Embed Size (px)
Citation preview
Epilepsy & Behavior 29 (2013) 178–183
Contents lists available at ScienceDirect
Epilepsy & Behavior
j ourna l homepage: www.e lsev ie r .com/ locate /yebeh
Emotional dysregulation, alexithymia, and attachment in psychogenicnonepileptic seizures
Richard J. Brown a,⁎, Julia F. Bouska a, Anna Frow a, Antonia Kirkby b, Gus A. Baker c, Steven Kemp d,Christine Burness c, Markus Reuber e
a School of Psychological Sciences, University of Manchester, UKb Department of Clinical Neuropsychology, Salford Royal Hospital, Salford, UKc Department of Neuropsychology, Walton Centre for Neurology and Neurosurgery, Liverpool, UKd St James' University Hospital, Leeds, UKe Academic Neurology Unit, University of Sheffield, Sheffield, UK
⁎ Corresponding author at: School of Psychological ScienBrunswick Street, University ofManchester, ManchesterM0406.
E-mail address: [email protected]
1525-5050/$ – see front matter © 2013 Elsevier Inc. All rihttp://dx.doi.org/10.1016/j.yebeh.2013.07.019
a b s t r a c t
a r t i c l e i n f oArticle history:
Received 15 April 2013Revised 17 July 2013Accepted 22 July 2013Available online xxxxKeywords:Psychogenic nonepileptic seizuresConversion disorderAlexithymiaEmotional regulationAttachment
Objectives: Psychogenic nonepileptic seizures (PNESs) are poorly understood and difficult to treat. Research andtheory suggest that problems with recognizing, acknowledging, and regulating emotional states (i.e., emotionaldysregulation) may contribute to the development and maintenance of PNESs. However, there is a lack of well-controlled studies using dedicatedmeasures of emotional regulationwith patientswith PNESs. The current studysought to address this gap.Methods: Forty-three patients with PNESs and 24 with epilepsy completed a postal survey comprisingmeasuresof emotional dysregulation (Difficulties in Emotion Regulation Scale), alexithymia (Toronto Alexithymia Scale),attachment (Relationship Scales Questionnaire), and psychopathology (Generalized Anxiety Disorder-7; PatientHealth Questionnaire-9; Somatoform Dissociation Questionnaire-20). Cluster analysis was used to identify pos-sible subgroups of patients with PNESs characterized by distinct patterns of emotional dysregulation.Results: Two clusters of patients with PNESs were identified. The first (n = 11) was characterized by higher
levels of psychopathology, somatization, alexithymia, and difficulties with most aspects of emotional regulation(including identifying, accepting, and describing feelings, accessing adaptive regulatory strategies, performinggoal-directed behaviors, and controlling feelings and actions) compared with the group with epilepsy. The sec-ond (n = 32)was characterized by relatively high somatization and depression scores but comparatively normallevels of alexithymia and emotional regulation.Conclusions: The findings suggest that patients with PNESs can be divided into at least twomeaningful subgroupscharacterized by distinct psychological profiles, only one of which is characterized by significant problems withemotional dysregulation. Further research is needed to determine whether the relatively normal emotionaldysregulation and high somatization scores of some patients with PNESs are due to emotional avoidance ormore basic problems with perceptual and behavioral control.© 2013 Elsevier Inc. All rights reserved.
1. Introduction
Psychogenic nonepileptic seizures (PNESs) superficially resembleepileptic seizures but result from psychological processes rather thanabnormal electrical activity in the brain. Psychogenic nonepileptic sei-zures are common [1] and associated with considerable personal, eco-nomic, and social costs [2]; however, they remain poorly understood,and there is a lack of evidence-based treatments [3]. As such, there is a
ces, 2nd Floor Zochonis Building,13 9WL, UK. Fax:+44 161 306
k (R.J. Brown).
ghts reserved.
pressing need for research that further elucidates the psychologicalbasis of PNESs.
Several studies using variants of the Minnesota Multiphasic Person-ality Inventory (MMPI) have suggested that patients with PNESs arecharacterized by a “conversion V” personality profile, associated withhigh scores on the hypochondriasis and hysteria subscales coupledwith comparatively low (but still elevated) depression scores [4–8].Following Freud [9], these findings have often been interpreted as evi-dence for a defensive process (i.e., conversion) in patients with PNESs,inwhich thedistress associatedwith emotional conflict and/or traumat-icmemories is converted into physical symptoms to protect the individ-ual. Others have suggested that this pattern reflects a general inabilityor unwillingness to perceive, process, and communicate emotionalstates (so-called alexithymia) [10], with the resulting buildup of tension
179R.J. Brown et al. / Epilepsy & Behavior 29 (2013) 178–183
leading to PNESs. Such difficulties are often thought to stem from mal-adaptive attachments with early caregivers, which undermine thedevelopment of appropriate ways of understanding and relating toemotional states in oneself and others [11].
Despite the apparent popularity of such concepts, several studieswith groupswith PNESs have failed to replicate the conversion V profileon the MMPI [12,13], and the few published studies on alexithymia inPNESs have not demonstrated higher levels of alexithymia in thesepatients compared with those with epilepsy [10,14]. One possibility isthat these mixed findings reflect the heterogeneity of patients withPNESs in terms of etiology and comorbid psychopathology. While anxi-ety, affective, personality, somatoform, and dissociative disorders are alldisproportionately common in samples with PNESs compared withthose with epileptic seizures [15], the type and extent of psychopathol-ogy vary considerably between patients. In some cases, PNESs may beone aspect of a much wider set of psychiatric complaints; in others,PNESs may occur in seemingly well-adjusted individuals with minimalcomorbidity. This interpretation would be in keeping with the observa-tion that there is similar variation in the psychosocial histories of pa-tients with PNESs, ranging from the apparently ordinary to a lifetimeof serious adversity and potentially traumatizing events [16].
