Cause-Specific Mortality in Epilepsy


  • Samden D. Lhatoo,

    1. Department of Neurology, Institute of Clinical Neurosciences, Frenchay Hospital, Bristol BS16 I LE, United Kingdom
    2. Department of Clinical and Experimental Epilepsy, Institute of Neurology, University College London, Queen Square, London WC 1N 3BG, United Kingdom
    Search for more papers by this author
  • Josemir W. A. S. Sander

    1. Department of Clinical and Experimental Epilepsy, Institute of Neurology, University College London, Queen Square, London WC 1N 3BG, United Kingdom
    Search for more papers by this author

  • The commission wishes to acknowledge the support of UCB Pharma for logistical support in the organization of this workshop.

Address correspondence and reprint requests to S. D. Lhatoo at Department of Neurology, Institute of Clinical Neurosciences, Frenchay Hospital, Bristol BS16 1LE, U.K. E-mail:


Summary:  Epilepsy is associated with a two- to three-fold increase in mortality. Studies of cause-specific mortality show that deaths may be classified into those that are directly or indirectly related to epilepsy, those that are related to the underlying pathology giving rise to epilepsy, and those that are unrelated to both epilepsy and its causes. Overall, direct epilepsy related deaths are infrequent. Pneumonia, especially in the elderly, central nervous system (CNS) and non-CNS neoplasias, and cerebrovascular disease are frequent causes of death. Suicides, accidental deaths, and ischemic heart disease do not appear to be significant contributors to mortality in community-based studies. In hospital/institution-based analyses, epilepsy-related deaths are common and sudden unexpected death in epilepsy (SUDEP) may account for up to 17% of all deaths in epilepsy. A small proportion of these deaths may be witnessed and most such witnessed deaths occur in relation to convulsive seizures. The exact pathogenetic mechanisms are unknown although it is very probable that lack of seizure control is an important risk factor. Patients who continue to suffer seizures appear to have an almost 40 times higher risk of mortality than those in remission.

It is reliably established that mortality in epilepsy is increased as much as threefold compared to an age-matched general population (1–4). Prospective, community-based analyses of mortality in newly diagnosed epilepsy indicate that the underlying pathology responsible for seizures is often the cause of death (1). Seizures as an agonal manifestation of serious background illnesses thus constitute, quite literally, a large fraction of early mortality in epilepsy. In addition, common causes of death in epilepsy include unrelated conditions such as respiratory infections and malignancies outside the nervous system. Seizure-related deaths on the other hand, appear to be uncommon in the general population although these account for between 24–62% of mortality figures in hospital-based populations of persons with epilepsy (5). Community-based studies also suggest that cause of death varies according to the duration of the seizure disorder, and deaths caused by seizures themselves, probably, largely account for late increases in mortality (1,2). Thus, the presence of continued active seizures may be important (reference) determinants of mortality in the long term. Cause-specific mortality will, therefore, differ according to patient populations studied as well as the duration and control of epilepsy.


Studies of mortality in people with epilepsy have been both hospital-based (6–11) with attendant biases for chronically and severely afflicted patients with epilepsy and community-based, therefore capturing the full spectrum of epilepsy. This review focuses upon community-based studies but also includes what we consider to be the better hospital-based studies.

The standardized mortality ratio (SMR) for all-cause mortality in both hospital- and community-based studies ranges from 1.6 to 3.6 (Table 1). While the range of SMRs reported in both types of studies is similar, it does not imply that identical causes of death contribute toward these SMRs in both groups.

Table 1. Studies of mortality in epilepsy
 CountrySMRa(95% CI)
  1. aStandardized mortality ratio = observed/expected deaths.

