Acta Obstetricia et Gynecologica.

2006; 85: 49
/55

ORIGINAL ARTICLE

Maternal and fetal outcome after severe anemia in pregnancy in rural

Ghana

DIEDERIKE GEELHOED, FLORENCE AGADZI, LUCIA VISSER, EMELIA

ABLORDEPPEY, KOFI ASARE, PETER OROURKE, JULES SCHAGEN VAN LEEUWEN &
JOS VAN ROOSMALEN

Holy Family Hospital, Berekum and Mathias Hospital, Yeji, Ghana

Abstract
Background . Anemia in pregnancy contributes to poor outcome for mother and child in low-income countries. This study
analyzes adverse maternal and fetal outcome after severe anemia in pregnancy in rural Ghana. Methods. A cohort study in
two (sub)district hospitals, including 157 pregnant women exposed to severe anemia (HbB/8.0 g/dl) and 152 nonexposed
pregnant women (Hb ]/10.9 g/dl), matched for age and parity strata. Adverse outcomes analyzed were postpartum
hemorrhage, need for blood transfusion, maternal mortality, low birth-weight, and perinatal mortality. Results. Compared
to nonexposed women, exposed women had an increased risk of maternal death (5/157 versus 0/152). Fetal outcome did not
significantly differ between the study groups, although perinatal mortality was increased with exposure to Hb B/7.0 g/dl (OR
3.1; 95%CI 1.0
/9.4), and low birth-weight was increased with exposure to Hb B/6.0 g/dl (OR 2.5; 95%CI 1.2
/5.4). Overall
fetal outcome was significantly better when hemoglobin prior to childbirth was at least 8.0 g/dl (OR 3.9; 95%CI 1.6
/9.6),
body mass index at least 20 kg/m2 (OR 2.8; 95%CI 1.5
/5.3), and number of antenatal visits at least 4 (OR 2.0; 95%CI
1.1
/3.7). Conclusions . Severe anemia in pregnancy results in relatively poor maternal and fetal outcome. Apparently
maternal risks increase prior to fetal risks. In order to improve maternal and fetal outcome, it is recommended that district
hospitals in low-income countries make prevention, early diagnosis, and treatment of severe anemia in pregnancy a priority.

Key words: Anemia in pregnancy, maternal outcome, fetal outcome, low-income countries

Abbreviations: Hb: hemoglobin

Anemia in pregnancy is a common problem in low-
income countries. It substantially contributes to
poor outcome in both mother and child. Its effects
on maternal health include reduction of immune
response, danger of heart failure, and aggravation of
the risks of childbirth (1,2). Maternal anemia during
pregnancy may lead to poor fetal outcome through
growth retardation or perinatal death, while the risks
of infant morbidity and mortality are also increased
(3
/7). However, it is not yet clearly known to what
extent these maternal and fetal risks are affected by
anemia in pregnancy in low-income countries. Simi-
larly, little knowledge is available on the relation
between pregnancy outcome and severity of anemia,
gestational age at onset or duration of anemia,
although anemia in early pregnancy seems to be
associated with an increased risk of low birth-weight
or preterm birth (8,9). This study analyzes the effect
of severe anemia in pregnancy (HbB/8.0 g/dl) on
maternal and fetal outcome in a cohort of women in
rural Ghana.
Methods
This cohort study was carried out at the district
hospital in Berekum and the subdistrict hospital in
Yeji, both in rural Ghana. From January 1, 1999 till
January 31, 2000, women who were treated for
severe anemia in pregnancy (HbB/8.0 g/dl) were
entered into the study, matched for age and parity

Correspondence: Jules H. Schagen van Leeuwen, Department of Obstetrics and Gynaecology, St. Antonius Hospital, PO Box 2500, 3430 EM Nieuwegein,
The Netherlands. E-mail: jsvleeuw@knmg.nl

(Received 25 April 2005; accepted 20 June 2005)

