Incidence, Risk Factors and Consequences of Epilepsy-Related Injuries and Accidents: A Retrospective, Single Center Study

Introduction: This study was designed to evaluate risk factors and incidence of epilepsy-related injuries and accidents (ERIA) at an outpatient clinic of a German epilepsy center providing healthcare to a mixed urban and rural population of over one million inhabitants. Methods: Data acquisition was performed between 10/2013 and 09/2014 using a validated patient questionnaire on socioeconomic status, course of epilepsy, quality of life (QoL), depression, injuries and accidents associated with seizures or inadequate periictal patterns of behavior concerning a period of 3 months. Univariate analysis, multiple testing and regression analysis were performed to identify possible variables associated with ERIA. Results: A total of 292 patients (mean age 40.8 years, range 18–86; 55% female) were enrolled and analyzed. Focal epilepsy was diagnosed in 75% of the patients. The majority was on an antiepileptic drug (AEDs) polytherapy (mean number of AEDs: 1.65). Overall, 41 patients (14.0%) suffered from epilepsy-related injuries and accidents in a 3-month period. Besides lacerations (n = 18, 6.2%), abrasions and bruises (n = 9, 3.1%), fractures (n = 6, 2.2%) and burns (n = 3, 1.0%), 17 mild injuries (5.8%) were reported. In 20 (6.8% of the total cohort) cases, urgent medical treatment with hospitalization was necessary. Epilepsy-related injuries and accidents were related to active epilepsy, occurrence of generalized tonic-clonic seizures (GTCS) and drug-refractory course as well as reported ictal falls, ictal loss of consciousness and abnormal peri-ictal behavior in the medical history. In addition, patients with ERIA had significantly higher depression rates and lower QoL. Conclusion: ERIA and their consequences should be given more attention and standardized assessment for ERIA should be performed in every outpatient visit.


INTRODUCTION
Epilepsy is a common and chronic neurological disorder that affects about 39 million people worldwide (1,2). People with epilepsy are subject to social and vocational stigma, have only restricted access to the labor market and significantly reduced employment opportunities (3). Moreover, quality of life (QoL) is significantly reduced for themselves and their caregivers (4)(5)(6)(7)(8)(9). Active epilepsy with persisting seizures is associated with loss of consciousness, uncontrolled movements, falls or periictal abnormal behavior may predispose to accidents and injuries such as burns, contusions, lacerations or fractures (10)(11)(12).
A prospective longitudinal analysis from Finland on 245 children with epilepsy since 1964 showed a significantly increased age-and sex-adjusted mortality. During the 40-year follow up, 60 (24%) subjects died and 33 (55%) of these events were attributed to the underlying epilepsy. Besides sudden unexpected death in epilepsy (SUDEP) and status epilepticus, epilepsy-related injuries and accidents (ERIA), such as peri-ictal drowning, were common causes of death. Within this cohort, pneumonia, cardiovascular diseases and suicide have been reported as most frequent causes of death not related to seizures or epilepsy (13)(14)(15). ERIA were shown to be a major cost factor for hospitalizations among patients with epilepsy (16). One major problem with ERIA is unreported cases in daily practice. If not investigated in detail, it can be assumed that approximately 50% of ERIA are falsely not documented as seizure-related in medical documentation (17).
The main aim of our study was to assess the frequency and types of ERIA in a cohort of consecutive patients with epilepsy using a validated questionnaire (18) and to search for variables associated with ERIA. Improved screening parameters for ERIA may help in reducing the frequency and severity of ERIA. A second aim was to assess QoL and depression as possible consequences of ERIA.

