Journal of Autism and Developmental Disorders

, 36:1039

Etiologies of Autism in a Case-series from Tanzania

Authors

  • Raymond E. Mankoski
    • Tufts University School of Medicine and Sackler School of Graduate Biomedical Sciences
  • Martha Collins
    • Department of PsychiatryTufts-New England Medical Center
  • Noah K. Ndosi
    • Department of PsychiatryMuhimbili University College of Health Sciences of the University of Dar es Salaam
  • Ella H. Mgalla
    • Autism UnitMsimbazi Mseto Primary School
  • Veronica V. Sarwatt
    • Autism UnitMsimbazi Mseto Primary School
    • Johns Hopkins University School of Medicine
Original Paper

DOI: 10.1007/s10803-006-0143-9

Cite this article as:
Mankoski, R.E., Collins, M., Ndosi, N.K. et al. J Autism Dev Disord (2006) 36: 1039. doi:10.1007/s10803-006-0143-9

Abstract

Most autism has a genetic cause although post-encephalitis cases are reported. In a case-series (N = 20) from Tanzania, 14 met research criteria for autism. Three (M:F = 1:2) had normal development to age 22, 35, and 42 months, with onset of autism upon recovery from severe malaria, attended by prolonged high fever, convulsions, and in one case prolonged loss of consciousness. In four other cases (M:F = 3:1), the temporal relationship between onset of autism and severe infection was close, but possibly spurious since malaria is common in Tanzania and there were indications of abnormal development in the child or a family member. In seven cases, (M:F = 6:1) autism onset was unrelated to malaria. The excess of non-verbal cases (N = 10) is related local diagnostic practice.

Keywords

AutismMalariaAfricaInfection

Introduction

Autism in the Western world has a primarily genetic etiology, although the genes responsible for the majority of cases are still not known (Folstein & Rosen-Sheidley, 2001). A number of claims about exogenous causes related to vaccines have been advanced recently, but these claims have little support (IOM, 2004). Two exogenous etiologies have, however, been well documented: congenital rubella (Chess, 1971) and encephalitis. In particular, herpes simplex virus (HSV) encephalitis has been clearly identified as an etiologic agent of autism in at least six published cases. It is not clear how the neuropathology of herpes simplex encephalitis results in autism, but a number of cases have demonstrated extensive bilateral lesions of the medial temporal lobes. In two cases the damage was more pronounced on the left side (Ghaziuddin, Tsai, Eilers, & Ghaziuddin, 1992; Gillberg, 1991) and parietal lobe involvement was also seen in a third (Gillberg, 1986). In one recently described case there was primarily bilateral medial frontal lobe involvement (Ghaziuddin, Al-Khouri, & Ghaziuddin, 2002).

We had occasion to systematically examine 20 children in a school in Dar es Salaam, Tanzania who had been clinically diagnosed as autistic. Several of the children we examined had symptom onset following a severe cerebral infection, most commonly malaria. Even though malaria is endemic in sub-Saharan Africa, only about 2% of clinical infections with Plasmodium falciparum result in severe disease (Greenwood, Marsh, & Snow, 1991). Features of severe malaria include prostration, multiple seizures, respiratory distress, and severe anemia (Marsh et al., 1995). Cerebral malaria is diagnosed when a patient cannot localize a painful stimulus, has peripheral asexual falciparum parasitemia, and has no other causes of an encephalopathy (Warrell, Molyneux, & Beales, 1990). The favored hypothesis about the cause of cerebral malaria is that parasitized red blood cells clog small vessels in the brain, especially in relatively hypoxic areas such as venous beds (Newton, Hien, & White, 2000). Although most children are not left with severe neurological impairments, a number of small studies have found associations between severe malaria and language impairment (Carter, Murira, Ross, Mung’ala-Odera, & Newton, 2003; Holding, Stevenson, Peshu, & Marcsh, 1999). The authors did not attempt to evaluate the children for autism.

