Address correspondence to Dr Markus Reuber, Academic Neurology Unit, Royal Hallamshire Hospital, Glossop Road, Sheffield S10 2JF, U.K. E-mail: firstname.lastname@example.org
Purpose: Most neurologists endorse psychotherapy as the treatment of choice for psychogenic nonepileptic seizures (PNES), but its effectiveness remains unproven, and there are no previous reports of long-term outcome after psychotherapy. This study aimed to establish the outcome of brief augmented psychodynamic interpersonal therapy (PIT) for 47 patients with PNES in terms of seizures and health care utilization 31–65 months (median 50 months) after diagnosis.
Methods: Participants completed questionnaires before starting therapy (Clinical Outcomes in Routine Evaluation Outcome Measure [CORE-OM]; Patient Health Questionnaire [PHQ15]; Short-Form Health Survey [SF-36]). Forty-seven of 66 consecutive patients (71%) also completed a follow-up questionnaire about current seizure frequency, employment status, and health care utilization 42 months after the end of therapy (range 12–61 months). Factors associated with seizure outcome and predictors of seizure cessation were evaluated.
Results: At follow-up, 25.5% of patients had become seizure-free; a further 40.4% achieved a seizure reduction of >50%. Logistic regression showed “economic activity” status to be the only significant baseline predictor of seizure cessation (p < 0.021). Health care utilization declined significantly from baseline to follow-up (p < 0.039), suggesting minimum expected annual health care expenditure savings of £245 ($408).
Discussion: These results indicate that this intervention is associated with a significant improvement in seizure frequency and health care utilization, suggesting that a randomized controlled study of the intervention is justified.
Psychogenic nonepileptic seizures (PNES) resemble epileptic seizures but are not caused by epileptic discharges. PNES can be interpreted as an impairment of self-control associated with a range of motor, sensory, and mental manifestations that occur as a response to emotional distress (Reuber, 2008). With an estimated prevalence of 2–33 per 100,000, PNES are one of the most common medically unexplained neurologic symptoms (Benbadis et al., 2000). Twelve percent to 18% of patients first presenting to doctors with blackouts have PNES (Kotsopoulos et al., 2003; Angus-Leppan, 2008), and approximately one-third of patients referred to specialist epilepsy centers with refractory seizure disorders are found to have PNES rather than epilepsy (Cragar et al., 2005).
Despite the facts that PNES are common, disabling, and costly, there is little evidence regarding the best management of the disorder, and much of the current available evidence lacks scientific rigor (Brooks et al., 2007). Although psychological treatment is considered the intervention of choice by the majority of neurologists (LaFrance et al., 2008), the effectiveness of psychological interventions for PNES remains to be proven. The best evidence for psychological treatment comes from two small pilot studies describing the use of cognitive behavioral therapy (Goldstein et al., 2004; LaFrance et al., 2009). In addition, we have described the effect of brief augmented psychodynamic interpersonal therapy (PIT) in patients with a range of functional symptoms on HRQoL, physical symptoms and levels of distress (Reuber et al., 2007). Although more than two-thirds of the patients included in this study had PNES, we have not previously reported the seizure outcome of this patient group. In fact, there are no previous reports of long-term seizure outcome after any intervention for PNES.
This study intends to describe outcome in terms of seizure frequency and health care utilization in a consecutive group of patients with PNES treated with our brief augmented PIT approach (Howlett & Reuber, 2009). It explores whether baseline demographic or clinical factors predict seizure frequency outcome at follow-up 31–65 months (median 50 months) after diagnosis.
This study describes consecutive patients with PNES seen at the Royal Hallamshire Hospital and the Barnsley District General Hospital in the United Kingdom between November 2003 and July 2006, and referred for outpatient psychotherapy to a specialized service within the neurology department. All patients attended at least one psychotherapy appointment. Patients were included only if there was a minimum follow-up period of 12 months after their last psychotherapy appointment. The diagnosis of PNES was based on the clinical impression of a fully trained neurologist with a particular interest in epilepsy, direct observation of seizures in clinic, video recordings, or a history or witness accounts highly suggestive of PNES. Diagnoses were confirmed by video/EEG if necessary and possible. Patients with proven or suspected additional epilepsy were excluded from this study. Patients were also excluded if the neurologist was unable to ascertain the frequency of PNES at the time of referral for psychotherapy, or if they failed to return the follow-up postal questionnaire.
