Background: Differentiating between first seizure, epilepsy and a non-epileptic event is a challenging clinical exercise for many physicians as it may lead to different therapeutic implications. This study aims to investigate the agreement between the initial diagnosis at the accident and emergency (A&E) department and the final diagnosis following inpatient neurological evaluation of seizure disorders.
Method: A prospective observational study between April 2004 and June 2005 in a regional hospital in Hong Kong recruited 1701 patients from the A&E to neurology/medical wards with initial diagnoses/labels matching any one of 12 predefined keywords which were categorised as either “seizure specific” or “non-specific”.
Results: Among the 1170 patients with “non-specific” initial diagnoses/labels, 58 (5%) were finally diagnosed as having had a first seizure or epilepsy. Among 531 patients with “seizure specific” initial diagnoses/labels, 27 (5.1%) were subsequently diagnosed as having had non-epileptic events. The κ value for agreement between the initial and final diagnosis was 0.88. Of the 154 patients with a final diagnosis of first seizure, 34 (22%) had “non-specific” initial labels. Among these patients, components of the evaluation contributing to revision of diagnosis included retrieval of witness accounts (47%), epileptiform discharges on EEG (47%), short term monitoring in patients suspected of acute symptomatic seizures (28%) and panel discussion of cases (22%).
Conclusion: There was generally a high degree of agreement between the initial and final diagnosis, but first seizures were often missed initially. Careful history taking, judicious use of EEG, selective short term monitoring and liaison with specialists are important in reaching an accurate diagnosis.
Statistics from Altmetric.com
Up to 5% of the population will experience at least one seizure in their lifetime.1 Reaching a diagnosis can be challenging in a person presenting with a suspected seizure because of its varied clinical presentation.2–4 A number of crucial questions need to be addressed, including differentiation of non-epileptic events from epileptic seizures, unprovoked from provoked seizures and ascertaining whether it was the first seizure or was in fact a recurrent event.
In many healthcare systems, the accident and emergency (A&E) department is the first point of medical contact for patients who have had a suspected seizure disorder. Frontline medical staff need to decide on the extent of evaluation and/or admission. In Hong Kong, over 90% of the emergency care service is provided by public hospitals.5 Patients presenting with seizure disorders are often admitted to hospitals because the primary care in Hong Kong, mostly in the form of care paid out-of-pocket, only shoulders a small proportion of acute medical service.6 This setting confers a unique opportunity for the study of first seizures, epilepsy and non-epileptic events.
The present study aimed to investigate the agreement between the initial diagnoses/labels made at the A&E department and the subsequent final diagnoses after inpatient neurological evaluation. We identified components of the evaluation contributing to the revision of diagnosis.
Setting and patients
This was a prospective cohort study of patients with suspected seizures admitted from the A&E department to the adult medical/neurology wards of Prince of Wales Hospital, Shatin, Hong Kong SAR, China, between 1 April 2004 and 30 June 2005 (population served 0.64 million5). The adult medical/neurology wards have 350 beds and admit patients 18 years or older.
On every day except Sundays, we routinely reviewed the initial diagnosis/label given by the A&E department of every patient admitted to the adult medical/neurology wards in the preceding 24 h (48 h for patients admitted on Saturdays). The initial diagnosis/label was cross checked with a list of “seizure specific” and “non-specific” keywords, selected based on our experience and in consultation with A&E staff. “Seizure specific” keywords were “seizure”, “epilepsy”, “grand mal”, “convulsion”, “fit”, “generalised tonic–clonic seizure” and “status epilepticus ”. “Non-specific” keywords were “loss of consciousness”, “syncope”, “collapse”, “confusion” and “dizziness”.