There are few published reports about autism in Africa. We could find only two attempts at systematic surveys of indigenous African populations (Kahn & Hombarume, 1996; Lotter, 1978). Lotter visited schools and hospitals for mentally handicapped children in nine cities in six sub-Saharan African countries. He found 22 cases that met the diagnostic criteria, as operationalized by Creak (1961). The clinical features were fairly typical except that 16 of the 22 were non-speaking and the excess of males was only 2:1. Kahn and Hombarume found 18 children in schools for mentally handicapped children in three regions of Zimbabwe who met DSM III-R criteria for autism. Most were non-verbal; the sex ratio was 5:1. The prevalence of autism in African populations is unknown.

We report here a case series of 14 children who met ADI-R criteria for autism. In three, autism began after severe malaria, following a long period of normal development.

Methods

Sample

Twenty autistic children and adults were examined. At the time of ascertainment, 10 attended the Autism Unit at the Msimbazi Mseto Primary School in Dar es Salaam, which had a total enrollment of 12 students. Four of the eight children from the school’s waiting list were examined, and an additional six were recruited from the community by school staff from among 14 families who they knew had a child with clinically diagnosed autism. All of the families contacted agreed to participate, but we had only enough time to examine 20. Being enrolled in the autism unit or being known to the unit staff is the only known bias of ascertainment. The staff considered these children to be typically autistic and they had no knowledge of the children’s medical history.

Of the 20 subjects examined, 14 meet our diagnostic criteria for autism. Three were estimated to have a non-verbal functioning of less than 18 months, one had a mucopolysaccharide storage disease, one did not meet criteria on the social domain of the ADI-R algorithm, and one did not meet criteria on the repetitive/stereotyped behavior domain of the ADI-R algorithm. Case B met all behavioral ADI-R criteria for autism but is diagnosed as PDD-NOS because onset of symptoms began following malaria at 42 months. The subjects ranged in age from 8 years to 22 years; mean age 12.3 years. Permission to publish brief descriptions of the subjects (Appendix) was obtained from the parents or caregivers in all except Case H.

The protocol was approved by the Tanzanian Commission for Science and Technology in Dar es Salaam and the Ilala District Education Authorities. Consent forms approved at Tufts-New England Medical Center were translated into Kiswahili. The Tanzanian woman (VS) who translated the consent forms explained the study to the parents or caregivers of the subjects. Each participant received 10,000 Tanzanian shillings (US $13) to cover expenses.

Diagnosis

The Autism Diagnostic Interview-Revised (ADI-R) (Lord, Rutter, & Le Couteur, 1994) was translated into Kiswahili by a psychiatrist (NN) who is fluent in English and clinically experienced with autism. ADIs were administered by one of two teams: a Kiswahili-speaking psychiatrist (MC) paired with the social worker in the autism unit or a child psychiatrist (SF) paired with the head teacher (EM) who translated as needed. Some parents spoke good English and often answered in English to the questions, which were asked in Kiswahili.

Subjects are classified as “verbal” according to the definition used on the ADI-R. This requires the “functional use of spontaneous, echoed, or stereotyped language that, on a daily basis, involves phrases of 3 words or more, that at least sometimes includes a verb, and is comprehensible to other people.”

SF and RM observed the children, performed an examination of the motor system using a standardized coding system, and obtained measurements of head circumference. Head circumferences (HC) are scored as age- and sex- matched percentile ranks (Gordon et al., 1988; Roche, Mukherjee, Guo, & Moore, 1987). The head circumference norms are those used at the Muhimbili Medical Center in Dar es Salaam, Tanzania. These are derived from a Caucasian population since there are no norms for the East African population. Estimations of level of non-verbal functioning were based on this examination, the ADI, and their school performance. We also obtained a medical and developmental history, as well as a family history of developmental disorders. Diagnosis is based on the ADI-R diagnostic algorithm, history, and the clinical evaluation. Mental Age was estimated using the Daily Living Scale from the Vineland (Sparrow & Cicchetti, 1985).

The parents of seven children reported that autism symptoms appeared immediately after recovery from an acute febrile infection that had required hospitalization. We could not get access to medical records except in two cases in which the parents had them, but parents reported that all but one of the children had severe malaria as confirmed by examination of a blood smear. Case “A” met criteria for cerebral malaria. One child for whom we had medical records had documented Salmonella meningitis, described in the record by an attending physician as a “very severe central nervous system infection”. Parents for all hospitalized cases reported extremely high fever (often 104°F) that persisted for a week or more, accompanied by multiple febrile convulsions and postictal loss of consciousness.

