Genetic attribution and perceived impact of epilepsy in multiplex epilepsy families

Abstract

Objective:

Studies have found that affected individuals who believe the cause of their disorder is genetic may react in various ways, including optimism for improved treatments and pessimism due to perceived permanence of the condition. This study assessed the psychosocial impact of genetic attribution among people with epilepsy.

Methods:

Study participants were 165 persons with epilepsy from multiplex epilepsy families who completed a self-administered survey. Psychosocial impact of epilepsy was assessed with the Impact of Epilepsy Scale, containing items about relationships, employment, overall health, self-esteem, and standard of living. Genetic attribution was assessed using a scale derived from three items asking about the role of genetics in causing epilepsy in the family, the chance of having an epilepsy-related mutation, and the influence of genetics in causing the participant’s epilepsy. We estimated prevalence ratios (PRs) for impact of epilepsy above the median using Poisson regression with robust standard errors, adjusting for number of lifetime seizures and time since last seizure.

Results:

Participants’ age averaged 51 years; 87% were non-Hispanic white, 63% were women, and 54% were college graduates. The genetic attribution scale was significantly associated with having a high impact of epilepsy (adjusted PR = 1.4, 95% confidence interval = 1.07–1.91, P = .02). One of the three genetic attribution questions was also significantly associated with a high impact of epilepsy (belief that genetics had a big role in causing epilepsy in the family, adjusted PR = 1.8).

Significance:

These findings reflect an association between the psychosocial impact of epilepsy and the belief that epilepsy has a genetic cause, among people with epilepsy in families containing multiple affected individuals. This association could arise either because belief in a genetic cause leads to increased psychosocial impacts, or because a greater psychosocial impact of epilepsy leads some to believe their epilepsy is genetic.

1 |. INTRODUCTION

Use of clinical genetic testing has increased dramatically in the epilepsies due to rapid progress in genetic research, and this trend is expected to continue as more genes are identified.^1‒3 The increase in clinical use of genomic information raises the need to assess the potential consequences, for people with epilepsy, of conceptualizations of their disorder as genetic. Research in other disorders has assessed the psychosocial impacts of genetic causal attribution,^4‒6 that is, the belief that the cause of a disorder is genetic (regardless of the actual genetic contribution to the disorder); however, little is known about these impacts among people with epilepsy.

Evidence has shown that believing a disorder is caused by genetics might produce a range of reactions among affected persons, including optimism for improved prevention and treatments^7,8 and alternatively, a heightened perception of disease severity and intractability, denoted “prognostic pessimism.”⁹ Similarly, surveys of the general population have found that genetic attributions of mental illnesses lead to increases in their perceived severity.¹⁰

The impacts of epilepsy on those affected extends beyond seizures, medications, and side effects, and includes self-image, relationships with others, felt stigma or discrimination, fear of having seizures, and mental health issues such as anxiety and depression.^11‒14 These effects are likely worse when seizures have occurred recently or frequently, but evidence shows that the impacts can linger even when seizures are controlled.¹⁵

To investigate the relations of genetic attribution to perceived psychosocial impact of epilepsy, we analyzed data from a study of families with multiple individuals with epilepsy who had previously participated in genetic research. The current analysis is part of a larger exploration of the psychosocial impact of genetic attribution and receipt of actual genetic test results in families with multiple individuals with epilepsy. To date, results indicate that interest in genetic testing for oneself or one’s offspring is strong, especially if genetic findings are likely to have clinical utility.^16,17 Also, felt stigma was higher in people with epilepsy who attributed their disorder to a genetic cause than in others.¹⁸ Additionally, among family members without epilepsy, the perceived chance of having an epilepsy-related mutation was associated with increased prevalence of depressive symptoms.¹⁹

Based on previous findings, we hypothesized that the psychosocial impact of epilepsy would be higher among people who attributed their epilepsy to a genetic cause than among others. To test this hypothesis, we examined the relations of responses on the Impact of Epilepsy Scale (IES),¹² a previously validated measure of the impact of epilepsy, with measures of genetic causal attribution of epilepsy. Given our past findings on stigma, we also aimed to explore whether felt stigma and perceived impact of epilepsy act independently in individuals with epilepsy.

