Prevalence and Temporal Trends of Epilepsy in Children: A Retrospective Birth Cohort Study

Epilepsy, a chronic neurological disease characterized by recurrent unprovoked seizures [1], has been recognized as a major public health concern by the World Health Organization [2]. Although it is a condition that affects individuals of all ages, the incidence is higher in children, with the highest rate in the first year after birth [3]. Epilepsy in children has wide-ranging, long-term consequences, including negative impacts on physical, cognitive, emotional, and social development [4].

Recent evidence on the prevalence of epilepsy in children is scarce, and studies on temporal trends are further limited. An international meta-analysis of studies published from 1985 to 2013 reported a pooled prevalence estimate in children under 18 years ranging from 5.2 (95% CI: 3.5–7.6) to 8.9 (6.6–11.9) per 1,000 children [5], with higher prevalence in developing countries. In North America, the reported prevalence of epilepsy ranged from 2.5 to 5.5 per 1,000 children in Canadian studies [6‒10] and from 4.71 to 7.31 per 1,000 children in the United States (US) [11‒15]. In Europe, incidence rates have shown a decreasing trend until 2008 [16‒18], particularly among children aged under 10 years [3, 16]. In North America, a study conducted between 1940 and 1980 showed that the prevalence of epilepsy decreased among children aged 0–9 years, while it increased among adolescents aged 10–14 years over the same period [19].

It has been shown that the prevalence varies according to different characteristics: (a) demographic characteristics, with higher prevalence among boys [6, 20‒22] or children of young mothers [23]; (b) birth characteristics, with higher rates among children with a gestational age of 25–26 weeks or over 41 weeks [24]; and (c) maternal characteristics, with higher prevalence among those from lower socioeconomic backgrounds [8, 11, 20]. However, none of these studies has examined these disparities over time and some are subject to inconsistent results or limited data. According to ethnicity and race, no association was found in some studies [20] while others showed higher prevalence among African Americans [21]. Besides, only two studies have looked at the situation of children from immigrant backgrounds and reported inconsistent results, probably due to different populations and methodologies. While no significant change in the prevalence of epilepsy by countries of birth (Israel vs. not) was found in Israel in a restricted 17–18-year male population [25], Wandell et al. [26] found a lower risk of epilepsy in second-generation immigrant children compared to children from Swedish-born parents after adjustment on education and deprivation. In a country like Canada, where the immigrant population represents a quarter of the total population [27], these questions seem even more relevant. In addition, all these studies have never examined these disparities over time.

With a better understanding of epilepsy trends, overall and according to individual- or area-level characteristics, we speculate that it would be easier to target actions to reduce the burden of the disease and thus reduce its impact on the child’s development and social integration. This study aims to systematically examine (1) the prevalence of epilepsy diagnosed by age 18 among children born between 2002 and 2020, (2) temporal trends in the prevalence of epilepsy diagnosed by age 8, by their year of birth (2002–2012), overall, and by maternal and birth characteristics, in a large population-based birth cohort of children born in Ontario, Canada.

We created a retrospective birth cohort of children born between 2002 and 2020 in Ontario, the most populous province in Canada, identified in the MOMBABY database and linked to other 7 administrative datasets housed in ICES (see details in online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000540528) [28]. We followed each child from birth to death or the end of follow-up, on March 31, 2020, for the data linkage. ICES is an independent, nonprofit research institute whose legal status under Ontario’s health information privacy law allows collection and analysis of healthcare data, without consent, for health system evaluation. These datasets were linked using unique encoded identifiers and analyzed at ICES. We followed the REporting of studies Conducted using Observational Routinely collected health Data (RECORD) guidelines for observational studies [29] using routinely collected health data.

