+Model
ARTICLE IN PRESS
EPIRES-5215; No. of Pages 11 Epilepsy Research (2014) xxx, xxx—xxx
journal homepage: www.elsevier.com/locate/epilepsyres
Biopsychosocial predictors of psychogenic non-epileptic seizures John O. Elliott a,b,∗, Christine Charyton c a
Ohio Health Riverside Methodist Hospital, 3535 Olentangy River Road, Columbus, OH 43214, United States Ohio State University, College of Social Work, 1947 Stillman Hall, Columbus, OH 43210, United States c Department of Neurology, Ohio State University Wexner Medical Center, 395W, 12th Avenue, 7th Floor, Columbus, OH 43210, United States b
Received 29 May 2014; received in revised form 22 August 2014; accepted 6 September 2014
KEYWORDS Non-epileptic seizures; Pseudoseizures; Biological; Psychological; Social
Summary Background: Previous studies have identified numerous biological, psychological and social characteristics of persons with psychogenic non-epileptic seizures (PNES) however the strength of many of these factors have not been evaluated to determine which are predictive of the diagnosis compared to those that may only be stereotypes with limited clinical utility. Method: A retrospective chart review of persons admitted to our epilepsy monitoring unit over a 6-year period was conducted to examine predictors of a video-EEG confirmed PNES diagnosis. Results: A total of 689 patients had events leading to a diagnosis, 47% (n = 324) with PNES only, 12% (n = 84) with PNES & Epilepsy and 41% (n = 281) with Epilepsy only. Five biological predictors of a PNES only diagnosis were found; number of years with events (OR = 1.10), history of head injury (OR = 1.91), asthma (OR = 2.94), gastro-esophageal reflux disease (OR = 1.72) and pain (OR = 2.25). One psychological predictor; anxiety (OR = 1.72) and two social predictors; being married (OR = 1.81) and history of physical/sexual abuse (OR = 3.35). Two significant biological predictors of a PNES & Epilepsy diagnosis were found; migraine (OR = 1.83) and gastroesophageal reflux disease (OR = 2.17). Conclusions: Our findings support the importance of considering the biopsychosocial model for the diagnosis and treatment of PNES or PNES with concomitant epilepsy. © 2014 Elsevier B.V. All rights reserved.
Introduction ∗ Corresponding author at: Ohio Health Riverside Methodist Hospital, 3535 Olentangy River Road, Columbus, OH 43214, United States. Tel.: +1 614 566 3643; fax: +1 614 566 1073. E-mail addresses: [email protected], [email protected] (J.O. Elliott), [email protected], [email protected] (C. Charyton).
Psychogenic non-epileptic seizures (PNES) are a common type of non-epileptic event that clinically resemble a seizure but are psychologically based. Prevalence of PNES has been estimated between 1 per 50,000 and 1 per 3000 (Benbadis and Allen Hauser, 2000) with estimated annual costs of PNES, misdiagnosed as epilepsy ranging between $650
http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003 0920-1211/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
2
J.O. Elliott, C. Charyton
million and $4 billion (Nowack, 1997). In addition lifetime cost of diagnostic tests, procedures and medications for persons with PNES have been estimated at $100,000 (LaFrance and Benbadis, 2006). Of persons with epilepsy, between 5 and 20% are thought to have PNES (LaFrance and Benbadis, 2006). Diagnosis and management is complicated by the difficulty in distinguishing PNES from epilepsy (Chung et al., 2006). The current gold standard is based on a lack of ictal electroencephalogram (EEG) activities during the event, via a continuous video-EEG study (LaFrance and Benbadis, 2006). Interest in clinical signs that distinguish PNES from epilepsy have primarily focused on physical signs during events such as motor features, closed eyes, tongue biting and urinary incontinence (Avbersek and Sisodiya, 2010) however a more comprehensive approach such as one proposed by the biopsychosocial model may help identify more robust predictors of a PNES diagnosis. In conceptualizing the biopsychosocial model, George Engel sought to use General Systems theory to improve the understanding of the bi-directional relationship between the body and mind, as well as to reconcile the dualist concepts that separate health and disease (Engel, 1977). In General Systems theory no system exists in isolation and every system is influenced by its environmental configuration (Richter, 1999). In the medical domain, Engel felt General Systems theory provided a conceptual approach for studying the biopsychosocial approach but also for studying disease and medical care as interrelated processes (Engel, 1977). By contrast in the traditional biological—biomedical approach, the causes, diagnosis, prognosis, treatment and outcomes of disease are largely based on physical or somatic components (McCollum and Pincus, 2009) where the focus is on etiologic agents, pathological processes and biological, physiological or clinical outcomes (Wilson and Cleary, 1995). Furthermore, the biological—biomedical approach separates the mind and body in the causation of disease and this has lead to health outcomes that are primarily driven by health professionals and the medical system with little input from the individual patient (McCollum and Pincus, 2009). Overall, this focus on pathology, to the exclusion of processes of health and recovery, has resulted in a fragmented and incomplete understanding of the person and their disease (Davidson and Strauss, 1995). In the present PNES literature many disparate factors have been examined, oftentimes in isolation from interacting biological, psychological or social domains. The biological—biomedical factors previously established include female predominance (Duncan and Oto, 2008), antecedent mild head injuries (Barry et al., 1998; Westbrook et al., 1998; Mökleby et al., 2002) and a later onset of events (Brown et al., 1991). Studies have shown that an early diagnosis of PNES results in a better prognosis (Walczak et al., 1995), yet a delay of more than seven years is often found before an official diagnosis is made by video-EEG (Reuber et al., 2002). Persons with PNES also present with a large number of somatic comorbidities. For example previous research found chronic pain (Fleisher et al., 2002), headaches (Ettinger et al., 1999a), sleep disturbances (Benbadis, 2005; Zhang et al., 2009), asthma (de Wet et al., 2003) and obesity
