Psychogenic non-epileptic seizures—Diagnostic issues: A critical review

3.1. Description of the studies 

Some comments on the results in Table 1, Table 2 are in order. The dominant type of design in studies on PNES is the open non-randomized comparative study. The studies are therefore not protected against the effects of bias, especially selection bias. Patients with epilepsy are mostly used as the comparator. Although this seems obvious in the light of the fact that the symptoms resemble seizures, but is not logical when studying for example the psychopathology or outcomes in daily life in patients with PNES. In many cases patients with PNES have been in the diagnostic process as ‘epileptic patient’ for many years. The effect may therefore not be different and patients with epilepsy may not be helpful to study the specific effects or etiologies in patients with PNES. The sample size is mostly rather limited; the majority in the range of 20–30 patients. Given the high variability of the symptoms and underlying characteristics in these patients, it is doubtful whether any of the studies achieves sufficient power. Formal power analyses, have however not been performed. The larger studies are retrospective studies and mostly studies on patient files. The only exceptions are postal questionnaire studies with such a high non-response rate that bias cannot be excluded.

3.2. Epidemiology 

The epidemiological data have two sources of origin:

When evaluating the specific characteristics, most studies found that 75–80% of the patients with PNES are female [8], [24]. Holmes et al. [25], however, suggest that the gender factor may be less pronounced than given by the previous figures. Generally, fewer men than women seek medical attention. This may result in an evaluation of only the most severely affected men. This is in line with some reports that suggest more severe psychogenic factors in men compared to women. The true incidence of men with PNES may thus be significantly unrecognized and underreported. The PNES episodes tend to begin in early adulthood in most patients [3], [8], [26]. Wyllie et al. [27] have demonstrated that PNES are rare in children under 10 years old. When occurring in the younger age range, the PNES are seldom a ‘stand-alone phenomenon’ and more often occur as a symptom in a range of other symptoms [28], [29], mostly mannerisms, parasomnias, hyperventilation attacks, breath-holding spells, syncope or movement disorders [30]. On the other hand Behrouz et al. [31] and Abubakr and Wambacq [32] have demonstrated that PNES should also be considered as a diagnostic possibility in older patients (60 years of age or older). The actual prevalence of PNES in the elderly is however not known. Moreover a higher incidence is found with lower educational level [26]. Significant psychiatric comorbidity is found in at least 70% of the patients, although often the exact definition of the psychiatric disorder is lacking or nonconclusive [9], [14], [21].

Among the conversion types of symptoms, PNES is the second most common conversion symptom category exceeded in frequency only by all forms of movement disturbances, including paralysis, weakness, impaired gait, and tremor [15].

3.3. Diagnosis including differential diagnosis 

In a first phase PNES must be distinguished from organic non-epileptic seizures, such as syncope, migraine or tension headache, hyperventilation, transient ischemic attacks, hypoglycaemia, benign myoclonia, myasthenia gravis, paroxysmal choreoathethosis, seizures due to alcohol abuse and one of the parasomnias [7], [16], [33]. In children especially the parasomnias are a diagnostic challenge [29], [34]. All these conditions require specific diagnostic procedures. Next, attention must be paid to non-epileptic emotionally based states that resemble seizures [12]. Especially panic attacks and related anxiety states such as impulse control problems and flashbacks in the context of post-traumatic stress disorder are a major diagnostic confound that must be distinguished from epilepsy [7], [35]. Moreover ictal and postictal fear are reported to be common anxiety states in epilepsy and are reported to occur in approximately 10–15% of the patients with complex partial seizures with a temporal origin [15]. Therefore not all anxiety states are non-epileptic by definition.

Most of the attention during this phase must, however, be paid to distinguishing epileptic from non-epileptic seizures. Such differential diagnosis may be challenging in many cases [14], [34], [35], [36], [37] and especially in those cases mimicking idiopathic generalized seizures, such as myoclonic or absence seizures [38]. Various studies demonstrated that many signs that have been considered typical for PNES, appeared not to be specific and can also be found in epileptic seizures, especially in those seizures that originate from the frontal lobe [3], [39]. Hence, caution is needed in the clinical interpretation of ictal features that have been considered pathognomonic for epileptic seizures such as tongue biting, or complex movements [2], [10]. Sometimes injuries are seen as indicative for epilepsy; however, impressive degrees of self-injury can occur in PNES as well. The frequent occurrence of fractures in patients with PNES are in fact evidence for the capacity of some PNES patients for self-inflicted injury and discomfort. This capacity is perhaps explained by the tolerance of pain conferred by dissociation [15], which is one of the most common psychological mechanisms in PNES. Autonomic changes can occur with epileptic seizures, but also with PNES (e.g. coughing, palpitations and pupillary dilatation). Many of these signs are simply part of the generalized arousal response attached to panic or other extreme emotional states [23], [40].

