Acta Tropica, 54(1993)105-115 105

© 1993 Elsevier Science Publishers B.V. All rights reserved 0001-706X/93/$06.00

ACTROP 00297

Classification of clinical falciparum malaria and its

use for the evaluation of chemosuppression in
children under six years of age in Liberia, West
Africa

Birthe Hogh a'b, Nuahn T. Marbiah a, Eskild Petersen "'d, Eugene

Dolopaye a, Michael Willcox c, Anders Bj6rkman d, Aloysius P.
Hanson a and Adam Gottschau e
aLiberian Institutefor Biomedical Research, Liberia, bDepartment of lmmunology, University of Stockholm,
Stockholm, Sweden, ~Department of Clinical Bacteriology, Giivle Hospital, Sweden, dDepartment of
Infectious Diseases, Karolinska Institute, Stockholm, Sweden and eDepartment of Biostatistics, Statens
Seruminstitut, Copenhagen, Denmark
(Received l0 February 1993; accepted 31 March 1993)

The possible role of malaria as cause of morbidity was assessed during one year in 262 children aged 6
months to 6 years living in two villages in a rural area of Liberia. The study population was followed by
weekly clinics and three-monthly surveys and the children were randomly allocated to receive either
chloroquine or placebo every 3 weeks. The morbidity of the children was evaluated by criteria based on
the history and the clinical condition into four different stages, in order to describe the probability that an
observed clinical event could be attributed to malaria infection, based on the presence of detectable
parasites in the blood, the history the previous week, and the clinical status of the child. The level of
anaemia, splenomegaly and measured body temperature supported that malaria was the major contributor
to the overall morbidity observed. Based on the stage classification of clinical illness, children were
classified as having 'possible clinical malaria' or 'probable clinical malaria'. Malaria appeared to be an
important cause of febrile episodes during both dry and rainy seasons. During the rainy season more than
60% of the children experienced at least one clinical malaria episode, and during the dry season more than
50% of the children experienced at least one clinical attack of malaria. Children receiving chemosuppres-
sion had overall fewer clinical malaria attacks, and the effect of the chemosuppression was most pro-
nounced in the dry season, the odds ratio comparing children receiving regular chemosuppression with
children receiving presumptive treatment only was estimated to 0.39 (0.25-0.62).

Key words: Plasmodiumfalciparum; Malaria; Clinical malaria; Children; Chemosuppression

Introduction

T h e r e a r e n o u n i v e r s a l i n d i c a t o r s o f m a l a r i a m o r b i d i t y in s e m i - i m m u n e i n d i v i d u a l s .
I n a m a l a r i a e n d e m i c a r e a a p r e s u m p t i v e d i a g n o s i s o f m a l a r i a is o f t e n m a d e in a n y
i n f a n t o r y o u n g c h i l d , w h o h a s a f e b r i l e illness. E v e n w h e n t h e d i a g n o s t i c c r i t e r i a is

Correspondence to." B. Hogh, Laboratory of Parasitology, Statens Seruminstitut, Artillerivej 5, DK-2300

Copenhagen S, Denmark. Phone +45 32 68 37 22; Fax +45 32 68 38 71.
106

based on the history of symptoms, careful clinical examination of the patient to

exclude non-malarial aetiology, and competent examination of the blood to detect
parasitemia, the presence of parasites in the peripheral blood of semi-immune
individuals does not necessarily mean that the illness manifested by the patient is
due to a malarial infection (Greenwood, 1988; Gilles, 1988).
Even the severe and complicated cases of malaria can be difficult to diagnose, as
there are no clinical signs or symptoms which are specific for malaria. For hospitali-
sed patients a definition of severe and complicated falciparum malaria and the
differences between severe malaria in adults and children has been comprehensively
considered by the W H O (Warrell et al., 1990).
Reports on malaria morbidity have suggested different pathogenic or pyrogenic
thresholds of Plasmodiumfalciparum parasites ranging from 1000-15 000 parasites
la1-1 (Trape et al., 1985; Baudon et al., 1986; Benassani et al., 1987; Snow et al.,
1988; Menon et al., 1990; Lyiomo et al., 1991; Velema et al., 1991). The pyrogenic
threshold will, however, vary according to the degree of immunity of the population,
and reflect the age of the host and the transmission characteristics in the area, and
it is therefore difficult to compare data on malaria morbidity between different areas.
In endemic areas, most cases of malaria are diagnosed on clinical grounds without
laboratory confirmation of parasitemia. A clinical case definition of malaria based
on fever or a history of fever and the condition of the child is often the only
information available to the health worker to decide whether treatment with an anti-
malarial drug is needed. Recognising that no single definition will be satisfactory or
relevant in all clinical situations, we describe the likelihood that an observed clinical
event can be attributed to malaria infection, based on the presence of detectable
parasites in the blood, the history of fever the previous week and the clinical status
of the child. We use the level of anaemia, splenomegaly and measured body temper-
ature to support our stage classification.
The stage classification was used to measure the burden of clinical malaria episodes
and the impact of 3 weekly chemosuppression with chloroquine.