At the time of referral for psychotherapy, neurologists completed a screening questionnaire providing clinical details. The seizure frequency was determined by questioning patients and carers in all cases, and by checking seizure diaries when available. Patients received a postal questionnaire to their home 1–3 weeks after their meeting with the neurologist. They were asked to return this to the psychotherapy department to indicate their interest in attending for treatment. The questionnaire comprised demographic questions and three validated self-report measures.
The SF-36 Health Survey (Ware & Sherbourne, 1992) is a 36-item measure of eight health concepts that can be combined to produce a physical health component score (made up of four concept scores: Physical Functioning, Role Limitation Physical, Bodily Pain, and General Health) and a mental health component score (made up of four concept scores: Vitality, Social Functioning, Role Limitations Emotional, and Mental Health). The SF-36 raw data were transformed to 0–100 scores, with low scores indicating poor health and functioning.
The Clinical Outcomes in Routine Evaluation (CORE-OM) is a 34-item measure scoring self-reported global level of psychological distress that can be compared with clinical thresholds. It taps subjective well being, commonly experienced problems or symptoms, and life/social functioning, including items on risk to self and others. The mean score of the 34 items is calculated with higher scores indicating greater distress. The measure has been extensively validated in large clinical and nonclinical samples and shows strong correlation with different well-established measures of depression such as the Beck Depression Index (Meierkord et al., 1991; Leach et al., 2006).
The Patient Health Questionnaire (PHQ)-15 is composed of 15 physical symptoms that have been extracted from the PHQ. The PHQ-15 is part of the self-administered PRIME MD diagnostic instrument for common mental disorders (Spitzer et al., 1994). The 15 symptoms do not include blackouts or seizures. Symptoms over the last 4 weeks are rated as 0 (“not at all bothered”), 1 (“bothered a little”), or 2 (“bothered a lot”) (Kroenke et al., 2002). A number of studies suggest that higher scores on the PHQ-15 correlate with greater psychological distress (Interian et al., 2006; Kroenke et al., 2007; Tietjen et al., 2007).
Patients were considered “economically active” if they were employed, self-employed, in full-time education, or older than 60 years of age and receiving an old-age pension. Patients were considered “economically inactive” if they were unemployed, receiving disability benefits, or retired on health grounds.
Patients were given four response options to provide information about health care utilization (appointments with doctors, outpatient appointments, visits to hospital emergency departments) in the previous 3 months (1: none, 2: 1–5 contacts, 3: 6–10 contacts, or 4: over 10 contacts). Medical contact costs were looked up in a manual (Curtis, 2008).
Patients’ seizure status at the end psychotherapy was extracted from psychotherapy notes and discharge letters. Longer term outcome was determined by another postal questionnaire sent 12–61 months after the end of psychotherapy comprising information about current employment status, health care utilization, and seizure frequency (seizure-free for over a year, seizure-free for over 6 months, or stated number of seizures per day/week/month).
The data for this study were collected in the context of our ongoing Service Evaluation. The Service Evaluation, all questionnaires and letters sent to patients were approved by the Service Evaluation Manager in the Clinical Effectiveness Unit of the Sheffield Teaching Hospitals NHS Foundation Trust (Project number 3127). Given that this project was conducted as a service evaluation, there was no requirement for evaluation by an ethics review board. Patients were informed that participation was voluntary and that the failure to participate would have no effect on their care. Return of the questionnaire was interpreted as implied consent.
Patients were offered up to 20 sessions of augmented PIT. The treatment was delivered by a single therapist (SH) and has been described in detail elsewhere (Howlett et al., 2007; Howlett & Reuber, 2009). Treatment consisted of a 2-h initial semistructured assessment interview followed by up to 19, 50-min therapy sessions at weekly or fortnightly intervals. Individualized treatment was based on a formulation developed in the first session, which took account of relevant predisposing, precipitating, and perpetuating factors. If appropriate, patients were shown cognitive and behavioral strategies such as techniques for warding off threatened PNES, panic attacks, and flashbacks, and offered relaxation tapes.
