Prevalence and incidence of epilepsy
`A systematic review and meta-analysis of international studies
`
`Kirsten M. Fiest, PhD*
`Khara M. Sauro, PhD*
`Samuel Wiebe, MD, MSc
`Scott B. Patten, MD,
`PhD
`Churl-Su Kwon, MD
`Jonathan Dykeman, MD
`Tamara Pringsheim, MD,
`MSc
`Diane L. Lorenzetti, PhD
`Nathalie Jetté, MD, MSc
`
`Correspondence to
`Dr. Jetté:
`Nathalie.jette@ahs.ca
`
`Supplemental data
`at Neurology.org
`
`ABSTRACT
`Objective: To review population-based studies of the prevalence and incidence of epilepsy world-
`wide and use meta-analytic techniques to explore factors that may explain heterogeneity
`between estimates.
`Methods: The Preferred Reporting Items for Systematic Reviews and Meta-Analyses standards
`were followed. We searched MEDLINE and EMBASE for articles published on the prevalence
`or incidence of epilepsy since 1985. Abstract, full-text review, and data abstraction were con-
`ducted in duplicate. Meta-analyses and meta-regressions were used to explore the association
`between prevalence or incidence, age group, sex, country level income, and study quality.
`Results: A total of 222 studies were included (197 on prevalence, 48 on incidence). The point
`prevalence of active epilepsy was 6.38 per 1,000 persons (95% confidence interval [95% CI]
`5.57–7.30), while the lifetime prevalence was 7.60 per 1,000 persons (95% CI 6.17–9.38).
`The annual cumulative incidence of epilepsy was 67.77 per 100,000 persons (95% CI 56.69–
`81.03) while the incidence rate was 61.44 per 100,000 person-years (95% CI 50.75–74.38).
`The prevalence of epilepsy did not differ by age group, sex, or study quality. The active annual
`period prevalence, lifetime prevalence, and incidence rate of epilepsy were higher in low to middle
`income countries. Epilepsies of unknown etiology and those with generalized seizures had the
`highest prevalence.
`Conclusions: This study provides a comprehensive synthesis of the prevalence and incidence of
`epilepsy from published international studies and offers insight into factors that contribute to het-
`erogeneity between estimates. Significant gaps (e.g., lack of incidence studies, stratification by
`age groups) were identified. Standardized reporting of future epidemiologic studies of epilepsy
`is needed. Neurology® 2017;88:296–303
`
`GLOSSARY
`CI 5 confidence interval.
`
`Epilepsy is a serious neurologic condition associated with stigma,1 psychiatric comorbidity,2 and
`high economic costs.3 The WHO’s 2010 Global Burden of Disease study ranks epilepsy as the
`second most burdensome neurologic disorder worldwide in terms of disability-adjusted life years.4
`Studies investigating the prevalence and incidence of epilepsy are increasingly common, par-
`ticularly in low- and middle-income countries. Estimates of the prevalence and incidence of epi-
`lepsy worldwide vary considerably, likely reflecting differences in measurement and reporting,
`along with clinical characteristics such as etiology and seizure type. Previous systematic reviews
`of the prevalence of epilepsy focused on specific regions (China,5 Europe,6 Latin America,7 and
`Arab countries8) and prior reviews on the incidence of epilepsy did not use meta-analyses to
`explore associated factors.9,10 Few of these studies explored potential sources of heterogeneity
`between estimates or they examined both prevalence and incidence globally.
`
`*These authors contributed equally to this work.
`From the Department of Community Health Sciences, O’Brien Institute for Public Health (K.M.F., K.M.S., S.W., S.B.P., T.P., D.L.L., N.J.), and
`the Department of Clinical Neurosciences & Hotchkiss Brain Institute (K.M.S., S.W., J.D., T.P., N.J.) and Department of Psychiatry, Mathison
`Centre for Mental Health Research & Education (S.B.P.), Hotchkiss Brain Institute (K.M.S, S.W., S.B.P., T.P., N.J.), Department of Critical Care
`Medicine (K.M.F.), Cumming School of Medicine, University of Calgary, Canada; Department of Neurosurgery (C.-S.K.), King’s College
`Hospital, London, UK; and Institute of Health Economics (D.L.L.), Edmonton, Canada.
`Go to Neurology.org for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.