Study Settings, Patients and Design
This study was performed as a survey at the epilepsy outpatient clinic of the University Hospital Marburg. The University Hospital Marburg is a large multispecialty tertiary care hospital in the center of Germany that provides healthcare to a population of over 1,000,000 inhabitants. Marburg is located within the postal code area 35, which was used previously for a population-based estimation of the incidence of status epilepticus and different cost-of-illness studies (3,(19)(20)(21). After receiving written informed consent, all adult epilepsy patients aged 18 years or older were eligible. Socioeconomic data and information on injuries or accidents as well as on QoL were accessed between 10/2013 and 09/2014 using a validated questionnaire (18). Patients were asked to complete the questionnaire concerning their individual epilepsy related experiences during the last 3 months. The diagnosis was based on the definitions proposed by the International League Against Epilepsy and the International Bureau for Epilepsy (22) and was provided by the treating physician. Any patient with at least one seizure during the last 12 months at enrolment was classified as having an "active epilepsy." Patients were excluded when the diagnosis of epilepsy could not be determined without doubt. The study had the approval of the local ethics committee.

Variables Associated With Epilepsy-Related Injuries and Accidents
Specification of epilepsy syndromes as well as information regarding intake of AEDs was assessed by the attending physician. The following factors were identified as possible parameters influencing injuries or accidents in patients with epilepsy according to published literature (18): patient age, duration of epilepsy, sex, epilepsy syndrome, postictal patterns of behavior as well as seizure frequency and semiology (loss of consciousness, nocturnal seizures). Abnormal postictal behavior patterns were classified in line with the guidelines of the German employers' liability insurance association (Unfallversicherung, DGUV 250-001), which classifies seizures with abnormal postictal behavior as "high risk" for related injuries or accidents at work (23). Refractory epilepsy was classified according to ILAE definitions (24). Moreover, different sociodemographic and clinical characteristics were documented.

Changes in Mood and Quality of Life Due to ERIA
To evaluate the psychological aspects of ERIA, various established neuropsychological inventories were assessed using already established questionnaires: To measure health related QoL, the QOLIE31 score (Quality of Life in Epilepsy Inventory) (25) and EQ5D score (Euro Quality of Life) were used (26). For EQ5D, both relevant parameters, i.e., overall TTO (time trade-off) and VAS (visual analog scale), were determined (26). To evaluate health-related depression, NDDI-E (Neurological Disorders Depression Inventory for Epilepsy) (27) and BDI-II (Becks Depression Inventory) were used (28). As an independent parameter, ABNAS (A-B Neuropsychological Percentage of "old" AEDs a 28.8% Percentage of enzyme-inducing AEDs a 13.7% a Enzyme-inducing AEDs: carbamazepine, phenobarbital, primidone, and phenytoin; "old" AEDs: valproate, carbamazepine, phenobarbital and phenytoine. Assessment Schedule) was assessed (29). Liverpool Adverse Events Profile (LAEP) was measured to detect AED-specific aspects in QoL (30,31).

Data Entry and Statistical Analysis
Data entry and analysis were performed using the File Maker Pro 8.5 database (Filemaker Inc., Santa Clara, CA, USA). A double-entry procedure was employed to assure a high level of data accuracy. Statistical analyses were performed using IBM SPSS Statistics 22 (IBM Corporation, Armonk, NY, USA). For the calculation of variables that potentially influence injuries and accidents in patients with active epilepsy, a chi-square test following Pearson and Benjamini-Hochberg-Adjustment was used to exclude multiple testing bias (32,33). In addition, regression analysis was performed. P-values < 0.05 were considered as significant.

Sociodemographic and Clinical Characteristics
We enrolled 292 patients with epilepsy and a mean age of 40.8 years (SD ± 15.6; range 18-86 years). Gender ratio was nearly balanced with 55% female patients. Two-thirds of patients (67.8%) had an active epilepsy with at least one seizure within the last 12 months. The majority of 75% (n = 219) suffered from focal epilepsy (FE), 19.5% (n = 57) from an idiopathic generalized epilepsy (IGE) and, in 5.5%, the epilepsy syndrome was reported as unclassified; details are presented in Table 1. Between the two cohorts with IGE or FE, there were no significant differences in gender distribution, number of AED or seizure freedom. The mean age in the IGE cohort was significantly younger. Detailed information of sociodemographic and clinical characteristics for all patients and the FE and IGE cohorts are displayed in Table 2.

