Owusu-Agyei, S; Fryauff, DJ; Chandramohan, D; Koram, KA; Binka,
FN; Nkrumah, FK; Utz, GC; Hoffman, SL (2002) Characteristics of
severe anemia and its association with malaria in young children of
the Kassena-Nankana District of northern Ghana. The American
journal of tropical medicine and hygiene, 67 (4). pp. 371-7. ISSN
0002-9637
Downloaded from: http://researchonline.lshtm.ac.uk/16969/
DOI:
Usage Guidelines
Please refer to usage guidelines at http://researchonline.lshtm.ac.uk/policies.html or alterna-
tively contact researchonline@lshtm.ac.uk.
Available under license: Copyright the publishers
Am. J. Trop. Med. Hyg., 67(4), 2002, pp. 371–377
Copyright © 2002 by The American Society of Tropical Medicine and Hygiene
CHARACTERISTICS OF SEVERE ANEMIA AND ITS ASSOCIATION WITH
MALARIA IN YOUNG CHILDREN OF THE KASSENA-NANKANA DISTRICT OF
NORTHERN GHANA
SETH OWUSU-AGYEI, DAVID J. FRYAUFF, DANIEL CHANDRAMOHAN, KWADWOA. KORAM, FRED N. BINKA,
FRANCIS K. NKRUMAH, GREG C. UTZ, AND STEPHEN L. HOFFMAN
Navrongo Health Research Center, Ministry of Health, Navrongo, Ghana; Naval Medical Research Center, Silver Spring, Maryland;
London School of Hygiene and Tropical Medicine, London, United Kingdom; Noguchi Memorial Institute for Medical Research,
Accra, Ghana; School of Public Health, University of Ghana, Accra, Ghana; United States Naval Medical Research Unit No. 3,
Cairo, Egypt; Celera Genomics, Rockville, Maryland
Abstract. Severe anemia is thought to be the principal underlying cause of malaria death in areas of intense seasonal
malaria transmission such as the Kassena-Nankana District of northern Ghana. Factors associated with severe anemia
in young children, 6–24 months old, were elucidated by analyzing results of 2 malaria-associated anemia surveys (1996,
2000), separated by 4 years, but conducted in the same community and at the same seasonal time point. Age-adjusted
comparison confirmed that the proportion of severely anemic children and overall mean hemoglobin (Hb) levels in the
November 2000 sample were significantly improved over those of the 1996 sample (17.5 versus 26.4%, P 
 0.03; Hb 7.5
versus 6.9 g/dL, P 
 0.002). Weight-for-age Z-scores also indicated a significant improvement in the 2000 sample (−1.93
versus −2.20, P < 0.05). Independently, each survey identified statistically significant associations between severe anemia
and age, parasite rate, fever, and sex. Relative to children with Hb 
 6.0 g/dL, those with severe anemia (Hb < 6.0 g/dL)
were older, more frequently parasitemic (odds ratio [OR], 1.60; 95% confidence interval [CI], 1.08–2.35), more often
febrile (OR, 2.44; 95% CI, 1.71–3.48), and predominantly male (OR, 1.50; 95% CI, 1.05–2.13). An association was
identified in both surveys between severe anemia and residence in the northern part of the district, but no clear link was
observed in relation to irrigation. Blood transfusions, a likely surrogate index of severe anemia in young children,
followed a distinct seasonal pattern. Evidence suggests that dramatic peaks and troughs of severe anemia are regular and
possibly predictable events that may be used to gauge the health and survival of young children in this area.
INTRODUCTION this district by reexamining the previously collected data and
by expanding the database with newly acquired results. We
Severe anemia may be the principal underlying cause of
hypothesized that a repeat study of malaria-associated ane-
malaria death in areas of intense seasonal malaria transmis-
mia of the same age groups and locations as previously stud-
sion such as the holoendemic Kassena-Nankana District
ied, but at a time corresponding to the end of the 2000 wet
(KND) of northern Ghana.1,2 A recent study of malaria-
season, would again show large numbers of young children
associated anemia there of young children 6–24 months old
with severe anemia. Relative to the cohort with noncritical
revealed that 22% of those sampled at the end of the wet
Hb, we also hypothesized that severe anemia would be asso-
season (November 1996), a time corresponding to agricultural
ciated with parasitemia, higher parasite density, more febrile
and nutritional abundance, had hemoglobin (Hb) concentra-
illness, and residence proximal to perennial breeding sites of
tions < 6.0 g/dL. In contrast, a survey of the same age cohort
anopheline mosquitoes. Owing to natural decline, by 6
6 months later at the end of the dry season, and at a time of
months of maternally transferred protection,3 and the greater
dwindling food supply, found that only 1% of the children fell
vulnerability to infection that malnutrition, low birth weight,
into this category of severe anemia.2 With entomological in-
and stunting impart, we further expected that severe anemia
oculation rates in nonirrigated and irrigated sectors calculated
might be more prevalent at the lower end of the 6–24-month
to be, respectively, 72 and 800 infective bites per person-year
range studied and would be associated with higher rates of
(Binka FNB, unpublished data), and a clear pattern of ma-
1 parasitemia, greater parasite densities, more febrile illness,laria deaths that mirrored rainfall, it was reasoned that ane-
female sex, and low body weight for age.
