SEARCH

SEARCH BY CITATION

Keywords:

  • Sudden unexpected death;
  • Sudden unexpected death in epilepsy;
  • Sudden death;
  • Epilepsy

Summary

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Purpose

Patients with epilepsy are at increased risk of premature death from all causes and likely also from sudden unexplained death (SUD). Many patients with epilepsy have significant comorbidity, and it is unclear how much of the increased risk can be explained by epilepsy itself. We aimed to chart the incidence of sudden unexpected death in epilepsy (SUDEP) and estimate the risk of death from all causes and SUD conferred by epilepsy independently.

Methods

We conducted a historical cohort study using data from Danish registries and a complete manual review of all death certificates. The population studied consisted of all Danish residents in the age group 1–35 years, in the period 2000–2006 (inclusive), and the main outcome measures were risk of death and SUD.

Key Findings

We identified 33,022 subjects with epilepsy (median follow-up 3.7 years) and 3,001,952 subjects without (median follow-up 7.0 years). Among 685 deaths in the population with epilepsy, we identified 50 cases of definite and probable SUDEP corresponding to an incidence rate of 41.1 (95% confidence interval [CI] 31.6–54.9) per 100,000 person-years. Incidence rates increased with age from 17.6 (95% CI 9.5–32.8) in the age group 1–18 years to 73.8 (95% CI 52.5–103.8) for the age group 24–35 years. Having epilepsy increased the crude risk of death with a hazard ratio (HR) of 11.9 (95% CI 11.0–12.9). When adjusting for sex and comorbidities often encountered in patients with epilepsy (neurologic disease including cerebral palsy, psychiatric disease including mental retardation, and congenital disorders), as well as the Charlson comorbidity score, the HR fell to 5.4 (95% CI 4.9–6.0). The crude HR for SUD was 27.5 (95% CI 18.1–41.8) and fell to 16.3 (95% CI 9.8–26.9) when adjusted for the same covariates as above.

Significance

Epilepsy in and of itself carries a significant risk of premature death and SUD. These findings highlight the potential gains of risk factor modification for the prevention of premature death and SUDEP in patients with epilepsy.

Although epilepsy is often thought of as a relatively benign disease, it carries an increased risk of premature death from all causes (Cockerell et al., 1994; Gaitatzis et al., 2004; Neligan et al., 2011; Ding et al., 2013). Sudden unexpected death (SUD) is the occurrence of a nontraumatic sudden death with no obvious cause of death, even after autopsy. In patients with epilepsy, this phenomenon is termed sudden unexpected death in epilepsy (SUDEP), and there are often signs of an associated epileptic seizure (Nashef, 1997). Although some risk factors are established, the underlying mechanism of SUDEP remains unclear. Several theories have been suggested, including cerebral autonomic dysfunction, respiratory compromise, and primary cardiac arrhythmias (Nashef & Ryvlin, 2009; Devinsky, 2011). In addition, it seems to be accepted that patients with epilepsy have an increased incidence of SUD; this has, however, to our knowledge, been shown only once in a small study (Ficker et al., 1998). Furthermore, patients with epilepsy often have substantial comorbidity and both with regard to death from all causes and SUD, previous designs have not allowed adjustment for comorbidity in order to elucidate the independent risk of SUD associated with epilepsy.

Studies of SUDEP incidence have shown rates per 100,000 person-years ranging from 0.09 to 2.65 in community samples to 1.2 to 5.9 in tertiary care centers, to 6.0 to 9.3 in patients treated with epilepsy surgery (Devinsky, 2011). Previous studies of SUDEP in the community have studied only limited geographical areas and often examined only deaths referred to a coroner.

We have previously identified all cases of SUD and all sudden cardiac deaths (SCDs) in the young (1–35 years) in Denmark from 2000 through 2006 (Winkel et al., 2011). In this study, we aimed to chart the incidence of SUDEP in the same population. Furthermore, we wanted to estimate the risk of death from all causes and SUD conferred by epilepsy independently using Danish registry data.

This could have implications for risk factor modification for the prevention of premature death and SUDEP in patients with epilepsy.

Method

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Registries

We employed three different national Danish registries, briefly introduced below, for this study. The uses of the registries are described in the following sections, where appropriate. All registries were created for administrative purposes.

The Danish National Patient Registry contains information on all hospital, ambulatory, and emergency room discharge diagnoses (International Classification of Diseases and Related Health Problems, revision 8 [ICD-8] from 1977 to 1993 and revision 10 [ICD-10] from 1994 and onward) (Lynge et al., 2011).

