Epilepsy and Toxocariasis: A Case-Control Study in Burundi


Address correspondence and reprint requests to Alessandra Nicoletti, Department of Neurosciences, University of Catania, Via S. Sofia no. 78, 95123 Catania, Italy. E-mail: anicolet@unict.it


Summary: Purpose: A case-control study to assess the relationship between epilepsy and toxocariasis was carried out in the Kiremba population, Burundi.

Methods: People with epilepsy (PWE) were diagnosed according to the definition proposed by the International League Against Epilepsy (ILAE). Seizures were classified according to the classification proposed by ILAE in 1981. One control per case was selected matched by age (±5 years). Control subjects also lived in Kiremba, had neither neurological disorders nor kinship with the PWE. Cases and controls were assessed serologically for antibodies against Toxocara canis by an immunoblotting assay. Odds ratios (ORs) and 95% CI were determined using conditional regression analysis for matched case-control study.

Results: One hundred ninety-one PWE (99 men and 92 women) and 191 age-matched controls (72 men and 112 women) were enrolled in the study. Of the 191 PWE, 113 presented partial seizures while 73 generalized seizures and five were unclassifiable. Antibodies anti T. canis were found in 114 PWE (59.7%) and in 97 controls (50.8%). Multivariate analysis (conditional logistic regression) showed a significant association between positivity for T canis and epilepsy with an adjusted OR of 2.13 (95% CI 1.18–3.83; p-value 0.01).

Conclusions: We found a significant association between toxocariasis and epilepsy. In agreement with a previous study, our finding suggests that toxocariasis may increase the risk of developing epilepsy in endemic areas and could participate to the high burden of epilepsy in tropical areas.

Human toxocariasis is a parasitic zoonosis caused by the larval stages of Toxocara canis, the common roundworm of dogs, and probably by the roundworm of cats (T. cati) too. Humans are infected by ingestion of embryonated T. canis eggs present in the soil or on contaminated hands and fomites (Schantz and Glickman, 1978). Most humans infected by T. canis do not develop overt clinical disease (Glickman, 1990). On the other hand, migrating T. canis larvae can cause, particularly in young children, inflammatory tissue reactions with multisystem involvement leading to the clinical syndrome of visceral larva migrans. The larvae can locate in the central nervous system (CNS) leading to a variety of neurological disorders (Schantz and Glickman, 1978). How often Toxocara larvae provoke neurological disorders in humans is not clear. Toxocariasis occurs whenever the man-soil-dog relationship is particularly close. High prevalence of Toxocara infection have been found in low-resource tropical countries where the humid climate favors the survival of parasite eggs in the soil and poor hygiene increases the probability of human infection (Thompson et al., 1986; Lynch et al., 1988; Magnaval et al., 1994).

A possible association between Toxocara infection and epilepsy has been hypothesized and toxocariasis has been suggested as a cofactor for epilepsy (Arpino et al., 1992). Further evidence of association was reported by our group in a population based case-control study carried out in rural Bolivia (Nicoletti et al., 2002).

A case-control study was carried out in 2001 to evaluate the relationship between cysticercosis and epilepsy in an area known as Kiremba located in the northern part of Burundi (Nsengiyumva et al., 2003). In the present paper we report the results of a study aimed of evaluating, as secondary endpoint, the relationship between epilepsy and seropositivity for antibodies against T. canis.


Study population

The survey was carried out during March and April 2001, in the territorial division of Kiremba in the north of Burundi. The Kiremba division comprises 46 hills with 70,000 inhabitants. This is a rural area with five primary health care centers and one hospital with 200 beds.

We used a case-control design where cases were people with epilepsy (PWE) defined according to standardized criteria. The medical personnel working in these towns had, in the weeks preceding the study, talked to the village leaders, PWE, and their families to obtain their participation in the study. One healthy control per case, matched by age (±5 years), was also selected from the general population. Controls were subjects coming to the hospital for vaccination or were neighbors of PWE; they were free from neurological disorders and had no kinship with the PWE, in order to prevent any genetic propensity to epilepsy among the control subjects. Cases and controls had been living in the Kiremba area for at least two years. PWE and controls were enrolled during the same period (March–April 2001). From the original sample selected to investigate the relationship between epilepsy and cysticercosis (324 PWE and 648 age-matched controls), we included in the present study only those PWE and their age-matched controls for whom an additional aliquot of serum was available.