`
`296
`
`© 2016 American Academy of Neurology
`ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
`
`Neurelis - EX. 2005
`Aquestive Therapeutics, Inv. v. Neurelis, Inc. - IPR2019-00450
`
`

`

`Our aim was to estimate the prevalence and
`incidence of epilepsy from international stud-
`ies, and to quantify the burden of epilepsy
`using meta-analytic
`techniques. We
`also
`explore the sources of heterogeneity between
`estimates, assessing factors such as age, sex,
`country income level, epilepsy syndrome, sei-
`zure type, epilepsy etiology, and study quality.
`
`METHODS Search strategy. The systematic review was con-
`ducted according to a predetermined protocol and adhered to the
`Preferred Reporting Items for Systematic review and Meta-Analysis.11
`The search strategy (appendix e-1 at Neurology.org) was
`developed by experts in epilepsy and epidemiology and an
`academic research librarian (D.L.L.). Search terms included
`prevalence, incidence, and epidemiology in conjunction with
`epilepsy, seizure, and convulsion. The search was conducted
`from 1985 to October 22, 2013, in MEDLINE and EMBASE.
`Articles in English or French were included. The reference lists of
`included articles were also hand searched. References were
`managed using EndNote X5.12
`
`Study selection. Abstracts and titles of all references were
`screened in duplicate by 2 independent reviewers to identify orig-
`inal, population-based studies on the prevalence or incidence of
`epilepsy. Two independent
`reviewers
`screened the full-text
`articles of abstracts identified in the first stage of review.
`Articles were included if they met the following criteria: (1)
`original research,
`(2) population-based (selecting the entire
`population or using probability-based sampling methods), (3)
`reported a prevalence or incidence of epilepsy (or raw numbers
`that allowed the calculation of an estimate). Disagreements
`pertaining to the inclusion of articles were resolved by
`consensus or involvement of a third author as necessary.
`
`Data extraction and study quality. Data abstraction was com-
`pleted in duplicate by 2 independent reviewers using a standardized
`data collection form. When multiple articles reporting data from
`the same study population were identified, the most comprehensive
`data were used. When studies reported on different data collection
`years or subgroups (sex, age), all nonoverlapping data were
`included. Age, sex, study location, sources of ascertainment, and
`definitions/diagnostic criteria for epilepsy were extracted. Epilepsy
`prevalence or incidence estimates, raw numbers, and confidence in-
`tervals (CIs) (when provided) were recorded along with any strati-
`fied results by age, sex, or year of data collection.
`The quality of included studies was evaluated using standard
`assessment tools13,14 (appendix e-2), and included sample repre-
`sentativeness, condition assessment, and statistical methods. Each
`study was given a quality score of 0 to 8 based on fulfillment of
`the quality criteria.
`
`Data synthesis and analysis. Prevalence estimates were divided
`into 2 groups: point prevalence and annual period prevalence. Point
`prevalence is the number of existing cases of epilepsy in a population,
`over the total population at one specific point in time (e.g., on June
`30, 2013). Period prevalence includes both existing and new cases
`of epilepsy in a population over the total population over a defined
`period of time (e.g., between January 1 and December 31, 2013).
`Estimates of prevalence were additionally categorized into 2
`mutually exclusive groups based on the definitions provided within
`individual articles: active and lifetime. Lifetime prevalence was con-
`ceptualized differently than active epilepsy prevalence, as it can be
`considered a type of period prevalence, conditional on survival,
`
`where the period is the time between birth and assessment. We re-
`ported on the point prevalence of active epilepsy, the annual period
`prevalence of active epilepsy, and the lifetime prevalence of epilepsy.
`These different categories take into account (1) the clinical differen-
`ces of those with active vs inactive epilepsy and (2) how the time
`period of assessment may influence reported estimates.
`Incidence estimates were stratified into cumulative incidence
`and incidence rate. Cumulative incidence is the number of new
`cases of epilepsy over the total number of people in the popula-
`tion at risk for developing epilepsy during a specified period of
`time (e.g., 64.90 persons with epilepsy per 100,000 persons dur-
`ing 1 year). We used the following formula to calculate annual
`cumulative incidence based on estimates provided in the articles:
`cumulative incidencen years 5 1 2 exp (2 annual rate 3 n). The
`incidence rate of epilepsy is the number of new cases of epilepsy
`over the total amount of person-time at risk for developing epi-
`lepsy during a specified period of time (e.g., 68.40 persons with
`epilepsy per 100,000 person-years). In both cases the numerator
`is the number of new cases, while the denominator will differ as
`the incidence rate also incorporates time as a unit.
