Psychological and behavioural treatments for adults with non‐epileptic attack disorder

Abstract

Background

Psychogenic non‐epileptic seizures, also known as non‐epileptic attack disorder (NEAD), have the outward appearance of epilepsy in the absence of physiological or electroencephalographic correlates. Non‐epileptic seizures can occur in isolation or in combination with epileptic seizures. The development and maintenance of non‐epileptic seizures has been well documented and there is a growing literature on the treatment of non‐epileptic seizures which includes non‐psychological (including anti‐anxiety and antidepressant pharmacological treatment) and psychological therapies (including cognitive behavioural therapy (CBT), hypnotherapy and paradoxical therapy). Various treatment methodologies have been tried with variable success. The purpose of this Cochrane review was to establish the evidence base for the treatment of non‐epileptic seizures with behavioural and psychological therapies only.

Objectives

To assess whether behavioural or psychological treatments for non‐epileptic seizures or NEAD result in a reduction in the frequency of seizures or improvement in quality of life, or both, and whether any treatment is significantly more effective than others.

Search methods

We searched the Cochrane Epilepsy Group's Specialised Register (4 February 2013), the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2013, Issue 1) (January 2013), MEDLINE (1946 to 4 February 2013), PsycINFO (4 February 2013) and SCOPUS (4 February 2013). No language restrictions were imposed. We checked the reference lists of retrieved studies for additional reports of relevant studies

Selection criteria

Randomised controlled trials (RCTs) and before and after controlled and non‐controlled studies were eligible for inclusion. Studies were required to assess one or more types of behavioural or psychological interventions, or both, for the treatment of non‐epileptic seizures. Studies of childhood non‐epileptic seizures were excluded from our review.

Data collection and analysis

Two review authors (JM, JP) independently assessed the trials for inclusion and extracted data. Outcomes included reduction in seizure frequency and improvements in quality of life.

Main results

Twelve studies, with a total of 343 participants, met our inclusion criteria (four RCTs and eight before and after non‐controlled studies). Of the four RCTs, one examined patients with non‐epileptic seizures and three had a mixed diagnosis (pseudoseizures, conversion disorder and somatisation disorder). Most of the non‐randomised studies used non‐epileptic seizure patients exclusively. Overall, five studies examined the effectiveness of psychotherapy, three examined CBT, two investigated hypnosis, one assessed paradoxical intention and one had a mixed intervention design. We classified two included studies as low risk of bias, one as unclear and nine as high risk of bias. Meta‐analysis could not be undertaken due to the heterogeneity of design and interventions. Most included studies reported improved outcomes for the intervention under investigation. One RCT investigating the effectiveness of CBT in this patient group found a significant reduction in seizure frequency compared to controls (P < 0.001).

Authors' conclusions

There is little reliable evidence to support the use of any treatment, including CBT, in the treatment of non‐epileptic seizures. Further randomised controlled trials of CBT and other interventions are needed.

Plain language summary

Psychological and behavioural treatments for non‐epileptic attack disorder

Non‐epileptic attacks look like epileptic seizures but they are not caused by epilepsy. There have been many investigations of the causes, but evidence about successful treatment is less available. We reviewed the existing studies treating people having non‐epileptic attacks. We found 12 studies looking at different types of treatment such as psychotherapy, cognitive behavioural therapy and hypnosis. Three hundred and forty three participants were recruited to these 12 studies. Four studies were randomised controlled trials, and 8 were non‐randomised studies. Most of the controlled trials included patients with other diagnoses as well as non‐epileptic attacks. Most of the non‐randomised studies included patients with mainly non‐epileptic attacks. Most included studies reported improved outcomes for the treatment they were investigating. One randomised trial investigating Cognitive Behavioural Therapy found that seizures were significantly reduced. Due to the variety of treatments and designs of the included studies, it was not possible to combine the results to produce an overall outcome for our review.

Many of the studies did not use satisfactory methods which meant that the evidence was rated as high risk of bias. The overall evidence for the main outcome of reducing seizures as a result of treatment is not considered reliable except in one study. Our conclusion is there is little reliable evidence to support the use of any treatment for people with non‐epileptic attacks. The evidence in this review is up to date as from 4 February 2013.

Summary of findings

Summary of findings for the main comparison. RCT evidence for people with non‐epileptic attacks.

Behavioural treatments compared with control interventions for people with non‐epileptic attack disorder
Patient or population: people with non‐epileptic attack disorder Settings: hospital Intervention: behavioural treatments Comparison: control intervention
Outcomes	No of participants (studies)	Quality of the evidence (GRADE)	Comments
Seizure frequency: > 50% seizure reduction	66 (1)	⊕⊕⊕⊕ high	Only one study compared a behavioural treatment (CBT) to a control intervention (standard medical care). This study was a well‐conducted RCT (no evidence of bias) and reported a significant decrease in seizures in the CBT intervention group compared to the standard medical care group. Study reports median monthly seizure frequency; no dichotomous data available to provide an overall effect estimate for the outcome of 50% seizure reduction.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. RCT: randomised controlled trial

Summary of findings 2. Non‐randomised evidence for people with non‐epileptic attacks.

Behavioural treatments for people with non‐epileptic attack disorder
Patient or population: people with non‐epileptic attack disorder Settings: hospital Intervention: behavioural treatments
Outcomes	No of participants (studies)	Quality of the evidence (GRADE)	Comments
Seizure frequency: > 50% reduction in seizures	120 (6)	⊕⊝⊝⊝ very low	Providing an overall effect estimate from the non‐randomised evidence for the effectiveness of behavioural interventions in people with NEAD is problematic due to the heterogenous nature of the studies. Studies investigated differing behavioural treatments and patients receiving a behavioural intervention were often receiving other forms of therapy simultaneously. Therefore we rated the quality of evidence provided by these studies as very low.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. NEAD: non‐epileptic attack disorder

Summary of findings 3. RCT evidence for people with conversion disorder.

Behavioural interventions compared with control interventions for people with conversion disorder
Patient or population: people with conversion disorder Settings: hospital Intervention: behavioural interventions Comparison: control interventions
Outcomes	No of participants (studies)	Quality of the evidence (GRADE)	Comments
Seizure frequency: > 50% reduction in seizures	128 (3)	⊕⊕⊕⊝ moderate	Evidence for this outcome downgraded once for heterogeneity between studies due to differences in intervention investigated. On the whole we rated all three studies as low/unclear risk of bias due to lack of detailed methodology reported in the papers.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

Summary of findings 4. Non‐randomised evidence for people with conversion disorder.

Behavioural interventions compared with control interventions for people with conversion disorder
Patient or population: people with conversion/dissociative/hysterical seizures Settings: hospital Intervention: behavioural interventions Comparison: control interventions
Outcomes	No of participants (studies)	Quality of the evidence (GRADE)	Comments
Seizure frequency > 50% seizure reduction	29 (2)	⊕⊕⊕⊝ low	The evidence for the effectiveness of behavioural interventions used for conversion disorder patients is low quality due to the inconsistency between study interventions and certain aspects of risk of bias
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

Background

Description of the condition

There have been many names given to this condition over the years. In this review we refer to non‐epileptic seizures or NEAD, but such attacks have variously been called pseudo seizures, psychogenic seizures, somatoform disorders, dissociative seizures, conversion disorders and hysterical seizures, to name but a few. Whatever terminology is used, they are describing attacks which look very similar to seizures caused by epilepsy, but they do not have the abnormal electrical activity in the brain that characterises epilepsy. Instead they are unconsciously produced, beyond the control of the person having the attacks, in response to overwhelming distress and can be triggered by many situations, emotions, thoughts, etc. People with non‐epileptic seizures are not always aware of what triggers their attacks and may say they "just happen". Psychologists may formulate the development and maintenance of the attacks in terms of emotional trauma or stress but identifying a particular causative factor may not always be straightforward.

There are various sources citing the incidence of non‐epileptic seizures. The NEAD Trust website states that the incidence of NEAD is approximately three people per 100,000 per year in England alone. However, they add that these data were obtained from neurology centres and only video/electroencephalogram (EEG)‐proven cases were counted, therefore the figure is likely to represent a significant underestimate. The same source indicated that on average it took five years for patients to be correctly diagnosed with non‐epileptic seizures. The website Non‐Epileptic Attacks cites a higher incidence rate of two or three people per 10,000. More recent papers have indicated a general population incidence from 1.5per 100,000 persons per year (Bodde 2009) to 4.6 per 100,000 per year (Szaflarski 2000). In the UK specifically 4.9 per 100,000 has been reported (Duncan 2011). In addition, it is indicated that 25% to 30% of patients referred to tertiary epilepsy centres for refractory epilepsy are found to have non‐epileptic seizures. Of these, 5% to 40% have a dual diagnosis of epilepsy or have a past history of epilepsy (Bodde 2009). The prevalence of non‐epileptic seizures has been cited to be between 5% and 20% in the outpatient epilepsy population (Buchanan 1993).

Description of the intervention

There are many interventions that have been tried for non‐epileptic seizures. It is unclear as yet whether one particular intervention is suitable for all people given that these seizures occur in many different contexts and develop in response to many different situations.

Behavioural and psychological treatments have been used to help people with non‐epileptic seizures with varying levels of documented success. Some of the most common forms of intervention include the following:

• Cognitive behavioural therapy (CBT): aims to identify and link thoughts, emotions, physical sensations and behaviour. The therapist works with the patient to address difficulties such as thinking errors to help the person deal with problems more effectively. One of the key components is self monitoring (completing behavioural diaries) and analysing these to look for alternative ways of thinking about or managing the situation.

• Psychodynamic therapy: this approach looks at the impact of early life events on the way people view themselves today, their current relationships and how they handle emotions. Difficult experiences from the past may be repressed and people may not realise how they have been affected by the experiences until they start thinking about it in therapy.

• Cognitive analytical therapy: aims to help people identify unhelpful patterns of actions and how they have developed over time. Links are made between childhood behaviour patterns and current behaviour and thoughts.

• Interpersonal therapy: aims to identify the connections between symptoms experienced and current interpersonal problems in key relationships. These interpersonal difficulties are seen as central to the development and maintenance of the symptoms. Therapy focuses on current social relationships and how expectations within these relationships may be causing a patient’s symptoms.

• Family therapy: involves working with several members of a family at the same time. Relationships within the family unit are the focus, as are the ways in which the family views and solves problems ‐ or makes them worse.

• Hypnotherapy: this approach involves the person experiencing deep relaxation and their attention is focused on the suggestions made by the therapist. These suggestions are made with the aim of helping people make positive changes within themselves and in the behaviour outside sessions.

• Paradoxical therapy: the therapist encourages the patient to engage in the unwanted behaviour, in this case doing something to deliberately bring on a non‐epileptic seizure.

How the intervention might work

Much research has been devoted to the aetiology of non‐epileptic seizures and it is from this that some of the hypotheses about treatment methodologies arise. Significant links have been found between certain life events and the occurrence of non‐epileptic seizures. In particular there have been consistent findings of co‐occurrence of experiences of abuse in childhood or early adulthood and the development of non‐epileptic events (Francis 1999a; Reilly 1999). An understanding of how these early life events contribute to the development of maladaptive behaviours that are triggered in later life is seen as crucial in some treatment methodologies (e.g. psychodynamic) and less so in others (e.g. CBT). Psychological factors such as anxiety, stress, anger and other emotions, as well as mental tasks and thoughts (Francis 1999b), can trigger an attack, as can physiological states such as over‐exertion and pain (Stone 2004). It is on the basis of such studies that much literature has evolved regarding potential treatments. For example, eye movement desensitisation and reprocessing (Chemali 2004) is thought to ameliorate symptoms by re‐processing the memory of the event so that it no longer feels as bad as the first time it was experienced. Following eye movement desensitisation and reprocessing, the person still remembers the event, but the associated images, sounds and feelings that were evoked are no longer triggered to the same extent. CBT, on the other hand, is usually more focused on the link between thoughts, emotions, physical sensations and behaviour (Rusch 2001). Therefore, if the presenting symptoms were anxiety or panic the patient would be treated with CBT by looking at strategies such as thought challenging and exposure of the feared thoughts, memories or situations, perhaps in a graded way. It is hypothesised that the way we interpret these events or situations can be modified by challenging negative automatic thoughts that arise in these contexts and replacing such thoughts with a more appropriate response.