mia trends in this vulnerable age group were primarily influ-
enced by the intensity of malaria transmission and that dra-
matic troughs and peaks of severe anemia are regular MATERIALS AND METHODS
seasonal events.
The findings presented by Koram and others2 suggest that Subjects and informed consent process. The study site and
naturally declining levels of malaria transmission, seen during population have been described in previously published re-
6 months of dry season, augmented only by their own pro- ports.1,2,4–7 As with previous studies, use was made of the
gressive malaria immunity, enable most young Ghanaian chil- Navrongo Demographic Surveillance System, a continually
dren to stabilize and improve their Hb status. This reversal updated database that records virtually all births, deaths, and
appears to occur without any recourse to improved diet, nu- movements in the district’s population of 140,000. An accu-
trient supplement, or anti-infective therapy. If this is the case, rate name, age, and home location listing was made from this
then proportions of young children with severe anemia, database of children aged 6–24 months on the first of Novem-
rather than proportions dying, might serve as a more conser- ber, 1996, and again in November 2000. Respective parents
vative and accessible measure of malaria vaccine effect. and community leaders were identified, contacted, and given
Toward this objective, we sought to elucidate factors asso- a detailed explanation of the study plan. Informed parents
ciated with or influencing severe anemia in young children of wishing to have their children tested for malaria and Hb gave
371
372 OWUSU-AGYEI AND OTHERS
their assent in writing or provided a thumbprint, then brought age, sex, blood type, malaria status, and Hb level, as well as
their infants to a central location for registration and testing. the identity and results of laboratory screening performed on
Sample collection and screening. Children were assigned a donors.
study number on the basis of their consecutive order of ap-
pearance, and they were given a brief physical examination. RESULTS
Axillary temperature was recorded and temperatures 

37.5°C were designated febrile. A sterile lancet was used to Full data set comparisons between times and cohorts. Pa-
prick the heel or toe sufficient to make thick and thin blood rental consent was provided nearly universally in both sur-
films on a clean labeled slide. Approximately 5 L additional veys. Severe anemia was significantly more prevalent at the
blood was obtained for Hb determination by means of the time of the 1996 survey than in 2000 (22 versus 12.5%; P <
Hemocue photometer (Leo Diagnostics, Helsinborg, Swe- 0.0002) (Table 1). The ages of children in both the Hb < 6.0
den). Hemoglobin readouts were transcribed into the child’s and Hb 
 6.0 cohorts of the 1996 survey were significantly
record. Parents were informed when their child’s Hb readout greater than those of their respective 2000 cohorts (P <
was < 6.0 g/dL, and treatment was initiated. On the bases of 0.0001), and mean Hb levels in these two 1996 cohorts were
physical condition and Hb level, children were either admin- significantly lower (P 
 0.005) than their 2000 counterparts.
istered daily treatment with orally administered ferrous sul- Moreover, significantly greater numbers of children overall
fate solution or typed in preparation for transfusion by a ma- (69.2 versus 57.1%, P < 0.001), as well as in the two Hb
laria-free family member. In accordance with national health cohorts, were parasitemic in the 1996 survey. Analysis of each
policy, orally administered chloroquine (25 mg/kg) was pro- sample population subdivided into severely anemic (Hb < 6.0
vided as first-line treatment for cases of uncomplicated clini- g/dL), moderately anemic (Hb 6.0–7.9 g/dL), and normal (Hb
cal malaria. Malaria slides were stained with Giemsa and ex- 
 8.0 g/dL) groups, revealed that severely anemic children in
amined via high-power ×1,000 oil immersion microscopy for both surveys were older (1996: 16.0 months versus 14.3
the presence of malaria parasites. Parasite species were iden- months, P 
 0.03; 2000: 12.6 months versus 11.4 months, P 

tified by morphology, and parasite density per microliter of 0.02) and contained fewer girls than normal Hb groups (1996:
blood was estimated from counts per 200 white blood cells 39.5 versus 53%, P 
 0.0002; 2000: 38.7 versus 54%, P 

and an assumption of 8,000 leukocytes/L. Three hundred 0.01). Relative to the normal Hb group, point prevalence of
thick-film fields were examined before assigning a negative parasitemia was significantly higher in both the moderately (P
malaria diagnosis. < 0.001) and severely anemic (P 
 0.05) children.