A small number of studies have addressed the problem of heteroge-neity by using cluster analysis to identify subgroups of patients withPNESs with distinct psychological profiles. Cragar et al. [17], for exam-ple, clustered patients with PNESs (n = 66) according to their scoreson a measure of normal personality then compared the clusters on theMMPI-2 and a neuropsychological battery. Three clusters of patientswith PNESs emerged: two characterized by high levels of neuroticism,elevated scores on several MMPI dimensions, and low-average intellec-tual ability and a third (33% of the sample) characterized by averagescores on theneuropsychological battery and an apparently normal per-sonality. Only this third cluster exhibited a conversion V profile on theMMPI-2. Reuber et al. [18], in contrast, found that the majority of theirsamplewith PNESs (80/85 patients)were classifiedwithin two clusters,the first characterized by elevated scores on all four of the higher ordertraits of personality pathology studied, insecure attachment, a greaterlikelihood of inpatient psychiatric treatment, and a relatively poor prog-nosis and the second (43.5% of the sample) characterized by a relativelygood prognosis and normal scores on all personality measures apartfrom compulsivity. A similar pattern was found by Uliaszek et al. [19],who identified two clusters of patients with PNESs, one (n = 14) char-acterized by emotional dysregulation, poor quality of life, and highscores on measures of depression, anxiety, stress, somatization, anddissociation and a second (n = 41) characterized by low emotionaldysregulation, higher quality of life, and minimal psychopathology.
These cluster analytic findings have been interpreted as confirma-tion that most patients with PNESs have problems in regulating theiremotional states and that there are at least two different types of mal-adaptive emotional regulation that may predispose to PNESs. The firsttype appears to be characterized by emotional reactivity, poor arousaltolerance, and difficulties in controlling affect; this “undermodulation”of emotion [20] is akin to that seen in patients with borderline person-ality disorder [18], which is commonly comorbid with PNESs [21] andconfers vulnerability to all types of psychopathology. In contrast, the ap-parently normal personality and minimal comorbidity of the secondgroup have been interpreted as evidence of “overmodulated” affect inthese individuals [20], resulting from a “somatic defense” [17] charac-terized by emotional avoidance, excessively controlled behavior, and atendency to use physical symptoms as a way of expressing emotionalconflicts [18].
Of the few studies in this area, only one [19] has studied emotionaldysregulation in patients with PNESs using a dedicated measure ofthis construct, but no data from the cluster analysis are presented, mak-ing it difficult to judge its validity. Moreover, the emotional dysreg-ulation scores of clusters with PNESs in this study were classified aslow or high relative to normative scores on the Difficulties in Emotion
Regulation Scale (DERS) obtained fromundergraduate students. It is un-certain how these profiles compare with a more meaningful clinicalcontrol group, such as patients with epilepsy. This is particularly impor-tant given the results of Reuber et al. [18], who found that their seem-ingly overmodulated group with PNESs actually had scores that werevery similar to those of a control group with epilepsy on two of thethree facets of emotional dysregulation where they had scored lowerthan healthy controls. As such, we are some way from being able todraw firm conclusions regarding emotional dysregulation in patientswith PNESs.
In order to shed light on these issues, we collected data concerningemotional dysregulation, alexithymia, attachment, and psychopatholo-gy in groups of patients with either PNESs or epilepsy and conductedcluster analysis to determine whether there are meaningful subgroupswith PNESs characterized by different kinds of emotional dysregulationdifficulties.
2. Methods
2.1. Design and procedure
The study was carried out in accordance with the Declaration ofHelsinki concerning experiments with human participants. Patientswith PNESs from neurology or neuropsychology services in the northor northwest of England (Salford Royal NHS Foundation Trust [Site 1];The Walton Centre for Neurology and Neurosurgery NHS Trust [Liver-pool; Site 2]; Sheffield Teaching Hospitals NHS Foundation Trust [Site3]; The Leeds Teaching Hospitals NHS Trust [Site 4]) were approachedto take part in a cross-sectional postal survey. Patients with epilepsywere recruited from a single neurology service (Sheffield Teaching Hos-pitals NHS Foundation Trust). Consultant neurologists or neuropsychol-ogists identified potential participants from their existing caseloads(retrospective recruitment) and new patients as they came into thestudy sites (prospective recruitment). Secretarial staff sent potentialparticipants a study pack, consisting of an invitation letter from thepatient's consultant, an information sheet, a consent form, and a ques-tionnaire booklet. Participants who wished to take part returned thecompleted questionnaires and consent form in the stamped addressedenvelope provided. These participants were entered into a £150 prizedraw.
2.1.1. Inclusion/exclusion criteriaThe groupwith PNESs consisted of adult (≥18 years) patientswith a
diagnosis of PNESs made by an experienced consultant neurologist orconsultant neuropsychologist based on clinical history/presentationand neurological examination in all cases, EEG recordings, brain imag-ing, and video-EEG telemetry findings when available. The group withepilepsy consisted of adult (≥18 years) patients with a diagnosis ofepilepsy made by an experienced consultant neurologist on the samebasis. Individuals with any form of epilepsy were included in order tomaximize power. All diagnoses were verified by a consultant neurolo-gist or neuropsychologist review of patients' medical records. Individ-uals still under investigation or without a clinically definite diagnosisof PNESs or epilepsy were excluded. Patients with likely mixed seizuredisorders were excluded. Other exclusion criteria were inability togive informed consent, inability to read or speak English, and moderateto severe learning disabilities as indicated by patients' medical recordsor by the responsible clinician. We did not specifically exclude patientswith a history of traumatic brain injury, other neurologic diseases, or co-morbid psychiatric disorders.