`Seizure type and epilepsy etiology were categorized according
`to the most recent International League Against Epilepsy classifi-
`cation15; these subgroup analyses were only available for estimates
`of active point prevalence of epilepsy. Country income level was
`dichotomized into low–middle and high based on the World
`Bank’s classification.16
`Age was stratified into 2 broad groups for age-specific analy-
`sis: (1) those younger than 18 years and those 18 years and older
`and (2) in 10-year groups: 0–9, 10–19, 20–29, 30–39, 40–49,
`50–59, and 60 and above, if available, and using only studies that
`reported on all 7 categories.
`Studies were included in the meta-analysis if they reported the
`number of cases and sample denominator, the estimate with 95%
`CI, or the information with which to calculate the estimated prev-
`alence or incidence. Age, sex, country income level, seizure type,
`and epilepsy etiology were examined categorically. Only estimates
`that included persons of all ages were included in the pooled anal-
`yses (except for age-specific analyses). The association of age, sex,
`country income level, and study quality with prevalence and inci-
`dence estimates was assessed using meta-regression if there were 2
`studies or more per grouping. Boxplots assessed the presence of
`outliers (defined as an estimate more than 1.5 times the interquar-
`tile range beyond the first [p25] or third [p75] quartiles).17 We
`reported overall median prevalence or incidence and accompany-
`ing first and third quartiles (p25–p75).
`The I2 was used to quantify the magnitude of between-study
`heterogeneity and the Cochrane Q statistic was calculated to
`determine significance. A priori, we decided to report the pooled,
`weighted estimate generated by random effects models, because
`we hypothesized a high degree of between-study heterogeneity.
`Publication bias was investigated visually using funnel plots and
`statistically using Begg18 and Egger19 tests.
`R version 2.14 was used for all meta-analyses and meta-
`regressions.20 The meta package was used to generate the
`forest plots, pooled estimates, and to assess for publication
`bias.21 Meta-regression using restricted maximum likelihood
`estimation was conducted using the metafor package.22 Box-
`plots were generated using STATA v12.1.23 A p value ,0.05
`was deemed statistically significant.
`
`RESULTS Identification and description of studies.
`The search strategy yielded a total of 16,479 abstracts:
`8,798 from MEDLINE and 7,681 from EMBASE
`(figure 1). We screened 13,305 unique abstracts
`
`Neurology 88 January 17, 2017
`297
`ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
`
`Neurelis - EX. 2005
`Aquestive Therapeutics, Inv. v. Neurelis, Inc. - IPR2019-00450
`
`

`

`1
`
`Figure 1
`
`Study flow diagram
`
`Medline
`(N=8,798)
`
`EMBASE
`(N=7,681)
`
`Total abstracts screened
`(with duplicates removed)
`(n=13,305)
`
`Full-text review
`(n=470)
`
`V
`Included
`(n=222)
`
`V
`Incidence
`(n=48*)
`
`V
`Prevalence
`(n=197*)
`
`1
`
`Not selected for
`full-text review
`(n=12,835)
`
`Excluded (n=248):
`• Not original (61)
`• Pre-1985 data (41)
`• No estimate provided (36)
`• Foreign language (33)
`• Not population-based (29)
`• Status epilepticus only (13)
`• Neonatal seizures only (13)
`• Abstract only (7)
`• Febrile seizures only (5)
`• Nonepileptic seizures only (1)
`
`*Incidence and prevalence studies equal greater than 222 because 24 articles reported both incidence and prevalence.
`
`and 470 articles met the criteria for full-text review, of
`which 248 were excluded. Hand searching did not
`contribute additional articles. A total of 197 articles
`reported on prevalence of epilepsy, 48 on incidence,
`and 24 on both (tables e-1 and e-2, and appendix e-3
`for reference list).
`
`Prevalence of active epilepsy. Seventy-three studies re-
`ported on the point prevalence of active epilepsy
`and of those, 67 estimates (63 unique studies) were
`eligible for
`inclusion in the meta-analysis. The
`pooled point prevalence of active epilepsy was 6.38
`per 1,000 persons (95% CI 5.57–7.30) (table 1,
`figure e-1). Heterogeneity existed between estimates
`(I2 5 99.6%, Q p value ,0.0001). The median point
`prevalence of active epilepsy was 5.40 per 1,000
`persons (p25–p75, 3.90–9.99). Four outliers were
`identified: 104.97 per 1,000 persons
`(95% CI
`68.60–160.63) from Cameroon,24 57.23 per 1,000
`persons (95% CI 36.98–88.56)
`from Panama,25
`29.46 per 1,000 persons (95% CI 21.16–41.03)
`from Ethiopia,26 and 22.62 per 1,000 persons (95%
`CI 9.51–53.82) from Ecuador.27
`Twelve studies reported on the annual period
`prevalence of active epilepsy, and 11 were eligible
`for inclusion in the meta-analysis. The pooled annual
`period prevalence of active epilepsy was 2.83 per
`
`1,000 persons (95% CI 1.53–5.26) (table 1, figure 2).