Blood transfusion records.Daily record of blood transfu- Age-stratified analysis. Figure 1, which plots the frequency
sions performed in the single district hospital during the 18- distribution of ages sampled and proportions of severely ane-
month period October 1999 to March 2001 were evaluated as mic children by age, shows uniformity in sampling over age
a complementary index of severe anemia in the district popu- groups in 1996, and no indication of declining rates of severe
lation. This activity recorded each recipient’s name, address, anemia in the 18–24 month age range. The 2000 survey, in
TABLE 1
Characteristics of severe anemia (hemoglobin [Hb] < 6.0) and noncritical anemia (Hb > 6.0) among 2 populations of Ghanaian children, 6–24
months old, surveyed November 1996–November 2000†
November 1996 November 2000
Characteristic Hb < 6.0 g/dL Hb 
 6.0 g/dL Hb < 6.0 g /dL Hb 
 6.0 g/dL
Number (%) 75 (22.0)** 266 87 (12.5)** 608
Sex
Male 45 (25.4) 132 51 (14.6) 299
Female 30 (18.3) 134 36 (10.4) 309
Age (mo) (mean ± SD) 16.0 ± 1.9** 15.0 ± 0.7** 12.8 ± 0.7** 12.0 ± 0.3**
Male 16.4 ± 1.5 15.3 ± 0.2 12.5 ± 1.0 12.1 ± 0.5
Female 15.4 ± 0.4 14.6 ± 0.9 13.8 ± 1.2 11.8 ± 0.5
Hemoglobin (g/dL) (mean ± SD) 4.8 ± 0.2** 7.8 ± 0.2** 5.2 ± 0.1** 8.2 ± 0.1**
Male 4.9 ± 0.3 7.8 ± 0.2 5.3 ± 0.2 8.1 ± 0.1
Female 4.8 ± 0.2 7.9 ± 0.2 5.1 ± 0.2 8.3 ± 0.1
Parasitemia, n (%) 55 (76.0)** 181 (68.0)** 49 (65.3)** 296 (55.9)**
Male 32/45 (71.1) 90/132 (68.2) 30/47 (63.8) 141/256 (55.1)
Female 23/30 (83.3) 91/134 (67.9) 19/28 (67.8) 155/273 (56.8)
GM parasitemia/L (95% CI) 1288 (860–1928) 1106 (842–1452) 1644 (1062–2547) 1312 (1066–1614)
Male 1248 (737–2115) 904 (614–1330) 2203 (1202–4036) 1306 (955–1977)
Female 1345 (709–2551) 1349 (920–1977) 1037 (601–1791) 1315 (984–1758)
Febrile, n (%) 16/76 (21.0)* 22/265 (8.3)* 18/74 (24.3)* 62/528 (11.7)*
Febrile and parasitemic, n (%) 12/55 (21.8)* 15/181 (8.3)* 12/48 (25.0)* 39/295 (13.2)*
Male 7/32 (21.9)* 7/90 (7.8)* 7/30 (23.3)* 20/141 (14.2)*
Female 5/23 (21.7)* 8/91 (8.8)* 5/18 (27.8)* 19/154 (12.3)*
Weight (kg) (mean ± SD) 7.9 ± 0.3 8.0 ± 0.2 7.7 ± 0.3 7.6 ± 0.1
Male 8.2 ± 0.5* 8.3 ± 0.3 8.1 ± 0.4* 7.9 ± 0.1*
Female 7.4 ± 0.4* 7.6 ± 0.3 7.1 ± 0.4* 7.3 ± 0.1*
* Statistically significant differences between groups or cohorts within a survey.
** Statistically significant differences between surveys for a group or cohort.
† 95% CI 
 95% confidence interval; GM 
 geometric mean parasitemia; Hb 
 hemoglobin.
SD 
 standard deviation.
