Eight population-based studies of mortality among people with epilepsy have been reported (Table 1). Three of these studies (Zielinski, 1974; Loiseau et al., 1999; Ding et al., 2006) have only short follow-up periods, so no conclusions about changes in mortality rates with duration of epilepsy can be drawn from these. The oldest study, a prevalent cohort from Poland (Zielinski, 1974), had a follow-up of 3 years; the Gironde study looked at the SMR in people with epilepsy 1 year after diagnosis (Loiseau et al., 1999), and in the other, a prospective study of a predominantly prevalent cohort in rural China, the median follow-up was 25 months (Ding et al., 2006).
Table 1. Population studies of mortality in people with epilepsy with standardized mortality ratios (with 95% confidence intervals)
|Reference||Country||SMR||Length of follow-up||Ages||Comments|
|Zielinski (1974)||Poland||1.8||3 years||All||Retrospective prevalent cohort|
|Hauser et al. (1980)||United States||2.3 (1.9–2.6)||Mean 13 years (0–29 years)||All||Historic incident cohort (Rochester)|
|Cockerell et al. (1994)a||United Kingdom||3.0 (2.5–3.7)b 2.5 (2.1–2.9)c||Median 6.9 years||All||Prospective incident cohort (NGPSE) 1984–1992|
|Olafsson et al. (1998)||Iceland||1.6 (1.2–2.2)||Mean 28 years||All||Historic incident cohort|
|Loiseau et al. (1999)||France||4.1 (2.5–6.2)||1 year||All||Prospective, incident cohort 1 year mortality|
|Lindsten et al. (2000)||Sweden||2.5 (1.2–3.2)||Mean 7.9 years||≥17 years||Prospective incident cohort with first seizure|
|Lhatoo et al. (2001)a||United Kingdom||2.6 (2.1–3.0)b 2.1 (1.8–2.4)c||Median 11.8 years||All||NGPSE cohort 1984–1997|
|Camfield et al. (2002)||Canada||5.3 (2.3–8.3)||Median 13.9 years for survivors||<17 years||Historic incident cohort|
|Ding et al. (2006)||China||3.9 (3.8–3.9)||Median 25 months||>2 years||Prospective incident and prevalent cohort|
A Swedish study identified 107 patients, newly diagnosed with unprovoked epileptic seizures between 1985 and 1987, and followed them until death or until 1996 (Lindsten et al., 2000). The overall SMR was significantly increased [2.5; 95% confidence interval (CI) 1.2–3.2]. In the first year the SMR was 7.3 (95% CI 4.4–12.1) and in the second year it was 3.6 (95% CI 1.6–8.1), following which the SMR decreased to a plateau level of between 1 and 2 for years 4–9 of follow-up (apart from year 4 at which the SMR rose slightly above 2 and year 6 at which it dropped below 1). In years 9–11, however, the SMR was 5.4 (95% CI 2.7–11.2).
In The National General Practice Study of Epilepsy and Epileptic Seizures (NGPSE), an incident cohort of 792 patients with definite or possible epilepsy, identified between 1984 and 1987, was followed for 14 years. The initial mortality data were reported at a median follow-up of 6.9 years, when the SMR for patients with definite epilepsy was 3.0 (95% CI 2.5–3.7) and for those with definite or possible epilepsy was 2.5 (95% CI 2.1–2.9). The SMR for those with definite or possible epilepsy combined was maximum at 1 year since diagnosis at 5.1 (95% CI 3.8–6.5), decreasing to 2.5 (95% CI 1.5–3.9) at 4 years and 1.3 (95% CI 0.7–2.0) at >5 years (Cockerell et al., 1994).
