Clinical Predictors of Psychogenic Nonepileptic Seizures a Critically Appraised Topic

CRITICALLY APPRAISED TOPICS

Clinical Predictors of Psychogenic Nonepileptic Seizures A Critically Appraised Topic

Matthew T. Hoerth, MD,* Kay E. Wellik, MLS, AHIP,† Bart M. Demaerschalk, MD, MSc, FRCP(C),* Joseph F. Drazkowski, MD,* Katherine H. Noe, MD, PhD,* Joseph I. Sirven, MD,* and Dean M. Wingerchuk, MD, MSc, FRCP(C)*

Key Words: nonepileptic seizures, seizures, epilepsy, diagnosis, Background: Psychogenic nonepileptic seizures (PNES) are often evidence-based medicine, critically appraised topic disabling and usually associated with psychiatric disorders and reduced quality of life. Although often suspected based on historical (The Neurologist 2008;14: 266–270) and clinical features, the gold standard for diagnosis of PNES is video electroencephalography. Identification of clinical features that reliably distinguish PNES from ES would be valuable in acute care settings, for patients that have coexisting disorders, and those with 22-year-old woman with chronic posttraumatic leg pain multiple event types. Adeveloped spells. The spells begin with sharp head pain Objective: To determine the diagnostic value of putative clinical and blurred vision and then sequential tightening of facial symptoms or signs of PNES against the gold standard of video muscles, right arm stiffening, and arrhythmic body jerking. electroencephalography. Her eyes are closed for most of the spell duration, which averages approximately 90 seconds, after which and she is Methods: We addressed the objective through development of a drowsy for up to 30 minutes. The patient cannot recall the structured critically appraised topic that included a clinical scenario, events. She is admitted for continuous video electroencepha- structured question, search strategy, critical appraisal, results, evi- lography (EEG) monitoring to determine whether the spells dence summary, commentary, and bottom-line conclusions. Partic- represent epileptic seizures (ES) or psychogenic nonepileptic ipants included consultant and resident neurologists, a medical seizures (PNES). librarian, clinical epidemiologists, and content experts in the field of epileptology. Results: There were wide variations in the rates of coexisting PNES BACKGROUND and epilepsy and study methodology. Ictal stuttering and the “teddy PNES, also termed pseudoseizures, have been defined bear” sign were associated with moderate specificity for PNES. as “paroxysmal changes in behavior that resemble ES but are However, the presence of pelvic thrusting or ictal eye closure did not without organic cause and are not accompanied by the ictal, accurately distinguish PNES from ES. peri-ictal and interictal EEG changes (spikes and slowing) that characterize epilepsy.”1 PNES are often suspected in Conclusions: The presence of either ictal stuttering or the teddy patients with a history of somatization, abuse, or psychiatric bear sign is moderately specific but poorly sensitive for PNES. comorbidity or when spells are characterized by unusual Pelvic thrusting and ictal eye closure are not reliable indicators of features such as emotional triggers, prolonged duration, a PNES. Future studies should establish more precise and reliable stop/start quality, or pelvic thrusting. However, the diagnosis definitions of clinical signs and evaluate combinations of such signs of PNES can be difficult based on clinical criteria alone. The in a broad spectrum of patients with PNES and ES spell phenotypes gold standard for diagnosis is recording a typical event with that may be difficult to distinguish, such as spells of unresponsive- video EEG to confirm the absence of electrographic changes ness with motor manifestations. Because PNES and ES may coexist, on the ictal EEG recording. In 1 study, 20% of patients analysis of diagnostic accuracy of clinical features should be per- referred for video-EEG with a clinical diagnosis of PNES formed for individual spells. were determined to have epilepsy or other physiologic cause for their events.2 The incidence of PNES has been reported to be 1.5 From the *Department of Neurology, and †Department of Library Services, per 100,000 per year, which is approximately 4% of the 3 Division of Education Administration, Mayo Clinic, Scottsdale, Arizona. incidence of epilepsy. In addition, series from patients Reprints: Dean M. Wingerchuk, MD, MSc, FRCP(C), Department of Neu- referred to neurology clinics have found that 5% to 20% of rology, Mayo Clinic, 13400 E Shea Boulevard, Scottsdale, AZ 85259. patients referred for epilepsy have pseudoseizures.3,4 The E-mail: [email protected]. interval from first event to PNES diagnosis averages more Copyright © 2008 by Lippincott Williams & Wilkins 4 ISSN: 1074-7931/08/1404-0266 than 8 years. Treatment differs greatly between ES and DOI: 10.1097/NRL.0b013e31817acee4 PNES and misdiagnosis can be detrimental. For example,