(Marquez et al., 2004) are more common in persons with PNES. Investigations examining other somatic comorbidities (hypertension, heart disease, lung disease and ulcers) suggest additional comorbidities, but these studies have been limited by very small sample sizes (Tojek et al., 2000) or have only examined the association between PNES and one condition (de Wet et al., 2003; Marquez et al., 2004). In terms of the psychological domain, persons with PNES often present with significant psychological comorbidities in comparison to populations with or without epilepsy (Goldstein et al., 2000; Binzer et al., 2004). Previous studies report higher rates of post-traumatic stress disorder (PTSD), a higher prevalence of somatoform disorders and anxiety in persons with PNES (as well as those with both PNES and concomitant epilepsy), compared to those with epilepsy only (Kuyk et al., 2003; Owczarek, 2003). In terms of the social domain, a history of physical or sexual abuse has been reported in 11—84% of cases (Bowman and Markand, 1996; Dikel et al., 2003; LaFrance and Syc, 2009). In addition, up to 50% of persons with PNES are disabled — a level equal to those with epilepsy (Krawetz et al., 2001) which highlights the severity of the condition on overall well-being. A continued focus on the biological—biomedical aspects of disease (including a purely ‘‘psychiatric’’ view of poor mental health rooted solely in the use of psychotropic medications for symptomatic treatment) works to further perpetuate psychosocial disparities in persons with epilepsy and/or PNES. The more recent literature has suggested epilepsy treatment focus on broad strategies that addresses the needs of the whole person (Kramer, 2003) by taking into account social, vocational and psychological function (Sander, 2005) however this approach has not been examined in persons with PNES. In many clinical populations the biopsychosocial model has mostly remained an unmet challenge for research (understanding the etiology and development of disease or disorder) and practice (diagnosis and treatment) (Kiesler, 1999). The current PNES literature highlights many unique characteristics however these factors have not been incorporated in a larger understanding of the whole person. The purpose of this research was to gain a better understanding of the unique biological, psychological and social factors associated with a continuous video-EEG confirmed diagnosis of PNES.
Materials and methods A retrospective chart review was conducted of patients admitted to the Ohio State University Wexner Medical Center epilepsy monitoring unit (EMU) data over a 6-year period (2002—2007). Participants were identified via administrative billing records using the current procedural terminology (CPT) code 95951 for video-EEG monitoring. Data were obtained via a review of electronic medical records (demographic data used for billing purposes, inpatient history and physical exam assessments completed as part of the EMU admission, visit notes and hospital discharge summaries), as well as outpatient medical records (history and physical, EEG reports and correspondence with referring physicians).
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
Biopsychosocial predictors of psychogenic non-epileptic seizures The diagnosis of PNES was made by the attending epileptologist based on the absence of electrographic changes associated with a typical event. Events of impaired consciousness or generalized shaking were considered PNES if they were not associated with electrographic changes. Evidence of epilepsy was defined by unequivocal epileptiform discharges (sharp waves, spike, polyspike or spike-wave complexes) in any previous EMU visit or a routine EEG conducted at our center. Benign EEG variants were not considered epilepsy. For persons with PNES with concomitant epilepsy, a non-epileptic event was captured during a videoEEG monitoring session in addition to epileptiform activity or a captured seizure. Use of induction methods was limited to hyperventilation, photic stimulation, sleep deprivation and tapering/discontinuation of medications. Conceptually the data were categorized by biological, psychological and social domains that are inter-related in treatment of the entire person. We define each domain separately to address how they can be integrated into the diagnosis and treatment of PNES. For this study, the biological—biomedical domain was defined to include: age, gender, race/ethnicity, clinical factors: age of onset, number of years with events/seizures (the difference between age of onset and definitive videoEEG diagnosis), history of head injury, body mass index and a diagnosis of intellectual disability. These factors were defined and coded based on administrative data used for medical billing purposes (age, gender and race/ethnicity) as well as from the history and physical completed at the patient’s initial visit. Age is typically viewed as a significant factor in medical treatment especially in the medical specialties of pediatrics and geriatric medicine. Age at epilepsy diagnosis can have a profound long-term effect across the lifespan (Tidman et al., 2003). In addition, cognitive problems from seizures and/or their treatment can have problematic effects on the changing psychological and medical contexts and life adjustment to epilepsy as an adult (Velissaris et al., 2009). Due the large gap between onset of events and definitive diagnosis via video-EEG these same issues apply to persons with a PNES diagnosis. Gender differences are also well recognized in the biological—biomedical aspects of clinical care. Several studies suggest females are more likely to be seen by a neurologist vs. a general practitioner (Jette et al., 2008; Mattsson et al., 2010). The Institute of Medicine report titled Exploring the Biological Contributions to Human Health: Does Sex Matter? highlights the importance of understanding gender differences at the societal level based on individual behaviors, lifestyle and surroundings (Institute of Medicine Committee on Understanding the Biology of Sex and Gender Differences et al., 2001). The biological—biomedical domain also included comorbid somatic disease states that are predominately diagnosed and/or treated via medication and/or surgery approaches: migraines, stroke, cardiovascular disease, high cholesterol, hypertension, diabetes, hypothyroidism, asthma, gastro-esophageal reflux disease (GERD), irritable bowel syndrome (IBS), cancer, arthritis, chronic pain (includes refractory back pain as well as pain syndromes, such as fibromyalgia and reflex sympathetic dystrophy) and insomnia. For these conditions patients may have a history of
3
