Prevalence and factors associated with convulsive status epilepticus in Africans with epilepsy

Objective: We conducted a community survey to estimate the prevalence and describe the features, risk factors, and consequences of convulsive status epilepticus (CSE) among people with active convulsive epilepsy (ACE) identified in a multisite survey in Africa. Methods: We obtained clinical histories of CSE and neurologic examination data among 1,196 people with ACE identified from a population of 379,166 people in 3 sites: Agincourt, South Africa; Iganga-Mayuge, Uganda; and Kilifi, Kenya. We performed serologic assessment for the presence of antibodies to parasitic infections and HIV and determined adherence to antiepileptic drugs. Consequences of CSE were assessed using a questionnaire. Logistic regression was used to identify risk factors. Results: The adjusted prevalence of CSE in ACE among the general population across the 3 sites was 2.3 per 1,000, and differed with site (p < 0.0001). Over half (55%) of CSE occurred in febrile illnesses and focal seizures were present in 61%. Risk factors for CSE in ACE were neurologic impairments, acute encephalopathy, previous hospitalization, and presence of antibody titers to falciparum malaria and HIV; these differed across sites. Burns (15%), lack of education (49%), being single (77%), and unemployment (78%) were common in CSE; these differed across the 3 sites. Nine percent with and 10% without CSE died. Conclusions: CSE is common in people with ACE in Africa; most occurs with febrile illnesses, is untreated, and has focal features suggesting preventable risk factors. Effective prevention and the management of infections and neurologic impairments may reduce the burden of CSE in ACE.

most of the available hospital-based studies probably underestimate the burden. The incidence of CSE in children admitted to a Kenyan hospital was 2-5 times that in London and 6% of these cases occurred in children with established epilepsy 6,7 ; many do not attend hospital.
Risk factors for CSE are widely studied, 8 but it is unclear whether these differ in ACE in Africa. The few available studies from developed countries have focused on epilepsyrelated risk factors of CSE 1 ; other infectious risk factors of CSE may be important among people with epilepsy from low-income countries, where the etiology of epilepsy is different and the treatment gap is high. 9 The clinical features of CSE and medical and psychosocial consequences may also differ.
We documented the medical history of CSE among people with ACE in 3 African sites to estimate the lifetime prevalence, determine associations with risk factors and clinical features, and examine case fatality, burns, and social factors. We further examined whether there was heterogeneity in clinical features and risk factors across different ecological sites.
METHODS Study population. Cases of CSE were identified in community surveys of ACE conducted between August 2008 and April 2011 as part of a multisite study of epilepsy. 5 The present analysis includes 3 sites: Agincourt, South Africa; Iganga, Uganda; and Kilifi, Kenya (http://www.indepth-network.org/), where reliable histories of CSE among people with ACE were obtained. Histories of CSE among people with ACE were obtained from the participants, relatives, or witnesses, with further questions on whether CSE was associated with a febrile illness or fever. A clinician obtained data on sociodemographic characteristics and medical history and performed a neurologic examination. Experienced EEG technicians performed a 16 leadchannel EEG using the standard 10-20 montage system (figure 1), read by an experienced EEG technician (E.C.) in consultation with a neurophysiologist (S.W.), who reviewed some recordings to confirm consistency. We followed up individuals with CSE for a further 3 years to observe survivorship.
Definition of terms. We defined CSE as a history of convulsive seizures lasting for 30 minutes or more with impaired consciousness, or intermittent seizures lasting for 30 minutes or more without regaining consciousness in between the seizures. 10 The definitions utilized culturally appropriate events to time the seizures for those without watches or mobile phones, such as whether a seizure lasted longer than the time taken to boil a pot of maize, duration of a news program, or religious sermon, all of which last $30 minutes.
We classified epileptic seizures as focal, generalized, or others according to recent International League Against Epilepsy recommendations. 11 We categorized seizure frequency into daily, weekly, monthly, or yearly.
We defined children as those aged 0-18 years. We defined a perinatal adverse event as delay in crying, breathing, or breastfeeding after birth, but was included only in those aged 18 years and below, to reduce recall bias. We defined acute encephalopathy as coma related to an acute malarial, viral, or bacterial illness. Cognitive impairment was based on ability to follow standardized instructions during neurologic examination. We defined malnutrition as height for age z scores below 22 for those 0-10 years, body mass index (BMI) (weight/height 2 ) in the lower 5th percentile for those 11-19 years, and BMI ,18.5 for adults. 2,12 Investigations. In a random proportion of people [selected by RAND () command in MySQL (Oracle, Redwood Shores, CA)] with ACE and a history of CSE (figure 1), G.K. analyzed blood for antibodies to Onchocerca volvulus, HIV, Taenia solium, Toxocara canis, Plasmodium falciparum schizonts, and Toxoplasma gondii. The commercial kits, technique used, cut-off criteria, and sensitivities and specificities associated with detection of each organism were described previously. 13 Adherence to AEDs was measured in blood using a technique described previously. 14