At the end of therapy, the intervention was categorized by the therapist as “completed,”“terminated by client,” or “terminated by therapist.”“Completion” was completion of a number of sessions prospectively agreed between patient or therapist, or agreement that no more sessions were needed. Therapy was “terminated by the client” if treatment had started but was discontinued before the therapist regarded it as complete. The therapist may have terminated therapy if the patient was already receiving psychological treatment elsewhere, or was considered more suitable for treatment in a mental health or learning disability setting.
Data were analyzed using SPSS (version 15; SPSS Inc., Chicago, IL, U.S.A.). Mann-Whitney U-tests were used to consider any differences in baseline measures between the patients who did not return the follow-up questionnaire and the patient group reported here. Seizure outcome data were expressed as percentage of improvement in seizure frequency from baseline to follow-up (a 50% improvement represented 50% fewer seizures at follow up than at baseline). An increase of the seizure frequency was expressed as a negative percentage. Two cases experienced a worsening of seizure frequency by more than 100% (−500% and −900%, respectively). For the correlational analyses, these were replaced by −100% (in view of the fact that patients could only get 100% better).
To examine possible relationships between clinical, self-report, and demographic features and seizure outcome, the total patient group was split into four outcome categories depending on percentage improvement in seizure frequency: Seizure free (0 seizures for at least 6 months), improved (>50% reduction in seizure frequency but not seizure-free), unchanged (+49 to −49% change in seizure frequency), and worse (>50% increase in seizure frequency). Frequencies and descriptive statistics were calculated for each outcome category. In view of the small group sizes, no statistical comparisons were made between the four categories.
The relationship between continuous variables and the percentage of seizure frequency improvement was examined using Spearman correlations. To examine potential predictors of complete seizure-control after psychotherapy, the group was also split into those who were seizure-free at follow-up and those who continued to have seizures. Potentially differentiating features in these two groups were compared using the Mann-Whitney U-test. Logistic binary regression analysis was carried out to identify predictors of seizure cessation. Mann-Whitney U-tests were used to examine differences across the groups between baseline and follow-up and to examine differences between economic activity groups.
To measure changes in the number of health care contacts in the last 3 months between baseline and follow-up, Wilcoxon signed ranks tests were used. The mean number of contacts was calculated by taking the midpoint of each category (no contact = 0, 1–5 contacts = 2.5, 6–10 contacts = 7.5, and 10+ contacts = 12.5).
Of 108 consecutive patients with PNES who attended for psychotherapy and 66 patients who were approached to provide follow-up information (12–65 months after end of treatment), 47 (71%) returned their questionnaires (see Fig. 1 for further details).
The patients who returned the follow-up questionnaire did not differ significantly from those who failed to return the questionnaire in terms of age, age of onset, gender, duration of PNES, seizure frequency at diagnosis, or other baseline measures: employment status, health care utilization, PHQ-15, CORE, or SF-36 mental health component scores. The patients who did not return the follow-up questionnaires had a shorter follow-up period [mean 40.9 months, 95% confidence interval (CI) 31–49 months, compared to mean 47.8, 95% CI 37–57 months, p < 0.014] and lower SF-36 physical health component scores (mean 45.3, 95% CI 36–54, compared to 36.8, 95% CI 22–50, p < 0.008).
Table 1 lists the demographic and clinic characteristics of the patient group at baseline. Because the scores on the measures were not normally distributed, median values and ranges are listed rather than means and standard deviations.
