Exposure to parasitic infections determines features and phenotypes of active convulsive epilepsy in Africa [version 2; peer review: 1 approved, 1 not approved]

Background : Epilepsy affects 70 million people worldwide, 80% of whom are in low-and-middle income countries (LMICs). Parasitic infections contribute considerably to the burden of epilepsy in LMICs, but the nature and presentation of epilepsy the proportion of severe clinical and electroencephalographic features of epilepsy attributable to parasitic infections. Results: A total of 997 participants with active convulsive epilepsy from the five African sites were analyzed, 51% of whom were males. Exposure to parasitic infections was associated with more frequent seizures in adult epilepsy (relative risk (RR)=2.58, 95% confidence interval (95%CI):1.71-3.89). In children, exposure to any parasite was associated with convulsive status epilepticus (RR=4.68, (95%CI: 3.79-5.78), intellectual disabilities (RR=2.13, 95%CI: 1.35-3.34) and neurological deficits (RR=1.92, 95%CI: 1.42-2.61). Toxoplasma gondii and Onchocerca volvulus interacted synergistically to increase the risk of status epilepticus (RERI=0.91, 95%CI=0.48-1.35) in the data pooled across the sites. Exposure to parasitic infections contributed to 30% of severe features of epilepsy as shown by PAF. Conclusions: Parasitic infections may determine features and phenotypes of epilepsy through synergistic or antagonistic interactions, which can be different in children and adults. Interventions to control or manage infections may reduce complications and improve prognosis in people with epilepsy. Competing Interests: I have no competing interests to report. The authors aimed to examine if exposure to common central nervous system (CNS) infections may play a role in the early onset of epilepsy, convulsive status epilepticus, frequent seizures, abnormal electroencephalograph (EEG) findings, intellectual disabilities, neurological deficits as well as other medical comorbidities of epilepsy e.g. malnutrition. The population consisted of 1,711 patients diagnosed with epilepsy in the five African countries. 997 patients were defined as respondents for test of infection. Respondents were tested for Toxacara canis , Onchocerca volvulus , Plasmodium falciparum , Toxoplasma gondii , HIV and Taenia solium (exposed). infection status was determined through analysis of blood samples to detect antibodies to the infectious parasites and HIV. These assays to the infectious parasites and HIV were performed on 986 people with epilepsy. Although these infections could affect the central nervous system, they could not be defined as (CNS) cases


Introduction
Epilepsy is a common serious neurological disorder, with an estimated 70 million people being affected, 90% of whom live in low-and-middle-income countries (LMICs) (Ngugi et al., 2010). Global incidence of epilepsy is up to 278 per 100,000 people, while global prevalence is up to 11 per 1,000 people (Fiest et al., 2017), with both incidence and prevalence being higher in LMICs than high-income countries (HICs) (Newton & Garcia, 2012). The higher burden of epilepsy in LMICs can be attributed to the higher incidence of infections such as malaria and human immunodeficiency virus (HIV), birth-related injuries and low accessibility to health care (Newton & Garcia, 2012).
A study carried out in selected African populations indicated that exposure to multiple parasites was associated with the risk of epilepsy (Kamuyu et al., 2014). However, this study did not determine the contribution of infections to features and phenotypes of epilepsy and examine response to anti-seizure medications (ASM) (Kariuki et al., 2014). It is important to describe the features and phenotypes of epilepsy ascribable to exposure to infections since most infections are preventable through simple public health interventions. Infections such as malaria and HIV can increase the risk for status epilepticus (Kariuki et al., 2015) and frequent seizures, both of which worsen prognosis of epilepsy. Early onset epilepsy, which is common in Africa, may be caused by infections, which affects children because of undeveloped immunity and brain. Infections may also damage the brain resulting in comorbidities of severe epilepsy such as intellectual disability and neurological deficits, which can affect adherence to ASM (Ibinda et al., 2017;Mbuba et al., 2012). Additionally, exposure to infections may complicate biomedical treatment of epilepsy. It important to determine the burden of severe epilepsy that could be prevented by control of infections. Findings from such studies may be useful in informing policy makers on the impact that control of these infections may have on the burden of epilepsy, particularly towards efforts to reduce the clinical and psychosocial burden of epilepsy.
We performed an analysis of people with epilepsy with well-defined clinical features to examine if exposure to common infections with a propensity for the central nervous system (CNS) may play a role in the early onset of epilepsy, convulsive status epilepticus, frequent seizures, abnormal electroencephalograph (EEG) findings, intellectual disabilities, neurological deficits as well as other medical comorbidities of epilepsy e.g. malnutrition.
We computed the proportion of severe epilepsy that is attributable to these infections, which allows us to estimate the clinical and psychosocial burden of epilepsy that would be eliminated following the control of these infections.