266 The Neurologist • Volume 14, Number 4, July 2008 The Neurologist • Volume 14, Number 4, July 2008 Clinical Predictors of PNES

TABLE 1. Summary of Study Designs No. Patients Multiple Independent Raters Evaluation Blinded Study Clinical Feature Study Type ES PNES of Clinical Sign From Video EEG Geyer et al5 Pelvic thrusting Retrospective 150 100 Yes Yes Vossler et al6 Ictal stutter Prospective 113 117 No No Chung et al7 Ictal eye closure Retrospective 155 52 Unclear Unclear Syed et al8 Ictal eye closure Prospective 314 spells 151 spells Unclear Yes Burneo et al9 “Teddy bear” sign Retrospective 453 381 Yes No treating a PNES patient for epilepsy exposes them unneces- and the MeSH term seizures. This search was limited to sarily to the potentially serious side effects of antiepileptic humans and English language and combined with the saved drugs and such therapies may even exacerbate PNES symp- search hedge for diagnosis (comprehensive retrieval). The toms.4 Current estimates of the rate of coexisting ES and purpose of this initial search was to survey the available PNES range from 5% to 50%; such scenarios are complex literature regarding the types of clinical features reported to with treatment required for both disorders.4 assist with PNES diagnosis. Titles and abstracts of 239 Although video EEG monitoring is the gold standard potentially relevant articles were examined. Clinical features for diagnosis of PNES, identification of clinical signs that linked to PNES from this survey included tongue biting, fibro- can reliably distinguish PNES from ES would be useful for myalgia, pelvic thrusting, stertorous breathing, weeping, ictal clinicians evaluating these patients in acute care settings, stuttering, ictal eye closure, presence of a teddy bear or similar for patients with new-onset events or multiple seizure object brought to the video EEG unit, alexithymia, and preictal types, for clinicians without access to video EEG services, pseudosleep. Pelvic thrusting, ictal stuttering, ictal eye closure, and for patients with coexisting PNES and ES. Features and the “teddy bear sign” were selected for review because they such as progression of symptoms in a manner consistent with were positive features present during the examination, which neuroanatomy, tongue biting, and urinary incontinence are would be more easily identified by observers of the event. For reported to suggest an epileptic rather than nonepileptic each of these signs, an individual Ovid MEDLINE search was etiology for spells. We sought to critically appraise the conducted by combining the term with the prior PNES-related evidence supporting use of other putative clinical predictors search terms, yielding the following results: pelvic thrusting (7 in the diagnosis of pseudoseizures when compared with the articles), ictal stuttering (1 article), ictal eye closure (2 articles), gold standard of video EEG. teddy bear, teddy bears, play, and playthings (20 articles). The articles selected for appraisal represented the most valid diag- CLINICAL QUESTION nostic studies of each clinical feature with direct comparison to For patients presenting with spells, do specific clinical video EEG as the reference standard for diagnosis. EMBASE observations, compared with the gold standard of video EEG, was also searched but no additional articles of high quality were accurately differentiate PNES from ES? detected.

SEARCH STRATEGY EVIDENCE, RESULTS, AND The Ovid MEDLINE database was searched from 1950 CRITICAL APPRAISAL to the ﬁfth week in January 2008 using the keywords: Table 1 describes the primary study design character- pseudoseizure, psychogenic nonepileptic seizure, and PNES istics and Table 2 summarizes the results. Table 3 is a guide