the disease (i.e., stroke, cancer, IBS) and/or they were active conditions: defined as being currently treated with FDA approved medications (acute, preventative or on a long-term basis), off-label use medications (i.e., antihypertensives and antidepressants used for migraine prevention or trazodone for insomnia) and/or non-medication approaches (i.e., dietary management for diabetes). The comorbid somatic disease states were identified in the EMU preadmission history and physical, the discharge summary report from the EMU admission and/or the clinician’s notes of current medical problems just prior to the EMU admission. The psychological domain was conceptualized as conditions patients had a documented history of and/or psychological/psychiatric conditions that were currently active: defined as being currently treated with FDA approved medications (i.e., antidepressants, anxiolytics, mood stabilizing AEDs, lithium, antipsychotics), off-label medications (antihypertensives for anxiety) and/or non-medication approaches (i.e., psychological therapy). These conditions included: depression, anxiety, bipolar disorder, posttraumatic stress disorder, personality disorder, schizophrenia and history of suicidality or suicide attempts. The psychological/psychiatric conditions were also identified in the EMU pre-admission history and physical, the discharge summary report from the EMU admission and/or clinician notes of current medical problems just prior to the EMU admission. The social domain was conceptualized as factors influenced by the social environment these include: marital status, disability, history of abuse (physical and/or sexual), current smoking status, alcohol use and illicit drug use. These factors were documented based on administrative data used for medical billing (marital status and disability) or through documentation in the medical record by the clinician as part of the history or description of the patient’s current problem statement. Statistical analyses were conducted using SPSS v 19. Three groups based on the continuous video-EEG confirmed diagnosis are presented in the univariate analyses: PNES only, PNES & Epilepsy and Epilepsy only (dependent variables). Next, univariate and multivariate logistic regression analyses were utilized to determine the biological, psychological and social predictors of PNES only and PNES & Epilepsy (independent variables). In these analyses persons in the Epilepsy only group were the comparator. In the univariate models we examined the predictive ability of each independent variable via odds ratios (ORs) and 95% CIs. ORs and 95% CIs provide a parsimonious assessment of both the strength and reliability of the associations. Last, we conducted two multivariate logistic regression analyses which included all independent variables that had significant predictive value in the univariate model at the p < 0.05 level. Since the univariate analyses which are unable to assess whether an association is direct, or is due to the covariance with another variable (Duncan and Oto, 2008) the multivariate model incorporates the correlations and interactions between all the variables in the model, i.e., the relationship between female gender can be examined simultaneously as both an individual predictive factor and as a control variable.
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
4
J.O. Elliott, C. Charyton
Results A total of 689 patients admitted to our EMU had events leading to a diagnosis, 47% (n = 324) participants had their typical events diagnosed as PNES only, 12% (n = 84) had their typical events diagnosed as PNES & Epilepsy and 41% (n = 281) were diagnosed with Epilepsy only by video-EEG, see Table 1.
PNES only There were significant differences between the PNES only and the Epilepsy only groups based on biological—biomedical factors (gender, age at onset of events, number of years having events or seizures, history of head injury, BMI (obesity) and all somatic conditions except stroke, hypothyroidism and cancer. There were also differences in psychological factors (all except history of suicidality/suicide attempts) and social factors (marital status, disability status, history of physical or sexual abuse and smoking status), see Table 1. Univariate logistic regression analyses revealed 24 significant predictors of PNES only, see Table 2. These included biological—biomedical factors (female gender, years with events or seizures, age of onset, history of head injury, a diagnosis of intellectual disability, migraine, cardiovascular disease, diabetes, asthma, GERD, IBS, arthritis, chronic pain and insomnia), psychological factors (depression, anxiety, bipolar disorder, PTSD, personality disorder and schizophrenia) and social factors (marital status and history of physical/sexual abuse). Once these factors were entered into a multivariable logistic regression model, eight predictors of the PNES only group remained significant. Five were biological—biomedical: numbers of years with events OR = 1.10 (95% CI: 1.07—1.13), history of head injury OR = 1.91 (95% CI: 1.21—3.01), asthma OR = 2.94 (95% CI: 1.69—5.12), GERD OR = 1.72 (95% CI: 1.05—2.82) and chronic pain OR = 2.25 (95%CI: 1.31—3.86), one psychological factor: anxiety OR = 1.72 (95% CI: 1.09—2.71) and two social factors: being married OR = 1.81 (95% CI: 1.15—2.85) and history of physical/sexual abuse OR = 3.35 (95% CI: 1.23—9.10), see Table 2.
PNES & Epilepsy There were significant differences between the PNES & Epilepsy and the Epilepsy only groups based on biological—biomedical factors (female gender, migraines, asthma, GERD and chronic pain). There were also differences in psychological factors (depression, anxiety, bipolar disorder and personality disorder) and social factors (disability status, history of physical or sexual abuse and smoking status), see Table 1. Univariate logistic regression analyses revealed eleven significant predictors of PNES & Epilepsy, see Table 3. These included biological—biomedical factors (female gender, migraine, asthma, GERD and pain), psychological factors (depression, anxiety, bipolar disorder and personality disorder) and social factors (disability status, history of physical/sexual abuse and smoking status). Once these factors were entered into a multivariable logistic regression model two biological—biomedical factors
migraine OR = 1.83 (95% CI: 1.04—3.23) and GERD OR = 2.17 (95% CI: 1.18—3.99) remained significant predictors of a PNES & Epilepsy diagnosis, see Table 3.
Discussion The purpose of this research was to gain a better understanding of the biological, psychological and social factors associated with a continuous video-EEG confirmed diagnosis of PNES. The current study focused on a more comprehensive look at PNES than many previous investigations. The multivariate models provide more insight than individual factors or stereotyped characteristics (i.e., female predominance) when a diagnosis of PNES is suspected. Our findings support the importance of understanding the complex interaction of biological—biomedical factors (number of years with events, previous head injuries and certain medical diagnoses such as migraines, asthma, GERD and chronic pain) as well as psychological (anxiety) and social factors (marital status and physical/sexual abuse) in the PNES population.
