Am. J. Trop. Med. Hyg., 82(2), 2010, pp. 231–234
Copyright © 2010 by The American Society of Tropical Medicine and Hygiene
Distinguishing children with cerebral malaria (CM) from
those with non-malarial comas or an incidental parasitemia is
difficult, especially in malaria-endemic areas with high rates
of asymptomatic parasitemia (e.g., no imaging facilities, lim-
ited laboratory support, and no recourse to electroencepha-
lography). An autopsy study of children meeting the standard
clinical case definition for CM (a Blantyre Coma Score of
≤ 2 in a child with asexual Plasmodium falciparum parasitemia
and no other evident etiology for coma 1 – 3 ) who were admit-
ted to Queen Elizabeth Central Hospital (QECH) in Blantyre,
Malawi found that 23% (7/31) did not have sequestration of
parasites in central nervous system (CNS) vasculature , the hall-
mark of CM pathology, but had other causes of death identified
including pneumonia, Reye’s syndrome, and head trauma. 4 The
clinical expertise and diagnostic facilities available at QECH,
a tertiary care hospital, undoubtedly exceed those generally
available in most African district hospitals and clinics. Rates
of asymptomatic parasitemia in sub-Saharan Africa (SSA)
may be as high as 80% in heavily affected regions. 5 – 7 Therefore,
the overdiagnosis of CM, or misclassification of coma cases as
being caused by malaria when in fact another underlying cause
exists, likely results in substantial limitations for clinical trials
of antimalarial medications and other malaria-specific inter-
ventions as well as “under-diagnosis” (or delayed recognition)
of non-malarial causes of coma in parasitemic children.
Studies aimed at quantifying the neuropsychiatric seque-
lae of CM may also be adversely impacted by the problem
of overdiagnosis or misclassification. In a retrospective cohort
study conducted in Kilifi, Kenya, researchers found that chil-
dren with CM (using the standard clinical case definition) had
an odds ratio of 4.4 (95% CI = 1.4–13.7) for the later devel-
opment of epilepsy. 8 Seizures are a common occurrence in
malaria, typically identified in > 60% of pediatric CM cases, 2 , 9
so the Kilifi study also examined the risk of later epilepsy
development among non-comatose children who had malaria
complicated by seizures. Paradoxically, this non-comatose
“malaria and seizure” group had an odds ratio for epilepsy of
6.1 (95% CI = 2.0–18.3), somewhat higher than that found in
the CM population, many of whom had experienced seizures
during their acute illness. These findings could be caused by the
misattribution of coma to CM in the setting of incidental par-
asitemia among children with a pre-disposition to prolonged
or recurrent seizures when challenged with an acute illness.
Alternatively, malaria-induced seizures, even in the absence of
coma, could be the epileptogenic etiology in this group of chil-
dren who develop epilepsy after a malaria infection.
Recent advances in CM diagnosis based on identifica-
tion of a malaria retinopathy may provide an opportunity to
improve on the standard clinical case definition of CM. 10 – 12
Malaria retinopathy is characterized by retinal whitening, ves-
sel changes, and/or hemorrhages (frequently with white cen-
ters) in patients meeting the standard clinical case definition
of CM ( Figures 1 – 4 ). Papilledema may accompany any of
these features when they are severe, but when present without
the other findings, papilledema is not specific. Children with
malaria retinopathy differ significantly in a number of clinical
and laboratory parameters compared with children who meet
the standard clinical case definition of CM but do not have
the retinopathy. 12 At autopsy, eyes containing retinal vessels
with malaria retinopathy changes seen on clinical examination
have retinal blood vessels containing many dehemoglobin-
ized parasitized red blood cells on histopathological examina-
tion, 11 and angiography has confirmed that retinal whitening is
associated with impaired perfusion. 13 A validation study using
autopsy as the gold standard comparison found that the pres-
ence of malaria retinopathy has 100% specificity and 95%
sensitivity for the identification of CM in fatal disease. 14
In an ongoing, prospective, exposure-control study aimed
at evaluating CM as a risk factor for epilepsy, we recruited
CM survivors from the Pediatric Research Ward at QECH
in Blantyre, Malawi who, on admission, had met the standard
Identification of Malaria Retinopathy Improves the Specificity of the Clinical
Diagnosis of Cerebral Malaria: Findings from a Prospective Cohort Study
Gretchen L. Birbeck, * Nicholas Beare , Susan Lewallen , Simon J. Glover ,