Ethical considerations
The study was approved by the ethics committees or ethics review boards in each of the participating countries (Clearance number: M080455 for Agincourt, South Africa; IHI/IRB/No A 70-2009 for Ifakara, Tanzania; Reference number: HS 663 for Iganga, Uganda; SSC number 787 for Kilifi, Kenya; FWA number 00011103 for Kintampo, Ghana;). All participants or guardians gave written informed consent before responding to study-related questions.

Study site and population
The study was carried out in sites in five African countries with Health Demographic Surveillance Systems (HDSS), which are part of the International Network for the Demographic Evaluation of Populations and Their Health (INDEPTH): Agincourt, South Africa; Kintampo, Ghana; Kilifi, Kenya; Ifakara, Tanzania; and Iganga-Mayuge, Uganda. These sites collect health and vital demographic data from residents on a regular cyclical basis. The study population for each HDSS was described in detail previously (Ngugi et al., 2013).

Study design
People with epilepsy were identified from a 3-stage community cross-sectional survey carried out between 2008 and 2011 (Ngugi et al., 2013). Screening for history of convulsions was conducted in stage I, followed by administration of a detailed questionnaire to describe seizures in stage II, and a neurological assessment by a clinician to diagnose epilepsy in stage III. Formal validation of this three-stage methodology is published elsewhere (Ngugi et al., 2012), and findings supported need for continued efforts to develop and improve case-ascertainment methods in population-based epidemiological studies of epilepsy in LMIC. Community members diagnosed with epilepsy were tested for antibodies to HIV and infectious parasites (Toxacara canis, Onchocerca volvulus, Plasmodium falciparum, Toxoplasma gondii, and Taenia solium) and formed the sample for this case-comparison study. The exposure was any of the listed infections and the outcomes were the presenting features of epilepsy. All study materials can be found as extended data (Langat et al., 2021).

Amendments from Version 1
We have revised the useful comments from the reviewers and made the following changes to the revised article. We have clarified that some parasitic infections may not directly impact the brain, and so have used an appropriate term "infections with a propensity for the central nervous system" to refer to such parasitic infections. The strengths and limitations section now acknowledges that it may be difficult to make causal inference in cross-sectional studies, but findings from these studies may be very useful in generation of new or important hypothesis for understanding pathophysiology or prognosis of health conditions when this is not previously studied or understood. The statistical analysis section also clarifies that relative risks approximate prevalence ratios in cross-sectional studies, and may be considered appropriate when outcomes are common and short-lived e.g. exposure to infections. Severity and survival bias, which may be possible in cross-sectional studies, are now acknowledged in the strengths and limitations section. Other responses to the reviewers' comments, include clarifications that multivariable modified Poisson regression model was used to compute relative risks; that samples for assays were collected when participants were reportedly healthy and that the variable "focal seizures" considers "generalised seizures" as the comparison group.

Sampling
The population consisted of 1,711 patients diagnosed with epilepsy in the five sites. Respondents were tested for Toxacara canis, Onchocerca volvulus, Plasmodium falciparum, Toxoplasma gondii, HIV and Taenia solium (exposed). HIV was tested because it is often cormobid with these parasitic infections. Sample size calculation showed this study needed 386 participants to determine the association between features of epilepsy and infections at 80% power and 95% level of confidence (supplementary appendix 1). However, we utilized all data available (997 patients) in this study.

