Mortality due to epilepsy is a significant concern. Patients with epilepsy have a mortality rate significantly higher than that of the general population. The standardized mortality rate (SMR) is shown to be 1.6-9.3 times higher in this population. [1, 2, 3, 4, 5]
Epilepsy-related causes of death account for 40% of mortality in persons with epilepsy and include the following  :
Death due to the underlying neurologic disorder in symptomatic epilepsy
Sudden unexpected death in epilepsy (SUDEP)
Accidents during epileptic attack (ie, trauma, drowning, burning, choking)
SUDEP is defined as sudden, unexpected, nontraumatic, nondrowning death in an individual with epilepsy, witnessed or unwitnessed, in which postmortem examination does not reveal an anatomic or toxicologic cause for the death.
Although studies on SUDEP are heterogenous in methodology and the accuracy of mortality data available, consistent patterns in incidence are obvious.  SUDEP accounts for 8-17% of deaths in people with epilepsy.
In an attempt to standardize the definition of SUDEP, the US Food and Drug Administration (FDA) and Burroughs-Wellcome developed criteria for SUDEP in 1993. These criteria, now used in most SUDEP studies, are as follows:
The patient has epilepsy, which is defined as recurrent, unprovoked seizures
The patient died unexpectedly while in a reasonable state of health
The death occurred suddenly (ie, within minutes)
The death occurred during normal and benign circumstances
An obvious medical cause of death could not be determined at autopsy
The death was not the direct result of a seizure or Status epilepticus
Notably, evidence of a recent seizure does not exclude the diagnosis of SUDEP as long as death did not occur during the seizure.
The FDA/Burroughs-Wellcome also defined the following categories:
Definite SUDEP - Cases meet all criteria and have sufficient descriptions of the circumstances of the death and a postmortem report
Probable SUDEP - Cases meet all criteria but lack postmortem data
Possible SUDEP - SUDEP cannot be ruled out but evidence is insufficient regarding the circumstances of death and no postmortem report is available.
Not SUDEP - Other causes of death are established clearly or the circumstances make the diagnosis of SUDEP highly improbable
Witnessed cases of SUDEP
Only a small portion of definite SUDEP cases have been documented as having been witnessed. Langen et al have reported 15 cases of witnessed SUDEP; 80% of these patients had a seizure immediately before death.  Terrence  reported 24% and Leetsma  reported 38% of witnessed deaths to be an immediate consequence of a seizure attack. Kloster reported evidence of recent seizures (ie, witnessed, oral trauma, cyanosis) in 67% of victims. 
However, in all witnessed deaths, seizures stopped before death, and in many cases, patients regained consciousness. In a few witnessed cases, the immediate event before death was respiratory arrest (obstructive and central). Most victims were reported to have had difficulty breathing before death. Attempts at cardiopulmonary resuscitation were unsuccessful.
An ongoing study has documented SUDEP and has helped with understanding this phenomenon. The MORTality in Epilepsy Monitoring Unit Study (MORTEMUS) is a collaborative project that aims to quantify the risk of death, SUDEP, and near-SUDEP in patients with drug-resistant partial epilepsy during long-term video EEG monitoring. One hundred forty-eight qualified eligible epilepsy centers participated in this study. Responses reflect an estimated total of 133,371 video EEGs and 2,814 patient-years of monitoring. Nineteen deaths were identified, of which 14 were determined to be definite or probable SUDEP. This study is ongoing and the final results are not yet published. A certain pattern has been established by observation in these patients. In most cases within 3 minutes after a seizure, a so-called shutdown of the EEG was noted, followed by asystole and, finally, apnea was the terminal phenomenon.
If SUDEP was shown to have occurred during sleep, a variety of circumstances might have contributed. A recent unwitnessed seizure with or without bradyarrhythmias and lack of sympathetic tone to oppose bradyarrhythmias might have contributed. In addition, obstruction of airways and asphyxia are more probable during sleep.
The interaction between the autonomic control of the cardiovascular functions and the seizure phenomenon is very complex.
Autopsy Findings in SUDEP
Autopsy plays the core role in determining the diagnosis of SUDEP. Knowing the circumstances surrounding the death will contribute. Information obtained from family and friends is extremely important. This information, also known as verbal autopsy, is often overlooked. [11, 12]
Autopsy, per its definition, fails to reveal the underlying cause of death; however, several autopsy reports confirm the following findings in the brain, lungs, heart, and liver of patients with SUDEP. (All of these findings were more frequent in patients with SUDEP than in other patients with epilepsy.)