Hypnosis or deep relaxation allows the conscious mind to rest and go into a neutral state. By doing so, it allows communication with the subconscious mind directly. This process is powerful because the subconscious mind can create positive changes by accepting positive suggestions. Through this communication with the subconscious mind, hypnotherapy can help the patient to release negative feelings and beliefs that may restrict their ability to achieve their potential to live life to the full.

Other methodologies are thought to work by different mechanisms, e.g. interpersonal psychotherapy involves resolving relationship problems or finding new relationships or activities as compensation.

Why it is important to do this review

This review is important to inform clinicians of the best, most effective and evidenced treatment options for their patients.

Objectives

To assess whether behavioural or psychological treatments for non‐epileptic seizures or NEAD result in a reduction in the frequency of seizures or improvement in quality of life, or both, and whether any treatment is significantly more effective than others.

Methods

Criteria for considering studies for this review

Types of studies

1. Randomised controlled trials (RCTs) or quasi‐randomised studies (e.g. where randomisation is according to the day of the week or date of birth). The studies may be single or double‐blind or unblinded.

2. Before and after studies with or without a control group (outcome measures taken pre‐ and post‐intervention).

Types of participants

Adult male or female with any type of non‐organic, non‐epileptic seizures, with or without learning disabilities.

Types of interventions

Any behavioural or psychological intervention, such as use of cognitive behaviour therapy, relaxation therapy, biofeedback, counselling, hypnotherapy, conditioning, physical therapies, massage or aromatherapy.

Types of outcome measures

Primary outcomes

Seizure reduction

• Fifty per cent or greater reduction in seizure frequency

• Seizure freedom

• Percentage change in seizure frequency

Secondary outcomes

• Quality of life

• Seizure severity ‐ provided a standardised and validated scale is used (for example, Quality Of Life in Epilepsy (QOLIE‐31))

Search methods for identification of studies

Electronic searches

We searched:

• the Cochrane Epilepsy Group's Specialised Register (4 February 2013) using the search strategy outlined in Appendix 1;

• the Cochrane Central Register of Controlled Trials (CENTRAL 2013, Issue 1, The Cochrane Library, January 2013) using the search strategy outlined in Appendix 2;

• MEDLINE (Ovid, 1946 to 4 February 2013) using the search strategy outlined in Appendix 3;

• PsycINFO (4 February 2013) using the search strategy outlined in Appendix 4; and

• SCOPUS (4 February 2013) using the search strategy outlined in Appendix 5.

We did not impose any language restrictions.

Searching other resources

We reviewed the reference lists of included studies to search for additional reports of relevant studies.

Data collection and analysis

Two review authors (JM and JP) independently assessed studies for inclusion, with disagreements resolved by mutual discussion.

The same review authors independently extracted the following data and again any differences of opinion were resolved by mutual discussion:

1. Study methods:

   1. Study design (e.g. RCT, non‐randomised, parallel or cross‐over design)

   2. Randomisation method (allocation concealment and list generation)

   3. Blinding method

   4. Duration (treatment and follow‐up period)

2. Participants:

   1. Number (total/per group)

   2. Age and sex distribution

   3. Exclusions and withdrawals

   4. Type of seizures

   5. Duration of attack disorder

   6. Aetiology of attack disorder

   7. Presence or absence of learning disability

3. Type of intervention and control

   1. Psychological components

   2. Mode of delivery

   3. Setting

4. Results

   1. Seizure data

   2. Quality of life data

Selection of studies We managed search results using EndNote referencing software where examination for duplicates took place. Two authors reviewed the titles and abstracts of articles highlighted by the searches and removed studies that obviously did not meet the inclusion criteria. The two authors used full text reports to determine eligibility in cases not obvious from the abstract. We contacted corresponding authors of reports to clarify eligibility where necessary. We discussed any disagreements and if not resolved we sought the opinion of a third author. Eligibility for inclusion in this review was based on study design, treatment and outcome.

Selection of studies

We managed search results using EndNote referencing software where examination for duplicates took place. Two authors reviewed the titles and abstracts of articles highlighted by the searches and removed studies that obviously did not meet the inclusion criteria. The two authors used full text reports to determine eligibility in cases not obvious from the abstract. We contacted corresponding authors of reports to clarify eligibility where necessary. We discussed any disagreements and if not resolved we sought the opinion of a third author. Eligibility for inclusion in this review was based on study design, treatment and outcome.

Data extraction and management

We extracted data using paper‐based data extraction forms specifically designed for the review. Data extracted is outlined in Data collection and analysis. Two authors piloted the form initially and made amendments where necessary. One review specialist and a specialist in the review subject area completed data extraction.

Assessment of risk of bias in included studies

For RCT evidence, each study underwent a 'Risk of bias' assessment using the Cochrane 'Risk of bias' tool. The tool examines selection bias (sequence generation, allocation concealment), performance bias (blinding), attrition bias (incomplete outcome data, blinding), detection bias (blinding, other potential threats to validity) and reporting bias (selective outcome reporting).

For non‐randomised evidence, we assessed each study for bias using the extended Cochrane 'Risk of bias' tool developed by the Cochrane Non‐Randomised Studies Methods Group. This tool assesses all domains mentioned above with the addition of confounding variables. We rated blinding, incomplete outcome data, selective outcome reporting, confounding variables and other bias on a five‐point scale ranging from low to high risk of bias according to the risk for the outcome. The review authors determined the parameters of the scale. See Appendix 6 for a table of scale parameters.

We made an overall summary judgement of risk of bias for each study followed by an overall judgement across studies. We could not incorporate the 'Risk of bias' judgements into the analysis using sensitivity analysis as meta‐analysis was not undertaken. We created four 'Summary of findings' tables, one consisting of RCT evidence and one non‐randomised evidence. We graded each outcome accordingly using the GRADE approach.

Measures of treatment effect

The primary outcome of seizure reduction and the secondary outcome of quality of life is presented for each study as it was reported in the individual study papers. We judged the studies included in the review to have high clinical heterogeneity therefore no meta‐analysis was undertaken. Results presented within this review are discussed in a narrative format.

Unit of analysis issues

We expected quality of life measures to be heterogenous in which case we would have used the standardised mean difference. However, as no meta‐analysis was undertaken there was no requirement to undertake this analysis.

Assessment of heterogeneity

We assessed clinical heterogeneity by examining the differences in study characteristics in order to inform decisions regarding the combination of study data. Sources of clinical heterogeneity hypothesised a priori were population differences (learning difficulties), differences in treatments and loss to follow‐up. As mentioned above, studies had high clinical heterogeneity due to differences between intervention techniques and measures used to assess quality of life.

Assessment of reporting biases

We investigated outcome reporting bias using the ORBIT classification system (Kirkham 2010) We requested all protocols from study authors to enable comparison of outcomes of interest, however few protocols were supplied.

We addressed publication bias by carrying out a comprehensive search of multiple sources to identify unpublished data and by requesting any unpublished data from authors.

Data synthesis

We synthesised data in a narrative format as we considered meta‐analysis to be inappropriate given the differences across studies in intervention characteristics and outcome measures. Data provided under each comparison listed below were minimal and could not be combined across outcomes.

Comparisons we planned to carry out included:

• intervention versus control (no treatment or usual care);

• intervention versus other intervention.

Results

Description of studies

Results of the search

The search revealed 987 records identified from the databases outlined in Electronic searches. We identified 26 further records through the reference lists of included studies. After duplicates were removed, 649 records remained. We screened all of them for inclusion in the review. We excluded 603 studies at this point leaving 46 full‐text articles to be assessed for eligibility. Following this we excluded 34 texts (see Figure 1 and Characteristics of excluded studies for reasons for exclusion). We included 12 studies (12 reports) in a narrative synthesis.

1.

Study flow diagram.

Included studies

Cognitive behavioural therapy (CBT) studies

Three studies investigated the effects of CBT on seizure frequency (Goldstein 2004; Goldstein 2010; LaFrance 2009).

Goldstein 2004 and colleagues evaluated the effectiveness of CBT for adults with dissociative seizures using a before and after design. Twenty patients were recruited consecutively from the Maudsley Hospital, London. Inclusion criteria included experience of at least one characteristic attack during EEG. Only 16 completed the treatment (14 female; mean age 34.9, standard deviation (SD) 13.4). Participants had 12 sessions of weekly, individual CBT; sessions lasted one hour except for the initial session which was allocated two hours. The experienced CBT therapist who delivered the treatment followed a detailed, session‐by‐session treatment schedule prepared for the study. Outcome measures were administered pretreatment, at discharge and at six‐month follow‐up.

Goldstein 2010 and colleagues used a randomised controlled trial to compare CBT to standard medical care for people with psychogenic non‐epileptic seizures. Sixty‐six participants were recruited from the South London and Maudsley NHS Foundation Trust, London. Inclusion criteria included diagnostic confirmation by video‐EEG telemetry in most patients. Patients were randomised to CBT or standard medical care. The CBT treatment manual was based on the previously reported 2004 study. Patients were offered up to 12 weekly or fortnightly hour‐long outpatient sessions with a trained therapist. Thirty‐three participants were allocated to each group; of these 31 received standard medical care (two withdrew post‐randomisation) and 24 received at least nine CBT sessions (nine participants received fewer than nine sessions). Three of the CBT group patients were lost to follow‐up. An intention‐to‐treat analysis captured 33 data sets for the CBT group (24 females, mean age 37.4, SD 12.6) and 31 for the standard medical care group (26 females, mean age 35.9, SD 15.1).

LaFrance 2009 evaluated the effect of CBT on reduction of psychogenic non‐epileptic seizures (PNES) using a before and after design. Twenty‐one participants (17 female, 81%; mean age 36, SD 10.4) with video/EEG‐confirmed PNES were enrolled into the Rhode Island Hospital study and 17 completed the 12 sessions of manualised CBT intervention. Three people with mixed epileptic seizures and PNES who could clearly distinguish between their events were included in the group. One of the 21 participants was not included in the modified intention‐to‐treat analysis as they dropped out the day after completing the baseline measures and so did not receive any intervention. Twenty were therefore included in the treatment analyses. Patients recorded their seizures on a daily seizure calendar in one‐week intervals. Secondary outcome measures included the Beck Depression Inventory II (BDI), Modified Hamilton Rating Scale for Depression (MHRSD), Davidson Trauma Scale (DTS) and the Quality of Life in Epilepsy 31 (QOLIE). After completing the CBT sessions, a follow‐up questionnaire was administered via the phone at four, eight and 12 months post‐enrolment.

Hypnosis studies

Two randomised controlled trials investigated the effects of hypnosis on conversion symptoms (Moene 2002; Moene 2003).