`Heterogeneity existed between estimates (I2 5 100%,
`Q p value ,0.0001). The median annual period prev-
`alence of active epilepsy was 3.91 per 1,000 persons
`(p25–p75, 1.14–5.15). One outlier from Tanzania
`(13.56 per 1,000 persons [95% CI 10.68–17.21])
`was identified.28
`
`Lifetime prevalence of epilepsy. Sixty-seven studies re-
`ported on the lifetime prevalence of epilepsy, and
`56 were eligible for inclusion in the meta-analysis.
`The pooled lifetime prevalence of epilepsy was 7.60
`per 1,000 persons (95% CI 6.17–9.38) (table 1,
`figure e-2). Heterogeneity existed between estimates
`(I2 5 99.7%, Q p value ,0.0001). The median
`lifetime prevalence of epilepsy was 7.06 per 1,000
`persons (p25–p75, 4.74–11.23). Three studies were
`identified as outliers: from Panama (75.30 per 1,000
`[95% CI 51.65–109.78]),25 Cameroon
`persons
`(49.00 per 1,000 persons
`[95% CI 40.19–
`59.74]),29 and Honduras (23.33 per 1,000 persons
`[95% CI 19.93–27.31]).30
`
`Cumulative incidence of epilepsy. Thirty-one studies re-
`ported on the cumulative incidence of epilepsy, 14 of
`which were included in the meta-analysis. The pooled
`annual cumulative incidence of epilepsy was 67.77
`
`298
`
`Neurology 88 January 17, 2017
`ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
`
`Neurelis - EX. 2005
`Aquestive Therapeutics, Inv. v. Neurelis, Inc. - IPR2019-00450
`
`

`

`1
`
`Table 1
`
`Pooled estimates for each type of prevalence (per 1,000)
`
`Corresponding
`figure
`
`Subgroup
`
`No. included
`estimates
`
`Estimate per
`1,000 (95% CI)
`
`1
`
`Active point prevalence
`
`Overall
`
`By sex
`
`By age
`
`By country income
`
`By seizure type
`
`e-1
`
`NA
`
`e-3
`
`e-4
`
`NA
`
`Male
`
`Female
`
`0–9
`
`10–19
`
`20–29
`
`30–59
`
`601
`
`Low and middle
`
`High
`
`Active generalized
`
`Active focal seizures
`
`Active unknown seizures
`
`By epilepsy etiology
`
`NA
`
`Presumed genetic
`
`Structural/metabolic
`
`67
`
`27
`
`28
`
`12
`
`12
`
`12
`
`12
`
`12
`
`50
`
`13
`
`10
`
`10
`
`7
`
`5
`
`5
`
`6.38 (5.57–7.30)
`
`7.31 (6.06–8.81)
`
`6.85 (5.55–8.47)
`
`5.19 (3.54–7.62)
`
`8.86 (6.58–11.92)
`
`9.14 (7.17–11.64)
`
`7.94 (6.20–10.15)
`
`7.17 (4.67–11.01)
`
`6.68 (5.45–8.18)
`
`5.49 (4.16–7.26)
`
`4.33 (2.55–8.32)
`
`2.99 (1.39–6.42)
`
`0.81 (0.28–2.32)
`
`1.70 (0.75–3.90)
`
`2.70 (1.12–3.81)
`
`3.15 (2.57–3.87)
`
`Active period prevalence
`
`Overall
`
`By sex
`
`By age
`
`By country income
`
`Lifetime prevalence
`
`Overall
`
`By sex
`
`By age
`
`By country income
`
`2
`
`NA
`
`NA
`
`e-5
`
`e-2
`
`NA
`
`NA
`
`e-6
`
`Unknown origin
`
`Male
`
`Female
`
`#18
`
`191
`
`Low and middle
`
`High
`
`Male
`
`Female
`
`#18
`
`191
`
`Low and middle
`
`High
`
`3
`
`11
`
`6
`
`6
`
`22
`
`22
`
`3
`
`8
`
`56
`
`20
`
`18
`
`30
`
`24
`
`41
`
`15
`
`2.83 (1.53–5.26)
`
`3.47 (1.22–9.84)
`
`2.92 (1.36–6.26)
`
`4.80 (4.17–5.52)
`
`5.43 (3.93–7.50)
`
`6.79 (2.77–16.65)
`
`2.06 (1.00–4.25)
`
`7.6 (6.17–9.38)
`
`6.99 (5.3–9.20)
`
`7.62 (5.52–10.50)
`
`7.24 (5.74–9.14)
`
`8.59 (5.92–12.46)
`
`8.75 (7.23–10.59)
`
`5.18 (3.75–7.15)
`
`Abbreviations: CI 5 confidence interval; NA 5 not available.