The evaluation consisted of standardised history taking from the patients and eyewitnesses of the suspected seizure event, full physical/neurological examination, baseline haematological, renal/liver function tests and ECG. Toxicology screening, 24 h Holter ECG, electroencephalogram (EEG) within 48 h (72 h for patients admitted on Saturday), CT and/or MRI were performed as clinically indicated. A panel of neurologists with expertise in epilepsy (PK, ACFH, HL) discussed any ambiguous cases. A final diagnosis of the event was made usually within 1∼2 days, which was either first seizure (further classified into provoked or unprovoked), epilepsy or non-epileptic event. The diagnosis of seizure and epilepsy was made in accordance with international guidelines.3 7 8 Detailed definitions were given under internet information. In some patients the final diagnosis remained indefinite.
To estimate the completeness of patient inclusion, we reviewed the A&E department’s records of adult attendees who had any one of the 12 initial diagnoses/labels but who were directly discharged home, or admitted to wards other than the medical/neurology wards. We determined whether these patients could also have had a first seizure, epilepsy or non-epileptic events. The number of patients (hence sampling period) required was calculated “a priori” by setting an 80% power of confirming an inclusion rate of >80% for first seizure patients.
Agreement between initial/final diagnoses was calculated using kappa statistics (“seizure specific” with first seizure/epilepsy and “non-specific” with non-epileptic event). Pairwise comparisons between different groups were performed using the Student’s t test, and the completeness of patient inclusion using the z test for proportions. Two tailed p values <0.05 were considered significant. The Statistical Package for the Social Sciences (Windows V.11.5; SPSS Inc, Chicago, Illinois, USA) was used.
During the 15-month period, 18 030 patients were admitted from the A&E department to the adult medical/neurology wards. A total of 1701 patients were found to have initial diagnoses/labels matching the 12 predefined keywords (“seizure specific” in 531 patients, “non-specific” in 1170 patients, mean age 59 (SD 22.1) years, 52.7% men). After evaluation, the final diagnoses were first seizure in 154 patients, epilepsy in 408 patients and non-epileptic events in 1119 patients. The final diagnoses for 20 patients remained indefinite (fig 1).
Among the 1170 patients with “non-specific” initial diagnoses/labels, 58 (5%) were finally diagnosed as having had a first seizure (n = 34) or epilepsy (n = 24). Among the 531 patients with “seizure specific” initial diagnoses/labels, 27 (5.1%) were subsequently diagnosed as having had non-epileptic events (table 1). The κ value for agreement between the initial/final diagnosis was 0.88 (online information). Of the 154 patients with final diagnoses of first ever seizure, 34 (22%) had “non-specific” initial labels. Twenty-four of 408 (5.9%) patients with final diagnoses of epilepsy had initial “non-specific” labels and 27 of 1119 (2.4%) patients with final diagnoses of non-epileptic events had initial “seizure specific” labels.
Revision of initial “non-specific” diagnoses/labels to seizure/epilepsy
Among the 58 patients with initial “non-specific” diagnoses revised to either first seizure or epilepsy, components of the neurological evaluation contributing towards the revision were retrieval of witness accounts of events in 27 patients (47%), epileptiform discharges on EEG recording in the appropriate clinical context in 27 (47%), a history of previous seizures in 24 (41%), short term monitoring of patients for 48 h registering seizure recurrence in 16 (28%) and panel discussion using peri-/post-event information in 13 (22%). More than one component might have contributed to the revision of diagnosis in a given patient.
The underlying aetiologies for these 58 patients were ischaemic cerebrovascular disease in 19 patients (33%), dementia/neurodegenerative diseases in five (9%) and CNS infection in five (9%). The cause was unidentified (cryptogenic) in nine (16%). In 21 patients, the first seizures were provoked by an acute symptomatic cause requiring urgent treatment. A comparison of first seizure/epilepsy patients with initial “non-specific” diagnoses/labels (n = 58) and those with initial “seizure specific” diagnoses/labels (n = 504) showed no difference in age (p = 0.32), sex (p = 0.27) or proportion of generalised/partial seizures (p = 0.41).