Procedure
Clinical assessments and definitions. Most clinical and EEG features, treatment and detrimental consequences were obtained through clinical examination and EEG recordings performed by trained clinicians, and questionnaires. The clinical and neurophysiological features of epilepsy have been reported previously (Kariuki et al., 2014;Kariuki et al., 2016). Epilepsy was defined as ≥2 unprovoked seizures occurring at least 24 hours apart and was classified as active if at least one seizure occurred within the previous 12 months (Thurman et al., 2011).
Briefly, convulsive status epilepticus was defined as seizures lasting >30 minutes or intermittent seizures lasting for 30 minutes without recovery of consciousness. Seizure frequency was categorized into daily (at least one each day), weekly (at least one a week), monthly (at least one a month), and yearly (at least one a year). An EEG was categorized as abnormal if there was evidence of an abnormal background, focal changes, interictal epileptiform activity or an abnormal response to either of the activation procedures (hyperventilation or photic stimulation). Seizures were classified as focal, generalized, or other. Intellectual disability was assessed by a clinician observing young children who had problems performing the standardized test of a locally adapted developmental inventory, or by assessing an adult's awareness of person, place and time. Malnutrition was defined as a weight-for-age z score value of −2 or lower or a mid-upper arm circumference less than 11.5 cm. Neurological deficits were defined as an inability to do things and walk or sit upright if of an appropriate age. Severe epilepsy was considered in a person with epilepsy plus focal seizures, frequent seizures, abnormal EEG, neurological deficits or intellectual disability.
Serological assay analysis. Data on assays for parasitic and HIV infections were obtained from the work by Kamuyu and colleagues (Kamuyu et al., 2014). They determined infection status through analysis of blood samples to detect antibodies to the infectious parasites and HIV. The kits and techniques used, criteria for cut-off and sensitivity and specificity have been described (Kamuyu et al., 2014). These assays were performed on 986 people with epilepsy.

Data management and statistical analysis
The data was exported from MS Excel to STATA version 15 for sampling and analysis. Data cleaning was done to assess completeness of data and exclude duplicates. For continuous data (e.g. age), histograms were plotted to show distribution. Mean (standard deviation (SD))/median (Inter-quartile range (IQR)) were reported depending on the nature of the distribution of data (i.e. whether or not the data is normally distributed). Histogram plots and Shapiro-Wilk's test of normality were performed to check for normality of data, followed by transformation where appropriate. Categorical data such as sex, education, HIV status were illustrated using bar charts, with associated frequencies and proportions presented in tables.
Student's t-test was used to compare means of titers of infection in those with severe features of epilepsy and those without, if the raw or transformed data was normally distributed. For non-parametric continuous data, Mann-Whitney U test was used. The association between parasitic infections (predictors) and features of epilepsy (outcomes) was determined using a mixed effects modified Poisson regression model with robust standard errors. Since our outcomes are binary, using robust standard errors rectified overdispesion and produced reliable estimates. Modified Poisson regression model was chosen over log-binomial model for two reasons: (i) it is not largely affected by convergence problems, and (ii) it allows for addition of random effect(s) variable while log-Binomial model does not. The model uses a log function to link a binary outcome (feature of epilepsy) to a set of predictors (infections and confounders), while accounting for clustering effect of site. To obtain crude relative risks, simple modified Poisson regression model with site as a random effect was used. Adjusted relative risks were obtained using multiple modified Poisson regression models with site as a random effect. We chose relative risk over odds ratio (OR) since the prevalence of some features was >10%, to avoid exaggerating risks with OR. Relative risks approximate prevalence ratios in cross-sectional studies, and are considered appropriate when outcomes are common and short-lived e.g. exposure to infections. In this model, the outcomes were the various clinical, or neurophysiological features of epilepsy (focal seizures, status epilepticus, frequent seizures [daily/weekly], neurological deficits, abnormal EEG, cognitive difficulties, malnutrition) and predictors were the parasitic infections most commonly associated with epilepsy (Toxacara canis, Onchocerca volvulus, Plasmodium falciparum, Toxoplasma gondii, Taenia solium, and HIV (included because increases risk for these parasites)) combined or considered individually together. The model was accounted for age, sex, other socioeconomic confounders and ASM use. Biological (additive) interaction was assessed to determine public health implications of co-infections on features of epilepsy. Biologic interaction is more informative than the multiplicative interaction in the sense that the former gives the additive effect (synergism/antagonism) of exposures on the feature of epilepsy, by determining the relative excess risk due to interaction (RERI). RERI enabled us to know whether the combined effect of the exposures exceeds (synergism) or is less than (antagonism) the effect of each considered individually. RERI was determined as follows: Additive interaction (RERI) = (p 11 -p 00 )-[(p 10 -p 00 )+(p 01 -p 00 )] (1) Where; p 00 -Prevalence of the epileptic feature when both infections are absent; p 11 -Prevalence of epileptic feature when both infections are present; p 10 and p 01 -Prevalence of epilepsy feature when one infection is present.