2.1.2. ParticipantsParticipants were recruited over eight months. During this time, a
total of 273 possible patientswith PNESswere identified and approached,of whom 51 returned completed questionnaire packs (response rate =18.7%; Site 1 = 9/50 [18%]; Site 2 = 25/65 [38.5%]; Site 3 = 16/147
180 R.J. Brown et al. / Epilepsy & Behavior 29 (2013) 178–183
[10.9%]; Site 4 = 1/11 [9.1%]). Of these, 43 (88%) had confirmed PNESsonly according to their medical records and were entered into the anal-ysis. Thirty-one (72.1%) of these participants had video-EEG confirma-tion of their diagnosis; in the remaining cases, confident diagnoses ofPNESsweremade in the absence of video-EEG confirmation. Completedpackswere also returned by 28 out of 157 possible patients with epilep-sy identified and approached at Site 3 (response rate = 17.8%); analy-sis was restricted to those cases (n = 24; 85.7%)where diagnoses couldbe confirmed from records. Of these, 11 patients had focal epilepsy, andsix had idiopathic generalized epilepsy; the remaining seven patientshad unclassifiable epilepsy.
Table 1 presents characteristics of the sample. All of the partici-pants self-identified as white British. There were no significant differ-ences between the groups in age (U(67) = 492.5, p = 0.758), gender(χ2
(1) = 0.303, p = .582), marital status (χ2(5) = 8.811, p = .117),
employment status (χ2(5) = 4.443, p = .487), qualifications (χ2
(6) =2.965, p = .813), or seizure frequency (U(66) = 353.0, p = 0.074). Inboth groups, the majority of participants were female and high propor-tions of participants were unemployed and/or had no formal qualifica-tions, although degree-level education or higher was comparativelycommon in the group with PNESs.
2.1.3. MeasuresThe following self-report measures were included in the question-
naire booklet, which took approximately 45 min to complete:
Demographics and seizure questionnaire. A measure was constructedto obtain information about gender, age, marital status, employmentstatus, highest qualification received, and ethnic group. Participantswere also asked to indicate whether they had experienced a seizurein the last six months and to estimate approximately howmany sei-zures they had experienced within this time.
Table 1Demographic characteristics and seizure frequency of total samples with PNESs andepilepsy.
Demographic variable PNES Epilepsy
Gender (n = 43) (n = 24)Male 15 (34.9%) 10 (41.7%)Female 28 (65.1%) 14 (58.3%)
Median age [IQR] (n = 45) (n = 24)42.0 [27.0] 43.5 [26.0]
Marital status (n = 45) (n = 24)Single 12 (27.9%) 10 (41.7%)Cohabiting 9 (20.9%) 1 (4.2%)Married 15 (34.9%) 7 (29.2%)Separated 3 (7.0%) 0 (0%)Divorced 1 (2.3%) 3 (12.5%)Widowed 3 (7.0%) 3 (12.5%)
Employment status (n = 43) (n = 24)Unemployed 19 (44.2%) 7 (29.2%)Employed part-time 4 (9.3%) 3 (12.5%)Employed full-time 4 (9.3%) 6 (25.0%)Self-employed 1 (2.3%) 0 (0%)Retired 11 (25.6%) 5 (20.8%)Full-time student 4 (9.3%) 3 (12.5%)
Qualifications received (n = 40) (n = 22)No formal qualifications 9 (20.9%) 5 (22.7%)Entry level certificate or equivalent 1 (2.3%) 1 (4.5%)GCSE/O-level/NVQ levels 1–2 or equivalent 13 (30.2%) 6 (27.3%)A-level/NVQ level 3 or equivalent 3 (7.0%) 4 (18.2%)Higher education below degree level 4 (9.3%) 3 (13.6%)Degree or degree equivalent 8 (18.6%) 2 (9.1%)Postgraduate or postgraduate equivalent 2 (4.7%) 1 (4.5%)
Seizure in last 6 months? (n = 43) (n = 24)Yes 35 (81.4%) 13 (54.2%)No 8 (18.6%) 11 (45.8%)
Median number of seizures in last six months [IQR] (n = 40) (n = 24)12.5 [96.5] 1.5 [40.25]
N.B. Variation in sample sizes indicates missing data for certain variables.IQR = interquartile range.