Malcolm E. Molyneux , Peter W. Kaplan , and Terrie E. Taylor
International Neurologic and Psychiatric Epidemiology Program, Michigan State University, East Lansing, Michigan; St. Paul’s Eye Unit,
Royal Liverpool University Hospital, Liverpool, United Kingdom; Kilimanjaro Centre for Community Ophthalmology, Tumaini
University/KCMC, Moshi, Tanzania; Anatomy Department, University of Malawi College of Medicine, Blantyre, Malawi; Malawi-
Liverpool-Wellcome Trust Clinical Research Programme, College of Medicine, Malawi and The Liverpool School of Tropical Medicine
and Hygiene, University of Liverpool, United Kingdom; Department of Neurology, Johns Hopkins Health Systems, Baltimore,
Maryland; College of Osteopathic Medicine, Michigan State University, East Lansing, Michigan; Blantyre Malaria Project,
University of Malawi College of Medicine, Blantyre, Malawi
Abstract. The diagnosis of cerebral malaria (CM) is difficult to confirm in endemic regions with limited neurodiag-
nostics. Accurate diagnoses are critical for trials and outcomes studies. Findings from an autopsy-based study suggest that
identifying malaria retinopathy in children satisfying the standard clinical case definition of CM improves our ability to
accurately diagnose CM in vivo . In a post hoc analysis of a prospective exposure-control study to evaluate CM as a risk
factor for epilepsy, we stratified children meeting the standard case definition by their retinopathy status (presence versus
absence) and compared these groups for pre-existing risk factors for epilepsy. We also compared them to the concurrently
enrolled, non-comatose controls. Children meeting the standard case definition of CM who lacked malaria retinopathy
had a higher prevalence of pre-existing developmental problems and family history of epilepsy. This subset of patients
may represent children with a pre-existing propensity to adverse neurologic symptoms and outcomes.
* Address correspondence to Gretchen L. Birbeck, #324 West Fee
Hall, East Lansing, MI 48824. E-mail: Gretchen.Birbeck@ht.msu.edu
BIRBECK AND OTHERS
clinical case definition of CM. As part of an ongoing autopsy
study of clinicopathological correlates of fatal malaria, all chil-
dren also underwent an ophthalmological assessment looking
for evidence of malaria retinopathy. We undertook a post hoc
analysis of children with CM (per the standard clinical case
definition) to determine whether or not there was a differ-
ence in pre-existing risk factors for epilepsy between those
with malaria retinopathy and those without it. Our purpose
was to assess what proportion of children presenting to the
hospital with apparent CM have evidence of a pre-existing
susceptibility to seizures in order to take this into account in
our ongoing study of subsequent epilepsy risk. Our hypothesis
was that children with such a pre-disposition would be over-
represented in the retinopathy-negative group.
MATERIALS AND METHODS
Study design. This was a post hoc analysis of admission
characteristics among children participating in a prospective,
exposure-control study of CM survivors. Analytic groups
included CM with malaria retinopathy, CM without malaria
retinopathy, and a non-comatose, age-matched, concurrently
admitted control group of children from the general pediatrics
ward. Admitting diagnoses among the control group included
uncomplicated malaria, pneumonia, and gastroenteritis.
Setting and participants. On the QECH Pediatric Research
Ward, children who met the standard clinical case definition of
CM were evaluated for enrollment in a prospective exposure-
control study of CM outcomes aimed at determining whether
or not CM was a risk factor for later epilepsy. Recruitment
occurred between March and December of 2005 and then
from January to June annually (peak malaria season) in 2006
and 2007. Each enrolled CM child was age-matched to a
concurrently admitted child from the general pediatrics ward.
Matching strategy was 1:1 in 2005–2006. When the importance
of including CM retinopathy in our diagnostic criteria became
evident in 2007, we changed our matching strategy to 2:1 to
try and increase our power for the ongoing prospective study,
because requiring this additional inclusion criterion would
inevitably decrease our planned recruitment. Exclusion criteria
for all groups included having any history of unprovoked
seizure before the index illness and residing outside of the
Blantyre area. The Epilepsy Screening Questionnaire was used
to identify children with a history of previous unprovoked
Figure 1. CM patient with retinal whitening in the central mac-
ula and subtle orange discoloration of the blood vessels. This figure
appears in color at www.ajtmh.org .