Table 1. Median values and ranges of demographic, clinical, psychological, and health care utilization variables at baseline in all patients and in different outcome categories
Proportion of female patients
Age in years (range)
Age at onset in years (range)
Duration in years (range)
Seizure frequency/month (range)
Proportion of “economically active” patients
Health care contacts in last 3 months (percent)
PHQ-15 score (range)
CORE score (range)
SF-36 Mental Health Component Summary (range)
SF-36 Physical Health Component Summary (range)
Twelve of the 47 respondents (25.5%) had been seizure-free for at least 1 month at conclusion of psychotherapy. Only seven of these patients had been seizure-free in the 12 months before receiving the follow-up questionnaire. However, an additional five patients not seizure-free at the end of therapy achieved seizure cessation subsequently.
Table 2 provides further information about the details of psychotherapy and health care utilization in the whole group and different outcome categories at the time of follow-up. At this time, a median of 50 months (range 31–65 months) after the baseline data were collected, and a median 42 months (range 12–61 months) after end of treatment, 12 (25.5%) of 47 patients included in the study were seizure free: 11 had been free of seizures for at least a year and 1 for at least 6 months. A further 19 (40.4%) reported a >50% reduction in seizures frequency compared to baseline. The median seizure frequency across the groups had improved from six to one per month (p < 0.007). Despite the improvements in seizure frequency, the percentage of patients categorized as “economically active” declined from 46.8% at baseline to 38.3% at follow-up (difference not significant, p < 0.058).
Table 2. Median values and ranges for follow-up data
Time from referral to follow-up in months (range)
Time from end of psychotherapy to follow-up in months (range)
Seizure frequency at follow-up per month (range)
87.5 (−11 to 100)
0 (−33.33 to 40)
−100 (−100 to 50)
Proportion of “economically active” patients
Health care contacts in last 3 months (percent)
Number of sessions (range)
Reason for ending therapy
There was a significant decline in health care utilization at follow-up. The median health care utilization category dropped from 3 (5–10 contacts in 3 months) to 2 (1–4 contacts) (p < 0.039). The number of health care contacts in 3 months dropped by a mean of 1.7 between baseline (mean 6.2) and follow-up (mean 4.5). Based on the minimum cost for a health care contact as £36 ($60, based on an 11.7 min general practitioner consultation without prescription costs) and assuming that this effect was related to the therapeutic intervention and maintained for 1 year, the direct savings for outpatient health care costs during this time were £244.80 ($408).
No significant correlations were found between percentage of seizure frequency improvement and continuous baseline variables (age, age at onset, duration, baseline seizure frequency, PHQ-15, CORE, SF-36 MHC, SF-36 PHC scores). When patients who had achieved complete seizure control were compared to those still experiencing PNES at follow-up, only baseline “economic activity” status (active vs. inactive) was found to differ between the two groups. Patients who became seizure-free after psychotherapy were more likely to have been “economically active” at baseline (p < 0.025).
A binary logistic regression was carried out to identify baseline variables predicting seizure cessation. “Economic activity” status at baseline was the only variable included in the model that significantly predicted seizure cessation at follow-up. The model explained 74.5% of variance (p < 0.021). Patients categorized as “economically active” at baseline were 5.08 times more likely to become seizure free. All other baseline variables failed to contribute to the model at p < 0.05 significance level. To learn more about the reasons that “economic activity” status was so strongly predictive of seizure cessation after psychotherapy, we carried out some additional comparisons between “economically active” and “inactive” patients. At the time of referral for psychotherapy the “economically active” group had a shorter history of PNES (median duration of 1.75 years vs. 9 years, p < 0.023), better SF-36 PHC scores (median 42.25 vs. 27.72, p < 0.003), and a lower seizure frequency (median 2 per month vs. 20 per month, p < 0.003). “Economically active” patients were also more likely to complete psychotherapy (70.8% vs. 43.5%, p < 0.048).
Correlations between follow-up data and percentage of seizure frequency improvement showed only shorter time from referral (rs = 0.28 p < 0.028) and end of psychotherapy to follow-up (rs = 0.28 p < 0.03) to be significant. Comparisons of follow-up data (economic activity status, health care utilization, number of sessions, and reason for ending psychotherapy) between seizure-free versus nonseizure-free groups yielded no significant findings.