Difficulties in Emotion Regulation Scale (DERS) [22]. The DERS is a 36-item questionnaire designed to assess different aspects of emo-tional dysregulation. Participants were presented with statementsdescribing how they might relate to their feelings or react whenthey are upset (e.g., “I have no idea how I feel”; “When I'm upset, Ibecomeout of control”) and indicated howmuch each statement ap-plied to them on a five-point Likert scale ranging from 1 (“almostnever [0–10%]”) to 5 (“almost always [91–100%]”). The DERS yieldsa total emotional dysregulation score as well as scores on six scalespertaining to six different aspects of the construct: Nonacceptance(nonacceptance of emotional responses), Goals (difficulties engag-ing in goal-directed behavior), Impulse (impulse control difficul-ties), Awareness (lack of emotional awareness), Strategies (limitedaccess to emotional regulation strategies), and Clarity (lack of emo-tional clarity). Higher DERS total and subscale scores representgreater emotional regulation difficulties on that scale. Evidenceconcerning the reliability and validity of the scale is described bythe scale authors [22]. Subscale reliabilities across the entire samplein this study ranged from very good to excellent (Nonacceptance =.91; Goals = 0.86; Impulse = 0.87; Awareness = 0.79; Strategies =0.87; Clarity = 0.80).Toronto Alexithymia Scale-20 (TAS-20) [23]. The TAS-20 is a 20-itemself-report questionnaire designed to measure three different facetsof the alexithymia construct: Difficulty Identifying Feelings, Difficul-ty Describing Feelings, and Externally Oriented Thinking. Partici-pants were asked to indicate how much they agreed or disagreedwith 20 statements on a five-point Likert scale from 1 (“strongly dis-agree”) to 5 (“strongly agree”). Evidence concerning the reliabilityand validity of the scale is described in the original reports [23,24].Scale reliability across the entire sample in this study was verygood for the Difficulty Identifying Feelings subscale (α = 0.81)and good for Difficulty Describing Feelings (α = 0.73) but less satis-factory for Externally Oriented Thinking (α = 0.59).Relationship Scales Questionnaire (RSQ) [25]. The RSQ is a 30-itemself-report questionnaire designed tomeasure four common attach-ment styles (Secure, Dismissing, Fearful, and Preoccupied). Partici-pants were asked to rate the extent to which each of the 30statements described their characteristic style in close relationshipson a five-point Likert scale ranging from 1 (“not at all like me”) to 5(“very much like me”). Evidence pertaining to the reliability andvalidity of the measure is described by Kurdek [26]. As subscale reli-abilities across the entire sample in the current studywere generallypoor (Secure = 0.24, Dismissing = 0.50, Fearful = 0.66, Preoccu-pied = 0.38), we combined the items from the Dismissing, Fearful,and Preoccupied subscales into a single insecure attachment mea-sure with moderate internal consistency (α = 0.63).Generalized Anxiety Disorder-7 (GAD-7) [27]. The GAD-7 was used tomeasure current anxiety symptoms. Participants rated how muchthey had been bothered by seven common anxiety symptoms(e.g., “Feeling nervous, anxious, on edge, or worrying a lot about dif-ferent things”) in the preceding two weeks on a four-point Likertscale ranging from0 (“not at all”) to 3 (“nearly every day”). Evidenceconcerning the reliability and validity of the scale is described inSpitzer et al. Scale reliability across the entire sample in this studywas excellent (α = .92).Patient Health Questionnaire 9 (PHQ-9) [28]. The PHQ-9 was used tomeasure current symptoms of depression. Participants rated howmuch they had been bothered by seven common depressive symp-toms (e.g., “Little interest or pleasure in doing things”) in the preced-ing two weeks on a four-point Likert scale ranging from 0 (“not atall”) to 3 (“nearly every day”). Evidence concerning the reliabilityand validity of the scale is described in the original scale report(2001). Scale reliability across the entire sample in this study wasexcellent (α = .90).Somatoform Dissociation Questionnaire-20 (SDQ-20) [29]. The SDQ-20 was used as a measure of somatization in this study. Participants
Table 2Comparison of median (interquartile range) questionnaire scores for complete groupswith PNESs and epilepsy.
Variable Epilepsy(n = 24)
PNES(n = 43)
U value Effect size(r)
GAD-7 4.5 (11.75) 11.0 (10.0) 337.5 0.29PHQ-9 4.5 (8.75) 13.0 (11.0) 244.5 0.43⁎⁎
SDQ-20 24.0 (7.0) 37.0 (17.0) 187.5 0.53⁎⁎
DERS—Nonacceptance 15.5 (12.25) 19.0 (12.0) 396.5 0.19DERS—Goals 12.0 (9.0) 17.0 (7.0) 255.5 0.42⁎⁎
DERS—Impulse 10.0 (7.5) 15.0 (11.0) 290.0 0.36⁎
DERS—Awareness 15.0 (5.75) 18.0 (9.0) 389.5 0.20DERS—Strategies 15.0 (10.75) 20.0 (13.0) 342.0 0.28DERS—Clarity 10.0 (6.5) 13.0 (8.0) 328.0 0.30TAS—DIF 18.0 (12.5) 24.0 (11.0) 307.5 0.33TAS—DDF 12.0 (7.0) 16.0 (8.0) 323.0 0.31TAS—EOT 21.0 (9.0) 23.0 (13.0) 478.5 0.06RSQ—Insecure 36.5 (12.75) 37.0 (11.0) 495.5 0.03
Key: GAD = Generalized Anxiety Disorder; PHQ = PatientHealth Questionnaire; SDQ =Somatoform Dissociation Questionnaire; DERS = Difficulties in Emotion RegulationScale; TAS = Toronto Alexithymia Scale; DIF = Difficulty Identifying Feelings; DDF =Difficulty Describing Feelings; EOT = Externally Oriented Thinking; RSQ = RelationshipScales Questionnaire.⁎ p ≤ .005.⁎⁎ p ≤ .001.
181R.J. Brown et al. / Epilepsy & Behavior 29 (2013) 178–183
rated howmuch they had been bothered by twenty physical symp-toms or body experiences (e.g., “Sometimes I am paralyzed for awhile”) in the past year on a five-point Likert scale ranging from 1(“this applies tome not at all”) to 5 (“this applies tome extremely”).Participants were also asked to indicate whether a physician hadconnected any of their symptoms with a physical disease. Evidenceconcerning the reliability and validity of the scale is described byNijenhuis [30]. Scale reliability across the entire sample in thisstudy was excellent (α = .87).