ISSN 0001-6349 print/ISSN 1600-0412 online # 2006 Taylor & Francis
DOI: 10.1080/00016340500334794
50 D. Geelhoed et al.
with nonexposed women (Hb ]/10.9 g/dl) who at-
tended the study hospitals for childbirth. This paper
deals with the results regarding maternal and fetal
outcome; the pregnancy component of the study as
well as study setting and population are described in
detail elsewhere (10).
Information concerning pregnancy outcome was
collected from hospital records. When an exposed
woman did not report at one of the hospitals for
childbirth, despite the advice to give birth in
hospital, she was traced and visited at home.
Information concerning pregnancy outcome was
gathered from the woman, her relatives or husband,
or from closely connected people in the same village.
In case of birth at a health facility other than the
study hospitals, that facility was visited to obtain
details from their records. Nonexposed women were,
due to practical reasons, entered in the study before
or after their childbirth in the two hospitals, but
always exclusively based upon their nonanemic
status during pregnancy, age, and parity.
History of present and any previous pregnancies,
mode of delivery, place of birth, and birth attendant
were recorded, as well as postpartum hemorrhage,
defined as a blood loss after birth ]/500 ml. Blood
loss was measured in hospitals or maternity clinics,
but estimated by comparison with examples of
known volumes as more or less than 500 ml in
home births. Maternal deaths, circumstances of
death, and their main cause were recorded. Fetal
outcome was recorded as live birth, stillbirth, or
neonatal death. Birth weight was recorded in grams
in hospitals and maternity clinics, and estimated by
comparison with other babies as above or below
2500 g in home births. Gestational age at birth was
not recorded, as reliable information on this was
generally lacking.
Data were entered in Epi-Info, version 6.04, and
validated. Differences in proportions between the
exposed and nonexposed groups were assessed by
x2-tests and odds ratios were calculated adjusted for
age. The study design made use of stratum matching
between exposed and nonexposed women for both
age and parity, but as these two variables were found
to be colinear, analysis was stratified for age only.
Differences in means were assessed with Students t -
test for normally distributed variables. A logistic
regression model (SPSS 10.0) was fitted to assess
which variables significantly contributed to the
variability in outcome of singleton babies of exposed
and nonexposed women. The model was fitted
through stepwise elimination of variables not sig-
nificantly contributing to the fit of the model, after
which interactions between the remaining variables
were tested in a similar manner. The significance
level was set at 5%.
Results
In the study period 175 women had been treated for
severe anemia in pregnancy, but follow-up of preg-
nancy outcome was complete for only 157 women
(89.7%). The nonexposed group, with 100% com-
plete follow-up, consisted of 152 women.
Median age among women in the exposed group
was 22 years (range 16
/45) and 24.8% were
younger than 20 years of age. Median parity was 0
(range 0
/9), with 57.3% nullipara. Among parous
women, 24.2% reported a previous cesarean section,
postpartum hemorrhage, or perinatal death. Ex-
posed women had been severely anemic with a
median lowest Hb of 6.5 g/dl. After treatment the
median Hb had increased to 9.5 g/dl when last
measured in pregnancy. The median age among
nonexposed women was 22 years (range 16
/42) and
17.8% were younger than 20 years of age. Median
parity was 0 (range 0
/8), with 57.2% nullipara.
Among parous women, 27.7% reported a previous
cesarean section, postpartum hemorrhage, or peri-
natal death. Median lowest Hb among nonexposed
women was 11.0 g/dl. Median Hb measured prior
to childbirth was 11.5 g/dl and 4.6% had a Hb ]/
14 g/dl. The study groups did not significantly differ
in age, parity, or poor obstetric history. The 18
women lost to follow-up were not significantly
different from the other exposed women considering
their age, parity, residence, antenatal care visits,
body mass index, or lowest hemoglobin. They did,
however, have a significantly lower hemoglobin level
when last measured (mean 8.2 g/dl, SD 1.7; p B/
0.05).
Maternal outcome
Of 157 women exposed to severe anemia, two
aborted before a gestational age of 28 weeks. Details
of childbirth were unknown for one woman,
although it was known that she did deliver a live
baby. One woman died while pregnant. Of the
remaining 153 women, 141 (92.2%) gave birth
vaginally (133 spontaneously, 8 assisted), while 12
(7.8%) were delivered by cesarean section. Of the
152 controls, 132 (86.8%) had a vaginal birth (123
spontaneously, 9 assisted) and 20 (13.2%) were
delivered by cesarean section. The mode of delivery
was not significantly different between the two
groups. Births were supervised by a trained atten-
dant (doctor, midwife) in 80.4% of women exposed
to severe anemia: 64.7% gave birth in hospital,
 Severe anemia in pregnancy in rural Ghana 51

