Perceived Impact of Childhood-onset Epilepsy on Quality of Life as an Adult
Address correspondence and reprint requests to Dr. M. Sillanpää at Department of Public Health, 20014 University of Turku, Finland. E-mail: email@example.com
Summary: Purpose: Childhood-onset epilepsy is a common disorder. The long-term impact of having childhood epilepsy on quality of life (QOL) as an adult, whether or not seizures are in remission, has not been systematically studied.
Methods: A population-based cohort of 245 children younger than 16 years with active epilepsy between 1961 and 1964 residing in the catchment area of Turku University Hospital was followed up prospectively until 1997. Of the 99 surviving cases with uncomplicated epilepsy and 99 matched population controls, 91 subjects and controls completed questionnaires on QOL and psychosocial outcomes.
Results: Of the 91 subjects, 61 (67%) were in remission off medication, 13 (14%) in remission on medications, and 17 (19%) were not in remission. Subjects on medication, whether in remission or not, had worse scores on both general measures of QOL and epilepsy-specific measures than did either controls or subjects in remission off medications. They also had significantly higher rates of unemployment (p < 0.001) and lower socioeconomic status. These differences could not be accounted for by differences in education or seizure frequency. Subjects in remission off medication had rates of employment and socioeconomic status similar to those of controls. All subjects, regardless of remission status, had lower rates of marriage and of having children than did controls (p < 0.001).
Conclusions: Childhood-onset epilepsy has a persistent long-term adverse impact on health-related quality of life. The major impact is on those still on medications as adults, whether or not they are in remission. The impact on those in remission off medications is relatively modest.
In clinical practice, the treatment of epilepsy usually focuses on control of seizures and minimizing the adverse effects of antiepileptic drug (AED) therapy. However, the factors contributing to a person's well-being and overall health-related quality of life (HRQOL) are a more complex network of issues than just seizures and medications, but include many psychosocial, economic, and societal dimensions. It has been stated that childhood-onset epilepsy might have broad indirect effects on individual's personality and psychosocial maturation (1). The indirect adverse impact can be conveyed through parental lower expectations compared with their healthy children and ambivalent overprotection (2). The socioemotional development also might be affected by negative peer group attitudes (3). In addition, academic underachievement among children with epilepsy has been noted by several authors (4). Some of these indirect adverse effects (e.g., parental and social attitudes) might persist even though the epilepsy would be in remission.
In recent years, many scales measuring QOL have been designed to be incorporated into the treatment of different patient groups (5). Generic instruments are intended to measure an individual's general perception of the position in life related to the living circumstances and environment (6,7). The SF-36 health survey (8) is one of the most commonly used generic scales (5). Generic scales have been developed for general populations and do not consider specific diseases (9,10). However, generic scales have often been used as a core structure in disease-specific QOL, for example, in epilepsy, by adding epilepsy-specific questions or an additional seizure-specific instrument (11). Despite a relatively short history, the research in formal assessment of QOL in epilepsy has produced a large number of well-established scales (6,8,10,12–14). In addition, an Impact of Epilepsy Scale, which uses a patient-centered approach, was developed (15) and has been successfully used in a large European QOL in epilepsy study (16).
We prospectively followed up a population-based cohort of 245 subjects with childhood-onset epilepsy in Finland for >30 years (17). We previously reported that, although the long-term seizure outcomes in this cohort were generally favorable, compared with population-based age-matched controls, persistent adverse effects on psychosocial function (proportion married, having children, and driving) were associated with having childhood-onset epilepsy that persisted to adult life (17). This study reports on further investigations into psychosocial outcomes in this cohort, including formal measures of HRQOL.
Patients and controls
The original study group consisted of 245 patients who were living in the catchment area of Turku University Hospital, Turku, Finland, at the end of 1964, and who had had two or more unprovoked seizures. The patients were identified, not only on the basis of hospital and institution files all over Finland but also from the National Health Service records, a registry of all persons residing in Finland at that time. All the 245 patients were examined in 1972 by one pediatric neurologist (M.S.) and enrolled in a longitudinal study of outcomes. Details of the cohort characteristics have been previously reported (17). In 1992, 176 of 245 original subjects were traceable and participating; 44 patients had died, seven emigrated, 10 could not be traced, and eight refused further participation in the study (17).