Studies by Kloster et al,  Earnest et al,  and Engelskjon indicated that only 16% of victims had no significant pathologic findings. In addition to the underlying cerebral pathology in patients with symptomatic epilepsy, cerebral edema was reported in the majority of childhood and adult cases. However, none of the cases showed mass effect due to edema.
In addition, signs of hypoxia in the hippocampal area were noticed in a few instances. Sclerosis of the amygdala has been documented in patients with SUDEP; no significant difference in the amygdala is seen in other patients with chronic epilepsy.
The lungs were heavier than expected in all patients in different studies; lung weights were 110-190% of normal in a study by Terrence et al.  Mild to moderate pulmonary edema with protein-rich fluid and alveolar hemorrhage were seen in all specimens in this study. Other investigators have confirmed the presence of pulmonary edema per weight and histology in 62-84% of cases.
Nonfatal pathologic findings, including fibrosis of the conductive system, have been reported in 33-40% of patients.
Increases in weight and venous congestion, indicating right cardiac failure, were documented in the majority of cases.
Various and overlapping pathophysiologic events may contribute to SUDEP in different patients, and the mechanism is probably multifactorial. Respiratory events, including airway obstruction, central apnea, and neurogenic pulmonary edema, are probable terminal events. In addition, cardiac arrhythmia during an ictal event, leading to arrest and acute cardiac failure, plays an important role.  Furthermore, the autonomic regulation of the heart is impaired in most patients with epilepsy and at baseline, also known as the interictal phase; this can lead to cardiac arrhythmias. Specifically, bradyarrhythmias are shown to be associated with apnea and more pronounced risk of sudden death. Antiepileptic drugs (AEDs) may be another risk factor.  Finally, recent postmortem reviews of SUDEP cases have found an increased frequency of genetic proof for certain channelopathies, especially those that involve both the brain and the heart.
In few observed cases of SUDEP in epilepsy centers, EEG, ECG, and breathing had been documented. The recordings showed the sequence of the events to be an EEG shutdown, followed by asystole and, finally, apnea.
These interactions can be summarized as the events directly related to seizures (see the first diagram below) and the baseline (see the second diagram below).
Central apnea syndromes are characterized by cessation of spontaneous respiratory drive during sleep. Seizures are known to cause central apnea by direct propagation of the electrical discharge to the respiratory center. Episodes of apnea lasting 10-63 seconds, accompanied by a significant fall in oxygen saturation, have been documented. In addition, cardiac arrest can cause secondary cardiopulmonary arrest. So et al documented a case of near-SUDEP due to postictal apnea.  These 2 mechanisms may contribute significantly to the pathophysiology of SUDEP.
Asphyxiation secondary to an obstructive cause has been postulated to play a role in the deaths of patients who were found in a prone position at the time of death. Several investigators reported the prone position more frequently than statistically expected (eg, 81%,  71%  ). The prone position may affect ventilation by obstructing the upper respiratory tract as well as increasing the chances of aspiration. In addition, cases of laryngeal spasm and stridor after a seizure has subsided have been reported.  In unsupervised patients, this might have contributed to death after a seizure.
Neurogenic pulmonary edema has been reported following a variety of neurologic conditions, including epilepsy. Neurogenic pulmonary edema has been documented in 84% of patients in whom autopsy was done following SUDEP. The underlying mechanism for neurogenic pulmonary edema seems to include a massive alpha-adrenergic response, generalized vasoconstriction, and pulmonary hypertension. In addition, the high protein content of the alveoli is indicative of severe damage to the endothelial membranes that leads to increased pulmonary permeability. These findings also indicate that the terminal event lasted longer than a few minutes.
Cardiac arrhythmias also may play an important role as an underlying mechanism of SUDEP. Fatal arrhythmias can occur during the ictal attack and interictally. Many markers of cardiac autonomic dysregulation have been shown to indicate impairment in the baseline in patients with epilepsy. However, one small study has linked this impaired function specifically to SUDEP. 