2 |. MATERIALS AND METHODS

2.1 |. Study sample

This report presents a cross-sectional analysis of the base-line phase of a larger investigation, initiated in 2012 with follow-up still in progress, on the psychosocial impacts of genetic conceptualizations of epilepsy and receipt of actual genetic test results among members of families containing multiple individuals with epilepsy. The sample was drawn from individuals who previously participated in the Epilepsy Family Study of Columbia University (EFSCU), a long-term investigation of genetics and epilepsy that began in the mid-1980s as a familial aggregation study and developed into a genetic linkage study.²⁰ Families were eligible for the EFSCU linkage study if they included a sibling pair or three or more individuals with nonacquired epilepsy, with onset before the age of 25 years. Families were recruited from 1993 to 2006 from all over the United States, using physician referrals and self-referrals in response to advertisements through the Epilepsy Foundation and a study website. Recruitment involved an initial eligibility screen with an index person from each family. Once a family was deemed eligible, we attempted to interview each family member (or parent, for children) using a standardized screen for possible seizure occurrence. Those who screened positive were given a thorough diagnostic evaluation including a diagnostic interview, review of medical records, and in some cases, an electroencephalogram and neurologic examination. Epilepsy was defined as a lifetime history of two or more unprovoked seizures (excluding febrile seizures). More details on the diagnostic and recruitment procedures have been described previously for families recruited from 1997 to 2002 (EFSCU Phase 2);²¹ data collection methods were the same for families recruited before and after those dates.

Among 1274 previous EFSCU participants (ie, individuals who were interviewed or donated a blood sample), eligibility for the current study was defined by current age of 18–79 years, ability to complete a self-administered questionnaire in English, and willingness to be contacted for future studies. From the total pool, 342 persons were excluded because they did not meet these criteria, leaving 932 eligible persons (see Figure S1). These individuals were asked to complete a self-administered, 30-minute questionnaire either online through Survey Monkey (www.surveymonkey.com) or on paper. The current analysis was restricted to individuals who themselves have epilepsy. All protocols were approved by the institutional review board of the Columbia University Medical Center.

2.2 |. Determination of epilepsy status

For the current study, we classified participants as having epilepsy based on self-reported data from the questionnaire rather than clinical data from the original EFSCU study. We made this decision for two reasons. First, participants’ perception of their own status was most relevant for the questions addressed here. Second, because of the long-time interval since our previous contacts with these participants (up to 20 years), some could have had onset of epilepsy after we last had contact with them. Participants were classified as having epilepsy if they responded positively to one or both of two survey questions. The first question asked, “Which of your biological relatives have had epilepsy or a seizure disorder?” with a list of relatives, including “yourself.” The second question was, “Have you ever been told that you had epilepsy or a seizure disorder?” with possible answers “yes,” “no,” and “don’t know.”

2.3 |. Impact of epilepsy scale

To assess our main outcome, perceived psychosocial impact of epilepsy, we used the IES.¹² This is a validated 10-item scale that includes questions pertaining to relationships, employment, overall health, self-esteem, and standard of living. For each item, respondents are asked how much each aspect of life is impacted by the experience of having epilepsy, with values from 1 = “not at all” to 4 = “a lot.” An individual’s score was taken to be the mean response of all 10 items. The scale’s developers previously reported that scale values were associated with measures of self-esteem, life fulfillment, perceived quality of life, frequency of minor seizures, and marital status, supporting its construct validity.¹² The scale had high internal consistency in our sample (Cronbach’s alpha = .93). Because IES scores were substantially right-skewed (mean = 1.7, median = 1.5), for analysis we dichotomized scores at the median to create a binary outcome variable coded as 0 = low IES (score < 1.5) and 1 = moderate to high IES (score = 1.5–4). For a sensitivity analysis, we performed analyses with other specifications of the outcome: continuous and untrans-formed, natural log transformed, square root transformed, and cube root transformed.