The Danish Registry of Causes of Death holds information on any in-country death since 1970 (ICD-10 classified since 1994) (Helweg-Larsen, 2011).

The Danish Registry of Medicinal Product Statistics holds information on the dispensing date of the prescription, the strength of the drug, and the total quantity (e.g., number of tablets) dispensed, but not the prescribed dose of the drug (Kildemoes et al., 2011).

Definitions

We defined SUD as the sudden, natural unexpected death and with an autopsy being performed and not revealing alternative causes of death; in unwitnessed cases as a person last seen alive and functioning normally <24 h before being found dead; and in witnessed cases as an acute change in cardiovascular status with the time to death being <1 h.

SUDEP was defined as SUD in a subject with epilepsy and allowing signs of seizure in association with death but excluding documented status epilepticus (Nashef, 1997). SUDEP cases were divided into the following three subgroups on the basis of the certainty of the diagnosis: (1) definite SUDEP when an autopsy was performed that revealed no alternative cause of death, although there may be evidence of a seizure; (2) probable SUDEP when all clinical criteria were met, but no autopsy was performed; and (3) possible SUDEP if an alternative explanation existed or there was a lack of clinical criteria and/or information (Annegers, 1997).

An epilepsy population was defined according to diagnosis codes from the National Patient Registry. The diagnosis codes were ICD-10 codes G40 and G41, and ICD-8 code 345. In addition, we used medication data to determine periods of antiepileptic drug (AED) treatment (see below). A subject belonged to the epilepsy population 5 years from the date of receiving an epilepsy diagnosis (Christensen et al., 2007a,b). If a subject was treated with AED within and beyond this period, the 5-year period was extended until the end of AED treatment. This means that a subject could have multiple separate periods in the epilepsy population.

Medication data

Periods of medication with a drug were estimated in the following way: An estimated daily dose was calculated on the basis of strength, quantity, and overall time span for all dispensed prescriptions in a period for a specific drug. The first day in the medication period was defined as the day when the first prescription was dispensed. The last day in the medication period was defined as the day of the last dispensed prescription plus the number of days an individual was expected to take the medication, as estimated from the estimated daily dose.

AED was defined as belonging to Anatomical Therapeutic Chemical Classification System groups N03 and N05BA09.

Identification and characterization of SUDEP cases

We examined all deaths in the age group 1–35 years in the period 2000–2006 in Denmark using a design described in detail previously (Winkel et al., 2011). In brief, we used data from death certificates, which include a supplemental information field describing the circumstances surrounding the death including interview with eyewitnesses and relatives, previous medical conditions, an external examination of the body, and the preliminary conclusion before autopsy. Furthermore, we used complete extracts from the Danish National Patient Registry and the Danish Causes of Death Registry.

For the specific aim of identifying all SUDEP cases, we first identified the subgroup of decedents that belonged to the epilepsy population at time of death. The epilepsy subgroup was then reviewed by a single investigator (AGH) for cases of obvious non-SUDEP (trauma, suicides, murders, and so on). All remaining cases were reviewed independently by a neurologist (AS) and by a child neurologist (PU) and classified into the subgroups described above. In cases of disagreement, the two investigators reevaluated the death certificate together to reach a consensus. A flow chart of the process can be seen in Fig. 1.

image

Figure 1. Flowchart outlining the SUDEP selection process.

Download figure to PowerPoint

All cases of SUDEP were reviewed again, and data on circumstances surrounding the death were extracted.

Cases of SUDEP were characterized with regard to the following comorbidity using ICD-8 and ICD-10 diagnosis codes retrieved from the National Patient Registry: (1) Neurological disease including cerebral palsy (ICD-8: 314–31599, 322–344, 346–34990, 741–74399; ICD-10: G00-39, G45-47, G60-99); (2) Psychiatric disease including mental retardation (ICD-8: 310–314, 30909–30929; ICD-10: F00-09, F71–79, F84.0–84.4); (3) Congenital disorders (ICD-8: 740–74419, 74609–75209, 753, 75520–759; ICD-10:Q00–07, Q20–28, Q32–34, Q38–45, Q75, Q85–99), and (4) Charlson comorbidity score predicting 10-year mortality for a subject, taking into consideration a range of comorbid conditions according to the ICD-10 codes listed in Table S1 (Charlson et al., 1987).

