Diagnosis of psychogenic nonepileptic seizures (PNES) is confirmed via video-electroencephalography (v-EEG) monitoring in nearly one-quarter of patients evaluated at epilepsy referral centers.1 Etiologically, PNES have been linked to a dysfunction in the processing of psychological or social distress.2 Treatment referrals to mental health specialists are common practice in the management of PNES;3 however, little is known about whether PNES patients follow such recommendations.
Effective psychotherapeutic and psychopharmacological interventions for PNES are emerging.4–6 As more of these interventions receive empirical support, it is relevant to understand treatment context and what factors may predict and facilitate patient participation.
PNES patients have been regarded as a difficult-to-engage population.7 There is a previous report showing adherence to at least one psychotherapeutic session after presentation of the diagnosis; however, sustained engagement in a structured clinical program has not been formally evaluated.8
In this retrospective, observational study, we aim to 1) compare levels of initial adherence in PNES subjects referred to our clinical program, based on the recommended treatment setting; and 2) delineate baseline subject differences between those patients who stayed in treatment for the first several weeks and those who did not. We hypothesized that 1) receiving all mental health services within the same clinic (integrated treatment) would render higher rates of adherence during the initial phase of treatment, as compared with receiving different aspects of treatment by providers at different institutions (divided treatment); and 2) certain psychosocial factors and clinical variables, such as being single, delay in diagnosis, more severe psychopathology, and certain comorbidities, such as epilepsy, would be more common in nonadherent patients.
Patients were recruited as they underwent a comprehensive clinical evaluation at the Non-Epileptic Seizures Intervention Clinic (NESIC) at the University of Illinois at Chicago Medical Center. All patients had been diagnosed with PNES by the referring epileptologist through v-EEG monitoring capturing a typical attack or, in a few instances, through routine EEG during the typical attack. IRB approval and subject consent for participation were obtained.
After the initial outpatient neuropsychiatric evaluation, patients and their families attended a feedback session where the psychogenic origin of the attacks and treatment recommendations were discussed. The feedback session took place in an outpatient setting 1 to 2 weeks after the initial neuropsychiatric evaluation. Delay between discharge from the Epilepsy Monitoring Unit and feedback session was variable, based on first patient contact and appointment availability, usually ranging from 2 to 6 weeks. A neuropsychiatrist (G.B.), a psychologist (E.P.) and a neuropsychiatric social worker were all present during the feedback session. Treatment recommendations were always discussed with the referring epileptologist before being presented to the patients and their families. During the feedback session, the discussion offered an explanation for the psychogenic origin of the events, using information from the history and risk factors obtained during the initial evaluation, and the discussion encouraged engagement in treatment to help reduce vulnerability to the attacks.
To analyze follow-up data, patients were classified into two subgroups on the basis of the treatment offered: 1) the "integrated" treatment subgroup (psychotherapy and psychiatric management offered through the same clinic at the University of Illinois Medical Center); or 2) the "divided" treatment subgroup (psychiatric management was offered at the University of Illinois Medical Center, but psychotherapy was offered elsewhere). The decision to offer "divided" instead of "integrated" treatment was based on geographical limitations (i.e., patients could come every 2–4 weeks for a psychiatric appointment, but not weekly for psychotherapy) or the patient having an established relationship with a therapist willing to work in cooperation with our team.
Adherence was defined in the Integrated-treatment subgroup as attending at least five psychotherapy sessions over the course of 2 months. Adherence was defined in the Divided-treatment subgroup as attending at least two follow-up appointments with the psychiatrist while remaining active in outside psychotherapy over the course of 3 months. These definitions of adherence were established by clinician consensus. The number of attended appointments needed to achieve adherence status was regarded as a reasonable indicator that patients were actively participating in treatment. For instance, in the Integrated group, five sessions were necessary to complete the first two modules of the psychotherapeutic intervention conducted. In the Divided group, two psychiatric appointments were consistently paralleled to five or more psychotherapy sessions with outside providers. Classification of each subject as Adherent or Nonadherent was discussed in weekly meetings based on the agreed-upon criteria.
