Introduction

Epilepsy is a common neurologic condition estimated to affect up to 70 million people globally. Its incidence ranges from 45 per 100,000 individuals annually in high-income countries to about 82/100,000/year in low and middle-income countries. [13].

People with epilepsy are at increased risk for premature mortality compared to the general population [411]. Epilepsy ranks as the fifth-highest cause of years life lost before age 75 for males and eighth highest for females [12], while symptomatic epilepsy may reduce life expectancy by up to 18 years [13]. Causes of death in epilepsy may be stratified as those attributable to epilepsy, resulting from the underlying pathology, and unrelated [14]. Pneumonia, central nervous system (CNS) and non-CNS neoplasms, and cerebrovascular disease are common causes of death [6, 15]. Deaths resulting from accidents, status epilepticus, accidents, and sudden unexpected death in epilepsy (SUDEP) are considered epilepsy-related [16].

SUDEP epitomizes the commonest epilepsy-related cause of premature death [17], and it is the most upsetting outcome of epilepsy. Little is known about strategies to reduce SUDEP risk. This review aims to characterize SUDEP, describe pathophysiology and risk factors, and delineate preventative strategy to reduce it. We also discuss gaps in existing knowledge and indicate areas for future research.

Definition of SUDEP

The definition of SUDEP is death in a seemingly healthy individual with epilepsy not resulting from accident, trauma, status epilepticus, or other identifiable causes [1821]. A definitive SUDEP diagnosis is when clinical criteria are met, and no other cause of death is identified on a full anatomical and toxicological post-mortem examination, through evidence of seizures may be present [22]. SUDEP is probable when clinical criteria are met, but an autopsy is lacking and possible when there is an alternative cause of death or a lack of clinical data [23]. SUDEP definition is widely accepted, but it has shortcomings. The definition encompasses heterogeneous cases, preventing differentiation of pathophysiology of SUDEP cases occurring in seizures from that of SUDEP cases that occur without evidence for seizures [21]. The definition excludes potentially life-threatening simultaneous pathological processes, preventing the determination of increased risk associated with epilepsy in the presence of concomitant disease [20]. The classification of many cases as “possible SUDEP” in clinical practice is due to a scarcity of information or other plausible explanations for death.

Epidemiology of SUDEP

The sudden death rate in young people with epilepsy is nearly 24 times higher than the general population [24]. The risk of SUDEP varies by population, criteria, SUDEP definition and study methods [20, 25]. Table 1 shows the incidence of SUDEP for different populations [8, 9, 24, 2653]. Unselected cohorts report the lowest incidence rates, at 0.09 to 0.35 per 1000 person-years [8, 24]. More commonly, the incidence of SUDEP is estimated and based on the assumed prevalence of epilepsy in the region after SUDEP cases are identified after a review of post-mortem records [20]. Most frequently, SUDEP cases are identified from among cohorts of people with epilepsy from medical records or databases [20]. SUDEP is the leading cause of premature death among children and adults, comprising 10–50% of all deaths in people with chronic epilepsy seen at tertiary care centers [9, 3942, 44]. A cohort study determined children had a 7% cumulative risk of dying suddenly within 40 years, accounting for 38% of all deaths, nearly all in adulthood [11]. SUDEP is less common in children than in adults. In children SUDEP, occurs mostly in those with major neurologic impairment or genetics syndromes such as Dravet’s [11, 5456]. Additionally, epilepsy occurs in 6.6% of individuals with sudden arrhythmic death syndrome [57]. Given similarities of SUDEP and sudden arrhythmic death syndrome, the true SUDEP incidence may be larger than previously described [58].

SUDEP Risk Factors

Studies have sought to identify SUDEP risk factors. Early uncontrolled studies identified male sex, young age, poor adherence with antiseizure medication (ASM) treatment, and alcohol overuse [17]. These studies are difficult to interpret given selection bias and the absence of controls [20]. Newer studies have employed control groups. These studies have yielded different results and risk factors through variation in SUDEP definition, population, risk factors examined, methodology, and study design [20, 59]. Nonetheless, identifying risk factors is vital to improving risk stratification and SUDEP prevention.

