Article Text
PDF
Abstract
Background Diagnosing psychogenic non-epileptic seizures (PNES) remains challenging. The majority of ‘PNES status’ cases are likely to be seen in the emergency department or similar non-specialised units, where patients are initially assessed and managed by physicians of varying expertise in neurology.
Methods 216 participants including medical students and doctors of all grades from a wide range of medical disciplines were shown video recordings of six patients with PNES and six other patients with convulsive epileptic seizures (ES). Participants were asked to choose between PNES and ES as a diagnosis and to rate their confidence in each diagnosis, both before and after a 15-minute teaching presentation on PNES and ES.
Results Pre-teaching sensitivity for diagnosing PNES was 0.77, specificity 0.55. The positive predictive value (PPV) of diagnosing PNES was 0.63, and was 0.7 for ES. Diagnostic accuracy increased with increasing clinical grades (p=0.022), as did clinical confidence (p<0.0005). Clinical accuracy and clinical confidence increased post-teaching (p<0.0005). Sensitivity for diagnosing PNES post-teaching improved to 0.88, specificity to 0.67. The PPV of diagnosing PNES increased to 0.72, and to 0.84 for ES.
Conclusions Diagnosing PNES can be improved by clinical experience in neurology and focussed teaching interventions.
- Psychogenic non-epileptic seizures
- video-EEG
- education
- epilepsy
- sensitivity
- specificity
Statistics from Altmetric.com
Introduction
Psychogenic non-epileptic seizures (PNES) are episodes of altered movement, sensation, or experience that resemble epileptic seizures (ES) but have a psychological origin. Diagnosing PNES remains challenging and can often be delayed by over 7 years.1 The misdiagnosis of PNES is associated with iatrogenic risk to patients from anti-epileptic drugs use and invasive interventions such as intubation and ventilation in intensive care units.2 The delay in diagnosis represents a lost opportunity to initiate treatment of the underlying cause, and a prolonged history of PNES is associated with worse outcome.3 The financial costs of misdiagnosing PNES as epilepsy are high for both the patient and the society. US-based studies estimate up to US$900 million is unnecessarily spent annually on repetitive diagnostic tests, anti-epileptic drugs and emergency department utilisation in relation to PNES.4
Despite estimated incidence rates of PNES being much lower than epilepsy at about 3/100 000 per year,4 disproportionately high rates of PNES patients are seen in ‘PNES status’.5 The majority of cases are likely to present to the emergency department or similar non-specialised units, where patients are initially assessed and managed by physicians of varying expertise in neurology and in PNES in particular. Not uncommonly, the neurologist is involved only after the patient has been intubated and sedated in intensive care units following a cascade of treatments for apparently refractory status epilepticus. This scenario is more likely when patients present in PNES-status and no reliable history is available. Research has largely focused on the inter-rater reliability measures of epileptologists when analysing the ‘gold-standard’ video-EEG recordings and diagnosing PNES.6 There are no data available on PNES diagnostic accuracy among non-specialists, and if this is amenable to improvement through education.
We examine the diagnostic accuracy and confidence in the diagnoses reached among medical students, general medical physicians and non-consultant neurologists with varying grades of neurology experience when presented with video recordings of video-EEG confirmed PNES or convulsive ES. No clinical details were given, to approximate the real-life scenario of a convulsive seizure and unconscious patient being witnessed in a casualty setting. A simple teaching programme on recognition of PNES was instigated, and we tested participants before and after this teaching intervention.
Methods
Patients and videos
The study was approved by the institutional review board at the Royal London Hospital. Written informed consent for education and research was obtained from all patients (or guardians of patients) whose videos were shown in the study. Video recordings of six patients with PNES and six other patients with ES were obtained from EEG-video monitoring at the Royal London Hospital. The diagnoses of PNES or ES were confirmed in our epilepsy group meeting. The recordings all demonstrated the onset of the seizure episode, the seizure and immediate aftermath. Video recordings were selected based on the presence of generally accepted semiological clues indicating either PNES or ES, and were all less than two minutes duration for practical teaching purposes.