Because no written developmental records were kept, either by the parents or by the physicians who attended them, we took particular care to document, using life events, seasons, etc., the age at which the children achieved various developmental milestones and the details of their language and social development. Most of the parents appeared to be reliable informants, and many were well educated and worked at responsible jobs. They were good advocates for their children, as evidenced by their success in getting their children admitted to, or on the wait list for, the only school for autistic children in Dar es Salaam.

Results

The series includes 10 males (m) and four females (f) (m:f ratio 2.5:1). The case summaries (except Case H) are included as an appendix. Table 1 summarizes the salient demographic and clinical features. Notably, the majority of cases are non-verbal. Only four children had sufficient speech at age 4–5 to meet the ADI-R criterion for “verbal”; an additional two now meet this criterion. Four cases have mild craniofacial dysmorphology. Head circumferences greater than the 75th percentile are seen in more than half of the cases, five have non-febrile seizures, and five have a positive family history of developmental difficulties.
Table 1

Characteristics of cases

 

Number of cases N = 14

Relevant case summary

m:f ratio

2.5:1

 

Language present

4 (age 4–5) 6 current

C, H, M, N (D,F now verbal)

Positive family history of developmental abnormalities

6

D,E, G, I, N

Craniofacial dysmorphology

4

E,K,L,M

Head circumference ≥ 75th percentile

8 (N = 13)

B,D,E,F,I,K,L,N

Non-febrile seizures

5

C,D,E,F,L (M atypical febrile)

The cases are arranged in Table 2 and in the appendix according to the relationship between onset of autism and onset of severe infection. They can be divided into three groups: Cases A–C for which the relationship between onset of autism and severe malaria seems clear, Cases D–G for which there may have been a relationship, and Cases H–N for which there is no relationship.
Table 2

Characteristics of cases by likely etiology

 

Clear infectious etiology N = 3

Possible infectious etiology N = 4

Non-infectious etiology N = 7

m:f ratio

1:2

3:1

6:1

Language present (verbal)

1

0 age 4–5; 2 current

3

Positive family history of developmental abnormalities

0

1

3

Cranial dysmorphology

0

1

3

Head circumference ≥ 75th percentile

1 (of 2)

3

4 (of 6)

Non-febrile seizures

1

3

1 (+1 atypical febrile)

Cases A–C

In these three children (one male, two females), the period of normal development was sufficiently long (35, 42, and 22 months) to be reasonably sure that development was truly normal at the time their infectious illnesses occurred. They became ill after the time at which regressive autism is common. Two of three were using phrases and the other had many self-initiated single words. None had a positive family history for developmental disorders. Only one of the three met criteria for cerebral malaria, but all three were extremely ill and were hospitalized for a week or more with extremely high fevers, repeated convulsions and postictal loss of consciousness.

Cases D–G

The evidence in these cases (three males, one female) for a relationship between autism and infection is not as strong, although in Cases D–F there was an acute change in development following the infectious illness. They were too young at the time of the infection to be sure that development was truly normal, and some had mild developmental delays prior to their infection. These include lack of phrase speech at 18 months in Case E and lack of babble at nine months in Case F. Case F also had albinism but was otherwise not dysmorphic. Three of the four children had family members with developmental delays or disorders.

The most difficult to characterize is Case G, who had phrase speech at 15 months and contracted malaria at about 17 months. It was not a serious case (no loss of consciousness following seizures), and his mother said that there was no acute change in his development after his recovery. However, afterwards he became non-verbal and remains so and developed symptoms of autism.

Cases H–N

In the remaining seven cases (six males, one female) there was no relationship between the time at which abnormal development was noticed and any serious infection. Except for the excess of non-verbal cases (four of seven), they are all typical autistic children. Three of the four non-verbal cases had mildly dysmorphic craniofacial features (see Table 3).
Table 3

Summary of case information

Case

Age, Sex

Age 1st Sx.

Infection/epilepsy

Development before infection

ADI Algorithm

HC %tile

Comment

A

10 years F

35 m

Hospitalized at 35 m for cerebral malaria with prolonged LOC. Confirmed by smear.

No Epilepsy.