The main exposure in study was the presence of antibodies against T. canis detected using an immune blot assay. Clinical and paraclinical examinations were carried out on PWE and the controls in the Kiremba hospital. Informed verbal consent was obtained from each included subject.

The study was carried out under the auspices of Burundi's Ministry of Public Health. Materials and methods have been described elsewhere in details (Nsengiyumva et al., 2003).

Diagnostic criteria of epilepsy

We accepted the definition proposed by ILAE in 1993: “Epileptic seizure is a clinical manifestation presumed to result from an abnormal and excessive discharge of a set of neurons in the brain”; “Epilepsy is a condition characterized by recurrent (two or more) epileptic seizures, unprovoked by any immediate identified cause. Multiple seizures occurring in a 24-h period are considered a single event. Individuals who have had only febrile or neonatal seizures are excluded from this category”; “Active case of epilepsy is defined as a person who has had at least one epileptic seizure in the previous five years, regardless of any antiepileptic drug (AED) treatment” (Commission on Epidemiology and Prognosis, ILAE 1993).

In an attempt to determine the accuracy of the classification an EEG was performed on PWE with a portable digital device (Medatec Brainnet II). Those recordings were made according to the same standardized procedure and by a technician trained in the use of such a device. The recording was 15–20 min long. Electrodes were placed according to the international 10–20 system using standard procedures, with reference electrode as C1 or bilateral ear electrodes. Hyperventilation was performed in all cases. However, photic stimulation was unavailable and thus was not performed. The results of the recordings were digitally archived for later evaluation by reviewers who were blinded to clinical information, in the Limoges University Hospital's Neurophysiologic Department (Limoges, France).

Seizure types were identified on the basis of the classification proposed by ILAE in 1981 (Commission on Classification and Terminology of the ILAE 1981).

Clinical examination

A neurologist confirmed the diagnosis after a thorough history-taking and a clear neurological clinical examination. Motor and sensory problems, cerebella, pyramidal and extrapyramidal syndromes, as well as all cranial nerve palsies were thoroughly evaluated in both the cases and the controls. In order to exclude the presence of neurological disorders, for control subjects, the presence of abnormal physical signs was considered an exclusion criterion.

The physician looked for the dates of onset of seizure disorders, the date of the latest seizure, checked if the epilepsy was active (at least one seizure in the last five years), and listed the antiepileptic drugs used. In order to exclude the presence of neurological disorders also control subjects underwent a complete standard neurological examination.

Collecting data

Local physicians in Burundi gathered the data from both cases and controls. A standardized data collection sheet was translated into the local dialect and was applied to both cases and controls. This data sheet was highly inspired from the questionnaire for investigation of epilepsy in tropical countries (Preux et al., 2000). It gathered information on the subjects’ identity; demographics; history; family history, and occurrence of epilepsy in a first or second degree relative; the epilepsy risk factors (cranial traumas, severe childhood illnesses or adulthood illnesses); general information concerning lifestyle, sanitary habits, water supply, etc.

Serological evaluation

In the original case-control study carried out to assess the relationship between cysticercosis and epilepsy a 10-ml blood sample was obtained from each enrolled subject using nonanticoagulated Vacutainer tubes (Fisher Bioblock Scientific, Illkirch, France) in order to perform serologies (Nsengiyumva et al., 2003). Each sample was then centrifuged and the sera transferred into cryofreezing Nunc tubes (Fisher Bioblock Scientific, Illkirch, France) and immediately put into liquid nitrogen containers. Then, they were preserved in dry ice in an isothermal container ready to be flown to Limoges, France. After performing cysticercosis serology, the preserved aliquots of serum sample were shipped to the Infectious and Tropical Diseases Unit of the University of Florence, Italy, for the toxocara serological analysis. Laboratory personnel were blind to the case-control status of serum samples.