2.4 |. Genetic causal attribution

Our main predictor of interest, genetic causal attribution, was defined based on three questions.¹⁹ The first asked, “In your opinion, how big a role has genetics had in causing the epilepsy in your family?” with possible responses “none,” “small,” “medium,” or “big.” The second asked, “In your opinion, what do you think the chances are that you have a change or mutation in a gene that affects risk for epilepsy?” with possible responses: “none,” “small,” “moderate,” “high,” or “don’t know.” The third question asked: “How much do you think each of the following influenced your risk of developing epilepsy? (Please answer what you think caused your epilepsy in the first place, rather than what you think might trigger your seizures.),” followed by 14 items including “genetics or inheritance” in the list. Possible responses were “no influence,” “some influence,” and “strong influence.” Responses to each of the three genetic attribution questions were recoded into values from 1 to 3 (1 = no/small/none, 2 = moderate/medium/some, 3 = high/strong/big). As we have reported previously,¹⁹ responses to these questions indicating high levels of genetic attribution were significantly associated with higher number of affected family members, supporting their concurrent validity, and Cronbach’s alpha for the three questions was .77 in participants with epilepsy, reflecting good internal consistency. We also created a genetic attribution scale variable defined as the mean of an individual’s responses to the three genetic attribution questions. This summary scale value was generated even if one or two of the genetic attribution questions had missing data. We used Student’s t test to examine the relationship between the genetic attribution scale and the total number of affected family members reported.

2.5 |. Stigma

We used the previously validated, 10-item Epilepsy Stigma Scale (ESS) to measure felt stigma in our sample of individuals with epilepsy.²² An individual’s score was calculated as the mean of all 10 items. We explored the relationship between perceived impact of epilepsy and felt stigma using analysis of variance (ANOVA) and Spearman correlations. For ANOVA, we dichotomized responses to each item of the IES and used continuous ESS scores. We maintained both the IES and ESS measures as continuous variables to examine the correlation between the two.

2.6 |. Potential confounders

We considered several potential confounders of the relationship between genetic attribution and impact of epilepsy based on previous literature and theory, including age (<40, 40–59, and ≥60 years, as well as continuous age), sex, education (college graduates vs others), employment (employed, unemployed/retired), and number of family members with epilepsy (<4 vs ≥4). Two measures of epilepsy “severity” were also considered potential confounders: time since last seizure (<5 vs ≥5 years) and lifetime number of seizures (≤20, 21–100, >100).

2.7 |. Statistical analysis

To evaluate potential confounding, we carried out bivariate analyses to assess (1) the relationship of each potential confounder to the outcome variable (binary IES) and (2) the relationship of each potential confounder to the predictor (genetic attribution, measured as both individual categorical variables and the scale). Covariates associated with both the exposure and outcome using an alpha level of .2 were considered for inclusion in the final model if they were not hypothesized to be on the causal pathway between exposure and outcome. Only the two epilepsy severity measures (time since last seizure and total lifetime number of seizures) met our criteria to be potential confounders and were included in the adjusted model. We also ran a supplemental adjusted model that additionally included an individual’s mean score on the ESS to assess whether the results for the IES reflected a similar construct as felt stigma, as reported previously.¹⁸

To assess the associations between genetic attribution and impact of epilepsy, we used generalized linear models with robust Poisson regression to compute prevalence ratios (PRs) evaluating the relative likelihood of having a “high” score on the IES. Poisson models were used to predict PRs rather than odds ratios, given that our outcome was common.²³ We used generalized estimating equations to account for nonindependence resulting from the inclusion of multiple members per family. Statistical analyses were carried out using SPSS Statistics for Windows 22.0 (IBM) and SAS 9.4 (SAS Institute).