In addition, using information from all available sources we recorded if cases had a history of substance abuse.

Association of epilepsy with risk of death and SUD

We wanted to examine the risk of death and SUD in subjects with and without epilepsy. In this regard, we defined SUD in a subject without epilepsy as being equivalent to SUDEP in a patient with epilepsy.

The entire Danish population being between 1 and 35 years of age in the period 2000–2006 was included. Deaths were retrieved from The Danish Registry of Causes of Death.

As noted earlier, we previously identified all cases of SUD (in that study denoted sudden unexplained death) in the same population and period as used in the present study (Winkel et al., 2011). In our previous study we excluded cases of SUD where a seizure was reported in association with death. In the present study we combined all cases of SUD, from the previous study, with all cases of SUDEP, from the present study, and used this as the primary endpoint.

Baseline characteristics are reported for an epilepsy population consisting of all subjects that at some time in the period belonged to the epilepsy population.

Statistics

Data were analyzed using survival analyses with age as timescale. Subjects were followed from the age of 1 year or January 1, 2000, whichever came last, until death, the age of 36 years, or until December 31, 2006, whichever came first. To describe the association between having epilepsy and the outcomes of death or SCD, we applied extended Cox regression models. In these models, epilepsy was entered as a time-dependent variable. In this regard, individuals were during follow-up allowed to alternate between a status of having epilepsy and a status of not having epilepsy. The other covariates were obtained at the time of study inclusion and were adjusted for as time independent (baseline) covariates. These covariates comprised the following: (1) sex, (2) neurologic disease including cerebral palsy, (3) psychiatric disease including mental retardation, (4) congenital disorders, and (5) Charlson comorbidity index (categorized in score = 0, score = 1, score >1).

Incidence rates of SUDEP with corresponding 95% confidence limits (CIs) were calculated with the use of Poisson regression (Cameron & Trivedi, 1998).

We considered a two-tailed p-value below 0.05 as statistically significant. Proportional hazard assumptions were checked and accepted for non–time-dependent covariates. All analyses were conducted with the use of the STATA 12.0 software package (StataCorp LP, College Station, TX, U.S.A.).

Ethical approval

The study was approved by the local Ethics Committee (KF 01 272484), The Danish Data Protection Agency (2005–41–5237), and the Danish National Board of Health (7–505–29–58/1–5).

Results

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Identification and characterization of SUDEP cases

We identified a total of 166 cases of SUD, among these there were 26 cases of definite SUDEP. In addition, we identified 24 cases of probable SUDEP and 37 cases of possible SUDEP. (See Table 1 for characteristics regarding the SUDEP cases.)

Table 1. Characteristics of SUDEP cases
 Definite and probable SUDEP n = 50Possible SUDEP n = 37p-Value
Median age, in years (IQR)23.9 (17.3–29.4)22.7 (14.5–27.5)0.57
Male, sex (%)35 (70)28 (76)0.56
Witnessed deaths (%)6 (12)2 (5)0.001
Antecedent symptoms, no (%)   
Malaise4 (8)2 (5)0.64
Fever2 (4)2 (5)0.76
Died in, no (%)   
Own/parents' home35 (71)23 (62)0.75
Institution1 (3)2 (4)
Unknown14 (28)12 (32)
Comorbidities, no (%)   
Neurologic disease14 (28)15 (41)0.22
Psychiatric disease4 (8)9 (24)0.04
Mental retardation5 (10)5 (13)0.61
Cerebral palsy3 (6)8 (22)0.03
Congenital disease7 (14)13 (35)0.02
Charlson score >07 (14)13 (27)0.13
Substance abuse5 (10)13 (35)0.004
No comorbidities27 (54)3 (8)<0.001

There was a tendency to higher prevalence of comorbidity in the possible SUDEP cases than in the definite and probable cases, although this was only significant with regard to psychiatric disease, congenital disease, and substance abuse.

When looking at the age group 1–18 years, there was only one witnessed death (11%) in the definite and probable SUDEP cases and none in the possible cases (p = 0.11). Furthermore, there were 5 (50%) without any comorbidity in the definite and probable SUDEP cases and 1 (7.7%) in the possible cases (p = 0.02).

The age distribution at time of death for the definite and probable SUDEP cases can be seen in Fig. 2. We observed a bimodal distribution with peaks at the lowest and highest age group, although the highest number of cases was seen in the older age groups.

image

Figure 2. Age distribution of definite and probable SUDEP cases.