Differences Between Retention Groups
Subjects whose adherence was previously determined were classified as Adherent and Nonadherent regardless of treatment modality. First, adherence was analyzed based on treatment modality. Adherence was treated as the dependent variable, with treatment modality being considered one special characteristic upon which adherence was analyzed. Then demographic, and clinical variables obtained during the initial neuropsychiatric assessment were also analyzed to determine whether they differed on the basis of subjects’ initial adherence. Clinical variables included characteristics of the PNES (i.e., frequency, age at onset, duration of each episode, delay in diagnosis), detailed history of medical, neurological, and psychiatric comorbidities, and questionnaires measuring psychopathology severity. Variables that were significantly different by adherence group and treatment modality were tested in a logistic-regression model to determine ability to predict initial adherence.
SPSS software, Version 2.0 was used to perform chi-square analysis for categorical variables, and ANOVA was used for continuous variables between the two treatment modalities as well as between the adherent and nonadherent groups. Logistic regression, with adherence as the dependent variable, was used to determine factors predictive of adherence.
Demographic and Follow-Up Data
Of the 104 patients who completed the initial evaluation, 4 never returned for a feedback session, and 15 were referred to receive all mental health treatment elsewhere; we were therefore unable to track their adherence. Of the remaining 85 subjects, 69 were offered Integrated treatment, and 16 were offered Divided treatment. Demographic data for the whole sample and by treatment group are shown in Table 1. There was no statistically significant difference in demographic data between the two treatment modalities. Regarding adherence, 37 of the 69 patients in the Integrated treatment subgroup (53.62%) and 5 of the 16 patients in the Divided treatment subgroup (31.25%) initially followed our recommendations based on the parameters established above for each subgroup. The chi-square value comparing adherence rates by treatment group was x2=2.83 (p=0.09).
TABLE 1.Demographic Variables for the Whole Sample (N=104) and by Treatment Modality Group
| Add to My POL
|Demographic Variable||Integrated Treatment (N=69)
N (%) or Mean (SD)||Divided Treatment (N=16)
N (%) or Mean (SD)||Whole Samplea (N=104)
N (%) or Mean (SD)|
| Male||13 (18.8)||4 (25%)||82 (78.8)|
| Female||56 (81.2)||12 (75%)||22 (21.2)|
|Age, years||37.38 (14.43)||32.44 (13.00)||36.19 (13.68)|
| White||44 (63.8)||9 (56.3)||69 (66.3)|
| Nonwhite||25 (36.2)||7 (10.1)||35 (33.7)|
| Married (or live-in partner)||24 (34.8)||6 (37.5)||37 (35.6)|
| Single||32 (46.4)||8 (50)||50 (48.1)|
| Divorced||12 (17.4)||2 (12.5)||16 (15.4)|
| Widowed||1 (1.5)||0 (0%)||1 (1)|
| Employed||30 (43.5)||7 (10.1)||41 (39.4)|
| Unemployed||14 (20.3)||4 (25%)||22 (21.2)|
| Disabled||18 (26.1)||3 (18.8)||30 (28.9)|
| Student||7 (10.1)||2 (12.5)||11 (10.2)|
| ≤12 years||26 (37.7)||6 (37.5)||35 (33.7)|
| 13–16 years||34 (49.3)||9 (56.3)||56 (53.9)|
| > 16 years||9 (13)||1 (6.3)||13 (12.5)|
Differences Between Retention Groups
The 85 subjects whose initial adherence was tracked were classified as Adherent (N=41) and Nonadherent (N=44) regardless of treatment modality offered. Table 2 shows the values of all variables analyzed by adherence groups. Rates of marital status (or having a live-in partner), current exposure to antiepileptic drugs, and subjective cognitive complaints were different between the adherence groups. A logistic-regression analysis was conducted to predict adherence, using treatment modality and the three characteristics found to differ based on adherence. The model was significant (R2=0.26; p <0.01). The Wald criterion demonstrated that only marital status (Wald=4.56; p <0.05) and subjective cognitive complaints (Wald=4.16; p <0.05) made a significant contribution to prediction of adherence. Exp (β) values (odds ratio) indicate that PNES patients who are married or have a live-in partner are 3.14 times more likely to be adherent and that PNES subjects without cognitive complaints are 2.85 times more likely to be adherent.