Our cohort included all in-hospital deliveries between April 1, 2002, and March 31, 2020 (N = 2,343,482). Of those, we excluded stillbirths, children with gestational age outside of the 21–43 weeks range, and children with data recording errors (e.g., dates of diagnosis that occurred before birth), yielding 2,335,568 children for our analytic sample (online suppl. Fig. S1). For temporal trend analyses, we restricted our population to children born up to March 31, 2012 (N = 1,405,271), to have an equal follow-up time of 8 years for all children by year of birth because more recent births had shorter follow-up times resulting in lower prevalence. The age cut-off of 8 years was determined to capture a significant proportion of cases diagnosed in childhood (84.4% of cases in our cohort, online suppl. Table S2), consistent with other studies [30‒33].

Epilepsy diagnosed before the age of 18 years was defined either as (1) an inpatient hospitalization with a diagnosis of epilepsy or as (2) three or more outpatient diagnoses of epilepsy based on physician billing codes, separated by at least 30 days within 2 years [34‒36], based on the International Statistical Classification of Diseases-Tenth Revision (ICD-10) codes (G40, G41, or F80.3). The algorithm has been used in pediatric studies [37] but validated in the adult population and has a sensitivity of 73.7%, a specificity of 99.8%, a positive predictive value of 79.5%, and a negative predictive value of 99.7% [36].

We examined child sex (male/female), plurality (single/multiple), presence of congenital malformations based on one hospitalization or one outpatient visit with a corresponding code between ages 0–4 years (yes/no), gestational age at birth in completed weeks (<28, 28–31, 32–33, 34–36, 37–38, 39–40, >40 weeks) as well as a binary indicator of preterm birth (<37), and birthweight for sex and gestational age categorized into small (<10th centile), appropriate (10th–90th), and large (>90th) for gestational age [38]. We merged the <28 and 28–31 weeks for temporal analyses to ensure a sufficient number in each category (online suppl. Table S3).

Maternal individual-level characteristics included parity (0, 1, and 2 or more previous live births), age at delivery (<20 years, 20–24, 25–29, 30–34, 35–39, or at least 40), maternal history of neurodevelopmental disorder (epilepsy, autism spectrum disorder, and cerebral palsy) before gestational period (none, epilepsy, or other disorders) and receipt (or not) of the Ontario Drug Benefit (ODB), the provincial drug program benefits for low-income individuals [39] (online suppl. Table S3).

Based on the mother’s residential postal code recorded at the child’s birth date, we included the neighborhood income quintiles and two dissemination area-specific Ontario Marginalization (ON-Marg) indices – material deprivation and ethnic diversity (online suppl. Table S1). Quintile 1 refers to the least deprived/diverse and quintile 5 to the most deprived/diverse [40]. We also considered the immigration characteristics of the mother based on data from Immigration, Refugees and Citizenship Canada, which captures immigrants landing in Ontario since January 1985 [41] (online suppl. Table S1): immigration status (yes/no), immigration admission category: economic class (immigrants selected based on skills and abilities), family class (immigrated for family union), resettled refugees or protected persons in Canada, or other immigrants, duration of residence in Canada (<10/10 years or more), immigrant mother’s country of origin classified by geographic areas and by income level according to the 2006 World Bank country classification among immigrants [42].

We estimated the prevalence of epilepsy in children before 18 years, born between 2002 and 2020 per 1,000 live births with a 95% confidence interval, overall, and by the birth and maternal characteristics, and the crude risk differences (RDs) by those characteristics, using multiple univariable analyses on complete cases. We used Poisson regression, with the number of births per year as an offset term, to estimate temporal trends of the prevalence of epilepsy diagnosed by age 8, by year of birth (2002–2012). We visually presented these prevalence trends by year of birth using figures. We accounted for the correlation between children born to the same mother using a clustered sandwich estimator for the variances [43]. We examined both linear and nonlinear trends over time and compared the fit of a linear model with a spline model (natural cubic spline with 3 knots identified and placed at the default quantiles) using the Akaike Information Criterion and Bayesian Information Criterion [44] (linear model showed a better fit), but we also tested the overall association between prevalence and the year of birth for the linear pattern of temporal trends with a likelihood test ratio.

