Seizure: European Journal of Epilepsy 73 (2019) 1–8

Contents lists available at ScienceDirect

Seizure: European Journal of Epilepsy

journal homepage: www.elsevier.com/locate/seizure

Personality profiles differ between patients with epileptic seizures and T

patients with psychogenic non-epileptic seizures
Michelle Leonga, Albert D Wanga, David Trainorb, Ben Johnstonec, Genevieve Raynerd,e,f,g,
Tomas Kalincika,h, Izanne Roosa,h, Patrick Kwanc,f,g,h, Terence J O’Brienc,f,g,h,
⁎
Dennis Velakoulisb,i, Charles B Malpasa,e,f,g,h,
a
Clinical Outcomes Research Unit (CORe), Department of Medicine, Royal Melbourne Hospital, The University of Melbourne, Parkville, Victoria, 3010, Australia
b
Department of Psychiatry, Royal Melbourne Hospital, Parkville, Victoria, 3050, Australia
c
Department of Medicine, Royal Melbourne Hospital, The University of Melbourne, Parkville, Victoria, 3010, Australia
d
Department of Medicine, Austin Health, The University of Melbourne, Victoria, 3010, Australia
e
Melbourne School of Psychological Sciences, The University of Melbourne, Parkville, Victoria, 3010, Australia
f
Department of Neuroscience, Monash University, Clayton, Victoria, 3800, Australia
g
Department of Neurology, Alfred Health, Melbourne, Victoria, 3004, Australia
h
Department of Neurology, Royal Melbourne Hospital, Parkville, Victoria, 3010, Australia
i
Department of Psychiatry, University of Melbourne, Parkville, Victoria, 3010 Australia

A R T I C LE I N FO A B S T R A C T

Keywords: Purpose: Psychogenic non-epileptic seizures (PNES) and epileptic seizures (ES) are often difficult to differentiate,
Epileptic seizures leading to incorrect or delayed diagnosis. The aim of the study was to determine whether patients of these two
Psychogenic non-epileptic seizures diagnostic groups possess different personality profiles, and whether they could be used to efficiently screen for
Personality PNES in clinical settings.
Five factor model
Methods: Collection of data was conducted on 305 patients who completed the NEO-Five Factor Inventory
Big 5 personality traits
questionnaire during a Video EEG Monitoring admission to the Royal Melbourne Hospital between 2002–2017.
Personality differences were investigated using Bayesian linear mixed effects models, with receiver operating
characteristic curve analysis computed to evaluate diagnostic accuracy.
Results: The ‘openness to experience’ domain (BF10 = 21.55, d = −0.43 [95% CI −0.71, −0.17]) and the
‘aesthetic interest’ facet (B10 = 7.98, d = −0.39 [95% CI −0.66, −0.12]) were the only personality factors
demonstrating strong evidence for a group difference, with patients with PNES having higher scores compared to
the ES group. ES patients had lower scores on these measures compared to the normal population, while PNES
patients did not. Both openness to experience and aesthetic interest showed poor sensitivities (53%, 46% re-
spectively) and specificities (69%, 46% respectively) for classifying PNES and ES patients.
Conclusion: While openness and aesthetic interests differ greatly between PNES and ES groups, low sensitivity
and specificity suggests their use is limited in a clinical setting. Nevertheless, these findings open up new avenues
of research using modern personality models to further understand patients with epilepsy and related pre-
sentations.

Abbreviations: AED, anti-epileptic drugs; AUC, area under the curve; BPPV, benign paroxysmal positional vertigo; BFI, bear-Fedio Inventory; DAPP-BQ, dimensional
assessment of personality pathology- basic questionnaire; ES, epileptic seizures; Extra-TLE, extra-temporal lobe epilepsy; FFM, five-factor model; ILAE, international
league against epilepsy; Hs, hypochondriasis; Hy, hysteria; MMPI, minnesota multiphasic personality inventory; NEO-FFI, neo five-factor inventory; NES indicator,
non-epileptic seizure indicator; PAI, personality assessment inventory; PMD, psychogenic movement disorder; PNES, psychogenic non-epileptic seizures; PTSD, post-
traumatic stress disorder; SOM-C, somatic-conversion; TLE, temporal lobe epilepsy; RMH, royal Melbourne Hospital; ROC, receiver operating characteristic; VEM,
video-EEG monitoring
⁎
Corresponding author at: Clinical Outcomes Research Unit (CORe), Department of Medicine, Royal Melbourne Hospital, University of Melbourne, Melbourne,
3010 Australia.
E-mail address: [email protected] (C.B. Malpas).

https://doi.org/10.1016/j.seizure.2019.10.011
Received 24 August 2019; Received in revised form 10 October 2019; Accepted 14 October 2019
1059-1311/ © 2019 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.
M. Leong, et al. Seizure: European Journal of Epilepsy 73 (2019) 1–8