15.9% in a clinic (including one birth attended by a Fetal outcome

trained traditional birth attendant), and 19.6% at
Of 157 women exposed to severe anemia, 154
home. Births of nonexposed women were, related to
women gave birth; 163 children were born (includ-
the study design, more often supervised: 98.0% gave
ing nine sets of twins), among whom seven were
birth in hospital, 1.3% in a clinic, and 0.7% at home.
stillbirths (4.3%) and 12 were neonatal deaths
In the exposed group, blood loss at childbirth was
(7.4%). In 152 nonexposed women 153 children
unknown in eight cases. Of these eight women, three
were born (including one set of twins), among whom
had cesarean section, none needed manual removal
four were stillbirths (2.6%), and two were neonatal
of placenta or transfusion of blood, and none died,
deaths (1.3%). The perinatal mortality of the 10 sets
though two of their children were perinatal deaths.
of twins was high, with five deaths (25.0%), all
Postpartum hemorrhage occurred after 16 vaginal
among women exposed to severe anemia. When
births (11.3%), of which seven were at home.
Removal of placenta was done manually in four birth was not supervised by a trained attendant,
cases (2.8%). At seven of the 153 births (4.6%) a perinatal mortality was also high, with eight deaths
blood transfusion was necessary, usually two units in 32 births (25.0%).
(range 1
/5). In the nonexposed group postpartum To assess age-adjusted OR for perinatal mortality,
hemorrhage occurred after 16 births (12.1%). Man- but avoid confounding by differences in supervision
ual removal of placenta and blood transfusion were of birth and multiple pregnancy between exposed
both necessary in one birth. Risks of postpartum and nonexposed women, OR was calculated for
hemorrhage and manual removal of placenta did not supervised singleton births only. Compared to non-
differ between the groups. Women exposed to severe exposed women, perinatal mortality of supervised
anemia, however, tended to have a higher need for singleton births was not significantly different among
blood transfusion at childbirth (age-adjusted OR the entire group of women exposed to severe anemia
7.5; 95%CI 0.9
/63.1). in pregnancy. However, perinatal mortality was
Maternal death occurred in five women, all among significantly higher among women who had been
women exposed to severe anemia (Table I). The case exposed to a hemoglobin level below 7.0 g/dl (OR
fatality rate was 3.2%. Risk factors for death other 3.1; 95%CI 1.0
/9.4; Table II). Perinatal mortality of
than exposure to severe anemia in pregnancy were supervised singleton births was also significantly
not identified. higher among those exposed women who remained
severely anemic at the end of their pregnancy (OR
Table I. Case histories of maternal deaths 9.2; 95%CI 1.7
/50.6), or who had a body mass
index B/20 kg/m2 (OR 11.8; 95%CI 2.9
/48.9;
G2P1, 25 years, with three antenatal visits and one week Table II). Teenage pregnancy was not significantly
admission at 34 weeks gestation. Lowest Hb was 7.0 g/dl, associated with perinatal mortality among exposed
recovered to 11.0 g/dl. She was not seen after discharge, and died
or nonexposed women.
of unknown causes in a prayer camp, still pregnant.

/ G2P1, 20 years, with three antenatal visits, lowest Hb 7.0 g/dl, In the exposure group, birth weight of ten children
recovered to 10.0 g/dl. Assisted vaginal hospital delivery after was unknown. Of the remaining 153 children of
prolonged trial of labor at home, perinatal death, birth weight women exposed to severe anemia, 39 (25.5%) had
3,290 g. No postpartum hemorrhage, but did need manual
removal of placenta. Cause of death: postpartum sepsis, despite Table II. Perinatal mortality in exposed and nonexposed groups
treatment with antibiotics.