TABLE 2.Demographic and Clinical Variables by Adherence Group (N=85)
| Add to My POL
|Demographic and Clinical Variable||Adherent Group (N=41)
N (%) or Mean (SD)||Nonadherent Group (N=44)
N (%) or Mean (SD)|
| Male||11 (26.8)||7 (15.9)|
| Female||30 (73.2)||37 (84.1)|
|Age at initial evaluation, years||37.20 (13.56)||35.41 (14.93)||NS|
| White||27 (65.9)||26 (59.1)|
| Nonwhite||14 (34.1)||18 (40.9)|
| Married (or live-in partner)||20 (48.8)||10 (22.7)|
| Not currently married||21 (51.2)||34 (77.3)|
| Employed||19 (46.3)||18 (40.9)|
| Unemployed||11 (26.8)||7 (15.9)|
| Disabled||7 (17.1)||14 (31.8)|
| Student/ Retired||4 (9.8)||5 (11.4)|
| ≤12 years||14 (34.1)||17 (38.6)|
| 13–16 years||21 (51.2)||23 (52.3)|
| > 16 years||6 (14.6)||4 (9.1)|
| Age at onset, years||32.80 (14.36)||27.89 (14.97)||NS|
| Attacks per month||42.05 (99.56)||19.66 (39.34)||NS|
| Delay in diagnosis (months)||40.46 (44.03)||70.18 (99.97)||NS|
| Duration (minutes)||12.58 (23.48)||16.14 (29.84)||NS|
| Epilepsy||7 (17.1)||11 (25)||NS|
| Intellectual disability||3 (7.3)||5 (11.4)||NS|
| Depressive disorder||32 (78)||31 (70.5)||NS|
| Posttraumatic stress disorder||8 (19.5)||16 (36.4)||NS|
| Panic disorder||10 (24.4)||14 (31.8)||NS|
| Other anxiety disorders||12 (29.3)||12 (27.3)||NS|
| Dissociative disorder||9 (22)||10 (22.7)||NS|
| Somatoform disorder||8 (19.5)||13 (29.5)||NS|
| Active substance abuse||3 (7.3)||6 (13.6)||NS|
| Borderline personality disorder||4 (9.8)||8 (18.2)||NS|
| Cognitive complaints||17 (41.5)||30 (68.2)||p<0.05*|
| Eating disorder||3 (7.3)||2 (4.5)||NS|
| Other clinical features|
|Current AED||12 (29.3)||23 (53.3)||p<0.05**|
|History of abuse||24 (58.5)||30 (68.2)||NS|
|Past psychiatric treatment||26 (63.4)||29 (65.9)||NS|
|Past suicide attempts||7 (17.1)||12 (27.3)||NS|
|History of traumatic brain injury||14 (34.1)||13 (29.5)||NS|
|Beck Depression Inventory-II||17.99 (13.3)||19.43 (12.57)||NS|
|DASS-D||10.38 (11)||10.77 (11.35)||NS|
|DASS-A||10.49 (7.9)||11.05 (9.73)||NS|
|DASS-S||12.83 (10.61)||14.27 (12.5)||NS|
|DES||12.52 (14.68)||19.24 (20.13)||NS|
|PHQ-15||11.21 (5.85)||11.17 (5.24)||NS|
|DFI||3.63 (2.33)||3.57 (2.16)||NS|
A previous report established 100% attendance to at least one psychotherapy session 2 years after the diagnosis of PNES was presented to 48 patients in a structured way. Level of adherence over the initial course of treatment and relationship to baseline characteristics was not specified in this study.8
Our data indicate that initial treatment retention in this population is higher when integrated mental health treatment is offered at the same institution than when divided treatment is offered, with the difference showing a trend toward statistical significance. Early treatment dropout in other outpatient psychiatric clinical samples usually ranges between 20% and 50%,9,10 placing our "integrated" treatment group at the high end of this range of dropout, but the "divided" group at a much higher rate.