1. Introduction differences in a clinical setting, which has often been overlooked in

previous research. Following Cragar et al., we hypothesised that PNES
Psychogenic non-epileptic seizures (PNES) are episodes of abnormal patients would score higher on neuroticism, and lower on extraversion
limb movements that resemble epileptic seizures (ES), but are not as- compared to patients with ES.
sociated with organic aetiologies, and presumably manifest due to
psychological distress [1]. Similarities in presentation to ES mean that 2. Methods
patients with PNES are often misdiagnosed with epilepsy, resulting in
multiple financial, social, emotional and health implications [2–6]. 2.1. Participants
A notable area of research has been the differential psychological
profiles of PNES and ES patients. For example, a body of research has This was a retrospective case control study. Data were collected
demonstrated different rates of personality disorders in these two from patients as part of routine psychological care who were admitted
groups. There is evidence that PNES is more commonly associated with to the Royal Melbourne Hospital (RMH), Australia, VEM unit between
cluster B personality disorders (e.g. borderline, histrionic) [7–14], the years 2002-2017. The RMH is a major epilepsy centre in Australia
while ES is more often associated with cluster C disorders (e.g. avoi- and utilises the use of VEM and a multidisciplinary team involving
dant, obsessive-compulsive) [7,13,15]. It is possible, however, that epileptologists, psychiatrists, EEG scientists, and neuropsychologists in
some patients experience ‘subthreshold’ personality features that are diagnosing epilepsy. Inclusion criteria for the current study included
not severe enough to cause functional impairment and to be diagnosed patients who had completed the NEO Five-Factor Inventory (NEO-FFI)
as a personality disorder [13]. It is therefore possible that personality questionnaire during their VEM admission and those who had a 6th
assessment that includes coverage of the less pathological range of grade reading level of English and above. Exclusion criteria included
personality traits might reveal differential profiles between PNES and patients who were under the age of 17 and patients with 10 or more
ES patients. responses missing. Of the 344 patients who were initially recruited, 8
Along this line, multiple psychological questionnaires have been patients were underage and 31 had missing responses, leading to 305
utilised to investigate these differences with significant results. patients being finally included in the study. This study received ap-
Numerous studies using the Minnesota Multiphasic Personality proval from the Melbourne Health Ethics Committee (MH HREC#
Inventory (MMPI) highlighted an increase in hypochondriasis (Hs) and QA2012044). All data are available at http://doi.org/10.17605/OSF.
hysteria (Hy) scales in PNES groups [16–21]. The Personality Assess- IO/T3A9G.
ment Inventory (PAI) has also been used extensively in PNES and ES
research with the Somatic-Conversion (SOM-C) and Non-epileptic sei- 2.2. Clinical diagnosis
zure indicator (NES indicator) being used to differentiate PNES and ES
patients [22–24]. Additionally, the Dimensional Assessment of Per- Diagnosis of the patients admitted for VEM was determined at a
sonality Pathology- Basic Questionnaire (DAPP-BQ) has shown PNES large multidisciplinary meeting consisting of EEG scientists, epileptol-
patients to be elevated in emotional dysregulation in comparison to ogists, neuropsychologists, and psychiatrists. Neuropsychiatric assess-
epileptic and healthy controls [25]. While these measures assess some ments, neurological examinations, and imaging was also heavily con-
aspects of personality, they also focus on common symptoms of psy- sidered in making the final diagnosis. A diagnosis of Epilepsy was made
chiatric disorders (such as depression, or anxiety). For example, while when a patient met the International League Against Epilepsy (ILAE)
the MMPI does include the assessment of some personality character- criteria of epilepsy [31,32], while a diagnosis of PNES was made based
istics, the psychopathological variables make it hard to determine on abnormal movements in the absence of EEG changes [33]. This
personality correlations between patients with ES and PNES. Further- definition included ‘swoon’ attacks (when the patients sinks to the floor
more, the PAI’s clinical scales have been derived based on DSM criteria and becomes unresponsive) [34]. Patients were classified into 4 main
[26]. Relatively little research has used modern models of non-patho- categories: ES, PNES, ES + PNES and non-diagnostic. Patients were
logical personality traits such as the five-factor model (FFM). The FFM categorised as non-diagnostic when a definite diagnosis could not be
is a statistically-derived model that describes personality along five made or if they had a condition which did not meet either ES or PNES
domains, including neuroticism, extraversion, openness to experience, criteria. Conditions reported were migraines, postural hypotension,
agreeableness, and conscientiousness [27]. anxiety, cardiac syncope, arrhythmia, alcohol excess, post-traumatic
Few notable studies using the five-factor model in the differentia- stress disorder (PTSD), benign paroxysmal positional vertigo (BPPV)
tion of PNES and ES has been performed thus far. In the process of and sick sinus syndrome. For patients with ES, the type of epilepsy was
cluster analysing PNES using the Revised NEO Personality Inventory also classified according to the ILAE classification of epilepsies [35].
(NEO PI-R), Cragar et al. [28] compared the PNES patients to an epi- Specifically, the focal epilepsies, temporal lobe epilepsy (TLE) and
lepsy control group and found that PNES patients scored higher on extra-temporal lobe epilepsy (extra-TLE), and generalised epilepsy.
neuroticism and on the agreeableness facet of modesty and lower on the Laterality of seizure focus was also recorded for TLE patients. The NEO-
extraversion facet of gregariousness and on the agreeableness facet of FFI results were not known to the VEM team and did not inform the
trust compared to epilepsy controls. Although these findings are re- diagnosis of PNES.
levant, the study sizes were small and personality differences between
PNES and epilepsy were not the focus of the study, but rather a sec- 2.3. Personality assessment
ondary finding. A study investigating personality traits between pa-
tients with PNES and psychogenic movement disorder (PMD), also re- The NEO-FFI is a 60-item self-administered measure of personality
ported patients with PNES having greater levels of neuroticism that describe an individual’s degree of standing across five higher order
compared to healthy controls [29]. More recently, Vilyte et al. reported factors: neuroticism, extraversion, openness to experience, agreeable-
no differences in FFM traits between patients with PNES and ES in a ness, and conscientiousness [27]. These domains and their definitions
small sample of South African patients [30]. Lack of clinically focused, are included in Table A1 in the Supplementary material. The five fac-
high-quality research in this area shows the need for a more in-depth tors were identified via the factor analysis of all the trait adjectives
study using the modern personality models. found in English and other natural languages [36] and has received
This large retrospective study aimed to utilise the FFM model of widespread empirical support [27,37,38]. Construct validity and in-
personality to determine profile differences between patients with ternal consistency studies have shown the NEO-FFI to be a valid and
PNES and ES admitted to a specialist VEM unit in a large tertiary centre. reliable psychometric tool for the assessment of personality [27]. Fur-
It also aimed to directly investigate the classification accuracy of these thermore, the five-factor model has shown to have good cross-cultural