2.1.4. Data analysisFive patients (three with PNESs, two with epilepsy) did not indicate
their highest qualification, and three patients with PNESs did not pro-vide usable information regarding their seizure frequency. In the fewremaining cases where individual item responses were incomplete,the missing items were replaced by the participant's mean item scorefor that scale, allowing total scores to be calculated. Five participantswith PNESs completed ≤ 3 SDQ-20 items; in these cases, the partici-pants' total score was replaced by the mean of the group with PNESsfor that scale. Shapiro–Wilk tests indicated that several of the variables(seizure frequency— both groups; GAD— epilepsy; SDQ-20— epilepsy;PQQ-15— epilepsy; DERS—Impulse— both groups; DERS—Strategies—epilepsy; TAS—EOT — both groups) were non-normally distributed inone or both of the groups. For the sake of simplicity, nonparametricanalyseswere, therefore, conducted for all variables, with the exception
Table 3Median questionnaire scores (interquartile range in parentheses) for clusters with PNESs and
Cluster 1 PNES(n = 11)
Cluster 2 PNES(n = 32)
Epilepsy(n = 24)
DERS—Nonacceptance 24.0 (10.0) 17.5 (11.5) 15.5 (12.25)DERS—Goals 21.0 (4.0) 16.0 (8.0) 12.0 (9.0)DERS—Impulse 25.0 (11.0) 12.0 (6.0) 10.0 (7.5)DERS—Awareness 23.0 (9.0) 17.0 (7.0) 15.0 (5.75)DERS—Strategies 34.0 (7.0) 19.0 (9.5) 15.0 (10.75)DERS—Clarity 19.0 (5.0) 11.0 (5.0) 10.0 (6.5)TAS—DIF 31.0 (5.0) 20.0 (11.0) 18.0 (12.5)TAS—DDF 21.0 (4.0) 15.0 (7.75) 12.0 (7.0)TAS—EOT 24.0 (4.0) 22.5 (8.0) 21.0 (9.0)
Key: DERS = Difficulties in Emotion Regulation Scale; TAS = Toronto Alexithymia Scale; DIF =Oriented Thinking.⁎ p ≤ .005.⁎⁎ p ≤ .001.
of the cluster analysis. Initial analyses compared the complete groupswith epilepsy and PNESs on all of the questionnaire measures using aseries of MannWhitney U tests. Tominimize the likelihood of Type I er-rors, a conservative alpha value of p b .005 (two-tailed) was adoptedfor all analyses involving the main study variables. Effect size (r) esti-mates were calculated using the formula provided by Field [31]. Clusteranalysis was then performed on the DERS and TAS-20 data for the par-ticipants with PNESs only to examine whether there were clusters ofpatients with PNESs characterized by differences in emotional regula-tion. Comparisons were then made between the resulting clusters andthe subgroup with epilepsy using Mann Whitney U tests. All analyseswere conducted using SPSS for Windows version 20.0.
3. Results
3.1. Comparison of complete samples with PNESs and epilepsy
Table 2 indicates that the group with PNESs scored significantlyhigher than the group with epilepsy on depression (PHQ-9), somatiza-tion (SDQ-20), and some aspects of emotional dysregulation (Goalsand Impulse subscales of the DERS). There were nonsignificant trends(.005 b p b .05) for between-groups differences in anxiety (GAD-7)and some aspects of alexithymia (TAS Difficulty Describing Feelingsand Difficulty Identifying Feelings) and emotional regulation (DERS—Strategy and DERS—Clarity).
3.2. Cluster analysis
Agglomerative hierarchical cluster analysis based on within-grouplinkage using Squared Euclidean Distance as the distance measure wasperformed on the DERS and TAS subscale scores for the participantswith PNESs only. Difficulties in Emotion Regulation Scale—Impulseand TAS—EOT scores were transformed (logarithm and logarithm fol-lowing reflection, respectively) to ensure that all variables in the analy-sis were normally distributed. Z-transformed values were used toremove the effect of scaling differences between the variables. The ag-glomeration schedule revealed small and relatively uniform increasesin fusion coefficient until the final stage of the analysis, where a com-paratively sharp increase indicated that the optimal solution consistedof two clusters, one consisting of 11 participants and the other 32. Clus-ter 1 was somewhat younger than Cluster 2, but this difference onlyapproached significance (p = .034). There were no significant differ-ences between the clusters in terms of gender, seizure frequency, mar-ital status, highest qualification, or employment status.
Comparisons between the two groups and the sample with epilepsyindicated that the cluster analysis had identified a subgroup of patientswith PNESs (Cluster 1)with high levels of both emotional dysregulationand alexithymia compared with the other subgroups of patients withPNESs (Cluster 2) and controls with epilepsy, who were similar to one
epilepsy on clustering variables for total sample.
U value (and r) ofCluster 1 vs. Cluster 2
U value (and r) ofCluster 1 vs. epilepsy
U value (and r) ofCluster 2 vs. epilepsy
61.5 (0.49) ⁎⁎ 46.0 (0.52)⁎ 350.5 (0.07)56.6 (0.51)⁎⁎ 20.5 (0.67)⁎⁎ 235.0 (0.33)18.5 (0.67)⁎⁎ 16.0 (0.70)⁎⁎ 274.0 (0.24)94.0 (0.35) 62.5 (0.42) 327.0 (0.13)28.5 (0.63)⁎⁎ 18.0 (0.69)⁎⁎ 324.0 (0.13)28.5 (0.63)⁎⁎ 20.5 (0.67)⁎⁎ 307.5 (0.17)34.5 (0.60)⁎⁎ 13.5 (0.71) ⁎⁎ 294.0 (0.20)42.5 (0.57)⁎⁎ 16.0 (0.70)⁎⁎ 307.0 (0.17)95.5 (0.34) 80.0 (0.31) 369.5 (0.03)
Difficulty Identifying Feelings; DDF = Difficulty Describing Feelings; EOT = Externally
Table 4Median (interquartile range) scores for clusters with PNESs and group with epilepsy on nonclustering variables.