At a median follow-up of 11.8 years the overall SMR (definite and possible epilepsy) was 2.1 (95% CI 1.8–2.4). The SMR decreased to 1.3 (95% CI 0.9–1.7) at years 4–9 and was slightly higher at 1.6 (95% CI 1.1–2.2) at years 9–14 years after diagnosis. This suggestion of a trend was more marked for those with definite epilepsy (n = 564), for whom the SMR was significantly elevated at all points of follow-up (Lhatoo et al., 2001). It has also been shown in this cohort that people with epilepsy have a decreased life expectancy dependent on the severity of the epilepsy (Gaitatzis et al., 2004).
A Canadian study followed 692 children (up to 16 years of age) who developed epilepsy between 1977 and 1985, for a median of 13.9 years (range 0–22.5 years) from the time of diagnosis (Camfield et al., 2002). The SMR from 1980–1989 (corresponding to 0–12 years since diagnosis) was 5.3 (95% CI 2.3–8.3) and for 1990–1999 (5–22 years since diagnosis) was 8.8 (95% CI 4.2–13.4). Twenty years after the onset of seizures 6.1% (95% CI 3.0–9.2) of the cohort had died.
In a retrospective Icelandic population-based study of 224 children and adults, the overall SMR after 30 years was 1.6 (95% CI 1.2–2.2) (Olafsson et al., 1998). This risk was most marked in those with remote symptomatic epilepsy (SMR 2.3; 95% CI 1.4–3.5), whereas there was no significantly increased risk in any age group for those with idiopathic epilepsy at any point during follow-up (overall SMR 1.3; 95% CI 0.8–1.9). The overall risk of mortality was elevated for only the first 14 years of follow-up, returning to baseline risk thereafter. The SMR was 3.0 (95% CI 1.5–5.4) at 0–4 years follow-up, 2.1 (95% CI 1.0–3.9) at 5–9 years, 2.0 (95% CI 1.0–3.8) at 10–14 years follow-up, and between 0.7 and 1.2 thereafter.
In the Mayo Clinic Linkage study, a historical cohort of all residents in Rochester, Minnesota who had a first diagnosis of epilepsy between 1935 and 1974, the SMR for the total group after 29 years follow-up was 2.3 (95% CI 1.9–2.6) (Hauser et al., 1980). The SMR was initially high at 3.8 (95% CI 2.8–5.0) at 0–1 years follow-up, subsequently decreasing to 2.4 (95% CI 1.7–3.3) at 2–4 years and SMR 2.0 (95% CI 1.4–2.7) at 5–9 years. The SMR was not significantly elevated at 10–14 years follow-up (SMR 1.4; 95% CI 0.8–2.2), 15–19 years (SMR 1.4; 95% CI 0.7–2.5), and 20–24 years (SMR 1.8; 95% CI 0.7–3.0), but was again significantly increased at 25–29 years follow-up (SMR 3.9; 95% CI 1.8–7.6).
The SMRs by broad etiologic category for these studies are shown in Table 2. The SMR is either not increased, or is slightly increased in people with idiopathic/cryptogenic epilepsy, is between 2.2 and 3.7 for those with symptomatic epilepsy, but is much higher in people with a congenital neurologic deficit. Most studies did not have comparable data for specific causes of death.
Table 2. Population studies of mortality in people with epilepsy with standardized mortality ratios (with 95% confidence intervals) for the major etiological categories
|Reference||Idiopathic/cryptogenic SMR (95% CI)||Acute symptomatic SMR (95% CI)||Remote symptomatic SMR (95% CI)||Congenital deficit SMR (95% CI)|
|Hauser et al. (1980)a||1.8 (1.4–2.3)||2.2 (1.8–2.7)||11 (6.9–16.4)|
|Olafsson et al. (1998)b||1.3 (0.8–1.9)||N/A||2.3 (1.4–3.5)||N/A|
|Lhatoo et al. (2001)||1.3 (0.9–1.9)||3.0 (2.0–4.3)||3.7 (2.9–4.6)|| 25 (5.1–73.1)|
|Lindsten et al. (2000)||1.1 (0.5–2.4)||N/A||3.3 (2.4–4.5)||N/A|
|Camfield et al. (2002)c|| 1.51 (0.19–2.83)||N/A|
The largest hospital-based study was a Swedish study that identified 9,061 people aged 15–97 years who were admitted for inpatient care with a diagnosis of epilepsy between 1980 and 1989 (people with both incident and prevalent epilepsy) (Nilsson et al., 1997). Four thousand and one people in the cohort died. The overall SMR was 3.6 (95% CI 3.5–3.7), highest in young patients but significantly elevated in all age groups. The overall SMR was compared with that in patients who died more than 4 years after entering the cohort and thus at least 4 years after onset of epilepsy. The SMR in the group of patients who died later was 3.1 (95% CI 3.0–3.3), lower than the SMR of the whole cohort but within the same range. No comparison was made of the SMRs for the group of patients who died ≤4 years after entering the cohort and those who died >4 years after entering the cohort.