Figure 2. CM patient with several features of CM retinopathy
including macular and peripheral retinal whitening, hemorrhages, and
papilledema. This figure appears in color at www.ajtmh.org .
Figure 3. Typical CM retinopathy retinal hemorrhage with central
whitening. This figure appears in color at www.ajtmh.org .
CEREBRAL MALARIA RETINOPATHY
seizure. This nine-item questionnaire has been previously
used with 79.3% sensitivity and 92.9% specificity in similar
Variables. At enrollment, data obtained for the CM and
comparison populations included demographic data on
age and gender, biomedical characteristics including birth
weight, an Apgar proxy assessment, body mass index (BMI),
and admission temperature, and information regarding any
pre-existing risk factors for epilepsy including a history of
developmental problems and a family history of epilepsy.
Birth weight was abstracted from the health passport, medi-
cal record, or parental report. Mothers were asked whether
or not their child cried immediately after delivery, and the
answer was treated as a dichotomous Apgar proxy. BMI was
derived from height and weight measurements taken on the
day of discharge. A historical Ten Questions screen was used
to identify pre-index illness developmental problems. This
instrument has been shown to be valid in children as young
as 2 years 16 , 17 and has been evaluated in similar environments
with good reliability (κ = 0.67) and 85% sensitivity for detect-
ing moderate neurodevelopmental disabilities. The Epilepsy
Screening Questionnaire was used to determine whether or
not the child had a prior history of unprovoked seizure and
to ascertain whether or not the study subjects had any first-
degree relatives (parents or siblings) with a history of epilepsy.
Further detailed questions were asked by a physician if any
ambiguity remained after the Ten Questions and Epilepsy
screens. All data were recorded on standardized question-
naires before entry into Microsoft Access and analysis using
EPI INFO (http://www.cdc.gov/epiinfo/about.htm) .
Statistical methods. An analysis of variance (ANOVA) was
used to compare admission characteristics across the three
analytic groups (CM with retinopathy, CM without retinopathy,
and controls). Then, two additional analyses were conducted
comparing 1) the malaria retinopathy negative CM children to
controls and 2) the malaria retinopathy positive CM children to
controls to obtain odds ratios (with 95% confidence intervals).
Where any subgroup included < 5 cases, Fisher’s Exact tests
were used. If Bartlett’s test for inequality indicated the need,
the Kruskal–Wallis test was applied.
Role of the funding sources. The study sponsors played no
role in the study design, collection, analysis, or interpretation
of the data, in writing this manuscript, or in the decision to
submit this paper for publication.
During the observation period, 167 children meeting the
standard clinical case definition for CM were enrolled. Among
these, 132 (79.0%) had malaria retinopathy, and these children
were matched to controls (2:1 matching commenced in year
3 of enrollment, resulting in a total of 264 controls). Table 1
provides the comparisons across all three groups (CM with
malaria retinopathy, CM without retinopathy, and controls).
Compared with controls, the children who met the standard
clinical case definition of CM but lacked malaria retinopathy
were significantly more likely to have a pre-existing history
of developmental delays (OR = 5.75; 95% CI = 2.06–16.02)
and a positive family history of epilepsy (OR = 3.50; 95%
CI = 1.51–10.62). The CM children with malaria retinopathy
did not differ from the control group on these or any other
characteristics assessed ( P > 0.05).
Figure 4. Vessel whitening typical of CM retinopathy with involve-
ment of the capillary bed as well as larger retinal vessels. Retinal whit-
ening is also seen. This figure appears in color at www.ajtmh.org .
