Clinical characteristics of children with epilepsy managed at a tertiary hospital in Africa: a retrospective study

Background: Most children with epilepsy reside in resource-limited regions such as sub-Saharan Africa, where the majority of studies have been conducted in rural areas with limited investigations. Medical records from children with epilepsy seen at an urban hospital in Kenya were examined to provide a comprehensive description of epilepsy in children from this hospital. Methods: A retrospective observational study was conducted which involved reviewing medical records of 426 epilepsy patients (260 males and 166 females) aged 0 - 18 years, seen in Nairobi, Kenya between February 2011 and December 2014. Results: The most frequent age at presentation; documented in 29% was in infancy. Generalized seizures due to structural brain abnormalities were the most common form of epilepsy (28%). Lennox-Gastaut Syndrome was the most common electroclinical syndrome (7%). Focal seizures and focal seizures with loss of awareness were identiﬁed in 12% of the population. There were no cases of childhood absence epilepsy in this group. Brain atrophy was the most common MRI ﬁnding, occurring in a ﬁfth of the population (20%), while cystic encephalomalacia occurred in 13%. Half (50%) of all EEG recordings performed for this cohort were abnormal. Generalized seizures due to structural brain abnormalities and Lennox-Gastaut Syndrome (LGS) were signiﬁcant predictors of a treatment history of three or more AEDs. At the conclusion of the review period, 16% of the patients had not visited the clinic for more than 12 months and were considered to be lost to follow-up. Conclusion: The highest frequency of epilepsy cases was documented in children less than one year of age. Generalized seizures due to structural abnormalities and Lennox-Gastaut syndrome were the most common seizure type and syndrome. Improvement of public awareness of different types of seizures in children may increase identiﬁcation of children with childhood absence epilepsy.

Commission on Classification and Terminology [9], Schef-23 fer et al. proposed an organization of epilepsy diagno-24 sis based on electroclinical syndrome and aetiology [10]. 25 An electroclinical syndrome is a presentation of epilepsy 26 that demonstrates specific outcomes, clinical and electroen-27 cephalogram(EEG) characteristics [9] [10]. In resource-28 limited settings, various challenges present limitations for 29 delineating epilepsy according to electroclinical syndrome 30 and aetiology. Recent population and hospital-based stud- 31 ies conducted in Africa providing descriptions of epilepsy 32 in children based on these electro-clinical syndromes or 33 syndrome-associated outcomes are sparse to non-existent 34 [6] [10][11] [12]. This paper describes the static and dy- 35 namic demographic characteristics of the patient population 36 and seizure onset; electroclinical syndromes and seizure 37 types; structural and electrophysiological brain abnormal-  The patient medical records examined consisted of phys-63 ical files and as well as information stored on the hospital's 64 electronic databases. Physical files of patients seen at the 65 hospital during this period indicating a history of seizures 66 or diagnosis of epilepsy were obtained from the medical 67 records department using ICD-10 (International Classifica-68 tion of Diseases, Tenth Revision) codes utilized at the time of 69 archiving. From these records, all patients diagnosed with 70 epilepsy, according to the ILAE 2014 definition of epilepsy, 71 were selected for inclusion in the study [13]. According to 72 the ILAE, epilepsy is defined as two or more unprovoked 73 seizures occurring at least 24 hours apart, or one unpro-74 voked seizure with a probability of recurrence determined 75 to be greater than 60% [13]. Based on this definition, the 76 following patients were excluded from the study: patients 77 presenting with typical simple febrile seizures only; patients 78 with a single seizure episode without an epilepsy syndrome 79 diagnosis or identifiable structural brain abnormality; those 80 with a single provoked seizure in the context of fever or 81 hypocalcemia, for example; as well as those with episodes of 82 syncope only; or non-epileptic, psychogenic seizures solely.

83
Evaluation for co-morbidities such as attention deficity 84 hyperactivity disorder(ADHD)and autism spectrum disor-85 der(ASD) were made by the paediatric neurologist who 86 assumed care for the children. Screening for these co-87 morbidities was done using National Institute for Children's 88 Health Quality (NICHQ), Vanderbilt Assessment Scales for 89 ADHD and the Diagnostic and Statistical Manual of Men-90 tal Disorder, Fifth Edition (DSM V) for ASD. Where other 91 behavioural concerns were identified these children were 92 evaluated by a child psychiatrist. Results of these evalua-93 tions were included in patient records. During regular clinic 94 visits data on patient developmental progress was also in-95 cluded in the medical records. 96 Data on patient sex; age at seizure onset; electroclinical 97 diagnosis; structural brain abnormalities; number and type 98 of ASMs in treatment history; consistency of clinic follow-99 up and co-morbidities were extracted as previously docu-100 mented in the patient's clinical notes. Further data was ob-101 tained from clinical notes and technical reports describing 102 imaging and electrophysiological findings. The research as-103 sistant who collected and analysed the data was trained and 104 supervised by the paediatric neurologist who assessed and 105 managed the patients.

106
Electroclinical syndrome diagnoses were determined ac-107 cording to guidelines outlined in the Proposal for Revised 108 Classification of Epilepsies and Epileptic Syndromes, by the 109 Commission on Classification and Terminology of the ILAE 110 [9] [10]. Epilepsy that could not be classified as an electro-111 clinical syndrome was described according to seizure type 112 (i.e. generalized or focal) and cause (e.g. structural aetiol-113 ogy) [9].    Table 2.