CHARACTERISTICS OF SEVERE ANEMIA IN GHANAIAN INFANTS 373
times, analysis was reapplied, but only to children 12–18
months of age in the 2 survey populations. This age-adjusted
comparison confirmed that the proportion of severely anemic
children and overall mean Hb levels in the November 2000
sample were significantly improved over those of the 1996
sample (26.4 versus 17.5%, P 
 0.03; 6.9 versus 7.5 g/dL, P 

0.002). Differences between survey times for rates of parasit-
emia in children 12–18 months old (69.2 versus 57.1%) and
the geometric mean (GM) parasitemia of those with patent
infections were not statistically significant.
Comparisons between sexes. Despite uneven age-group
sampling, nearly equal numbers of boys and girls were
screened in the 2 surveys. Girls in both noncritical and se-
verely anemic cohorts weighed significantly less (P < 0.02)
than their male counterparts. Interestingly, boys accounted
for the majority of severely anemic children in both surveys,
a difference that attained statistical significance when each
survey population was subdivided into severely anemic, mod-
erately anemic, and normal cohorts or when the 2 study popu-
lations were combined (boys, 18.2% versus girls, 12.9%, P 

0.02). Figure 2, a paired frequency histogram of this combined
population comparing proportions by age of severely anemic
boys and girls, shows similar profiles and the suggestion of
age relatedness; lowest proportions of severe anemia in boys
and girls were measured in the younger, < 12-month age
groups. A positive correlation between age and proportion
with Hb < 6.0 g/dL in the combined 1996 and 2000 population
of 6–19-month-olds attests to a stronger relationship in girls
F 1. Comparative frequency distributions of ages sampled (boys, r
2 
 0.04; girls, r2 
 0.53).
IGURE
and the proportions of severely anemic (hemoglobin < 6.0 g/dL) chil- Effect of irrigation and sector of residence.Analysis of the
dren detected by age group in 2 community-wide surveys of malaria- 1996 and 2000 survey populations was stratified according to
associated anemia (a) November 1996 and (b) November 2000. whether children lived in irrigated or nonirrigated communi-
ties of the KND. Children residing in irrigated communities
accounted for 57% of the 1996 survey population but com-
contrast, preferentially sampled the 6–11-month age range. prised only 29% of the 2000 sample (P < 0.0001). Unexpect-
Children aged 20–24 months, who comprised 27% of the 1996 edly, no statistically significant differences were seen between
sample and 29% of its severely anemic cohort, comprised survey years in malaria point prevalence (66.2 versus 62.2%),
only 3% of the 2000 survey. To determine whether this bias frequency of severe anemia (19.9 versus 10.9%, P 
 0.11),
may have been responsible for differences between survey mean Hb levels (7.3 versus 7.7 g/dL), or GM parasitemia (891
FIGURE 2. Paired frequency histogram of the combined 1996 and 2000 survey populations comparing proportions by age of severely anemic
(hemoglobin < 6.0 g/dL) boys and girls 6–19 months old.
374 OWUSU-AGYEI AND OTHERS
FIGURE 3. Point prevalence of severe anemia (hemoglobin < 6.0 g/dL) in young children according to geographic sector of residence within
the Kassena-Nankana District, northern Ghana: comparison of 1996 and 2000 survey results.
versus 1,349/L) of children living in irrigated communities. appreciably among zones, but in both surveys, Hb levels for
Significant differences were seen in comparing between 1996 the larger cohorts of noncritical (Hb 
 6.0 g/dL) children
and 2000 “nonirrigated” cohorts of children; the expected followed the same zonal pattern seen with the severely ane-
higher frequencies in 1996 of parasitemia (72 versus 54.5%, P mic cohorts (Figure 4).

 0.0003) and severe anemia (25.3 versus 13.8%, P 
 0.05), Blood transfusions as plausible marker of severe anemia.