Download figure to PowerPoint

The incidence rate for SUD in the nonepilepsy population was 0.80 (95% CI 0.68–0.95). The incidence rate for definite and probable SUDEP in the epilepsy population was 41.1 (95% CI 31.6–54.9) per 100,000 person-years and 72.4 (95% CI 58.7–89.4) for all SUDEP cases. Incidence rates increased with age from 17.6 (95% CI 9.5–32.8) in the age group 1–18 years to 73.8 (95% CI 52.5–103.8) in the age group 24–35 years with regard to definite and probable SUDEP cases. The complete data can be seen in Table 2.

Table 2. Incidence rates of SUDEP in the epilepsy population
 Definite and probable SUDEPAll SUDEPPerson-years
No. of casesIncidence rateNo. of casesIncidence rate
  1. Incidence rates are per 100,000 person-years (95% CI).

1–35 years5041.1 (31.6–54.9)8772.4 (58.7–89.4)120,096
1–<18 years1017.6 (9.5–32.8)1933.5 (21.4–52.5)57,703
18–<24 years737.5 (17.9–78.6)1369.6 (40.4–119.8)18,688
24–<35 years3373.8 (52.5–103.8)55123.0 (94.5–160.2)44,704

Focusing on subjects likely to live with adult supervision (10– <18 years of age) versus subjects without (18–35 years of age), incidence rate for definite and probable SUDEP cases 10– <18 years of age was 21.2 (95% CI 9.5–47.3) and for 18–35 years 107.3 (95% CI 84.6–136.0).

Association of epilepsy with risk of death and SUD

The epilepsy and the nonepilepsy population differed highly significantly on nearly all described characteristics (Table 3). Accordingly, the epilepsy population had much higher prevalences of comorbidities (proportion with no comorbidity: 77 vs. 95%, p < 0.001).

Table 3. Baseline characteristics of the epilepsy versus nonepilepsy population
 Epilepsy n = 33,022Nonepilepsy n = 3,001,952p-Value
Median age, in years (IQR)14.7 (6.5–26.1)16.6 (5.0–27.0)0.25
Male, sex (%)17,071 (52)1,528,529 (51)0.005
Comorbidities, no (%)   
Neurologic disease4,242 (13)27,603 (0.9)<0.001
Psychiatric disease883 (2.7)2,693 (0.1)<0.001
Mental retardation745 (2)1,450 (0.1)<0.001
Cerebral palsy1,609 (5)4,799 (0.2)<0.001
Congenital disease2,558 (8)42,736 (1)<0.001
Charlson score >03,148 (10)93,453 (3)<0.001
No comorbidities25,493 (77)2,848,627 (95)<0.001

The combined population comprised 3,034,974 subjects and median follow-up time was 7.0 years, corresponding to 17,487,019 person-years. The epilepsy population comprised 33,022 subjects and median follow-up time was 3.7 years, corresponding to 120,096 person-years. This resulted in an epilepsy prevalence of 0.7%.

In total, there were 685 deaths in the epilepsy population. Definite and probable SUDEP cases thus comprised 7% and all SUDEP cases comprised 13% of all deaths in the population.

Having epilepsy was associated with an increased risk of death with a hazard ratio (HR) of 11.9 (95% CI 11.0–13.0) in a model adjusted only for sex (Table 4). Adjustment for comorbidities attenuated this risk, and in a model adjusted for sex, Charlson score, and neurologic, psychiatric, and congenital disease the HR was 5.4 (95% CI 4.9–6.0).

Table 4. Risk of death from all causes for the epilepsy versus nonepilepsy population
Adjusted byHazard ratio (95% CI)
  1. Cox proportional hazard models showing the hazard ratio of death from all causes as a function of having epilepsy.

Age11.9 (11.0–12.9)
Age, sex11.9 (11.0–13.0)
Age, sex, and Charlson score7.9 (7.2–8.6)
Age, sex, Charlson score, and psychiatric disease7.2 (6.6–7.9)
Age, sex, Charlson score, and congenital disease6.6 (6.0–7.2)
Age, sex Charlson score, and neurologic disease6.1 (5.6–6.8)
Age, sex, Charlson score, neurologic, psychiatric, and congenital disease5.4 (4.9–6.0)

Having epilepsy was also associated with a severely increased risk of SUD with an HR of 27.6 (95% CI 18.1–41.9) in a model adjusted only for sex (Table 5). Adjustment for comorbidities again attenuated this risk, and in a model adjusted for sex, Charlson score, and neurologic, psychiatric, and congenital disease the HR was 16.3 (95% CI 9.8–26.9).