We have previously shown that brief augmented psychodynamic therapy can improve health-related quality of life and emotional distress in patients with functional neurologic symptoms including PNES (Reuber et al., 2007). Our present study indicates that this intervention is also associated with significant improvements of seizure frequency and health care needs: 25.5% of patients were seizure-free a mean of 3.5 (1–5) years after psychotherapy, and a further 40% reported a >50% improvement of their seizure frequency. The rate of patients achieving seizure-freedom is a little lower than that reported in some studies charting the “natural” course of PNES, although it compares favorably with that reported in a pilot study of a cognitive behavioral intervention (Goldstein et al., 2004).
The most important reason for the apparent similarities in outcome reported here and in “natural history” studies is that these studies have captured consecutive patient cohorts at the time of diagnosis by a neurologist and obtained follow-up data from all patients (Lempert & Schmidt, 1990; Walczak et al., 1995; Selwa et al., 2000; Quigg et al., 2002; Carton et al., 2003; Bodde et al., 2007; O’Sullivan et al., 2007). The current study is based on a different group of patients, that is, patients who had completed at least one session of psychotherapy. Our study does not include patients who stopped having seizures after the diagnosis had been explained to them by a neurologist, or patients who decided that they needed no further therapeutic intervention after receiving the diagnosis (we have recently shown that, at least in the short term, 14% of patients cease to experience PNES at this point and 20% decide not to see a therapist although they told their neurologist that they wanted to be referred) (Howlett et al., 2007; Hall-Patch et al., 2010).
We have previously calculated that our brief augmented psychodynamic interpersonal intervention costs an average of £231 (E264, $369), and estimated that the cost of one quality adjusted life year gained through the intervention was modest, at £5,328 (E6,091, $8,511) (Reuber et al., 2007). The present study suggests that our intervention is not only cost-effective but may actually save health care expenditure overall. At baseline, many of our patients were heavy users of medical services, with >50% of patients reporting more than five health care contacts over 3 months. At follow-up, only 30% of patients were reporting as many contacts. Based on the average decline of 1.7 health care visits over 3 months at follow-up, the direct health care cost savings would be at least £244.80 ($408) per patient per year. This is likely to be a gross underestimate of the actual savings because this figure is based on the minimum possible cost of an 11.7 min primary care consultation without drug prescription. Some visits would have involved the prescription of medication, more than doubling the cost, and many of the contacts would have involved visits to hospital emergency departments, use of ambulance services, or hospital outpatient attendances to see specialist doctors, which would have generated higher costs. Although the potential of the psychotherapy to reduce health care utilization should be proven in a future randomized controlled study, our low estimate suggests that the direct health care cost savings could recoup the cost of psychotherapy within 1 year.
The only predictor of seizure cessation at follow-up was patients being “economically active” at baseline. However, this factor explained an impressive 74.5% of the variance. Although this may reflect how difficult it is for patients to give up financial and social benefits associated with having a disabling seizure disorder, our analysis also identified a number of baseline differences between the “active” and “inactive” patient groups, which may have had an effect on responsiveness to psychotherapy (shorter duration, lower seizure frequency better HRQoL in the active group). These findings are similar to those of a recent large scale outcome study (McKenzie et al., 2010), which suggests that receiving social security benefits is a significant predictor of outcome. The authors also suggest that seizure cessation was a significant predictor of return to employment. Intriguingly, unlike the study of McKenzie et al., in our study, patients whose seizures had stopped at follow-up were not more likely to be “economically active.” An even greater percentage of the patients included in this study were “economically inactive” at the time of follow-up than in the largest naturalistic long-term follow-up study in which many patients never had any psychotherapy (Reuber et al., 2003). In fact, the percentage of patients categorized as “economically active” declined between baseline and follow-up, although the median seizure frequency improved and more than one-fourth of patients were seizure-free at follow-up. These findings are in line with those of Quigg et al. (2002), who also showed that the number of patients on disability benefits increased from diagnosis to follow-up. The decline in the percentage of patients categorized as “economically active” was seen in all outcome categories. This is in keeping with a previous study by Walczak et al. (1995), who suggested that although a significant number of patients showed significant improvements in seizure outcome, improvements in occupational status were seen in only about 20% of PNES patients. This also fits with the findings of a recent study of predictors and outcome of PNES showing that individuals receiving social security payment at the time of diagnosis are unlikely to return to work, even if their seizures stop (McKenzie et al., 2010).