Material and Methods

Study area
The study was performed in two villages in a rural area of the north-western part
of Liberia, in the Mount Nimba region close to the borders of Guinea and Ivory
Coast. The climate is tropical with a rainy season from May to October and a dry
season from November to April. The annual precipitation varies between 2000 mm
and 2500 mm and the temperature is generally between 21°C and 32°C. Malaria
transmission is perennial, but is most intense during the rainy season. No health
care services were available in the study villages and the nearest drug store was
situated 30 km from the villages.

Study population
A census was performed and an open cohort of children aged 6 months to 6 years
in the study villages (Beytonwee n = 121, Bonah n-- 141) were included after informed
consent and followed by weekly clinics and three-monthly surveys from May 1987
to April 1988. The sex ratio male/female was 1.06.
 107

Parasitological examination
Every three-months and whenever the child was unwell and/or had a temperature
>37.5°C thick and thin blood films were made from finger-prick blood specimens
and stained with Giemsa. Parasite density was calculated assuming 200 fields equal
to 0.2 lal blood.

Spleen examination
Palpation of the spleen was performed for children below 2 years of age with the
child lying down, and for children above 2 years of age the palpation was performed
with the child standing.

Packed cell volume

Every 3 months blood was collected in two 75 Ixl capillary tubes. The packed red
cell volume was measured in percent of whole blood, recording the mean of the two
readings.

Haemoglobin typing
Sickle cell trait (HbAS) was identified by electrophoresis and 13-thalassaemia (13T) was
detected by an elevated HbA2 level as previously described Willcox et al. (1983).

Entomological investigations
Every second month, pyrethrum spray collection of indoor-resting mosquitoes in
10% of the houses was performed. The person-biting rate was calculated by dividing
the number of indoor-resting, human blood-fed mosquitoes by the number of humans
per room. The mosquitoes were examined for sporozoites by ELISA using monoclo-
nal antibodies against P. falciparum sporozoites. The entomological inoculation rate
being the number of bites per person multiplied by the fraction of infectious
mosquitoes.

Morbidity measurements
The cohort of children was followed by passive case detection during weekly clinics
and active case detection during three-monthly morbidity surveys by the research
team including a medical doctor. Both at the weekly clinics (3871 encounters) and
at the three-monthly surveys (730 encounters) a standardised clinical assessment was
made by the medical doctor based on the history obtained from the child's attendant,
usually the mother, by a registered nurse from the local tribe who asked specific
questions according to the questionnaire.
The mother or attendant was asked whether the child was (a) well, (b) unwell but
able to continue normal activities, or (c) unwell and unable to continue normal
activities; and whether the child had had fever, had been vomiting, had had diarrhoea
or cough during the previous week. At each visit the temperature was measured by
an axillary digital clinical thermometer (Terumo*).
Thick and thin blood films were made from finger-prick blood specimens whenever
the child was unwell and/or had a temperature > 37.5°C. Children found positive
were treated with chloroquine 25 mg/kg body weight.
108

Chemosuppression
262 children were randomly allocated in the proportion 3:2 to receive chemosuppres-
sion with chloroquine, 5 mg/kg body weight (group 1, n = 158) or placebo (group 0,
n = 104) every 3 weeks. Children in the placebo group were given chloroquine when
parasites were detected in a blood smear. In group 1 and group 0, 2045 and 428
doses of chloroquine were given, respectively.
All the children were actively searched for every 3 weeks in order to have a
supervised intake of the tablets. The fact that only 75% of the planned doses were
actually received by the children reflects the mobility of the rural population in these
communities, where stay at distant fields or with relatives for a couple of weeks at
a time is widely practised.