and the overall mean Hb level (7.0 versus 7.8 g/dL, P < 0.05). Figure 5, which plots the monthly numbers of transfusions in
Analysis between the 1996 irrigated and nonirrigated cohorts young children (age < 60 months) and rainfall, shows that
revealed a significantly lower mean Hb level among severely most transfusions were to infants < 25 months old (381 of 485,
anemic children of the irrigated cohort (4.6 versus 5.0 g/dL, P 78.6%), and there was a distinctly seasonal pattern in the

 0.03) and significantly greater GM parasitemia among frequency of occurrence. The monthly transfusion profile
those of the nonirrigated cohort (1,549 versus 891/L, P 
 during 2000 for children  2 years old appears to follow the
0.02). Measured characteristics of children in the November very sharply demarcated period of rains, with transfusions
2000 irrigated and nonirrigated cohorts were comparable. peaking 1 month after the month of greatest rainfall. As with
Analysis by sector was performed to determine whether severe anemia, boys accounted for the majority of young chil-
severe anemia and malaria infection in these children might dren receiving transfusions (57.5%); the frequency of trans-
be associated with residence in a particular geographic zone fusion for young girls was significantly lower than that ex-
within the district (north, south, east, west, central). Consis- pected under conditions of equality (50 versus 42.5%, P 

tent differences were seen between sectors in the prevalence 0.01). There was no difference between sexes in the age and
of severe anemia, with highest levels in both surveys associ- Hb levels of those receiving blood. Pairwise comparison by
ated with residence in the north (Figure 3). Differences be- age between boys and girls (Figure 6) shows that the majority
tween zones in the prevalence of severe anemia were most of transfusion recipients in both sexes were 5–12 months old
pronounced and statistically significant in the 2000 survey. and that relatively few children > 36 months old underwent
Mean Hb levels of severely anemic children did not differ transfusion.
FIGURE 4. Mean hemoglobin levels in young, noncritically anemic (hemoglobin 
 6.0 g/dL) children according to geographic sector of
residence within the Kassena-Nankana District, northern Ghana: comparison of 1996 and 2000 survey results.
CHARACTERISTICS OF SEVERE ANEMIA IN GHANAIAN INFANTS 375
FIGURE 5. Monthly distribution of transfusions in children (0–5 years old) during October 1999–March 2001, separated by age  24 months
(shaded bars) and 25–60 months (open bars).
DISCUSSION diminished from 18 to 24 months, and beyond, but the 2000
population was not structured to sample these age groups and
Two malaria-associated anemia surveys, separated by 4 focused preferentially on children aged < 12 months, in whom
years but conducted in the same community and at the same severe anemia rates had been lower. The 1996 population had
November time point corresponding to early dry season in the also been balanced in terms of children residing in irrigated
northern Ghana sahel, both document alarmingly high rates and nonirrigated communities, whereas the 2000 population
and levels of anemia in young children 6–24 months old. In- sampled predominantly children from nonirrigated com-
dependently, each survey identified associations between se-
pounds. However, even after correcting for age bias, differ-
vere anemia and age, parasite rate, fever, and sex. Relative to
children with Hb 6.0 g/dL, those with severe anemia were ences remained highly significant, suggesting that rates of se-

older, more frequently parasitemic, more often febrile, and vere anemia, moderate anemia, and parasitemia may have
predominantly male. Our initial hypotheses that severe ane- improved from 1996. Monthly rainfall profiles for the KND
mia was associated with higher parasitemias was not borne show remarkable similarity over 20 years of recording, sug-
out by analysis, nor was evidence obtained to suggest that gesting no aberrant pattern or volume as reasons for reduced
severe anemia was associated with being younger, lower malaria and anemia in 2000. In fact, total rainfall in 2000
weight, or female. exceeded that of 1996 and closely followed the classic profile
Sampling bias was initially considered to be the major cause of the 20-year average for the district. In this respect, the
underlying differences between our studies. There was evi- quality of rainfall, rather than its quantity, may have been
dence from the 1996 study that severe anemia continued un- decisive, because unusually heavy, infrequent rains may have
disturbed anopheline populations in the KND far more than
light, frequent rains. In addition, it is possible that widespread
community participation in malaria research, such as the large
vitamin A supplement trial4 and bed net studies,1,5–7 may
have stimulated greater parental awareness and practice by
November 2000. The 1996 Ghanaian Ministry of Health
policy of free medical care and medicines for pregnant
women and children < 5 years old also may have reduced
malaria morbidity and mortality in children of the KND by
November 2000.