Unmodifiable Risk Factors

Unmodifiable SUDEP risk factors have been scrutinized. Male sex is associated with a greater risk [29, 31, 60, 61]. The mechanism is unclear, and other studies have shown either a similar SUDEP incidence in males and females or increased incidence in females [31, 33, 40, 62]. History of a significant neurologic insult is also associated with an increased SUDEP risk [11, 54, 55]. Younger age of onset has also been associated with SUDEP [24, 25, 31, 59, 60, 63], indicating an earlier developmental age may provide a substrate for deleterious effects of seizures. Longer duration of epilepsy is a risk [171925325962, 63], perhaps due to cumulative neurological injury resulting from repeated seizures. People with a history of convulsions are also more likely to succumb to SUDEP [25, 31]. Others, however, have reported no association between SUDEP and history of convulsions [63, 64]. Symptomatic epilepsy is also associated with SUDEP [19, 59, 60]. This may serve as a proxy for severity [19], as symptomatic seizures may promote more significant neurologic damage. Associated intellectual disability appears to predispose individuals to SUDEP [28, 32, 43, 59, 65], although often studies have lacked a clear definition of the nature and degree of intellectual disability [65]. Non-ambulatory status is also associated with SUDEP [28], potentially explained by a concomitant insult in the CNS. Knowledge of unmodifiable risk factors is essential for clinicians when counseling people with epilepsy and caregivers.

Modifiable Risk Factors

Some modifiable risk factors are directly related to the character of epilepsy. The presence of convulsive seizures may be the most critical SUDEP risk factor [59, 62] associated with a 27-fold increase in risk, while individuals with other seizure types do not have such an increase [66]. Similarly, the presence of nocturnal convulsions is associated with a 15-fold risk increase [66, 67]. A greater frequency of seizures, particularly convulsive seizures, is likely associated with SUDEP via seizure-induced pathophysiological changes [17, 25, 31, 32, 59, 62, 64, 68, 69]. Seizure frequency may increase in the months preceding SUDEP by one-quarter [62, 70], perhaps reflecting progressively more inadequate seizure control anticipating a fatal outcome [31, 32, 43, 54].

Treatment-related factors have also been implicated. Increased time since the last ASM prescription is associated with a greater SUDEP risk. [62] Poor adherence may increase SUDEP risk by 50%, particularly if seizures persist [17, 62, 71, 72]. Treatment that is not evidence-based is also associated with SUDEP due to failure to decrease or stop seizures [62]. Sudden and frequent ASM dosage changes are associated with greater SUDEP risk, [31] likely reflecting poor seizure control or increased severity. ASM polytherapy has also been associated with a greater likelihood of SUDEP [15, 17, 25, 31, 32, 59]. Polytherapy may be a surrogate for seizure frequency [73], and studies have indicated that increased SUDEP risk with polytherapy persists when controlling for seizure frequency, suggesting greater risk cannot be explained by increased seizure frequency alone [31]. A recent study determined that polytherapy is associated with reduced SUDEP risk, particularly in combinations including lamotrigine, valproic acid, and levetiracetam [72].

Studies have provided conflicting evidence regarding the role of specific ASMs. One reported SUDEP incidence was greater in Norwegian women treated with lamotrigine, perhaps resulting from seizure type, coupled with a genetic predisposition [74]. In another study, lamotrigine therapy was associated with significantly greater SUDEP risk in individuals with generalized epilepsy [59]. No increase in SUDEP risk was found when controlling for the frequency of convulsions [59]. Still, the United States Food and Drug Administration has recently placed a warning on lamotrigine for individuals with possible cardiac co-morbidity [75]. Other studies have found carbamazepine therapy or high levels of carbamazepine in people on polytherapy or frequent dose changes increased the risk of SUDEP [64, 76, 77]. Another study determined that users of levetiracetam had the lowest risk of SUDEP in monotherapy [72]. Others have not found a difference in SUDEP risk based on the used ASM [32, 78]. There may be no difference in risk associated with monotherapy, polytherapy, or particular ASMs when controlling for the number of generalized tonic–clonic seizures [79]. Instead, failure to control for confounding variables may explain the apparent differences [25, 80].