Normal socialization; sentences. Acute change after malaria

S:28

C:14 (NV)

R:4

Onset: 1

FH negative

Not dysmorphic

Typical non-verbal autism

MA/VDL = 2 y,11 m

B

9 y M

42 mos

Malaria by report at 42 m. Hosp with high fever and convulsions × 1 wk with postictal LOC

No epilepsy

Normal socialization; sentences. Acute change after malaria.

S:26

C: 16 (NV)

R:6

Onset 0

75–90

FH negative

Not dysmorphic

Meets autism criteria except onset

MA/VDL = 1 y, 6 m

C

10 y F

22 m

Malaria by smear at 22 m with severe convulsions and pneumonia.

Postictal LOC

Has epilepsy

Normal socialization; single words. Acute change after malaria.

S:28

C:21 (V)

R:13

Onset: 3

5–10

FH negative

Not dysmorphic

Typical autism

MA/VDL = 4 y, 4 m

D

15 y M

6 m

Salmonella meningitis by CSF culture at 6 m. high fever and postictal LOC × 1 wk;

Has epilepsy

Normal, to 6 m. Acute change after meningitis.

S:27

C:14 (NV)

R:7

Onset: 4

90–95

Possible language delay in sister

Long face but no clear dysmorphic features

Typical autism

MA/VDL = 3 y, 3 m

E

17 y M

18 m

Malaria at 18 m; hospitalized × 1 wk with convulsions with postictal LOC

Has epilepsy

Single words before malaria; words gradually lost after repeated bouts

S:25

C:14 (NV)

R:5

Onset 3

75–90

Brother with abnormal development; mat. bleeding 2nd trimester

Possibly dysmorphic (protruding upper teeth)

MD/VDL = 18 m

F

8 y F

9 m

Malaria at 9 m by report; high fever  × 1 wk. with convulsions; brief hospitalizations for repeated fevers for seizure control.

Mild epilepsy

No babble at 9 m. Seemed normal socially until malaria.

S:26

C:14 (NV)

R:8

Onset: 2

75–90

FH negative

Albinism but not dysmorphic

Typical autism; now verbal

MA/VDL = 2 y, 8 m

G

12 M

17 m

Malaria at 17 m; hospitalized with high fever, convulsions and given quinine but no LOC

No epilepsy

Walk 17 m

Phrases 15 m. No acute change after malaria but speech gradually lost.

S:24

C:12 (NV)

R:9

Onset: 3

10–25

Pos.FH developmental disorders

Mother pre-eclampsia and child given O2 at birth

Not dysmorphic but h/o exptropia resolved

Low functioning autism MA/VDL = 19 m

H

13 M

18 m

Never had malaria

No epilepsy

NA

Verbal

 

FH Negative

Not dysmorphic

Typical autism

MA/VDL = 4 y, 9 m

I

11 M

1st year of life

Clearly affected before 1st malaria

No epilepsy

NA

S:22

C:11 (NV)

R:10

Onset: 3

75–90

MR in mat. cousin

Not dysmorphic

Typical low functioning autism.

MA/VDL = 2 y, 3 m

J

13 M

2nd year of life

Clearly affected before 1st malaria

No epilepsy

NA

S:25

C:14 (NV)

R:6

Onset: 3

10–25

FH negative.

Not dysmorphic.

Low functioning autism.

MA/VDL = 2 y, 3 m

K

22 M

1st yr of life

Clearly affected before 1st malaria

NA

S:27

C:12 (NV)

R:9

Onset: 3

95–97

FH negative

Large head, protruding upper teeth

Typical low functioning autism.

MA/VDL = 1 y, 10 m

L

12 M

12 m

Clearly affected before 1st malaria

Has epilepsy

NA

S:28

C:14 (NV)

R:6

Onset: 3

75–90

FH negative

Mild dysmorphic features

Low functioning autism

MA/VDL = 1 y, 8 m

M

9 M

1st yr of life

Clearly affected before 1st malaria

Febrile seizures only, but atypical unilateral convulsions

NA

S:27

C:19 (V)

R:4

Onset: 4

< 5

FH negative

Prominent occiput; worse gait than usual

Typical low functioning autism

MA/VDL = 1 y, 8 m

N

14F

30 m

Clearly affected before 1st malaria

No epilepsy

NA

S:26

C:19(V)

R:6

Onset: 1

90–95

Brother and father delayed language

Not dysmorphic

Typical autism.