Community-based studies
 Hauser et al. (1980)U.S.A.2.3(1.9–2.6)
 Cockerell et al. (1994)U.K.3.0(2.5–3.7)
 Olafsson et al. (1998)Iceland1.6(1.2–2.2)
Hospital/institution-based studies
 Klenerman et al. (1993)U.K.1.9(1.6–2.3)
 Nilsson et al. (1997)Sweden3.6(3.5–3.7)
 Shackleton et al. (1999)The Netherlands3.2(2.9–3.5)

The cause of epilepsy is related to mortality. When observed deaths are compared to those expected, symptomatic epilepsies carry a higher risk of death than the idiopathic or cryptogenic epilepsies. In community-based studies of death in idiopathic epilespy (defined as epilepsy with no apparent cause and distinct from the idiopathic generalized epilepsies) the SMRs indicate only a 30–80% increased mortality: 1.8 (95% CI: 1.4–2.3) in Rochester, Minnesota (2), 1.6 (95% CI: 1.0–2.4) in the United Kingdom (1), and 1.3 (95% CI: 0.8–1.9) in Iceland (3). In the symptomatic epilepsies, however, the SMRs are considerably and significantly higher (2.2 (1.8–2.7), 4.3 (3.3, 5.5), and 2.3 (1.4–3.5) in the Rochester (2), United Kingdom (1) and Iceland (3) studies, respectively).

Causes of death vary when population-based studies are compared to hospital-based studies. The commonest causes of death in community-based studies are usually pneumonia, cerebrovascular disease, and central nervous system (CNS)- as well as non-CNS-related neoplasia (Table 2); sudden unexpected death in epilepsy (SUDEP) is very infrequent. SUDEP is more common in hospital-based studies.

Table 2. Causes and categories of death in epilepsy
  1. aSudden unexpected death in epilepsy.

Epilepsy-related deaths
 Directly related
  Status epilepticus
 Indirectly related
  Accidents as a consequence of seizures
   Aspiration pneumonia after seizures
  Iatrogenic; drug toxicity and idiosyncratic reactions
Underlying disease related deaths
 Primary and secondary CNS neoplasia
 Cerebrovascular disease
 CNS infections
 Inherited neurodegenerative disorders
Unrelated deaths
 Non-CNS neoplasia
 Ischemic heart disease
 Accidents unrelated to seizures



SUDEP is defined as a sudden unexpected death in an individual with epilepsy with or without evidence of a seizure, and excluding documented status epilepticus, where postmortem examination does not reveal an anatomical or toxicological cause of death. Estimates of the frequency of SUDEP are heavily influenced by the study population and a more likely estimate is probably <1% (12). In community-based studies, the incidence of SUDEP ranges from 0.35 per 1,000 person-years in the Rochester database (13) to 2.7 per 1,000 person-years in a prospective, medical examiner-based study (Table 3). The National GP Study of Epilepsy (NGPSE) in the United Kingdom has confirmed one SUDEP death in a community-based, prospective cohort of 564 patients with definite epilepsy followed up for approximately 8,000 person-years (14). In the MRC Antiepileptic Drug Withdrawal Study, there are two deaths attributable to SUDEP after 5,000 person-years of follow-up (15).

Table 3. Some estimates of SUDEP incidence/1,000 person-years
Community-based studies
 Leestma et al. (1989)0.9–2.7Prospective study, medical examiner-based
 Tennis et al. (1995)0.5–1.4Retrospective study of death certificates
 Ficker et a1. (1998) 0.35Retrospective study of Rochester database and autopsy reports
 Dashieff (1991)9.3Prospective study, referrals for surgery
 Nashef et a1. (1995)3.4Retrospective, learning disability population
 Nashef et a1. (1995)5.9 
 Nilsson et al. (1999)1.5Retrospective, hospital admissions

In patients with refractory epilepsy, the incidence of SUDEP is high and studies of patients in epilepsy surgery programs have indicated figures as high as 9.3 per 1,000 person years (16). Estimates from other studies of hospital-based populations lie between 1.5–5.9 per 1,000 person-years (Table 3). In contrast, the incidence of SUDEP from population-based studies is far lower, consistent with the larger remission rates expected in such populations with epilepsy (17).