2
M. Leong, et al. Seizure: European Journal of Epilepsy 73 (2019) 1–8

replicability [39] and real-life applicability by demonstrating its ability 3.2. Comparison of PNES and ES patients to population normative data
to predict numerous important life outcomes [27]. The NEO-FFI scores
are calculated in the form of a normalised score, providing an indica- A Bayesian liner mixed effects model was computed to determine
tion as to where an individual lies on a spectrum compared to the whether there were differences between personality domains and facets
general population [27]. in the combined PNES and ES cohort. There was strong evidence for an
Facet scores were also computed for each of the 5 domains. While effect of domain (BFinc > 100). As shown in Fig. 1, the highest scores
domain scores allow for a global picture of an individual’s personality, were observed for neuroticism (M = 0.49, SD = 0.97), followed by
facets allow for a deeper understanding of the individual [27]. We openness (M = −0.13, SD = 1.10), agreeableness (M = -0.22,
computed scores for the 13 facets derived via an analysis performed by SD = 1.20), extraversion (M = −0.23, SD = 1.03), and con-
Saucier [40]. Table A2 in Supplementary material illustrates the do- scientiousness (M = −0.58, SD = 1.04). There was strong evidence
mains and its facets. that the levels of neuroticism were higher when compared to normal
population (BF10 ≥ 100). Levels of extraversion (BF10 = 10.84) and
consciousness (BF10 ≥100) were significantly lower than the normal
2.4. Statistical analysis
population.
All analyses were performed JASP [41] and R [42]. Bayesian linear
mixed effects models were used to determine differences in personality
3.3. Comparison of epilepsy and PNES groups
profiles between the PNES and ES groups. Separate analyses were
conducted for TLE and non-TLE groups, as well as laterality in TLE
A Bayesian linear mixed effects model was computed to determine
patients. For all analyses, the null model included gender and age as
differences in personality domains between diagnostic groups (PNES
predictors of personality and the alternative model contained these
and ES). While there was strong evidence for a main effect of domain
predictors plus the relevant personality domains and diagnostic group
(BFinc > 100), there was no evidence for a main effect of diagnostic
(e.g. PNES vs ES). The statistical importance of each term was de-
group (BFinc = 1.86). There was, however, evidence for diagnostic
termined using a model averaging approach. Specifically, the Bayes
group by domain interaction (BFinc = 9.92). This indicates that the two
factor of inclusion (BFinc) was computed for each term which represents
diagnostic groups had different personality profiles. Follow-up Bayesian
the sum of the posterior model probabilities for all models containing
independent samples T-tests were performed to investigate which spe-
the term divided by the sum of posterior probabilities for all models not
cific domains differed between groups (see Table 2 and Fig. 1). Open-
containing the term. Higher interactions are excluded. A BFinc > 3 was
ness to experience was the only domain found with a strong evidence
taken to indicate evidence for an effect, while BFinc < 1/3 was taken as
for a group difference (BF10 = 21.55, d = −0.43 [95% CI −0.71,
evidence for the null hypothesis. BFs between 1/3 and 3 were taken to
−0.17]). Patients in the PNES group reported greater openness to ex-
indicate insensitivity to either hypothesis. Follow-up Bayesian in-
perience compared to patients in the ES group. The null hypothesis was
dependent sample T-tests were computed to identify specific domain
supported for the agreeableness domain (BF10 = 0.21, d = 0.11 [95%
differences between the groups, with BFs interpreted as described
CI −0.16, 0.38]), while the evidence was insensitive for the domains of
above. Effect sizes were reported for all group comparisons in the form
neuroticism, extraversion and conscientiousness.
of Cohen’s [43] d, with 95% credible intervals. Receiver operating
A focused analysis was performed on the three facets of the open-
characteristic (ROC) curve were computed to generate sensitivities and
ness domain (see Table 2). There was evidence for group differences on
specificities of individual domains and facets in order to determine
the aesthetic interest facet (BF10 = 7.98, d = −0.39 [95% CI −0.66,
usefulness in a clinical setting.
−0.12]), with PNES patients reporting higher scores compared to ES
patients. The evidence for the facets of intellectual interests and un-
3. Results conventionality was insensitive. No evidence for a group difference was
observed for the remaining facets, and the null hypothesis was sup-
3.1. Sample characteristics ported for the facets of sociability, activity, nonantagonistic orientation,
prosocial orientation, goal striving, and dependability.
Full diagnosis and demographic variables are described in Table 1. To investigate this further, Bayesian single sample t-test were per-
Of the 305 patients who met eligibility criteria, 122 (40%) were diag- formed to compare each group to the general population in terms of
nosed with ES, 14 (5%) with both ES and PNES, and 90 (30%) with levels of openness to experience. Patients in the ES group reported
PNES. The remaining 79 (25%) patients were classified as either ‘non- lower openness to experience (BF10 = 94.07, d = −0.34 [95% CI
diagnostic’, or were diagnosed with an alternative, non-epileptic or −0.52, −0.160]) and aesthetic interests (BF10 > 100, d = −0.58
non-psychogenic cause of their presentation. Only the ES and PNES [95% CI −0.77, −0.39]). In contrast, patients in the PNES group had
patients were included in the main statistical analyses. Of the ES pa- scores on openness to experience (BF10 = 0.23, d = 0.12 [95% CI
tients, 23 (17%) had extra-temporal focal epilepsy, 74 (54%) had TLE, −0.09, 0.32]) and aesthetic interests (BF10 = 0.21, d = −0.11, [95%
31 (23%) had generalised epilepsy, and 8 (6%) had probable focal CI −0.32,0.09]) that were not different to the general population.
epilepsy of unspecified or unclear focus. Mean age was 38.79 (SD 15.33,
range 17–92) and the majority of participants were female (n = 202,
66%). There was no relationship between gender and VEM diagnosis 3.4. Comparison of all diagnostic groups
(BF10 = 0.02).
Psychiatric comorbidity was common with 113 (37%) patients A Bayesian linear mixed effects model was computed to determine
having a history of any major psychiatric disorder, and 12 (4%) patients differences in personality domains between all four diagnostic groups
having a history of borderline personality disorder. PNES participants (PNES, ES, ES + PNES and non-diagnostic). While there was strong
were more likely to have a history of depression (BF10 = 4.66), current evidence for a main effect of domain (BFinc > 100), the null hypothesis
depression (BF10 = 5.91), or any current major psychiatric disorder was supported for the main effect of diagnostic group (BFinc < .01). The
(BF10 = 16.95). There was no evidence for a relationship between di- null hypothesis was also supported for the domain by diagnostic group
agnostic category and history of any major psychiatric disorder interaction (BFinc = 0.02). Examination of the means for these groups
(BF10 = 1.52), history of anxiety disorder (BF10 = 2.23), current an- confirmed that the addition of the ES + PNES and non-diagnostic
xiety disorder (BF10 = 1.36), or history of borderline personality dis- groups diluted the statistical effects.
order (BF10 = 2.87).