Erickson systematically studied ictal ECG changes for the first time.  He reported tachycardia, cardiac arrhythmia, and T-wave flattening secondary to a right temporal lobe seizure. Initial bradycardia, followed by tachycardia, was documented in as many as 64% of petit mal and 100% of generalized tonic-clonic seizure attacks.
Subsequent recent studies, documenting simultaneous electroencephalograms (EEGs) and echocardiograms (ECG), reported tachycardia in 74-92% of complex partial seizures. Persistent bradycardia was less common, being reported in 3-7% of complex partial seizures.
Ictal cardiac rhythm and conduction abnormalities have been reported in 5-42% of patients with partial seizures.
Arrhythmias preceding SUDEP have been postulated to be the underlying cause of death. Lathers documented the synchronization of ictal and interictal spikes with cardiac sympathetic activity.
Ictal tachycardia has been documented in 83% of seizure attacks, and bradycardia can accompany as many as 4% of seizures. During the attack, patients presented with prolonged decreases in heart rate, which lasted beyond the seizure attack in most cases. The majority of patients had decreased brain perfusion, with potentially fatal outcome. Similar findings have been documented in other studies.
The potential mechanism for this is propagation of the electrical activity to the amygdala, which has efferent connections, via the central nuclei, to the cardioregulatory centers in the medulla. Arrhythmia can be a consequence of this event. Massive sympathetic surge during a seizure attack and vagal inhibition might be other potential mechanisms for increased ectopic ventricular activity. Also, extreme vagal stimulation might cause heart blocks.
Cardiac arrhythmia during the interictal state is another potentially fatal condition. The evaluation of autonomic cardiovascular reflexes in patients with epilepsy indicates dysfunction of sympathetic component and parasympathetic division. Furthermore, hypofunction of the autonomic cardiovascular reflexes is postulated to be more prominent in patients who also are at high risk for SUDEP and in patients with a more refractory seizure disorder.
Decreased heart rate variability is well known to increase the vulnerability of the cardioregulatory centers, leading to an increase in ventricular automaticity and thus to arrhythmias. The mechanism of dysfunction of the autonomic nervous system (ANS) in epileptic seizures may be multifactorial.
Interictal spikes have been shown to cause arrhythmias in animals. Also, the autonomic control centers may undergo physiologic or anatomic alterations. An example of these changes is the interictal hypometabolism seen in the area adjacent to the epileptic foci on positron emission tomographic (PET) scanning studies. In addition, autopsies of patients with SUDEP have shown fibrosis of the cardiac conductive system in 33% of patients. Repetitive exposure to catecholamines is known to cause myocardial fibrosis. These can act, per se, as new foci for cardiac arrhythmias.
Only 1 study has evaluated ictal tachycardia retrospectively in patients with SUDEP. Nei et al described a higher increase in heart rate during seizures in patients who would experience SUDEP in the future, in comparison with other patients with epilepsy.  To ascertain the true role of cardiac arrhythmia as an immediate event before death, further investigations must be carried out.
Autonomic cardiac arrhythmias may contribute significantly to SUDEP. Decreased baseline heart rate variability is indicative of impaired autonomic cardiovascular reflexes in epilepsy. This can cause an increase in ventricular automaticity, in turn predisposing to arrhythmias. Catecholamine surges during repeated seizures can cause cardiac conduction system fibrosis and arrhythmias. Anatomic and functional changes in the cardiac and pulmonary function are evident and might be a direct or indirect consequence of autonomic dysregulation. In particular, foci of fibrotic changes in the myocardium might contribute to arrhythmias. A recent seizure might cause central apnea, often associated with bradyarrhythmias. Use of carbamazepine is associated with impaired cardiac regulation and with increased risk of SUDEP.
Medication-associated risk factors
The subtherapeutic levels of antiepileptic agents in SUDEP patients might be a reflection of poor seizure control or poor compliance. In addition, hair samples of patients with SUDEP revealed greater variability of AED levels than have been found in other persons with epilepsy.  (However, studies that have compared SUDEP in children to that in adults report that children more often seem to have therapeutic blood levels of antiepileptic medications.)
Antiepileptic medications potentially play a role in modification of the autonomic nervous system functions. In a study of the cardiovascular reflexes in 24 patients with epilepsy, Devinsky et al documented increased heart rate variability that was, at least partially, attributed to carbamazepine.  Other researchers have reported similar findings. Lamotrigine has been studied as another medication with a sodium-channel modulating effect.