3 |. RESULTS

Among the 932 eligible individuals, 703 (75%) were reached by telephone to invite participation, and of those, 431 completed the survey for the study, for a final participation rate of 46% (see Figure S1). Participation rates were significantly higher among participants originally classified as having epilepsy (176/331, 53%) than among their biological relatives without epilepsy (198/443, 45%) or family members related through marriage (57/158, 36%; P = .001). Among people originally classified as having epilepsy, those who participated in the current study were significantly older (average = 52 years vs 44 years, P < .001) and more likely to be women (64% vs 53%, P = .04) than were nonparticipants. Participation rates were unrelated to education among persons originally classified as having epilepsy.

Agreement was very strong between our original diagnoses and self-reported epilepsy (based on the questions described above; kappa = 0.86, reflecting “almost perfect” agreement according to Landis and Koch).²⁴ Among 186 participants who self-reported epilepsy in the current survey, 18 had not previously been classified as having epilepsy. Of these, seven had new onset of epilepsy since our previous contact with them and the remaining 11 had histories of febrile seizures or other events that led them to perceive they had epilepsy, and thus to self-report epilepsy in the current study. Conversely, eight participants previously classified as having epilepsy did not self-report epilepsy and therefore were excluded from our current analyses. Review of our previous information for these eight participants revealed that diagnoses in the original study had been based on limited information and may have been incorrect.

After excluding participants with missing information for the IES scale or genetic attribution scale (n = 21), the final sample for the current study comprised 165 participants who self-reported epilepsy, from 89 families. These participants averaged 51 years of age (range = 18–79, SD = 14.5), and were 87% white non-Hispanic, 63% women, and 54% college graduates.

The overall mean IES score was 1.7 (SD = 0.79) on a scale of 1–4, with higher scores indicating greater perceived impact (Table 1). The item with the highest mean was “Your health overall” (2.0, Table 1); the item with the greatest number of individuals who responded, “a lot” was “The kind of paid work you can do,” whereas that with the least was “Your relationship with friends.”

TABLE 1.

Distribution of responses to Impact of Epilepsy Scale among individuals with epilepsy

How much do you think a particular aspect of your life is affected by your epilepsy and its treatment?
	Total n	Not at all n (%)	A little n (%)	Somewhat n (%)	A lot n (%)	Mean (SD)
Overall mean	165					1.7 (0.79)
Your relationship with your spouse/partner	146	87 (59.6)	27 (18.5)	16 (11.0)	16 (11.0)	1.7 (1.04)
Your relationship with other close members of your family	164	122 (74.4)	19 (11.6)	13 (7.9)	10 (6.1)	1.5 (0.88)
Your social life and social activities	164	81 (49.4)	44 (26.8)	19 (11.6)	20 (12.2)	1.9 (1.04)
Whether or not you are able to work in paid employment	157	110 (70.1)	18 (11.5)	12 (7.6)	17 (10.8)	1.6 (1.03)
The kind of paid work you can do	158	84 (53.2)	30 (19.0)	20 (12.7)	24 (15.2)	1.9 (1.12)
Your health overall	162	66 (40.7)	47 (29.0)	30 (18.5)	19 (11.7)	2.0 (1.03)
Your relationship with friends	161	114 (70.8)	24 (14.9)	18 (11.2)	5 (3.1)	1.5 (0.81)
The way you feel about yourself	166	76 (45.8)	51 (30.7)	19 (11.4)	20 (12.0)	1.9 (1.03)
Your plans and ambitions for the future	162	99 (61.1)	30 (18.5)	15 (9.3)	18 (11.1)	1.7 (1.03)
Your standard of living	162	109 (67.3)	21 (13.0)	17 (10.5)	15 (9.3)	1.6 (1.00)

Note: Cronbach’s α = .93.

Most of the demographic characteristics we investigated were not associated with IES score (Table 2). Notably, being unemployed was associated in bivariate analyses with having a higher IES score (PR = 1.4, 95% confidence interval [CI] = 1.06–1.84). This is expected, because being unable to work is likely to be affected by similar underlying mechanisms as self-esteem, ambitions for the future, and relationships, which are all measured on the IES. Increasing number of lifetime seizures and time since last seizure ≤ 5 years were both significantly associated with high IES score (Table 2). We previously showed that these seizure variables were also associated with genetic attribution,¹⁹ leading to their inclusion in our adjusted models as potential confounders.