Additionally, we must also discuss modifiable risk factors unrelated to epilepsy. SUDEP is associated with substance use disorders and alcohol dependence [27, 29, 59, 61, 62, 66]. A case–control study, however, failed to reinforce this association [31], while few studies examine this factor. Further studies are required to investigate potential mechanisms. Statin use is reported to be associated with reduced SUDEP risk [72], either due to physiological effects or as a proxy of improved compliance. Additionally, living alone is related to a fivefold increased SUDEP risk [66]. Similarly, a combination of not sharing a bedroom and convulsions associates with a 67-fold higher SUDEP risk [66]. The presence of supervision by sharing the same bedroom or employing precautions such as a listening device may be protective against SUDEP by allowing cohabitants to identify changes in seizure frequency or quality and seek medical care before SUDEP [64].

Pathophysiology of SUDEP

SUDEP mechanisms remain mostly undefined, but recent studies have yielded proposed mechanisms by which nonmodifiable and modifiable factors may increase SUDEP risk. The pathophysiology of SUDEP is likely multifactorial, consisting of seizure-induced cardiorespiratory changes, intrinsic pulmonary and cardiac dysfunction, individual positioning, loss of arousal, sleep, and genetic mutations [68]. The mechanisms of death in SUDEP may overlap considerably with sudden arrhythmic death syndrome [57, 58].

Direct Seizure-Induced Alteration in Cardiorespiratory Function

The MORTEMUS study examined 16 SUDEP cases and nine near SUDEP cases in EEG monitoring units [81]. The study determined that a convulsion prompted early postictal neurologically-mediated alteration in cardiorespiratory function, leading to immediate death or a period of restored cardiorespiratory function preceding cardiac arrest secondary to terminal apnea [81]. Though rare, non-seizure related SUDEP indicates the seizures are not necessary to trigger fatal cardiorespiratory dysfunction [82]. Autonomic dysfunction may predispose people with epilepsy to SUDEP due to alterations in cardiorespiratory function in seizure and non-seizure cases [83, 84].

Additional studies examining respiratory monitoring techniques have indicated that ictal respiratory dysfunction is common and may be severe [68]. Greater than half of the people undergoing long-term video EEG monitoring had apnea during seizures, of whom almost two-thirds had decreases in oxygen saturation below 85% [85]. Another determined oxygen saturation decreased below 90% in a third of seizures [86]. Measurements of nasal airflow and respiratory effort indicate that central apnea or hypopnea occurs in half of the seizures, and mixed or obstructive apneas occur in one-tenth of seizures [86]. Seizures were also associated with severe prolonged increased ETCO2 levels, greater than 50 mmHg in a third of people [86, 87]. Clearly, respiratory disturbances resulting from seizures may result in SUDEP.

Similarly, seizures impinge on cardiac function [68]. One study showed that almost all seizures increase heart rate, including sinus tachycardia in all convulsions and in three-quarters of focal seizures [88]. Another noted tachycardia in 87% of seizures, significantly more in mesial temporal epilepsy than other temporal or extratemporal epilepsy [89]. About a fifth of seizures and over a third of cases have associated ECG abnormalities other than sinus tachycardia, most of which are benign [90]. In contrast, seizure-induced bradycardia occurs in approximately 1.4% of all seizures [89], while ictal asystole occurs in 0.27% to 0.4% of cases [91, 92]. In addition to changes in heart rate, arrhythmias and asystole occur after seizures. A study reported only 6% of seizures, experienced by 10% of cases, were associated with malignant ECG abnormalities such as ST-segment depression and T-wave inversion [88]. Another study detected the ictal prolongation of the QTc interval in 16% of seizures [90]. QTc prolongation may increase the likelihood of polymorphic ventricular tachycardia and sudden cardiac death [93]. Additionally, alterations in heart rate variability in epilepsy may contribute [68]. People with temporal epilepsy have reduced heart rate variability, particularly at night [94]. Precise mechanisms have yet to be elucidated. One proposed mechanism is the activation of pathways that project to neurons in the brainstem controlling autonomic output to the cardiovascular system by seizures [95]. At the same time, another posits seizure-induced apnea and oxygen desaturation lead to autonomic responses that reduce heart rate variability [96]. Reduced heart rate variability is associated with more significant cardiac mortality and sudden cardiac death [97]. Cardiac alterations resulting from seizures may lead to SUDEP.