Participants
Participants were recruited from medical students attending Barts and the London School of Medicine and Dentistry, medical trainees (junior house officers/interns/FY1/FY2 ie, first 2 years post-qualification; senior house officers/junior residents/CT1-3 ie, years 3–5 post-qualification; registrars/senior residents/ST3-7 ie, years 5–9 post-qualification) and consultants attending London neurology training days in 2009–2011. Information was collected on each participant's specialty, grade, number of years of experience in neurology, and any prior experience of knowingly treating PNES and status epilepticus.
Presentations and questionnaires
Participants were shown the 12 video recordings in a random order and asked to choose between PNES or ES as a diagnosis for each on a pre-teaching questionnaire. They were asked to rate their confidence in each diagnosis for each video on a scale of 1–10, where 1= not confident, 10= very confident.
A 15-min teaching presentation followed the initial video presentations, delivered in a standardised form by SSOS or RR, highlighting semiological features that may suggest PNES instead of ES. See Devinsky et al 2011 for a review.4
After the training presentation, participants were shown the same 12 videos in different order, and were asked to complete a post-teaching questionnaire assessing diagnoses and confidence levels.
Analysis
Statistical analyses were performed using SPSS V.20 for Mac (IBM, 2011).
Clinical confidence was analysed separately for correctly and incorrectly diagnosed cases. Fisher's least significant difference (LSD) was used in post-hoc ANOVAs tests. Participants were grouped into five clinical grades for analyses (table 1).
Clinical Grade and groups of participants
Results
Participant demographics
A total of 216 participants completed the pre- and post-training questionnaires. Participants included medical students, 20 psychiatrists and doctors from 14 different medical disciplines. All of the qualified doctors included would have clinical responsibility for assessing patients in an emergency department setting as part of their duties. 76% of participants had never worked in a neurology post at any stage of their career. 50.3% of respondents had never knowingly treated PNES, while 80.3% had treated status epilepticus.
Clinical accuracy and confidence pre-teaching
The mean number of correctly diagnosed seizures pre-teaching was 7.9 (±1.8) /12 (65.8%), with a confidence score of 5.3±1.9/10 per correctly diagnosed case (figure 1A,B). The sensitivity for diagnosing PNES was 0.77 (95% CI 0.75 to 0.79), specificity 0.55 (95% CI 0.52 to 0.57). The positive predictive value of diagnosing PNES was 0.63 (95% CI 0.60 to 0.65), and was 0.7 (95% CI 0.67 to 0.73), for ES. Previous experience of treating PNES was associated with a higher accuracy of 8 compared with 7.24 in those without experience of treating PNES (t test, t=2.6, p=0.009), and higher confidence levels compared with those without experience of treating PNES (t test, t=2.5, p=0.012). Similarly, higher accuracy (t test, t=3.03, p=0.003) and confidence (t test, t=4.47, p<0.0005) were seen in participants with experience of treating status epilepticus compared with those without such experience (figure 2).
(A) Diganostic accuracy and confidence levels pre- and post-teaching. Mean and SD error bars. Comparisons with t-test. (B) Diganostic accuracy and confidence levels pre- and post-teaching according to the clinical grade of participants (mean, error bars = 1 SD).
Diagnostic accuracy and confidence levels pre-teaching according to previous experience of participants in treating psychogenic non-epileptic seizures (PNES) and status epilepticus (mean, error bars = 1 SD).
Diagnostic accuracy generally increased with increasing clinical grades (ANOVA, f=2.92, p=0.022), as did clinical confidence (ANOVA, f=5.97, p<0.0005). The CT3-ST3 grade and ST7/consultant grades both groups scored higher in terms of accuracy and diagnostic confidence than the students and FY1-CT2 grades (p<0.05, LSD correction). The number of years experience in neurology correlated with clinical accuracy (Spearman's ρ =0.239, p=0.001), but not with confidence levels. Diagnostic accuracy was greater for the PNES cases (4.6/6, 77%), than the ES cases (3.3/6, 55%) (paired t test, t=13.6, p<0.0005). Clinical confidence was higher for the PNES videos (5.9±2.2/10) than the ES videos (4.8±2.4/10), paired t test, t=6.4, p<0.0005.