TABLE 2. Diagnostic Accuracy of Clinical Signs for Distinguishing PNES From ES Clinical Sign Sensitivity Speciﬁcity LR؉ LR؊ Pelvic thrusting5 (patient data, dichotomous) 17% (10–24) 89% (81–98) 2.46 (1.32–4.58) 0.89 (0.81–0.98) Pelvic thrusting5 (patient data; PNES “thrashers” 78% (59–97) 76% (64–88) 3.24 (1.87–5.63) 0.29 (0.12–0.70) vs. frontal ES) Pelvic thrusting5 (spell data; PNES thrashers vs. 100% 20% (8–32) 1.25 (1.07–1.46) 0 frontal ES) Ictal stutter6 (patient data, dichotomous) 9% (3–14) 100 Inﬁnity 0.91 (0.86–0.97) Eye closure7 (patient data, dichotomous) 96% (91–100) 98% (96–100) 50.0 (16.3–153.5) 0.04 (0.01–0.15) Eye closure8 (spell data; dichotomous with eye 64% (56–71) 74% (69–79) 2.47 (1.98–3.08) 0.49 (0.39–0.61) closure Ͼ5% spell duration) Teddy bear sign9 (patient data, dichotomous) 5.2% (3.0–7.5) 99% (99–100) 7.93 (2.37–26.5) 0.95 (0.93–0.98)

Values in parentheses are 95% conﬁdence intervals. LRϩ indicates likelihood ratio for a positive test result; LRϪ, likelihood ratio for a negative test result.

words during a seizure, either spontaneously or in response to TABLE 3. Clinical Significance of Likelihood Ratios questioning, which was not present interictally.” The authors Likelihood Ratio Change in Probability Clinical Importance were careful to differentiate between aphasia potentially Ͼ10 or Ͻ0.1 Large Often very high caused by epileptic seizure activity versus stuttering. Expres- 5–10 or 0.1–0.2 Moderate Moderate to high sive or receptive aphasias including word substitutions and 2–5 or 0.2–0.5 Small Sometimes neologisms were not considered ictal stuttering. This was a 1–2 or 0.5–1.0 Very small Rare prospective study, however, diagnosis and presence of stuttering were assessed by review of video EEG data with final ascertainments made by consensus at a case conference. Therefore, assessment of ictal stuttering seems neither have to clinical interpretation of likelihood ratios, indicating the been independently evaluated by more than 1 reviewer nor impact of the likelihood ratio on posttest probability of assessed in blinded fashion with respect to video EEG results. PNES. Ten of 117 PNES patients displayed ictal stuttering Pelvic Thrusting during their spells compared with none of the 113 ES pa- A retrospective observational study performed at the tients. These data translate to perfect specificity but very low University of Michigan sought to “evaluate pelvic thrusting sensitivity (8.9%). Thus, the presence of ictal stuttering as a potential diagnostic sign and as a potential lateralizing or suggests a diagnosis of PNES but it is not commonly ob- localizing indicator of ictal onset” in 261 consecutively ad- served. mitted patients with PNES and ES as defined by video EEG monitoring.5 The groups were defined as having focal epi- Ictal Eye Closure lepsy ͓n ϭ 50 each for right temporal lobe epilepsy (TLE), Two articles provide somewhat contrasting data in ͔ evaluating ictal eye closure as a predictor of nonepileptic left TLE, and frontal lobe epilepsy , generalized epilepsy 7 (n ϭ 11), or PNES (n ϭ 100). Exclusion criteria were not spell etiology. In a single-center study, Chung et al retro- clear and pelvic thrusting was not further defined. A maxi- spectively acquired video EEG data from consecutively ad- mum of 5 seizures from each patient were reviewed to mitted epilepsy monitoring unit patients aged 6 to 65 years. A categorize each patient with the presence or absence of pelvic total of 221 patients’ video EEGs were reviewed (including thrusting. Two investigators independently reviewed videos 938 events), 52 (23.5%) received the diagnosis of PNES