General description
A total of 997 participants with active convulsive epilepsy from the five African sites were included in the analysis. The overall proportion of males was 51% across all the samples, ranging from 48% in Ifakara to 56% in Agincourt. The overall median (IQR) age was 20.0 years (13.0 -32.5), being lowest in Iganga (12.0, IQR: 5.0-21.0) and highest in Agincourt (29.0, IQR:18.0-45.0). There was a significant difference in age (P=0.0001) across the sites (supplementary figure 1). The distribution of these and other sociodemographic characteristics including education and employment status are shown in Table 1 and most characteristics varied by site.

Associations of parasitic and HIV infections with features of epilepsy among all participants
In the crude analysis, presence of infections as shown in Figure 1, was associated with convulsive status epilepticus (RR=2.79, 95% confidence interval (95%CI):2.53-3.08) and neurologic deficits (RR=1.73, 95%CI:1.35-2.22). After adjusting for age, sex, education and employment, presence of infections     (Table 7).

Interaction between infections and features of epilepsy
In the analysis for multiplicative interactions, several paired infections showed interactions (Table 8 and supplementary  table 1 In the analysis for biological interactions, T. gondii and T. canis interacted synergistically to increase the risk of

Discussion
This study provides important evidence that parasitic infections that have a propensity for the CNS are not only associated with epilepsy, but also determine the occurrence of features and phenotypes of epilepsy. In children with epilepsy, exposure to infections is particularly associated with convulsive status epilepticus, neurological deficits and intellectual disability, while in adults it is associated with frequent seizures. Individual infections may increase some features (e.g. falciparum malaria increasing risk for convulsive status epilepticus) but not others. These infections may interact synergistically or antagonistically to affect the features of epilepsy, and their prevention may reduce up to 26% severe epilepsy.
Antibody titres and prevalence for parasitic/HIV infections There was significant difference in median antibody titer levels between severe and non-severe epilepsy only for HIV (P<0.001) and Onchocerca volvulus (P=0.038), suggesting that these levels are ubiquitous representation of exposure, and reliable associations should be measured from cut-offs derived from high antibody titres. Using predefined cut-offs, infections were more prevalent across all features considered, indicating that these infections could be potential risk factors for development of these epilepsy features. The prevalence of infections were significantly greater than in the general population since they are based on an epilepsy cohort, whose exposure to infections is already known to be relatively greater (Carter et al., 2004).

Associations between any parasitic/HIV infections and features of epilepsy in all populations
Among all the study population, exposure to any of the infections were associated with increased risk of having frequent (daily/weekly) seizures and an increased risk of cognitive difficulties, after controlling for confounding effects, including the use of ASM. HIV-positive individuals have increased risk of abnormal EEG, which has been noted in previous studies (Kellinghaus et al., 2008). T. canis was associated with neurological deficits which is not surprising since neurological involvement has been reported (Finsterer & Auer, 2007). Associations that appeared protective were observed especially for some specific infections. This may be due to absence of these features and phenotypes when samples for positive assays were collected or may be a spurious association. Since susceptibility to infections and pathogenesis of epilepsy following these infections is different between children and adults, it warranted evaluations of these associations between two population strata i.e. adults and children/minors separately.