In addition, withdrawal from medications might contribute to SUDEP, by increasing a patient’s vulnerability to cardiac arrhythmia.  Seizure threshold also might diminish, leading to rebound effect and increase in seizure frequency.
Patients on multiple antiepileptic drugs had a significantly higher rate of SUDEP than patients with epilepsy who were not on multiple antiepileptic drugs. Also, a questionable role has been attributed to recent sudden changes in medication regimen of patients.
In a recent study, Tu et al undertook genetic studies in key familial long QT syndrome (LQTS) genes.  All autopsies performed from 1993-2009 at a forensic center in Sydney, Australia were reviewed, and SUDEP cases were identified and DNA was extracted from postmortem blood. The 3 most common LQTS genes, ie, KCNQ1, KCNH2 (HERG), and SCN5A were amplified and analyzed. Sixty-eight SUDEP cases were identified (mean age of 40 ±16 y). Genetic analysis revealed 6 (13%) nonsynonymous (amino acid changing) variants in KCNH2 (n=2) and SCN5A (n=4), all previously reported in LQTS patients.
In conclusion, it is possible that certain channelopathies contribute to predisposition to SUDEP in patients with epilepsy.
Frequency of SUDEP
The incidence of sudden unexpected death in epilepsy (SUDEP) in North America is believed to be comparable to that in Europe.
Statistics on the incidence of SUDEP are obtained mainly from community-based data and patient population-based data. The latter are obtained from hospitals, clinics, and referral centers and are therefore reflective of patients who are more tightly monitored but who probably also have more severe epilepsy. The median reported incidence rate of SUDEP is approximately one case per 1,000 people with epilepsy per year. [27, 28]
The incidence is estimated to be 0.09-3.5 cases of SUDEP per 1000 patient-years in the community-based studies. A higher incidence, of 2-10 cases of SUDEP per 1000 patient-years, is reported in population-based studies. [7, 29, 30, 31, 32] If applied to the US population, the estimated incidence of SUDEP indicates nearly 2,000 SUDEP deaths per year.
In a comprehensive study, Ficker reported the only large population-based study, comparing the incidence of SUDEP in patients with epilepsy to sudden unexplained death in the general population.  The incidence of SUDEP was estimated to be 0.35 cases per 1000 person-years of follow-up. The highest standardized mortality ratio (SMR) of sudden unexpected deaths was estimated to be 24 times higher in persons with epilepsy than it was in the general population.
Other reports have documented an incidence of 0.5-6 cases of SUDEP per 1000 person-years of follow-up. This translates to a range of 1 case of SUDEP in 370-1,100 in the general epileptic population.
This discrepancy reflects patient selection criteria, different study methods, and analysis methods. As an example, Neuspiel et al evaluated sudden deaths in a population of adolescents aged 14-21 years in Pennsylvania.  They reported an incidence of 5.6 per 1,000 person-years of follow-up and an SMR 40 times higher in people with SUDEP than in the general population. [9, 35, 36]
SUDEP has been shown to be associated with the risk factors listed in the Table. 
Table. Summary of Possible Risk Factors for SUDEP (Open Table in a new window)
-Young (25-35 y)
-IQ lower than 70
-Use of alcohol and recreational drugs
-Patient found in sleep or prone position
-Patient unsupervised after seizures
-Seizure type: Generalized tonic-clonic
-Younger age of seizure onset
-Duration of seizure disorder: Longer than 10 y
-Higher number of seizures
-Subtherapeutic serum level of antiepileptic medication
-Higher number of antiepileptic medications
-Treatments other than AEDs
-Frequent AED Changes
-Higher serum levels of carbamazepine
Developmental delay, defined as an intelligence quotient score (IQ) of less than 70 or a delay so severe that formal mental status examination is not possible, is significantly more common in the SUDEP group than in patients with epilepsy who do not experience SUDEP.
Excessive alcohol consumption and substance abuse are more frequently documented behaviors in patients with SUDEP than in the general population of patients with epilepsy. A higher proportion of patients have been found dead in bed. In such cases, either the patient having been in the prone position or the occurrence of sleep-related seizures may have contributed to SUDEP.
Patients who have been unsupervised after their seizures have had a higher rate of SUDEP.