and while blinded to EEG data and concordance for diagnosis was 156 (70.6%) ES; 13 patients were excluded because they did perfect. not have stereotypical events during their admission. The Pelvic thrusting was noted in both PNES (n ϭ 17/100) presence or absence of ictal eye closure, which was not and epilepsy (n ϭ 18 of 261; 12 frontal and 6 TLE), yielding clearly defined by the authors, was retrospectively deter- good specificity but low sensitivity (Table 2). However, most mined for each patient by reviewing video clips without the of the PNES patients had ictal unresponsiveness without corresponding EEG. motor manifestations. Therefore, the authors presented data The investigators found that 50 of 52 (96%) PNES concerning a subgroup of PNES patients whose spells were patients had ictal eye closure compared with only 3 of 156 described as predominantly “thrashing.” One important clin- (2%) of patients with ES (all partial seizures). In fact, they ical question is whether pelvic thrusting could differentiate noted that most patients in the PNES group closed their eyes such patients from those with frontal lobe epilepsy. Fourteen during the entire seizure. This yields high sensitivity (96%) of 18 thrashing PNES patients displayed pelvic thrusting and specificity (98%) values and clinically useful likelihood compared with 12 of 50 frontal lobe ES patients. Of the 185 ratios for both presence and absence of ictal eye closure. spells captured in those 14 PNES patients, all included pelvic Limitations of this study include its retrospective nature, lack thrusting compared with 80% (32 of 40) spells in the frontal of study entry criteria, and clear definition of ictal eye ES group. Therefore, the PNES group contributed many more closure, single-observer evaluation of ictal eye closure, lack spells per patient and they were likely rather stereotypic. The of description of the clinical events, and uncertain blinding authors do not mention whether any patients had coexisting status. Furthermore, despite collection of 938 separate events, PNES and ES, but assuming there were none, data from this the analysis and reporting is done at the level of the patient study (Table 2) show that the presence or absence of pelvic (ie, as long as 1 spell was associated with ictal eye closure, thrusting, whether expressed at the level of the patient or the that patient would be categorized as having ictal eye closure) spell, exerts little effect on the posttest probability of PNES. rather than the level of individual spells. No note is made regarding coexistence of PNES and ES. Ictal Stuttering A more recent publication on ictal eye closure reached Vossler et al6 sought to determine “...ifictal stutter- somewhat different conclusions.8 The investigators planned ing is more common among patients with PNES than patients to prospectively “assess whether observer or self-report of with ES.” This was a prospective observational study of 230 eye closure can be used to predict PNES, prior to video EEG consecutive patients admitted to the epilepsy monitoring unit evaluation.” Entry criteria included age Ն18 years, adequate at Swedish Medical Center in Seattle. Patients were excluded patient vision and adequate visualization of patient’s eyes in if they were under 18 years old, had evidence of mental the reviewed video. Spells were excluded from analysis if retardation or had interictal stuttering. Ictal stuttering was there was no alteration of awareness or if they had an organic defined as the “repeated utterances of initial phonemes or etiology. This study differed from that of Chung et al7 in that