`
`(95% CI 56.69–81.03).
`per 100,000 persons
`(I2 5
`Heterogeneity existed between estimates
`95.6%, Q p value ,0.0001) (table 2, figure 3A).
`The median cumulative incidence of epilepsy
`was 65.61 per 100,000 persons
`(p25–p75,
`48.00–81.00), and there was one outlier from
`Andean Ecuador (189.96 per 100,000 persons
`[95% CI 160.70–224.55]).31
`
`Incidence rate of epilepsy. Nineteen studies reported on
`the incidence rate of epilepsy, 13 of which were included
`in the meta-analysis. The pooled incidence rate of
`
`epilepsy was 61.44 per 100,000 person-years (95% CI
`50.75–74.38). Heterogeneity existed between estimates
`(I2 5 98.6%, Q p value ,0.0001) (table 2, figure 3B).
`The median incidence rate of epilepsy was 56.79
`per 100,000 person-years (p25–p75, 46.00–76.89),
`and there were 3 outliers (in the Assiut Governorate
`in Egypt,32 Chile,33 and West Uganda34).
`
`Sources of heterogeneity. Estimates of the prevalence
`and incidence of epilepsy by sex, age, country income
`level, and prevalence by seizure type and epilepsy eti-
`ology are presented in appendix e-4 and figures e-3
`
`Neurology 88 January 17, 2017
`299
`ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
`
`Neurelis - EX. 2005
`Aquestive Therapeutics, Inv. v. Neurelis, Inc. - IPR2019-00450
`
`

`

`Figure 2
`
`Active period prevalence of epilepsy
`
`Study* Country
`
`Period
`prevalence
`
`95% CI
`
`Tanzania
`90
`Singapore
`204
`India
`161
`171 United Kingdom
`192
`Turkey
`230
`Greece
`252 United Kingdom
`79
`Denmark
`127
`Croatia
`167 United States
`195 United States
`
`13.56
`0.75
`3.91
`6.80
`5.90
`2.26
`5.15
`1.14
`1.09
`0.94
`4.61
`
`(10.68; 17.21)
`(0.73; 0.77)
`(3.46; 4.42)
`(6.56; 7.05)
`(3.67; 9.48)
`(2.01; 2.55)
`(5.05; 5.25)
`(1.11; 1.17)
`(0.91; 1.30)
`(0.89; 1.00)
`(4.34; 4.90)
`
`(1.53; 5.26)
`2.83
`Pooled totals
`Heterogeneity: 12=100%, 0=24541.4, df=10, p<0.0001
`
`. 11810.-
`
`*Study numbers correspond to references in appendix e-3. CI 5 confidence interval.
`
`5
`20
`15
`10
`0
`Active period prevalence of epilepsy
`per 1,000 persons
`
`(pooled point prevalence of epilepsy in 10-year age
`groups), e-4 (active point prevalence of epilepsy by
`country income level), e-5 (active period prevalence
`of epilepsy by country income level), e-6 (lifetime
`prevalence of epilepsy by country income level), e-7
`(cumulative
`incidence of
`epilepsy by
`country
`income level), and e-8 (incidence rate of epilepsy by
`country income level).
`
`Table 2
`
`Pooled estimates for each type of incidence of epilepsy
`
`Corresponding
`figure
`
`Subgroup
`
`No. included
`estimates
`
`Estimate per
`100,000 (95% CI)
`
`1
`
`Cumulative
`incidence
`
`Overall
`
`By sex
`
`By age
`
`By country
`income
`
`Incidence rate
`
`Overall
`
`By sex
`
`By age
`
`By country
`income
`
`3A
`
`NA
`
`NA
`
`e-7
`
`3B
`
`NA
`
`NA
`
`e-8
`
`Male
`
`Female
`
`#18
`
`191
`
`14
`
`10
`
`8
`
`5
`
`3
`
`67.77 (56.69–81.03)
`
`58.13 (43.94–81.55)
`
`55.78 (41.09–75.72)
`
`85.29 (59.54–122.19)
`
`64.81 (13.90–302.24)
`
`Low and middle 9
`
`65.19 (41.65–102.02)
`
`High
`
`Male
`
`Female
`
`#18
`
`191
`
`5
`
`13
`
`8
`
`8
`
`1
`
`2
`
`70.24 (57.51–85.78)
`
`Estimate per 100,000
`person years (95% CI)
`
`61.44 (50.75–74.38)
`
`63.97 (47.96–85.32)
`
`57.43 (41.60–79.29)
`
`46.90 (42.29–52.01)
`
`34.63 (28.38–42.25)
`
`Low and middle 4
`
`138.99 (69.45–278.16)
`
`High
`
`9
`
`48.86 (39.05–61.13)
`
`Abbreviations: CI 5 confidence interval; NA 5 not available.