Of the 176 subjects, 100 had uncomplicated epilepsy, defined as epilepsy that was without any other initial neurologic impairment (17). In 1992, a stratified random sampling was used to choose a control subject for each of the 100 subjects with uncomplicated epilepsy, matched for sex, age, and domicile. One such subject–control pair was done technically incorrectly. The remaining 99 matched pairs constitute the sample for assessment of psychosocial outcomes and HRQOL (17,18). Patients with epilepsy and known initial neurologic impairment such as mental retardation or cerebral palsy are not included in this analysis, as their outcome is likely to be more a function of the underlying brain disorder than of their epilepsy.
In 1997, detailed questionnaires including measures of social function and HRQOL instruments described later were mailed to the 99 subjects with uncomplicated epilepsy and their matched controls. Of those eligible to participate, 91 subjects and 90 controls returned the questionnaires. The 90 control subjects participating in the present 1997 study were the same as those in our previous report of this follow-up study (17,18).
The study protocol was approved by the joint ethics review committee of the Turku University Medical School and the Turku University Central Hospital. All the subjects had given either oral or written informed consent.
Data collection in 1997
In 1997, a structured questionnaire about patients' present epilepsy, social accomplishment, and general life situation was sent by mail, including a cover letter with instructions. The questionnaire on the QOL, mailed to both the index and control subjects, included the 36-item SF-36 (8) and the Impact of Epilepsy Scale developed by Jacoby et al. (15,19). We chose these scales rather than an epilepsy QOL scale such as the QOLIE (6,10,13,14), as in our experience in other research studies, the QOLIE, although an excellent QOL measure of those with active epilepsy, was not a very good QOL measure in patients with epilepsy who were seizure free and off medications for many years, which was true for the majority of our subjects.
The SF-36 and the Impact of Epilepsy Scale were originally designed in English. Therefore they were translated into Finnish by a professional translator and back-translated by another professional translator. Thereafter, a meeting between the translators and one of the authors (M.S.) was arranged to ensure that the original questionnaire and the translation measured the same issues.
The social outcomes selected for comparison between index patients and controls were as follows: marital status, having own children, present employment status, and holding a driver's license. These particular social outcomes were earlier shown to be significant measures of social competence (17,20) and were therefore selected for comparison purposes in the present study. Socioeconomic status (SES) also was assessed. SES was defined according to the Central Statistical Office of Finland (21). The classification is based, according to the United Nations recommendations, on several aspects influencing an individual's life, including occupation and status in employment, nature of occupation, and work.
Educational outcomes were assessed in 1992. They are reported here again for purposes of comparison with the major social outcomes. As previously reported (17), primary education is the equivalent of completing sixth grade. Matriculation examination is taken after completion of 12 years of schooling and is approximately equivalent to 1 year of college in the United States. Vocational training can be taken after ninth grade and is intended to train for specific occupations.
Epileptic seizures and syndromes were classified according to International League Against Epilepsy (ILAE) guidelines (17,22–24). Any patient who had been seizure free with or without medication for ≥5 years at the time of last follow-up was considered to be in remission.
Statistical analyses were performed by using BMDP software (25) and SAS System for Windows, release 8.02. For the comparisons of SF-36, eight health concepts between subjects and controls, the impact of epilepsy scale between patients in remission and not in remission, a t test was used. For the comparisons of SF-36 in groups in remission off medication, in remission on medication, and not in remission, one-way analysis of variance was used. Tukey's test was used for pair-wise comparisons. For the comparison of social outcomes between subjects and controls, relative risks were used. Pearson's χ2 test, with Yates' correction when appropriate, and Fisher's exact test were used. All comparisons used two-tailed tests. A p value of <0.05 was considered to indicate statistical significance.