Among the 2,335,568 children, 18,936 were diagnosed with epilepsy, with an overall prevalence of 8.1 (95% CI: 8.0–8.2) per 1,000 live births among children between 0 and 17 years. The median age at diagnosis was 3 years, 2 months with an interquartile range of 13 months to 6 years, 11 months. Approximately 50% were diagnosed before the age of 2 years and almost 85% before the age of 8 years (online suppl. Table S2). Participant characteristics were nearly identical between those included in the study and the whole cohort (online suppl. Table S4).

The rates of epilepsy among children aged 0–8 years showed a statistically significant slight increase over time by years of birth (p value for linear trend = 0.04), from 6.9 (95% CI: 6.2–7.6) in the 2002 birth cohort to 7.3 (95% CI: 6.6–8.1) per 1,000 live births in children born in 2012 (Fig. 1).

Epilepsy prevalence was slightly higher in males compared with females (RD: 0.9 per 1,000 live births, 95% CI: 0.6–1.1) (Table 1); however, the difference appeared to decrease over time by birth year cohort (Fig. 2a). Epilepsy decreased with advancing gestational age at birth, following a gradient, with an RD between 3.4 (95% CI: 2.8–3.9) for late preterm and 17.6 (95% CI: 14.3–20.8) for children born between 21 and 27 weeks of gestation (Table 1). These disparities persisted over time, with no fluctuations across different gestational ages (Fig. 2b). Regarding weight for gestational age, children born small had a higher prevalence of epilepsy (RD: 2.4, 95% CI: 1.9–2.8), and this difference persisted over time (Table 1; online suppl. Fig. S2).

The overall prevalence was also higher for multiple births than singletons (Table 1), regardless of gestational age, but with a decrease over time among multiple births between 2002 and 2012, resulting in a narrowing gap between the two groups (Fig. 2c). As expected, children with congenital malformations had a higher prevalence (RD: 13.5, 95% CI: 13.0–14.1) (Table 1), and this substantial gap persisted over time (Fig. 2d).

Children born to mothers under the age of 25 at delivery, to nulliparous mothers, or to mothers diagnosed with neurodevelopmental disorders had a substantially higher prevalence of epilepsy, particularly among those born to mothers with diagnosed epilepsy (RD: 18.1, 95% CI: 15.5–20.6) (Table 2; online suppl. Fig. S3a). A slight increase over time was observed among children of mothers who delivered under the age of 20 or over 40 years (Fig. 3a) and for all categories of parity (Fig. 3b).

Children whose mothers were eligible for the ODB program had a higher prevalence (RD: 2.8 95% CI: 2.5–3.1). We also observed a graded relationship between neighborhood income quintiles and the ON-MARG indices, particularly about material deprivation. Epilepsy was higher among children residing in the lowest income (RD: 1.5, 95% CI: 1.2–1.9) and most deprived quintile (RD: 1.9, 95% CI: 1.5–2.2). These socioeconomic disparities remained stable over time (Fig. 3c, d; online suppl. Fig. S3b). For the ethnic diversity index, children living in areas with higher ethnic diversity had a lower prevalence of epilepsy (RD: −1.0, 95% CI: −1.4 to −0.6), but the gap between the most and the least diverse areas tended to widen (Table 2; Fig. 3e).

Children born to immigrant mothers had a lower prevalence of epilepsy in all children (Table 2), and the difference persisted over time (Fig. 3f). Of children born to immigrant mothers, epilepsy prevalence was lower among those born to economic-class immigrant mothers (6.1 vs. ∼7.1 per 1,000) than in other classes. While we found no clear disparities according to the income level of the maternal country of origin, we observed lower prevalence for those coming from East Asia, the Pacific, the Middle East, North Africa, Europe, and Central Asia (Table 2). These disparities remained stable over time (Fig. 3g, h; online suppl. Fig. S3c). The prevalence of epilepsy was not different by the maternal duration of residence in Canada among children born to immigrant mothers (Table 2; online suppl. Fig. S3d).