One study investigated the effects of hypnosis on non‐epileptic seizures (Moene 2002). This study included 49 adult inpatients with conversion disorder. Seven experienced paroxysmal myoclonic outbursts and eight had seizures or convulsions. Twenty‐six were allocated randomly to the experimental group and 23 to the control group. The average duration of conversion disorder was 3.9 years (range two months to 22 years; SD 4.5 months). Forty per cent (18) of the 45 had acute onset (within three days). Sixty‐two per cent (28) (an undisclosed proportion of whom had acute onset) developed conversion symptoms in connection with a previous physical complaint or affliction in the relevant part of the body. Thirty‐two (72.7%) of the patients used medication (benzodiazepines (43.2%), medication for somatic complaints (45.5%) and painkillers (44.4%) most frequently). No discussion of withdrawal of medication or otherwise was included. Patients in both treatment and control groups followed a group therapy programme with the aim of increasing problem‐solving skills. This programme consisted of group psychotherapy, social skills training, formulating and evaluating treatment goals, creative therapy and sports. In addition, conversion patients also had physiotherapy, individual exercise sessions and bed rest. The experimental group had the addition of hypnotic treatment. This involved one preparatory session followed by eight weekly sessions lasting one hour. Part of this treatment involved the patients learning self hypnosis and they were instructed to practise each day for 30 minutes. In the control group, instead of hypnosis, a treatment aimed at optimising non‐specific or common therapy factors was implemented. This involved a preparatory session followed by eight weekly sessions lasting one hour.The primary outcome measure used was the Video Rating Scale for Motor Conversion Symptoms (VRMC). Follow‐up in this study was eight months after pretreatment assessment.

A further study by Moene 2003 included 49 adult outpatients with conversion disorder (motor type) or somatisation disorder (with motor conversion symptoms). Of the 49, six experienced paroxysmal myoclonic outbursts and two had seizures or convulsions. Twenty‐four were randomly allocated to the treatment group and 25 to the waiting list control group. The treatment group received one preparatory session followed by 10 weekly sessions. Two hypnotic strategies were used: (1) direct symptom alleviation used suggestions designed to alter conditioned cues to motor symptoms; (2) emotional expression/insight involved age regression to explore factors implicated in the development of the symptoms. Self hypnosis was also taught and patients were asked to practise the symptom alleviation strategies for 30 minutes a day. A waiting list control group was used. Average duration of conversion disorder was 3.7 years (range two months to 16.7 years; SD 4.7 months). In terms of onset, 37.2% (16) patients had acute onset (within three days) and 47.8% (21) developed conversion symptoms in connection with a previous physical complaint or affliction in the relevant part of the body. Twenty‐one (48.8%) of the patients used medication (benzodiazepines (14%), medication for somatic complaints (30.2%) and painkillers (34.8%) most frequently). No discussion of withdrawal of medication or otherwise was included. The primary outcome measure used was the VRMC. Follow‐up for the treatment group was at six months after their 10th session. No follow‐up was arranged for the waiting list control group. It is noted that preliminary data on eight adults with motor conversion symptoms (one described as having pseudo epileptic seizures) was published by Moene in 1998 (Moene 1998).

Paradoxical intention therapy studies

We found one randomised controlled clinical trial investigating the effects of paradoxical intention therapy on non‐epileptic seizures (Ataoglu 2003). This study included 30 adults with conversion disorder, specifically pseudoseizures. Fifteen were randomly allocated to the experimental group and 15 to the control group after exclusions for abnormal EEG, organic disease, previous psychiatric treatment etc. In the experimental group, patients were hospitalised and given two paradoxical intention treatment sessions per day. During sessions, patients were encouraged to imagine anxiety‐provoking situations or experiences, or both. The aim was to help the patients to re‐experience their traumas and experience their conversion attacks. After three weeks, patients were discharged. Three weeks post‐discharge, a reassessment took place and comparisons of anxiety and conversion scores were made. In the control group, patients were prescribed diazepam as outpatients (5 to 15 mg). They were given appointments at 10, 20, 30 and 45 days to review their progress and were also reviewed at the end of treatment for anxiety and conversion symptoms. The overall mean duration of conversion disorder for the whole sample was 42 days (experimental group 34 days; control group 48 days). There was no discussion of medication use in these groups. The aetiology was not detailed; however anxiety‐provoking situations and experiences, and traumatic events were mentioned. The primary outcome measure was anxiety score as measured by the Hamilton Rating Scale for Anxiety (HRSA). Follow‐up in this study was three weeks after the study finished. The study duration was six weeks therefore follow‐up was approximately two months after the start of the study. It is noted that an earlier account of this research was published by Ataoglu in 1998 (Ataoglu 1998).

Psychotherapy studies

Five studies investigated the effect of psychotherapy on non‐epileptic seizures (Barry 2008; Kuyk 2008; Prigatano 2002; Ramani 1982; Zaroff 2004).

This pilot before and after study which used psychodynamically focused group therapy for people with psychogenic non‐epileptic seizures (Barry 2008) recruited 12 female patients to the study; seven (aged from 30 to 60) completed at least 75% of the weekly sessions. They met for 90 minutes once a week for 32 weeks. The authors state in their paper that the overall goal of the therapy was to develop conscious and verbal expression of emotional distress which would obviate the need for somatic displays of stress. In addition to the group therapy, five of the participants had intermittent individual cognitive behaviour therapy with one of the authors during the study period. Another patient saw a former therapist sporadically during the study. The primary outcome measures were the Beck Depression Inventory (BDI), the Symptom Checklist‐90 (SCL‐90) and the patient's diary of seizure frequency. The BDI and SCL‐90 were completed by each patient at the start of the study, 16 weeks into group therapy and finally at completion of the 32‐week trial. The seizure logs were completed daily by patients and documented the number of seizures, times they occurred and types of seizures.

Kuyk 2008 looked at the effect of an uncontrolled, prospective inpatient treatment programme for PNES using a before and after design. This treatment was offered in the Epilepsy Institute in Heemsteded, The Netherlands. The treatment programme ran five days per week and was provided by a multidisciplinary team and followed cognitive behavioural principles. At the weekends when the unit was closed, the patients went home. Twenty‐nine patients agreed to participate, of these 26 were included in the study (77% were females; mean age 30.6, SD 10.8). Four of the participants were excluded from the analysis because they had a dual diagnosis and experienced both epileptic seizures and psychogenic non‐epileptic seizures. Sixteen of these were followed up six months post‐treatment. The duration of the programme varied from two to six months with an average of 4.8 months. The main outcome measures included seizure frequency (observed and counted by nursing staff at the start of treatment and at discharge and reported by patients at the six‐month follow‐up), medication use, SCL‐90 and BDI. Mean weekly seizure frequency at the end of treatment and at six‐month follow‐up had decreased significantly compared to the previous measurement.

Prigatano 2002 used a before and after design and provided two six‐month group psychotherapy programmes for 15 patients (all female; mean age 36, range 20 to 48) diagnosed as having non‐epileptic seizures by EEG/video monitoring and brain imaging (one of the patients from the second group was thought to have both epileptic seizures and non‐epileptic seizures). These sessions took place in St Joseph's Hospital and Medical Centre, Phoenix, Arizona. Patients met for 1.5 hours per week for 24 group therapy sessions. Eight participants were treated during the first programme and seven during the second. No quality of life measures were included in this study. Instead the authors looked at various psychometric measures such as sub‐tests from the Wechsler Adult Intelligence Scale 3rd edition (WAIS‐III), California Verbal Learning Test (CVLT) and the Minnesota Multiphasic Personality Inventory‐2 (MMPI‐2). At the beginning of each session, patients completed a weekly log of seizures.

Ramani 1982 investigated the effectiveness of intensive and highly individualised behavioural psychotherapy using a before and after design in nine patients (eight female; average age 31.6, range 17 to 54 years) with video‐EEG confirmation of "hysterical" seizures as well as epileptic seizures. One of the nine was described as "severely retarded". Seven patients received behavioural therapy using supportive re‐educative methods. The "severely retarded" patient received a programme of behaviour modification, while the final patient received short‐term dynamic therapy. All treatments were completed as intensive inpatient treatments during a four to nine‐week hospital stay in the seizure unit at the Comprehensive Epilepsy Program (CEP) of Minnesota, Minneapolis.

Zaroff 2004 used a before and after design to investigate the effectiveness of an open‐ended group psychotherapy programme which included a disorder‐specific psychoeducational treatment component in the first 10 weeks. Ten patients were included (six female; mean age 25.7, SD 12.9, range 21 to 57). Before starting the group therapy sessions, all participants completed various self report measures including seizure frequency and QOLIE‐31. The sessions were one hour per week for a total of 10 weeks. The 10 patients were seen in three separate groups, although the authors state that identical treatments were used in each group. At the end of the 10‐week programme, group psychotherapy continued in a less structured and more supportive format, without specific session topics. Of the 10 patients with PNES that were recruited to the study, seven completed the majority of the psychoeducational sessions.

Mixed intervention study

One study investigated the effect of several behavioural interventions on non‐epileptic events (Rusch 2001).

Rusch 2001 investigated psychotherapy using a before and after design to treat 33 adults (25 female) diagnosed with non‐epileptic events by video EEG studies in the Medical College of Wisconsin. Fifteen patients had epileptic seizures as well as non‐epileptic events and 20 had received prior psychological or psychiatric intervention for depression, anxiety or other disorders. Seven of the patients terminated early. Treatment was modified as necessary to accommodate new information about underlying mechanisms, the functional consequences of non‐epileptic events and initial treatment response. Treatment length also varied according to patient need (mean number of sessions 9.5; SD 7.8; range 2 to 30).

Excluded studies

Thirty‐four studies were excluded, the majority of which examined behavioural interventions in somatisation patients as opposed to people with NEAD or conversion disorder. Other studies were excluded on the grounds of study design (see Characteristics of excluded studies).

Risk of bias in included studies

See Figure 2 and Figure 3 for illustrations of the risk of bias rating for each study and across studies for each domain. We gave each study an overall rating of high, low or unclear risk of bias based on the ratings of each domain. We rated only two studies as low risk of bias (Ataoglu 2003; Goldstein 2010), we rated one study as unclear (Moene 2003) and rated the rest of the studies as high risk of bias as they had one or more high risks in the key domains (Barry 2008; Goldstein 2004; Kuyk 2008; LaFrance 2009; Moene 2002; Prigatano 2002; Ramani 1982; Rusch 2001; Zaroff 2004).

2.

'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.

3.

'Risk of bias' graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

We rated all four RCTs (Ataoglu 2003; Goldstein 2010; Moene 2002; Moene 2003) as low risk of bias for sequence generation as all studies used adequate methods. For the non‐randomised evidence this domain was automatically rated as high risk of bias (as a five on the 'Risk of bias' scale). Only one study described allocation concealment methods (Goldstein 2010); all other RCTs and non‐randomised studies either did not describe any methods used or did not use methods to conceal allocation due to study design.

Blinding

Blinding participants and therapists in trials of therapy interventions is problematic. Therefore, even when a trial is well designed it is difficult to ascertain the relative contributions to the outcome of aspects of treatment and placebo effects.

Only one study described methods of blinding (Ataoglu 2003), which were adequate. We rated the other three RCTs as unclear due to lack of details as to whether the outcome assessor was blinded or not. Most non‐randomised studies did not use blinding methods for any of the study personnel or did not describe them. We sought any details regarding sequence generation, allocation concealment and blinding that were missing from the text from the study authors.