Biological In our study the number of years with events was predictive of a PNES only diagnosis. In the logistic model this translates to a 10% increased odds of a PNES diagnosis for each year of having events before video-EEG diagnosis. This supports the importance of receiving a clear and positive explanation of the PNES diagnosis based on video-EEG as early as possible (Kerr et al., 2011) which by itself may lead to a significant reduction and/or elimination of PNES events in some patients (McKenzie et al., 2010). One recent Canadian study found significant reductions in total emergency room visits, emergency room visits for neurological causes in the 2 years following PNES diagnosis which was sustained over a four year period (Jirsch et al., 2011). Prior research suggests that traumatic experiences play an important role in the development and expression of PNES (Fleisher et al., 2002; Selkirk et al., 2008). In particular prior head injury has been associated with the development of PNES (Westbrook et al., 1998; LaFrance and Syc, 2009). The current study provides additional support for inquiring about previous head injuries. We found higher rates of asthma in both the PNES and the PNES & Epilepsy groups. These findings are consistent with a previous investigation which suggested the association between asthma and PNES may be accounted by a combination of somatization, anxiety hyperventilation and dissociation (de Wet et al., 2003). That study was limited by the inclusion of non-confirmed video-EEG cases and made comparisons with a group of control patients with psychosis. The current study suggests asthma and anxiety are significant in the larger picture when many other factors have been accounted for. A recent paper found a very similar profile of comorbidities in PNES (Dixit et al., 2013), however persons with PNES & Epilepsy were excluded in that study. We also found GERD was associated with a PNES diagnosis. Previous research reported a higher rate of ulcers in persons with PNES compared to those with epilepsy (Tojek et al., 2000) and greater use of gastric reflux medications in those with PNES (Hantke et al., 2007). The literature also
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
Biopsychosocial predictors of psychogenic non-epileptic seizures Table 1
5
Univariate analysis of biopsychosocial characteristics.
Characteristic
PNES only (n = 324) Mean(sd)/%(n)
PNES & Epilepsy (n = 84) Mean(sd)/%(n)
Epilepsy only (n = 281) Mean(sd)/%(n)
37.3 (12.2) 71 (230)***
37.1 (12.5) 70 (59)**
38.6 (13.5) 54 (151)
89 (287) 11 (37) 28.2 (15.1)*** 9.1 (11.1)*** 37 (119)***
87 (73) 13 (11) 16.3 (15.0) 20.8 (14.9) 26 (22)
87 (244) 13 (37) 18.1 (16.7) 20.5 (14.9) 25 (69)
2 (6) 30 (97) 19 (62)** 49 (159)*** 7 (22)***
6 (5) 33 (28) 20 (17) 41 (34) 25 (21)
3 (7) 31 (88) 26 (74) 40 (112) 18 (51)
53 (171)*** 5 (17) 10 (31)** 19 (62) 32 (104) 17 (55)*** 11 (37) 33 (107)*** 41 (133)*** 12 (40)*** 6 (19) 16 (53)* 36 (117)*** 31 (100)***
51 (43)*** 7 (6) 4 (3) 12 (10) 24 (20) 11 (9) 14 (12) 23 (19) 45 (38)*** 7 (6) 4 (3) 13 (11) 27 (23)** 27 (23)
30 (85) 7 (20) 4 (11) 17 (47) 32 (89) 6 (17) 14 (38) 13 (37) 20 (57) 3 (9) 6 (16) 11 (30) 13 (35) 19 (54)
Psychological Depression Anxiety Bipolar disorder Post-traumatic stress disorder Personality disorder Schizophrenia History of suicidality/suicide attempts
67 (218)*** 50 (161)*** 17 (55)** 7 (21)*** 7 (21)* 9 (30)* 8 (25)
68 (57)*** 49 (41)*** 23 (19)** 5 (4) 8 (7)* 8 (7) 10 (8)
45 (127) 25 (70) 9 (25) 1 (3) 2 (6) 5 (13) 5 (13)
Social Marital status Single Married Separated Divorced Widowed Disabled History of abuse (physical/sexual) Current smoker Current alcohol use Current illicit drug use
40 (27)* 40 (131)* 4 (13) 14 (45) 2 (8) 39 (125) 12 (39)*** 45 (147) 25 (80) 13 (43)
51 (43) 29 (24) 4 (3) 13 (11) 4 (3) 56 (47)* 12 (10)** 50 (42)* 25 (21) 14 (12)
51 (144) 31 (86) 1 (4) 15 (42) 2 (5) 41 (115) 3 (8) 38 (106) 29 (82) 13 (36)
Biological Age Female gender Race/ethnicity Caucasian Non-Caucasian Age of onset Number of years with events or seizures History of head injury BMI Underweight Normal Overweight Obese Intellectual disability Somatic conditions Migraines History of stroke Cardiovascular disease High cholesterol Hypertension Diabetes Hypothyroidism Asthma Gastro-esophageal reflux disease Irritable bowel syndrome Cancer Arthritis Chronic pain Insomnia
Notes: Comparisons are made using Epilepsy only as the reference group. For continuous data one-way ANOVA was conducted or Wilcoxon rank sum tests were conducted. For categorical data 2 analyses were conducted. PNES = Psychogenic non-epileptic seizures. * p ≤ 0.05. ** p ≤ 0.01. *** p ≤ 0.001.
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
6
J.O. Elliott, C. Charyton Table 2
Logistic regression of persons with diagnosed events: PNES only vs. Epilepsy only.