`
`Publication bias. There was no evidence of publication
`bias for any of the estimates of prevalence and inci-
`dence using visual inspection of funnel plots or Begg
`or Egger test (all p . 0.05).
`
`Study quality. The median study quality score was
`6/8 (range 2–8) for the prevalence of epilepsy and
`7/8 for studies of incidence (range 4–8) (tables e-3
`and e-4). Meta-regression found no effect of study
`quality on the estimates of epilepsy prevalence or
`incidence, all p . 0.05.
`
`DISCUSSION This
`systematic review and meta-
`analysis of international studies on the prevalence
`and incidence of epilepsy used subgroups analyses
`to examine the relationship among socioeconomic,
`demographic, and clinical factors that may influence
`the prevalence and incidence of epilepsy.
`Age has commonly been associated with the prev-
`alence and incidence of epilepsy. Congruent with pre-
`the
`vious descriptive reports,35 we found that
`incidence of epilepsy was generally higher in the
`youngest and oldest age groups; however, there were
`insufficient studies to perform a meta-analysis. The
`trend in the point prevalence of active epilepsy by 10-
`year age groups is consistent with previous reports.35
`Prevalence is expectedly lowest early in life, increasing
`to its highest level during adolescence and early adult-
`hood, decreases after age 30, and remains fairly con-
`stant for the remainder of life. The number of studies
`included in this pooled analysis was limited by the
`reporting of common age groupings in individual
`studies; only 12 of 63 eligible studies used common
`age groups, and analysis by 10-year age groups was
`not possible for estimates of incidence due to the
`small number of studies. The prevalence of epilepsy
`was slightly higher in studies of persons over the age
`of 18 compared to those under 18, while the reverse
`was true for the incidence of epilepsy. This finding is
`consistent with previous studies of the epidemiology
`of epilepsy in Europe.6 Elevated mortality could pre-
`vent the lifetime prevalence of epilepsy from increas-
`ing significantly with age (particularly in older age
`groups) when the incidence of epilepsy is not zero,
`as is the case here.
`Sex, while not commonly thought to affect the
`occurrence of epilepsy, may contribute to differences
`in epilepsy incidence.35 The incidence of epilepsy
`tended to be higher in males than females. Some
`suggest that females may be more likely to conceal
`their epilepsy diagnosis if they live in a country where
`they would be considered unmarriageable or socially
`marginalized.35,36
`There was no difference between high and low–
`middle income countries for the point prevalence of
`active epilepsy or cumulative incidence. However, the
`active annual period prevalence was significantly
`
`300
`
`Neurology 88 January 17, 2017
`ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
`
`Neurelis - EX. 2005
`Aquestive Therapeutics, Inv. v. Neurelis, Inc. - IPR2019-00450
`
`

`

`1
`
`Figure 3
`
`Incidence of epilepsy
`
`A
`Study*
`
`Country
`
`Cumulative
`incidence
`
`95% CI
`
`(61.39; 87.53)
`73.30
`Tanzania
`210
`(32.50; 148.20)
`69.40
`Benin
`122
`(64.98; 100.97)
`81.00
`Tanzania
`256
`(34.87; 69.71)
`49.30
`India
`161
`(20.94; 39.54)
`28.77
`India
`213
`(29.40; 78.37)
`48.00
`97
`Egypt
`(29.59; 62.91)
`43.14
`96
`Egypt
`(45.32; 75.26)
`58.40
`230
`Greece
`(34.40; 62.98)
`46.54
`189
`Iceland
`(59.73; 63.96)
`61.81
`225
`Finland
`(76.97; 84.74)
`252 United Kingdom 80.76
`(93.08; 137.19)
`150
`Chile
`113.00
`(41.66; 206.25)
`164
`Honduras
`92.69
`198
`189.96 (160.70; 224.55)
`Ecuador
`
`67.77
`(56.69; 81.03)
`Pooled totals
`Heterogeneity: 12=95.6%, Q=298.5, df=13, p<0.0001
`
`B
`
`Study
`
`Country
`
`Incidence
`rate
`
`95% CI
`
`I
`I
`I
`I
`I
`I
`250
`200
`150
`100
`50