Anti-Toxocara-specific IgG were detected by an immunoblot using a commercial kit (Toxocara WB IgG, LDBIO Diagnostics, Lyon, France), as described by the manufacturer. Briefly, this assay uses Toxocara excretory/secretory antigens separated in two groups: low-molecular-weight antigen bands (LMW; 24–35 kDa) and high-molecular-weight bands (HMW; 132, 147, and 200 kDa) (Maizels et al., 1984; Magnaval et al., 1991). We considered a positive result only those samples reacting to two or more LMW bands.

Statistical analysis

Data was analyzed using Epi-Info 6.04 and STATA 6.0 software (Epi Info, 1994; STATA, 1999).

To estimate the sample size we considered a seroprevalence of Toxocara antibodies of 12%, that is the ratio that we found in controls in our previous study carried out in a rural area of Bolivia (Nicoletti et al., 2002). Considering a 5% level of significance and a power of 80% the minimum number of subjects needed for the study was 146 cases and 146 controls (ratio 1:1) to detect an odds ratio of 2.5.

Quantitative variables were described using mean and standard deviation (SD). The frequency comparisons were done with the Pearson's chi-square test.

Multivariate analysis using conditional logistic regression for a matched case-control study was performed to estimate adjusted ORs. Parameters associated with the outcome at the univariate analysis with a threshold of p = 0.10 were included in the model. The model was manually constructed using the likelihood ratio test (LRT) to compare the log-likelihood of the model with and without a specific variable. The possible interaction was also evaluated by the LRT (test of violation of proportional odds). For quantitative exposure, the test for linear trend was performed to evaluate the linear or trend effect. Subgroup analysis was conducted in those subjects with partial and with generalized epilepsy. Significance level was fixed to p ≤ 0.05.


Overall 191 cases and 191 age-matched controls were included in the case-control study to asses the relationship between epilepsy and toxocariasis. Out of the 191 PWE, 99 (51.8%) were men and 92 (48.2%) women, with a mean age at the time of the study of 25.1 ± 15.1 years (25.4 ± 15.9 for men and 24.8 ± 14.4 for women).

On the basis of the ILAE classification, considering both EEG and clinical data, 113 presented partial seizures (60.7%) while 73 (39.2%) generalized seizures. In five subjects seizures were unclassifiable. The mean age of subjects affected by partial seizures was 23.2 ± 15.9 years while it was slightly higher for those affected by generalized seizures 27.4 ± 12.7. Seventy-seven percent of PWE reported seizures onset before 20 years and the mean duration period was 9.0 ± 7.0 years. Out of the 191 selected PWE, only four presented abnormal signs at the neurological examination, one of whom presented mental retardation. The baseline characteristics of cases and controls are shown in Table 1.

Table 1. Baseline characteristics of cases and controls and univariate analysis
 CasesControlsOR*95% CIp-Value
  1. *OR, Matched OR.

Sex191100  191 100   
Sex (male) 9951.87237.71.721.15–2.590.008
 Christians (reference)15882.718596.91 
 Others 3317.3 6–16.40.001
 Farmer (reference)12766.513168.61 
 Others  9 4.74823.00.200.05–0.700.01 
 Not Active5528.812 6.35.991.99–24.50.002
 Family history of epilepsy64/16040.033/17219.23.051.74–5.350.000
Marital status
 Married (reference) 8142.410756.01 
 Widowed  9 4.7 6 3.12.530.79–8.780.11 
 Living alone  9549.77237.73.291.55–5.530.001
 Living with relatives  6 3.1 6–3.840.88 
 Indoor toilets1/190 0.57/184 3.770.85–5690.07 
 Seropositivity for T. canis114/19159.797/19150.81.50.97–2.310.07 
 Seropositivity for Cysticercosis113/19159.259/19130.93.41.91–6.04<0.001 


Among the 382 subjects examined (191 age-matched pairs), 211 (55.2%) presented antibodies against Toxocara. In particular out of the 191 PWE enrolled, 114 tested positive giving a seroprevalence for Toxocara antibodies of 59.7% (63.7% for partial epilepsy [73/113] and 54.8% for generalized epilepsy [40/73]), while 97 controls tested positive with a seroprevalence of 50.8%. The matched OR for T. canis seropositivity bordered on statistical significance (OR = 1.5; CI = 0.97–2.31; p-value = 0.07). Univariate analysis is shown in Table 1.