Characteristic
Univariate
Multivariate
Odds ratio (95% CI)
p-value
Odds ratio (95% CI)
p-value
Biological Female gender Number of years with events or seizures Age of onset History of head injury Overweight/obese Intellectual disability Somatic conditions Migraine Cardiovascular disease Diabetes Asthma Gastro-esophageal reflux disease Irritable bowel syndrome Arthritis Chronic pain Insomnia
2.11 1.07 0.96 1.78 0.91 0.33
(1.51—2.95) (1.05—1.09) (0.95—0.97) (1.25—2.54) (0.65—1.28) (0.19—0.56)
ARTICLE IN PRESS
EPIRES-5215; No. of Pages 11 Epilepsy Research (2014) xxx, xxx—xxx
journal homepage: www.elsevier.com/locate/epilepsyres
Biopsychosocial predictors of psychogenic non-epileptic seizures John O. Elliott a,b,∗, Christine Charyton c a
Ohio Health Riverside Methodist Hospital, 3535 Olentangy River Road, Columbus, OH 43214, United States Ohio State University, College of Social Work, 1947 Stillman Hall, Columbus, OH 43210, United States c Department of Neurology, Ohio State University Wexner Medical Center, 395W, 12th Avenue, 7th Floor, Columbus, OH 43210, United States b
Received 29 May 2014; received in revised form 22 August 2014; accepted 6 September 2014
KEYWORDS Non-epileptic seizures; Pseudoseizures; Biological; Psychological; Social
Summary Background: Previous studies have identified numerous biological, psychological and social characteristics of persons with psychogenic non-epileptic seizures (PNES) however the strength of many of these factors have not been evaluated to determine which are predictive of the diagnosis compared to those that may only be stereotypes with limited clinical utility. Method: A retrospective chart review of persons admitted to our epilepsy monitoring unit over a 6-year period was conducted to examine predictors of a video-EEG confirmed PNES diagnosis. Results: A total of 689 patients had events leading to a diagnosis, 47% (n = 324) with PNES only, 12% (n = 84) with PNES & Epilepsy and 41% (n = 281) with Epilepsy only. Five biological predictors of a PNES only diagnosis were found; number of years with events (OR = 1.10), history of head injury (OR = 1.91), asthma (OR = 2.94), gastro-esophageal reflux disease (OR = 1.72) and pain (OR = 2.25). One psychological predictor; anxiety (OR = 1.72) and two social predictors; being married (OR = 1.81) and history of physical/sexual abuse (OR = 3.35). Two significant biological predictors of a PNES & Epilepsy diagnosis were found; migraine (OR = 1.83) and gastroesophageal reflux disease (OR = 2.17). Conclusions: Our findings support the importance of considering the biopsychosocial model for the diagnosis and treatment of PNES or PNES with concomitant epilepsy. © 2014 Elsevier B.V. All rights reserved.
Introduction ∗ Corresponding author at: Ohio Health Riverside Methodist Hospital, 3535 Olentangy River Road, Columbus, OH 43214, United States. Tel.: +1 614 566 3643; fax: +1 614 566 1073. E-mail addresses: [email protected], [email protected] (J.O. Elliott), [email protected], [email protected] (C. Charyton).
Psychogenic non-epileptic seizures (PNES) are a common type of non-epileptic event that clinically resemble a seizure but are psychologically based. Prevalence of PNES has been estimated between 1 per 50,000 and 1 per 3000 (Benbadis and Allen Hauser, 2000) with estimated annual costs of PNES, misdiagnosed as epilepsy ranging between $650
http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003 0920-1211/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
2
J.O. Elliott, C. Charyton
million and $4 billion (Nowack, 1997). In addition lifetime cost of diagnostic tests, procedures and medications for persons with PNES have been estimated at $100,000 (LaFrance and Benbadis, 2006). Of persons with epilepsy, between 5 and 20% are thought to have PNES (LaFrance and Benbadis, 2006). Diagnosis and management is complicated by the difficulty in distinguishing PNES from epilepsy (Chung et al., 2006). The current gold standard is based on a lack of ictal electroencephalogram (EEG) activities during the event, via a continuous video-EEG study (LaFrance and Benbadis, 2006). Interest in clinical signs that distinguish PNES from epilepsy have primarily focused on physical signs during events such as motor features, closed eyes, tongue biting and urinary incontinence (Avbersek and Sisodiya, 2010) however a more comprehensive approach such as one proposed by the biopsychosocial model may help identify more robust predictors of a PNES diagnosis. In conceptualizing the biopsychosocial model, George Engel sought to use General Systems theory to improve the understanding of the bi-directional relationship between the body and mind, as well as to reconcile the dualist concepts that separate health and disease (Engel, 1977). In General Systems theory no system exists in isolation and every system is influenced by its environmental configuration (Richter, 1999). In the medical domain, Engel felt General Systems theory provided a conceptual approach for studying the biopsychosocial approach but also for studying disease and medical care as interrelated processes (Engel, 1977). By contrast in the traditional biological—biomedical approach, the causes, diagnosis, prognosis, treatment and outcomes of disease are largely based on physical or somatic components (McCollum and Pincus, 2009) where the focus is on etiologic agents, pathological processes and biological, physiological or clinical outcomes (Wilson and Cleary, 1995). Furthermore, the biological—biomedical approach separates the mind and body in the causation of disease and this has lead to health outcomes that are primarily driven by health professionals and the medical system with little input from the individual patient (McCollum and Pincus, 2009). Overall, this focus on pathology, to the exclusion of processes of health and recovery, has resulted in a fragmented and incomplete understanding of the person and their disease (Davidson and Strauss, 1995). In the present PNES literature many disparate factors have been examined, oftentimes in isolation from interacting biological, psychological or social domains. The biological—biomedical factors previously established include female predominance (Duncan and Oto, 2008), antecedent mild head injuries (Barry et al., 1998; Westbrook et al., 1998; Mökleby et al., 2002) and a later onset of events (Brown et al., 1991). Studies have shown that an early diagnosis of PNES results in a better prognosis (Walczak et al., 1995), yet a delay of more than seven years is often found before an official diagnosis is made by video-EEG (Reuber et al., 2002). Persons with PNES also present with a large number of somatic comorbidities. For example previous research found chronic pain (Fleisher et al., 2002), headaches (Ettinger et al., 1999a), sleep disturbances (Benbadis, 2005; Zhang et al., 2009), asthma (de Wet et al., 2003) and obesity