Twenty years of rainfall records show consistently that the
July–August–September quarter is the high point of the KND
wet season each year. Detailed studies in the KND have iden-
tified a corresponding pattern of child mortality in which rates
increased 4-fold from April, the last month of dry season to
August, the month of greatest rainfall. Rates of febrile illness
and parasitemia, GM parasite densities, and mean Hb levels
were similarly correlated with rainfall; infants 6–24 months
old were most affected.1 However, the regularity of the rain-
FIGURE 6. Pairwise comparison by age between boys and girls <
5 years old receiving at least one blood transfusion, for any reason, in fall pattern does not translate over to an invariable pattern of
the Kassena-Nankana District Hospital during October 1999–March malaria morbidity and death. Young children studied over a
2001. 3-year period in the KND experienced the highest death rates
376 OWUSU-AGYEI AND OTHERS
in the first 2 years (1992 and 1993) during the July–September Although > 98% of our surveyed children fit the clinical
quarter. By contrast, the highest infant death rate during 1994 definition of anemia (Hb < 11.0 g/dL) and 12.9% (75 of 581)
occurred in the early dry season quarter of October– of the parasitemias exceeded 10,000/L, only 0.6% (6 of 945)
December.5 Blood transfusions performed on young children fit the World Health Organization classification of severe ma-
during 2 consecutive years appear to follow a pattern closely laria-associated anemia.12 This low prevalence of severe ma-
dictated by rainfall and malaria transmission. However, de- laria-associated anemia is in contrast with rates of 5.2% re-
spite very comparable rainfall profiles in the 2 years, Novem- ported under conditions of fluctuating malaria in Malawi13
ber stood out in 1999, as opposed to September in 2000, as the and rates > 30% from areas of intense perennial transmis-
peak month for transfusions in children. We concede that sion.14,15 Notably, however, these high rates refer to predomi-
other factors may account for a portion of the transfusions nantly male hospital admissions, whereas our rate is derived
performed each month, but we believe that malaria-induced from a random cross-section of households with equal com-
anemia is the dominant factor underlying transfusions in chil- position of boys and girls.
dren. We further speculate that longitudinal profiling of in- The unfortunate fact of household-level sex bias, which
fant deaths as well as rates of severe anemia would closely exists over much of the malarious world, is assumed to ac-
mirror the transfusion profile each year. count for the reduced weight, health, and survival of infant
We obtained no evidence of positive or negative correla- girls and for the predominance of male hospital admissions
tion between GM parasitemia and age, weight, or Hb in either for all causes.13,16 For this reason, hospital-based studies are
individual or combined survey populations, but we did iden- unsuited and unable to detect important natural differences
tify a weak negative correlation between age and Hb value between boys and girls. Following this rationale, we hypoth-
and a statistically significant positive correlation between age esized a greater proportion and severity of anemia among
and proportion of children with Hb < 6.0 g/dL. Previous study infant girls and are impressed by random survey data indicat-
of children 0–5 years old at this location had reported a sig- ing just the opposite. We speculate that the reduced nutri-
nificant negative correlation between GM parasite density tional state of girls renders them less capable or less suitable
and Hb level.5 This relationship was not apparent in our sur- hosts for parasite superinfection and density-related pathol-
veys and may have resulted from focusing on younger chil- ogy of these infections. The high nutritional demand that ac-
dren of 6–24 months. Longitudinal study of Kenyan infants companies the faster growth of boys may disproportionately
from birth also determined an association between concur- contribute to, or even worsen, their malaria-associated ane-
rent parasitemia and lower mean Hb but required multiple mic condition over that of girls. Such protective mechanisms
slide readings over 90 days to show that significantly lower may compensate in a small way for the greatly increased risks
mean Hb levels were related to significantly higher mean of death that girls will later face from malaria in preg-
parasite densities.8 nancy.17,18
We hypothesized that residence in an irrigated area would In summary, severe anemia in the northern Ghana sahel
be a risk factor for malaria and severe anemia as a result of appears to be a regular and near-predictable seasonal event
increased density of infective mosquitoes and perennial trans- that is akin to a grim rite of passage for young children of
mission, but no clear pattern of increased risk was apparent in rural Africa. Under starkly seasonal conditions that rigidly
our analysis. This result is suggestive of a uniform condition of delimit malaria transmission, we identified high rates of se-
malaria transmission over the entire KND, virtually equaliz- vere anemia among boys and older children in the infant age
ing irrigated and nonirrigated communities for a time—and group studied and obtained evidence associating anemia se-
hence our failure to detect any November differences be- verity with residence in the northern sector of the KND. We
tween irrigated and nonirrigated cohorts. Alternatively, ma- believe that the daily record of blood transfusions performed
laria transmission affecting young children in irrigated sectors at the district hospital is a simple, valuable index that reflects
may be so much more intense and regular as to induce an the intensity of malaria, malaria-associated anemia, and ma-
earlier, broader, and stronger immunity than that triggered laria death in this district. From recent demonstrations of
only periodically by seasonal malaria in the nonirrigated sec- success against the high level of severe malaria-associated
tors. Such an advantage in the irrigated cohort might origi- anemia that occurs under conditions of continuous heavy
nate from immediately protective higher titer maternal anti- transmission,19–21 we advocate a similar attack against infant
bodies transferred during gestation and breast-feeding, fetal anemia as it occurs in the KND. This strategy, based on pro-
Hb, and subsequently from self-made cellular and humoral vision of orally administered iron syrup to mothers and simple
responses to a continuum of infections.9–11 Although we presumptive treatment of young children with sulfadoxine-
failed to detect clear differences between irrigated and non- pyrimethamine therapy, both timed in accordance with child-
irrigated cohorts in either survey, we observed that anemia hood immunizations and given through the existing infra-
was more pronounced among children of the northern sector. structure of the World Health Organization Expanded Pro-
Because this was seen in both surveys and manifested in mul- gram of Immunization, could save and improve many young
tiple factors (proportions of severely anemic children, mean lives.