It is unknown whether changes in heart rate, arrhythmias, and asystole are secondary to seizure-induced respiratory changes, given the lack of measurement of respiratory parameters in these studies [68]. Hypoxemia, hypoventilation, and apnea may promote secondary autonomic changes that affect cardiac function [68]. Seizure-induced QT changes are associated with concomitant hypoventilation and oxygen desaturation [98], indicating respiratory dysfunction may lead to abnormalities in cardiac tissue repolarization. Hypoxemia leads to bradycardia in people with apnea [99, 100]. Another reported peri-ictal bradycardia preceding asystole in an individual with concurrent oxygen saturation less than 50%, indicating bradycardia and asystole may have resulted from hypoventilation-induced hypoxemia [86]. Peri-ictal hypoxemia is associated with contralateral spread on EEG and postictal generalized EEG suppression [101, 102], reflecting the spread of seizures to brainstem respiratory centers [96]. Cardiac susceptibility to peri-ictal respiratory dysfunction may mediate the risk of SUDEP [68].

Pulmonary Impairment

SUDEP's likely pulmonary pathophysiology consists of an interplay between intrinsic pulmonary dysfunction, positioning, and alterations in breathing during sleep. Intrinsic pulmonary dysfunction occurs, demonstrated by hypercapnia and oxygen desaturation, during the postictal period, when respiratory effort may be preserved or increased [87, 101]. Seizure-induced changes in pulmonary blood flow or may result in a mismatch between ventilation and perfusion [68]. This evidence is, however, circumstantial given direct measurements of respiratory function [68]. Pulmonary edema is another possible manifestation of intrinsic pulmonary dysfunction resulting in SUDEP; however, the edema is usually mild and thought to be unable to cause death alone [81].

Ictal respiratory dysfunction may only be fatal under the appropriate conditions [68]. People who die of SUDEP are often found lying in the prone position. The mouth and nose may be occluded, increasing the risk of asphyxia in people with decreased arousal secondary to hypoventilation-induced hypercapnia [68, 81]. The prone position is infrequent among closely observed nonfatal convulsive seizures [103]. In the MORTEMUS study, there was no attempt to optimize the positioning of cases from the prone position between seizure termination and death, perhaps promoting hypoxia [81]. Suppression of brainstem arousal centers and potential loss of consciousness may result from decreased activity of serotonergic neurons in the midbrain. In contrast, impaired breathing and repositioning of the mouth and nose may result from the postictal reduction in serotonergic neurons in the medulla [104106].

Similarly, that most SUDEP occurs at night indicates circadian processes influencing breathing and arousal may be involved [81, 107]. Breathing is unstable at sleep onset due to cessation of respiratory stimuli involved in arousal [108], often leading to hypoventilation or apnea [68]. Additionally, sleep is associated with reduced ventilatory response to hypercapnia, hypercapnia, and ventilatory loading [109111]. Sleep-disordered breathing is more likely to occur in people with high SUDEP risk [112]. Similarly, sleep apnea may be more common in people with epilepsy, mainly when seizures are poorly controlled [113115]. In contrast, obstructive apnea may occur in susceptible individuals due to decreased upper airway dilator tone [116]. People with epilepsy who die during sleep are more likely to be in the prone position than those who die during wakefulness, while nocturnal seizures were associated with a 6.3 times greater likelihood of dying in the prone position than diurnal seizures [117]. Prone positioning and sleep may interact to increase risk synergistically.

Electrocerebral Shutdown

Another proposed mechanism is electrocerebral shutdown. The MORTEMUS study reported prolonged postictal generalized EEG suppression promoting electrocerebral shutdown might mediate cardiorespiratory dysfunction in SUDEP [81]. Other studies have provided inconsistent evidence regarding the role of this mechanism [118, 119]. Postictal generalized EEG suppression occurred more frequently at night, perhaps promoting central apnea and asystole [81]. Other studies have not confirmed this finding [120, 121].