(Marquez et al., 2004) are more common in persons with PNES. Investigations examining other somatic comorbidities (hypertension, heart disease, lung disease and ulcers) suggest additional comorbidities, but these studies have been limited by very small sample sizes (Tojek et al., 2000) or have only examined the association between PNES and one condition (de Wet et al., 2003; Marquez et al., 2004). In terms of the psychological domain, persons with PNES often present with significant psychological comorbidities in comparison to populations with or without epilepsy (Goldstein et al., 2000; Binzer et al., 2004). Previous studies report higher rates of post-traumatic stress disorder (PTSD), a higher prevalence of somatoform disorders and anxiety in persons with PNES (as well as those with both PNES and concomitant epilepsy), compared to those with epilepsy only (Kuyk et al., 2003; Owczarek, 2003). In terms of the social domain, a history of physical or sexual abuse has been reported in 11—84% of cases (Bowman and Markand, 1996; Dikel et al., 2003; LaFrance and Syc, 2009). In addition, up to 50% of persons with PNES are disabled — a level equal to those with epilepsy (Krawetz et al., 2001) which highlights the severity of the condition on overall well-being. A continued focus on the biological—biomedical aspects of disease (including a purely ‘‘psychiatric’’ view of poor mental health rooted solely in the use of psychotropic medications for symptomatic treatment) works to further perpetuate psychosocial disparities in persons with epilepsy and/or PNES. The more recent literature has suggested epilepsy treatment focus on broad strategies that addresses the needs of the whole person (Kramer, 2003) by taking into account social, vocational and psychological function (Sander, 2005) however this approach has not been examined in persons with PNES. In many clinical populations the biopsychosocial model has mostly remained an unmet challenge for research (understanding the etiology and development of disease or disorder) and practice (diagnosis and treatment) (Kiesler, 1999). The current PNES literature highlights many unique characteristics however these factors have not been incorporated in a larger understanding of the whole person. The purpose of this research was to gain a better understanding of the unique biological, psychological and social factors associated with a continuous video-EEG confirmed diagnosis of PNES.
Materials and methods A retrospective chart review was conducted of patients admitted to the Ohio State University Wexner Medical Center epilepsy monitoring unit (EMU) data over a 6-year period (2002—2007). Participants were identified via administrative billing records using the current procedural terminology (CPT) code 95951 for video-EEG monitoring. Data were obtained via a review of electronic medical records (demographic data used for billing purposes, inpatient history and physical exam assessments completed as part of the EMU admission, visit notes and hospital discharge summaries), as well as outpatient medical records (history and physical, EEG reports and correspondence with referring physicians).
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
Biopsychosocial predictors of psychogenic non-epileptic seizures The diagnosis of PNES was made by the attending epileptologist based on the absence of electrographic changes associated with a typical event. Events of impaired consciousness or generalized shaking were considered PNES if they were not associated with electrographic changes. Evidence of epilepsy was defined by unequivocal epileptiform discharges (sharp waves, spike, polyspike or spike-wave complexes) in any previous EMU visit or a routine EEG conducted at our center. Benign EEG variants were not considered epilepsy. For persons with PNES with concomitant epilepsy, a non-epileptic event was captured during a videoEEG monitoring session in addition to epileptiform activity or a captured seizure. Use of induction methods was limited to hyperventilation, photic stimulation, sleep deprivation and tapering/discontinuation of medications. Conceptually the data were categorized by biological, psychological and social domains that are inter-related in treatment of the entire person. We define each domain separately to address how they can be integrated into the diagnosis and treatment of PNES. For this study, the biological—biomedical domain was defined to include: age, gender, race/ethnicity, clinical factors: age of onset, number of years with events/seizures (the difference between age of onset and definitive videoEEG diagnosis), history of head injury, body mass index and a diagnosis of intellectual disability. These factors were defined and coded based on administrative data used for medical billing purposes (age, gender and race/ethnicity) as well as from the history and physical completed at the patient’s initial visit. Age is typically viewed as a significant factor in medical treatment especially in the medical specialties of pediatrics and geriatric medicine. Age at epilepsy diagnosis can have a profound long-term effect across the lifespan (Tidman et al., 2003). In addition, cognitive problems from seizures and/or their treatment can have problematic effects on the changing psychological and medical contexts and life adjustment to epilepsy as an adult (Velissaris et al., 2009). Due the large gap between onset of events and definitive diagnosis via video-EEG these same issues apply to persons with a PNES diagnosis. Gender differences are also well recognized in the biological—biomedical aspects of clinical care. Several studies suggest females are more likely to be seen by a neurologist vs. a general practitioner (Jette et al., 2008; Mattsson et al., 2010). The Institute of Medicine report titled Exploring the Biological Contributions to Human Health: Does Sex Matter? highlights the importance of understanding gender differences at the societal level based on individual behaviors, lifestyle and surroundings (Institute of Medicine Committee on Understanding the Biology of Sex and Gender Differences et al., 2001). The biological—biomedical domain also included comorbid somatic disease states that are predominately diagnosed and/or treated via medication and/or surgery approaches: migraines, stroke, cardiovascular disease, high cholesterol, hypertension, diabetes, hypothyroidism, asthma, gastro-esophageal reflux disease (GERD), irritable bowel syndrome (IBS), cancer, arthritis, chronic pain (includes refractory back pain as well as pain syndromes, such as fibromyalgia and reflex sympathetic dystrophy) and insomnia. For these conditions patients may have a history of
3