Hb levels among noncritical children), we consider these dif-
ferences to be valid and suggestive of increased risk. Greater Acknowledgments: We thank the chiefs and people of the Kassena-
risk of severe anemia in the north may derive from 1) distance Nankana District for their cooperation in making this study possible.
from medical treatment facility, 2) less effective or delayed Sincere appreciation also goes to the Navrongo Health Research
protective immunity, or 3) nutritional and economic stress. Centre (NHRC) Teams who carried out the fieldwork, especially Dr.
On the bases of these initial observations, we hypothesize that Kwame Amponsa-Achiano, Nathan Mensah, and Timothy Awine.We also thank Dr. George Mumuni, senior medical officer in charge
differences between locations would be even more apparent of the Navrongo War Memorial District Hospital, and Philip Amati
during the truly dry months of December to March. of the hospital laboratory for daily recording of transfusion data. We
CHARACTERISTICS OF SEVERE ANEMIA IN GHANAIAN INFANTS 377
thank Dr. Abraham Hodgson, the acting NHRC director, for his role of permethrin-impregnated bednets on child mortality in rural
in facilitating this work and NHRC staff members Kalifa Bugri and northern Ghana. Am J Trop Med Hyg 59: 80–85.
Charles Attiogbe for malaria microscopy. 8. McElroy PD, Ter Kuile FO, Lal AA, Bloland PB, Hawley WA,
Financial support: The work was supported by the Navrongo Health Oloo AJ, Monto AS, Meshnick SR, Nahlen BL, 2000. Effect of
Research Centre, the Noguchi Memorial Institute of Medical Re- Plasmodium falciparum parasitemia density on hemoglobin
search, U.S. Naval Medical Research Center Independent Research concentrations among full-term, normal birth weight children
Initiative funds, and the STO F6.1 work unit 61102AA0101.BXF.1431 in western Kenya, IV. The Asembo Bay Cohort Project. Am J
of the Military Infectious Disease Research Program. Trop Med Hyg 62: 504–512.
9. McGregor IA, 1965. The passive transfer of human immunity.
Disclaimer: The views of the authors expressed herein do not purport Am J Trop Med Hyg 13: 237–239.
to reflect those of the Ghanaian Ministry of Health, the U.S. Navy, or 10. Pasvol G, Weatherall DJ, Wilson RJM, 1977. Effects of foetal
the U.S. Department of Defense. This research was approved by hemoglobin on susceptibility of red cells to Plasmodium falci-
scientific and ethical review boards of the Ghanaian Ministry of parum. Nature 270: 171–173.
Health and the U.S. Navy and was conducted in accordance with 11. Snow RW, Nahlen B, Palmer A, Donnelly CA, Gupta S, Marsh
regulations governing the protection of human subjects in medical K, 1998. Risk of severe malaria among African infants: direct
research. evidence of clinical protection during early infancy. J Infect Dis
Authors’ addresses: Seth Owusu-Agyei, Navrongo Health Research 177: 819–822.
Center, Ministry of Health, Navrongo, Upper East Region, Ghana. 12. Warrell DA, Molyneux ME, Beales PF, 1990. Severe and com-
David J. Fryauff, Naval Medical Research Center, 503 Robert Grant plicated malaria. Trans R Soc Trop Med Hyg 84 (Suppl 2):
Ave., Silver Spring, MD 20910-7500. Daniel Chandramohan, London 1–65.
School of Hygiene and Tropical Medicine, Keppel St., LondonWCIE 13. Slutsker L, Taylor TE, Wirima JJ, Steketee RW, 1994. In-hospital
7HT, United Kingdom. Kwadwo A. Koram and Francis K. Nkrumah, morbidity and mortality due to malaria-associated severe
Noguchi Memorial Institute for Medical Research, University of anaemia in two areas of Malawi with different patterns of ma-
Ghana, P.O. Box LG581, Legon, Accra, Ghana. Fred N. Binka, laria infection. Trans R Soc Trop Med Hyg 88: 548–551.