TABLE 2.

Sample characteristics and demographic predictors of IES score

	N	% High IES	PR (95% CI)	P
Total	165	50.3
Age
<40 y	44	47.7	0.8 (0.57–1.15)	.24
40–59 y	70	45.7	0.8 (0.55–1.09)	.15
60+ y	51	58.8	1.0 (ref)	(ref)
Sex
Women	104	51.0	1.0 (0.76–1.42)	.82
Men	61	49.2	1.0 (ref)	(ref)
Education
College graduate	88	51.1	1.0 (0.76–1.37)	.88
< College graduate	76	50.0	1.0 (ref)	(ref)
Employment
Unemployed	47	63.8	1.4 (1.06–1.84)	.02
Employed/retired	116	45.7	1.0 (ref)	(ref)
Total number of relatives with epilepsy
4 or more	77	48.8	1.1 (0.80–1.41)	.66
<4	80	52.0	1.0 (ref)	(ref)
Total number of lifetime seizures
>100	33	81.8	2.4 (1.70–3.29)	<.001
21–100	32	56.3	1.6 (1.02–2.59)	.04
20	75	34.7	1.0 (ref)	(ref)
Time since last seizure
≤5 y	81	71.6	2.3 (1.54–3.41)	<.001
>5 y	64	31.3	1.0 (ref)	(ref)

Note: Total numbers vary based on missing data.

Abbreviations: CI, confidence interval; IES, Impact of Epilepsy Scale; PR, prevalence ratio, generated from bivariate Poisson regression models with robust standard errors; ref, referent.

Values on the genetic attribution scale averaged 2.4 on a scale of 1–3, reflecting moderate to strong beliefs among participants that their epilepsy was due to genetics. Overall, the scale was calculated using all three questions for 109 participants (66.1%), two questions for 50 participants (30.3%), and only one question for six participants (3.6%). The proportion of participants with missing data was considerably greater for one of the questions, “In your opinion, what do you think the chances are that you have a change or mutation in a gene that affects risk for epilepsy?” (31.5% “don’t know” responses) than for the other two (1.8% and 4.2%). However, average scale values were the same for participants with versus without missing data for this question (P = .61, t test). As expected, participants with four or more affected family members reported significantly higher levels of genetic attribution than those with fewer than four affected relatives (mean [SD]: ≥4 family members, 2.7 (0.44); <4 family members, 2.1 (0.68); P < .0001).

In unadjusted analyses, the genetic attribution scale variable was significantly associated with having a high IES score (Table 3). In addition, the associations were suggestive for two of the three individual questions (“high” vs “no/small” perceived chance of having an epilepsy-related mutation, and “big” vs “no/small” role of genetics in causing epilepsy in the family; Table 3). After adjustment for epilepsy severity measures, the relationship of genetic attribution to IES score persisted, and the relationship of perceived role of genetics in causing the epilepsy in the family to IES score became significant (PR = 1.8, 95% CI = 1.07–2.98). The PR for the genetic attribution scale was 1.4 (P = .02), suggesting that the likelihood of having a high IES score increased by 40% for each one-unit increase in the level of genetic attribution. This result was robust to different ways of specifying the outcome in our model (IES values on a continuous scale with no transformation, or natural log, cube root, or square root transformation), with significant prevalence ratios of similar magnitude.

TABLE 3.