3
M. Leong, et al. Seizure: European Journal of Epilepsy 73 (2019) 1–8

Table 1
Sample characteristics.
Diagnostic group – Mean (SD)

ES (n = 122) PNES (n = 90) ES & PNES (n = 14) Non-Diagnostic (n = 79)

Age, years 38.87 (14.73) 35.28 (12.70) 36.21 (11.91) 43.13 (18.37)
Females, N (%) 80 (65.57) 62 (68.89) 11 (78.57) 49 (62.03)
Males, N (%) 42 (34.43) 28 (31.11) 3 (21.43) 30 (37.97)

Domains
Neuroticism 0.39 (0.85) 0.63 (1.11) 0.88 (0.90) 0.22 (1.19)
Extraversion −0.13 (1.03) −0.35 (1.02) −0.26 (1.31) −0.19 (1.28)
Openness −0.34(0.98) 0.15 (1.19) −0.15 (0.94) −0.15 (1.03)
Agreeableness −0.16 (1.19) −0.30 (1.21) −0.43 (1.21) −0.15 (1.45)
Conscientiousness −0.47 (1.05) −0.73 (1.01) −1.11 (1.15) −0.56 (1.31)

Facets
Negative affect 0.35 (0.75) 0.57 (0.95) 0.66 (0.73) 0.25 (1.01)
Self-reproach 0.58 (1.01) 0.81 (1.27) 1.16 (1.06) 0.37 (1.33)
Positive affect 0.02 (0.92) −0.16 (1.02) 0.02 (0.93) −0.13 (1.01)
Sociability 0.26 (0.93) 0.14 (0.85) 0.12 (1.31) 0.16 (1.15)
Activity −0.20 (0.83) −0.35 (1.01) −0.33 (0.94) −0.07 (1.08)
Aesthetic interests −0.57 (0.96) −0.14 (1.17) −0.35 (0.90) −0.54 (1.00)
Intellectual interests −0.33 (0.84) −0.10 (0.90) −0.09 (0.70) −0.34 (0.95)
Unconventionality −0.38 (0.89) −0.06 (0.93) −0.16 (1.01) −0.01 (0.88)
Nonantagonistic orientation −0.41 (1.04) −0.57 (1.11) −0.52 (0.90) −0.57 (1.11)
Prosocial orientation −0.15 (1.32) −0.14 (1.14) −0.62 (1.61) −0.25 (1.33)
Orderliness −0.27 (1.00) −0.58 (0.97) −0.58 (1.06) −0.42 (1.10)
Goal striving −0.04 (0.93) −0.08 (0.99) −0.50 (1.12) 0.01 (1.02)
Dependability −0.55 (1.10) −0.72 (0.91) −1.33 (1.01) −0.60 (1.37)

Note: ES= Epileptic Seizures. PNES = Psychogenic non-epileptic seizures. Non-diagnostic = Patients for whom a diagnosis could not be determined, or for whom a
diagnosis of other non-epileptic events was given.

3.5. Comparison of TLE and non-TLE personality. Our primary finding was that patients with PNES and ES
differed in their personality profiles. Specifically, patients in the PNES
A Bayesian linear mixed effects model was computed to determine group exhibited higher scores on the openness to experience domain,
differences in personality domains between the ES patients with TLE which measures such personality facets as propensity for fantasy, aes-
and the non-TLE ES patients. There was strong evidence for a main thetics interests, openness to feelings, and acceptance of new ideas.
effect of domain (BFinc > 100). The null hypothesis was supported for Follow-up analyses showed that the strongest evidence was for the
the main effect of group (TLE vs non-TLE; BFinc = 0.11) and the group aesthetic interests facet. Patients with ES had lower scores on these
by domain interaction (BFinc = 0.10). variables compared to the general population, while PNES patients did
not differ from normative data. While these differences were large and
3.6. Comparison of left and right TLE patients supported by strong statistical evidence, the sensitivity and specificity
for the openness and aesthetic interests scores were poor, which sug-
Of the 70 TLE patients, 3 patients with bilateral TLE were excluded gests that their use as psychometric screening instruments for PNES
from the analysis, leaving 37 Left TLE and 26 Right TLE patients. A might be limited. Further analysis of the data revealed no differences
Bayesian linear mixed effects model was computed to determine dif- between personality profiles in TLE and non-TLE patients, and no dif-
ferences in personality domains between left and right TLE patients. ferences in personality profiles between patients with left or right TLE.
The evidence was insensitive to the main effect of domain in this Our finding of elevation in openness and aesthetic interests in PNES
smaller sub-grouping (BFinc = 2.36). The null hypothesis was support patients is not entirely consistent with the previous literature. In a
for the main effect of laterality (BFinc = 0.17) and the laterality by secondary analysis performed by Cragar et al., PNES patients scored
domain interaction (BFinc = 0.05). higher in the domain of neuroticism compared to ES patients, while
differences in the domain of openness were also found to be insignif-
3.7. Diagnostic accuracy icant [28]. This discrepancy could be accounted for by several factors.
One potential factor is the use of the NEO PI-R that was used in Cragar’s
Receiver operating characteristic (ROC) curves were computed to study as opposed to the NEO-FFI. The NEO PI-R assessment allows for
determine whether personality scores accurately classified PNES and ES personality to be described across 30 facet scales in addition to the 5
patients (Fig. 2 and 3). As shown in Table 3, the openness to experience domains [27]. This might permit more comprehensive coverage of
domain had an area under the curve (AUC) greater than .50, which personality, accounting for differences in results. The NEO-FFI was used
suggests that it performs better than chance at classifying diagnostic in our study due to time constraints and has been shown to be both a
group. The sensitivity (53%) and specificity (69%), however, were low. reliable and adequate measurement of a global assessment of person-
Similar results were seen for the aesthetic interests facet in terms of ality - the main focus of our study [27]. Results for neuroticism, con-
sensitivity (46%) and specificity (74%). None of the other personality scientiousness and extraversion were also insensitive in our study. It is
domains produced high sensitivity or specificity metrics. possible that our statistical power was insufficient to determine the
results for these specific domains. If true, however, the effect sizes are
4. Discussion likely to be small and of questionable clinical significance. Never-
theless, the finding of an elevation of openness is novel and adds to the
In this study, we investigated differences in personality profiles body of research reinforcing the association between personality pro-
between patients with PNES and ES using the five-factor model of files and diagnostic groups.