Cluster 1 PNES(n = 11)
Cluster 2 PNES(n = 32)
Epilepsy(n = 24)
U value (and r) ofCluster 1 vs. Cluster 2
U value (and r) ofCluster 1 vs. epilepsy
U value (and r) ofCluster 2 vs. epilepsy
Age 26.0 (26.0) 47.0 (22.5) 43.5 (26.0) 100.0 (0.32) 88.0 (0.26) 363.5 (0.05)Seizure frequencya 7.0 (201.75) 14.0 (94.5) 1.5 (40.25) 143.5 (0.03) 90.0 (0.20) 263.0 (0.23)GAD-7 15.0 (7.0) 10.0 (9.5) 4.5 (11.75) 99.0 (0.33) 54.0 (0.47)⁎ 283.5 (0.22)PHQ-9 16.0 (12.0) 10.0 (9.5) 4.5 (8.75) 97.0 (0.34) 29.5 (0.62)⁎⁎ 215.0 (0.37)⁎
SDQ-20 38.0 (16.0) 34.5 (18.0) 24.0 (7.0) 130.0 (0.20) 30.0 (0.61)⁎⁎ 157.5 (0.50)⁎⁎
RSQ—Insecure 41.0 (13.0) 35.5 (10.75) 36.5 (12.75) 96.5 (0.34) 81.5 (0.30) 354.0 (0.07)
Key: GAD = Generalized Anxiety Disorder; PHQ = Patient Health Questionnaire; SDQ = Somatoform Dissociation Questionnaire; RSQ = Relationship Scales Questionnaire.⁎ p ≤ .005.⁎⁎ p ≤ .001.a Number of seizures in last six months; data from one participant in PNES Cluster 1 and two participants in PNES Cluster 2 were missing for this variable.
1 Data available from the authors.
182 R.J. Brown et al. / Epilepsy & Behavior 29 (2013) 178–183
another on these variables (see Table 3). MannWhitney U tests indicat-ed that Cluster 2 and the group with epilepsy were not significantlydifferent from one another on any of the emotional dysregulation oralexithymia variables, although there was a nonsignificant trend forgreater problemswith goal-directed action in Cluster 2. Cluster 1 scoredsignificantly higher than Cluster 2 and the group with epilepsy on allof these variables apart from DERS—Awareness and TAS—ExternallyOriented Thinking.
To further validate and explore the PNES cluster differences, scoreson the remaining questionnaire variables (i.e., those not included inthe cluster analysis) were compared between the two clusters withPNESs and the group with epilepsy (Table 4). There were no significantdifferences between the two clusters, although there were nonsignifi-cant trends for higher anxiety, depression, and attachment insecurityin Cluster 1 compared with Cluster 2. Cluster 1 also scored significantlyhigher than the group with epilepsy on depression, anxiety, and soma-tization. Cluster 2 had significantly higher somatization and depressionscores than the group with epilepsy.
4. Discussion
When considered as a whole, the group with PNESs scored sig-nificantly higher than the group with epilepsy on some aspects of emo-tional dysregulation and alexithymia, but between-groups differenceswere absent on the majority of DERS and TAS-20 subscales. However,cluster analysis revealed two distinct groups of patients with PNESscharacterized by quite different profiles on these measures. Cluster 1patients scored significantly higher than both Cluster 2 and the controlgroup with epilepsy on all DERS and TAS subscales apart from DERS—Awareness and TAS—Externally Oriented Thinking and reported moreanxiety, depression, and somatization. There was also a trend for thisgroup to be younger than Cluster 2. Cluster 2 patients were comparablewith the group with epilepsy on all of the emotional regulation andalexithymia measures, although there was a nonsignificant trend forgreater problems with goal-directed action in the group with PNESs.This cluster of patients with PNESs also hadmore depression and soma-tization than the group with epilepsy.
Taken together, these results are consistentwith previous cluster an-alytic studies in this area, which have identified a cluster of patientswith PNESs with a relatively normal psychological profile and compar-atively low levels of general psychopathology as well as one or moregroups characterized by significant difficulties in multiple areas. Theexistence of two distinct groups of patients with PNESs, only one ofwhich is alexithymic, could account for the equivocal findings of previ-ous alexithymia studies [10,14] that have studied patientswith PNESs asa single group. As such, it underscores the value of using cluster analysiswhen studying the psychological characteristics of patients with PNESsand supports the notion that, in terms of psychopathology, PNESs are aheterogeneous syndrome.
In the current study, patients in Cluster 1 reported difficultieswith most aspects of alexithymia and emotional regulation, including
problems with identifying, accepting, and describing feelings, accessingadaptive strategies for managing upset, performing goal-directed be-haviors, and a tendency for their actions and emotions to feel out ofcontrol. Similar clusters, seemingly characterized by the “undermodu-lation” of affect [20] reminiscent of that seen in borderline personalitydisorder, were also found by Reuber et al. [18] and Uliaszek et al. [19].It may be that PNESs originate in these individuals as amechanism (un-conscious or otherwise) for interrupting intense emotional distress andare thenmaintained through negative reinforcement and a lack of alter-native coping strategies [32]. Perhaps surprisingly, Cluster 1 did notscore significantly higher than the other groups on the lack of aware-ness subscale of the DERS, and they also had comparable scores on theExternally Oriented Thinking subscale of the TAS-20. We believe thatthe most likely explanation for these discrepant results is problemswith the reliability or validity of these particular measures. The reliabil-ity of the Externally Oriented Thinking scale was relatively low (0.59) inthe current study and has been questioned previously [33]. Similarly,the awareness subscale was the only subscale of the DERS not to corre-late significantly with total scores on that measure (r = .26 comparedwith an average correlation of 0.80 between the other subscales andthe total).1
Low scores on measures of personality and psychopathology in asubgroup of patients with PNESs are typically interpreted as evidencefor emotional avoidance and/or a failure to recognize or acknowledgepsychological problems in these individuals, consistent with historicalideas about conversion and the psychodynamics of PNESs [17–19]. Al-though the elevated somatization scores in this cluster seem consistentwith this notion, it is possible that the SDQ-20 is actually capturing as-pects of the PNESs themselves rather than somatizationmore generally.For example, patients with PNESsmay bemore likely to endorse certainSDQ-20 items (e.g., “I am paralyzed for a while”, “I grow stiff for awhile”, “I cannot speak [or onlywith great effort] or I can onlywhisper”,“I cannot see for awhile”, “I cannot hear for a while”) than patients withepilepsy simply because they are more aware during their attacks. Fur-thermore, problemswith identifying, accepting, and describing feelings,which wemight have expected in an emotionally avoidant group, werenot found in Cluster 2. Moreover, this cluster of patients with PNESsreported significantly more depression than the group with epilepsy,suggesting that there is some recognition of emotional problems inthese individuals. While emotional suppression and avoidance remaina likely explanation for such low questionnaire scores, another, poten-tially more parsimonious, interpretation is that psychiatric disturbanceis relatively minimal in these patients. Such an explanation would be aradical departure from traditional theorizing in this area but is consis-tent with a recent model of somatization that describes how PNESsand many other medically unexplained symptoms can arise frommore basic problems with perceptual and behavioral control [34–37].Further research is clearly needed to elucidate these issues, preferably
183R.J. Brown et al. / Epilepsy & Behavior 29 (2013) 178–183
usingmore objective, performance-basedmeasures of emotional avoid-ance and a more general measure of somatization.