Incomplete outcome data

Only one study had no missing data (Ataoglu 2003). We rated several studies as unclear for risk of bias (Barry 2008; Kuyk 2008; Moene 2003; Prigatano 2002; Zaroff 2004) as it was clear that there were missing data which were not accounted for in the analysis, however the impact that this would have on the outcome remained unclear. Three studies had missing data and employed ITT analysis to account for it (Goldstein 2004; Goldstein 2010; LaFrance 2009); we rated these as either low risk of bias or allocated a number two on the 'Risk of bias' scale if non‐randomised evidence. We rated the last three studies (Moene 2002; Ramani 1982; Rusch 2001) as two out of five for risk of bias as they all had small amounts of missing data which were not subject to ITT, however this was deemed to have no effect on the outcome effect.

Selective reporting

We rated all studies for this domain as low risk of bias with one exception (Rusch 2001) which we rated as unclear due to lack of P values for a reported significant result. All studies described the outcomes in the methods section of the paper and went on to measure, analyse and report the results in the results section of the paper. We requested protocols for all studies from the study authors in order to compare the published report to an a priori list of outcomes, however none were supplied. We rated all non‐randomised studies as two out of five for risk of bias for this reason. We were not concerned about reporting biases.

Other potential sources of bias

In the non‐randomised evidence, we assessed the management of the influence of confounding variables on a scale of one to five. None of these studies incorporated any of the review's prespecified list of variables either in the planning stage of the study or at the analysis stage. Therefore, we allocated all non‐randomised studies a score of four or five and rated them as high risk of bias for this domain.

Other sources of bias varied across studies. We rated five studies as unclear (Ataoglu 2003; Goldstein 2010; LaFrance 2009; Moene 2003; Zaroff 2004) as there was not sufficient detail to make a decision on risk of bias. We rated five studies as high risk of bias (Barry 2008; Moene 2002; Prigatano 2002; Ramani 1982; Rusch 2001) due to methodological issues (see study 'Risk of bias' tables for reasons for suspecting high risk of bias: Characteristics of included studies). We rated two studies (one RCT, one non‐randomised study) as low risk of bias as no biases were identified (Goldstein 2004; Kuyk 2008).

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4

Cognitive behavioural therapy (CBT)

In a randomised controlled trial (Goldstein 2010), monthly seizure frequency reduced significantly (P = 0.002) at the end of treatment in the CBT group compared to the standard medical care group. Monthly seizure frequency was as follows (monthly median seizure frequency and interquartile range): start of treatment, CBT 12 (22.50), standard medical care 8 (29.25); end of treatment, CBT 2 (6), standard medical care 6.75 (38.63). There was a trend toward lower seizure frequency at follow‐up: CBT 1.5 (8), standard medical care 5 (24). Scores on the Hospital Anxiety and Depression Scale (HADS) showed no significant improvements.

In the Goldstein 2004 study, monthly seizure frequency reduced significantly (P < 0.01) from a pretreatment mean of 18.22 (standard deviation (SD) 43.70) to 2.88 (SD 4.73) at the end of treatment and this was maintained at six‐month follow‐up (2.59, SD 4.14). Scores on the HADS also showed some improvement (P < 0.05): anxiety at pretreatment 10.06 (SD 5.62) reduced to 7.81 (SD 5.52) at the end of treatment. The depression mean score was 6.75 (SD 3.55) pretreatment and reduced to 4.63 (SD 4.22) at the end of treatment. Both of these improvements were maintained at six‐month follow‐up (anxiety 8.13, SD 6.71; depression 4.63, SD 5.08).

In the LaFrance 2009 study, 16 participants out of the 21 reported a 50% reduction in seizure frequency and 11 of the 17 completers reported no seizures by the end of the 12 sessions (P = 0.001). Mean scores for depression, anxiety, somatic symptoms, quality of life and psychosocial functioning all showed improvement from baseline to final session. The mean Quality Of Life in Epilepsy (QOLIE) score at baseline of 46.7 improved to 53.3 at one month and further improved to 62.8 at final assessment (P = 0.0049).

Hypnosis

Moene 2002 provided no results specific to non‐epileptic seizures. It is stated that the frequency and duration of seizures was noted by staff and patients throughout the study. From this they calculated the mean percentage change in frequency and duration which was converted to a representative score on the Video Rating Scale for Motor Conversion Symptoms (VRMC) rating scale. However, none of these scores were provided in the paper. The authors' analyses were time and condition by time interactions. Their outcome measures were the VRMC, SCL‐90 (Symptom Checklist‐90; Dutch version) a self report measure of psycho‐neuroticism, the ICIDH (International Classification of Impairments, Disabilities and Handicaps) sub‐scale for physical activities) and the ICIDHP (the perceived problems sub‐scale). The authors reported significant symptom reduction in patients with a conversion disorder of the motor type; however, this was independent of the treatment condition. They reported no significant condition effects and no significant condition by time interactions for any of the outcome measures. However, a significant main effect for time was reported for all of the outcome measures VRMC, P = 0.001; SCL‐90, P = 0.05; ICIDH, P = 0.000; ICIDHP, P = 0.000). The authors concluded that the addition of hypnosis to the treatment programme did not affect outcome.

Moene 2003 provided no results specific to non‐epileptic seizures. As with the 2002 study, the outcome measures were VRMC, SCL‐90 and ICIDH. The authors reported significant treatment results for a hypnosis‐based treatment in patients with a conversion disorder, motor type. There was a statistically significant difference between the mean VRMC scores (treatment group 5.9, SD 1.3; control group 3.8, SD 1.4; P = 0.001). The analysis of the ICIDH results indicated that the treatment group improved more than controls on this interview measure of general motor impairment (treatment group, P < 0.01; control group, P = 0.29). The authors found no significant effect of treatment on the SCL‐90 (main effects: group, P = 0.54; time, P = 0.064; interaction, P = 0.56). At six‐month follow‐up the authors reported that improvement was maintained.

Paradoxical intention therapy

No specific non‐epileptic attack frequency or severity results were provided by Ataoglu 2003 as the primary outcome measure in this study was an anxiety score. However, the authors noted the percentage of the sample in each group who showed no conversion symptoms in the last two weeks at follow‐up: experimental group 14 (93.3%); control group 9 (60%). There was a significant improvement in recovery rate in the paradoxical intention therapy group when compared than the diazepam‐treated group (P = 0.034) although individual group significance levels were not provided. The authors analysed anxiety scores on the Hamilton Rating Scale for Anxiety (HRSA) before and after treatment. There was no significant difference between pretreatment anxiety scores. Both groups recorded significantly decreased anxiety scores by the end of treatment (diazepam‐treated group, P = 0.0012; paradoxical intention therapy group (P = 0.0007). A greater degree of significance was found in the paradoxical intention therapy group when comparing the difference between pre‐ and post‐treatment anxiety scores (P = 0.015).

Psychotherapy

Results found from Barry 2008 showed a decrease (improvement) in the Beck Depression Inventory (BDI) over the course of treatment, with a mean decrease from 16.6 ± 10.1 (SD) to 13.3 ± 7.9 (SD) post‐treatment. A mean reduction was found on 10 of the 12 SCL‐90 sub‐scales, with effect sizes ranging from 0.03 to 0.92 (e.g. Global Severity Index, effect size = 0.47; Obsessive–Compulsive, effect size = 0.92; Phobic Anxiety, effect size = 0.71; Depression, effect size = 0.67; and Somatization, effect size = 0.31). In addition, six of the seven patients experienced a decrease in non‐epileptic seizure frequency during the treatment period. Four of the seven (57%) had no non‐epileptic seizure events, except for occasional recurrence when they experienced increased emotional distress. Several months after the 32‐week trial, a follow‐up questionnaire was completed. The results showed that two of the seven remained seizure‐free. The rest reported a reduction in severity, intensity and duration. At the time of write‐up, the authors' long‐term follow‐up had shown that of the five participants who had continued in therapy with one of the authors, all had been either seizure‐free or had only a mild increase in seizure activity at times of stress.

Of the 22 participants in Kuyk 2008, 7.7% reported being seizure‐free at the start of treatment; 27.3% (P = 0.02) and 44% (P = 0.05) respectively were seizure‐free at the end of treatment and follow‐up. Fourteen per cent reported an increase in seizures at the end of treatment and 12.5% at follow‐up. With regard to antiepileptic drug use, 14 of the 22 patients (63.6%) were taking antiepileptic drugs at the start of treatment, but by the end of treatment only four patients were taking antiepileptic drugs. At follow‐up, of the 16 patients who responded, there had been no change in their medication use since the end of treatment, i.e. 14 used no antiepileptic drugs. Significant improvements were seen in the scores on the psychological measures including the BDI and SCL‐90. The scores improved at the end of treatment and either maintained or showed further improvement at follow‐up. For example, mean BDI scores were in the mild‐moderate severity category at the start of treatment and decreased at the end of treatment (P = 0.003) and follow‐up (P = 0.001) to the minimally depressed category.

Prigatano 2002 reported that of the 15 patients, nine completed at least 58% of the treatment sessions. Of those nine patients, six (66.6%) reported a decline in seizure frequency, two (22.2%) reported no change and one (11.1%) reported an increase.

In the Ramani 1982 study, at the time of admission, all patients were averaging two to three "hysterical" seizures per day. At the four‐year follow‐up seven of the eight had no further hysterical seizures, although three of these reported occasional "nervous spells", "shaking" or dizziness when tired or under emotional stress. One patient remained refractory to treatment and one was lost to follow‐up. Detailed tables or figures were not included in the paper.

Out of the seven patients in the Zaroff 2004 study who completed the intended therapy sessions, three were already seizure‐free at the start of treatment, one reported no change, two reported a decline and one an increase in seizure frequency. The authors reported a trend towards an improved overall score for quality of life (P = 0.07).

Mixed intervention

In the Rusch 2001 study, of the 26 patients who completed treatment, 21 were event‐free at the end of treatment and five were reported to demonstrate a significant reduction in frequency, but no P value for this was provided. Psychotherapy was reported to be ineffective for one of the patients. At six‐month follow‐up, three of the patients who had been episode‐free at the end of treatment reported recurrence of non‐epileptic events. They were given two to three more sessions and were reported to respond effectively. As a result of their study, Rusch 2001 grouped the 26 adults who completed treatment into six categories based on psychosocial history, non‐epileptic event aetiology and mechanisms of and response to specific psychotherapeutic interventions. For example, six of the 26 responded to accurate identification of anxiety and panic symptoms and cognitive therapy with exposure; and two of the 26 responded to insight‐oriented psychotherapy to recognise both the physical and emotional consequences of adverse conditions and experience etc.

Discussion

Summary of main results

Patients with non‐epileptic seizures represent a heterogeneous group with diverse psychological problems against a background of longstanding physical, psychological or sexual abuse, inadequate social skills and chronic adjustment problems (Devinsky 1998; Francis 1999a; Reilly 1999; Rusch 2001). Consequently, non‐epileptic seizures represent a serious problem for the patient, the family and the treating clinician. The costs to society can be significant with reported costs in the US of USD 100,000 per year per patient (Martin 2003). The challenge for the treating clinician relates to providing an accurate diagnosis and an effective treatment. The evidence for how best to manage and treat patients with this condition, however, is scarce.

This review considers the evidence provided by four randomised controlled trials and several before and after studies. Non‐randomised studies provide important indicators for future treatment options and research, however, they are not considered to be gold standard. Unfortunately in this population much of the evidence is provided by studies using a non‐randomised design. This review found that non‐randomised studies were the research methodology of choice, however, the interventions under investigation were heterogeneous as was the patient population. Deciding on the relative efficacy of different treatment options in some studies was complicated by the use of multiple interventions being used within the same research paradigm. These studies were not designed to compare the relative contributions towards the treatment outcomes.