Table 5. Risk of SUD for the epilepsy versus nonepilepsy population
Adjusted byHazard ratio (95% CI)
  1. Cox proportional hazard models showing the hazard ratio of SUD as a function of having epilepsy.

Age27.5 (18.1–41.8)
Age, sex27.6 (18.1–41.9)
Age, sex, and Charlson score23.4 (15.1–36.1)
Age, sex, Charlson score, and psychiatric disease20.3 (12.8–32.2)
Age, sex, Charlson score, and congenital disease22.4 (14.4–35.0)
Age, sex, Charlson score, and neurologic disease17.4 (10.7–28.3)
Age, sex, Charlson score, neurologic, psychiatric, and congenital disease16.3 (9.8–26.9)

Discussion

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

This was a nationwide study of all deaths in the age group 1–35 years in the period 2000–2006. In this population we identified all cases of SUD in both subjects with epilepsy, where SUD is denoted SUDEP, and in subjects without epilepsy. We found 50 definite and probable SUDEP cases, corresponding to an incidence rate of 41.1 per 100,000 person-years. Having epilepsy conferred a multivariable-adjusted risk of 5.4 of death from all causes and 16.3 of SUD.

Characteristics of SUDEP cases

The possible SUDEP cases had more comorbidity than the definite and probable cases. This was as expected, as the reason for many of the possible cases to be classified as such was competing causes of death, stemming from comorbidity. Also as expected, only in a few cases were the deaths witnessed and most cases were male (Leestma et al., 1984, 1989; Donner et al., 2001).

Incidence of SUDEP

A number of previous studies have tried to estimate the incidence of SUDEP in the community. Ficker et al. (1998) reported an incidence rate of 35 per 100,000 person-years based on nine cases. In a community study by Donner et al. (2001) of SUDEP in the age group 0–18 years, 27 cases were found corresponding to an incidence rate of 20 per 100,000 person-years. Another study by Ackers et al. (2011) in the same age group found 9 cases corresponding to an incidence rate of 33 per 100,000 person-years. A study by Weber et al. also in this age group found 4 cases and an incidence rate of 43 per 100,000 person-years. Two studies performed in the 1990s using referrals to the coroner office in Chicago found 60 and 66 cases, respectively, corresponding to an incidence rate of SUDEP of around 150 per 100,000 person-years in both studies (Leestma et al., 1984, 1989). Langan et al. studied the incidence of SUDEP in the community in Ireland, again by using coroner referrals, and found 15 cases and an incidence rate of 147 per 100,000 person-years. Opeskin et al. performed a similar study in Australia and found 50 cases and an incidence rate of 128 per 100,000 person-years (Opeskin & Berkovic, 2003). Except the study by Ficker et al., all studies estimated the denominator using a prevalence assumption of around 0.5%.

Our incidence rate of SUDEP (41.1 per 100,000 person-years) is in line with what was found by Ficker et al. and all the above studies performed in children, but significantly lower than what was reported from other studies. We find it unlikely that we should have missed a significant number of SUDEP cases, as we have thoroughly reviewed every death in patients with epilepsy. However, our denominator, the number of epilepsy person-years, was based on registry diagnoses, and the epilepsy prevalence of 0.8% could have been somewhat overestimated, and our prevalence is in line with the prevalence assumption of 0.5% used by most other studies.

We looked at a young population and noted an increasing incidence rate with increasing age. This could explain the lower rate of SUDEP compared to studies looking at all ages.

In line with this we also compared the group 10–<18 years of age with the group 18–35 years of age and found that the incidence rate was much higher in the older age group. This could, in addition to physiologic changes with age, be caused by a lack of parent supervision in early adulthood.

Association of epilepsy with risk of death and SUD

It is well established that epilepsy carries an increased risk of death (Cockerell et al., 1994; Gaitatzis et al., 2004; Neligan et al., 2011; Ding et al., 2013). However, how much of this excess risk is caused by comorbidity and how much is caused by epilepsy itself has, to our knowledge, not been examined. We found that much of the excess risk of death in epilepsy is explained by comorbidity, although after adjusting for comorbidity the HR was still 5.4, indicating a clinically significant excess risk of death with epilepsy itself.