171
Structural and electrophysiological brain abnormalities 172 A total of 8.5% (N = 36) of patients had no investiga-173 tion results available on file. The majority of the patients, 174 86.4% (N = 368) had a sleep EEG reported, while approx-175 imately half of the study population, 48.8% (N = 208) had 176 a brain MRI scan with contrast. Reported findings from 177 the brain MRI scans are summarized in Table 3. Ten chil-178 dren (4.8%) who had both brain MRI and head CT reports 179 had their findings included in the brain MRI group. Four 180 among these (0.9%) had both normal head CTs and brain 181 MRIs while three children (1.4%) had delayed myelination 182 on brain MRI. Two children (0.9%) had white matter hy-183 perintensities and one child (0.4%) had polymicrogyria on 184 brain MRI.   (Table 4). Among those on antiepileptic med-207 ication, 42.7% (N = 176) of patients been treated with 208 one ASM , 24.2% (N = 100) with two ASMs, 17.4% (N 209 = 72) with three ASMs and 9.2% (N = 38) with four or 210 more ASMs serially. Table 4 summarizes types of ASMs that 211 were utilized in this population in order of frequency. As il-212 lustrated above individual patients may have utilized more 213 than one medication. Valproic acid was the most commonly 214 used ASM with 69.4% usage amongst the study participants, 215 followed by Carbamazepine and Clonazepam with 24.2% 216 and 20.8% usage respectively. In contrast, Pregabalin and 217 Gabapentin were the least used (each having been utilized 218 by 0.3% of study participants).       Inability to determine the cause of epilepsy is largely due to 291 financial and diagnostic limitations at the point of care [18]. 292 Further, genetic and metabolic testing is not available locally 293 and performance of these tests by overseas healthcare part-294 ners is associated with significant financial costs and time 295 delays as has been our experience. This is particularly note-296 worthy when considering that Kenya's per capita GDP in 297 2016 was USD 1,410 [43], meaning that patients lacking 298 medical insurance cover would struggle to access requisite 299 tests such as the early infantile epileptic encephalopathy se-300 quencing panel that costs USD 2000 on average.

301
Since generalized epilepsy of unknown cause was a fre-302 quent observation amongst patients in this cohort, as in 303 other studies, and in light of potential benefits of deter-304 mining the cause of epilepsy where possible, interventions 305 to improve the quality of and access to local diagnostic 306 facilities such as neuro-imaging, electroencephalography, 307 metabolic and genetic testing would be beneficial to the 308 management of epilepsy patients in sub-Saharan Africa.

309
Positive family history of febrile seizures in children who 310 presented with atypical febrile seizures allowing a diagnosis 311 of Genetic Epilepsy with febrile seizures plus was rarely de-312 termined in this cohort. This could be due to low rates of di-313 agnosis of atypical febrile seizures in this setting or cultural 314 reasons where families rarely discuss seizures in children 315 openly due to associated stigma. Cultural reasons may also 316 contribute to the predilection for more significantly male 317 children presenting for care at this centre.

318
In this cohort we identified no children with childhood 319 absence epilepsy(CAE) which being a common form of 320 epilepsy, was an unexpected finding [40] [44]. This is most 321 likely due to under-recognition and lack of referral of such 322 cases to the neurology teams. It is possible CAE is managed 323 by paediatricians and being generally responsive to widely 324 available first line ASMs, referral in such cases would not be 325 required.

326
Critically, patients who are lost to follow-up are likely 327 to remain untreated. Long-standing, untreated epilepsy 328 has detrimental and enduring personal and social con-329 sequences, including impaired intellectual performance 330 [31][32] [33][34] [35]. For these reasons, there is an ur-331 gent need to determine the reasons for defaulting on clinic 332 follow-up [37]. Findings from these studies which may be 333 applicable to other similar hospital settings would enable 334 this institution to address these specific risk factors and po-335 tentially reduce the number of children with epilepsy who 336 remain untreated.    The study identified that males comprised the majority of 354 paediatric epilepsy patients seen at this paediatric neurol-355 ogy service while LGS was the most common electroclini-356 cal syndrome. Generalized seizures associated with struc-357 tural brain abnormalities were the most frequent form of 358 epilepsy overall. Half of all EEGs performed displayed ab-359 normalities including generalized spike wave discharges, fo-360 cal spike wave discharges and slow background activity in 361 decreasing occurrence respectively. Majority of patients had 362 been treated with one ASM only, with valproic acid being 363 most common utilized ASM. Psychiatric co-morbidities were 364 more commonly identified in children with focal seizures 365 while and those with Lennox Gastaut syndrome were more 366 likely to present with developmental delay. The findings 367 from this study hold significant relevance for the improved 368 diagnosis and management of epilepsy among paediatric 369 patients in urban SSA, including need for: greater preven-370 tive and diagnostic care predominantly targeting children 371 in the 1 to 10 year age group; increased efforts to address 372 patient attrition rates; and improvement of practitioner and 373 public awareness of common forms of epilepsy to facilitate 374 early detection and appropriate referral and management. 375 This paper's contribution to existing research concerns its 376 potential to address a dearth of recent studies conducted 377 in sub-Saharan Africa describing electro-clinical syndromes 378 or syndrome-associated outcomes among children in the re-379 gion. Prospective studies may seek to explore precise rela-380 tionships between specific syndromes and variable such as 381 psychiatric and neuro-developmental co-morbidities and re-382 sponses to anti-epileptic drugs.