A number of authors have recognised the benefits of cognitive behavioural therapy (CBT), including Betts 1993, Lesser 2003 and Ramani 1993. There was a clear recognition by these authors that patients with non‐epileptic seizures would benefit from cognitive behavioural treatment and this should be in the context of a multidisciplinary team (Betts 1993; Lesser 2003; Ramani 1993). We found one randomised clinical trial of CBT in the non‐epileptic seizure patient group (Goldstein 2010). The study was well designed and reported a significant reduction in seizure frequency at the end of treatment. However, at follow‐up reduction of seizures continued but this did not reach statistical significance. Although the authors concluded that CBT is effective in reducing psychogenic non‐epileptic seizures (PNES), the non‐significant result at follow‐up calls the long‐term effectiveness of this approach into question. A similar randomised controlled trial (RCT) of guided self help in patients with medically unexplained symptoms also found reduced effectiveness at follow‐up (Sharpe 2011). Despite this, the Goldstein 2010 study represents the strongest evidence we found of an effective intervention for people with non‐epileptic attacks.

With regard to review outcomes, our primary measure was seizure reduction and the secondary outcome measures were quality of life and seizure severity. Within the included studies, seizure frequency was the primary outcome in most. However, only four studies reported quality of life outcomes and non‐examined seizure severity.

This review, like previous reviews (Rusch 2001), clearly highlights that there is little evidence from well‐designed trials to inform treating physicians as to what therapeutic treatments exist and how effective they may be for this condition. According to Reuber and colleagues (Reuber 2005) lessons can, however, be drawn from the treatment of similar types of psychopathology (somatoform disorder, post‐traumatic stress disorder and hypochondria), where cognitive behavioural treatment is considered the psychological treatment of choice. Finally, it is difficult to disagree with the view of Devinsky who comments that while "our ability to diagnose NES has advanced significantly in the past two decades, our understanding of its pathophysiology and our ability to provide effective treatment has progressed in small tentative steps in the past century" (Devinsky 1998). It would seem that the status quo remains.

Quality of the evidence

The evidence provided by the included RCTs show benefits for the intervention under study in seizure reduction. One RCT investigated CBT in people with non‐epileptic attack disorder and we rated the evidence as high quality (see Table 1). We rated three other RCTs, looking at a variety of behavioural interventions in people with conversion disorder in synthesis, as moderate quality (see Table 3). We rated the non‐randomised evidence for behavioural interventions in people with non‐epileptic attack disorder (NEAD) as very low quality due to methodological weaknesses (see Table 2) and we rated the non‐randomised evidence for conversion disorder patients as low quality (See Table 4). Overall, this evidence was problematic due to the variety of interventions under study. There is a lack of both RCTs and non‐randomised studies which investigate the effectiveness of individual behavioural interventions delivered to people with NEAD. The evidence presented within this review therefore requires cautious interpretation.

Authors' conclusions

Implications for practice.

This review, with the inclusion of non‐randomised evidence, confirms that there continues to be very little high‐quality evidence on which to base treatment decisions for people with non‐epileptic attacks.

Implications for research.

Consensus is required regarding appropriate outcome measures, including measures of non‐epileptic seizure frequency and quality of life.

Pragmatic randomised controlled trials are required to assess the effectiveness of potential interventions.

What's new

Date	Event	Description
4 February 2013	New citation required but conclusions have not changed	New citation
4 February 2013	New search has been performed	Searches were updated on 4 February 2013

Appendices

Appendix 1. Cochrane Epilepsy Group's Specialised Register search strategy

(Title = depersonalisation, depersonalization, malingering, pseudo‐seizure, pseudoseizure, "pseudo seizure", somatization, somatisation) OR (Title = conversion, dissociative, factitious, psychophysiologic*, somatoform AND Title = disorder*) OR (Title = non‐epileptic, nonepileptic, "non epileptic", psychogenic, functional, hysteri*, pseudo, unintended AND Title = attack*, seizure*) OR (Keywords = depersonalisation, depersonalization, malingering, pseudo‐seizure, pseudoseizure, "pseudo seizure", somatization, somatisation) OR (Keywords = conversion, dissociative, factitious, psychophysiologic*, somatoform AND Keywords = disorder*) OR (Keywords = non‐epileptic, nonepileptic, "non epileptic", psychogenic, functional, hysteri*, pseudo, unintended AND Keywords = attack*, seizure*)

Appendix 2. CENTRAL search strategy

#1    MeSH descriptor Psychotherapy explode all trees

#2    MeSH descriptor Behavior Therapy explode all trees

#3    MeSH descriptor Desensitization, Psychologic explode all trees

#4    MeSH descriptor Suggestion explode all trees

#5    MeSH descriptor Psychoanalytic Therapy explode all trees

#6    MeSH descriptor Abreaction explode all trees

#7    MeSH descriptor Socioenvironmental Therapy explode all trees

#8    MeSH descriptor Conditioning (Psychology) explode all trees

#9    MeSH descriptor Adaptation, Psychological, this term only

#10   MeSH descriptor Counseling, this term only

#11   MeSH descriptor Directive Counseling, this term only

#12   MeSH descriptor Early Intervention (Education), this term only

#13   MeSH descriptor Exercise Therapy explode all trees

#14   MeSH descriptor Milieu Therapy explode all trees

#15   MeSH descriptor Psychotherapy, Group explode all trees

#16   MeSH descriptor Intervention Studies, this term only

#17   MeSH descriptor Meditation, this term only

#18   MeSH descriptor Patient‐Centered Care explode all trees

#19   MeSH descriptor Systems Theory, this term only

#20   abreaction OR "psychological adaptation" OR aromatherap* OR (autogenic NEXT train*) OR autosuggest* OR (behav* NEXT modification) OR bibliotherap* OR biofeedback OR catharsis OR ("compassionate mind" NEXT train*) OR conditioning OR counselling OR counseling OR "crisis intervention" OR desensiti?ation OR "early intervention" OR "emotional freedom tapping" OR ("eye movement" NEAR/2 (desensiti?ation OR reprocessing)) OR flooding OR "psychological feedback" OR "free association" OR hypnosis OR imagery OR "intervention studies" OR meditation OR mindfulness OR psychodrama OR psychodynamic OR psychotherap* OR "residential treatment" OR "rewind techniques" OR "stress management" OR "systems theory" OR "therapeutic community" OR "transactional analysis" in Trials

#21   ("acceptance commitment" OR art OR assertive OR autosuggestion OR aversive OR behav* OR ((client or patient) NEXT cent*) OR cognitive OR colour or color OR compassion* OR couples OR dance OR directive OR exercise OR family OR gestalt OR "human givens" OR humanistic OR implosive OR interpersonal OR marital OR mentali?ation OR milieu OR music) NEAR/2 therap* in Trials

#22   (non*directive OR paradoxical OR play OR psychoanalytic OR "rational emotive" OR reality OR socio*environmental OR suggestion OR systemic OR systems) NEAR/2 therap* in Trials

#23   (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22)

#24   MeSH descriptor Epilepsy explode all trees

#25   MeSH descriptor Seizures explode all trees

#26   (#24 OR #25 OR epilep* OR seizure* OR convuls*)

#27   MeSH descriptor Eclampsia explode all trees

#28   (#26 AND NOT #27)

#29   MeSH descriptor Dissociative Disorders explode all trees

#30   MeSH descriptor Factitious Disorders explode all trees

#31   MeSH descriptor Psychophysiologic Disorders explode all trees

#32   MeSH descriptor Somatoform Disorders explode all trees

#33   MeSH descriptor Depersonalization explode all trees

#34   MeSH descriptor Malingering explode all trees

#35   ((conversion or dissociative or factitious or psychophysiologic* or somatoform) NEXT disorder) or depersonali?ation or malingering in Trials

#36   ((non*epileptic or psychogenic) NEAR/2 (attack or seizure)) OR

((functional or hysteri* or pseudo or unintended) NEAR/2 seizure) OR

pseudo*seizure OR somati?ation in Trials

#37   ((#28 AND (#29 OR #30 OR #31 OR #32 OR #33 OR #34 OR #35) ) OR #36)

#38   (#23 AND #37)

Appendix 3. MEDLINE search strategy

1. exp Abreaction/ or abreaction.mp. or exp Adaptation, Psychological/ or Psychological Adaptation.mp. or exp aromatherapy/ or aromatherap$.mp. or exp art therapy/ or exp autogenic training/ or autogenic train$.mp. or exp autosuggestion/ or exp Aversive Therapy/ or exp behavior therapy/ or exp bibliotherapy/ or bibliotherap$.mp. or exp biofeedback, psychology/ or Biofeedback.mp. or exp catharsis/ or catharsis.mp. or exp "conditioning (psychology)"/ or conditioning.mp. or exp conditioning, classical/ or classical conditioning.mp. or exp conditioning, operant/ or operant conditioning.mp.

2. exp Cognitive Therapy/ or exp color therapy/ or exp Counseling/ or counsel?ing.mp. or exp Couples Therapy/ or exp crisis intervention/ or crisis intervention.mp. or exp dance therapy/ or exp Desensitization, Psychologic/ or Desensiti?ation.mp. or exp "Early Intervention (Education)"/ or Early Intervention.mp. or exp Exercise Therapy/ or exp Eye Movement Desensitization Reprocessing/ or (Eye Movement adj2 (Desensiti?ation or Reprocessing)).mp. or exp Family Therapy/ or exp feedback, psychological/ or exp free association/ or free association.mp. or exp gestalt therapy/ or exp hypnosis/ or hypnosis.mp. or exp "imagery (psychotherapy)"/ or imagery.mp. or exp implosive therapy/ or exp Intervention Studies/ or exp marital therapy/ or exp meditation/ or meditation.mp. or exp milieu therapy/ or exp music therapy/ or exp nondirective therapy/

3. exp play therapy/ or exp psychoanalytic therapy/ or exp psychodrama/ or psychodrama.mp. or exp psychotherapeutic processes/ or psychotherap$ process$.mp. or exp psychotherapy/ or psychotherap$.mp. or exp psychotherapy, brief/ or exp Psychotherapy, Group/ or exp psychotherapy, multiple/ or exp psychotherapy, rational‐emotive/ or exp reality therapy/ or exp residential treatment/ or residential treatment?.mp. or exp socioenvironmental therapy/ or exp suggestion/ or exp systems theory/ or exp therapeutic community/ or exp transactional analysis/ or transactional analysis.mp.

4. ((Acceptance commitment or Art or Assertive or autosuggestion or Aversive or Behav$ or Client cent$ or Cognitive or Colo?r or Compassion$ or couples or dance or Directive or Exercise or Family or gestalt or Human Givens or Humanistic or implosive or Interpersonal or marital or mentali?ation or milieu or music or nondirective or patient cent$ or play or psychoanalytic or rational?emotive or reality or socio?environmental or suggestion or systemic or systems or therapeutic community) adj2 therap$).mp.

5. (Behav$ modification or Compassionate Mind Train$ or Emotional freedom tapping or Flooding or Mindfulness or Psychodynamic or Rewind technique? or Stress manag$).mp.

6. 1 or 2 or 3 or 4 or 5

7. ((nonepileptic or non‐epileptic or psychogenic) adj2 (attack$ or seizure$)).mp.

8. ((functional or hysteri$ or pseudo or unintended) adj2 seizure$).mp.

9. (pseudoseizure$ or pseudo‐seizure$).mp.

10. somati?ation.mp.

