Epilepsia. 2015 Nov;56(11):1774-83. doi: 10.1111/epi.13187.

Childhood-onset epilepsy five decades later. A prospective population-based cohort study.

Sillanpää M1, Anttinen A2, Rinne JO3, Joutsa J3, Sonninen P4, Erkinjuntti M5, Hermann B6, Karrasch M7, Saarinen M8, Tiitta P9, Shinnar S10.
  • 1Department of Child Neurology, Turku University Hospital, Turku, Finland.
  • 2Division of Clinical Neurosciences, Turku University Hospital, Turku, Finland.
  • 3Turku PET Centre, Turku, Finland.
  • 4VSKK, TYKS-SAPA, Turku, Finland.
  • 5Department of Clinical Neurophysiology, Turku University Hospital, Turku, Finland.
  • 6Department of Neurology, University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin, U.S.A.
  • 7Department of Psychology, Abo Akademi University, Turku, Finland.
  • 8Department of Public Health, University of Turku, Turku, Finland.
  • 9Turku Teacher Training College, University of Turku, Turku, Finland.
  • 10Departments of Neurology, Pediatrics and Epidemiology and Public Health, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, New York, U.S.A.



OBJECTIVE: To study the impact of childhood-onset epilepsy on a variety of outcomes across the life span.

METHODS: A population-based cohort of 245 subjects with childhood-onset epilepsy was assessed for outcomes at 45 years. In addition, 51 of 78 surviving subjects with uncomplicated epilepsy and 52 of 99 originally matched controls participated in a detailed evaluation including electroencephalography (EEG), imaging, and laboratory studies at 50 years.

RESULTS: Of 179 surviving subjects, 61% were in terminal 10-year remission and 43% in remission off medications. At 45 years, 95% of the idiopathic group, 72% of the cryptogenic group, and 47% of the remote symptomatic group were in terminal remission (p < 0.001). Abnormal neurologic signs were significantly more common in subjects with uncomplicated epilepsy than in controls. Mortality during period 1992-2012 was higher in subjects than in controls (9% vs. 1%, p = 0.02). The rate of 3T MRI abnormalities was higher in subjects than in controls (risk ratio [RR] 2.0; 1.3-3.1) specifically including findings considered markers of cerebrovascular disease (RR 2.5; 1.04-5.9). Even subjects with idiopathic epilepsy had higher rates of imaging abnormalities than controls (73% vs. 34%, p = 0.002).

SIGNIFICANCE: Long-term seizure outcomes are excellent and a function of etiology. The presence of imaging abnormalities suggestive of vascular disease may put these subjects at higher risk for clinically evident stroke and cognitive changes as they age.

KEYWORDS: Childhood-onset epilepsy; Comorbidity of epilepsy; Long-term outcome; Prospective cohort study

PMID: 26434398





Matti Sillanpää, on behalf of the *TACOE study team.

Department of  Child Neurology, University of Turku and Turku University Hospital, 20500 Turku, Finland


Correspondence to:

Prof. Matti Sillanpää, MD PhD

Department of Public Health

University of Turku 20014, Turku, Finland

Tel. +358-2-23337368, Fax +358-2-333 8439

Email: matsilla@utu.fi


Running title: Fifty-year outcome of childhood epilepsy

Key words: childhood-onset epilepsy; long-term outcome; prospective cohort study; comorbidity of epilepsy

*The TACOE study tem includes Anttinen A, Erkinjuntti M, Hermann B, Joutsa J, Karrasch M, Rinne JO, Saarinen M, Shinnar S, Sillanpää M, Sonninen P, and Tiitta P.



The Turku Adult Childhood Onset Epilepsy (TACOE) study is based on follow-up of 50 years from 1964 of the well-known population-based cohort of children with epilepsy residing in the catchment area of Turku University Hospital, Finland[1-3]. The ultra-long follow-up and combination of population controls allows for a unique opportunity to life span studies of the impact of childhood-onset epilepsy on life in adulthood.



The study subjects, aged 0-15 years with a history of two or more unprovoked seizures residing in the catchment area of Turku University Hospital, Turku, Finland, at the end of 1964 two or more unprovoked seizures, were identified from the files of all relevant hospitals in southern Finland, primary health care records and the National Health Service records, a register of all patients residing in Finland. Follow-up data were collected uninterruptedly from semi-structured comprehensive questionnaires every fifth year and by written permission from healthcare documents between the questionnaire check-ups and in 1992 from in-depth clinical and laboratory examinations.