Suspected risk factors for SUDEP include age (20–40 years), male gender, refractory seizures, remote symptomatic epilepsy, treatment noncompliance, recent head injury, and alcohol, although most of these come from studies of selected populations without adequate controls (22). A case–control study (3 living controls for each case matched from the same population with epilepsy), suggested that relative risk for SUDEP was increased to 10.16 (95% CI: 2.94–35.18) in patients with more than 50 seizures per year compared to patients with two or less seizures per year (12). Estimated relative risk of SUDEP was 27.3 (95% CI: 3.7–203) when comparing patients in remission to those not in remission, although CIs are extremely wide. Polytherapy with antiepileptic drugs, frequent dose changes, early-onset epilepsy, and the use of psycho-pharmacological drugs were also significant risk factors. The relationship between antiepileptic drug use and SUDEP is interesting. A prospective case–control study is currently under way (23).

SUDEP is frequently unwitnessed and this is a consistent theme in many studies. Amongst those that are witnessed, however, the majority are associated with convulsions thus suggesting that seizures are directly and causally related to SUDEP (8,20,21,23–25).

Accident-related deaths

About 1–16% of deaths in epilepsy patients may be a consequence of accidents (20). These usually occur while swimming or bathing. In one Canadian study, 5% of all drownings were attributable to seizures, 60% of which occurred in bathtubs (28). The risk of injury among patients with epilepsy is estimated at 29.5 per 100,000 per year with a risk of death among the injured of 2.68 per 100,000 per year (29). These figures are based on visits to accident departments and may be an underestimation of the risk of injury in people with epilepsy. In the NGPSE there have so far been three accidental deaths (one bathtub drowning, one fall resulting in a cervical fracture, and one severe burns). One study reports that female sex, high seizure frequency, and interictal neurological impairment are risk factors for burns in people with seizures (30).


The proportion of mortality in epilepsy that is due to suicide is high in temporal lobe epilepsy (31). High suicide rates have also been found in a recent hospital-based study (9) and one community-based study (4). More recent studies suggest that suicide is probably not a major cause of death in epilepsy (1,2).


The high mortality of epilepsy due to primary and secondary cancers in the CNS is explained by the fact that these are causes of remote symptomatic seizures. The high SMRs for non-CNS neoplasia (SMRs range from 1.5 to 3.4), particularly lung cancer (SMRs range from 1.9 to 3.0) (1,2,9), is intriguing. A carcinogenic effect of antiepileptic drugs has been postulated although this has not really been substantiated and indeed, most neoplasia-related deaths occur soon after diagnosis, rendering this association unlikely (1,32). Some studies show a trend toward an increased incidence of lymphatic malignancies (8,32,33).


Pneumonia is a frequent cause of death in epilepsy (SMRs range from 3.5 to 7.2) in community-based studies (1,2) as well as hospital-based studies (4,8,9,27). This is a common terminal event in the elderly, and the mean age of patients who died of pneumonia in the NGPSE was 81.3 years (1).

Vascular disease

The significantly increased mortality due to cerebrovascular disease in epilepsy in community-based studies (SMRs ranging from 2.5 to 3.7 (1,2)) as well as hospital-based studies (SMRs ranging from 2.5 to 5.3 (9,10)) may be partly explained by cerebrovascular disease as a risk factor for remote symptomatic epilepsy.

Ischemic heart disease, however, does not appear to be a significant cause of mortality in epilepsy and in both the Rochester study (2) and the NGPSE (1). Other studies have found elevated SMRs, especially in patients under the age of 65 years (9,34).


Mortality in epilepsy can be due to epilepsy itself. Community-based studies show that mortality is more often due to underlying pathologies, especially in the early stages after diagnosis. The phenomenon of SUDEP is currently the subject of much study and it is becoming increasingly clear that seizure frequency and lack of seizure remission are important risk factors in mortality.