4
M. Leong, et al. Seizure: European Journal of Epilepsy 73 (2019) 1–8

One explanation for the elevation of openness and aesthetic inter-

ests might be related to the association of both openness and ES with
cognition. Openness is one of the more challenging of the five person-
ality factors to describe, due to the wide variety of traits that it seems to
encompass [44]. There has been a great deal of contention in the lit-
erature regarding naming this specific factor, with other contenders
being intellect and culture [44]. McCrae and Costa who coined the term
openness described it as essentiality traits relating to the exploration of
the world from a cognitive point of view [44]. It is, therefore, the factor
that has the greatest correlation with cognitive ability amongst the five
factors of personality [44,45]. Particularly in cognitive domains such as
working memory, latent inhibition and implicit learning [44]. Epilepsy
also has been known to have a strong association with cognition, with
chronic epileptic patients having higher degrees of cognitive impair-
ment [46]. There is evidence that patients who have poor seizure
control, longer seizure duration, and earlier onset of disease are more
adversely affected cognitively [46]. Given that this is most re-
presentative of our VEM cohort, it is possible that observed lower
openness scores might be partly reflective of prevalent cognitive im-
pairment in this group. However, it is important to note that this
finding is speculative and provides a clear direction for future research.
Our study showed no differences in personality profiles and later-
ality which is consistent with a previous study in the literature using the
same five-factor model [47]. In contrast, two other studies in the lit-
erature have shown to contradict these findings [48,49]. These studies
found that patients with left TLE were more dependent and less com-
posed, whereas patients with right TLE were found to have the ten-
dency to underestimate their own problems. However, these studies
were not performed using the five-factor model making them difficult
to be used as studies for comparison. Further research using modern
personality theory, such as the five-factor model, is warranted.
Fig. 1. Distribution of personality scores in the combined PNES and ES cohort. The strengths of our study include the use of a large sample size,
The dotted line shows the mean for the general population. As shown in A, gold standard VEM diagnosis, and the exclusion of non-diagnostic and
participants had higher scores on neuroticism, and lower scores in extraversion mixed groups of patients. This method of patient sampling produces a
and conscientiousness compared to the general population. As shown in B, the high level of certainty regarding classification and allows for well-de-
two groups had comparable scores for all domains except openness to experi- fined phenotypes, greatly increasing the accuracy of the data. Our study
ence, on which the PNES group had high scores compared to the ES group. also has an advantage over most in the area of statistical analysis. The
Scores for the ES group were below the mean for the general population. use of Bayesian statistics allows us to overcome the limitations of null

Table 2
Comparison between diagnostic groups (PNES and ES).
Mean (SD)

ES PNES BF10 BF01 d [95% CI]

Domains
Neuroticism 0.39 (0.85) 0.63 (1.11) 0.68 1.48 −0.23 [−0.50, 0.03]
Extraversion −0.13 (1.03) −0.35 (1.02) 0.47 2.12 0.20 [−0.06, 0.47]
Openness −0.34 (0.98) 0.15 (1.19) 21.55 0.05 −0.43 [−0.71, −0.17]
Agreeableness −0.16 (1.19) −0.30 (1.21) 0.21 4.79 0.11 [−0.16, 0.38]
Conscientiousness −0.47 (1.05) −0.73 (1.01) 0.68 1.47 0.23 [−0.03, 0.50]

Facets
Negative affect 0.35 (0.75) 0.57 (0.95) 0.77 1.31 −0.25 [−0.51, 0.02]
Self-reproach 0.58 (1.01) 0.81 (1.27) 0.42 2.40 −0.19 [−0.46, 0.08]
Positive affect 0.02 (0.92) −0.16 (1.02) 0.37 2.74 0.18 [−0.08, 0.45]
Sociability 0.26 (0.93) 0.14 (0.85) 0.23 4.39 0.12 [−0.14, 0.39]
Activity −0.20 (0.83) −0.35 (1.01) 0.30 3.36 0.15 [−0.11, 0.42]
Aesthetic interests −0.57 (0.96) −0.14 (1.17) 7.98 0.13 −0.39 [−0.66, −0.12]
Intellectual interests −0.33 (0.84) −0.10 (0.90) 0.88 1.13 −0.26 [−0.52, 0.02]
Unconventionality −0.38 (0.89) −0.06 (0.93) 2.99 0.33 −0.33 [−0.60. −0.07]
Nonantagonistic orientation −0.41 (1.04) −0.57 (1.11) 0.26 3.85 −0.14 [−0.13, 0.40]
Prosocial orientation −0.15 (1.32) −0.14 (1.14) 0.15 6.60 −0.01 [−0.28, 0.26]
Orderliness −0.27 (1.00) −0.58 (0.97) 1.81 0.55 0.30 [0.03, 0.57]
Goal striving −0.04 (0.93) −0.08 (0.99) 0.16 6.29 0.04 [−0.22, 0.30]
Dependability −0.55 (1.10) −0.72 (0.91) 0.29 3.51 0.15 [−0.12, 0.42]

Note: ES = Epileptic Seizures. PNES = Psychogenic non-epileptic seizures. BF10 = Bayes Factor for the alternative hypothesis. BF01 = Bayes Factor for the null
hypothesis. d = Cohen’s d and 95% credible intervals. BF10 > 3 considered evidence for an effect of the domain or facet. BF01 > 3 considered evidence for the null
hypothesis.