Associations between any parasitic/HIV infections and features of epilepsy in children
Among children with epilepsy, we found that exposure to infections was associated with higher risk of having status epilepticus, intellectual disabilities and neurologic deficits after adjusting for age and sex. Analysis of individual infections showed that P. falciparum increased the risk for convulsive status epilepticus, and T. canis that for neurological deficits, while interaction of several infections may explain the intellectual disability.
Children have low immunity to infections and also have a developing brain, and it is not surprising that infections Table 5   determine three features of epilepsy that are pathognomonic of severe childhood seizure disorders. Convulsive status epilepticus is common in children and previous studies showed that malaria is the most important cause (Sadarangani et al., 2008). Kariuki et al. in 2011 hypothesized that the non-malarial status epilepticus that did not decline over a period of malaria reduction was probably caused by epilepsy (Kariuki et al., 2011). It has been established in a recent multisite study that status epilepticus occurs in over 25% of people with epilepsy (Kariuki et al., 2015), and infections may contribute to this excess burden of prolonged seizures.

Associations between any parasitic/HIV infections and features of epilepsy in adults
Among adults with epilepsy, exposure to any infection was associated with increased risk for frequent seizures, but reduced risk for intellectual disability. Analysis of individual impairments shows that the increased risk for frequent seizures, status epilepticus was determined by falciparum malaria, which appeared to reduce the risk for intellectual disability in adults, perhaps suggesting these are related to other causes. Status epilepticus was also increased following O. volvulus, with other infections showing protection against some of these features.
The positive association of infections with frequent seizures is expected, since we have demonstrated in children that repetitive seizures are highly attributable to malaria (Kariuki et al., 2011). Presence of any infections were protective for intellectual disability in adults with epilepsy perhaps because of absence or low frequency of infections in this group of people movement of adults with cognitive difficulties is restricted indoors, resulting in less exposure to infections. Alternatively, it may be due to reporting bias whereby adults with epilepsy conceal their cognitive impairments, which is easily recognized as a condition for children (Carter et al., 2005) or due to counselling out of effect sizes of infections that confers susceptibility by those that reduces risk.

Association of individual parasitic/HIV infections with features of epilepsy
When we probed the effects of individual infections on the features of epilepsy, P. falciparum was protective for most epilepsy features, except for status epilepticus. The same case applied to children, but in adults, P. falciparum infection was associated with increased risk of both status epilepticus and daily/weekly seizures. This finding agrees with that of another study done in Africa, where infection with falciparum malaria were important risk factors for status epilepticus (Kariuki et al., 2015). We also found that exposure to falciparum malaria may in part determine the risk for malnutrition in those with epilepsy, supporting the previous association between malaria and malnutrition in general hospital admissions (Gone et al., 2017), but other factors involved in this association should be considered. HIV was a risk factor for frequent daily/weekly seizures, and abnormal EEG, suggesting it has neurological involvement. Seizures have been reported in HIV with or without epilepsy (Kellinghaus et al., 2008); frequent seizures in our study are likely due to epileptogenic nature of HIV since its negative association with neurological deficits appears to rule out contribution by brain damage. O. volvulus was protective for focal seizures and frequent seizures but increased the risk for status epilepticus. O. volvulus likely has diffuse brain involvement resulting in convulsive status epilepticus with a generalized onset in the brain, thus the negative association with focal seizures. Taenia solium increased the risk of having abnormal EEG but reduced the risk of early onset seizures. The finding that T. solium appeared to reduce the risk of early onset seizures may be because it is more common in adult onset epilepsy (Ngugi et al., 2013).