After adjustments on other variables according to a multivariable model using the conditional logistic regression, association between Toxocara seropositivity and epilepsy was stronger and significant with an adjusted OR of 2.13 (95% CI 1.18–3.83; p-value = 0.01). Results of the final model are shown in Table 2. When the analysis was restricted to those PWE with a normal neurological examination (187 subjects) and their matched-controls, we found a similar adjusted OR (OR 2.10; 95% CI 1.16–3.78; p-value = 0.01).

Table 2. Multivariate analysis (conditional logistic regression)
 Adjusted OR95%CIp-Value
Seropositivity for T. canis2.131.18–3.830.01 
 Farmer (reference)1 
 Others0.1300.2–0.66  0.02 
 Not Active10.5  2.14–51.80.004
 Christians (reference)1 
 Others5.7  1.77–18.40.004
 Seropositivity for Cysticercosis3.711.90–7.240.001

Family history of epilepsy, even if significantly associated with epilepsy, has not been included in the multivariate analysis due to the high percentage of missing values (50 observations). When the analysis was restricted to the 142 matched pairs without missing values and including “family history” in the model, we found a positive but not significant association between epilepsy and T. canis seropositivity (adjusted OR 1.63; 95%CI 0.77–3.4; p-value = 0.1). However it should be noted that the reduced sample size has probably led to a lack of power.

When the model was restricted to those subjects affected by generalized or partial epilepsies the OR for T. canis seropositivity (adjusted by occupation, religion, and cysticercosis seropositivity) bordered on statistical significance for generalized epilepsies (OR 2.52; 95%CI 1.01–6.26; p-value = 0.05), while was not significant for partial epilepsies (OR 1.83; 95%CI 0.82–4.07 p-value = 0.1).


Epilepsy is considered an important health problem in low-resource tropical countries where prevalence ranges from 2.47/1,000 to 57/1,000. The higher frequency of partial seizures in these countries compared to industrialized countries could be an indication of the high incidence of symptomatic epilepsy secondary to cortical damage (e.g., perinatal brain damage, head injury) (Senanayake and Román, 1993).

Toxocariasis is one of the most commonly reported zoonotic helminth infections in the world, occurring whenever the man-soil-dog relationship is particularly close. High seroprevalence of Toxocara infection have been found in developing countries where the humid climate favors the survival of parasite eggs in the soil and poor hygiene increases the probability of human infection (Magnaval et al., 1994).

The main objective of the study was to assess the relationship between epilepsy and toxocariasis.

Studies in animal models have demonstrated that T. canis larvae can enter the CNS. In small rodents larvae accumulate in the brain producing a variety of CNS disorders. Given these observations, neurological disorders in humans due to the presence of T. canis larvae in the CNS might not be an uncommon event (Magnaval et al., 1997). In fact, involvement of the CNS in human toxocariasis has been described and the invasion of the brain by nematode larvae has been demonstrated. Despite the higher seroprevalence of T. canis antibodies found in children with epilepsy versus controls in previous studies there has been doubt about whether this implicated causality (Woodruff et al., 1966; Glickman et al., 1979; Arpino et al., 1992). To our knowledge, few case-control studies have been carried out to assess a relationship between T. canis infection and epilepsy. In the first, a case-control study in children from the United States, a higher prevalence of antibodies against T.canis was found in PWE than in controls, but similar rates of Toxocara seropositivity were found in PWE with a known or suspected etiology (other than toxocariasis) and in PWE with unknown etiology (Glickman et al., 1979). In 1990 a case-control study was carried out in Italian children showing an association between T. canis seropositivity and epilepsy (Arpino et al., 1992). Further evidence of association has been found in a population based case-control study carried out by our group in a rural area of Bolivia where we found a significant association between T. canis seropositivity and epilepsy (OR 2.70; 95 CI %1.41–5.19). In this study a stronger association was found when the analysis was restricted to partial epilepsy (OR 4.70; 95% CI 1.47–15.10) while no evidence of association was found with generalized epilepsy (OR 1.79; 95% CI 0.76–4.17) suggesting that toxocariasis may in part explains the higher prevalence of epilepsy in developing countries (Nicoletti et al., 2002). These findings prompted us to add further information.