5
M. Leong, et al. Seizure: European Journal of Epilepsy 73 (2019) 1–8

Fig. 2. Receiver operating characteristic (ROC) curves for personality score domains in terms of classification of PNES and ES.

hypothesis significance testing which has come under considerable personality data and limited demographics. Further research should
criticism [50]. The built-in power analysis enables us to gather evidence investigate the incremental validity of using other commonly collected
for the null. For example, in our study, the domain agreeableness was variables in combination with the NEO-FFI. Another limitation is the
revealed to support the null hypothesis, indicating that there are no possible social desirability bias associated with self-report ques-
differences in agreeableness between ES and PNES patients. To our tionnaires and the characteristic nature of some PNES patients. Future
knowledge, our study has also been the first to directly investigate the research should aim to control for such bias by conducting comparison
classification accuracy of the five-factor model in a clinical setting. of self-reports with observer ratings as suggested by McCrae and Costa
While conventional statistical methods are used to identify group dif- [51].
ferences no matter how small, they do not address the clinically im- The novel finding of elevated openness and aesthetic interests in
portant question whether they assist in diagnosis. This is important to PNES patients compared to ES patients may provide a new avenue for
consider as often group differences and clinical classification are not management strategies in PNES patients. Currently, management in
aligned. Our study demonstrated considerable group differences pre- PNES patients involves a mixture of antidepressants and psychotherapy
sent in the domain of openness and facet of aesthetic interests, however, [52,53]. Evidence for these strategies is limited and remains a large gap
they did not perform sufficiently well to facilitate clinical use of the in the literature. Patients who tend to score higher on the domain of
NEO-FFI for diagnostic purposes. openness have been shown to be more receptive to unconventional
The main limitation of our study is the absence of other clinical forms of therapy such as imagery techniques [27]. Whilst no studies
information (for example, the number of years since first seizure-like have investigated the relation between the facet of aesthetic interests
event, length and frequency of seizures, and anti-epileptic drugs and therapy, people who score higher on this facet have a greater ap-
(AEDs)) in the collected patient data. Some of these variables have been preciation for art and may be more suited to creative therapeutic
used in combination with other psychometric testing in the past to in- modalities, such as art and music therapy. To our knowledge, there is
crease PNES classification accuracy [19,20,26]. However, this study no research studying the benefits of art and music therapy in PNES
was a clinical audit and it was necessary to limit the investigation to patients, therefore, this presents a direction for future research.

Fig. 3. Receiver operating characteristic (ROC) curves for personality score facets in terms of classification of PNES and ES.

6
M. Leong, et al. Seizure: European Journal of Epilepsy 73 (2019) 1–8

Table 3
Classification performance for personality domain and facet scores.
Psychometric marker AUC [95%CIs] Threshold Sensitivity Specificity PPV NPV

Domains
Neuroticism .57 [.49, .65] > 1.20 .36 .85 .65 .64
Extaversion .56 [.49, .64] < −1.40 .18 .93 .65 .61
Openness .62 [.54, .69] > 0.08 .53 .69 .56 .67
Agreeableness .53 [.45, .61] < −0.40 .50 .60 .48 .62
Conscientiousness .59 [.50, .66] < −0.01 .80 .36 .48 .71

Facets
Negative affect .58 [.50, .66] > 1.08 .32 .85 .60 .62
Self-reproach .55 [.48, .64] > 1.66 .32 .86 .63 .63
Positive affect .56 [.48, .63] < 0.19 .62 .48 .47 .63
Sociability .53 [.45, .61] < 0.51 .66 .40 .45 .62
Activity .56 [.48, .64] < −0.41 .52 .59 .48 .62
Aesthetic interests .61 [.53, .69] > 0.10 .46 .74 .57 .65
Intellectual interests .55 [.48, .63] > −0.87 .83 .30 .47 .70
Unconventionality .59 [.52, .69] > −0.13 .55 .62 .52 .65
Nonantagonistic orientation .54 [.46, .62] < −1.02 .39 .71 .50 .61
Prosocial orientation .50 [.42, .58] > −3.02 .99 .04 .43 .83
Orderliness .58 [.51, .66] < −0.01 .74 .44 .49 .69
Goal striving .48 [.40, .56] > 1.19 .11 .91 .48 .58
Dependability .44 [.37, .52] > −1.79 .88 .15 .44 .63

Note: AUC = area under the curve. CIs = 95% credible intervals. Threshold is the optimal cut-off for classifying PNES patients. PPV = positive predictive value.
NPV = negative predictive value.