Statistical methods
The effect of the chemosuppression on the occurrence of clinical malaria (variables:
clinical history of fever, clinical status, measured body temperature, packed cell
volume, spleen size) was analysed. The variables included were furthermore season
(dry, rainy), age groups ( < 2 years, > 2 years), haemoglobin type (AA,AS,[3T), sex
and village. The data included 3871 encounters from weekly clinics and 730
encounters from three-monthly surveys on 262 children included in the open cohort.
We expect samples on the same child to be more correlated than samples on two
different children, and an ordinary logistic regression analysis was not suitable.
The statistical software package E G R E T version 0.26.6 (Statistics and
Epidemiology Research Co-operation 1991) was used for the computations. Initially
all explanatory variables were included without any interaction, and non significant
variables were removed from the model according to Mauritsen (1984). In a two-
way analysis of variance the effect of clinical symptoms and parasitological groups
on mean body temperature, spleen rate and mean packed cell volume were assessed.
To test whether the effect of the chemosuppression depended on the different variables
considered, interactions were then added.

Results

Clinical measures of morbidity

Based on the history obtained from the mother/guardian concerning fever and the
general clinical status, the children were grouped into four different stages, depending
on the presence or absence of simple clinical criteria (Table 1).

Indicators of malaria morbidity

To validate the possible burden of malaria in causing the morbidity, the clinical
status and the history were related to the parasite rates and density classes, body
temperature, spleen rates and mean packed cell volume in children with parasitemia.
The likelihood and severity of clinical malaria increased from stage two to stage
four. The distribution of parasite rates and densities in the four stages are illustrated
in Fig. 1. Stages three and four had significantly higher parasite rates and densities
than stage one and two and all encounters for children in stage three, who had
 109

TABLE 1
Staging of general clinical condition of the child based on the history obtained from the mother/guardian

Stage condition

Stage I: Child is well, and had no fever during the past week
Stage II: Child is unwell, but able to continue normal activities,
and had no fever during the past week
Stage III: Child is unwell, but able to continue normal activities,
and had fever during the past week
Stage IV: Child is unwell, and unable to continue normal activities,
and had fever during the past week

100

80
no
60 • >0<5,000

" • 5-I0,000
~ 40
• :,1o,ooo

20

1 2 3 4
Clinical stages
Fig. I. On the basis of the history and the general condition, four morbidity stages were defined. The
distribution of parasite density groups within the four different stages is shown as follows 0, > 0 < 5000,
5-10000, > 10000 parasites/pl blood.

p a r a s i t e m i a , were r e g a r d e d as ' p o s s i b l e clinical m a l a r i a ' , a n d all children in stage

f o u r w h o h a d p a r a s i t e m i a , were defined as h a v i n g ' p r o b a b l e clinical m a l a r i a ' . Stages
one a n d two were u n d e f i n a b l e in terms o f clinical m a l a r i a as the p a r a s i t e rate a n d
p a r a s i t e d e n s i t y level d i s t r i b u t i o n are equal in the two stages.
T h e m e a n b o d y t e m p e r a t u r e for the f o u r different stages at the different levels o f
p a r a s i t e densities are s h o w n in Fig. 2. Stages three a n d f o u r h a d increased m e a n
b o d y t e m p e r a t u r e c o m p a r e d to stage one a n d two.
T h e spleen rate o f the f o u r different stages are shown in Fig. 3. T h e m a j o r
difference was seen b e t w e e n c h i l d r e n with a n d children w i t h o u t d e t e c t a b l e parasites.
T h e m e a n p a c k e d cell v o l u m e values for the four different stages are shown
in Fig. 4.
C h i l d r e n with p a r a s i t e m i a h a d l o w e r m e a n p a c k e d cell v o l u m e a n d those in stages
three a n d f o u r h a d lowest m e a n p a c k e d cell volume.