Figure 1
Investigations performed in convulsive status epilepticus in active convulsive epilepsy People with convulsive status epilepticus (CSE) in active convulsive epilepsy (ACE) who had EEG performed, and antiepileptic drug (AED) blood levels and antibody titers to parasitic infections and HIV determined. ACE 5 active convulsive epilepsy; IgG 5 immunoglobulin G.
Statistical analysis. We estimated the prevalence of CSE associated with ACE for each center by dividing the number of confirmed cases of CSE by the total number of individuals screened in the survey. The prevalence was adjusted for the sensitivity of the methodology used to identify people with ACE. 15 In Kilifi, we compared the community estimates with admissions during or prior to the survey using the Kilifi Health and Demographic Surveillance System database, which offers a framework for linking hospital and population data for each person. 16 We compared those with CSE in ACE with those with ACE without CSE. Pearson x 2 test was used to compare proportions while Student t test or Mann-Whitney, where appropriate, was used to compare continuous variables between groups. Up to 774 people with ACE without CSE were included in the risk factor analysis (table e-1 on the Neurology ® Web site at Neurology.org). We used a logistic regression model that included age, sex, site, education, and marital status of parents to compute odds ratios (ORs) for each risk factor. All infections were entered together into the adjusted logistic model to examine their independent association with CSE in ACE. We considered a p value of #0.05 significant. Consequences of CSE in those with ACE were compared first with those with ACE without CSE and then with age group-matched community controls without ACE (n 5 1,025 for the 3 sites) (    Prevalence of convulsive status epilepticus associated with active convulsive epilepsy by age group in Africa The cumulative prevalence of convulsive status epilepticus in active convulsive epilepsy increased up to 18 years then declined steadily across 3 sites in sub-Saharan Africa.
Risk factors associated with status epilepticus. Several factors were associated with a higher probability of having CSE in those with ACE compared to those with ACE without CSE in the pooled analysis across all the sites. These included history of acute encephalopathy, neurologic and learning impairments, and presence of exposure to malaria and HIV in plasma (tables 2 and 3