However, it is important to note that PNES patients are a heterogenous [3] Magee JA, Burke T, Delanty N, Pender N, Fortune GM. The economic cost of
group with different psychological profiles and defence mechanisms nonepileptic attack disorder in Ireland. Epilepsy Behav 2014;33:45–8.
[4] Nowack WJ. Epilepsy: a costly misdiagnosis. Clin Electroencephalogr
[9,54], and therapy should ultimately be tailored to the individual. 1997;28:225–8.
[5] Whitaker JN. The confluence of quality of care, cost-effectiveness, pragmatism, and
5. Conclusion medical ethics in the diagnosis of nonepileptic seizures: a provocative situation for
neurology. Arch Neurol 2001;58:2066–7.
[6] Karterud HN, Knizek BL, Nakken KO. Changing the diagnosis from epilepsy to PNES:
Overall, our findings show that patients with ES exhibit lower patients’ experiences and understanding of their new diagnosis. Seizure
openness to experience and aesthetic interest compared to patients with 2010;19:40–6.
[7] Galimberti CA, Ratti MT, Murelli R, Marchioni E, Manni R, Tartara A. Patients with
PNES and compared to the general population. These findings open up psychogenic nonepileptic seizures, alone or epilepsy-associated, share a psycholo-
new avenues of research using modern personality models to further gical profile distinct from that of epilepsy patients. J Neurol 2003;250:338–46.
understand patients with epilepsy and related presentations. https://doi.org/10.1007/s00415-003-1009-0.
[8] Scévola L, Teitelbaum J, Oddo S, Centurión E, Loidl CF, Kochen S, et al. Psychiatric
disorders in patients with psychogenic nonepileptic seizures and drug-resistant
Declaration of Competing Interest epilepsy: a study of an Argentine population. Epilepsy Behav 2013;29:155–60.
https://doi.org/10.1016/j.yebeh.2013.07.012.
All authors have no competing interests to declare. [9] Beghi M, Negrini PB, Perin C, Peroni F, Magaudda A, Cerri C, et al. Psychogenic
non-epileptic seizures: so-called psychiatric comorbidity and underlying defense
mechanisms. Neuropsychiatr Dis Treat 2015;11:2519.
Acknowledgements [10] Kanner AM, Schachter SC, Barry JJ, Hersdorffer DC, Mula M, Trimble M, et al.
Depression and epilepsy, pain and psychogenic non-epileptic seizures: clinical and
therapeutic perspectives. Epilepsy Behav 2012;24:169–81. https://doi.org/10.
We would like to thank all patients admitted to the Royal 1016/j.yebeh.2012.01.008.
Melbourne Hospital’s VEM unit who have taken the time to participate [11] Hovorka J, Nežádal T, Herman E, Neˇmcová I, Bajacˇek M. Psychogenic non-epi-
in this study. We also would like to extend our appreciation to all staff leptic seizures, prospective clinical experience: diagnosis, clinical features, risk
factors, psychiatric comorbidity, treatment outcome. Epileptic Disord 2017;9:7.
who have assisted in the collection of data. [12] Tojek TM, Lumley M, Barkley G, Mahr G, Thomas A. Stress and other psychosocial
characteristics of patients with psychogenic nonepileptic seizures. Psychosomatics
Funding 2000;41:221–6. https://doi.org/10.1176/appi.psy.41.3.221.
[13] Lacey C, Cook M, Salzberg M. The neurologist, psychogenic nonepileptic seizures,
and borderline personality disorder. Epilepsy Behav EB 2007;11:492–8.
This research did not receive any specific grant from funding [14] Harden CL, Jovine L, Burgut FT, Carey BT, Nikolov BG, Ferrando SJ. A comparison
agencies in the public, commercial, or not-for profit sectors. of personality disorder characteristics of patients with nonepileptic psychogenic
pseudoseizures with those of patients with epilepsy. Epilepsy Behav EB
2009;14:481–3. https://doi.org/10.1016/j.yebeh.2008.12.012.
Appendix A. Supplementary data [15] Krishnamoorthy ES, Brown RJ, Trimble MR. Personality and psychopathology in
nonepileptic attack disorder and epilepsy: a prospective study. Epilepsy Behav EB
Supplementary material related to this article can be found, in the 2001;2:418–22.
[16] Cragar DE, Berry DT, Fakhoury TA, Cibula JE, Schmitt FA. A review of diagnostic
online version, at doi:https://doi.org/10.1016/j.seizure.2019.10.011. techniques in the differential diagnosis of epileptic and nonepileptic seizures.
Neuropsychol Rev 2002;12:31–64.
References [17] Wilkus RJ, Dodrill CB, Thompson PM. Intensive EEG monitoring and psychological
studies of patients with pseudoepileptic seizures. Epilepsia 1984;25:100–7.
[18] Derry PA, McLachlan RS. The MMPI-2 as an adjunct to the diagnosis of pseudo-
[1] LaFrance Jr WC, Baker GA, Duncan R, Goldstein LH, Reuber M. Minimum re- seizures. Seizure 1996;5:35–40.
quirements for the diagnosis of psychogenic nonepileptic seizures: a staged ap- [19] Storzbach D, Binder LM, Salinsky MC, Campbell BR, Mueller RM. Improved pre-
proach: a report from the international league against epilepsy nonepileptic seizures diction of nonepileptic seizures with combined MMPI and EEG measures. Epilepsia
task force. Epilepsia 2013;54:2005–18. 2000;41:332–7. https://doi.org/10.1111/j.1528-1157.2000.tb00164.x.
[2] de Timary P, Fouchet P, Sylin M, Indriets JP, de Barsy T, Lefèbvre A, et al. Non- [20] Schramke CJ, Valeri A, Valeriano JP, Kelly KM. Using the Minnesota Multiphasic
epileptic seizures: delayed diagnosis in patients presenting with electroencephalo- Inventory 2, EEGs, and clinical data to predict nonepileptic events. Epilepsy Behav
graphic (EEG) or clinical signs of epileptic seizures. Seizure 2002;11:193–7. 2007;11:343–6. https://doi.org/10.1016/j.yebeh.2007.06.011.