A number of factors have been postulated as contributing to PNES subjects’ difficulty in engaging in treatment, including psychological factors and driving restrictions.7 We did not find specific psychopathological factors that distinguished the adherent versus nonadherent subgroups. Delay in diagnosis, also hypothesized to differ between adherent and nonadherent groups, was not different between the groups.
In our sample, being married or having a live-in partner was associated with higher initial treatment retention. This is consistent with the association between higher early dropout rates and lower social support in depression treatment.10 This finding highlights the need to involve available sources of social support to ensure understanding and compliance with treatment recommendations. Further research examining the nature of support for PNES patients may also result in clinical benefits regarding treatment outcome.
Presence of subjective cognitive complaints was significantly more frequent in the nonadherent group. Cognitive impairments in PNES have been proposed to represent a problem with motivation, not necessarily intentional, rather than a manifestation of neuropathology.11 Despite their origin, clinicians should understand cognitive complaints as a factor that limits treatment participation and another reason to maximize external resources to facilitate compliance.
Concurrent use of antiepileptic drugs (AEDs) was significantly more common in the nonadherent group, even when only the cases without comorbid epilepsy were selected in the analysis. Although, in some cases, ongoing use of AEDs is necessary to treat comorbid epilepsy, this may only constitute a minority of PNES patients.12 Fluid interdisciplinary collaboration and clear communication with patients regarding the indications of their AED may clarify any misconceptions that may lead to premature mental health treatment discontinuation. Although some AEDs may contribute to cognitive complaints, the two groups still differed in rate of cognitive complaints when only AED-free subjects were included in the analysis. Although significantly different between adherence groups, concurrent AED use did not significantly contribute to the suggested predictive model of adherence.
The following limitations should be considered when interpreting the results of this study. Our sample consists of PNES subjects who voluntarily agreed to a neuropsychiatric evaluation after the diagnostic long-term monitoring admission, creating a selection bias. Our sample did not account for those subjects who failed to follow through with the initial neuropsychiatric referral. Therefore, rates of adherence from this sample may represent an overestimation of the adherence observed if all patients referred from a long-term monitoring unit were considered. Our analysis also excluded those patients who never returned to discuss treatment options, making our evaluation of adherence restricted to those subjects who were clearly informed of what the treatment would consist of.
Another limitation is the use of different parameters to determine adherence in the Integrated versus the Divided treatment subgroups, because of the different nature of patient contact in each subgroup. Patients who were offered "divided" treatment may have chosen to seek psychiatric treatment elsewhere, especially when geographical limitations were the reason for the divided treatment recommendation, leading to an underestimation of adherence in this subgroup.
This was a retrospective, observational study from a clinical sample, and treatment recommendations were based on factors thought to facilitate clinical outcomes and adherence. As a result, subjects were not randomized to their respective recommended treatment, making the treatment subgroups unequal in size.
Finally, we had to limit our report to the initial phase of treatment because patients were at different stages of treatment at the time of the analysis.
Despite these limitations, our study highlights the difficulty in retaining PNES subjects in mental health treatment, even when they are agreeable to an initial neuropsychiatric evaluation. It also points out factors that should be taken into consideration to maximize adherence. As evidence-based treatments for PNES become available, efforts should be made to create a context conducive to treatment participation.