Relationship between genetic causal attribution and IES score

			Unadjusted		Adjusted
Predictor	% High IES	N	PR (95% CI)	P	PR (95% CI)	P
Chance you have a change or mutation in a gene that affects risk for epilepsy
High chance	56.5	62	2.1 (0.98–4.68)	.05	2.3 (0.85–6.11)	.10
Moderate chance	53.1	32	2.0 (0.89–4.57)	.09	2.1 (0.77–5.60)	.15
No/small chance	26.3	19	1.0 (ref)	(ref)	1.0 (ref)	(ref)
Total n		113
Role of genetics in causing epilepsy in your family
Big role	55.1	89	1.6 (0.95–2.54)	.08	1.8 (1.07–2.98)	.03
Medium	50.0	42	1.4 (0.81–2.44)	.22	1.6 (0.92–2.90)	.09
No/small	35.5	31	1.0 (ref)	(ref)	1.0 (ref)	(ref)
Total n		162
Influence of genetics in causing your epilepsy
Strong influence	53.6	97	1.4 (0.74–2.58)	.32	1.0 (0.53–1.92)	.97
Some influence	51.2	43	1.3 (0.69–2.51)	.41	0.9 (0.46–1.71)	.73
No influence	38.9	18	1.0 (ref)	(ref)	1.0 (ref)	(ref)
Total n		158
Genetic attribution scale	n/a	165	1.4 (1.04–1.77)	.03	1.4 (1.07–1.91)	.01

Note: Outcome is relative to “low IES” (<1.5). Total numbers vary based on missing data. Adjusted models include time since last seizure (≤5 years, >5 years) and total lifetime number of seizures (≤20, 21–100, >100).

Abbreviations: CI, confidence interval; IES, Impact of Epilepsy Scale; PR, prevalence ratio; ref, referent.

Correlation between the ESS and IES scores was strong and significant (Spearman correlation coefficient = 0.64, P < .0001). Each item of the IES was significantly associated with felt stigma (P ≤ .006 for each item, Table S1). When our model was further adjusted for felt stigma, the association between genetic attribution scale and perceived impact of epilepsy was slightly attenuated (PR of 1.4 declined to 1.3).

4 |. DISCUSSION

This study addressed the relations of the lived experience and psychosocial impact of epilepsy to the perception that its cause is genetic. The relationship between the genetic attribution scale and perceived impact of epilepsy was strong, reflecting a 40% increase in the likelihood of high IES for each one-unit increase in genetic attribution score among individuals with epilepsy. We also observed a strong positive association between participants’ belief that they had an epilepsy-causing mutation and the perceived impact of epilepsy. Both the unadjusted and adjusted models yielded a prevalence ratio greater than twofold for “high” versus “no/small” chance of having a mutation, although results were not significant. A substantial proportion of participants responded, “don’t know” to this question and were therefore excluded from the analysis, leading to reduced sample size and power. Some of this loss may have resulted from poor understanding or negative connotations of the word “mutation.”²⁵ However, a benefit of using the composite genetic attribution scale was to avoid the limitations of having missing information in one or more of the genetic attribution questions.

The questions are similar on the ESS and the IES, which likely caused the correlation we detected between the two measures. For example, the ESS asked, “Because of my epilepsy, I have problems developing intimate relationships,” and the IES asked about “Your relationship with your spouse/partner.” We previously reported that levels of felt stigma were associated with a “big” perceived role of genetics in causing the epilepsy in the subject’s family.¹⁸ When our analysis was adjusted for ESS, the PR for high impact of epilepsy score in relation to genetic attribution declined slightly, from 1.4 to 1.3. Because the prevalence ratio is borderline significant even with adjustment for felt stigma, we believe this provides preliminary evidence for an association of genetic attribution with impact of epilepsy that is independent of felt stigma.

These results suggest that genetic causal attributions are interconnected with the everyday experience of having epilepsy, such as relationships, employment, quality of life, and self-esteem. Importantly, those who believed more strongly that the cause of their epilepsy was genetic reported experiencing a higher impact of epilepsy.

The evidence that genetic attribution plays a role in the lived experiences of individuals with epilepsy was demon-strated despite our sample including few individuals with severe epilepsy, as measured by time since last seizure and total lifetime seizures. Given the strong associations we observed between epilepsy severity measures and impact of epilepsy, we suspect that the low proportion of individuals with severe epilepsy contributed to the low mean IES score in the sample. Some studies using the same scale have reported percentages of respondents who chose “a lot” to be twice as high as we found for some of the items.^26‒29 A sample including a diverse range of epilepsy severity levels might have had more normally distributed IES scores, avoiding the need to dichotomize the outcome.