Chemosuppression
T h e g r o u p o f c h i l d r e n receiving c h e m o s u p p r e s s i o n ( g r o u p 1) h a d less m a l a r i a t h a n
the c o n t r o l c h i l d r e n ( g r o u p 0). T h e effect was m o s t p r o n o u n c e d for ' p o s s i b l e clinical
I10

38.6 (141) (209) (254) (126)

38.4
38.2
F10
38.0
37.8 • >0<5,000
,<

37.6 • 5-10,000
37.4 • >10,000
37.2
37.0
36.8
1 2 3 4
Clinical stages

Fig. 2. The mean body temperature at different parasite densities in. the four different stages.

100 (141) (209)m (254) (12

90
1210

[] >0<5,000
z 80
• 5-10,000
,.d

• > I0,000
70

60 • !
I

1 2 3
Clinical stages

Fig. 3. The spleen rate at different parasite densities in the four different stages.

36 (141) (209) (254) (126)

34 m
DO

[] >0<5,000

]/i
• 5-10,000

• > I0,000
i 3o
28
2 3 4
Clinical stages
Fig. 4. The mean packed cell volume values at different parasite densities in the four different stages,
 ill

malaria'. The odds ratio comparing children receiving regular chemosuppression

with children receiving presumptive treatment only was estimated to 0.49 (0.35-0.69).
The difference between the chemosuppression and placebo groups did not depend
significantly ( P = 14.1%) on the season, but there was a tendency towards larger
effect in the dry season than in the rainy season, odds ratio in the chemosuppression
group for 'possible clinical malaria' was in the dry season 0.39 (0.25-0.62) and in
the rainy season 0.63 (0.39-1.04).
For 'possible clinical malaria' the prevalence was higher in the rainy season than
in the dry season odds ratio 1.41 (1.02-1.96), but for 'probable clinical malaria' no
significant season-effect was seen, odds ratio 1.31 (0.83-2.06).
The effect of chemosuppression is illustrated in Fig. 5 and Fig. 6; however, our
estimated odds ratio cannot be derived directly from the proportions given in Fig. 5
and Fig. 6, since we have used the random effect logistic regression model.

Inoculation rate

The entomological data showed seasonal difference in the inoculation rate. In the
rainy season there was at maximum 0.5 infectious bites per person per night, this
rate decreased to 0.02 infectious bites per person per night late in the dry season.
There was no difference in inoculation rates between the two villages.

Haemoglobinopathies

The sickle-cell trait was identified in 9.9% of the individuals, and 5.3% showed the
13-thalassaemia trait. Subjects with haemoglobin AS compared to A A had a lower,

RAINY SEASON

100

80

6O
r.)

~. 40

20

Group 0 Group 1 Group 0 Group 1

6 mths-2 yrs >2<6yrs
[] "probable clinical malaria" (stage IV)
[] "possible clinical malaria" (stage IlI)
[] "none or unlikely clinical malaria" (stages I and II)
Fig. 5. The impact of regular chloroquine chemosuppression in a rainy season in children 6 months - 2
years and > 2 years -< 6 years. Group 1: children receivingchemosuppression. Group0: control children.
Percentage of children who at one or more times of the samples had been having 'probable clinical
malaria' (stage IV), and percentage of children having 'possible clinical malaria' (stage III) (Stages I and
II) 'none and unlikely clinical malaria' have been combined.
ll2

DRY SEASON
100

80

6O
r,.)

~, 40

20

0
Group 0 Group1 Group 0 Group1
6 mths-2yrs >2<6 yrs
• "probableclinical malaria" (Stage IV)
[] "possibleclinical malaria" (Stage III)
[] "noneor unlikely clinical malaria" (Stages I and/I)
Fig. 6. The impact of regular chloroquine chemosuppression in a dry season in children 6 months - 2
years and > 2 years - < 6 years. Group 1: children receivingchemosuppression. Group 0: control children.
Percentage of children who at one or more times of the samples had been having 'probable clinical
malaria' (stage IV), and percentage of children having "possible clinical malaria' (stage III). (Stages I and
I1) 'none and unlikely clinical malaria' have been combined.

but not significantly lower risk of having experienced an attack of 'possible clinical
malaria', odds ratio 0.70 (0.38-1.30), and the risk for subjects with 13-thalassaemia
trait was higher, but not significantly higher, odds ratio 1.60 (0.77-3.33). The
differences between the subjects with different haemoglobin types for 'probable
clinical malaria' were even smaller. AS versus AA odds ratio 0.90 (0.37-2.19), 13-
thalassaemia versus AA odds ratio 1.44 (0.51-4.00).

