off at about 30 months of age. The
carefully controlled representativeness of the
sample gave homogeneous distributions of age
as
a
function of altitude (Kruskal-Wallis test
statistic
=
6.53,
P
=
0.70); children in the age
range 6-59 months were therefore included.
Several important differences in hemoglobin
levels between Andean and Himalayan studies
have been reported. Studies on Sherpa and
Tibetan adolescent and adult males residing
in
the Himalayas (Adams
&
Shresta, 1974; Adams
&
Strang, 1975; Samaja, Veicsteinas
&
Cerre-
tell¡, 1979; Beall
&
Reichsman, 1984) have
reported mean hemoglobin concentrations
between
1
O
and 20 gfl lower than those reported
in the Andes (Hurtado
et al.,
1945; Cosio, 1973;
Frisancho, Velasquez
&
Sanchez, 1975; Arnaud
et al.,
1979; Tufts
et al.,
1985), although some
results showed closer agreement (Son, 1979;
Clench
et al.,
1981: Garruto
&
Dutt, 1983).
Several hypotheses have been presented to
explain these differences, including:
(a)
differ-
ence in exposure to hypoxia due to variations in
atmospheric pressure with latitude, with higher
pressure in the Himalayas than in the Andes at the
same altitude (West
et
al.,
1983),
to
differences
in sleep apnea and hypoxemia (Beall, Stroh1
&
Brittenham. 1983) or to non-permanent residence
at higher altitude
for
certain groups (Arnaud
et
al.,
1979; Beall
&
Goldstein, 1987); (b) differ-
ence
in
response
to
similar levels of hypoxemia
in Himalayan and Andean populations (Reyna-
farje. 1957; Morpurgo
et
al.,
1972; Son, 1979:
Samaja
et
al..
1979; Tufts
et al.,
1985: Garruto
&
Dutt, 1983; Frisancho, 1988); within the same
Andean region, Arnaud
et al.
(1979) have
reported such differences in acquired adaptations
between two ethnic groups; (c) difference
in
disease pattems, particularly lung conditions
related to mining (Frisancho, 1988); and (d) a
series of methodological shortcomings plaguing
some
of
the Himalayan data, including very small
samples, failure to report age, gender, altitude of
residence, season, or size of the sample (Reyna-
farje, 1957; Pugh, 1966; Morpurgo
et
al.,
1972;
Adams
&
Shresta, 1974; Morpurgo
et al.,
1976;
Beall
&
ReichSman, 1984; Lanick
&
Topgyal,
1985; Beall
&
Goldstein, 1987). Obviously the
present study cannot provide any explanation for
,
.
,
.
Altitude
correction
.for
hrrnoXlobin
63
1
the difference in hemoglobin increase with
altitude reported in the Andean and the Hima-
layan regions.
The strength of the present study lies in the
large number
of
hemoglobin values and of
several other parameters of iron status which
have been measured in subjects living at
altitudes ranging from sea level to ahout
3400m, within the same study and under strict
methodological control. This led to the pro-
posed exponential model for hemoglobin
as
a
function
of
altitude. applicable to children of
both genders, aged 6 months to
5
years,
representing the cultural and ethnic mix of
.
Ecuadorian preschoolers, and including the
possible effects of short-term travel to other
altitudes for some of the children of the sample.
The striking parallelism between the present
results and the data reported by Hurtado for
apparently healthy adult males strongly sug-
gests that the hematopoietic response to alti-
tude-related hypoxia is about the same for all
ages, and possibly for both genders, at least in
the Andes; the fairly constant, large differences
in
the hemoglobin concentrations at each alti-
tude between both studies are attributable
mainly to hemoglobin increases with age;
differences in analytical methods used
in
the
two studies may
also
contribute.
For
the altitude
correction, however, the central point remains
the parallelism between the two studies.
The suggested correction of hemoglobin
values at different altitudes, presented
in
Table
,
2
for preschoolers, may be easily applied in two
ways.
For
assessing iron status of individuals or
for the daily work of pediatricians and health
workers,
an
altitude-adapted hemoglobin cnt-
off for anemia may be calculated by adding the
correction factor
of
Table
2
to the sea level
WHO-recommended cut-off, for example
110
g/l
for
6
months to
1
O
years (Dallman
et
ul.,
1980).
As
mentioned above, it is likely that the
same correction may he applicable to
all
age-
and gender-specific cut-Offs recommended by
WHO, at least for the Andean region. It is worth
noting that this altitude correction will retain the
well-known problems of lack of sensitivity and
specificity associated with the use
of
single
hemoglobin cut-Offs for defining anemia (Dal-
man,
1984;
Estrella
et
al.,
1987).
The second approach for correcting the
altitude effect consists of ‘projecting’ the hemo-
globin values to sea level; this is obtained by
subtracting from the measured hemoglobin
values at given altitudes. the corresponding
correction factors given in Table
2
for these
altitudes. This is the suitable approach when
comparing hemoglobin differences between
population groups living at different altitudes.
Although rarely used
in
community studies,
researchers may choose to use hematocrit
instead
of
hemoglobin. In view of the similar
increase
of
hemoglobin and hematocrit with
altitude
as
shown in Fig.
2,
the same exponen-
tial model may be applied for the hematocrit
correction for altitude.
I
Adams WH
&
Shresta
SM
(1974): Hemoglobin
IeveIs,
63,
301-306.
vitamin
BIZ.
and folate status in a Himalayan village.
Beall
CM.
Stroh1 KP
&
Brittcnham
GM
(1983): Reappraisal
Am.
J.
Clin.
Nutr.
27,
217-219. of Andean high altitude erythrocytoqis from a Himalayan
Adams WH
&
Strang
LJ
(1975):
Hemoglohin levels
in
perspective.
Sem.
Resp.
Med.
5(2),
195-201.
persons
of
Tibetan ancestry living at high altitude.
Proc.
Blumherg
WE.
Eiainger
J.
Lamola
AA
&
Zuckerman
DM
SOC.
fip.
nid.
Med.
149,
1036-1039.
(1977):
The
hematofluoromcter.
Clin.
Chem.
23(2),
Andrescn MI
&
Mugrage
ER
(1936):
Red blood cell values
for normal men and womcn.
Arch.
ht.
Med.
58.
BMDP Statistical Software Inc
(1990):
BMDP
PC-90
270-274.
(1979):
Melhaemoglohin and erythrocyte reducing
sys-
Hematocrit and hemoglobin. ATP, and DPG concentra-
tems
in high-altitude natives.
Ann. Hum.
Biol.
6(6).
tions
in
Andean man: The interaction of altitude and Irace
585-592.
metals with glycolytic and hematologic parameters in
Beall
CM,
Goldalein MC
&
the Tibetan Academy
of
Social man.
Prq
Clin.
Bid.
Res.
55,
747-766.
Sciences
(19871:
Hemoglobin conccntration
of
pastoral
Cook
JD,
Alvarado
J,
Gutniaky A, Jamra
M,
Labardini
J,
nomads pcrmanenlly resident at
4,850-5,450
meters in Layriase
M,
Linares
I,
Loria A, Maspes V, Restrepo A.
Tihet.
Am.
J.
Phys.
Anthr.
73.
433-438. Reynafarje
C,
Sheher-Medal
L.
Vélez
H
&
Viteri
F
Beall
CM
&
Reichsman AB (1984): Hemoglohin levels in
a
(1971):
Nutritional deficiency and anemia in Latin
Himalayan hieh altitude Doouletion.
Am.
J.
Phv$.
Anthr.
America: A collaborative
study.
Blood
38(5),
59
1-603.