the disease (i.e., stroke, cancer, IBS) and/or they were active conditions: defined as being currently treated with FDA approved medications (acute, preventative or on a long-term basis), off-label use medications (i.e., antihypertensives and antidepressants used for migraine prevention or trazodone for insomnia) and/or non-medication approaches (i.e., dietary management for diabetes). The comorbid somatic disease states were identified in the EMU preadmission history and physical, the discharge summary report from the EMU admission and/or the clinician’s notes of current medical problems just prior to the EMU admission. The psychological domain was conceptualized as conditions patients had a documented history of and/or psychological/psychiatric conditions that were currently active: defined as being currently treated with FDA approved medications (i.e., antidepressants, anxiolytics, mood stabilizing AEDs, lithium, antipsychotics), off-label medications (antihypertensives for anxiety) and/or non-medication approaches (i.e., psychological therapy). These conditions included: depression, anxiety, bipolar disorder, posttraumatic stress disorder, personality disorder, schizophrenia and history of suicidality or suicide attempts. The psychological/psychiatric conditions were also identified in the EMU pre-admission history and physical, the discharge summary report from the EMU admission and/or clinician notes of current medical problems just prior to the EMU admission. The social domain was conceptualized as factors influenced by the social environment these include: marital status, disability, history of abuse (physical and/or sexual), current smoking status, alcohol use and illicit drug use. These factors were documented based on administrative data used for medical billing (marital status and disability) or through documentation in the medical record by the clinician as part of the history or description of the patient’s current problem statement. Statistical analyses were conducted using SPSS v 19. Three groups based on the continuous video-EEG confirmed diagnosis are presented in the univariate analyses: PNES only, PNES & Epilepsy and Epilepsy only (dependent variables). Next, univariate and multivariate logistic regression analyses were utilized to determine the biological, psychological and social predictors of PNES only and PNES & Epilepsy (independent variables). In these analyses persons in the Epilepsy only group were the comparator. In the univariate models we examined the predictive ability of each independent variable via odds ratios (ORs) and 95% CIs. ORs and 95% CIs provide a parsimonious assessment of both the strength and reliability of the associations. Last, we conducted two multivariate logistic regression analyses which included all independent variables that had significant predictive value in the univariate model at the p < 0.05 level. Since the univariate analyses which are unable to assess whether an association is direct, or is due to the covariance with another variable (Duncan and Oto, 2008) the multivariate model incorporates the correlations and interactions between all the variables in the model, i.e., the relationship between female gender can be examined simultaneously as both an individual predictive factor and as a control variable.
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
4
J.O. Elliott, C. Charyton
Results A total of 689 patients admitted to our EMU had events leading to a diagnosis, 47% (n = 324) participants had their typical events diagnosed as PNES only, 12% (n = 84) had their typical events diagnosed as PNES & Epilepsy and 41% (n = 281) were diagnosed with Epilepsy only by video-EEG, see Table 1.
PNES only There were significant differences between the PNES only and the Epilepsy only groups based on biological—biomedical factors (gender, age at onset of events, number of years having events or seizures, history of head injury, BMI (obesity) and all somatic conditions except stroke, hypothyroidism and cancer. There were also differences in psychological factors (all except history of suicidality/suicide attempts) and social factors (marital status, disability status, history of physical or sexual abuse and smoking status), see Table 1. Univariate logistic regression analyses revealed 24 significant predictors of PNES only, see Table 2. These included biological—biomedical factors (female gender, years with events or seizures, age of onset, history of head injury, a diagnosis of intellectual disability, migraine, cardiovascular disease, diabetes, asthma, GERD, IBS, arthritis, chronic pain and insomnia), psychological factors (depression, anxiety, bipolar disorder, PTSD, personality disorder and schizophrenia) and social factors (marital status and history of physical/sexual abuse). Once these factors were entered into a multivariable logistic regression model, eight predictors of the PNES only group remained significant. Five were biological—biomedical: numbers of years with events OR = 1.10 (95% CI: 1.07—1.13), history of head injury OR = 1.91 (95% CI: 1.21—3.01), asthma OR = 2.94 (95% CI: 1.69—5.12), GERD OR = 1.72 (95% CI: 1.05—2.82) and chronic pain OR = 2.25 (95%CI: 1.31—3.86), one psychological factor: anxiety OR = 1.72 (95% CI: 1.09—2.71) and two social factors: being married OR = 1.81 (95% CI: 1.15—2.85) and history of physical/sexual abuse OR = 3.35 (95% CI: 1.23—9.10), see Table 2.
PNES & Epilepsy There were significant differences between the PNES & Epilepsy and the Epilepsy only groups based on biological—biomedical factors (female gender, migraines, asthma, GERD and chronic pain). There were also differences in psychological factors (depression, anxiety, bipolar disorder and personality disorder) and social factors (disability status, history of physical or sexual abuse and smoking status), see Table 1. Univariate logistic regression analyses revealed eleven significant predictors of PNES & Epilepsy, see Table 3. These included biological—biomedical factors (female gender, migraine, asthma, GERD and pain), psychological factors (depression, anxiety, bipolar disorder and personality disorder) and social factors (disability status, history of physical/sexual abuse and smoking status). Once these factors were entered into a multivariable logistic regression model two biological—biomedical factors
migraine OR = 1.83 (95% CI: 1.04—3.23) and GERD OR = 2.17 (95% CI: 1.18—3.99) remained significant predictors of a PNES & Epilepsy diagnosis, see Table 3.
Discussion The purpose of this research was to gain a better understanding of the biological, psychological and social factors associated with a continuous video-EEG confirmed diagnosis of PNES. The current study focused on a more comprehensive look at PNES than many previous investigations. The multivariate models provide more insight than individual factors or stereotyped characteristics (i.e., female predominance) when a diagnosis of PNES is suspected. Our findings support the importance of understanding the complex interaction of biological—biomedical factors (number of years with events, previous head injuries and certain medical diagnoses such as migraines, asthma, GERD and chronic pain) as well as psychological (anxiety) and social factors (marital status and physical/sexual abuse) in the PNES population.