MA/VDL = 4 y, 7 m

FA = family history, NV = non-verbal, MA = mental age, LOC = loss of consciousness, y = year, m = month wk = week S = Social domain C = Communication domain R = Restricted/repetitive behaviors NA = not applicable, MA/VDL = mental age from Vineland Daily Living Scale

Discussion

We describe 14 young people with autism who were seen at a special education unit for autism at the Msimbazi Mseto Primary School in Dar es Salaam, Tanzania. Each case was examined by a psychiatrist with extensive experience in the diagnosis and treatment of autism in the United States (SF), and each case meets ADI-R diagnostic criteria for autism (one case meets all criteria except for having a mental age < 18 months). Some cases also have additional features that are common in autism, such as sensory aversions, abnormal sensory interests, epilepsy, hypotonicity, and poor coordination.

Although the features of the individual cases are the same as those seen in developed countries, the proportion of non-verbal cases is higher (11/14;although two more are now verbal since going to school). The high proportion of non-verbal children is probably related to the fact that only the most severe cases are accepted to the unit at Msimbazi Mseto, which is the only educational unit for autistic children in Dar es Salaam and only one of two in all of Tanzania. Also, children who have functional language are less likely to be diagnosed with autism in Tanzania. The cases reported by Lotter (1978) and by Khan and Hombarume were also mainly non-verbal. The latter authors point out that in Zimbabwe the diagnosis of autism is not made separately from mental retardation and that “those of average or above average ability are not catered for.” Thus, it appears that autism, when it is diagnosed, is diagnosed mainly in mentally retarded children.

All four cases that had mild craniofacial dysmorphology were non-verbal. Kanner would not have diagnosed these children as having autism since he excluded children with dysmorphic features. However, mild dysmorphic features are relatively common in non-verbal autism as it is diagnosed using current criteria and practice. Cytogenetic or neuroimaging evaluations would be illuminating as to the cause of autism in these four cases, but these facilities are limited in Tanzania and utilized only when there are likely to be therapeutic implications.

We attempted to estimate non-verbal mental age, based on the parents’ histories and our evaluations, but these should be interpreted with caution and as minimum estimates. Most of the children had extremely limited schooling and their parents did not make efforts to teach them. It was remarkable how many parents commented that their children’s language, social, and daily care functioning had begun to improve after they started school.

The most interesting aspect of this series is that 3/14 (21%) children who had an entirely normal development through the first two years of life acquired autism immediately after having severe malaria. Autism cases having an infectious etiology are not unknown in developed countries, but they are rare. Agents in these cases include congenital cytomegalovirus (Markowitz, 1983; Stubbs, Ash, & Williams, 1984), herpes simplex virus (DeLong, Dean, & Brown, 1981; Ghaziuddin et al., 1992, 2002; Gillberg, 1986, 1991), and a stealth virus (Martin, 1995). To our knowledge, this is the first report to identify Plasmodium falciparum as an etiologic agent of autism. Persistent impairments following cerebral malaria have been described, including paresis, ataxia, aphasia/dysarthria, behavior problems, and developmental regression (van Hensbrook, Palmer, Jaffar, Schneider, & Kwiatkoski, 1997). Recent studies in Kenya have found an association between severe malaria and language impairment. In one study, children who had severe malaria with impaired consciousness had significantly worse scores on measures of syntax and articulation than matched controls (Holding et al., 1999). Another study reports that children who had severe malaria performed significantly worse than matched controls on tests of comprehension, syntax, and lexical semantics (Carter et al., 2003). None of these authors attempted to evaluate these children for autism, but the group in Kenya is now doing so in a large ongoing project.