Discussion

There are no universal indicators of malaria morbidity, and it is difficult to formulate

a general and practical case definition for malaria disease in semi-immunes. Infection
with P. falciparum is not always associated with clinical disease in subjects living
under conditions of intense transmission, because of the development of clinical and
parasitological immunity.
The classification of morbidity stages three and four with detectable parasites as
'possible clinical malaria' and 'probable clinical malaria', was supported by the
higher parasite rates and densities, higher mean body temperature and low packed
cell volume. The high levels of parasitemia and low packed cell volume in stages
three and four indicates that malaria probably is the most important cause of ill
health in this population of children.
The association between temperature and malaria parasitemia is not straightfor-
ward. In a previous study in Liberia, Miller (1958) found that in children parasite
counts of < 11 000 ~tl- 1 were rarely associated with clinical symptoms, but the study
 113

included only 10 children aged 3 to 7 years, and was performed during the dry
season from February through April.
In a study between 1978 and 1982 (911 children age range from new-born to 9
years) of biocultural perceptions of malaria illness (indigenous ethnomedical percep-
tions of malaria-caused ill health) vs. clinical malaria (presence of parasites plus
selected symptoms) of disease in Liberian children, a significant positive linear
correlation was observed between parasitological and clinical measures of malaria
(Jackson, 1985). Among children, the most frequently reported sign or symptom
experienced in connection with malaria illness was fever.
In The Gambia where malaria transmission is highly seasonal, there was found
no correlation between temperature and parasitemia during a cross-sectional survey
towards the end of the dry season, but at the end of the rainy season a significant
correlation was found, but still many children with high parasitemias were afebrile
and well. Many children aged 1 to 4 were anaemic and as expected, low packed cell
volumes were associated with malaria parasitemia, especially in children less than 3
years (Greenwood et al., 1987).
Individual indices of clinical malaria have been used in evaluation of different
control strategies. In the Garki project it was possible to show reduction in point
prevalence of fever > 37.5°C in the most susceptible group <9 years, reduction in
prevalence of P. fal¢iparum and lowering of the spleen rate (Molineaux and
Gramiccia, 1980). The criterion of temperature > 37.5°C in association with parasi-
temia was sensitive enough to detect an effect of bed nets on a population in The
Gambia (Bradley et al., 1986).
In a recent study in Benin it was shown that children with parasitemia exceeding
1000 parasites gl- 1 blood had fever ( > 37.9°C) significantly more often than children
with lower levels of parasitemia (Velema et al., 1991).
The clinical parameters chosen in this study showed a good correlation between
increasing severity of symptoms and increasing parasite density (Fig. 1), increasing
mean body temperature (Fig. 2), increasing spleen rate (Fig. 3), and decreasing mean
packed cell volume (Fig. 4), which supports that the likelihood that an event of
clinical malaria has been encountered increases from stage one to stage four.
In our study, a seasonal difference in the clinical condition stage four was found,
indicating that the increased transmission in the rainy season results in an increase
in clinical malaria indicated by the increase in 'probable clinical malaria' shown
in Fig. 5.
Applying the classification of clinical malaria developed here, we found that
chloroquine chemosuppression had most effect on 'possible clinical malaria', but
less effect on 'probable clinical malaria'. Mild attacks of clinical malaria will be
classified as "possible clinical malaria', and more severe clinical attacks will be
classified as 'probable clinical malaria'. The data thus shows that chloroquine chemo-
suppression in this area is most effective in suppressing mild attacks of malaria.
The data suggests that a certain degree of chloroquine resistance exists in the area,
which was supported by a later study performed in December 1988, where chlo-
roquine resistance was found in 34% of the isolates (Bjorkman et al., 1991).
Chloroquine and immunity probably act in synergy, which may explain why the
combined effect is more pronounced in the dry season with few new inoculations.
In the rainy season with more inoculations of parasites with presumably new anti-
genic varieties, the immunity is not yet as fully developed, and episodes of 'probable
clinical malaria' develops and chloroquine is less effective.
114