The present study reports data from all subjects at the 2007 questionnaires (N=133) (later referred to as “the 45-year follow-up study” data), and data on physical comorbidities from 51 subjects with uncomplicated epilepsy and 52 controls at 2010 (later referred to as “the 50-year follow-up study”.

The 45-year follow-up study

The 2007 study at 45 years included 245 subjects, of whom 66(27%) had deceased. Of 179 surviving subjects, 133(71%) satisfactorily completed the mailed extensive semi-structured questionnaires on health status, seizure history, antiepileptic medication, and comorbidities. Death data were available from the Finnish National Death Register. Remission of epilepsy was defined in accordance with the newer ILAE definition as a seizure-free period of 10 years off medications for the last 5 years[4]. Terminal remission was defined as remission at last follow-up. Epilepsy without major neurological comorbidity was defined as uncomplicated. Abnormal neurological status was defined as one or more abnormal findings on examination. The study design and some earlier results are previously reported[2,3,5-9].

The 50-year follow-up study

The 2012 study at 50 years comprised all 99 patients with uncomplicated epilepsy and the 99 controls, except for two subjects and one control whose mother tongue was other than Finnish. Fifty-one subjects and 78 controls participated. Dropout analysis showed no significant differences in the relevant seizure variables between the participating and non-participating subjects. The study, performed for both subjects and controls, included a detailed clinical and neurological examination; serum cholesterol, high density lipoprotein (HDL), low density lipoprotein (LDL), and triglycerides; conventional EEG recording; and 3T magnetic resonance imaging (MRI).

Fisher’s exact test and univariate and multivariate modified Poisson regression models[10] were used. The results are given as risk ratios (RRs) and 95% confidence intervals (95%CIs), Bonferroni-corrected in multi-level predictors. A p-value <0.05 (two-tailed) was considered significant. The Institutional Review Board approved the study design (Diary No. 120/2008/26.1.2009 §454). Written consent was obtained from both subjects and controls.



The 45-year follow-up study

At 45 years, outcomes were available on 199 (81%) of the original 245 subjects. Among 133(74%) of 179 surviving patients, the mean age was 47.4 years and mean calendar age 50.5 years. At least one 10-year remission had been entered by 99 (74%) of 133 patients and 62 (47%) had attained at least one 10-year remission with last 5 years without medications. Relapse following remission was found in 29 (29%) of those who entered 10-year remission and 3 (5%) of those who attained 10-year remission without medications. Eighty-eight (89%) of these 99 were in 10-year terminal remission and 67 (68%) were in 10-year terminal remission without medications for at least 5 years. Both the 5-year or 10-year terminal remission rates were the higher, the longer follow-up was. During extraordinary long follow-up, the most subjects with the 5-year remission rate approached that of the 10-year remission rate. At 45-year follow-up, 81 (61%) of the 133 surviving available subjects were in 10-year terminal remission on or off medications and 57 (43%) in remission off medications.

Of the 133, 69 (52%) had idiopathic/cryptogenic etiology and 64 (48%) had symptomatic etiology. With more than 40 years of follow-up, the terminal remission rates including remission off medications among subjects with cryptogenic etiology approached the rate of the idiopathic group. The terminal remission rates on or off medications of the idiopathic group and cryptogenic group were significantly higher (95% and 72%, respectively) than the rate of remote symptomatic group (47%). The corresponding rates are even more striking in comparison between the three groups off medications (81%, 63% and 25%, respectively). On multivariate analysis, the independent predictors of 10-year terminal remission on or off medications were idiopathic vs. symptomatic etiology (2.0; 1.4–2.8) and cryptogenic vs. symptomatic etiology (1.5; 1.01–2.3). On similar multivariate analysis of attaining 10-year terminal remission off medications >5 years, independent predictors were idiopathic vs. symptomatic etiology (RR 1.9; 1.005–3.7) and a childhood IQ of 71 or more (RR 2.5; 1.2–5.5).