268 © 2008 Lippincott Williams & Wilkins The Neurologist • Volume 14, Number 4, July 2008 Clinical Predictors of PNES it was prospective, ictal eye closure was defined (eye closure PNES from ES, only to raise the suspicion of PNES when duration greater than 5% of the event time), and video present. reviewers were blinded to the EEG and seizure classification. In addition to the video EEG recording analysis, the authors Clinical Bottom Lines gathered self and observer report data on eye closure. • Ictal stuttering is specific for PNES but is uncommon One hundred twelve subjects were enrolled and the (specificity 100%, sensitivity 9%). video EEG diagnosis of ES was made in 84 patients and • Video-recorded ictal eye closure is associated with PNES PNES in 43 patients; 15 patients (13.4%) had coexisting and is moderately specific if closure lasts Ͼ50% of the PNES and ES. A total of 465 evaluable spells were recorded duration of the spell (LRϩϭ5.2). However, self-report (87 were excluded based on entry criteria listed above), and observer reports of ictal eye closure are unreliable. including 306 focal seizures, 8 primarily generalized seizures, • Pelvic thrusting is more common in PNES than ES but and 151 PNES. Ninety-six of 151 PNES events displayed cannot satisfactorily distinguish PNES from frontal lobe ictal eye closure for at least 5% of the spell duration com- seizures ͓LRϩϭ1.25 (1.07–1.46)͔. pared with 81 of 314 from the ES group. This equates to 64% • Bringing an age-inappropriate toy animal to the epilepsy sensitivity and 74% specificity of recorded ictal eye closure monitoring unit (teddy bear sign) increases the likelihood for differentiating PNES from ES spell etiology. Eye closure ͓ ϩϭ Ͼ of PNES in an individual patient LR 7.93 (2.37– for 5% of the spell was the most sensitive definition 26.5)͔ but is weakly sensitive (5.2%). whereas eye closure for the entire spell was the most specific. • Although any or a combination of the described clinical The optimal results were obtained at about 50% to 55% eye signs may increase the likelihood of the existence of PNES closure duration, with sensitivities of 52% to 58% and spec- for an individual patient, none can displace video EEG ificities of 87% to 90%, equating to likelihood ratios of monitoring for diagnosis of individual spells. approximately 5.1 for presence of eye closure and 0.5 for absence of eye closure. Self-report (53.5% sensitive; 50.7% specific) was somewhat more accurate than observer report DISCUSSION (41.9% sensitive; 47.0% specific) but neither was satisfac- Of the 4 clinical features reviewed herein, ictal stutter- tory. For the aforementioned methodological reasons, this ing and the teddy bear sign are most strongly associated with study is more likely a valid assessment of the utility of ictal PNES, however, both are very insensitive, occurring in less eye closure for spell diagnosis. than 10% of such patients. Further study of ictal stuttering, especially its definition and agreement among independent raters, is needed to confirm its specificity for PNES. The Teddy Bear Sign teddy bear sign appears to be a marker for psychologic or Burneo et al9 observed that age-inappropriate behavior, psychiatric morbidity and could perhaps signal the need for such as bringing a teddy bear or similar item to the epilepsy more careful and extensive psychiatric evaluation. However, monitoring unit, was associated with PNES. Their retrospec- the sign is not a manifestation of the spells themselves and tive review of a single center’s video EEG database evaluated cannot serve to differentiate PNES from ES. the diagnostic accuracy of presence of a stuffed toy was Pelvic thrusting is commonly associated with PNES but compared with the reference standard of video EEG for the is also known to occur in ES, especially those of frontal diagnosis of PNES. Information data cards were created of all origin. In the study by Geyer et al,5 pelvic thrusting seems to patients admitted, collecting data regarding objects “deemed have some moderate diagnostic value for PNES, however, unusual for a neurologically intact older adolescent or adult- detailed evaluation shows that it cannot reliably distinguish aged patient to possess.” Objects brought by family members PNES from frontal lobe ES. Similarly, the observation or or other visitors were not considered age-inappropriate ob- self-report of ictal eye closure may not be very clinically jects and not included into the stuffed toy group. Exclusion useful. Although recorded ictal eye closure, especially if it criteria for this study were patients less than 14 years old and lasts for between 55% and 100% of the spell duration, is patients with an unclear diagnosis, both ES and PNES, or reasonably strongly associated with PNES, that information other nonepileptic events. can only be reliably gathered by video EEG. It is not known In total, 903 patients were initially included based on whether clinical observers examining the patient during a the patient age, 69 patients were excluded for reasons listed spell in an emergent care setting could perform more reliably above ͓including 17 (1.9%) who had coexisting PNES and than patient self-report or other observers. ES͔, 453 were diagnosed with ES and 381 with PNES. Video EEG monitoring remains the gold standard for Twenty patients (sensitivity ϭ 5.2%) in the PNES group had diagnosis of epileptic-like spells. Subtle (simple partial) sei- inappropriate stuffed toys compared with 3 (0.7%) in the ES zures may be missed when using surface electrodes to record group. Therefore, the presence of the teddy bear sign is the EEG and this possibility should be considered in the 99.3% specific for PNES. Furthermore, the 3 ES patients who evaluation process of the possible PNES patient.10 However, brought toy animals to the unit had coexisting psychiatric it is particularly important to confirm PNES in patients with disease and 3 of the 17 excluded patients with coexisting severe, disabling symptoms in whom lack of such evidence PNES and ES brought toy animals to the unit. This retrospec- may fail to convince the patient, family members, or other tive, unblinded observational study is interesting but the care providers, therefore exposing the patient to unnecessary clinical sign is insensitive and cannot be used to differentiate and potentially harmful effects of epilepsy therapies and