Our data are in agreement with a case-control study from the Republic of Niger,
where there was a highly significant relation between the likelihood of fever and the
parasite count during the rainy season, an association which was not found during
the dry season (Rougemont et al., 1991).
In a study in Tanzania chloroquine resistance increased between 1983 and 1989
in children under 5 years old, whereas in schoolchildren resistance decreased from
1986 1989. It is suggested that antigenic differences between resistant and sensitive
strains may explain this age-specifc pattern (Koella et al., 1990).
Peters (1987) suggested that chloroquine resistance is associated with the immuno-
logical properties of Plasmodiumfalciparum which is in accordance with the generally
accepted assumption that immunity against malaria is strain specific.
Evidence for an association between chloroquine resistance of Plasmodium falci-
parum and its immunological properties has recently been reviewed by Koella (1993).
The observations on the distribution of resistance suggest that selection on resistance
is frequency dependent. This may explain why resistance in areas with intense
transmission initially spreads very rapidly after which it is maintained at an interme-
diate level.
It is unknown whether chloroquine chemosuppression may actually be able to
induce enhanced transmission which would be counterproductive to the intended
control of malaria in the individuals. Indications for this have been seen in a recent
study in Tanzania, where a 2.5 fold increase in human infectiousness has developed
in the last 25 years. It has been argued that at least some of the increase may be
attributed to the widespread use of chloroquine, and that chloroquine would be
expected to select for increased infectivity in the parasite (Lines et al., 199l).
It is therefore of fundamental importance to determine the influence of drugs
on transmission dynamics, not only for the understanding of malaria as a disease,
but also as a prerequisite for the development of an appropriate malaria control
strategy.
The clinical stage classification used here based on the history and the clinical
condition of the child as perceived by the mother, provides a tool for easy surveillance
of malaria morbidity without invasive techniques, and is therefore of special use in
large scale operational community studies. Future malaria intervention strategies
need to carefully consider the malaria burden, both at the individual and at the
community level.

Acknowledgements

We wish to thank the parents and children in Beytonwee and Bonah without whose
support this study would not have been possible. We are grateful to the staff at the
malaria research unit of the Liberian Institute for Biomedical Research, Yekepa, for
their invaluable assistance. Our study would not have been possible without con-
tinuous support of the Liberian American-Swedish Minerals Company (LAMCO).
We thank Professor Mario Coluzzi, Department of Parasitology, University of
Rome, for assistance in the entomological aspects of this work. Professor Peter
Perlmann is thanked for his encouraging support during the study. This investigation
received financial support from the UNDP/World Bank/WHO Special Programme
for Research and Training in Tropical Diseases.
 115