Biological In our study the number of years with events was predictive of a PNES only diagnosis. In the logistic model this translates to a 10% increased odds of a PNES diagnosis for each year of having events before video-EEG diagnosis. This supports the importance of receiving a clear and positive explanation of the PNES diagnosis based on video-EEG as early as possible (Kerr et al., 2011) which by itself may lead to a significant reduction and/or elimination of PNES events in some patients (McKenzie et al., 2010). One recent Canadian study found significant reductions in total emergency room visits, emergency room visits for neurological causes in the 2 years following PNES diagnosis which was sustained over a four year period (Jirsch et al., 2011). Prior research suggests that traumatic experiences play an important role in the development and expression of PNES (Fleisher et al., 2002; Selkirk et al., 2008). In particular prior head injury has been associated with the development of PNES (Westbrook et al., 1998; LaFrance and Syc, 2009). The current study provides additional support for inquiring about previous head injuries. We found higher rates of asthma in both the PNES and the PNES & Epilepsy groups. These findings are consistent with a previous investigation which suggested the association between asthma and PNES may be accounted by a combination of somatization, anxiety hyperventilation and dissociation (de Wet et al., 2003). That study was limited by the inclusion of non-confirmed video-EEG cases and made comparisons with a group of control patients with psychosis. The current study suggests asthma and anxiety are significant in the larger picture when many other factors have been accounted for. A recent paper found a very similar profile of comorbidities in PNES (Dixit et al., 2013), however persons with PNES & Epilepsy were excluded in that study. We also found GERD was associated with a PNES diagnosis. Previous research reported a higher rate of ulcers in persons with PNES compared to those with epilepsy (Tojek et al., 2000) and greater use of gastric reflux medications in those with PNES (Hantke et al., 2007). The literature also
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
Biopsychosocial predictors of psychogenic non-epileptic seizures Table 1
5
Univariate analysis of biopsychosocial characteristics.
Characteristic
PNES only (n = 324) Mean(sd)/%(n)
PNES & Epilepsy (n = 84) Mean(sd)/%(n)
Epilepsy only (n = 281) Mean(sd)/%(n)
37.3 (12.2) 71 (230)***
37.1 (12.5) 70 (59)**
38.6 (13.5) 54 (151)
89 (287) 11 (37) 28.2 (15.1)*** 9.1 (11.1)*** 37 (119)***
87 (73) 13 (11) 16.3 (15.0) 20.8 (14.9) 26 (22)
87 (244) 13 (37) 18.1 (16.7) 20.5 (14.9) 25 (69)
2 (6) 30 (97) 19 (62)** 49 (159)*** 7 (22)***
6 (5) 33 (28) 20 (17) 41 (34) 25 (21)
3 (7) 31 (88) 26 (74) 40 (112) 18 (51)
53 (171)*** 5 (17) 10 (31)** 19 (62) 32 (104) 17 (55)*** 11 (37) 33 (107)*** 41 (133)*** 12 (40)*** 6 (19) 16 (53)* 36 (117)*** 31 (100)***
51 (43)*** 7 (6) 4 (3) 12 (10) 24 (20) 11 (9) 14 (12) 23 (19) 45 (38)*** 7 (6) 4 (3) 13 (11) 27 (23)** 27 (23)
30 (85) 7 (20) 4 (11) 17 (47) 32 (89) 6 (17) 14 (38) 13 (37) 20 (57) 3 (9) 6 (16) 11 (30) 13 (35) 19 (54)
Psychological Depression Anxiety Bipolar disorder Post-traumatic stress disorder Personality disorder Schizophrenia History of suicidality/suicide attempts
67 (218)*** 50 (161)*** 17 (55)** 7 (21)*** 7 (21)* 9 (30)* 8 (25)
68 (57)*** 49 (41)*** 23 (19)** 5 (4) 8 (7)* 8 (7) 10 (8)
45 (127) 25 (70) 9 (25) 1 (3) 2 (6) 5 (13) 5 (13)
Social Marital status Single Married Separated Divorced Widowed Disabled History of abuse (physical/sexual) Current smoker Current alcohol use Current illicit drug use
40 (27)* 40 (131)* 4 (13) 14 (45) 2 (8) 39 (125) 12 (39)*** 45 (147) 25 (80) 13 (43)
51 (43) 29 (24) 4 (3) 13 (11) 4 (3) 56 (47)* 12 (10)** 50 (42)* 25 (21) 14 (12)
51 (144) 31 (86) 1 (4) 15 (42) 2 (5) 41 (115) 3 (8) 38 (106) 29 (82) 13 (36)
Biological Age Female gender Race/ethnicity Caucasian Non-Caucasian Age of onset Number of years with events or seizures History of head injury BMI Underweight Normal Overweight Obese Intellectual disability Somatic conditions Migraines History of stroke Cardiovascular disease High cholesterol Hypertension Diabetes Hypothyroidism Asthma Gastro-esophageal reflux disease Irritable bowel syndrome Cancer Arthritis Chronic pain Insomnia
Notes: Comparisons are made using Epilepsy only as the reference group. For continuous data one-way ANOVA was conducted or Wilcoxon rank sum tests were conducted. For categorical data 2 analyses were conducted. PNES = Psychogenic non-epileptic seizures. * p ≤ 0.05. ** p ≤ 0.01. *** p ≤ 0.001.
Please cite this article in press as: Elliott, J.O., Charyton, C., Biopsychosocial predictors of psychogenic non-epileptic seizures. Epilepsy Res. (2014), http://dx.doi.org/10.1016/j.eplepsyres.2014.09.003
+Model EPIRES-5215; No. of Pages 11
ARTICLE IN PRESS
6
J.O. Elliott, C. Charyton Table 2
Logistic regression of persons with diagnosed events: PNES only vs. Epilepsy only.
Characteristic
Univariate
Multivariate
Odds ratio (95% CI)
p-value
Odds ratio (95% CI)
p-value
Biological Female gender Number of years with events or seizures Age of onset History of head injury Overweight/obese Intellectual disability Somatic conditions Migraine Cardiovascular disease Diabetes Asthma Gastro-esophageal reflux disease Irritable bowel syndrome Arthritis Chronic pain Insomnia
2.11 1.07 0.96 1.78 0.91 0.33
(1.51—2.95) (1.05—1.09) (0.95—0.97) (1.25—2.54) (0.65—1.28) (0.19—0.56)