School of Public Health, University of Ghana, Accra, Ghana. Greg C. 14. Lackritz EM, Campbell CC, Ruebush TK, Hightower AW,
Utz, United States Naval Medical Research Unit No. 3, PSC 452, Box Wakube W, Steketee RW, Were JB, 1992. Effect of blood
102, FPO AE 09835-007. Stephen L. Hoffman, Celera Genomics, transfusion on survival among children in a Kenyan hospital.
Rockville, MD. Lancet 340: 524–528.
15. Dorward JA, Knowles JK, Dorward IM, 1990. Treatment of se-
Reprint requests: David J. Fryauff, Naval Medical Research Center, vere anaemia in children in a rural hospital. Trop Doctor 19:
503 Robert Grant Avenue, Silver Spring, MD 20910-7500, Telephone: 155–158.
301-319-7587, Fax: 301-319-7545, E-mail: fryauffd@nmrc.navy.mil. 16. Schellenberg D, Menendez C, Kahigwa E, Font F, Galindo C,
Acosta C, Schellenberg JA, Aponte JJ, Kimario J, Urassa H,
Mshinda H, Tanner M, Alonso P, 1999. African children with
REFERENCES malaria in an area of intense Plasmodium falciparum transmis-
sion: features on admission to the hospital and risk factors for
1. Binka FN, Morris SS, Ross DA, Arthur P, Aryeetey ME, 1994. death. Am J Trop Med Hyg 61: 431–438.
Patterns of malaria morbidity and mortality in children in 17. Fleming AF, 1987. The aetiology of severe anaemia in pregnancy
northern Ghana. Trans R Soc Trop Med Hyg 88: 381–385. in Ndola, Zambia. Ann Trop Med Parasitol 83: 37–49.
2. Koram KA, Owusu-Agyei S, Utz G, Binka FN, Baird JK, Hoff- 18. Matteelli A, Donato F, Shein A, Muchi JA, Leopardi O, Astori
man SL, Nkrumah FK, 2000. Severe anemia in young children L, Carosi G, 1994. Malaria and anaemia in pregnant women in
after high and low malaria transmission seasons in the Kas- urban Zanzibar, Tanzania. Ann Trop Med Parasitol 88: 475–
sena-Nankana District of northern Ghana. Am J Trop Med 483.
Hyg 62: 670–674. 19. Menendez C, Kahigwa E, Hirt R, Vounatsou P, Aponte JA, Font
3. Akanmori BD, Afari EA, Sakatoku H, Nkrumah FK, 1995. A F, Acosta CJ, Schellenberg DM, Galindo CM, Kimarlo J,
longitudinal study of malaria infant morbidity, and antibody Urassa H, Brabin B, Smith TA, Kitua AY, Tanner M, Alonso
titres in infants of a rural community in Ghana. Trans R Soc PL, 1997. Randomized placebo-controlled trial of iron supple-
Trop Med Hyg 89: 256–257. mentation and malaria chemoprophylaxis for prevention of
4. Ghana VAST Study Team, 1993. Vitamin A supplementation in severe anaemia and malaria in Tanzanian infants. Lancet 350:
northern Ghana: effects on clinic attendances, hospital admis- 844–849.
sions and child mortality. Lancet 342: 7–12. 20. Tomashek KM, Woodruff BA, Gotway CA, Bloland P, Mbaruku
5. Binka FN, Kubaje A, Adjuik M, Williams LA, Lengeler C, G, 2001. Randomized intervention study comparing several
Maude GH, Armah GE, Kajihara B, Adiamah JH, Smith PG, regimens for the treatment of moderate anemia among refugee
1996. Impact of permethrin-impregnated bednets on child children in Kigoma region, Tanzania. Am J Trop Med Hyg 64:
mortality in Kassena-Nankana district, Ghana: a randomized 164–171.
controlled trial. Trop Med Int Health 1: 147–154. 21. Schellenberg D, Menendez C, Kahigwa E, Aponte J, Vidal J,
6. Binka FN, Adongo P, 1997. Acceptability and use of insecticide- Tanner M, Mshinda M, Alonso P, 2001. Intermittent treatment
impregnated bednets in northern Ghana. Trop Med Int Health for malaria and anaemia control at time of routine vaccinations
2: 499–507. in Tanzanian infants: a randomized, placebo-controlled trial.
7. Binka FN, Indone F, Smith T, 1998. Impact of spatial distribution Lancet 357: 1471–1477.