Symptomatic seizures are reported in 34-70% of SUDEP cases. The annual risk of SUDEP is estimated to be 1 per 100 for patients with symptomatic seizures and 1 per 1,000 for patients with idiopathic seizures.
A recent seizure increases the chances of SUDEP.
Generalized seizures, lower age of seizure onset, duration of seizure disorder longer than 10 years, total higher number of tonic-clonic seizures  and history of therapeutic surgery for epilepsy are other seizure-related risk factors. Lhatoo et al showed a correlation with length of epilepsy history.  That study projects the odds ratio for SUDEP with duration of seizures in primary generalized epilepsy as follows:
Length more than 10 years: 11.42
Length more than 20 years: 12.99
Length more than 30 years: 7.61
Length more than 40 years: 8.1
Length more than 50 years: 8.9
Length more than 60 years: 15.18
Most patients who have died of SUDEP have had poorly controlled seizure disorder, ie, a higher-than-average number of seizures per year. Sperling et al showed that in a population of patients with refractory epilepsy who underwent therapeutic surgery, only patients who continued to have seizures were at risk for SUDEP.  No patients who had no further seizures after surgery died.
Right-sided mesial temporal seizure seems to carry a greater risk than does left-sided mesial temporal seizure.
Race, sex, and age predilections
African Americans have higher rates of SUDEP than do whites.
Male-to-female ratios for SUDEP of as high as 7:4 have been reported.
Most cases of SUDEP have been observed in patients with epilepsy who are in their third to fifth decade (ie, age 20-40 y), with a higher incidence at the younger end of the age range. The average age is estimated to be 28-35 years.
Highest risk factors
In summary, the highest association with risk factors are proven to be in patients with the following:
Nocturnal generalized tonic-clonic seizures 
Uncontrolled generalized tonic-clonic seizures
Risk of generalized tonic-clonic seizureson no antiepileptic medications
Concomitant long QT syndrome and epilepsy
Deterrence and Prevention
Proper control of seizures is the hallmark of preventing sudden unexpected death in epilepsy (SUDEP).
Many studies have shown the risk of SUDEP to be higher in untreated patients or patients with poor compliance and subtherapeutic drug levels. In a recent study, Hessdorfer et al showed monotherapy with antiepileptic drugs (AEDs) was protective. Polytherapy showed no further benefit when adjusted for seizure frequency as a risk factor. Adjuvant treatment with AEDs might have lowered the risk of SUDEP 7 times in patients with previously uncontrolled seizures when compared with placebo. 
Clear evidence supports a protective effect of surgery in patients with refractory epilepsy. Patients who were candidates for surgery but did not undergo surgery had a higher risk for SUDEP when compared with the surgical group. 
Additionally, close monitoring of the patient during and in the minutes and hours after a seizure can be life-saving. Caregivers need to be trained in the acute management of tonic-clonic seizures, including in the positioning of patients during and after the attack and in the delivery of cardiopulmonary resuscitation. Respiration needs to be monitored during the postictal period. Stimulating patients postictally is believed to reduce the chances of apnea.
Nashef et al reported that SUDEP was far more common in an outpatient setting than in a group home setting where the staff had received vigorous training in first aid treatment of tonic-clonic seizures. 
Discussing mortality is a difficult subject both for physicians and patients. However, patient education plays a significant role in preventing sudden death. Sufficient information is now available to reassure most patients, identify high-risk patients, and suggest means to reduce risk of sudden unexpected death in epilepsy (SUDEP). Early discussion is beneficial to set routines. Furthermore, it is possible that patients next follow up in months or years. Communication with sensitivity minimizes mental trauma for patients.
The issue of SUDEP needs to be discussed specifically with patients and caregivers. Morton et al surveyed neurologists in the United Kingdom to determine how frequently they discuss SUDEP with their epilepsy patients.  Only 18 (4.7%) respondents discussed SUDEP with all of their patients. It is reasonable to assume that similar statistics apply to specialists in the United States. Increasing awareness of caregivers might improve the ease with which physicians discuss SUDEP with their patients and help to prevent this outcome.
An open discussion with the patient is essential.  Each case should be considered individually. Although better seizure control can be reached when the patient is aware of the risks of SUDEP, a disclosure may casually adversely affect the quality of life.