7
M. Leong, et al. Seizure: European Journal of Epilepsy 73 (2019) 1–8

[21] Owczarek K, Jedrzejczak J. Patients with coexistent psychogenic pseudoepileptic [36] John O. The “Big Five” factor taxonomy: dimensions of personality in the natural
and epileptic seizures: a psychological profile. Seizure 2001;10:566–9. language and in questionnaires. Handb. Personal. Res.. 1990. p. 66–100.
[22] Thompson AW, Hantke N, Phatak V, Chaytor N. The Personality Assessment [37] McCrae RR. The five-factor model and its assessment in clinical settings. J Pers
Inventory as a tool for diagnosing psychogenic nonepileptic seizures. Epilepsia Assess 1991;57. https://doi.org/10.1207/s15327752jpa5703_2. 399–314.
2010;51:161–4. https://doi.org/10.1111/j.1528-1167.2009.02151.x. [38] Digman J. Personality structure:emergence of the five-factor model. Annu Rev Clin
[23] Marc Testa S, Lesser RP, Krauss GL, Brandt J. Personality Assessment Inventory Psychol 1990;41:417–40.
among patients with psychogenic seizures and those with epilepsy. Epilepsia [39] McCrae RR. NEO-PIR data from 36 cultures. Five-factor model personal. Cult..
2011;52:e84–8. https://doi.org/10.1111/j.1528-1167.2011.03141.x. Springer; 2002. p. 105–25.
[24] Wagner MT, Wymer JH, Topping KB, Pritchard PB. Use of the Personality [40] Saucier G. Replicable item-cluster subcomponents in the NEO five-factor inventory.
Assessment Inventory as an efficacious and cost-effective diagnostic tool for none- J Pers Assess 1998;70:263–76.
pileptic seizures. Epilepsy Behav EB 2005;7:301–4. [41] JASP Team. JASP(Version 0.9.2)[Computer software]. 2019.
[25] Reuber M, Pukrop R, Bauer J, Derfuss R, Elger CE. Multidimensional assessment of [42] R Core Team. R: a language and environment for statistical computing. 2019.
personality in patients with psychogenic non-epileptic seizures. J Neurol Neurosurg [43] Cohen J. Statistical power analysis for the behavioral sciences. Routledge; 2013.
Psychiatry 2004;75:743–8. [44] DeYoung CG. Openness/Intellect: a dimension of personality reflecting cognitive
[26] Hill SW, Gale SD. Predicting psychogenic nonepileptic seizures with the Personality exploration. APA Handb Personal Soc Psychol Personal Process Individ Differ
Assessment Inventory and seizure variables. Epilepsy Behav EB 2011;22:505–10. 2014;4:369–99.
https://doi.org/10.1016/j.yebeh.2011.08.001. [45] Zillig LMP, Hemenover SH, Dienstbier RA. What do we assess when we assess a Big
[27] Costa PT, MacCrae RR. Revised NEO personality inventory (NEO PI-R) and NEO 5 trait? A content analysis of the affective, behavioral, and cognitive processes
five-factor inventory (NEO-FFI): professional manual. Psychological Assessment represented in Big 5 personality inventories. Pers Soc Psychol Bull 2002;28:847–58.
Resources, Incorporated; 1992. [46] Elger CE, Helmstaedter C, Kurthen M. Chronic epilepsy and cognition. Lancet
[28] Cragar DE, Berry DTR, Schmitt FA, Fakhoury TA. Cluster analysis of normal per- Neurol 2004;3:663–72.
sonality traits in patients with psychogenic nonepileptic seizures. Epilepsy Behav [47] Swinkels WAM, Van Emde Boas W, Kuyk J, Van Dyck R, Spinhoven P. Interictal
2005;6:593–600. https://doi.org/10.1016/j.yebeh.2005.03.007. Depression, Anxiety, Personality Traits, and Psychological Dissociation in Patients
[29] Ekanayake V, Kranick S, LaFaver K, Naz A, Frank Webb A, LaFrance WCJ, et al. with Temporal Lobe Epilepsy (TLE) and Extra-TLE: INTERICTAL
Personality traits in psychogenic nonepileptic seizures (PNES) and psychogenic PSYCHOPATHOLOGY IN TLE AND EXTRA-TLE. Epilepsia 2006;47:2092–103.
movement disorder (PMD): neuroticism and perfectionism. J Psychosom Res https://doi.org/10.1111/j.1528-1167.2006.00808.x.
2017;97:23–9. https://doi.org/10.1016/j.jpsychores.2017.03.018. [48] Bear DM, Fedio P. Quantitative analysis of interictal behavior in temporal lobe
[30] Vilyte G, Pretorius C. Personality traits, illness behaviors, and psychiatric co- epilepsy. Arch Neurol 1977;34:454–67.
morbidity in individuals with psychogenic nonepileptic seizures (PNES), epilepsy, [49] Feddersen B, Herzer R, Hartmann U, Gaab MR, Runge U. On the psychopathology of
and other nonepileptic seizures (oNES): differentiating between the conditions. unilateral temporal lobe epilepsy. Epilepsy Behav EB 2005;6:43–9.
Epilepsy Behav 2019;98:210–9. [50] Wasserstein RL, Lazar NA. The ASA’s statement on p-values: context, process, and
[31] Fisher RS, Boas WVE, Blume W, Elger C, Genton P, Lee P, et al. Epileptic seizures purpose. Am Stat 2016;70:129–33.
and epilepsy: definitions proposed by the International League Against Epilepsy [51] McCrae RR, Costa PT. Validation of the five-factor model of personality across in-
(ILAE) and the International Bureau for Epilepsy (IBE). Epilepsia 2005;46:470–2. struments and observers. J Pers Soc Psychol 1987;52:81–90.
[32] Fisher RS, Acevedo C, Arzimanoglou A, Bogacz A, Cross JH, Elger CE, et al. ILAE [52] LaFrance WC, Keitner GI, Papandonatos GD, Blum AS, Machan JT, Ryan CE, et al.
official report: a practical clinical definition of epilepsy. Epilepsia 2014;55:475–82. Pilot pharmacologic randomized controlled trial for psychogenic nonepileptic sei-
[33] Association AP. Diagnostic and statistical manual of mental disorders (DSM-5®). zures. Neurology 2010;75:1166–73. https://doi.org/10.1212/WNL.
American Psychiatric Pub; 2013. 0b013e3181f4d5a9.
[34] Betts T, Boden S. Diagnosis, management and prognosis of a group of 128 patients [53] Goldstein LH, Chalder T, Chigwedere C, Khondoker MR, Moriarty J, Toone BK, et al.
with non-epileptic attack disorder. Part II. Previous childhood sexual abuse in the Cognitive-behavioral therapy for psychogenic nonepileptic seizures: a pilot RCT.
aetiology of these disorders. Seizure 1992;1:27–32. Neurology 2010;74:1986–94. https://doi.org/10.1212/WNL.0b013e3181e39658.
[35] Berg AT, Berkovic SF, Brodie MJ, Buchhalter J, Cross JH, van Emde Boas W, et al. [54] Marchetti RL, Kurcgant D, Neto JG, von Bismark MA, Marchetti LB, Fiore LA.
Revised terminology and concepts for organization of seizures and epilepsies: report Psychiatric diagnoses of patients with psychogenic non-epileptic seizures. Seizure
of the ILAE Commission on Classification and Terminology, 2005–2009. Epilepsia 2008;17:247–53.
2010;51:676–85.