In this large population-based cohort study of live births in Canada spanning an 18-year period, the prevalence of epilepsy was 8.1 (95% CI: 8.0–8.2) per 1,000 live births. We found a slight increase in prevalence over time. Extremely premature infants and those born with congenital malformations had a considerably higher prevalence of epilepsy than those born at term or later, or without malformations. Boys, children born from multiple pregnancies, those born to mothers born in Canada or under 25 years of age, and those living in lower socioeconomic neighborhoods also had a higher prevalence of epilepsy compared with other groups. These disparities persisted over time, except for plurality and sex, where the gaps were reduced.

Our estimated prevalence falls within the range of the pooled estimates reported by the international meta-analysis based on previous studies conducted over earlier periods all over the world. However, our estimate is at the upper end of this range, predominantly corresponding to developing countries in the meta-analysis, thus exceeding prevalence rates typically found in high-income countries [5]. Our estimated prevalence was also higher than those of previous Canadian studies, reporting prevalence from 2.5 to 5.5 per 1,000 children [6‒10]. It could be due to different definitions (inclusion of epileptic seizures), case ascertainment based on self-report and diagnosis codes, different coding practices over time [45], or the use of an algorithm validated in adults in our study. The change in the definition of epilepsy in 2014, including a single unprovoked seizure as a part of the global definition [46], could also explain this discrepancy, but other studies reported no impact on the pediatric population [47]. Alternatively, the higher prevalence estimated in our data might reflect the increasing trend over time observed in our study.

This increasing trend also differs from that observed in the literature. In the UK, a study of children aged 0–7 born between 2001 and 2008 that identified epilepsy using records on antiepileptic drug prescription, epilepsy diagnoses, or epilepsy symptoms found a decrease in annual incidence of up to 9% [18]. Other studies conducted in the US or Denmark have also shown a decrease in prevalence between 1940 and 2002 [3, 16, 17, 19]. This difference with our results could be explained by differences in the sources of data to identify epilepsy cases (e.g., administrative dataset, register) and/or in the definitions of epilepsy between studies (e.g., based on antiepileptic drugs prescription) and that they do not use the same algorithms. Another explanation could be the fact that our study is based on more recent data and that our results may therefore indicate a slight resurgence of epilepsy, requiring an understanding of the reasons behind this increase.

The disparities observed in our study according to sex [6, 20‒22], gestational age [24], maternal age at childbirth [23], and socioeconomic status [8, 11, 18, 20] are consistent with those found in the literature. However, to our knowledge, this study is the first to explore differences in epilepsy prevalence according to the immigration status of the mother. Previous research in this area has been both scarce and inconsistent, focusing primarily on the immigration status of the child rather than that of the mother. For instance, one study found no significant difference in epilepsy prevalence based on the child’s country of birth [25], whereas other research reported either similar or improved neuro-developmental outcomes, including reduced epilepsy risk, in immigrant children [26]. The observed lower risk of epilepsy among children of immigrant mothers in our study could potentially be attributed to the “healthy migrant effect” [48], or to cultural variations in the diagnosis of epilepsy, which is highly stigmatized in some countries [49].

Finally, our results regarding disparities by ethnicity are difficult to compare with the literature. Most studies focused on individual characteristics such as race, showing higher prevalence rates in African American children compared to white children [21, 50], while we used an aggregated indicator, founding higher rates in less ethnically diverse areas (fewer ethnic minorities or recent immigrants). Moreover, direct comparability is difficult knowing that in Canada visible minorities are often immigrants, and African Americans in the US are often US born.