Comparison of children meeting the standard clinical case definition of CM with and without malaria retinopathy vs. controls
( N = 264)
positive ( N = 132)
negative ( N = 35)
negative vs. control*
positive vs. control
Sex (% male)
P = 0.99
OR = 0.52
(CI = 0.25–1.07)
P = 0.89
P = 0.43
P = 0.88
OR = 0.52
(CI = 0.11–2.53)
OR = 0.62
(CI = 0.08–4.90)
OR = 5.75
(CI = 2.06–16.02)
OR = 3.50
(CI = 1.51–10.62)
P = 0.85
OR = 1.32
(CI = 0.87–2.00)
P = 0.88
P = 0.68
P = 0.99
OR = 1.01
(CI = 0.30–3.41)
OR = 1.00
(CI = 0.37–2.73)
OR = 0.48
(CI = 0.20–1.13)
OR = 0.58
(CI = 0.24–1.38)
Admit temp (°C)
Body mass index
3.2 ( N = 173)
3.4 ( N = 80)
3.3 ( N = 30)
History of severe
History of developmental
Family history of
12.0 (4.5%)6.0 (4.5%)21.0 (2.9%)0.90
11.0 (4.2%)11.0 (8.3%)7.0 (20.0%)
12.0 (4.5%)10.0 (7.6%)5.0 (14.3%)0.06
* t test or χ 2 test used.
† Mother was asked, “Did the child cry immediately after delivery?”
‡ Malaria requiring prior hospital admission per the health passport.
§ History of developmental problems as per the Ten Question Screen.
234 Download full-text
BIRBECK AND OTHERS
Children who meet the standard clinical case definition of
CM but who lack malaria retinopathy are more likely to have
pre-index illness risk factors for seizures. In this post hoc anal-
ysis, children with pre-existing neurodevelopmental abnor-
malities and/or a genetic pre-disposition to seizures are more
likely to experience prolonged and/or complicated seizures
when challenged with a toxic, metabolic, or infectious pertur-
bation. 18 – 20 Unfortunately, inclusion of these children in CM
clinical trials or outcomes studies may produce misleading
results, because these children already have pre-determined
neurologic outcomes before the index CM illness of interest.
Interventions aimed at decreasing long-term neurologic seque-
lae of CM that include children with a pre-disposition to
adverse neurologic outcomes will, at the very least, be sub-
stantially underpowered to find a beneficial effect. Outcomes
studies of such children may incorrectly attribute adverse out-
comes to CM. Adding the presence of malaria retinopathy to
the diagnostic criteria for pediatric CM will increase diagnos-
tic specificity and may help researchers avoid potential con-
founding factors related to CNS injuries that preceded the
acute CM infection. Future clinical trials and outcomes stud-
ies of CM should include the CM retinopathy criterion.
Received September 9, 2009. Accepted for publication November 16,
Acknowledgment: The authors thank the Blantyre Malaria Project
Epilepsy Study (BMPES) nurses for their important role in this work.
Financial support: Funding for this work was provided by National
Institutes of Health Grant NIH/NINDS K23 NS046086-01, National
Institutes of Health Grant NIH/NIAID R01 AI034969-10AI, and The
Authors’ addresses: Gretchen L. Birbeck , Michigan State University,
International Neurologic and Psychiatric Epidemiology Program,
East Lansing, MI, E-mail: Gretchen.Birbeck@ht.msu.edu. Nicholas
Beare, St. Paul’s Eye Unit, Royal Liverpool University Hospital, Liv-
erpool, E-mail: firstname.lastname@example.org. Susan Lewallen, Kilimanjaro
Centre for Community Ophthalmology, Tumaini University/KCMC,
Moshi , Tanzania , E-mail: email@example.com. Simon J. Glover,
University of Malawi, Anatomy Department, Blantyre , Malawi, E -mail:
firstname.lastname@example.org. Malcolm E. Molyneux, Malawi-
Liverpool-Wellcome Trust Clinical Research Programme, College of
Medicine, Blantyre , Malawi, E -mail: email@example.com. Peter W.
Kaplan, Johns Hopkins Health System, Department of Neurology,
Baltimore , MD, E-mail: firstname.lastname@example.org. Terrie E. Taylor, Michigan
State University, College of Osteopathic Medicine and University of
Malawi College of Medicine, Blantyre Malaria Project, East Lansing,
MI, E-mail: email@example.com.
1. Brewster DR , Kwiatkowski D , White NJ , 1990 . Neurological seque-
lae of cerebral malaria in children . Lancet 336 : 1039 – 1043 .
2. Molyneux ME , Taylor TE , Wirima JJ , Borgstein A , 1989 . Clinical
features and prognostic indicators in paediatric cerebral
malaria: a study of 131 comatose Malawian children . Quarterly
Journal of Medicine 71 : 441 – 459 .