Concerning the validity of the study the Kiremba study was initially designed to investigate the association between cysticercosis and epilepsy and, in the initial sample of 972 subjects enrolled, no differences were found between the selected subjects and the characteristic of the Kiremba population (Nsengiyumva et al., 2003). From the initial sample of 972 subjects, the relationship between toxocariasis seropositivity and epilepsy was studied only on 191 age-matched pairs. The sample was not randomly selected from the original study group, in fact, we included in the study only those PWE and their age-matched controls for whom an additional aliquot of serum was available. When a single aliquot was stored, in fact, it was utilized to perform cysticercosis serology, primary endpoint of the study; the fact that one or two aliquots were stored was totally independent of the disease or the studied exposure (T. canis). Regardless, the 191 cases were otherwise similar to the original sample with respect to the main characteristics such as age, sex and religion; the only difference recorded was the higher frequency of partial epilepsies in the present sample.

Serologic examination for Toxocara infection was carried out using an immunoblotting method in all PWE and control subjects. Definitive diagnosis of CNS toxocariasis is provided only by histological demonstration of the parasite in biopsy material (Taylor and Holland, 2001). The most frequent entity involved in the differential diagnosis of subjects with late-onset epilepsy or inflammatory brain nodules is the larval stage of Taenia solium, etiological agent of neurocysticercosis (NCC). In a study carried out by our group, using the porcine cysticercosis infection model, we found a lack of cross-reactivity with anti-T. solium antibodies in the T. canis assay utilized in the present study (García et al., 2006). However, we would underline, that the presence of antibodies against T. canis detected in serum does not provide evidence of either an active systemic infection or a CNS involvement.

One of the major weaknesses of the study is the not population based design. In fact the medical personnel working in these towns had, in the weeks preceding the study, talked to the village leaders, the PWE, and their families about participation in the study. Thus PWE and controls actively went to hospital to participate in the study.

Furthermore, due to the retrospective nature of the study, the only design feasible in a rural area of a low resource country, the use of prevalent, rather than incident, cases does not allow us to be certain that exposure (T. canis infection) occurred before the outcome (epilepsy). Other studies have reported some evidence of association between T. canis and mental retardation. As mental retardation is frequent among PWE this could predispose them to Toxocara infection (Genchi et al., 1990; Huminer et al., 1992). Nevertheless in our sample the only one PWE who presented mental retardation did not presented antibodies anti T. canis.

In agreement with our previous study carried out in rural Bolivia we found a significant association between T. canis seropositivity and epilepsy (OR 2.13; 95% CI 1.18–3.83). However in the present study the association was not significant when analysis was restricted to partial epilepsies and bordered on statistical significance for the generalized ones probably due to a lack of power. In fact, concerning the sample size, due to the lack of data regarding the seroprevalence of T. canis in the general population of Kiremba, sample size was calculated assuming the seroprevalence (12%) found in a rural area of another low-resource country (Nicoletti et al., 2002). Consequently the higher seroprevalence found in the control group (54.8%) could have resulted in a loss of statistical power probably leading to failure to find significant differences when subgroup analysis was performed.

Our results, in agreement with those obtained in rural Bolivia, suggest that toxocariasis, a preventable disease, could play a role in the incidence of epilepsy in endemic areas. Prospective surveys using incidence rather then prevalent cases should be performed in order to confirm this association.


Acknowledgment:  The authors wish to thank Annie Prado-Jean for her logistical help.