`0
`Overall cumulative incidence of epilepsy
`per 100,000 persons
`
`215.00 (157.76; 293.01)
`129
`Uganda
`76.89
`(70.30; 84.09)
`Kenya
`178
`152.00
`(69.85; 330.78)
`133
`Egypt
`72.00
`(64.51; 80.36)
`Netherlands
`220
`33.90
`(31.86; 36.07)
`41
`Sweden
`158 United Kingdom 46.00
`(35.63; 59.39)
`163 United Kingdom 50.10
`(47.10; 53.30)
`79
`Denmark
`68.80
`(68.34; 69.25)
`120
`51.73
`(44.88; 59.63)
`England
`191
`Iceland
`56.79
`(52.03; 61.99)
`250
`Peru
`162.40
`(73.01; 361.25)
`63 United States
`38.60
`(33.71; 44.20)
`46 United States
`35.50
`(30.87; 40.82)
`
`(50.75; 74.38)
`61.44
`Pooled totals
`Heterogeneity: 12=98.6%, Q=848.8, df=12, p<0.0001
`
`•
`
`I
`I
`I
`I
`I
`I
`1
`100 150 200 250 300
`50
`0
`Overall incidence rate of epilespy
`per 100,000 person-years
`
`(A) Cumulative incidence of epilepsy. (B) Incidence rate of epilepsy. Study numbers correspond to references in appendix e-3.
`CI 5 confidence interval.
`
`higher in low–middle income countries. Interest-
`ingly, more low–middle income countries (14/50;
`28%) reported an active point prevalence of greater
`than 10 per 1,000 persons, compared to high-income
`countries (2/13; 15%). The incidence rate of epilepsy
`was also higher in low–middle income countries
`compared to high–middle income countries, which is
`corroborated by others.10 This trend was reversed for
`estimates of cumulative incidence. Factors such as
`premature mortality, etiology (e.g., CNS infections),
`variations in treatment, and study methodology (e.g.,
`
`case ascertainment and case definition) may differ
`between high and low–middle income countries,37
`which may partially explain our results. Regardless,
`the higher estimates in low- and middle-income
`countries are noteworthy from a public health stand-
`point as these low resource areas are also those with
`the highest treatment gap.38
`Single pooled estimates of prevalence and inci-
`dence should be interpreted with caution, given the
`amount of heterogeneity between studies. Given this
`heterogeneity, we present median estimates along
`
`Neurology 88 January 17, 2017
`301
`ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
`
`Neurelis - EX. 2005
`Aquestive Therapeutics, Inv. v. Neurelis, Inc. - IPR2019-00450
`
`

`

`with the pooled estimates. Though we explored
`a number of factors that may partially explain high
`levels of heterogeneity, including age, sex, and coun-
`try income level, we were unable to employ a single
`model to account for these factors together. A num-
`ber of studies with high prevalence and incidence
`were identified as outliers (all from Latin America,
`Africa, or the Middle East), which may further influ-
`ence heterogeneity. The authors of these studies com-
`monly hypothesize that
`the presence of CNS
`infections or antibodies (e.g., neurocysticercosis, cys-
`ticercosis antibody),25,27–30 consanguinity or family
`history of epilepsy,24–26,29 and perinatal/prenatal risk
`factors28–30 may explain the higher prevalence and
`incidence.
`The current study pooled over 124 million per-
`sons and over 655,000 persons with epilepsy, and
`used meta-regression to statistically examine the effect
`of many sources of heterogeneity. However, our study
`is not without limitations. There was heterogeneity
`between estimates of prevalence and incidence, which
`could be due to variable sampling methods, case
`ascertainment, and diagnostic methods. The quality
`of the included studies varied and some studies pro-
`vided little information on sampling and data collec-
`tion methodologies, though study quality was not
`associated with prevalence and incidence estimates.
`It was also impossible to conduct meta-analyses
`between some groups due to a smaller number of
`studies assessing those factors (e.g., under vs over
`age 65 years).