While we examined only a single case (Case A) with medically documented cerebral malaria, it raises the interesting question of how this infection can cause the same clinical features seen in autism cases with a genetic etiology. We speculate that the concordance between them is due to a shared anatomic focus of pathology. The temporal lobes have been implicated as an important region of dysfunction in several neuroimaging and histological studies of idiopathic autism (Bauman & Kemper, 1994; Bolton, Park, Higgins, Griffiths, & Pickles, 2002; Casanova, Buxhoeveden, Switala, & Roy, 2002; Mountz, Tolbert, Lill, Katholi, & Liu, 1995; Ohnishi et al., 2000; Zilbovicius et al., 2000), and like herpes simplex virus encephalitis (Craighead, 2000), the damage caused by cerebral malaria may also have a predilection for this region. A clinical and electrophysiological study of 65 children with cerebral malaria found that most seizures arose from the posterior temporo-parietal region (Crawley, Smith, Muthinji, Marsh, & Kirkham, 2001). As pointed out by Crawley et al., this “watershed” area is “particularly vulnerable to hypoxia when oxygen delivery to the brain is compromised as a result of (parasite) sequestration, severe anemia, or inadequate cerebral perfusion caused by hypotension or raised intracranial pressure.” While Cases B and C did have severe malaria, they did not meet criteria for cerebral malaria (loss of consciousness was not sufficiently prolonged). However, their autism did begin immediately upon recovery. Thus, the question arises whether slightly less severe malaria can also cause damage to the temporo-parietal region.

Case D had onset after documented Salmonella meningitis. Since the child was only 6 months old when he became ill, it is not certain that this was the cause of his autism, even though he was extremely ill and social impairment was impaired after he recovered. However, he was extremely ill and when he recovered, social interaction was impaired. Salmonella, like malaria has also been associated with neurologic sequelae (Huang, 1996; Unhanand, Mustafa, McCracken, & Nelson, 1993), but it does not necessarily cause damage to any particular brain region. However, Case D had an ensuing anemia severe enough to require blood transfusions, and this anemia may have exacerbated any temporal lobe damage he suffered from the meningitis. Thus, the disruption of neural circuits in the temporal lobes early in life, regardless of mechanism, could be the unifying pathologic mechanism of both autism caused by severe Plasmodium falciparum malaria or Salmonella and the more commonly caused genetic disorder.

It is possible that the onset of autism symptoms in cases E–G is in fact related to a first malarial infection as reported by the parents. None of these children had cerebral malaria, although most were quite ill with prolonged high fever and febrile convulsions. Many children in an endemic region are exposed to Plasmodium falciparum malaria before they are five years old, and like the hypothesized link between MMR and autism, the link between mild malaria and autism may be specious. Without better evidence, the assertions by the parents that their child’s autism was acquired after a their first episode of malaria must for the time being be considered with caution.

If we conservatively consider only cases A, B, and C as having convincing histories for onset after malaria, it represents 21% of the cases that we saw. We have no reason to suspect that there was any bias of ascertainment toward seeing cases with malaria. The teachers at the school who found the cases had no belief that there any such relationship existed, nor did they know the children’s early medical history.

One logical consequence of the frequency of cases with infectious etiologies is that the prevalence of autism should be higher in Africa than it is in the West. This assumes that the prevalence of idiopathic autism is the same in Africa as in the countries in which it has been estimated, mainly Western Europe, Japan and North America. The prevalence of autism in Africa is, however, unknown.

We would like to propose the hypothesis that severe malaria, when contracted in the first few years of life, can cause autism. The major limitation of this case-series as a hypothesis-generating study is the lack of medical documentation in most cases. Medical documentation is not retained indefinitely in Tanzania and, except in the University Hospitals, is not very detailed. Parents had kept their own copies of case notes in the two cases for which we had documentation. In other cases, parents were told that the diagnosis of malaria was confirmed by smear when their children were hospitalized.

Acknowledgments

We wish to gratefully acknowledge all of the participants, their families, and their friends for committing their efforts and energy to autism research. We also wish to thank the staff at the Msimbazi Mseto Primary School for their valuable assistance in coordinating this research project. We are grateful to Professor Japhet Minjas at the Muhimbili Medical Center for generously providing us with laboratory space and resources. We also wish to acknowledge the Tanzania Commission for Science and Technology for permitting us to conduct this research in Dar es Salaam. Mr. Mankoski is supported by a grant from the National Alliance for Autism Research and by the Medical Scientist Training Program of the National Institutes of Health (USA). Dr. Folstein is supported (in part) by a grant from the National Alliance of Autism Research and by the National Institutes of Mental Health (USA).

Copyright information

© Springer Science+Business Media, Inc. 2006