Many studies have reported that the clinical diagnosis and careful history making is still essential for the differential diagnosis. The following symptoms have found to distinguish epileptic seizures from PNES:

Although such seizure characteristics can be important, Reuber and Elger [7] warn that often the past medical, social and psychiatric history is more helpful in the differentiation of PNES from epilepsy than the description of the seizures themselves. Others warn that a history with clear psychogenic factors does not protect against developing organic problems such as epilepsy [48]. Therefore such a positive history only has meaning in the context of excluding medical causes for the seizures [49]. Wagner et al. [50] describe the use of a psychological test (the Personality Assessment Inventory, PAI) as an effective psychological screening tool to aid in the differential diagnosis of epileptic seizures versus PNES prior to hospital admission for VEEG. Cragar et al. [17] conclude that, despite a plethora of research on methods for differentiating PNES from epilepsy, seizure recording with simultaneous video EEG monitoring remains the gold standard for differential diagnosis. However, some studies report that despite the most modern investigatory equipment it sometimes may be impossible to decide what one is dealing with [12], [45], [51]. Parra et al. [36] state that during the initial years of video EEG monitoring neurologists were more prone to misdiagnose PNES as epilepsy [52], [46], but in the last few years clinicians are more likely to suspect and recognize PNES but also ‘overdiagnose’ atypical paroxysmal events as PNES. In their own study they also found that epileptic seizures were misdiagnosed as PNES more frequently than the reverse (57% versus 12%). A worrisome observation is that, despite the current available technical facilities, the mean latency between onset of PNES and final diagnosis as being non-epileptic and psychogenic is approximately 7 years [7], [9], [37], [53]. Müller et al. [16] therefore conclude that early admission of so-called pharmacoresistent epilepsy to an epilepsy centre (establishing a standard work-up and clarifying the medical terminology) will improve diagnosis (and lead to adequate therapy of PNES as well prevent unnecessary drug treatment). They describe a diagnostic algorithm in which video EEG and clinical observation as well as a close review of the general medical and drug history must be considered as a minimal diagnostic standard. One of the reasons for diagnostic delay is outlined by Dekkers and Domburg [54]: the medical diagnosis of PNES is often limited to a ‘negative’ process and consequently PNES is characterized as a ‘non-disease’ (i.e. ‘not epilepsy’, ‘no cardial disease’, ‘not an anxiety attack’, etc.), which may obstruct a ‘positive diagnosis’, evaluating the exact psychological mechanisms that have caused PNES in an individual patient.

The psychological diagnosis is thus an important, although not a conclusive, ‘second phase’ aspect of medical decision making in which the focus will gradually shift to etiological factors. Iriarte et al. [10] (see also Mökleby et al. [55]) emphasize that the diagnosis of PNES requires a multidisciplinary approach. It is not enough to determine that the seizures have a non-epileptic origin, but also to obtain a comprehensive clinical psychological, psychiatric and neuropsychological evaluation that may shed light on potential psychogenic mechanisms of PNES [56], [57], [58]. Bowman and Markand [59] advise that after PNES are confirmed a complete assessment should be carried out, requiring at least three components: (a) screen for co-existent neurological disorder; (b) psychological examination; (c) assessment of social/interpersonal problems. Three general types of psychological tests have been used to assess PNES: personality inventories, neuropsychological tests and forced choice malingering detection tests. They are useful methods, and each probably contributes to obtaining unique information that can identify patients who should be monitored. Also multiple measures should provide a more accurate prediction of the seizure type than single measures [23], [60].