/ G4P2, 24 years, no antenatal visits, admitted 2 weeks, lowest
Hb 6.0 g/dl, recovered to 10.5 g/dl. Cause of death: postpartum Supervised births of Perinatal (Age-adjusted)
hemorrhage, retained placenta after home birth, during transfer to singletons only death (%) OR (95%CI)
hospital.
Nonexposed 6/151 (4.0) 1

/ G1P0, 27 years, five antenatal visits, lowest Hb 6.5 g/dl,
Exposed, lowest Hb B/8.0 g/dl 9/117 (7.7) 2.2 (0.7
/6.3)
recovered to 8.5 g/dl. Positive sickle test, (weakly) reactive HIV-
Exposed, lowest Hb B/7.0 g/dl 8/78 (10.3) 3.1 (1.0
/9.4)
test. Spontaneous vaginal hospital birth, live child, birth weight
Exposed, lowest Hb B/6.0 g/dl 8/47 (17.0) 5.4 (1.7
/16.7)
2,570 g. Discharged in good condition. Cause of death: sickle cell
Exposed, lowest Hb B/5.0 g/dl 4/25 (16.0) 5.6 (1.4
/22.2)
crisis 1 month postpartum, despite hospital treatment.
Exposed, last Hb ]/10.9 g/dl 1/32 (3.1) 0.7 (0.1
/5.8)

/ G6P5, 33 years, three antenatal visits, admitted for antepartum
Exposed, last Hb 8.0
/10.8 g/dl 5/68 (7.4) 2.2 (0.6
/7.5)
hemorrhage. Previous history of uterine rupture and repair,
refused tubal ligation. Hb reduced from 11.0 g/dl to 6.0 g/dl, had Body mass index B/20 kg/m2, 3/30 (10.0) 4.1 (0.8
/21.6)
2 units of blood, recovered to 11.2 g/dl. Elective cesarean section nonexposed
for placenta previa, live child, birth weight 2,750 g. Cause of Body mass index ]/20 kg/m2, 1/79 (1.3) 0.5 (0.1
/5.2)
death: hypovolemic shock, coagulation disorder, despite 5 units of exposed
blood and emergency hysterectomy for persistent bleeding from Body mass index B/20 kg/m2, 8/38 (21.1) 11.8 (2.9
/48.9)
placenta site. exposed
52 D. Geelhoed et al.