Clinical Characteristics of Patients With Epilepsy-Related Injuries and Accidents
In total, 41 patients (14.0%) suffered from epilepsy-related injuries and accidents (ERIA) within the 3-month observation period. Of these, 20 patients (6.8% of total cohort) required hospitalization. Out of a total of 292 patients, two independent injuries were reported in two cases and one patient suffered from three independent injuries. Besides lacerations (n = 18, 6.2%), abrasions and bruises (n = 9, 3.2%), fractures (n = 6, 2.1%) and burns (n = 3, 1.0%), 17   periictal abnormal behavior in association with ictal falls and ictal loss of consciousness (n = 12, 29.3%). Only two patients with ERIA reported to suffer from sleep-bound seizures only (4.9%). For a detailed list of all patients that reported ERIA see Table 4.

Variables Associated With Epilepsy-Related Injuries and Accidents
Reported sociodemographic and clinical characteristics were analyzed as potential variables associated with ERIA. Comparing patients with (n = 41, 14.0%) to patients without (n = 251, 86.0%) ERIA, we identified an active epilepsy (p < 0.001), regular GTCS (p < 0.001) and a refractory course of disease (p = 0.005) as variables associated with ERIA using a univariate analysis. Moreover, ictal falls (p = 0.012), ictal loss of consciousness (p = 0.022) and inadequate postictal behavior (p = 0.017) were significantly increased in patients suffering ERIA. After post hoc multivariate analysis all reported parameters remained significant. No difference was determined for patient age, duration of epilepsy, gender, epilepsy syndrome or sleep-related seizures in our cohort; for details please refer to