Some authors use a classification system for the clinical manifestations of PNES such as the DSM-IV or ICD-10 [20], [55], [61]. Seizures are then diagnosed as either ‘dissociative disorders’ (ICD-10) [62] or on the DSM on either axis I, or axis II or both. The most frequent DSM-IV diagnosis for PNES appears to be somatoform disorder (conversion disorder) [24], [63]. The second most common diagnosis was anxiety disorder [5], [64]. It is questionable though, whether the seizures should be ‘relabeled’, since these diagnostic categories on their own tell us comparatively little about the management of patients with PNES [65]. Wilkus et al. [66] even used MMPI profiles to distinguish epilepsy form PNES and report 80% success score. This has never been confirmed, though. Moreover Arnold and Privitera [67] could not distinguish patients with epilepsy and patients with PNES using the DSM classification in a double blind study.

Induction protocols are controversial (e.g. suggestive techniques such as tuning fork or intravenous injection of normal saline solution; Dericioglu et al. [68]) because it may negatively influence the patient–physician relationship, but also because induction tests can also induce epileptic seizures or can even cause PNES in patients who might not have had PNES before [10], [16], [44]. The diagnostic accuracy of serum prolactin levels is less than that of provocation with suggestion during video EEG monitoring [15], [69], [70]. Although Chen et al. [71] recommend that elevated serum prolactin assay, when measured in the appropriate clinical setting (at 10–20min after a suspected event) is a useful adjunct for the differentiation of generalized tonic–clonic or complex partial seizure from PNES among adults and older children.

As mentioned before, the coexistence of epileptic and non-epileptic seizures in the same patient presents a specific diagnostic challenge. Some studies have found that there may be relationships between the two types of seizures, such as symptom modeling and the susceptibility to behavioral dysfunction conferred by neurological illness and behavioral toxicity of AEDs [15]. Two types of seizures may be especially difficult to differentiate: frontal seizures [12], [39] and minor non-convulsive psychogenic seizures which simulate complex partial epileptic seizures [1]. Bazil et al. [72] suggest that monitoring the sleep structure may be an helpful additional aid in the differential diagnosis. Patients with PNES have a sleep architecture similar to patients with major depression with an increased percentage of REM sleep.

3.4. Seizure characteristics 

Although PNES can imitate any type of seizure event [5], Abubakr et al. [8] suggest a division into two seizure types: (a) major motor manifestations and (b) limpness, unresponsiveness, flaccidity. Sometimes this division is made in parallel with the epileptic seizures that they resemble: ‘Major’ seizures are defined as attacks that resemble most convulsive-type seizures such as the generalized tonic or clonic seizures. ‘Minor’ seizures resemble absence-like or short partial seizures [1]. A similar division is proposed by Riggio [24], distinguishing between: a primary motor component type and the second non-motor type in which the event is often characterized by a change in behavior. Meierkord et al. [73] report that 66% of the patients presented with excessive motor manifestations and unresponsiveness is a more seldom reported symptom. This is confirmed by McDade and Brown [74]. Leis et al. [2], however, caution against such conclusions as unresponsiveness may go undetected more frequently. Betts and Boden [12], [75] distinguish three types of seizures: (a) ‘Swoons’: a relaxed fall to the ground without injury, followed by lying still without convulsion, with eyes closed for various periods of time and apparently unconscious, usually followed by rapid recovery with no post-ictal confusion; (b) ‘Tantrums’: the patient emits a cry, falls, thrashes about with a convulsive struggle if restrained, kicks and may bite (either himself of spectators), and is often noisy, shouting, roaring or crying; (c) ‘Abreactive attacks’; there may be initial over-breathing, which may pass unrecognized, followed by sudden movement and stiffening of the body, which is followed by breath holding, gasping, uncoordinated jerking of the body with pelvic thrusting and back arching. There may be screaming or crying, spitting or retching. It bears superficial resemblance to sexual intercourse and may continue for many hours. Alper [15] distinguishes PNES that are conversion symptoms (conversion PNES) from PNES that are non-conversion and are paroxysmal behavioral features of other syndromes (non-conversion PNES). Another important distinction proposed by Alper [15], [35] is that between consiousness/unconsiousness and intentional/nonintentional. This distinction leads to the following subgroups: (a) PNES as conversion disorder: the patient is not consciously aware of intentionally producing symptoms or of the presumed unconscious conflict or unmet need underlying their occurrence. (b) PNES as factitious disorder: patients are consciously aware of intentionally producing symptoms but not consciously aware of their reason for pursuing the sick role as a dominant motivating theme of their lives. (c) Non-epileptic seizures in the context of malingering: the patient is consciously aware of intentional symptom production and clearly understands his or her agenda, which is specifically identifiable in external incentives. Intentional PNES production distinguishes these disorders from other axis I disorders, which may occasionally manifest unintentional paroxysmal behavioral features that may be suspected of being epileptic seizures, such as anxiety, dissociative or impulse control disorders.