3. WHO , 1999 . SEAR malaria incidence . WHO . Available at: http://
4. Taylor TE , Fu WJ , Carr RA , Whitten RO , Mueller JS , Fosiko NG ,
Lewallen S , Liomba NG , Molyneux ME , 2004 . Differentiating
the pathologies of cerebral malaria by postmortem parasite
counts . Nat Med 10 : 143 – 145 .
5. Bronzan RN , McMorrow ML , Kachur SP , 2008 . Diagnosis of
malaria: challenges for clinicians in endemic and non-endemic
regions . Journal of Molecular Diagnostics 12 : 299 – 306 .
6. Dal-Bianco MP , Koster KB , Kombila UD , Kun JF , Grobusch MP ,
Ngoma GM , Matsiegui PB , Supan C , Salazar CL , Missinou MA ,
Issifou S , Lell B , Kremsner P , 2007 . High prevalence of asymp-
tomatic Plasmodium falciparum infection in Gabonese adults .
Am J Trop Med Hyg 77 : 939 – 942 .
7. Smith T , Schellenberg JA , Hayes R , 1994 . Attributable fraction
estimates and case definitions for malaria in endemic areas . Stat
Med 13 : 2345 – 2358 .
8. Carter JA , Neville BG , White S , Ross AJ , Otieno G , Mturi N ,
Musumba C , Newton CR , 2004 . Increased prevalence of epi-
lepsy associated with severe falciparum malaria in children .
Epilepsia 45 : 978 – 981 .
9. Newton CR , Warrell DA , 1998 . Neurological manifestations of
falciparum malaria . Ann Neurol 43 : 695 – 702 .
10. Lewallen S , Harding SP , Ajewole J , Schulenburg WE , Molyneux ME ,
Marsh K , Usen S , White NJ , Taylor TE , 1999 . A review of the
spectrum of clinical ocular fundus findings in P. falciparum
malaria in African children with a proposed classification and
grading system . Trans R Soc Trop Med Hyg 93 : 619 – 622 .
11. Lewallen S , White VA , Whitten RO , Gardiner J , Hoar B , Lindley J ,
Lochhead J , McCormick A , Wade K , Tembo M , Mwenechanyana J ,
Molyneux ME , Taylor TE , 2000 . Clinical-histopathological cor-
relation of the abnormal retinal vessels in cerebral malaria .
Arch Ophthalmol 118 : 924 – 928 .
12. Lewallen S , Bronzan RN , Beare NA , Harding SP , Molyneux ME ,
Taylor TE , 2008 . Using malarial retinopathy to improve the
classification of children with cerebral malaria . Trans R Soc
Trop Med Hyg 102 : 1089 – 1094 .
13. Beare NA , Harding SP , Taylor TE , Lewallen S , Molyneux ME ,
2009 . Perfusion abnormalities in children with cerebral malaria
and malarial retinopathy . J Infect Dis 199 : 263 – 271 .
14. Beare NA , Taylor TE , Harding SP , Lewallen S , Molyneux ME ,
2006 . Malarial retinopathy: a newly established diagnostic sign
in severe malaria . Am J Trop Med Hyg 75 : 790 – 797 .
15. Placencia M , Sander JW , Shorvon SD , Ellison RH , Cascante SM ,
1992 . Validation of a screening questionnaire for the detection
of epileptic seizures in epidemiological studies . Brain 115 :
783 – 794 .
16. Durkin MS , Hasan ZM , Hasan KZ , 1995 . The ten questions screen
for childhood disabilities: its uses and limitations in Pakistan .
J Epidemiol Community Health 49 : 431 – 436 .
17. Durkin MS , Wang W , Shrout PE , Zaman SS , Hasan ZM , Desai P ,
Davidson LL , 1995 . Evaluating a ten questions screen for child-
hood disability: reliability and internal structure in different
cultures . J Clin Epidemiol 48 : 657 – 666 .
18. Berg AT , 1992 . Febrile seizures and epilepsy: the contributions of
epidemiology . Paediatr Perinat Epidemiol 6 : 145 – 152 .
19. Berg AT , Shinnar S , 1991 . The risk of seizure recurrence following
a first unprovoked seizure: a quantitative review . Neurology
41 : 965 – 972 .
20. Berg AT , Shinnar S , 1996 . Unprovoked seizures in children
with febrile seizures: short-term outcome . Neurology 47 :
562 – 568 .