Acm zyxwvutsrqponmlkjihgfed Neurol zyxwvutsrqponmlkji Scund zyxwvutsrqponm IY93’ 87: 335-34/ Printed zyxwvutsrqponm in BelKium zyxwvutsrqponmlk - ull righi, reserved Copyright 0 Munksguard 1993 zy ACTA NEUROLOGICA SCANDINAVICA ISSN 0001-6314 Social competence of people with epilepsy: a new methodological approach I Sillanpaa M, Helenius H. Social competence of people with epilepsy: a I zyxwvuts M. Sillanpal’, H. Helenius’ new methodological approach. Acta Neurol Scand 1993: 87: 335-341. zyxwvut 0 Munksgaard 1993. A new multivariate method, a stepwise cumulative regression analysis, which allows a polychotomous outcome variable was used to analyse a representative population sample of children with epilepsy as adults, after a follow-up of 30 years from onset of seizures. On follow-up, 63% of subjects had been in terminal remission for 2 3 years and 48% were both seizure-free and off medication. More than a half (58%) felt completely independent in daily activities and almost as many (57%) were successfully employed. Good communication ability, good intelligence and seizure freedom were significant predictors of a good social competence. Studies on the outcome of epilepsy and its determi- nants are liable to many pitfalls which may bias the results. The definition of epilepsy may vary from study to study. The study population may be prima- rily selected, or the methods of case finding and identification may be defective or biassed. The pro- cess of data collection on the study population may show methodological weaknesses. Analysis of the data may be inappropriate or inappropriate used, or no steps may have been taken to control for con- founding factors. The time from the last seizure as a marker of seizure freedom may vary from six months to five years (1-9). The duration of follow-up may affect prognosis (10). In addition, the end var- iable or variables of many reports are dichotomous, for example between freedom or non-freedom from seizures. Even more confusing is the linkage of cases with improvement in seizure control with those showing total seizure control. To avoid these problems, we studied the long- term prognosis of epilepsy in an unselected popula- tion using well-defined criteria for case finding, in- tensive case identification methods and a long follow-up period. The novel feature of the present study was the application of a statistical control method for confounding which allowed the use of polychotomous end variables to study trends in sei- zure control during therapy. Study population and zyxwvutsrqpo methods The original study population consisted of all chil- dren, who were aged 15 years or less and living in Departments of ’ Child Neurology, Biostatistics, University of Turku, Finland i 2 Key words: people with epilepsy; long-term outcome; social competence; cumulative regression model Matti Sillanpaa, Department of Child Neurology, TYKS, 20520 Turku, Finland Accepted for publication November 9, 1992 the catchment area of the Turku University Hospi- tal in Southwestern Finland at the end of the year 1964. The total population of the area is about 470000 or almost one tenth of the total Finnish pop- ulation. The original study population comprised all children with “active epilepsy” (recurrent seizures during the last three years) during 1961 to 1964. Patients were ascertained at first evaluation as suf- fering from epilepsy personally by the author on the basis of previous history, clinical examination and laboratory investigations. Children whose seizures were occasional or situational or neonatal only, non- convulsive and without any disturbance of con- sciousness, associated with a progressive brain pro- cess, or less than three in number were excluded from the study. Patient identification was based on hospital, in- stitution and outpatient unit records, EEG studies and other laboratory investigations as well as out- patient and inpatient documents of the National Health Service, a registry of all persons resident in Finland. Thus, the study was population-based, its patient information supplemented with hospital- based data. The coverage of cases of interest was considered excellent and the study highly represent- ative of the child population with active epilepsy in the area. There were 245 children (134 boys and 11 1 girls) meeting the preset criteria. The design of the original study has been described in detail previously At the beginning of follow-up, the mean age of the children was 4.8 years (SD 1.6). The patients were followed for 30 years with serial cross-sectional stud- (1 1). 335