The 91 subjects included 34 (37%) male and 57 (63%) female subjects. The mean age at onset was 4.4 years (median, 4.0 years). At last follow-up in 1997, the mean age was 41.5 years. At the time of the present study, of the 91 subjects with uncomplicated epilepsy who responded, 17 (19%) subjects with epilepsy were not in 5-year remission, and 74 (81%) subjects were in remission. Of those in remission, 61 (67%) were in remission off medication, and the mean age of cessation of regular AED therapy was 19 years (median, 14.5 years). Thirteen (14%) subjects were in remission but were still taking AEDs. Among the patients with active epilepsy, one patient had daily seizures, three patients had seizures once a month, seven patients had had only one seizure, and five patients had no seizures at all during the past year. One patient with active epilepsy did not report the seizure frequency.
General QOL measures
The results of the responses of subjects and controls on each of the eight domains of the SF-36 are summarized in Table 1. No significant differences were found between subjects and controls on any of the eight subscales or in the total SF-36 scores. If we examine the scores within subgroups of subjects, only two statistically significant differences appear between the index patients and controls. The 61 subjects in remission without medication reported significantly more physical pain than did controls (p < 0.05). In addition, the 13 subjects in remission but still taking regular AED therapy reported more limitations in usual role activities because of emotional problems. This difference was highly significant (p < 0.005) despite the small sample size of this group.
Table 1. Health-related quality of life as measured by the SF-36 in Finnish childhood epilepsy cohort versus population controls matched for age, sex, and place of birth
|Controls||90||79.2||15.4|| ||92.1||14.4|| ||91.2||23.1|| ||75.5||21.3|| ||74.8||20.4|| |
|Subjects in remissiona||74||82.1||15.5||0.24||91.8||17.0||0.91||88.1||27.0||0.44||82.0||22.6||0.06||77.3||20.7||0.44|
|In remission off medsa||61||82.5||15.7||0.22||93.1||15.3||0.69||88.5||27.8||0.52||82.9||20.7||0.04||79.5||18.9||0.16|
|In remission on meds||13||80.7||14.9||0.75||86.2||22.7||0.38||86.5||24.2||0.50||78.2||30.2||0.70||67.9||25.8||0.28|
|Not in remission||17||72.6||22.1||0.25||85.3||23.0||0.25||89.1||24.1||0.73||74.8||24.4||0.90||68.0||29.3||0.37|
|Controls||90||73.8||17.5|| ||85.2||30.8|| ||82.8||15.5|| ||60.3||19.3|| |
|All subjectsa||91||74.6||20.0||0.76||86.1||30.7||0.84||82.9||14.4||0.97||64.0||20.4||0.22|| |
|Subjects in remissiona||74||77.6||16.2||0.17||89.6||27.3||0.36||85.2||9.4||0.22||66.1||18.6||0.06|| |
|In remission off medsa||61||77.0||17.3||0.28||87.7||29.9||0.64||85.5||9.7||0.20||66.6||18.6||0.06|| |
|In remission on meds||13||79.9||10.7||0.22||97.4||9.2||0.004||84.2||8.3||0.61||64.2||19.3||0.50|| |
|Not in remission||17||62.8||28.6||0.14||72.5||39.5||0.14||73.4||24.4||0.14||55.6||25.5||0.38|| |
Within the patient subgroups, the number of differences also were few. The patients in remission regardless of medication status reported better scores on the emotional well-being subscale role limitations than did patients with active epilepsy (p < 0.02). The limitations in social activities because of physical or emotional problems were significantly more marked in patients with active epilepsy compared with patients in remission off medication (p < 0.008), but no significance was found when compared with patients in remission on medication.
Impact of Epilepsy Scale
The results of the Impact of Epilepsy Scale were available for 69 (76%) subjects and are summarized in Table 2. Not surprisingly, subjects in remission had lower scores (less impact) than did those not in remission (p = 0.004). However, the differences were limited to those in remission off medications. The subgroup of subjects in remission off medication had the lowest scores and did significantly better than either the group not in remission (p = 0.006) or those in remission but still taking medications (p = 0.03). No differences were noted in the scores between those in remission on medication (mean, 14.2) and those not in remission (mean, 14.1).