delaying the appropriate psychologic and psychiatric man- REFERENCES agement options. Futhermore, PNES and ES may coexist in a 1. Bowman ES, Coons PM. The differential diagnosis of epilepsy, pseudo- minority of cases, highlighting the need to determine the seizures, dissociative identity disorder, and dissociative disorder not number of types of spells patients experience and to capture otherwise specified. Bull Menninger Clin. 2000;64:164–180. 2. Parra J, Iriarte J, Kanner AM. Are we overusing the diagnosis of each of them during video EEG monitoring. psychogenic non-epileptic events? Seizure. 1999;8:223–227. The use of video EEG does not diminish the need for 3. Krumholz A, Hopp J. Psychogenic (nonepileptic) seizures. Semin Neu- careful history-taking and examination nor the potential util- rol. 2006;26:341–350. ity of accurate clinical symptoms or signs of PNES. World- 4. De Timary P, Fouchet P, Sylin M, et al. Non-epileptic seizures: delayed wide access to video EEG monitoring is limited and in many diagnosis in patients presenting with electroencephalographic (EEG) or clinical signs of epileptic seizures. Seizure. 2002;11:193–197. instances, clinical diagnosis must suffice. Accurate clinical 5. Geyer JD, Payne TA, Drury I. The value of pelvic thrusting in the indicators of PNES could be useful in acute care settings, for diagnosis of seizures and pseudoseizures. Neurology. 2000;54:227–229. evaluation of patients with new onset or multiple types of 6. Vossler DG, Haltiner AM, Schepp SK, et al. Ictal stuttering: a sign events, and for assessment of those challenging patients with suggestive of psychogenic nonepileptic seizures. Neurology. 2004;63: both PNES and ES. 516–519. 7. Chung SS, Gerber P, Kirlin KA. Ictal eye closure is a reliable indicator It is unlikely that 1 clinical sign will prove sensitive and for psychogenic nonepileptic seizures. Neurology. 2006;66:1730–1731. specific enough to apply in most clinical encounters. How- 8. Syed TU, Arozullah AM, Suciu GP, et al. Do observer and self-reports ever, prospective evaluation of combinations of such signs of ictal eye closure predict psychogenic nonepileptic seizures? Epilepsia. could lead to the development of validated clinical models or 2008;49:898–904. 9. Burneo JG, Martin R, Powell T, et al. Teddy bears: an observational decision-support algorithms. This would require methodolog- finding in patients with non-epileptic events. Neurology. 2003;61:714– ical improvements beyond what has been accomplished by 715. most investigations to date. Ideally, future studies should 10. Devinsky O, Sato S, Kufta CV, et al. Electroencephalographic studies of prospectively evaluate a number of well-defined clinical signs simple partial seizures with subdural electrode recordings. Neurology. that have high inter-rater reliability or agreement. Patient 1989;34:527–533. samples should include a broad representation of spells with common features which cause reasonable uncertainty about the etiology, for example, spells of loss of awareness with APPENDIX motor features. They should use independent assessors who Mayo Clinic Evidence Based Clinical Practice, Research, are blinded to the reference standard of video EEG diagnosis. Informatics, and Training (MERIT) Center Cofounders and Finally, evaluation should be done at the level of individual Codirectors: Dean M. Wingerchuk, MD, and Bart M. spells, rather than patients, to account for the fact that spell Demaerschalk, MD. semiology may vary from occasion to occasion and because Medical Library Sciences and Evidence Based Search- PNES and ES may coexist. Consistent incorporation of these ing and Informatics Director: Kay E. Wellik, MLS, AHIP. and other methodological approaches could allow the clinical Evidence Appraisers: Matthew T. Hoerth, MD; Bart M. assessment of “spells” to advance toward more sophisticated models that accurately represent their true complexity. In the Demaerschalk, MD; Dean M. Wingerchuk, MD. meantime, detection of ictal stuttering, prolonged ictal eye Content Experts: Joseph F. Drazkowski, MD, Kather- closure, or the teddy bear sign should elevate concern about ine Noe, MD, Joseph I. Sirven, MD. psychologic or psychiatric etiologies or comorbidities in Evidence Appraisal Date: March 4, 2008. these challenging patients. Suggested Topic Reappraisal Date: March 4, 2010.