There were no differences between the groups in terms of attach-ment insecurity, although there was a nonsignificant trend for Cluster1 to exhibit more attachment insecurity than Cluster 2. Previous re-search on attachment in PNESs is similarly equivocal [38,39], althoughit is unclear whether this is a methodological problem or somethingmore substantive. The RSQ subscales were clearly unreliable in the cur-rent study, and the pooled measure of attachment insecurity used heremay be insufficiently precise to reveal meaningful between-groupsdifferences, particularly if only some types of insecure attachment(and, therefore, emotional dysregulation) [38] are relevant in thiscontext. Theory suggests that the kinds of emotional regulationdeficit seen in Cluster 1 are likely to be associated with problematicattachment [10], suggesting that further research using morereliable measures is warranted.
The current study has a number of limitations including a relativelysmall sample size, a mixed prospective–retrospective recruitmentprocedure, a reliance on self-report measures, and a low return ratefor the postal survey (albeit comparable with previous studies of thissort). Demographic data indicate that the proportion ofmen in our sam-ple with PNESs was slightly higher compared with the large cohortstudy described by Duncan and Oto (35% vs. 23%; [40]) and that theeducational level was also comparatively high, which raises questionsabout the representativeness of the sample. The fact that not all diagno-ses had been confirmed with video-EEG means that this study presentsfindings in a broader group of patients (not only those with frequentseizures who have attacks during video-EEG monitoring) but may beassociated with an increased risk of misclassification. We also did notsystematically collect data concerning traumatic brain injury, other neu-rologic diseases, or comorbid psychiatric disorders, and the distributionof these features across our patient groups is unfortunately unknown.To our knowledge, however, this study represents the first to comparethe emotional regulation profiles of patients with PNESs and epilepsyusing a dedicatedmeasure of the construct and adds to the growing lit-erature suggesting that there are at least two subgroups of patientswithPNESs with distinct psychological profiles.
References
[1] Szaflarski JP, Ficker DM, Cahill WT, Privitera MD. Four-year incidence of psychogenicnonepileptic seizures in adults in Hamilton County, OH. Neurology 2000;55:1561–3.
[2] LaFrance Jr, Alper K, Babcock D, Barry J, Benbadis S, Caplan R, et al. Nonepileptic sei-zures treatment workshop summary. Epilepsy Behav 2006;8(3):451–61.
[3] Carson AJ, Brown R, David AS, Duncan R, Edwards MJ, Goldstein LH, et al. Functional(conversion) neurological symptoms: research since the millennium. J NeurolNeurosurg Psychiatry 2012;83:842–50.
[4] Owczarek K, Jedrzejczak J, Owczarek F. Comparison of personality profiles ofpatients with psychogenic non-epileptic seizures and patients hospitalized forother dissociative disorders. Acta Neuropsychol 2008;6(3):253–67.
[5] Johnson A, Storzbach D, Binder L, Barkhuizen A, Kent Anger W, Salinsky M, et al.MMPI-2 profiles: fibromyalgia patients compared to epileptic and non-epileptic sei-zure patients. Clin Neuropsychol 2010;24(2):220–34.
[6] Kalogjera-Sackellares D, Sackellares J. Personality profiles of patients with pseudo-seizures. Seizure 1997;6(1):1–7.
[7] Drake ME, Pakalnis A, Phillips BB. Neuropsychological and psychiatric correlates ofintractable pseudoseizures. Seizure 1992;1(1):11–3.
[8] Derry P, McLachlan R. The MMPI-2 as an adjunct to the diagnosis of pseudoseizures.Seizure 1996;5(1):35–40.
[9] Breuer J, Freud S. Studies on hysteria. In: Strachey J, Strachey A, editors. The standardedition of the complete psychological works of Sigmund Freud. London: HogarthPress and the Institute of Psycho-Analysis; 1893-1895/1955.
[10] Bewley J, Murphy P, Mallows J, Baker G. Does alexithymia differentiate between pa-tients with nonepileptic seizures, patients with epilepsy, and nonpatient controls?Epilepsy Behav 2005;7(3):430–7.
[11] Waller E, Scheidt CE. Somatoform disorders as disorders of affect regulation: a devel-opment perspective. Int Rev Psychiatry 2006;18(1):13–24.