Revision of “seizure specific” initial diagnoses/labels to non-epileptic events
Twenty-seven patients with “seizure specific” initial diagnoses/labels had final diagnoses of non-epileptic events. Components of the evaluation contributing towards the revision of diagnosis were further history taking in 23 patients (85%), results from special investigations (including video EEG monitoring, 24 h ECG, autonomic function tests, overnight oximetry, carotid Doppler ultrasound and pituitary function tests) in 14 (52%) and panel review in four (15%). Many events were found to be vasovagal syncope/hypotensive episodes (44%,12/27). Psychogenic seizures and cardiac arrhythmia were found in 26% (7/27) and 11% (3/27), respectively. Other causes included panic attack, Parkinson’s disease, sepsis presenting with chills/rigor, obstructive sleep apnoea and carotid stenosis (1% each). Patients with final diagnoses of non-epileptic events but who had “seizure specific” initial diagnoses/labels (n = 27, mean age 53.4 years) were younger than those who had “non-specific” initial diagnoses/labels (n = 1092, mean age 65.4 years, p = 0.002). There was no difference in sex distribution (p = 0.64).
Completeness of patient inclusion
The sampling period for first seizure patients required to make an estimation with >80% power was ∼9 months (∼100 first seizure patients). Fifteen patients during the period April–December 2004 were considered to have had first seizures, 115 epilepsy and 370 non-epileptic events but were not included in our cohort during this 9 month period. We estimated that the completeness of patient inclusion in the study was 87% for first seizure (103/118), 70% for epilepsy (272/387) and 67% for non-epileptic events (746/1106).
Several studies have investigated patients presenting with seizures to the A&E department, but few have specifically focused on the issue of possible misdiagnosis.9–15 According to one study, the inter-rater agreement of first seizures based on clinical judgement is approximately 0.58 but discussion among neurologists may improve it to 0.86.16 Use of explicit descriptive criteria may also improve the value to 0.73, although in that study EEG information was not considered. In a study involving a paediatric population, the inter-rater agreement of first seizure was also moderate, with κ values ranging from 0.41 to 0.69.17
The optimal management of seizure disorders at the initial point of medical contact remains controversial. Guidelines from the American College of Emergency Physicians (ACEP) provide level B recommendations for performing laboratory tests and CT of the brain in patients who have returned to normal baseline after a first seizure. The ACEP provide level C recommendations for EEGs, possibly in the outpatient setting.2 Guidelines from the Scottish Intercollegiate Guidelines Network (SIGN) advocate referral of patients to a dedicated first seizure clinic for assessment.3 Results from the present study confirm that clear history taking, early EEG and short term monitoring in selected patients, with input from expertise in epilepsy, may help in reaching an accurate diagnosis promptly.
The methodology of this study exploited the characteristics of the healthcare system of Hong Kong, where patients rarely consult the primary care service for seizure disorders but may attend A&E departments directly, from which the great majority are admitted for management. Nonetheless, the proportion of patients who never presented to the A&E department was unknown. To address this issue, a community based study is needed, which may be difficult in the local setting because most primary care physicians operate individually in the private sector.
In conclusion, despite the high degree of agreement between the initial diagnosis at the A&E department and final diagnosis following inpatient neurological evaluation of seizure disorders, first seizures were often missed. Differentiating between epileptic and non-epileptic events is only the first step in the evaluation of patients with epilepsy and such patients should be referred to specialists for accurate classification of the seizure types and underlying epilepsy syndromes which have implications on long term prognosis and management.12 18–21
We thank all staff members of the A&E of Prince of Wales Hospital for their time and interest. We express our gratitude to all the members of the neurology team at the Chinese University of Hong Kong, Prince of Wales Hospital for their support on this research project.
HL and PK took part in conducting, planning and reporting the screening service. The specialist panel comprised HL, PK and ACFH. CYM assisted in the logistics at the A&E Department. HL and PK designed the study and wrote the manuscript. All authors participated in reviewing and editing the manuscript.
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.