11. exp Conversion Disorder/ or exp Dissociative Disorders/ or exp Factitious Disorders/ or exp Psychophysiologic Disorders/ or exp Somatoform Disorders/ or exp Depersonalization/ or exp Malingering/

12. (((conversion or dissociative or factitious or psychophysiologic$ or somatoform) adj disorder$) or depersonali?ation or malingering).mp.

13. 10 or 11 or 12

14. exp Epilepsy/ or exp Seizures/

15. (epilep$ or seizure$ or convuls$).mp.

16. 14 or 15

17. 13 and 16

18. 7 or 8 or 9 or 17

19. 6 and 18

20. limit 19 to (humans and (clinical trial, all or clinical trial, phase i or clinical trial, phase ii or clinical trial, phase iii or clinical trial, phase iv or clinical trial or comparative study or controlled clinical trial or meta analysis or multicenter study or randomized controlled trial or "review"))

Appendix 4. PsycINFO search strategy

abreaction or "psychological adaptation" or aromatherap* or (autogenic W2 train*) or autosuggest* or (behav* W2 modification) or bibliotherap* or biofeedback or catharsis or ("compassionate mind" W2 train*) or conditioning or counselling or "crisis intervention" or desensitization or "early intervention" or "emotional freedom tapping" or ("eye movement" W2 desensitization) or ("eye movement" W2 reprocessing) or flooding or "psychological feedback" or "free association" or hypnosis or imagery or "intervention studies" or meditation or mindfulness or psychodrama or psychodynamic or psychotherap* or "residential treatment" or "rewind techniques" or "stress management" or "systems theory" or "therapeutic community" or "transactional analysis" or (("acceptance commitment" or art or assertive or autosuggestion or aversive or behav* or (client W1 cent*) or cognitive or colour or compassion* or couples or dance or directive or exercise or family or gestalt or "human givens" or humanistic or implosive or interpersonal or marital or mentalization or milieu or music or nondirective or (patient W1 cent*) or play or psychoanalytic or "rational emotive" or reality or socio#environmental or suggestion or systemic or systems or "therapeutic community") W2 therap*) OR DE "Acceptance and Commitment Therapy" OR DE "Aromatherapy" OR DE "Art Therapy" OR DE "Assertiveness Training" OR DE "Autogenic Training" OR DE "Aversion Therapy" OR DE "Behavior Modification" OR DE "Behavior Therapy" OR DE "Bibliotherapy" OR DE "Biofeedback" OR DE "Brief Psychotherapy" OR DE "Catharsis" OR DE "Classical Conditioning" OR DE "Client Centered Therapy" OR DE "Cognitive Behavior Therapy" OR DE "Cognitive Therapy" OR DE "Conditioning" OR DE "Conjoint Therapy" OR DE "Counseling" OR DE "Countertransference" OR DE "Couples Therapy" OR DE "Creative Arts Therapy" OR DE "Crisis Intervention" OR DE "Dance Therapy" OR DE "Early Intervention" OR DE "Exercise" OR DE "Eye Movement Desensitization Therapy" OR DE "Family Therapy" OR DE "Free Association" OR DE "Gestalt Therapy" OR DE "Group Psychotherapy" OR DE "Guided Imagery" OR DE "Humanistic Psychotherapy" OR DE "Hypnotherapy" OR DE "Implosive Therapy" OR DE "Insight Psychotherapeutic Process" OR DE "Interpersonal Psychotherapy" OR DE "Marriage Counseling" OR DE "Meditation" OR DE "Milieu Therapy" OR DE "Mindfulness" OR DE "Music Therapy" OR DE "Negative Therapeutic Reaction" OR DE "Operant Conditioning" OR DE "Play Therapy" OR DE "Poetry Therapy" OR DE "Psychoanalysis" OR DE "Psychodrama" OR DE "Psychodynamic Psychotherapy" OR DE "Psychotherapeutic Breakthrough" OR DE "Psychotherapeutic Counseling" OR DE "Psychotherapeutic Processes" OR DE "Psychotherapeutic Resistance" OR DE "Psychotherapeutic Techniques" OR DE "Psychotherapeutic Transference" OR DE "Psychotherapy" OR DE "Rational Emotive Behavior Therapy" OR DE "Reality Therapy" OR DE "Recreation Therapy" OR DE "Sociotherapy" OR DE "Stress Management" OR DE "Systematic Desensitization Therapy" OR DE "Systems Neuroscience"OR DE "Systems Theory" OR DE "Therapeutic Alliance" OR DE "Therapeutic Community" OR DE "Transactional Analysis"

AND

(non#epileptic W2 (attack or seizure)) or (psychogenic W2 (attack or seizure)) or (functional W2 seizure) or (hysteri* W2 seizure) or pseudo#seizure or (unintended W2 seizure) or ((somati?ation or "conversion disorder" or "dissociative disorder" or "factitious disorder" or (psychophysiologic* W2 disorder) or "somatoform disorder" or depersonali?ation or malingering) and

(MM "Epilepsy" OR MM "Epileptic Seizures" OR DE "Grand Mal Seizures" OR DE "Petit Mal Seizures" OR DE "Status Epilepticus" OR epilep* OR seizure* OR convuls*))

Appendix 5. SCOPUS search strategy

(TITLE‐ABS‐KEY(abreaction OR "psychological adaptation" OR aromatherap* OR "autogenic train*" OR autosuggest*OR "behav* modification"OR bibliotherap* OR biofeedback OR catharsis OR "compassionate mind train*" OR conditioningOR counselling OR "crisis intervention" OR desensitization OR "early intervention" OR "emotional freedom tapping"OR ("eye movement"PRE/2 desensitization) OR ("eye movement" PRE/2 reprocessing) OR flooding OR "psychological feedback" OR "free association" OR hypnosis OR imageryOR "intervention studies"OR meditation OR mindfulness OR psychodrama OR psychodynamicOR psychotherap* OR "residential treatment" OR "rewind techniques" OR "stress management" OR "systems theory" OR "therapeutic community" OR "transactional analysis" OR ("acceptance commitment"PRE/2 therap*) OR (art PRE/2 therap*) OR (assertivePRE/2 therap*) OR (autosuggestion PRE/2 therap*) OR (aversive PRE/2 therap*) OR (behav* PRE/2 therap*) OR ("client cent*" PRE/2 therap*) OR (cognitivePRE/2 therap*) OR (colour PRE/2 therap*) OR (compassion*PRE/2 therap*) OR (couples PRE/2 therap*) OR (dancePRE/2 therap*) OR (directive PRE/2 therap*) OR (exercise PRE/2 therap*) OR (family PRE/2 therap*) OR (gestalt PRE/2 therap*) OR ("human givens"PRE/2 therap*) OR (humanistic PRE/2 therap*) OR (implosive PRE/2 therap*) OR (interpersonal PRE/2 therap*) OR (marital PRE/2 therap*) OR (mentalization PRE/2 therap*) OR (milieu PRE/2 therap*) OR (music PRE/2 therap*) OR (nondirective PRE/2 therap*) OR ("patient cent*"PRE/2 therap*) OR (play PRE/2 therap*) OR (psychoanalyticPRE/2 therap*) OR ("rational emotive" PRE/2 therap*) OR (reality PRE/2 therap*) OR (socio*environmental PRE/2 therap*) OR (suggestion PRE/2 therap*) OR (systemic PRE/2 therap*) OR (systems PRE/2 therap*) OR "therapeutic community"))

AND (TITLE‐ABS‐KEY((non*epilepticPRE/2 (attack OR seizure)) OR (psychogenic PRE/2 (attackOR seizure)) OR (functional PRE/2 seizure) OR (hysteri* PRE/2 seizure) OR pseudo*seizure OR (unintended PRE/2 seizure)

OR ((somati*ation OR "conversion disorder" OR "dissociative disorder" OR "factitious disorder" OR "psychophysiologic* disorder"OR "somatoform disorder"OR depersonali*ation OR malingering)  AND (*epilepOR epilep* OR seizure* OR *seizure OR convuls*))))

AND (LIMIT‐TO(EXACTKEYWORD, "Human") OR LIMIT‐TO(EXACTKEYWORD, "Humans")) AND (LIMIT‐TO(SUBJAREA, "MEDI") OR LIMIT‐TO(SUBJAREA, "NEUR") OR LIMIT‐TO(SUBJAREA, "PSYC") OR LIMIT‐TO(SUBJAREA, "HEAL") OR LIMIT‐TO(SUBJAREA, "SOCI") OR LIMIT‐TO(SUBJAREA, "NURS") OR LIMIT‐TO(SUBJAREA, "MULT"))

Appendix 6. Table of scale parameters

	1	2	3	4	5
Confounding	All important1 confounders considered2 and suitable method of adjustment3 employed Outcome unlikely to be affected	Most important4 confounders considered and suitable method of adjustment employed Outcome unlikely to be affected	Some confounders5 considered and full or partial adjustment employed6 Possible influence on outcome	Some confounders considered and no adjustment employed Likely to affect outcome	No important confounders considered and no adjustment employed Likely to affect outcome
Blinding	Assessors blinded to participant's drug regime and participants blinded to drug regime Outcome unlikely to be affected	Assessors blinded to participants drug regime Outcome unlikely to be affected	Partial blinding7 involved in study Possible implication on outcome	Partial or no blinding involved in study Outcome likely to be affected	No blinding involved in study Outcome likely to be affected
Incomplete outcome data	No missing data and/or appropriate analysis8 used to deal with missing data Unlikely to affect outcome	Smaller amount (< 25%) of missing data with reasons given, balanced across groups Unlikely to affect outcome	Larger amount of missing data (> 25%) with or without reasons given, balanced across groups Possible implication on outcome	Larger amount (> 25%) of missing data, imbalance across groups Outcome likely to be affected	No information provided regarding missing data Likely to affect outcome
Selective outcome reporting	A priori outcomes measured, analysed and reported in main report Protocol available Unlikely to affect outcome	A priori outcomes measured, analysed and reported in main report9 Protocol not available Unlikely to affect outcomes	Limited information regarding a priori outcomes and measures Possible influence on outcome	Outcomes measured but not analysed or reported Outcome likely to be affected	Outcomes measured but not analysed or reported and clinical judgement infers the presence of an unreported measured outcome10
Other bias	No bias identified	Bias identified Unlikely to affect outcome	Bias identified Possible implication on outcome	Bias identified Likely to affect outcome	Bias identified Extremely likely to affect outcome

Footnotes

¹Important confounders include: duration of treatment, types of treatment, different aetiologies, mixed diagnosis patients, inpatient and outpatient data grouped together.

²Reported demographic information and other confounders.

³Matching scores, multiple regression, analysis of co‐variance, stratification.

⁴At least two out of five of important confounders including: types of treatment, mixed diagnosis patients.

⁵At least one out of five of confounders.

⁶Full adjustment of confounding variables, e.g. see footnote 2 or partial adjustment, e.g. researchers select limited number of variables to adjust for.

⁷Assessors of outcome are only blinded to certain groups, e.g. blinded to intervention group but not controls.

⁸Intention‐to‐treat analysis.

⁹An a priori statement is made in the methods section of the main report regarding measurement and analysis of outcome.

¹⁰For example, failure to report overall quality of life when all other domains are reported. 

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Ataoglu 2003.