The 50-year follow-up study

The study included 51 (57%) of 90 surviving subjects with uncomplicated epilepsy with the mean age of 56.1 and 52 controls with the mean age of 56.0. Compared with controls (53% of 98 surviving controls), there was no significant differences in basic or vocational education, civil status, employment status or smoking habits, yet there were significantly fewer subjects with children (61% vs. 90%, p<0.001), fewer subjects consuming alcohol more than once a week (18% vs. 31%, p=0.01), and fewer holders of driver’s licenses (69% vs. 90%, p=0.02).

All subjects participating in 50-year follow-up had uncomplicated epilepsy with no initial neurological impairment at onset[2] but, 50 years later, abnormal neurological signs were two times as common in subjects as in controls (RR 2.0; 1.3-3.1, p=0.02). There was also a trend toward more neurological disorders in the subjects but no differences in the rate of the non-neurological disorders (29% vs. 13%, p=0.06). Subjects not in terminal remission had a slightly higher frequency of abnormal neurological findings (71% vs. 35%, p=0.05) than subjects in remission and any somatic comorbidity (p=0.09). In addition, there was a higher mortality among subjects than controls during 1992–2012 (9% vs. 1%, p=0.02). 

Congenital abnormalities and atrophy were non-significantly more common in the subjects. However, findings considered as markers of cerebrovascular disease were significantly more common in the subjects than controls (RR 2.5; 1.04–5.9), even when adjusted for other neurological diseases and abnormal neurological status. The frequency of MRI findings including markers of cerebrovascular disease were present even in the idiopathic group more often than among controls (73% vs. 34%, p=0.002). Eleven subjects not in terminal remission had more often abnormal MRI findings than the 36 who were in remission (100% vs. 58%, p=0.009). Even those in remission had higher rates of MRI abnormalities than controls (58% vs 34%, p=0.03).

EEG was normal in 43 (84%) of 51 subjects and in all 52 controls. All eight patients with EEG abnormalities had imaging abnormalities, were still on medications and five still having seizures. None of the chemical laboratory examinations revealed any significant differences in the proportion of subjects and controls with abnormal laboratory measurements.



This prospective 50-year follow-up of a population cohort with childhood-onset epilepsy and matched controls also followed for many years not only provided further information on epilepsy outcome, but comprehensively characterized somatic and neurological comorbidity in late middle age. Prior studies including this one have focused on five-year remission data which were previously used[11,12]. In our study, within 30 years, the most dominant predictor of outcome is etiology. Most subjects with idiopathic etiology of epilepsy achieve “resolved epilepsy” status. By 45–50 years, but not prior to, the remission rate on or off medications is approached by cryptogenic cases approaches that of idiopathic cases. In subjects with symptomatic epilepsy, the remission rate remains low even after 45–50 years despite the high mortality in the more severely affected cases[3,9] and the availability of new drugs. Even use of the very stringent definition of “resolved epilepsy”[4] did not  mean absence of subsequent relapse. Our conclusion is that patients with childhood-onset epilepsy who survive until late middle age have excellent seizure outcomes. This is particularly in those who have idiopathic epilepsy, but also in cryptogenic cases in the long run. Knowledge of the specific syndrome, however, is eventually decisive as a predictor of relapse even after having attained a 10 year remission with 5 years off medications.

The classification of etiologies used was current at the time the cohort was assembled and at the time of the major prior analysis of seizure outcomes[2]. Our symptomatic cases do correspond to the “structural/metabolic” etiology in the new classification and our “cryptogenic cases” do correspond to the “unknown cause” in the new classification[13]. We have previously described the epilepsy syndromes in this cohort[14].

The concept of “uncomplicated” epilepsy basically included the idiopathic and cryptogenic group with no major initial neurological abnormality. No imaging was available at onset of epilepsy in this cohort which was assembled well before the availability of modern imaging techniques. When re-examined in middle age, subjects had a significantly higher rate of clinical neurological abnormalities than did controls. A trend was also found towards more other neurological comorbidities but not of non-neurological somatic comorbidities in subjects than controls. The reasons are unclear but may be related to the epilepsy, its underlying cause or drug therapy.

On imaging, the rate of abnormalities in the controls was not unexpected given the mean age of 56[15]. What was striking was the high rate of abnormalities in the subjects. In this cohort of presumed idiopathic and cryptogenic cases, few abnormalities were better related to cerebrovascular disease than to potential etiological factors. The abnormalities were present even in idiopathic cases in terminal remission for many years. This is particularly striking as no differences were noted neither in serum markers of cerebrovascular disease such as cholesterol, triglycerides, HDL, and LDL nor in other risk factors such as smoking, obesity, diabetes or hypertension. This finding suggests that it is the epilepsy, including the underlying etiology, seizures, and treatment that may be associated with these early changes. Both subjects and controls are approaching the age with greater cerebrovascular risk and a trend towards more cerebral atrophy. Whether these imaging findings will translate into an increased rate of clinically evident disease such as stroke or cognitive decline can only be seen in during extended follow-up.