low birth-weight (B/2,500 g). In the nonexposed Table III. Low birth-weight in exposed and nonexposed groups
group 21 (13.7%) babies were born with low birth-
weight. Mean birth-weight of singletons in the Singletons only Low birth (Age-adjusted)
weight (%) OR (95%CI)
exposed group was 2,802 g (SD 565), and in the
nonexposed group 2,996 g (SD 444), a difference of Nonexposed 20/151 (13.2) 1
194 g (p B/0.05). Birth weight of twins was low at a Exposed, lowest Hb B/8.0 g/dl 26/138 (18.8) 1.4 (0.7
/2.7)
mean of 2,120 g (SD 385); 14/17 (82.4%) twins Exposed, lowest Hb B/7.0 g/dl 17/90 (18.9) 1.4 (0.7
/2.8)
with a known birth weight weighed under 2,500 g. Exposed, lowest Hb B/6.0 g/dl 16/56 (28.6) 2.5 (1.2
/5.4)
Exposed, lowest Hb B/5.0 g/dl 13/30 (43.3) 4.2 (1.8
/10.1)
Supervision of birth was not significantly related to Exposed, last Hb ]/10.9 g/dl 3/34 (8.8) 0.8 (0.2
/2.7)
birth weight. Exposed, last Hb 8.0
/10.8 g/dl 13/80 (16.3) 1.1 (0.5
/2.4)
Confounding by differences in multiple pregnancy Exposed, last Hb B/8.0 g/dl 10/24 (41.7) 3.3 (1.3
/8.5)
between exposed and nonexposed women was Body mass index ]/20 kg/m2, 14/121 (11.6) 1
nonexposed
avoided by calculating age-adjusted OR of low
Body mass index B/20 kg/m2, 6/30 (20.0) 2.3 (0.8
/7.2)
birth-weight for singletons only. Compared to sin- nonexposed
gleton babies of nonexposed women, singleton Body mass index ]/20 kg/m2, 12/94 (12.8) 3.1 (0.5
/2.5)
babies of all women exposed to severe anemia were exposed
not significantly more likely to have low birth- Body mass index B/20 kg/m2, 14/44 (31.8) 3.1 (1.3
/7.6)
exposed
weight. However, when the lowest maternal hemo-
Age ]/20 years, nonexposed 11/121 (8.9) 1
globin had been below 6.0 g/dl, singleton babies of Age B/20 years, nonexposed 9/27 (33.3) 5.1 (1.7
/16.0)
exposed women had a significantly higher risk of low Age ]/20 years, exposed 17/104 (16.3) 2.0 (0.8
/4.9)
birth-weight (OR 2.5; 95%CI 1.2
/5.4; Table III). Age B/20 years, exposed 9/34 (26.5) 3.7 (1.2
/11.1)
Low birth-weight was also significantly more often
seen in singleton babies of women who remained
severely anemic at the end of their pregnancy (OR schistosomiasis, severe anemia, hemoglobin prior to
3.3; 95%CI 1.3
/8.5), or who had a body mass index childbirth, body mass index, number of antenatal
B/20 kg/m2 (OR 3.1; 95%CI 1.3
/7.6). Teenage visits, gestational age at first contact) and entered in
pregnancy was associated with low birth-weight in the model. After stepwise removal and testing of any
both exposed (OR 3.7; 95%CI 1.2
/11.1) and interactions, the number of antenatal visits, body
nonexposed groups (OR 5.1; 95%CI 1.7
/16.0). mass index, and hemoglobin prior to childbirth
To assess which combination of factors signifi- significantly contributed to the variability of the
cantly influenced fetal outcome, it was categorized as singletons outcome (at a 5% significance level).
good (live baby with a birth weight ]/2,500 g) or Together these variables explained approximately
adverse (perinatal death and/or birth weight 15% of the variability in outcome. Singleton babies
B/2,500 g). Possible explanatory variables were of women whose hemoglobin prior to childbirth was
identified. Multiple pregnancy was excluded, as at least 8.0 g/dl, were 3.9 times more likely to have a
this variable was found to be colinear with outcome, good outcome than babies of those who had a lower
and also would violate the assumption of indepen- hemoglobin at the end of pregnancy (mutually
dence of the binomial distribution. As the number of adjusted OR 3.9; 95%CI 1.6
/9.6). Singleton babies
twins was relatively small (20/316; 6.3%), the of women with a body mass index of at least 20 kg/
sensitivity of the study was not seriously affected. m2 were 2.8 times more likely to have a good
Malaria, intestinal parasitosis, supervision of birth, outcome than babies of women with a lower body
and lowest hemoglobin were all colinear with ex- mass index (mutually adjusted OR 2.8; 95%CI 1.5
/
posure to severe anemia in pregnancy (Hb B/8.0 g/ 5.3). Singleton babies of women who attended an
dl), and thus not used in the modeling. Sickle cell antenatal clinic at least four times during their
and HIV status, both untreatable in the setting of the pregnancy were 2.0 times more likely to have a
study, were combined into one variable, as these good outcome than babies of women who visited less
disorders each had a low prevalence in the study often (mutually adjusted OR 2.0; 95%CI 1.1
/3.7).
group. For the same reason urinary tract infection
and schistosomiasis, both treatable disorders, were
Discussion
combined into one variable. Number of antenatal
visits was dichotomized as B/4 visits or ]/4 visits, as The study results provide evidence that severe
a minimum of 4 antenatal visits is locally recom- anemia in pregnancy is associated with a poor
mended. Of the remaining variables, seven were maternal outcome. Although mode of delivery, risk
significantly related to the singletons outcome at a of postpartum hemorrhage, and risk of manual
10% significance level (age, urinary tract infection/ removal of placenta were similar in exposed and