Clinical accuracy and confidence post-teaching
Clinical accuracy increased to 9.4 (±1.2)/12 (78.3%) post-teaching (paired t test, t=−12.74, p<0.0005), and confidence increased to 7.7±1.5/10 (paired t test, t=−18.86, p<0.0005) (figure 1A,B). Sensitivity for diagnosing PNES post-teaching improved to 0.88 (95% CI 0.86 to 0.89), specificity to 0.67 (95% CI 0.64 to 0.69). The positive predictive value of diagnosing PNES increased to 0.72 (95% CI 0.70 to 0.75), and to 0.84 (95% CI 0.82 to 0.86) for ES. Similar improvements in accuracy were observed for both the PNES and the ES cases, with confidence and diagnostic accuracy remaining higher for PNES than ES cases (paired t test, t>4.3, p<0.0005 for all comparisons). This improvement in diagnostic accuracy and confidence was also seen at all clinical grades when analysed separately (paired t tests, all p<0.0005). Participants with and without prior experience of treating PNES did not differ significantly on accuracy or confidence measures post-teaching. There was a trend for differing accuracy scores post teaching according to clinical grades of participants (ANOVA, f=2.36, p=0.055). A difference in confidence (ANOVA, f=4.41 p=0.002) was seen between clinical grades with the CT3-ST3 grade scoring highest, and significantly higher than the FY1-CT2 and ST4-ST6 grades (p<0.05, LSD correction). No correlations were found between number of years experience in neurology and clinical accuracy or diagnostic confidence after teaching. There was no change in the average clinical confidence levels per case for the videos that were misdiagnosed comparing pre-teaching (5.1±1.5/10) with post-teaching scores (5.0±2.3/10).
Discussion
Our study demonstrates the diagnostic uncertainty that often surrounds PNES among all grades of physicians likely to encounter this condition in the emergency department. This difficulty has also been shown by Benbadis and colleagues, who found only moderate agreement in diagnosing PNES using video monitoring even within an experienced group of neurologists and epileptologists.6 The authors acknowledge that diagnostic confidence in experts is reduced when denied access to a previous medical history, or EEG recordings. These vital clues are often unavailable in emergency departments, where up to 40% of patients with PNES may present to with recurrent PNES-status.5
An encouraging finding of our study is that diagnostic accuracy and confidence can be improved by brief, focused educational interventions. Diagnostic accuracy increased to 78.3%, which compares favourably with a rate of 73.9% seen in epilepsy specialist clinic referrals from mainly GPs and medical physicians, particularly when considering the high proportion of PNES cases in our study.7 An increased awareness of the difficulty in PNES diagnoses should encourage non-neurologists to consider PNES earlier, and to seek help from specialists. Unlike previous studies where experienced epileptologists over-called PNES as epilepsy,8 we found our cohort over-diagnosed ES as PNES. This is likely to represent a bias in respondents who may have been ‘primed’ to diagnose cases as PNES, despite their not being told what proportion of videos were PNES.
The concept of ‘neurophobia’ among medical trainees and practitioners describes a lack of knowledge and confidence in neurology leading to the reputation of neurology being a frightening subject.9 Most patients with PNES and other neurological illnesses in the UK setting are seen first by a general practitioner or a non-neurologically trained general medical team. In these settings, fear surrounding the diagnosis of PNES may outweigh the understanding of the potential harms done by not considering this diagnosis. Our study demonstrates the importance of clinical experience in neurology. The fact that 76% of participants had never worked in a neurology post is also a likely major factor in neurophobia, which will need addressing as part of training policies for doctors. This may be relevant also for training of emergency department specialist physicians, who were not included in our study. While the results of our brief educational intervention are encouraging, longer-term studies are required to assess if the improved outcomes measured endure without regular re-enforcement. We suggest that through clinical experience and increased teaching with re-assessments in neurology at regular intervals at least some aspects of neurophobia are curable across all grades of medical care providers.
Footnotes
-
Competing interests None.
-
Ethics approval The study was approved by the institutional review board at the Royal London Hospital. Written informed consent for education and research was obtained from all patients (or guardians of patients) whose videos were shown in the study.
-
Provenance and peer review Not commissioned; externally peer reviewed.