DISCUSSION
The adjusted prevalence of CSE in people with ACE in Africa is high (2.3 per 1,000), and the true burden of all CSE may be greater, as these estimates only represent that associated with ACE, which does not include acute symptomatic seizures and deaths after an episode of CSE. CSE was associated with history of acute encephalopathy, an absence of previous hospitalization, underlying neurologic impairments, and presence of antibody titers to falciparum malaria and HIV. Risk factors associated with CSE in ACE differed across the sites. CSE is more common in Africa than in other parts of the world. 17 Our estimates of CSE associated with epilepsy are comparable with those reported in studies from the United States (0.5-2.5 per 1,000), 1,18 where nonconvulsive epilepsies may have been included in some studies and case ascertainment may have been better as they are less hampered by logistical constraints. The prevalence was highest in Iganga, where family history of seizures was an important risk factor, suggesting shared environmental or genetic factors. Use of AEDs and hospital services in this study were low, suggesting the need to apply educational interventions to encourage use of health services for management of prolonged seizures. 19 Among people with CSE in ACE, seizures began earlier in those with febrile illnesses than in those without, probably related to infectious causes of CSE in childhood. 7,20 An early occurrence of CSE for those with a history of febrile illness may also suggest a genetic susceptibility for acute febrile seizures. 21,22 The difference in age at onset between sites suggests different age-dependent risk factors, some of which may increase case fatality rate, particularly in children with CSE. 4 People with ACE with CSE in Agincourt and Iganga had a particularly high proportion of generalized seizures, but a third had focal EEG features and could have rapidly generalized. Generalized prolonged seizures are thought to be associated with a good outcome with appropriate treatment, 23 while focal seizures may be a marker of underlying preventable causes or regional atrophy on MRI caused by prolonged seizures. 24 Neurologic and cognitive deficits were important risk factors of CSE in those with ACE, and could be markers of underlying brain damage. A history of acute encephalopathy was associated with CSE, as would be expected in those hospitalized with CNS infections, 7 also supported by significant presence of antibodies to falciparum malaria and HIV. Previous use of hospital facilities appeared protective against CSE, and this may be explained by some people who are likely to live near the hospitals or adhere to AEDs. Visiting a traditional healer may have delayed conventional treatment, with seizures leading to CSE in some people with ACE.
Exposure to falciparum malaria and HIV was associated with CSE in ACE and these 2 infections may interact. 25 We previously found that over half of hospitalized CSE cases in children was related to malaria 7 and recently demonstrated that over 90% episodes of CSE in parasitemic children are caused by malaria. 26 The differences in risk factors across sites suggest that specific preventative strategies and treatment options should be applied for different sites; alternatively, this Table 3 Pooled univariate and adjusted analysis of the medical and laboratory risk factors associated with convulsive status epilepticus among people with active convulsive epilepsy for all sites combined may be explained by the small numbers at some sites, different field assistants asking questions, and differences in medical services. These results may only apply in other African countries with similar ecological, cultural, and sociodemographic characteristics. Medical and psychosocial detrimental consequences were similar in those with ACE with and without CSE but more common than in the general population. The mortality was low because this represents a cohort who have survived CSE, which occurred during childhood and was associated with febrile illnesses. 7 Case fatality was lower than in CSE in Richmond, Virginia, where there may have been more accurate documentation and longer follow-up period than in our study. This African cohort included CSE survivors during the survey. 4 Case fatality was increased in those with cognitive impairment, suggesting that underlying comorbidities, including psychiatric problems, 27 may increase the risk of death in people with CSE in ACE. Adherence to AEDs appeared to increase the risk of some social consequences of CSE, supporting the hypotheses that these occur in severe epilepsy. Those using AEDs and with severe epilepsy may be stigmatized, affecting marriage prospects.
The strength of this work is that we used community data from multiple sites, and there is detailed description of risk factors and consequences of CSE. The study reported estimates for CSE in ACE only, and yet non-CSE may account for over 50% of the cases of status epilepticus. 28 Recall bias and case fatality prior to the survey could have caused underestimation of prevalence. This is a cross-sectional study and future prospective community incidence studies of CSE and its consequences are required to provide reliable estimates.
CSE is common in people with ACE in Africa, with estimates greater than in high-income countries, and is associated with focal clinical features suggestive of underlying preventable causes. Underlying neurologic impairments and CNS infections such as falciparum malaria are important associations with CSE in people with ACE. Some associations differ across countries, suggesting different strategies and control measures are required for specific sites. CSE in ACE is associated with detrimental effects, similar to that seen in ACE without CSE, but greater than in the general population and thus should be addressed. Future prospective studies are required to provide reliable estimates and explore feasible control and treatment measures.

AUTHOR CONTRIBUTIONS
Symon M. Kariuki: drafting/revising the manuscript, study concept or design, analysis or interpretation of data, accepts responsibility for conduct of research and final approval, statistical analysis. Angelina Kakooza-Mwesige: drafting/revising the manuscript, study concept or design, analysis or interpretation of data, accepts responsibility for conduct of research and final approval, acquisition of data, study supervision. Ryan G. Wagner: drafting/revising the manuscript, accepts responsibility for conduct of research and final approval, acquisition of data, study supervision. Eddie Chengo: drafting/revising the manuscript, study concept or design, accepts responsibility for conduct of research and final approval, acquisition of data, study supervision. Steven White: drafting/ revising the manuscript, analysis or interpretation of data, accepts responsibility for conduct of research and final approval. Gathoni Kamuyu: analysis or interpretation of data, accepts responsibility for conduct of research and final approval, acquisition of data, study supervision. Anthony Ngugi: drafting/revising the manuscript, study concept or design, accepts responsibility for conduct of research and final approval. Ley Sander: drafting/revising the manuscript, study concept or design, analysis or interpretation of data, accepts responsibility for conduct of research and final approval. Brian Neville: drafting/revising the manuscript, study concept or design, accepts responsibility for conduct of research and final approval, study supervision. Charles Newton: drafting/revising the manuscript, study concept or design, analysis or interpretation of data, accepts responsibility for conduct of research and final approval, acquisition of data, statistical analysis, study supervision, obtaining funding.