Limitations of this study include the validity of epilepsy identified using administrative data. Although our definition has the highest sensitivity (73.7%) and positive predictive value (79.5%) among existing definitions [51, 52] and has been used in other studies [37], it has been validated only in adults [36]. Second, we have arbitrarily defined a cut-off at the age of 8 to ensure an equal follow-up time and to maximize the analytical sample for the temporal trends analyses. These analyses therefore did not include cases where the onset of epilepsy occurred at age 9 or older. Nevertheless, most cases of epilepsy are diagnosed before the age of 5 years [30‒33, 53], and it is unlikely that the age of onset of epilepsy changed substantially over time. Therefore, under-ascertainment of cases with the age cut-off would have been non-differential over time. In addition, the age at onset varies according to epilepsy type [30‒33], but we could not examine the prevalence and temporal trend by subtype owing to the lack of information on epilepsy subtypes.

Strengths of our study include the almost complete inclusion of children born in Ontario, which minimized potential selection biases and improved the generalizability of our results. Our large sample size enabled us to examine the prevalence of epilepsy in children not only overall but also by birth and maternal characteristics and investigate the temporal trends of these prevalence trends. Moreover, in a country where almost 1 in 4 persons is an immigrant [27], we had access to information on the mother’s immigration status and other detailed characteristics of immigrant mothers such as duration since immigration, economic category, and region of origin, which enabled us to describe the prevalence of epilepsy according to these criteria, which have been understudied in the literature.

This study provides updated estimates of the prevalence of epilepsy in children in a high-income country, as well as the first examination of rates according to birth and maternal characteristics over time in Canada. We found a slight increase in overall rates over time in children aged 0–8 years and persistent disparities by gestational age, socioeconomic characteristics, and maternal immigration status. Our results highlight the need for continued surveillance of epilepsy occurrence to monitor whether the increasing trend observed in our study would persist, to ensure efforts are made to limit the increase, to understand potential causes behind it, and to understand the ongoing disparities by characteristics. Further studies exploring the mechanisms underlying differences in risk will be needed to answer these questions.

Ethics approval was received from the Institutional Review Board of the Faculty of Medicine and Health Sciences at McGill University. ICES is a prescribed entity under Ontario’s Personal Health Information Protection Act (PHIPA). Section 45 of PHIPA authorizes ICES to collect personal health information, without consent, for the purpose of analysis or compiling statistical information with respect to the management of, evaluation or monitoring of, the allocation of resources to or planning for all or part of the health system. Projects that use data collected by ICES under section 45 of PHIPA, and use no other data, are exempt from REB review. The use of the data in this project is authorized under section 45 and approved by ICES’ Privacy and Legal Office.

The authors have no example conflicts of interest to disclose.

This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health (MOH) and the Ministry of Long-Term Care (MLTC). This study also received funding from the Canadian Institutes of Health Research (CIHR). Parts of this material are based on data and information compiled and provided by MOH, CIHI, and IRCC. The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred.

B.D. conceptualized and designed the study, carried out the analyses, drafted the initial manuscript, and reviewed and revised the manuscript. S.Y. conceptualized and designed the study, and reviewed, and revised the manuscript. E.B. coordinated and supervised data collection and participated in data creation and data management. L.R. coordinated and supervised data collection, participated in the conceptualization of the study, and reviewed the manuscript. A.M.A. and J.A.H. participated in the conceptualization of the study and reviewed the manuscript. All authors critically reviewed the manuscript, approved the final version as submitted, and agreed to be accountable for all aspects of the work.

Parts of this material are based on data and/or information compiled and provided by Immigration, Refugees and Citizenship Canada (IRCC) current to 2020. However, the analyses, conclusions, opinions, and statements expressed in the material are those of the authors and not necessarily those of IRCC.

The dataset from this study is held securely in coded form at ICES. While legal data sharing agreements between ICES and data providers (e.g., healthcare organizations and government) prohibit ICES from making the dataset publicly available, access may be granted to those who meet pre-specified criteria for confidential access, available at www.ices.on.ca/DAS (email: das@ices.on.ca). The full dataset creation plan and underlying analytic code are available from the authors upon request, understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.