nonexposed groups, women in the exposed group
tended more often to need a blood transfusion at
birth, compared to nonexposed women. The study
group (n /309) was rather small to assess differ-
ences in maternal mortality, and the number of
adverse maternal events was not adequate for further
statistical analysis. Even so, there was a significant
difference in the number of deaths among exposed
and nonexposed women. At least one maternal death
could have been avoided, had a blood transfusion
been available. Only one death was possibly directly
due to severe anemia in pregnancy, while four
women died of other causes. This shows the hidden
contribution of anemia in pregnancy to maternal
mortality (11). The case fatality rate among women
exposed to severe anemia was high at 3.2%. Re-
markably, although they had been exposed to low
hemoglobin levels (5/7.0 g/dl), all five women re-
covered quite well (Hb ]/8.5 g/dl). It appears that
severe anemia in pregnancy is, even after treatment,
an important risk factor for maternal death.
In this study severe anemia in pregnancy ( B/8.0 g/
dl) was not associated with poor fetal outcome.
Women exposed to Hb B/7.0 g/dl, however, had a
threefold increased risk of perinatal mortality, and
women exposed to Hb B/6.0 g/dl had a doubling of
the risk of low birth-weight. Women whose hemo-
globin level had not recovered to at least 8.0 g/dl
before childbirth also had increased risks of perinatal
mortality and low birth-weight. Low birth-weight is
an important determinant of infant mortality (6). In
addition, maternal anemia and low birth-weight
contribute to infant anemia, due to insufficient iron
storage in the prenatal period (5,12).
The relationship between hemoglobin concentra-
tion in pregnancy and fetal outcome follows a
U-shape, with an increasingly poor outcome at
both extremes. The association of poor fetal out-
come with high maternal hemoglobin levels, related
to (pre)-eclampsia, has been described in high
income countries (4,13). This relationship of out-
come and high hemoglobin was not confirmed by
our findings, possibly because of the low prevalence
of high hemoglobin levels.
Many studies have described the increased risk of
perinatal mortality (3,4,14
/16) and low birth-
weight (3
/5,7,12,17) associated with low maternal
hemoglobin concentration. Comparison is difficult,
however, due to different definitions of anemia.
It has not yet been clarified at what hemoglobin
level anemia in pregnancy starts to affect maternal
and fetal outcome, and how seriously outcome is
affected (18). Very few prospective studies concern-
ing maternal and fetal outcome after severe anemia
in pregnancy have been performed, and even fewer
Severe anemia in pregnancy in rural Ghana 53
in low-income countries. In our study exposure to
hemoglobin levels of 8.0 g/dl or less, despite sub-
sequent recovery, significantly increased maternal
risks. Fetal risks started to increase significantly with
exposure to hemoglobin levels below 7.0 g/dl (peri-
natal death) or below 6.0 g/dl (low birth weight).
Women who were not severely anemic at childbirth
(Hb]/8.0 g/dl) had a greater chance of giving birth
to a live baby of normal birth weight. Apparently
maternal risks increase prior to fetal risk, which
suggests that the definition of severe anemia as a
hemoglobin concentration below 7.0 g/dl, as given
by the WHO, may be more useful for assessing risks
to the child than to the mother. A cut-off point for
severe anemia of 8.0 g/dl has been suggested, based
upon seasonal variation in the prevalence of anemia
(B/8.0 g/dl) related to malaria and iron deficiency
(19). The findings of our study seem to support a
higher cut-off point, in particular in view of maternal
risks.
Exposure to severe anemia in pregnancy (Hb
B/8.0 g/dl) in itself did not significantly contribute
to the variability of fetal outcome in this analysis, but
rather recovery of hemoglobin levels to at least 8.0 g/
dl. Therefore, in addition to exposure to severe
anemia in pregnancy, its duration and gestational
age at onset could be important. This assumption is
supported by other findings, which detected an
association between low birth-weight and maternal
anemia diagnosed in the first trimester (8). It is
recommended that severe anemia in pregnancy is
swiftly treated, and hemoglobin levels corrected
before childbirth, to limit negative effects on fetal
outcome.
When women regularly attended antenatal care
their chance of good fetal outcome was increased.
Lack of a trained supervisor to attend birth was
associated with a very poor fetal outcome. Inade-
quate antenatal and birth care played a role in three
of the five maternal deaths in this study. These
results clearly point towards the importance of
competent care during pregnancy and childbirth.
Presently emergency obstetric care is emphasized as
the key for reduction of maternal and fetal risks. It
remains important, however, to appreciate that
nonemergency antenatal and birth care also form
part of essential obstetric services, as this provides
opportunities for prevention and early detection of
complications (20,21).
The maternity units of both study hospitals
generally attend to women with complications of
pregnancy or birth. Even the nonexposure group
consisted of hospital attenders at rather high risk,
indicated by the high percentage of women with a