There are some limitations in this study. The main limitation is the small original target population (n=409,000 at the end of 1964[1] with approximately 20% of the population aged less than 16 years) to study such a rare disorder as epilepsy.  However, the sampling of the initial study subjects was very successful and, the participation rate of 71% of surviving study subjects with only 0.64% annual attrition rate was very satisfactory. While the participants of the 50-year followed sub-sample of subjects with uncomplicated epilepsy were fewer than expected, they did not significantly differ from the participants. The most reasons for non-participation were understandable, such as very heavy two-day study program; loss of two-day salary of employed subjects without compensation; other work obstacles; and tiredness to many medical visits. 

Our data provide further evidence of the long term effects on childhood onset epilepsy and its treatment even in remitted subjects who are without medications for years. Except for the symptomatic group, seizure outcomes are excellent and consistent with what was reported at 30 and 40 years though we have now utilized the new ILAE definition of 10 years terminal remission. The truly novel imaging findings, suggestive of vascular disease, may put the subjects to additional risk at age of naturally higher risk for stroke and cognitive impairment.



We owe to Leila Kesäläinen for her excellent work as a study coordinator and technical assistant. The study was financially supported by the national Governmental Research Grant (EVO) and a CURE Innovators Award (USA).



[1] Sillanpää M. Medico-social prognosis of children with epilepsy. Epidemiological study and analysis of 245 patients. Acta Paediatr Scand 1973;237 (Suppl 68):1-104.

[2] Sillanpää M, Jalava M, Kaleva O, Shinnar S. Long-term prognosis of seizures with onset in childhood. N Engl J Med 1998;338:1715-22.

[3] Sillanpää M, Shinnar S. Long-term mortality in childhood-onset epilepsy. N Engl J Med 2010;363:2522-9.

[4] Fisher RA, Acevado C, Arzimanoglou A et al. ILAE official report: A practical clinical definition of epilepsy. Epilepsia 2014;55:475-82.

[5] Sillanpää M, Shinnar S. Status epilepticus in a population-based cohort with childhood-onset epilepsy in Finland. Ann Neurol 2002;52:303-10.

[6] Sillanpää M, Schmidt D. Prognosis of seizure recurrence after stopping antiepileptic drugs in seizure-free patients. A long-term population-based study of childhood-onset epilepsy. Epilepsy Behav 2006;8:713-19.

[7] Sillanpää M, Schmidt D. Early seizure frequency and aetiology predict long-term medical outcome in childhood-onset epilepsy. Brain 2009;132:989-98.

[8] Sillanpää M, Saarinen M, Schmidt D. Long-term risks following first remission in childhood-onset epilepsy. A population-based study. Epilepsy Behav 2012;25:145-52.

[9] Sillanpää M, Shinnar S. SUDEP and other causes of mortality in childhood onset epilepsy. Epilepsy Behav 2013;28:249-55.

[10] Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004;159:702-6.

[11] Geerts A, Arts WF, Stroink H et al. Course and outcome of childhood epilepsy: A 15-year follow-up of the Dutch study of epilepsy in childhood. Epilepsia 2010;51:1189-97.

[12] Berg AT, Testa TM, Levy SR. Complete remission in nonsyndromic childhood-onst epilepsy. Ann Neurol 2011;70:566-73.

[13] Berg AT, Bercovic SF, Brodie MJ et al. Revised terminology and concepts for organization of seizures and epilepsies: Report of the ILAE commission on classification and terminology, 2005-2009. 2010;51:676-85.

[14] Sillanpää M, Jalava M, Shinnar S. Epilepsy syndromes in patients with childhood-onset seizures in Finland. Pediatr Neurol 1999;21:533-7.

[15] Morris Z, Whiteley WN, Longstreth WTJ et al. Incidental findings on brain magnetic resonance imaging: Systematic review and meta-analysis.  BMJ 2009;339.



Multiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier SchönmannMultiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier Schönmann