Total number of relatives with epilepsy was associated with our genetic attribution questions, providing validation, because individuals with many affected family members would be expected to hold stronger beliefs that the cause of their epilepsy is genetic.¹⁹ Similarly, one might expect that having many relatives with epilepsy could influence perceived impact of epilepsy in both affected and unaffected people through items such as “Your relationship with other close members of your family” and possibly “Your relationship with your spouse/partner,” but contrary to this expectation, number of affected relatives was not associated with the IES (Table 2). One explanation might be that the impact of epilepsy is influenced by the relationship of a person with epilepsy to other affected family members, rather than by the number of family members affected. Future analyses should explore this.

Because of the cross-sectional nature of these data, the association of genetic attribution with high IES score does not necessarily imply a causal effect of genetic attribution on psychosocial impact of epilepsy. It could arise either because (1) belief in a genetic cause leads to greater psychosocial impacts or (2) a greater psychosocial impact of epilepsy leads some persons to believe their epilepsy is genetic (eg, because of a belief that severe epilepsy is more likely to be genetic). We cannot distinguish between these explanations with our cross-sectional data, and both explanations are plausible. In either case, the association between beliefs about genetics and psychosocial impact of epilepsy are intertwined.

We also explored the possibility that the association of genetic attribution with psychosocial impact of epilepsy was explained by a small subset of participants with severe, genetic epilepsies associated with cognitive difficulties. These persons would be expected to have both high levels of genetic attribution and high psychosocial impacts of epilepsy related to low educational attainment or unemployment. The data did not support this explanation, however. The association of genetic attribution with IES was not restricted to a subgroup of persons with severe epilepsies, and was actually somewhat stronger in persons who were employed than in those who were unemployed, and in persons whose last seizure was >5 years ago than in those who had seizures in the past 5 years.

This study has several limitations in addition to our cross-sectional design. First, generalizability is limited because the sample is unrepresentative of all persons with epilepsy. Families with multiple individuals with epilepsy are uncommon, and the participants are likely to have higher genetic causal attribution than would generally be observed in people with epilepsy. These high levels of genetic attribution facilitated study of the relationships of interest in a way that might have been more difficult in a less selected sample. However, it would be important to see whether an analysis of a more representative sample would yield similar results.

The sample is also unrepresentative of all persons with epilepsy because of its ethnic homogeneity (87% white non-Hispanic), relatively high education (54% college graduates), and recruitment based on previous participation in genetic research. This recruitment approach means participants were un-usually knowledgeable about epilepsy genetics, and may help to explain their low numbers of lifetime seizures and long time interval since the last seizure, because people with fewer illness-related difficulties were probably more likely to participate.

We classified participants as having epilepsy based on self-report rather than medical diagnoses, because our primary interest was in the participants’ own perceptions of their epilepsy status. Nevertheless, self-reports agreed very well with our previous diagnoses of epilepsy (defined as re-current unprovoked seizures), which were based on detailed clinical information. The total number of lifetime seizures is an imperfect measure of epilepsy “severity,” because with this approach, some syndromes involving frequent seizures would automatically be classified as “severe” when they might not be experienced this way. Also, difficulties in recall may have led to errors in reporting of the total number of seizures; hence, to minimize misclassification, we used very broad categories for this variable (≤20, 21–100, >100).

The impact of epilepsy is complex and characterized by many factors other than clinical symptom management. As clinical use of genetic testing continues to grow, investigations into the relationships of genetic causal attribution with psychosocial impacts of epilepsy will be exceedingly important.

Key Points.

• In this sample of multiplex epilepsy families, individuals reported moderate to high genetic attribution of epilepsy

• Psychosocial impact of epilepsy was increased among people with epilepsy who had recent seizures or greater than 20 lifetime seizures

• People who attributed their epilepsy to a genetic cause were more likely than others to report a high psychosocial impact of epilepsy

• Psychosocial impact of epilepsy and felt stigma are highly correlated among individuals with epilepsy