Table 2. Impact of Epilepsy Scale (IMPACT8) in Finnish childhood-onset epilepsy cohort
|Subjects in remission||54|| 9.9||4.4|
| In remission on/off medication||42|| 8.7||2.8|
| In remission off medication||12||14.2||6.2|
|Subjects not in remission||15||14.1||6.3|
We examined the correlation between the generic SF-36 and the epilepsy-specific Impact of Epilepsy Scale in this population. A strong correlation existed between the two measures, with a Pearson's Correlation Coefficient of 0.62 (p < 0.001)
Selected social outcomes
Despite the relatively minor differences between groups in the HRQOL measures such as the SF-36 reported by the subjects and controls, major differences were found in measures of social outcome. These are summarized in Table 3. The most striking differences is in employment. The group in remission off medication was only slightly less likely than controls to be employed, and the difference was not statistically significant. However, few subjects taking medications, whether in remission or not, were employed, and despite the relatively small samples sizes in these groups, the differences were highly significant. The differences in employment between subjects in remission off medications and both those in remission on medication (RR, 1.5; 95% CI, 1.0–2.4; p < 0.003) and those not in remission (RR, 1.6; 95% CI, 1.1–2.4; p < 0.001) also was highly significant. For SES, the group in remission off medications was not different from the control group. However, both those in remission on medications and those not in remission were more likely to have a lower SES than were controls. Note that these differences in employment and SES are not attributable to differences in educational status. We have previously reported that the groups with uncomplicated epilepsy were more likely than controls to have only a primary education, although the proportion with a matriculation examination and with vocational training was similar (17). As shown in Table 3, no significant differences were found between the subgroups with uncomplicated epilepsy in educational achievement, and thus this is not the reason for the differences in employment. If one further examines the groups in remission off medication, differences are seen between those who were employed and those not employed.
Table 3. Social and educational outcomes of Finnish childhood epilepsy cohort versus population controls matched for age, sex, and place of birth
|Number of cases||90||91|| ||61|| ||13|| ||17|| |
|Primary education or less||20 (22)||46 (52)||2.3||1.5–3.6||<0.001||28 (47)||2.1||1.3–3.4||0.001|| 8 (38)||2.8||1.5–4.9||0.003||10 (58)||2.6||1.5–4.6||0.002|
|No matriculation exam||67 (74)||71 (80)||1.1||0.9–1.3||0.40||46 (78)||1.0||0.9–1.2||0.62||11 (85)||1.1||0.91.5||0.42||14 (82)||1.1||0.9–1.4||0.49|
|No vocational training||45 (49)||60 (67)||1.3||1.0–1.7||0.02||39 (66)||1.3||1.0–1.7||0.05||11 (85)||1.7||1.2–2.3||0.02||10 (59)||1.2||0.7–1.8||0.50|
|Not married or cohabiting||13 (15)||38 (43)||2.9||1.7–5.1||<0.001||21 (36)||2.4||1.3–4.4||0.003|| 8 (62)||4.2||2.1–8.1||<0.001|| 9 (53)||3.6||1.9–7.1||<0.001|
|No children||13 (15)||41 (47)||3.2||1.9–5.6||<0.001||21 (36)||2.5||1.4–4.6||0.002|| 7 (54)||3.7||1.8–7.6||0.003||13 (76)||5.3||3.0–9.3||<0.001|
|No driver's license||10 (11)||33 (38)||3.4||1.8–6.5||<0.001||12 (21)||1.9||0.9–4.1||0.10|| 8 (62)||5.5||2.7–11||<0.001||13 (76)||6.9||3.6–9.3||<0.001|
|Not employed|| 2 (2)||15 (17)||7.5||1.8–32||<0.001|| 3 (5)||2.2||0.4–13||0.35|| 5 (38)||17||3.7–80||<0.001|| 7 (41)||19||4.2–82||<0.001|
|Lower SES||46 (51)||58 (64)||1.3||0.9–1.6||0.07||34 (57)||1.1||0.8–1.5||0.50||11 (85)||1.6||1.2–2.2||0.02||13 (76)||1.5||1.1–2.1||0.05|
This cohort whose mean age was in the late thirties is already at the age at which one can reliably assess difference in the rates of marriage and having children. Interestingly, the differences between subjects and controls are significant and present even in those in remission off medications who, on average, have been off medication since age 19 years. No significant differences were seen on these two items between subjects in remission off medication and either those in remission on medication (p = 0.084) or those not in remission (p = 0.20) in the rates of marriage. In terms of the rates of having children, no significant differences appeared between subjects in remission off or on medication in the rates of having children (p = 0.24). However, subjects not in remission had significantly lower rates of having children than did subjects in remission off medications (p = 0.003).