Interaction of parasitic/HIV infections and features of epilepsy
More interactions were noted in the additive scale than in the multiplicative scale, but the later would only be helpful in examining the contribution of a risk factor to a disease. Interaction on the additive scale has public health importance in that it is easier to quantify the frequency of a disease that would be controlled by the removal of risk factors. The many significant combinations of interaction in the multiplicative scale suggests that multiple infections determine the features and phenotypes of epilepsy in Africa. Interaction tests performed on additive scale revealed that infections interact either synergistically or antagonistically in influencing the features of epilepsy. For instance, Toxoplasma gondii and Toxocara canis interacted synergistically to increase the risk of status epilepticus, implying that the number of status epilepticus cases due to the coinfection is higher than the total number of cases contributed by the two infections separately. In contrast, P. falciparum and HIV interacted antagonistically towards status epilepticus, implying that co-infections worsen the prognosis of epilepsy, and can also be found in those with milder features. These interaction findings support Table 7. Adjusted associations of infections and features of epilepsy among adults (≥18 years).

Strengths and limitations
Strengths: This study had more than the minimum required sample size, meaning the study had enough power to answer our research questions. Also, being a multisite design, the findings can be generalized across ecological zones considered. Additionally, the mixed effects model used in this study took accounted for the correlations (homogeneity) within sites and heterogeneity across sites. Limitations: We only studied active convulsive epilepsy. The infections were assayed for exposure and not really a measure of impact of severe disease on epilepsy, which would be done with longitudinal cohort studies. Another cause of concern was the possibility of recall bias since participants were interviewed for some epilepsy features and the inherent challenge of inferring direction of causality for cross-sectional studies. Cross-sectional studies, however, are easy and inexpensive, and may be very useful in generation of new or important hypothesis for understanding pathophysiology or prognosis of health conditions if these are not previously studied or fully understood. Unmeasured risk factors could contribute to the association between infections and epilepsy features i.e. residual confounding. Some of those who agreed to participate in the follow-up evaluations may have had severe forms of epilepsy than those who dropped out, introducing severity bias which can affect the associations. Some participants who did not participate may have been unwell or even died, which can introduce survival bias.

Conclusions
Features and phenotypes of epilepsy in Africa may be determined by infections and the direction of association depends on the type, frequency of infection and their interaction. Exposure to parasitic infections determine different types of features and phenotypes for children and adults, with status epilepticus and cognition being important in children and frequent seizures in adults. Taken together these findings suggests that not only are infections associated with epilepsy, they also determine the presentation of epilepsy in terms of features and phenotypes. Control of infections may reduce the risk of epilepsy becoming severe in addition to preventing epilepsy directly caused by infections in the community. Findings from this cross-sectional study are likely to inform set up of cohort studies to explore differential progression of epilepsy following exposure to infections, and to inform mass prophylaxis against exposure to infections among people already living with epilepsy.

Previous publications
Part of this data was published in PLoS Negl Trop Dis (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4038481/) and in Lancet Neurol (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3581814/) as case-control studies to examine exposure to infections as risk factors for developing epilepsy. However, none of these two articles explored the role of exposure to infections in the occurrence of features, outcomes and phenotypes among people with epilepsy, which is now presented in this paper. More details on justification for the current analysis is provided in the introduction section of the paper. This project contains the following underling data:

Data availability
(i) epilepsydata_deidentified_ID.tab (used for calculating the association of exposure to infections with epilepsy features and phenotypes) For ethical considerations and to comply with the language in the informed consent forms, the access to the dataset is controlled for general research use. Please request to access the data by completing the forms in the provided link to the data, following which permissions will be granted using procedures governed by the Data Governance Committee (DGC) of the Centre for Geographic Medicine Research Coast, Kenya Medical Research Institute. Data will be made available with the approval of the KEMRI Wellcome Trust Research Programme Data Governance Committee, only where anonymization can be adequately achieved to protect the privacy and confidentiality of the participants/respondents and any mentioned individuals and institutions, and where the proposed use is seen as relevant to the nature of the data. Where the DGC recommend this, the national KEMRI Science and Ethics Review Unit may also be asked to approve the proposed use. Conditions for data sharing are outlined in a KWTRP Data Sharing Agreement, including that the requestor shall use the data only for the agreed purpose as stipulated in the application form; shall agree to keep the data strictly confidential and shall not in any way attempt to seek to discover the identity of data subjects, to compromise or infringe on their privacy and confidentiality of their information.