`Ideally,
`a multivariable meta-
`regression would have been employed to deal with
`the possible confounding effects of variables such as
`age and location, though this would have required
`a very large number of studies, and as such only strat-
`ified estimates are provided. Our finding that the
`annual period prevalence of epilepsy was lower than
`the point prevalence was unexpected and should be
`interpreted with caution. This finding was likely
`due to the large amount of heterogeneity (.99%
`for prevalence studies) that existed between these 2
`groups of studies.
`This systematic review and meta-analysis presents
`information on the burden of epilepsy from interna-
`tional studies. The focus on population-based studies
`allows for the results to be more applicable to primary
`care settings, where much of epilepsy care is provided.
`Few studies reported on the prevalence and incidence
`of epilepsy by seizure type or etiology, and there were
`surprisingly no studies from high-income regions
`such as Australia that were identified or met our eligi-
`bility criteria. These are considerable gaps that must
`be addressed in future work. The possible effect of
`stigma and cultural differences in epilepsy reporting
`(or in seeking medical attention) also needs to be
`explored in future studies as it may explain some of
`
`the lower estimates reported in certain regions, where
`the burden of epilepsy would be expected to be much
`higher. Methodologic factors contributing to study
`heterogeneity should also be explored, including data
`collection methods, sources of case ascertainment,
`and criteria used for assessment. Future epilepsy epi-
`demiologic studies should consider following the
`Standards of Reporting of Neurological Disorders
`checklist39 and published Standards for Epidemio-
`logic Studies and Surveillance of Epilepsy40
`to
`enhance the quality of reporting of such studies and
`decrease the heterogeneity between studies to facili-
`tate international comparisons.
`
`AUTHOR CONTRIBUTIONS
`K.M. Fiest: substantial contribution to acquisition, analysis, and interpre-
`tation of data, drafting the work, and revising it critically for intellectual
`content. K.M. Sauro: substantial contribution to acquisition, analysis,
`and interpretation of data, drafting the work, and revising it critically
`for intellectual content. S. Wiebe: substantial contribution to conception
`or design of the work, interpretation of data, revising the work critically
`for intellectual content. S.B. Patten: substantial contribution to concep-
`tion or design of the work, interpretation of data, revising the work crit-
`ically for intellectual content. C.S. Kwon: substantial contribution to
`acquisition of data and revising the work critically for intellectual content.
`J. Dykeman: substantial contribution to acquisition, analysis, and inter-
`pretation of data and revising the work critically for intellectual content.
`T. Pringsheim: obtained study funding and provided substantial contri-
`bution to conception or design of the work, interpretation of data, revis-
`ing the work critically for
`intellectual content. D.L. Lorenzetti:
`substantial contribution to conception or design of the work, interpreta-
`tion of data, revising the work critically for intellectual content. N. Jette:
`obtained study funding and provided substantial contribution to concep-
`tion or design of the work, acquisition, analysis, and interpretation of
`data, drafting the work, and revising it critically for intellectual content.
`
`ACKNOWLEDGMENT
`This study is part of the National Population Health Study of Neurological
`Conditions. The authors thank the membership of the Neurological
`Health Charities Canada and the Public Health Agency of Canada for their
`contributions and Dr. Emilio Perucca and Dr. Solomon L. Moshé for
`feedback.
`
`STUDY FUNDING
`Funding provided by the Public Health Agency of Canada to N. Jetté
`and T. Pringsheim. The opinions expressed in this publication are those
`of the authors/researchers and do not necessarily reflect the official views
`of the Public Health Agency of Canada.
`
`DISCLOSURE
`K. Fiest held a studentship from Alberta Innovates Health Solutions during
`the study period. K. Sauro held a studentship from Alberta Innovates
`Health Solutions during the study period. S. Wiebe holds the Hopewell
`Professorship of Clinical Neurosciences Research from the Hotchkiss Brain
`Institute. S. Patten is a Senior Health Scholar with Alberta Innovates Health
`Solutions. C. Kwon, J. Dykeman, T. Pringsheim, and D. Lorenzetti report
`no disclosures relevant to the manuscript. N. Jette holds a Canada Research
`Chair in Neurologic Health Services Research and held an Alberta Inno-
`vates Health Solutions Population Health Investigator Award during the
`study period. Go to Neurology.org for full disclosures.
`
`Received March 15, 2016. Accepted in final form October 12, 2016.
`
`REFERENCES
`1.
`Fiest KM, Birbeck GL, Jacoby A, Jette N. Stigma in epi-
`lepsy. Curr Neurol Neurosci Rep 2014;14:444.
`
`302
`
`Neurology 88 January 17, 2017
`ª 2016 American Academy of Neurology. Unauthorized reprod