poor obstetric history (27.7% of parous women) and
54 D. Geelhoed et al.
with cesarean sections (13.2%). Still, this group
performed significantly better than women exposed
to severe anemia in pregnancy, as far as maternal and
fetal outcomes were considered. To ensure an
economic recruitment of women who had not been
exposed to any degree of anemia during pregnancy,
they were selected near childbirth, and thus gener-
ally entered later in the study than exposed women.
Therefore no conclusions can be drawn about
differences in abortion rate. The late selection of
nonexposed women might have led to an uncon-
scious preference to include women with a good
pregnancy outcome. Among attenders of both study
hospitals, however, anemia in pregnancy is very
common, especially among young nulliparous wo-
men. It was quite difficult to find a sufficiently large
number of nonanemic women of suitable age and
parity to comprise an adequate nonexposure group.
This difficulty would have prevented any possible
preference for women with a good pregnancy out-
come.
Follow-up of exposed women was high in this
study at 89.7%, certainly in view of the relatively
poor infrastructure in the study districts. Among
exposed women with home births, however, the
information obtained is likely to have been less
reliable than in the case of health facility births.
Particularly when information had to be gathered
from people other than the woman herself, and
when it concerned nonfatal events such as birth
weight or blood loss, trustworthiness may have been
compromised. To remedy this problem extensive
double-checking was done with as many people as
closely concerned as possible, and in cases of
persistent lack of clarity the particular information
was recorded as unknown, and thus excluded from
the analysis.
No data were collected on economic status or
educational level of the women, and therefore the
results can unfortunately not account for any poten-
tial confounding by these variables. In view of the
size of the odds ratios it is, however, unlikely that
confounding would explain the entire association.
Nevertheless, insufficient access to care to facilitate
recovery from severe anemia in pregnancy or to have
skilled care at childbirth, as well as poor nutri-
tional status, are probably associated with poverty,
and it seems likely that most women with poor
maternal or fetal outcome suffer from socioeconomic
deprivation.
The association between pregnancy anemia and
outcome may be more complex than has often been
considered. More research is required to compre-
hend fully the relation between severe anemia in
pregnancy and maternal and fetal outcome in low-
income countries.
Conclusions
In view of poor maternal and fetal outcome after
severe anemia in pregnancy, it is recommended that
district hospitals in low-income countries make
prevention, early diagnosis and treatment of severe
anemia in pregnancy a priority. The current WHO
definition point for severe anemia in pregnancy
(HbB/7.0 g/dl) might be inadequate for this pur-
pose; a cut-off point of 8.0 g/dl could be more
appropriate. Adequate referral services for the man-
agement of emergency complications must be com-
plemented by nonemergency maternity care to
prevent and detect complications during pregnancy
and childbirth, in order to improve maternal and
fetal outcome.
References
1. Guidotti RJ. Anaemia in pregnancy in developing countries.
Br J Obstet Gynaecol. 2000;107:437
/8.
/ /
2. van den Broek N. Anaemia in pregnancy in sub-Saharan
countries. Eur J Obstet Gynecol Reprod Biol. 2001;96:4
/6. / /
3. Lone FW, Qureshi RN, Emanuel F. Maternal anaemia and its
impact on perinatal outcome. Trop Med Int Health. 2004;9: / /
486
/90.
4. Scholl TO, Hediger ML. Anemia and iron-deficiency anemia:
compilation of data on pregnancy outcome. Am J Clin Nutr.
1994;59:492S
/500S.
/ /
5. Verhoeff FH, Brabin BJ, Chimsuku L, Kazembe P, Broadhead
RL. Malaria in pregnancy and its consequences for the infant
in rural Malawi. Ann Trop Med Parasitol. 1999;93:S25
/33.
/ /
6. World Health Organization. Low birth weight: a tabulation of
available information. Geneva: World Health Organization,
WHO/MCH/92.2, 1992.
7. Zhou LM, Yang WW, Hua JZ, Deng CQ, Tao X, Stoltzfus RJ.
Relation of hemoglobin measured at different times in
pregnancy to preterm birth and low birth weight in Shanghai,
China. Am J Epidemiol. 1998;148:998
/1006.
/ /
8. Hamalainen H, Hakkarainen K, Heinonen S. Anaemia in the
first but not in the second or third trimester is a risk factor for
low birth weight. Clin Nutr. 2003;22:271
/5. / /
9. Xiong X, Buekens P, Alexander S, Demianczuk N, Wollast E.
Anemia during pregnancy and birth outcome: a meta-
analysis. Am J Perinatol. 2000;17:137
/46.
/ /
10. Geelhoed D, Agadzi F, Visser L, Ablordeppey E, Asare K,
ORourke P et al. Severe anemia in pregnancy in rural Ghana:
A case
/control study of causes and management. Acta Obstet
Gynecol Scand 2005 (in press).
11. van den Broek N. Anaemia in pregnancy in developing
countries. Br J Obstet Gynaecol. 1998;105:385
/90. / /
12. Rusia U, Madan N, Agarwal N, Sikka M, Sood SK. Effect of
maternal iron deficiency anemia on foetal outcome. Indian J
Pathol Microbiol. 1995;38:273
/9.
/ /
13. Steer P, Ash Alam M, Wadsworth J, Welch A. Relation
between maternal haemoglobin concentration and birth
weight in different ethnic groups. BMJ. 1995;310:489
/91. / /