Both subjects in remission on medication and those not in remission are far less likely to hold a driver's license than are controls or those in remission off medication. Surprisingly, the impact of epilepsy on driving is similar in those in remission on medication and those not in remission. Even the group in remission off medication is somewhat less likely to hold a driver's license even many years later, but these differences are not statistically significant.
That epilepsy has an adverse impact on HRQOL and on disease-specific measures of QOL is well established (9–17,26–28). However, the focus of most studies has been on studying those with active epilepsy. This study is unique in that it examined the long-term impact of childhood-onset epilepsy in a population-based cohort including those in remission off medications for many years. Focusing on those with uncomplicated epilepsy is important, as in cases with epilepsy and other comorbidity, it is difficult to separate the impact of the epilepsy from that of the comorbid disorders.
The study was designed to gain a broad understanding of the long-term (>30 years) consequences of childhood-onset epilepsy. The original study population represented the whole child population in a geographically defined area. The matched healthy controls were a random sample from the same area. The method used for case identification had a very high sensitivity and specificity (17). The participation rate has, in all follow-up studies, been ∼90%. Thus this study is examining an unselected population-based sample. The relatively small numbers of subjects not in remission reflect the favorable prognosis of childhood-onset epilepsy in neurologically intact children (17,19,24,29–31).
Three major findings emerged in this study. The impact of childhood-onset epilepsy on long-term psychosocial function is substantial, even in those who are seizure free off medication for many years. Those most affected are those not in remission or in remission but those still on medication. Strikingly, little difference was noted in this population between those not in remission and those in remission but still on medication. In contrast, the effects of childhood-onset epilepsy on HRQOL, as reported by the subjects, are generally modest and are primarily seen in those not in remission or in remission but still taking medications.
Health-related quality of life
Although some differences were found between subjects and controls, they were relatively minor. In the group in remission off medication, the only significant or nearly significant differences were in the reports of bodily pain (p = 0.04) and energy/fatigue (p = 0.06). This may be consistent with our prior report that, compared with controls, subjects with childhood-onset epilepsy had a higher rate of psychiatric and psychosomatic disorders, whereas the rate of somatic disorders was similar (18). However, given the number of analysis performed, these findings should be considered of borderline significance and need replication in another population. In the group in remission but still on medication, the major reported impact was an increased level of role limitation due to emotional factors (p = 0.004). These subjects, even though seizure free, felt significant limitations. Interestingly this group also reported an impact of epilepsy scores that were as high as reported by those with epilepsy not in remission.
The most striking finding is the relative lack of effect of epilepsy on HRQOL, as measured by the SF-36, despite its clear impact on aspects of psychosocial function such as employment, marriage, etc. Generic scales provided a useful tool for clinical trials and medical practice, and summed up scores that mostly measure objective functioning. They also can be applied for comparisons with healthy controls. Wagner et al. (27) demonstrated that, although not disease specific, the SF-36 has proved at least as sensitive as the specific epilepsy scales. They studied 31 newly developed and validated generic and epilepsy-specific scales to evaluate their usefulness in the assessment of impact of epilepsy and AEDs on HRQOL. Both generic and epilepsy-specific scales satisfied, with few exceptions, psychometric tests of hypothesized item groupings and differentiated well between patient groups differing in time since the last seizure. Generic measures, especially measures of social and role functioning and mental health, were best at discriminating among groups differing in disease severity. However, in a clinical setting, the SF-36 data shown to the patients had little impact on either their perceptions about their epilepsy or patient management (27).