Yuelian Sun
Department of Neurology, Aarhus University Hospital, Aarhus, Denmark The authors aimed to examine if exposure to common central nervous system (CNS) infections may play a role in the early onset of epilepsy, convulsive status epilepticus, frequent seizures, abnormal electroencephalograph (EEG) findings, intellectual disabilities, neurological deficits as well as other medical comorbidities of epilepsy e.g. malnutrition.
The population consisted of 1,711 patients diagnosed with epilepsy in the five African countries. 997 patients were defined as respondents for test of infection. Respondents were tested for Toxacara canis, Onchocerca volvulus, Plasmodium falciparum, Toxoplasma gondii, HIV and Taenia solium (exposed). infection status was determined through analysis of blood samples to detect antibodies to the infectious parasites and HIV. These assays to the infectious parasites and HIV were performed on 986 people with epilepsy.
The main concern of this study is that the authors might be unable to make causal inference (between parasitic infections and risk of epilepsy) using the data they had.
Although these parasitic infections could affect the central nervous system, they could not be defined as central nervous system (CNS) infections.
Even for a descriptive study, for example, parasitic infections among these epilepsy cases, the authors should be aware of the representativeness of the 1,711, response rate (why some cases responded and some did not), and differences among countries.

Minor comments:
In the background of the abstract, should 'epilepsy foutcomes' be 'epilepsy outcomes'? ○ Table 2. The total number, for example according to feature of focal seizure, is 996. How about those with missing value? ○ Is the work clearly and accurately presented and does it cite the current literature? seropositivity may vary if measured at a slightly different than when we did, and because it is difficult to pinpoint when the exposure occurred. However, crosssectional studies, which easy and inexpensive, may be very useful generation of new or important hypothesis for understanding pathophysiology or prognosis or health conditions when this is not previously studied or understood 1 . For instance, many previous studies only explored the etiological role of infections in development of new onset epilepsy, with no literature on how the persistent exposure after development of epilepsy changes the course of the disease, including even increasing the odds for severe disability and premature mortality. In addition, the analytical approach in this cross-sectional study aims at computing prevalence ratios that approximates relative risks in cohort studies, when outcomes are common and short-lived e.g. acute infections 2 . Findings from this cross-sectional study are likely to inform set up of cohort studies to explore differential progression of epilepsy among persons exposed to infections, and to inform mass prophylaxis against exposure to infections among people already living with epilepsy. These observations are included in the statistical analysis section, limitations section and conclusion of the revised manuscript.
Although these parasitic infections could affect the central nervous system, they could not be defined as central nervous system (CNS) infections. Response: We agree that some parasites e.g. Onchocerca volvulus might not directly affect the brain, but rather it is the inflammation provoked by the death of the parasite and the subsequent release of Wolbachia, for example 3 , that may impact the brain indirectly. We have therefore revised the term "central nervous system (CNS) infections" to "infections with a propensity for the central nervous system (CNS)", which we hope is more acceptable.
Even for a descriptive study, for example, parasitic infections among these epilepsy cases, the authors should be aware of the representativeness of the 1,711, response rate (why some cases responded and some did not), and differences among countries. Response: We agree that selection bias may be introduced by non-participation of participants. Some of those participated in the follow up evaluations may have had severe forms of epilepsy than those who dropped out, introducing severity bias which affect the associations that premised on severe forms of epilepsy as outcomes. Although there are many reasons for choosing not to participate in follow ups evaluations e.g. non-preference for biomedical facilities, other possibilities may be that some participants were unwell or even died, which can introduce survival bias. These possibilities are acknowledged in the limitation section of the revised manuscript.

Minor comments:
In the background of the abstract, should 'epilepsy foutcomes' be 'epilepsy outcomes'? ○ Response: Sorry for this typographical error, which has been revised as suggested.