Some specific relationships were reported between seizure presentation and underlying psychological mechanisms. Meierkord et al. [73] report that pelvic thrusting can occur as prominent feature only in women with a history of childhood sexual abuse. Galimberti et al. [26] report that PNES mimicking generalised tonic–clonic seizures was associated with a low educational level. Ramchandani and Schindler [76] found that patients with pseudo complex partial seizures had dissociative symptoms, whereas patients with pseudo tonic–clonic seizures had their illness developed in the context of longstanding personality disorders. Alper [15] emphasize the importance of distinguishing the subgroup with non-conversion PNES as it may allow identification of a discrete underlying psychiatric disorder, whose specific treatment differs from that of conversion PNES. This is particularly true for axis I disorders that allows specific pharmacotherapeutic approaches such as panic disorder or schizophrenia. Moore and Baker [77] observed that many of the symptoms described by their patients appear to be anxiety related. Holmes et al. [25] found a relationship between seizure presentation and gender. Men were more likely than women to have ‘convulsive-like’ psychogenic seizures, while women were more likely to have ‘non-convulsive’ attacks. This semiological difference is attributed to ‘a more male form of acting out’ during forms of PNES associated with dramatic motor activity.

3.5. Psychological etiology 

Freud saw PNES as a ‘hysterical fit’, being the expression of normal sexual desires, repressed from conscious awareness due to the unbearable affects with which they had become associated. The phenomenological similarity of many patients’ seizures to movements common in the context of the sexual act was interpreted as evidence that the seizure was the symbolic expression of a repressed unconscious sexual conflict [4], [15], [58]. After the abolition of the term ‘hysterical neurosis’ from the current diagnostic systems, PNES is no longer seen as one discrete disorder (‘hysteria’) and the term ‘hysterical seizures’ [19], [78] vanished from the literature, although a survey among British neurologists showed that in 1991 the term hysteria was still used in the ‘informal contacts’ [79]. Sirven and Glosser [23] conclude that a satisfactory etiologically referenced classification system has been elusive because of many current problems, of which the issue of homogeneity is dominant. PNES are almost infinitely heterogeneous and are quite different from person to person. Even if the PNES behaviors of very different people are morphologically similar, clinical experience reveals that the psychogenic causes may be quite divergent [20], [44], [80].

It is helpful when describing psychological etiology also to define the used terms, because one of the major problems is the incoherent use of terminology:

A wide range of psychogenic factors have been identified that may underlie the occurrence of PNES in individual patients: a history of sexual and physical abuse and post-traumatic stress disorder, malingering, depression or chronic anxiety, dissociation, somatization disorder, including Briquet's syndrome, behaviorally oriented concepts of secondary gain and assumption of the sick role (mainly in intellectually impaired persons), personality disorders, organicity and age [8], [12], [15], [27], [55], [61], [63], [75], [77]. Elsewhere [83], we propose a model distinguishing 5 different levels. This model is based on the previous mentioned factors specifically found for PNES. (a) Psychological etiology: factors that are involved in the causation of PNES, such as sexual abuse; (b) Vulnerability: factors that predispose a person to develop psychosomatic symptoms, such as PNES. Many authors have pointed to the specific vulnerability of patients with PNES both in terms of the emotional ‘make-up’ and their neuropsychological functioning; (c) Shaping factors: factors that can specifically shape the symptoms in the form of ‘seizures’ (in contrast to for example movement disorders or ‘headache-like symptoms’). A shaping factor may be a relative with epileptic seizures (symptom modeling); (d) Triggering factors: create circumstances or situations that provoke PNES such as factors that refer to primary gain. Also psychological mechanisms that transfer an emotional state into a seizure can be part of these triggering factors such as dissociation; (e) Prolongation factors: the previous four factors are specifically important in the development of PNES. Prolongation factors are important in explaining why the seizures persist. Examples are the coping strategy of a patient or secondary gain aspects. This model resembles other models used to explain PNES or other somatisation disorders [84], [85], [86].