Adenosine Dysfunction

Disruption of adenosine has also been implicated in SUDEP [68, 122]. Adenosine is involved in the brainstem-mediated cardiorespiratory function and anticonvulsive action via modulation of synaptic transmission and neuronal activity [120123]. This suggests that adenosine dysfunction may reduce seizure thresholds and increase the probability of cardiorespiratory function secondary and unrelated to seizures. High-risk individuals have decreased distribution of adenosine A2A receptors in the temporal lobe, indicative of glial dysfunction [122]. Greater neuronal A1A receptor distribution in high-risk cases may promote peri-ictal amygdala dysfunction in SUDEP [122]. Adenosine receptor antagonists, including caffeine, may protect against SUDEP if administered at the beginning of the seizure [124].

Genetic Susceptibility

Some people who die from SUDEP may have a genetic susceptibility. Table 2 delineates genes implicated in SUDEP. People with Dravet Syndrome, most often associated with mutations in the SCN1A gene, experience greater rates of SUDEP than the general epilepsy population, with half of all deaths attributable to SUDEP [56, 125]. This may result from postictal hypoventilation and bradycardia due to increased cardiac conduction disease susceptibility [126, 127]. People with KCNT1 mutations have mortality from SUDEP akin to Dravet syndrome, possibly resulting from pulmonary and cardiovascular abnormalities [128]. There is conflicting evidence regarding whether people with mutations in the SCN8A may have an increased SUDEP risk [129]. Mutations may directly affect the cardiorespiratory function or providing an additional risk factor for SUDEP [129]. Isodicentric chromosome 15 syndrome is also associated with SUDEP through the high frequency of associated refractory epilepsy, intellectual disability, and non-ambulatory status [130].. Mutations in genes associated with epilepsy such as SCN2A, DEPDC5, KCNA1, HCN2, and PRRT2 and genes associated with either long-QT syndrome such as KCNQ1, KCNH2, and SCN5A or other cardiac arrhythmias such as RYR2 have been implicated in SUDEP [127, 131]. Future studies are necessary to precisely characterize the scope of pathophysiological mechanisms of SUDEP.

Prevention of SUDEP

SUDEP's prevention requires delineation of its epidemiology in specific populations, accurate SUDEP classification, and further investigation of risk factors and pathophysiological mechanisms. Clinicians, coroners, epidemiologists, health officials, researchers, administrators, and people with epilepsy and caregivers must be involved in the initiatives to prevent SUDEP. Broad themes include raising awareness, addressing risk factors, and implementing hospital protocols. Specific algorithms may be instrumental in preempting SUDEP.

Building Awareness of SUDEP

Increasing awareness among the medical community, people with epilepsy, and caregivers are essential to address SUDEP. Increased knowledge of SUDEP will ensure that cases of potential SUDEP are referred to a coroner for determination [73]. Accurate reporting and classification of SUDEP may enable investigators to identify risk factors and prevention strategies [73]. Many adults with epilepsy and people who lost a loved one from SUDEP are unaware of SUDEP [132, 133]. There is a discrepancy between physicians' provided information and what individuals seek. The majority of clinicians do not routinely discuss SUDEP and are uncertain about the effect of this information on people with epilepsy or caregivers [134136]. In contrast, most people with epilepsy and caregivers want to receive information regarding SUDEP [132, 133, 137, 138]. Low information provision by clinicians may result from the perception of adverse reactions from individuals and caregivers [134]. Having epilepsy or neurophysiology training is associated with an increased likelihood of perceived negative response, but years in practice and seeing adults and children are associated with decreased negative response likelihood [134]. People with epilepsy feel it is their right to be informed regardless of increased fear [138]. Training and practice-related factors affect the provision of SUDEP information. The number of people with epilepsy seen annually and having seen a person who had SUDEP within the preceding two years independently predict discussing SUDEP nearly all of the time [134]. Besides increasing awareness, providing information about SUDEP to people with epilepsy and caregivers improve medication adherence and management of lifestyle factors that lower seizure threshold without decreasing mood or quality of life [138, 139].