Optimal seizure management with effective monotherapy seems to be the goal for decreasing the risk for SUDEP. Compliance with medication and avoiding periods of decreased coverage during changes in medication regimens are essential. The importance of avoiding alcohol, drugs, seizure-provoking situations, and high-risk situations (eg, driving, swimming) needs to be emphasized.
Multiple recent movements have attempted to improve the awareness of the public and of healthcare professionals of sudden unexpected death in epilepsy (SUDEP).
The National Institutes of Health (NIH)/National Institute of Neurological Disorders and Stroke (NINDS) sponsored a 3-day international workshop in 2008, the results of which were published in 2011. Participants included those with expertise in multiple disciplines. [48, 49]
The first Partners Against Mortality in Epilepsy (PAME) conference took place in 2012. The meeting focused on SUDEP as a major source of mortality resulting from seizures. The partnership that organized this conference grew out of the collaboration known as the SUDEP Coalition. The partners that served as the Organizing Committee included the American Epilepsy Society, Centers for Disease Control and Prevention, Citizens United for Epilepsy Research, Epilepsy Foundation, Epilepsy Therapy Project, NINDS, and SUDEP Aware. However, additional advocacy, education, support, and commercial organizations supported the meeting with participation and resources.
A new multisite collaborative research consortium, the Center for sudden unexpected death in epilepsy (SUDEP) Research (CSR), has received major funding from the National Institutes of Health (NIH) to examine the possible biologic mechanisms underlying this potentially preventable comorbidity and develop predictive biomarkers for interventions that could lower SUDEP incidence. 
The results of the 2008 workshop by the National Institutes of Health (NIH)/National Institute of Neurological Disorders and Stroke (NINDS) on sudden unexpected death in epilepsy (SUDEP) recommends further investigational directions.  Creating a research consortium and SUDEP registry may be the most effective approach. A multicenter study of high-risk patients could be performed (eg, those with refractory convulsions, especially in sleep), enrolling subjects and studying the following while in the epilepsy monitoring unit:
Blood/DNA for banking - Blood for DNA can be banked via blood-spot cards that are easy to store and can be kept at room temperature
Baseline ECG; cardiac monitoring, including ictal and postictal
Autonomic evaluation, including heart rate variability, baroreceptor sensitivity, and response to Valsalva
Respiratory evaluation, including oxygen saturation in the interictal, ictal, and postictal states; nasal airflow, chest, and abdominal wall movement; sighs/yawns/arousability measures (as in sudden infant death syndrome studies); possibly full polysomnography
Standardized history including family history of sudden death, in utero and postnatal smoke exposure, sleep habits/environment, and alcohol and drug use
Check C-reactive protein, postictal troponin, and postictal brain natriuretic peptide
Consider including a volunteer high-risk subgroup in which a device would be implanted to obtain long-term recordings of the oxygen level, ECG, EEG, and respiratory effort
Annual phone follow-up and questionnaire, including information about medications, illicit drugs, alcohol, compliance, sleep habits, and SUDEP awareness
In the event of near-SUDEP or SUDEP, provide readily accessible information to first responders, emergency department staff, or medical examiner office on how to contact the study center (ie, prior educational programs, medical alert identification or bracelet)
If probable or possible SUDEP occurs, perform a standardized autopsy, preferably with select tissues analyzed at a central or regional site; of note, one should not rely on formalin-fixed, paraffin-embedded tissue for genetic studies but rather blood-spot cards, blood in ethylenediaminetetraacetic acid (EDTA), or frozen tissue; in addition, detailed cardiopulmonary examination; specialized brainstem neuropathology, including serotonin evaluation; further genetic studies on tissue
Create a SUDEP registry and central tissue bank; an 2,000 SUDEP deaths occur per year in the United States and perhaps 400–500 per year in the United Kingdom; include cases with and without known seizures at the time of death, and possibly include prolonged seizures/status epilepticus if no obvious cause of death
Role of future devices
Potential devices can help reduce the risk of patients for death by warning of ongoing convulsions, identifying patient at risk, and assisting with resuscitation in patients with SUDEP. By helping understand the mechanisms for SUDEP, they can help prevent it.
To detect a seizure, obviously an EEG is the standard of care. However, simpler technologies that use other biomarkers, for example to detect abnormal motions or ECG changes, might be more available and practical.