DISCUSSION
This study analyzed in detail a cohort of epilepsy patients in Germany showing the high burden imposed on patients due to ERIA. In addition, patients with ERIA showed a decreased QoL and increased depression rates. Our results are in line with other studies on ERIA, showing that number and frequency of injuries and accidents are significantly increased in epilepsy patients compared to matched controls. However, there are certain limitations of the study design (e.g., a missing epidemiologic approach) that may display limitations of the acquired data and findings. The finding that most enrolled patients had focal and active epilepsy may also affect the comparability with other studies.
In a large database study published in 2010, 8,890 subjects with epilepsy were analyzed, resulting in a 1-year incidence of one or more injuries in 20.6% of patients with and 16.1% of patients without epilepsy (p < 0.001) (34). Active epilepsy and refractory course of disease, as demonstrated in our cohort, are variables associated with ERIA. Within the 3-month observation period, 41 of 292 patients (14.0%) reported ERIA. The most common symptoms were lacerations (n = 18, 6.2%) followed by abrasions and bruises (n = 9, 3.1%), fractures (n = 6, 2.1%) and burns (n = 3, 1.0%). In 20 cases, urgent medical treatment and hospitalization was necessary. These results are well in line with previously published data on ERIA from a pilot study from 2008 among our outpatients (18,35) and other international publications reporting ERIA in children, adolescents and adults with epilepsy in Nigeria, the United States, Australia and the UK (34,(36)(37)(38)(39)(40)(41). In addition, a study reporting on 52 patients with refractory focal epilepsy led to comparable results with epilepsy-related injuries in 21% of cases with TLE and 8% with other focal epilepsies within a mean disease duration of 22 and 17 years. Lifetime prevalence of ERIA was estimated to 57%. Moreover, this study reported that only 45% of ERIA in general, and especially severe ones, had been documented (17). This is in contrast to another study with 247 enrolled patients showing a rather low frequency for ERIA of 1 per 44 patient years and which claimed a minor severity for the most reported cases (41).
Another prospective study analyzed ERIA using a control group of healthy relatives, showing that epilepsy patients have a 21% higher likelihood to harm themselves over a 1-2 year observation period, however, mainly mild and trivial injuries were reported (42). In children aged 2-16 years, ERIA were reported in 4.7% (43) and children with epilepsy were shown to have a 18-23% increased risk of fracture, a 49% increased risk of thermal injury and more than twice the risk of poisoning from medicinal products (39). Differences between the mentioned studies are probably due to diverging study designs (prospective vs. retrospective), different cohort characteristics and may be influenced by local factors as well as awareness and ascertainment issues. ERIA are commonly underreported and there is a high estimated number of underreported cases (17). The high frequency of ERIA in our study can be attributed to the targeted questions and precise documentation reducing underascertainment.
The vast majority of patients reporting ERIA (96.0%) presented specific ictal features and we were able to identify a significant correlation between ERIA with epilepsy characteristics and semiological aspects. Variables associated with ERIA were active epilepsy, drug-refractory course as well as reported ictal falls and ictal loss of consciousness or abnormal peri-ictal behavior in the medical history. These findings are similar to the results of past studies (18). Moreover, the presence of GTCS was significantly associated with ERIA, which is in line with previous studies, in which over 80% of ERIA occurred during GTCS (41). Additionally, we were able to show a significant coherence between ERIA and abnormal peri-ictal behavior, which was previously controversially discussed (18). Patients with frequent and prolonged ictal auras seem to be at lower risk, which is probably based on the possibility of informing bystanders, calling for help and creating a secure surrounding. As another high-risk subgroup, individuals with active epilepsy could be identified. There was no significant difference in the occurrence of ERIA between patients with FE or IGE, which is surprising because of the common presence of GTCS in patients with IGE. Overall, 75% of IGE patients presented with GTCS compared to 51% with FE in an ERIA cohort (40).
A second focus of our study was to assess QoL and depression as possible consequence of ERIA using a battery of neuropsychological inventories (BDI-II, NDDIE, EQ5D, QOLIE-31, ABNAS, and LEAP) that have been already established and validated in patients with epilepsy (25-29). Subjects reporting ERIA showed significantly increased depression rates and had a lower QoL, which has not been reported so far. The association of lower QoL and a trend to depression with drug-refractory epilepsy has been already shown (12,27,28,31,44,45) and ERIA might be associated or rather aggravate these aspects. However, this study was not designed to further analyse and correlate these aspects. Also, adverse effects due to AEDs were significantly higher in patients with ERIA.
The most practical solution to injury prevention is a better seizure control (40,41). However, this may not be achieved, especially in patients with drug-refractory epilepsies who have been shown to be a high-risk group for ERIA. Therefore, another focus should aim at reducing the risk for ERIA. We propose a prophylactic pathway focusing on three major aspects: (1) identifying high risk patients by a standardized assessment of ERIA, (2) optimising antiepileptic therapy, and (3) developing individual safety assessments to reduce the future risk of ERIA (see Table 7). Individual safety assessments may contain information and education on the risks and consequences of ERIA, possibilities to avoid injuries (behavior, early detection) and suggestions for auxiliary means (helmets, hip protectors, in-house emergency calls, wearables). Even if the majority of epilepsy-related injuries and accidents occur at home (88%) and not in a public space (10%), a professional and epilepsyspecific assessment of the actual or aspired profession and the concomitant work environment seems to be unavoidable, hence, the fact that only 2% of epilepsy-related injuries and accidents occur at work (46,47).

CONCLUSION
ERIA, as well as their clinical consequences, are an underreported but highly relevant aspect in the treatment of patients with active epilepsy. Moreover, ERIA might be associated or rather aggravate neuropsychological symptoms such as depression or decreased QOL in this severely affected subgroup of patients with epilepsy. However, ERIA are only sporadically considered in the development of tailored anti-epileptic therapies. More effort should be spent on the identification and assessment of ERIA or patients at risk as well as on prevention and general education on this topic to increase patient safety, satisfaction and QoL.

AUTHOR CONTRIBUTIONS
LW and AS generated the research idea, study design, and concept. NW, SR, and AS acquired the data. LW and AS analyzed the data and drafted the work. LW, NW, SR, LK, AH, SK, FR, and AS made critical revisions for important intellectual content and interpreted the data. LW and AS wrote the manuscript. LW, NW, SR, LK, AH, SK, FR, and AS approved the final manuscript.