Impact of Epilepsy Scale
A major finding in this cohort is that the impact of epilepsy, as measured by the self-report Impact of Epilepsy Scale or by outcome measures such as employment and driving, was not related to whether the subjects were in remission or having seizures but rather to whether they were taking long-term AED therapy. This is discussed in more detail in the discussion of social outcomes, where the results are similar. Although one would expect that the concerns of those in remission, especially given the stringent 5-year seizure-free criterion used, would be different from those who are not in remission, needing to take daily medications appears to have a very substantial adverse impact.
In the data from the Medical Research Council randomized study of AED withdrawal (32), differences also were found in the perception of subjects who were off AEDs versus those who were in remission on AEDs (33). Those on AEDs, even if in remission, were more likely to feel stigmatized by their epilepsy, more likely to feel that epilepsy limited their social activities, and more likely to feel that epilepsy affected employment than were those in remission off AEDs (33). In that study, some of the adverse impact on perceived QOL was related to self-reported side effects (34).
Having childhood-onset epilepsy has long-term adverse consequences on employment, having a driver's license, and the prospect of either marrying or having children. What is surprising is the breakdown by remission status. In 1997, the risk of being unemployed in subjects in remission off medications in this cohort was not significantly higher than that in controls. However, in those still taking medications, the risk of unemployment was extremely high and was the same whether or not the subject was in remission, in terms of seizures. This implies that, even in patients with uncomplicated epilepsy, seizures per se may not be the dominant reason that patients with epilepsy have difficulty finding employment (35–37). The educational outcomes are shown to demonstrate that whereas differences exist between subjects and controls (17), no substantial differences are noted between the different subgroups of patients with uncomplicated epilepsy, and that the differences in employment status are therefore not attributable to differences in education. The Medical Research Council study, although finding a difference in perceived impact on employment, did not find an actual difference in employment between those in remission on AEDs and off AEDs (32,33). However, in that study, the subjects off AEDs had been off AEDs for only a few years, whereas in the current study, those off AEDs had typically been off AEDs since adolescence. This may have accounted for the much greater differences between the two groups in the current study.
The adverse impact of being on medications on driving is even more surprising. Subjects in remission off medications had a slightly but not statistically significant higher rate of not having a driver's license. In contrast, both those in remission on medication and those not in remission had similar and markedly increased rates of not having a driver's license. This is surprising, as in Finland, as in most Western countries, patients with epilepsy who are seizure free are eligible to obtain a driver's license. Note that we addressed the issue of having a driver's license. That some patients with epilepsy who should not be driving, do so, is well established (38), but not the subject of this study.
The adverse impact of childhood-onset epilepsy on future marriage or having children as an adult was previously reported based on the earlier 1992 outcomes (17) and persists in this analysis. Interestingly, here the effect of remission or medication status is more modest. An increased rate of not being married or cohabiting and of not having children was found in all cases, but the differences between those in remission off medications and those on medications, whether in remission or not, are far more modest than the differences between the subgroups in terms of employment and driving. A lower rate of being married was reported in adults with active epilepsy attending a seizure clinic (35). Decreased fertility and lower marriage rates have been described in a survey of men and women with active epilepsy, particularly in those with childhood onset (39). As was the case with employment, the differences cannot simply be explained based on seizure status or educational attainments.
Childhood-onset epilepsy has a long-term adverse impact into adulthood. The major impact is in the groups that remain on medications regardless of whether or not they are in remission. Encouragingly, it appears that those in remission off medications, although having a higher rate of educational problems and lower marriage and fertility rates than population-based controls, perform similar to the general population in terms of employment, holding a driver's license, and SES. It is unclear whether the differences between those in remission off medication versus those in remission on medication reflect the adverse impact on QOL of being on medication (40) or the different epilepsy syndromes in the two groups (24). However, the fact that the group in remission on medications seems to have a QOL very similar to that of the group not in remission argues for at least some role for the adverse impact of being on long-term medications. Alternatively, even if medications do not have a direct adverse impact, being on medications, even if seizure free, may be a marker for a variety of other adverse effects, including stigma. Further studies are needed on the effects of being on medication on QOL in patients with epilepsy who are seizure free.
Acknowledgment: This study was supported in part by grants from the Finnish Epilepsy Research Foundation and TS Group Printing House (M.S.) and grant NS26151 from National Institute of Neurological Disorders and Stroke (S.S.).