In a retrospective Scottish study of people attending epilepsy services between 1981 and 2001, those with newly diagnosed epilepsy (N = 890) had a 42% increased risk of premature mortality compared with the general population (SMR 1.42; 95% C.I 1.16– 1.72) (Mohanraj et al., 2006). People who did not respond to treatment had a greater increase in mortality (SMR 2.54; 95% CI 1.83–3.44), whereas people in remission had no increased risk (SMR 0.95; 95% CI 0.68–1.29).
In a Danish study (Henriksson et al., 1970) cited elsewhere, of 2,450 patients with prevalent epilepsy aged 15–89 attending four epilepsy outpatient clinics from 1950–1963, and excluding patients with acute symptomatic causes, the overall SMR for the cohort was 2.8, with male patients having a higher SMR (3.5) than female patients (2.0) (Hauser et al., 1980).
In a retrospective Dutch study, the mortality of all patients attending an institution (both inpatients and outpatients) between 1953 and 1967 was followed, with a mean follow-up of 28 years (range 6 months to 41 years) (Shackleton et al., 1999). The overall SMR for the cohort with incident epilepsy was 3.2 (95% C.I 2.9–3.5). The SMR was significantly elevated soon after diagnosis (SMR 16; 95% CI 11–20) in the first 2 years, decreasing thereafter to an SMR of approximately 7.0 for the next eight years. The SMR decreased further to 3.9 and subsequently plateaued at between 2.0 and 2.4 between 15 and 34 years. The SMR was no longer significantly elevated in the last years of follow-up (years 35–41: SMR 0.9 [95% CI 0.4–1.4]).
Two studies carried out in institutions for people with mental disorders in the 1930s and 1940s were identified, one from the United States and the other from Sweden (Malzberg, 1934; Alstrom, 1942). Both of these studies comprised patients with learning disability and prevalent epilepsy (Shackleton et al., 2002), and both reported high SMRs: 8.0 (United States) and 8.7 (Sweden).
In a study from the Chalfont Centre for Epilepsy in the United Kingdom, deaths in 2,099 patients with prevalent epilepsy who were admitted between 1931 and 1971 were investigated. Between 1951 and 1977, 636 people died (both at the center and after discharge), giving an SMR of 3.0 (95% CI 2.8–3.3) (White et al., 1979). In another study from the Chalfont Centre, the death certificates of all patients who died between January 1980 and December 1990 were examined; there were 113 deaths giving an SMR of 1.9 (95% CI 1.6–2.3) (Klenerman et al., 1993). The deaths occurring in people still resident in the Chalfont Centre in 5-year periods from 1896–1965 were also examined (O’Donoghue & Sander, 1997). During this period, there were 416 deaths, with an overall SMR of 2.34 (95% CI 2.12–2.56). The SMR remained relatively stable at between 1 and 3 throughout most of the period of follow-up, with the lowest reported SMR (0.91) occurring during the earliest period of observation (1896–1900) and the highest occurring between 1916 and 1920 (4.41), which may in part be explained by the Spanish Flu pandemic preferentially affecting death rates in institutions.