The great impact of “economic activity” status on patients’ chances of achieving seizure freedom suggests that it may be important to diagnose PNES and initiate treatment early. One study showed that 69% of patients were working when they first developed PNES, although only 20% were still working at the time of diagnosis (Martin et al., 2003). In our study, the median duration of PNES before diagnosis was 4 years, with one patient being referred for treatment only 38 years after the manifestation of PNES. Although we did not find duration to be a significant predictor of seizure outcome overall, patients with a longer duration of symptoms were more likely to have an “inactive” employment status at baseline.
The sizes of the different outcome categories were too low to make any definitive statements about links between the completion of psychotherapy and long-term outcome, but it is notable that a particularly high proportion of patients whose seizure frequency was worse at follow-up did not complete psychotherapy.
Our study has a number of important limitations. First, we did not only include patients with “gold standard” diagnoses of PNES (video-EEG proven cases). However, we excluded all cases in which the diagnosis of PNES was in any clinical doubt or in which there was the slightest suggestion of additional epilepsy. Furthermore, because our data are based on a service evaluation capturing a consecutive patient group referred to one outpatient psychotherapy service, our findings may actually demonstrate in a more naturalistic way what psychological treatment can (and cannot) achieve in clinical practice.
Secondly, we cannot be certain that the improvements observed in seizure frequency and health care needs are really due to the intervention. It is not possible to separate the direct effects of psychotherapy from those of the patient’s encounter with the neurologist at time of diagnosis, after which antiepileptic drug treatment would have been withdrawn (if it was previously prescribed). In some cases, the neurologist or the patient’s primary care physician may also have prescribed antidepressant or anxiolytic medication, which may have had beneficial effects.
Next, it is possible that our outcomes were affected by selection bias. We are unable to present long-term outcome data on patients who were diagnosed with PNES but who failed to attend for at least one appointment with the psychotherapist. Although the reason for nonattendance may have been “spontaneous” seizure-cessation (suggesting that our outcomes were achieved in a more difficult patient group), it is also possible that patients with particularly entrenched PNES disorders were not motivated to start psychotherapy and produced a positive bias of our results. Furthermore, there were small differences in the median SF-36 physical health component scores and time to follow-up between our patient group and nonrespondents. However, the small differences suggest that the respondents were more severely disabled at baseline than the nonrespondents, suggesting that it is unlikely that selection bias increased treatment effects.
Nevertheless, we cannot be absolutely certain that there were no systematic differences between these two groups affecting seizure outcome because the lack of statistical significance may be due to the relatively small group sizes in this study.
Furthermore, the interpretation of the seizure outcome data provided is made more complicated by the fact that (in keeping with most studies of the “natural history” of PNES) our follow-up questionnaire did not capture patients at a predefined time-point after therapy.
Our measures of self-assessment of psychosocial functioning at baseline (SF-36, CORE-OM, and PHQ-15) were not found to correlate with seizure outcome. However, it is possible that following psychological treatment these variables could be related to seizure outcome. Without repeat questionnaires after treatment and at follow-up, we cannot be certain of patients’ long-term psychosocial outcome and whether this is related to seizure outcome or employment status. In contrast to previous outcome studies, we did not find any association between other factors (such as age at onset or gender) and seizure cessation (Meierkord et al., 1991; O’Sullivan et al., 2007).
Despite its limitations, this study demonstrates that brief, augmented PIT can be associated with complete seizure cessation in one in four patients and a >50% improvement for a further 40% of patients. The intervention is associated with significant improvements in health care utilization. Our findings justify a randomized control study of the intervention.
This study was supported by the Ryder Briggs Trust. We thank Professor John Brazier for his advice on the calculation of the health care utilization costs.
We confirm that we have read the Journal’s position on issues involved in the ethical publication and affirm that this report is consistent with those guidelines. None of the authors has any conflict of interest to disclose.