Methods	Randomised controlled trial, single centre (Psychiatry Department in Turkey) Treatment duration: 3 weeks of paradoxical therapy or 6 weeks of drug treatment
Participants	Diagnosis: pseudoseizures/conversion disorder using DSM‐IV criteria No information provided regarding previous seizures or comorbid conditions 30 patients were randomised 29 female, 1 male Aged 16 to 35 years
Interventions	Intervention group: paradoxical intention (N = 15): inpatient treatment in psychiatric ward, 2 sessions a day for 3 weeks Control group: diazepam (N = 15): outpatient treatment (5 to 15 mg/day). Appointments at 10, 20, 30 and 45 days
Outcomes	1) Frequency of conversion attacks in past week noted for each patient with changes in these scores converted to percentages 2) Anxiety Measurements taken at baseline and end of treatment
Notes	Patients primarily female and low in education
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated method of randomisation used; adequate method
Allocation concealment (selection bias)	Unclear risk	No details on allocation concealment
Blinding (performance bias and detection bias) All outcomes	Low risk	Outcome assessor was blinded to treatment conditions
Confounding variables (for non‐randomised studies) All outcomes	Low risk	NA
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data
Selective reporting (reporting bias)	Low risk	Outcomes outlined in the methods section were measured, analysed and reported in the results
Other bias	Unclear risk	The patients were mostly female and of low education

Barry 2008.

Methods	Before and after non‐controlled study, single centre (US) Treatment duration: 32 weeks
Participants	12 patients with non‐epileptic seizures enrolled in study, 7 received intended treatment (completed 75%) and 7 analysed Mean age of 7 analysed: 45.4 years (30 to 60 years) All 7 female
Interventions	Psychodynamic group therapy 5 patients also received individual cognitive behavioural therapy during the study and 1 saw a former therapist sporadically
Outcomes	1) Depression 2) Symptoms 3) Seizure frequency 4) Quality of life Measurements taken at baseline, 16 weeks into therapy and at 32 weeks on completion of therapy
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomised study therefore automatically at high risk of bias
Allocation concealment (selection bias)	High risk	No methods for concealed allocation
Blinding (performance bias and detection bias) All outcomes	High risk	Rated as 5 on 'Risk of bias' scale as no blinding techniques used in study
Confounding variables (for non‐randomised studies) All outcomes	High risk	Rated as 5 on 'Risk of bias' scale as several confounding variables from our prespecified list were apparent in the study and no methods were employed to deal with the variables (mixed aetiologies, extra individual therapy sessions, duration of treatment differed)
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Rated as 3 on 'Risk of bias' scale as it was clear there were missing data; only 7 participants completed 75% of the intended treatment (> 25% missing) therefore there is a possible influence on the outcome
Selective reporting (reporting bias)	Low risk	Rated as 2 on 'Risk of bias' scale as outcomes were outlined in the methods and measured, analysed and reported in the results, however it was unclear which outcomes were primary and which were secondary. Also, there was no a priori protocol to compare these outcomes to the published report
Other bias	High risk	Rated as 5 on 'Risk of bias' scale as 1 patient was tapering off diazepam during the study and 1 patient was being treated for hepatitis C

Goldstein 2004.

Methods	Before and after non‐controlled study, single centre (UK) Treatment duration: 12‐24 weeks. Follow‐up period: 6 months
Participants	20 patients with dissociative seizures were enrolled in the study, 16 patients received the intended treatment and 16 were analysed Mean age 34.9 years 14 females and 2 males
Interventions	Individual cognitive behavioural sessions. 12 sessions, first session 2 hours, all other sessions 1 hour in duration, either weekly or fortnightly
Outcomes	1) Seizure frequency 2) Work and social adjustment 3) Fear 4) Anxiety and depression 5) Employment 6) Health locus of control 7) Illness perception Measurements taken at baseline, post‐treatment and at follow‐up
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomised study therefore automatically at high risk of bias.
Allocation concealment (selection bias)	High risk	No methods of concealing allocation used.
Blinding (performance bias and detection bias) All outcomes	High risk	Rated as 5 on RoB scale as no methods of blinding were used.
Confounding variables (for non‐randomised studies) All outcomes	High risk	Rated as 4 on RoB scale as patients had mixed aetiologies (abusive history) and no methods used to control for any of the pre‐specified confounding variables.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Rated as 2 on RoB scale as there was a small amount of missing data which was incorporated into ITT analysis for two outcomes.
Selective reporting (reporting bias)	Low risk	Rated as 2 on RoB scale as outcomes outlined in methods were measured, analysed and reported however there was no a priori protocol available to compare these outcomes.
Other bias	Low risk	Rated as 1 on RoB scale as no other biases were identified.

Goldstein 2010.

Methods	Randomised controlled trial, single centre (UK) Treatment duration: 4 months. Follow‐up period: 6 months
Participants	66 patients with psychogenic non‐epileptic seizures were randomised 50 female, 16 male Mean age in intervention group 37.4 years, mean age in control group 35.9 years
Interventions	Intervention group: cognitive behavioural therapy and standard medical care for up to 12 weeks, 1‐hour sessions Control group: standard medical care only
Outcomes	1) Monthly seizure frequency 2) Seizure freedom 3) Work and social adjustment 4) Anxiety and depression 5) Health service use and employment Measurements taken at pretreatment, end of treatment and at 1, 3 and 6 months follow‐up
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Used random number table to generate randomisation sequence; adequate method used
Allocation concealment (selection bias)	Low risk	Sequentially numbered, sealed envelopes administered to patients; adequate method used
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Unclear if outcome assessor blinded
Confounding variables (for non‐randomised studies) All outcomes	Low risk	NA
Incomplete outcome data (attrition bias) All outcomes	Low risk	Small amount of missing data, however ITT analysis employed; adequate method
Selective reporting (reporting bias)	Low risk	Primary and secondary outcomes clearly outlined in the methods section; all measured, analysed and reported in the results section. No protocol with which to compare a priori outcomes
Other bias	Unclear risk	Standard medical care sessions varied according to the needs of the patients, number of CBT sessions varied

Kuyk 2008.

Methods	Before and after non‐controlled study, single centre (The Netherlands) Treatment duration: unclear. Follow‐up period: 6 months
Participants	29 patients with non‐epileptic seizures were enrolled in the study, 26 completed the intended treatment and 22 were analysed Mean age 30.6 years 77.3% female, 22.7% male
Interventions	Psychological treatment including individual psychotherapy, psychomotor and creative therapy, family therapy and group therapies
Outcomes	1) Seizure frequency 2) Seizure duration 3) Number of AEDs taken 4) Symptoms 5) Depression 6) Anxiety 7) Coping behaviour 8) Quality of life Measurements taken at the start of treatment, on discharge and at 6 months follow‐up
Notes	4 patients excluded due to occurrences of epileptic seizures; 3 excluded due to stopping prematurely
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomised study therefore automatically rated as high risk of bias.
Allocation concealment (selection bias)	High risk	No methods of concealing allocation used.
Blinding (performance bias and detection bias) All outcomes	High risk	Rated as 5 on RoB scale as no one was blinded in this study.
Confounding variables (for non‐randomised studies) All outcomes	High risk	Rated as 4 on RoB scale as mixed seizure patients were included in the study however excluded from the analysis altogether. The duration of treatment was unclear, reported average of 4.8 months. No methods used to account for any pre‐specified confounding variables.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Rated as 3 on RoB scale as there was missing data and no methods used in the analysis to account for it.
Selective reporting (reporting bias)	Low risk	Rated as 2 on RoB scale as secondary outcomes were unclear and there was no protocol to compare a priori list of outcomes.
Other bias	Low risk	Rated as 2 on RoB scale as group was self‐selected, unsure how much this could affect outcomes.

LaFrance 2009.

Methods	Before and after, non‐controlled study, single centre (US) Baseline period: 2 weeks. Treatment duration: 12 weeks. Follow‐up period: 12 months
Participants	21 patients with psychogenic non‐epileptic seizures (some with epileptic seizures also) enrolled in the study, 17 completed the intended treatment and 20 were analysed Mean age 36.0 years 17 female, 4 male
Interventions	12 sessions of cognitive behavioural therapy (1 hour per session)
Outcomes	1) Seizure frequency 2) Depression 3) Anxiety 4) Symptoms 5) Impulsivity 6) Family functioning 7) Somatic symptoms 8) Global functioning 9) Disability 10) Psychosocial functioning 11) Coping techniques 12) Quality of life Measurements taken at baseline, month 1 of treatment, final week of treatment and at 4, 8 and 12 months follow‐up
Notes	1 patient dropped out prior to receiving any treatment and was therefore excluded from the intention‐to‐treat analysis
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomised study therefore automatically rated as high risk of bias
Allocation concealment (selection bias)	High risk	No methods used to conceal allocation
Blinding (performance bias and detection bias) All outcomes	High risk	Rated as 5 on 'Risk of bias' scale as no methods of blinding were employed
Confounding variables (for non‐randomised studies) All outcomes	High risk	Rated as 5 on 'Risk of bias' scale as patients has mixed diagnoses and aetiologies and no adjustments were made in the analyses for any of the prespecified confounders
Incomplete outcome data (attrition bias) All outcomes	Low risk	Rated as 2 on 'Risk of bias' scale as there were missing data and a modified ITT analysis was employed
Selective reporting (reporting bias)	Low risk	Rated as 2 on 'Risk of bias' scale as outcomes outlined in the methods were measured, analysed and reported in the results, however there was no a priori protocol available with which to compare these outcomes
Other bias	Unclear risk	Rated as 3 as there were insufficient details to judge other bias in this study

Moene 2002.

Methods	Randomised controlled trial, 2 outpatient psychiatric departments (The Netherlands) Treatment duration: hypnosis group 8 weeks, rehab group 12 weeks. Follow‐up period: 8 months from before treatment started
Participants	49 patients with conversion disorder (motor type) or somatisation disorder (with motor conversion symptoms) were enrolled at the start of the study, 45 received the intended treatment and 45 were analysed 34 female, 11 male Mean age 36.8 years, ranging from 18 to 56 years
Interventions	Intervention group (N = 26) included an introductory session followed by 1 hour per week for 8 weeks of hypnosis. Also encouraged to practise self hypnosis for 0.5 hours per day with audiotape Control group (N = 23) included 8 weeks of 1‐hour sessions encouraging patients to talk about their experience and homework to write about sessions Both groups consisted of inpatient treatment programme (group work, individual physiotherapy, exercise and bed rest)
Outcomes	1) Motor conversion symptoms 2) Disability 3) Symptoms
Notes	4 patients dropped out prior to completion of the treatment, 2 from the intervention group and 2 from the control group; ITT analysis not performed
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Block randomisation method used to generate randomisation sequence; adequate method
Allocation concealment (selection bias)	Unclear risk	Allocation concealment stated but methods not explained
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Unclear description of blinding methods
Confounding variables (for non‐randomised studies) All outcomes	Low risk	NA
Incomplete outcome data (attrition bias) All outcomes	Low risk	Small amount of missing data, so although ITT not employed, outcomes unlikely to be affected
Selective reporting (reporting bias)	Low risk	Clearly described outcomes in the methods section, measured, analysed and reported in the results. No protocol with which to enable comparison
Other bias	High risk	Patients were either inpatients or outpatients and all data were combined together in analysis

Moene 2003.