[12] Bodde N, Bartelet D, Ploegmakers M, Lazeron R, Aldenkamp A, Boon P, et al. MMPI-IIpersonality profiles of patients with psychogenic nonepileptic seizures. EpilepsyBehav 2011;20(4):674–80.
[13] Mazza M, Marca G, Martini A, Scoppetta M, Vollono C, Valenti M, et al. Non-epilepticseizures (NES) are predicted by depressive and dissociative symptoms. Epilepsy Res2009;84(2–3):91–6.
[14] Tojek T, LumleyM, Barkley G,Mahr G, Thomas A. Stress and other psychosocial char-acteristics of patients with psychogenic nonepileptic seizures. Psychosomatics2000;41(3):221–6.
[15] Bodde N, Brooks J, Baker G, Boon P, Hendriksen J, Mulder O, et al. Psychogenicnon-epileptic seizures—definition, etiology, treatment and prognostic issues: a crit-ical review. Seizure 2009;18(8):543–53.
[16] Fiszman A, Alves-Leon S, Nunes R, D'Andrea I, Figueira I. Traumatic events andposttraumatic stress disorder in patients with psychogenic nonepileptic seizures: acritical review. Epilepsy Behav 2004;5(6):818–25.
[17] Cragar D, Berry D, Schmitt F, Fakhoury T. Cluster analysis of normal personality traits inpatients with psychogenic nonepileptic seizures. Epilepsy Behav 2005;6(4):593–600.
[18] ReuberM, Pukrop R, Bauer J, Derfuss R, Elger C. Multidimensional assessment of per-sonality in patients with psychogenic non-epileptic seizures. J Neurol NeurosurgPsychiatry 2004;75(5):743–8.
[19] Uliaszek AA, Prensky E, Baslet G. Emotion regulation profiles in psychogenic non-epileptic seizures. Epilepsy Behav 2012;23:364–9.
[20] Baslet G. Psychogenic non-epileptic seizures: a model of their pathogenic mecha-nism. Seizure 2011;20(1):1–13.
[21] Lacey C, Cook M, Salzberg M. The neurologist, psychogenic nonepileptic seizures,and borderline personality disorder. Epilepsy Behav 2007;11(4):492–8.
[22] Gratz KL, R. L. Multidimensional assessment of emotion regulation and dysregulation:development, factor structure, and initial validation of the Difficulties in EmotionRegulation Scale. J Psychopathol Behav Assess 2004;26:41–54.
[23] Bagby R, Parker JD, Taylor GJ. The twenty-item Toronto Alexithymia Scale: I. Item se-lection and cross-validation of the factor structure. J Psychosom Res 1994;38:23–32.
[24] Bagby R, Taylor GJ, Parker JD. The twenty-item Toronto Alexithymia Scale: II. Con-vergent, discriminant, and concurrent validity. J Psychosom Res 1994;38:33–40.
[25] Griffin DW, Bartholomew K. Models of the self and other: fundamental dimensionsunderlying measures of adult attachment. J Pers Soc Psychol 1994;67:430–45.
[26] Kurdek LA. On being insecure about the assessment of attachment styles. J Soc PersRelatsh 2002;19:811–34.
[27] Spitzer RL, Kroenke K,Williams JB, Löwe B. A brief measure for assessing generalizedanxiety disorder: the GAD-7. Arch Intern Med 2006;166:1092–7.
[28] Kroenke K, Spitzer RL, Williams JB. The PHQ-15: validity of a new measure for eval-uating the severity of somatic symptoms. Psychosom Med 2002;64:258–66.
[29] Nijenhuis ER, Spinhoven P, Van Dyck R, VanDer Hart O. The development and psycho-metric characteristics of the Somatoform Dissociation Questionnaire (SDQ-20). J NervMent Dis 1996;184:688–94.
[30] Nijenhuis ER. Somatoform dissociation: phenomena, measurement, & theoreticalissues. New York: Norton; 2004.
[31] Field A. Discovering statistics using SPSS. 2nd ed. London: Sage Publications Ltd.;2005.
[32] Goldstein L, Deale A, Mitchell-O'Malley S, Toone B, Mellers J. An evaluation of cogni-tive behavioral therapy as a treatment for dissociative seizures: a pilot study. CognBehav Neurol 2004;17(1):41–9.
[33] Kooiman CG, Spinhoven P, Trijsburg RW. The assessment of alexithymia: a criticalreview of the literature and a psychometric study of the Toronto AlexithymiaScale-20. J Psychosom Res 2002;53:1083–90.
[34] Brown RJ. Psychological mechanisms of medically unexplained symptoms: an inte-grative conceptual model. Psychol Bull 2004;130:793–812.
[35] Brown RJ. Different types of “dissociation” have different psychological mechanisms.J Trauma Dissociation 2006;7:7–28.
[36] Brown RJ. Dissociation and somatoform disorders. In: Kennedy F, Kennerley H,Pearson D, editors. Dissociation and cognitive therapy. London: Routledge; 2013.p. 133–47.
[37] Brown RJ, Syed TU, Benbadis S, LaFrance JrWC, ReuberM. Psychogenic non-epilepticseizures. Epilepsy Behav 2011;22:85–93.
[38] Holman N, Kirkby A, Duncan S, Brown RJ. Adult attachment style and childhood in-terpersonal trauma in non-epileptic attack disorder. Epilepsy Res 2008;79(1):84–9.
[39] Lally N, SpenceW, McCusker C, Craig J, Morrow J. Psychological processes and histo-ries associated with nonepileptic versus epileptic seizure presentations. EpilepsyBehav 2010;17(3):360–5.
[40] Duncan R, OtoM. Psychogenic nonepileptic seizures in patientswith learning disability:comparison with patients with no learning disability. Epilepsy Behav 2008;12:183–6.