References

Baudon, D., Gazin, P., Sanou, J.M., Ouedraogo, L., Ouedraogo, l., Guiguemde, T. and Carnevale, P.
(1986) Morbidit~ palustre en milieu rural au Burkina Faso - - &ude de 526 cas f6briles. M6dicine
d'Afrique Notre, 33, 767-776.
Benasseni, R., Gazin, P., Carnevale, P. and Baudon, D. (1987) Le paludisme urbain a Bobo-Dioulasso
(Burkina Faso) 3. Etude de la morbidite palustre. Cahiers ORSTOM, Seri6 Ent. M6d. Par. 25, 165-170.
Bjorkman, A., Willcox, M., Marbiah, N. and Payne, D. (1991) Susceptibility of Plasmodiumfaleiparum
to different doses of quinine in vivo and to quinine and quinidine in vitro in relation to chloroquine
in Liberia. Bull. WHO. 69, 459-465.
Bradley, A.K., Greenwood, B.M., Marsh, K., Byass, P., Tulloch, S. and Hayes, R. (1986) Bednets
(mosquito nets) and morbidity from malaria. Lancet it, 204-207.
Gilles, H.M. (1988) The differential diagnosis of malaria. In: Malaria, Wernsdorfer, W.H. and McGregor,
I. (eds.) Edinburgh, Churchill Livingstone 1,769-779.
Greenwood, B.M. (1988) Asymptomatic malaria infections - - Do they matter? Parasitol. Today. 3,
206-214.
Greenwood, B.M., Bradley, A.K., Greenwood, A.M., Byass, P., Jammeh, K., Marsh, K., Tulloch, S.,
Otdfield, F.S.J. and Hayes, R.J. (1987) Mortality and morbidity from malaria among children in a
rural area of The Gambia, West Africa. Trans. R. Soc. Trop. Med. Hyg. 81,478-486.
Jackson, L.C.,(1985) Malaria in Liberian children and mothers: biocultural perceptions of illness vs
clinical evidence of disease. Soc. Sci. Med. 20, 1281-1287.
Koella, J.C., Hatz, C., Mshinda, H., de Savigny, D., Macpherson, C.N., Degremont, A.A. and Tanner, M.
(1990) In vitro resistance patterns of Plasmodium falciparum to chloroquine a reflection of strain
specific immunity? Trans. R. Soc. Trop. Med. Hyg. 84(5), 662-665.
Koella, J.C. (1993) Epidemiological evidence for an association between chloroquine resistance of
Plasmodium falciparum and its immunological properties. Parasitol. Today 9, 105-108.
Lines, J.D., Wilkes, T.J., Lyimo, E.O. (1991) Human malaria infectiousness measured by age-specific
sporozoite rates in Anopheles gambiae in Tanzania. Parasitology. 102, 167-177.
Lyimo, E.O., Msuya, F.H.M., Rwegoshora, R.T., Nicholson, E.A., Mnzawa, A.E.P., Lines, J.D., Curtis,
C.F. (1991) Trial of pyrethroid impregnated bednets in an area of Tanzania holoendemic for malaria,
Part 3. Effects on the prevalence of malaria parasitaemia and fever. Acta Trop. 49, 157-163.
Mauritsen, R.H. (1984) Logistic Regression with random effects. PhD Thesis, Department of Biostatics,
University of Washington.
Menon, A., Snow, R.W., Byass, P., Greenwood, B.M., Hayes, R.J., N'Jie, A.B.H.N. (1990) Sustained
protection against mortality and morbidity from malaria in rural Gambian children by chemoprophy-
laxis given by village health workers. Trans. R. Soc. Trop. Med. Hyg. 84, 768-772.
Miller, M.J. (1958) Observations on,the natural history of malaria in the semi-resistant West African.
Trans. R. Soc. Trop. Med. Hyg. 52, 152-168.
Molineaux, L. and Gramiccia, G. (1980) The Garki Project. Geneva: World Health Organization.
Peters, W. (1987) Chemotherapy and Drug Resistance in Malaria (Vol.2) Academic Press.
Rougemont, A., Breslow, N., Brenner, E., Motet, A.L., Dumbo, O., Dolo, A., Soula, G. and Perrin, L.
(1991) Epidemiological basis for clinical diagnosis of childhood malaria in an endemic zone in West
Africa. Lancet, 338, 1292-1295.
Snow, R.W., Lindsay, S.W., Hayes, R.J., Greenwood, B.M. (1988) Permethrin-treated bed nets (mosquito
nets) prevent malaria in Gambian children. Trans. R. Soc. Trop. Med. Hyg. 82, 838-842.
Statistics and Epidemiology Research Corporation. (1991) EGRET: User's manual. Seattle: Statistics
and Epidemiology Research Corporation.
Trape, J.F., Peelman, P. and Morault-Peelman, B. (1985) Criteria for diagnosing clinical malaria among
a semi-immune population exposed to intense and perennial transmission. Trans. R. Soc. Trop. Med.
Hyg. 79, 435-442.
Velema, J.P., Alihonou, E.M., Chippaux, J.-P., van Boxel, Y., Gbedji, E. and Adegbini, R. (1991) Malaria
morbidity and mortality in children under 3 years of age on the coast of Benin, West Africa. Trans.
R. Soc. Trop. Med. Hyg. 85, 430-435.
Warrell, D.A., Molyneux, M.E. and Beasles, P.F. (1990) Severe and complicated malaria. Trans. R. Soc.
Trop. Med. Hyg. 84, 1-65.
Willcox, M., Bjorkman, A., Brohult, J., Pehrson, P.O., Rombo, L. and Bengtsson, E. (1983) A case-
control study in northern Liberia of Plasmodium falciparum malaria in haemoglobin S and ]3-
thalassaemia traits. Ann. Trop. Med. Parasitol. 77, 239-246.