14. Diallo MS, Diallo TS, Diallo FB, Diallo Y, Camara AY,
Onivogui G, et al. Anemie et grossesse. Rev Fr Gynecol
Obstet. 1995;90:138
/41.
/ /
15. Duthie SJ, Ven D, King PA, To WK, Lopes A, Ma HK. A case
controlled study of pregnancy complicated by severe maternal
anemia. Aus NZ J Obstet Gynaecol. 1991;31:125
/7.
/ /
16. Meda N, Dao Y, Toure B, Yameogo B, Cousens S, Graham
WJ. Evaluer lanemia maternelle severe et ses consequences:
la valeur dun simple examen de la coloration de conjunctives
palpebrales. Cah Sante. 1999;9:12
/7.
/ /
17. Brabin BJ, Ginny M, Sapau J, Galme K, Paino J. Con-
sequences of maternal anaemia on outcome of pregnancy in a
malaria endemic area in Papua New Guinea. Ann Trop Med
Parasitol. 1990;84:11
/24.
/ /
Severe anemia in pregnancy in rural Ghana 55
18. Rooney C. Antenatal care and maternal health: how effective
is it? Geneva: World Health Organization, WHO/MSM/92.4,
1992.
19. Verhoeff FH, Brabin BJ, Chimsuku L, Kazembe P, Broad-
head RL. An analysis of the determinants of anaemia in
pregnant women in rural Malawi
/ a basis for action. Ann
Trop Med Parasitol. 1999;93:119
/33.
/ /
20. Measham DM, Kallianes VD (eds) Issues in Essential
Obstetric Care. Report of a Technical Meeting of the Inter-
Agency Group for Safe Motherhood (1995). New York:
Population Council, 1996.
21. World Health Organization. Antenatal Care. Report of a
Technical Working Group (1994). Geneva: World Health
Organization, WHO/FRH/MSM/96.8, 1996.