Methods	Randomised controlled trial, 2 outpatient psychiatric clinics (The Netherlands) Treatment duration: 3 months (follow‐up at 6 months for treatment group)
Participants	49 patients with a diagnosis of conversion disorder (motor type) or somatisation disorder (with motor conversion symptoms) were enrolled at the start of the study, 44 received the intended treatment and 44 were analysed Mean age 36.6 years, range 18 to 61 years 75% female
Interventions	Intervention group (N = 24) 1‐hour introductory session explaining rationale for using hypnosis followed by hypnosis sessions 1 hour per week for 10 weeks. Also encouraged to practise self hypnosis for ½ hour per day with audiotape Control group (N = 25) waiting list for hypnosis
Outcomes	1) Motor conversion symptoms 2) Disability 3) Symptoms
Notes	5 patients dropped out prior to completion of the treatment, 4 from the intervention group and 1 from the control group
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Block randomisation method used to generate randomisation sequence; adequate method
Allocation concealment (selection bias)	Unclear risk	No methods indicated in the text
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Unclear description of blinding methods
Confounding variables (for non‐randomised studies) All outcomes	Low risk	NA
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Evidence of missing data, however unclear how outcome likely to be affected as ITT not employed
Selective reporting (reporting bias)	Low risk	All outcomes outlined in the methods were measured, analysed and reported in the results section. No protocol available
Other bias	Unclear risk	Insufficient details to judge bias in this study

Prigatano 2002.

Methods	Before and after non‐controlled study, single centre (US) Treatment duration: 6 months
Participants	15 patients with psychogenic non‐epileptic seizures enrolled at the start of the study, 8 were treated first followed by 7 treated second. 13 patients received the intended treatment and 13 were analysed Mean age 36.0 years, ranging from 20 to 48 years All participants female
Interventions	6‐month psychotherapy programme delivered in 1.5‐hour sessions over 24 weeks
Outcomes	1) Seizure frequency 2) Knowledge of previous session 3) Cerebral functioning 4) Intelligence 5) Verbal learning and memory 6) Personality Measures were reported following the first 12 sessions and following the last 12 sessions. Data reported are descriptive only
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomised study therefore automatically rated as high risk of bias
Allocation concealment (selection bias)	High risk	No methods of allocation concealment used
Blinding (performance bias and detection bias) All outcomes	High risk	No methods of blinding used
Confounding variables (for non‐randomised studies) All outcomes	High risk	Rated as 4 on 'Risk of bias' scale as one group had no exclusion criteria applied whilst other had a list of exclusion criteria and therefore no confounding variables were accounted for in one group. No methods for accounting for prespecified confounding variables in either groups within the analysis
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Rated as 3 on 'Risk of bias' scale as some data missing from study and not incorporated in the analysis
Selective reporting (reporting bias)	Low risk	Rated as 2 on 'Risk of bias' scale as outcomes outlined in the methods were measured, analysed and reported in the results, however there was no a priori protocol available with which to compare these outcomes
Other bias	High risk	Rated as 4 on 'Risk of bias' scale as both groups were handled very differently; one therapist took a planned vacation during the study

Ramani 1982.

Methods	Before and after non‐controlled study, single centre (US) Treatment duration: 4 to 9 weeks inpatient treatment. Follow‐up period: 4 years
Participants	9 patients with hysterical seizures and epileptic seizures enrolled in the study and 8 were analysed at follow‐up Mean age 31.6 years, ranging from 17 to 54 years 1 male and 8 female
Interventions	Individualised psychotherapy 7 patients received behavioural therapy, 1 received behaviour modification therapy and 1 received short‐term dynamic therapy
Outcomes	1) Seizure frequency 2) Medications 3) Living conditions 4) Employment 5) Hospitalisation 6) Attitude towards self 7) Significant events Measurements taken at baseline and at follow‐up
Notes	1 patient did not complete the follow‐up questionnaire
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomised study therefore automatically rated as high risk of bias
Allocation concealment (selection bias)	High risk	No methods of allocation concealment used
Blinding (performance bias and detection bias) All outcomes	High risk	No methods of blinding used
Confounding variables (for non‐randomised studies) All outcomes	High risk	Rated as 5 on 'Risk of bias' scale as treatments were highly individualised, patients had mixed diagnoses and there was no indication of treatment lengths
Incomplete outcome data (attrition bias) All outcomes	Low risk	Rated as 2 on 'Risk of bias' scale as there was a small amount of missing data which was unlikely to affect the outcomes
Selective reporting (reporting bias)	Low risk	Rated as 2 on 'Risk of bias' scale as outcomes outlined in the methods were measured, analysed and reported in the results, however there was no a priori protocol available with which to compare these outcomes
Other bias	High risk	Rated as 5 on 'Risk of bias' scale as in a sample size of 9, IQ ranged from 30 to 116 and 1 patient was described as "severely retarded"

Rusch 2001.

Methods	Before and after non‐controlled study, single centre (US) Treatment duration: unclear. Follow‐up period: 6 and 12 months
Participants	33 patients with non‐epileptic events enrolled at the start of the study, 26 received the intended treatment and 26 were analysed Mean age 33.8 years, ranging from 6 to 56 years 25 female and 8 male patients
Interventions	Patients were grouped according to clinical characteristics; patients received 6 separate forms of psychotherapy: (1) Accurate identification of anxiety and panic symptoms; cognitive therapy with exposure (N = 6) (2) Intensive psychotherapy (N = 6) (3) Insight‐orientated psychotherapy (N = 2) (4) Cognitive behavioural therapy (N = 2) (5) Exposure‐based therapies (N = 7) (6) Behavioural management strategies (N = 3) Mean number of sessions (9.5) ranging from 2 to 30 sessions
Outcomes	1) Seizure freedom 2) Seizure frequency 3) Seizure characteristics 4) Psychiatric history Measurements reported at less than 12 months and more than 12 months
Notes	7 patients terminated treatment early
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomised study therefore automatically rated as high risk of bias
Allocation concealment (selection bias)	High risk	No methods of allocation concealment used
Blinding (performance bias and detection bias) All outcomes	High risk	No methods of blinding used
Confounding variables (for non‐randomised studies) All outcomes	High risk	Rated as 4 on 'Risk of bias' scale as therapy was modified as it went on; 15 patients had both non‐epileptic and epileptic seizures, mixed aetiologies
Incomplete outcome data (attrition bias) All outcomes	Low risk	Rated a 2 on 'Risk of bias' scale as there was a small amount of missing data with no ITT analysis, unlikely to affect the outcomes
Selective reporting (reporting bias)	Unclear risk	Rated as unclear on 'Risk of bias' scale as main outcome reported as significant, however no corresponding P value reported in the text
Other bias	High risk	Rated as 4 on 'Risk of bias' scale as there is little information to determine any other risk of bias; sample size was small; very few patients in each group

Zaroff 2004.

Methods	Before and after non‐controlled study, single centre (US) Treatment period: 10 weeks
Participants	10 patients with psychogenic non‐epileptic seizures were enrolled at the start of the study, 7 completed the majority of the intended treatment and 7 were analysed Mean age 35.7 years, ranging from 21 to 57 years 6 women and 4 men
Interventions	Group psychotherapy 10 sessions for 1 hour per week carried out in 3 separate groups
Outcomes	1) Seizure frequency 2) Curious experiences 3) Quality of life 4) Avoidance 5) Anger 6) Emotion 7) Task Measurements were taken at baseline and post‐treatment. Pre‐ and post‐treatment scores were also compared to normative samples for each outcome
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomised study therefore automatically rated as high risk of bias
Allocation concealment (selection bias)	High risk	No methods of allocation concealment used
Blinding (performance bias and detection bias) All outcomes	High risk	No methods of blinding used
Confounding variables (for non‐randomised studies) All outcomes	High risk	Rated as 4 on 'Risk of bias' scale as patients had mixed aetiologies; no methods used to deal with prespecified list of variables
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Missing data and no ITT analysis
Selective reporting (reporting bias)	Low risk	Rated as 2 on 'Risk of bias' scale as outcomes outlined in the methods were measured, analysed and reported in the results, however there was no a priori protocol available with which to compare these outcomes
Other bias	Unclear risk	Rated as 4 as there is little information to determine any other risk of bias and 3 patients were actually seizure‐free at the start of treatment

AED: antiepileptic drug
 CBT: cognitive behavioural therapy
 DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders IV
 ITT: intention‐to‐treat
 NA: not applicable

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Allen 2006	Somatisation patients; no differentiation between patients with NEAD or PNES
Betts 1992	Retrospective study and case study data only
Bhatia 2005	Not an intervention study; no before and after measures taken
Bleichhardt 2004	Somatisation patients; no differentiation between patients with NEAD or PNES
Buchanan 1993	Retrospective study; no intervention
Chinta 2008	Not an intervention study
Elsas 2011	Epilepsy patients; not NEAD or PNES patients
Flom 1999	Not an intervention study; retrospective design
Ghosh 2007	Retrospective study; no intervention
Gupta 2011	Somatisation patients; no differentiation between patients with NEAD or PNES
Hiller 2003	Not an intervention study; no PNES patient data
Hovorka 2007	Not an intervention study; no before and after measures taken
Kashner 1995	Somatisation patients; no differentiation between patients with NEAD or PNES
Katsamanis 2011	Patients with medically unexplained physical symptoms; no differentiation between patients with NEAD or PNES
Khattak 2006	Somatisation patients; no differentiation between patients with NEAD or PNES
Lidbeck 1997	Somatisation patients; no differentiation between patients with NEAD or PNES
Lidbeck 2003	Somatisation patients, no differentiation between patients with NEAD or PNES
Magallon 2008	Protocol for study
Martin 2007	Somatisation patients; no differentiation between patients with NEAD or PNES
Mayor 2010	Follow‐up data collected retrospectively
McDade 1992	No before and after comparisons made
McLeod 1997	Somatisation patients; no differentiation between patients with NEAD or PNES
Miller 1983	Case study design
Mims 1994	Not defined behavioural treatment; presents case study data
Nanke 2000	Somatisation patients; no differentiation between patients with NEAD or PNES
Nanke 2003	Somatisation patients; no differentiation between patients with NEAD or PNES
Posse 2004	Somatisation patients; no differentiation between patients with NEAD or PNES
Schweickhardt 2007	Somatisation patients; no differentiation between patients with NEAD or PNES
Share 1996	Not an intervention study; letter to editor
Smith 1986	Somatisation patients; no differentiation between patients with NEAD or PNES
Smith 2009	Somatisation patients; no differentiation between patients with NEAD or PNES
Speckens 1996	Somatisation patients; no differentiation between patients with NEAD or PNES
Timmer 2004	Somatisation patients; no differentiation between patients with NEAD or PNES
Timmer 2006	Somatisation patients; no differentiation between patients with NEAD or PNES

NEAD: non‐epileptic attack disorder
 PNES: psychogenic non‐epileptic seizures

Characteristics of studies awaiting assessment [ordered by study ID]

La France.

Methods	Prospective, multi‐centre, multi‐arm, combined treatment pilot RCT. Data collected at baseline, week 2, week 8 and week 16
Participants	38 PNES patients enrolled, 35 patients completed baseline measurements and were randomised
Interventions	1) Medication (flexible‐dose sertraline) 2) CBT only 3) CBT and medication combined 4) Standard medical care
Outcomes	Seizure frequency Psychosocial and function variables
Notes	Awaiting publication

RCT: randomised controlled trial
 PNES: psychogenic non‐epileptic seizures

Differences between protocol and review

We included before and after controlled and before and after non‐controlled study designs.

Contributions of authors

Jayne Martlew is responsible for the update of this review and took part in updating the protocol, assessing studies for eligibility, data extraction, assessing risk of bias and writing up the review.

Jennifer Pulman updated plans for the review methodology, assessed studies for eligibility, carried out data extraction, assessed risk of bias, carried out data analysis and took part in the write‐up of the review.

Anthony Marson has supervised this review and provided expert opinion and comments.

Sources of support

Internal sources

• No sources of support supplied

External sources

• National Institute for Health Research, UK.

  This review presents independent research commissioned by the National Institute for Health Research (NIHR). The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.

Declarations of interest

None known.

New search for studies and content updated (no change to conclusions)