Knowledge of Unmodifiable Risk Factors

Unmodifiable risk factors cannot, by definition, be altered. Awareness of the role history of neurologic insult, younger age of epilepsy onset, longer duration of epilepsy, a history of generalized tonic–clonic seizure during the night, symptomatic epilepsy, intellectual disability, and non-ambulatory status may provide numerous advantages in reducing the risk of SUDEP [11, 17, 19, 24, 25, 28, 29, 31, 32, 43, 54, 55, 5963, 65]. This knowledge will allow clinicians to counsel people with epilepsy and caregivers. In turn, increased awareness among cohabitants and caregivers may promote greater vigilance and a lower threshold for seeking medical care. Additionally, knowledge of these factors will allow clinicians to enact specific imaging and treatment algorithms. Lastly, this will promote communication and coordination among care providers.

Reducing or Stopping Seizures

Reducing the frequency of convulsions is central to efforts to prevent SUDEP [59, 62]. Comprehensive education may help improve adherence to ASMs. Managing lifestyle factors that may contribute to SUDEP, increasing knowledge regarding treatment medication interactions, improve reaction to seizures, and increasing self-efficacy [138140]. Similarly, utilization of evidence-based ASMs treatment strategies, avoidance of sudden and frequent changes in ASM dosing, and minimization of polypharmacy to the extent possible may also reduce SUDEP [15, 17, 25, 31, 32, 59, 62]. Using levetiracetam in suitable cases may be optimal [72]. If polypharmacy is necessary, ASM combinations including lamotrigine, valproic acid, and levetiracetam should be considered [72]. Epilepsy surgery, vagus nerve stimulation, or responsive neurostimulation in well-selected patients may reduce seizure frequency in patients refractory to ASM [42, 4851].

Lifestyle

Addressing lifestyle factors will reduce the risk of SUDEP. Management of co-morbidities, such as hyperlipidemia, may prevent SUDEP [72]. Similarly, addiction care for individuals with epilepsy who with substance use disorders or alcohol dependence may reduce the risk of SUDEP [27, 29, 59, 61, 62, 66]. Additionally, limiting prone positioning and encouraging supine or lateral positioning during sleep may prevent SUDEP given that positioning at seizure onset determines postictal body position [68, 81, 141, 142]. Whether counseling people with epilepsy to sleep on their backs will reduce risk is unclear due to natural turning during sleep throughout the night [143]. Nocturnal surveillance is also a protective factor against SUDEP and may be utilized in select individuals [62, 142]. Nocturnal supervision and listening devices can alert caregivers to a seizure, allowing them or medical personnel to administer rescue medication, provide cardiopulmonary resuscitation or tactile stimulation, or turn the person to the recovery position to prevent hypoxia and hypercapnia [16, 66, 140]. Seizure alert alarms and anti-suffocation pillows may be beneficial but have not been studied for SUDEP [144], while lattice pillows may reduce airway obstruction but are not validated for use in epilepsy [19, 140, 144]. Further studies are necessary to provide high-quality evidence regarding the role of nocturnal supervision [145].

Hospital Setting

Interventions within the hospital can reduce seizure duration, respiratory dysfunction, and EEG suppression [73]. Constant supervision in the epilepsy monitoring unit and associated oxygen saturation and electrocardiogram monitoring can alert clinicians to catalyze early intervention [140, 146]. Repositioning the individual from the prone position, oral suctioning, oxygen administration, prompt and regular ASM administration, and immediate resuscitation may mitigate direct and indirect factors that increase the SUDEP risk [81, 144]. Education of providers in the epilepsy monitoring unit and otherwise in care enables adequate early intervention to minimize cardiorespiratory dysfunction [73, 140].

Conclusion

SUDEP is the most common cause of epilepsy-related mortality in people with epilepsy. Unmodifiable and modifiable risk factors have been identified. Potential direct pathophysiological determinant includes altering cardiorespiratory function, pulmonary impairment, electrocerebral shutdown, adenosine dysfunction, and genetic susceptibility. Methods to prevent SUDEP centers on increasing awareness broadly, addressing unmodifiable and modifiable risk factors, and enacting hospital protocols. Additional studies are necessary to comprehensively characterize risk factors and pathophysiology to develop evidence-based algorithms to minimize SUDEP.

Table 1: Incidence of SUDEP by Population
Table 2: Genes Implicated in SUDEP