It also seems to be accepted that epilepsy confers an increased risk of sudden unexpected death (Tomson et al., 2008). However, to the best of our knowledge, this notion is based on a single, small study: the study by Ficker et al. (1998) compared the rate of SUD and SUDEP in the age group 20–40 years. To calculate the standardized mortality rate of SUD, they used data from a study by Shen et al. (1995) on the incidence of SUD. They based their analyses on 5 SUDEP cases and presumably (not reported in detail) 14 SUD cases. Using these data, they found a standardized mortality ratio of 23.7 (95% CI 7.7–55.0). Their design did not allow them to adjust for comorbidities, and as such no conclusion could be made with regard to the independent effect of epilepsy on the risk of SUD. In addition, the confidence interval reported only reflects the uncertainty with regard to the incidence rate of SUDEP and not the uncertainty with regard to the incidence rate of SUD, which was based on, presumably, 14 cases.

We based our data on 26 SUDEP cases and 140 SUD cases, and as such our study allowed for more reliable estimates.

When comparing the epilepsy population with the nonepilepsy population, we see that the epilepsy population had significantly more comorbidity as witnessed by the percentage with no comorbidities (77% vs. 95%, p < 0.001).

We saw an attenuation of the risk when adjusting for this comorbidity, especially when adjusting for the presence of neurologic disease. This means that some of the excess risk is likely to be explained through underlying comorbidity; however, after adjusting for what is likely to be the most significant comorbidities, the HR was still 16.3, indicating a strong independent effect of epilepsy on the risk of SUD and a stronger effect than on risk of death from all causes. This is also supported by the fact that 54% of the definite and probable SUDEP cases were without known comorbidity.

Strengths and weaknesses

Using data from Danish registries, which covers all resident contacts to all secondary health care services, allowed us to define an epilepsy population and, retrospectively, follow this population with complete follow-up data. Besides giving us a very large volume of data, this enabled us to do survival analyses with adjustment for comorbidities, on the entire Danish population in the respective age group and period, instead of the indirect way of comparing mortality rates used by previous studies. This approach allowed us to study the direct effect of epilepsy on the risk of SUD. Furthermore, our study was nationwide and we reviewed every death in the population for SUD in a prior study (Winkel et al., 2011). In addition, all decedents with epilepsy not suffering an obvious non-SUDEP death were reviewed by two epilepsy experts.

Our reliance on registry diagnoses did not allow us to differentiate between idiopathic and symptomatic epilepsy. Instead we adjusted for many of the conditions usually associated with symptomatic epilepsy.

The validity according to the guidelines developed by the International League Against Epilepsy (ILAE) of the epilepsy diagnosis in the Danish National Patient Registry has been examined and was found to have a positive predictive value for epilepsy of 81% (Christensen et al., 2007a,b). This is only moderate to high, and it does most likely lead to significant bias, in the form of over diagnosing, in our study. However, subjects stayed in the epilepsy population for only 5 years after receiving an epilepsy diagnosis in the National Patient Registry or for the duration of AED treatment, and this probably leads to some underdiagnosing. Furthermore, patients who were treated outside the hospital system in Denmark are not included. This might cause a too high estimate of the SUDEP incidence rate, since it would be expected that the epilepsy population consists of the more severe cases.

Some subjects experiencing transient loss of consciousness caused by cardiac arrhythmias could have been misdiagnosed as having epilepsy. This would lead to an overestimation of the risk of death, SUD in epilepsy, and of the incidence of SUDEP.

Although we tried to adjust for significant comorbidities, there is, without doubt, some degree of residual confounding. In addition, for most other covariates studied, we do not know the validity of the diagnose codes used to define these. However, any inaccuracy in these should be the same in both the epilepsy and nonepilepsy populations.

We studied the age group 1–35 years only, and thus our findings may not be applicable to older age groups.

Implications

Our findings highlight the need for increased vigilance on potential modifiable risk factors for premature death from all causes and SUDEP in epilepsy. Our study was not designed to look at such risk factors, but from other studies we know that these include seizure burden (Nilsson et al., 1999; Opeskin & Berkovic, 2003; Hitiris et al., 2007; Tomson et al., 2008) frequent night time checks, and sleeping with others in the room (Langan et al., 2005). However, currently we do not know if intervention of these risk factors will have any effect (Devinsky, 2011).

Conclusions

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

In this study, of >30,000 subjects with a diagnosis of epilepsy, we identified 50 cases of definite and probable SUDEP corresponding to an incidence rate of 41.1 per 100,000 person-years. In addition, we found that much of the excess risk of death and SUD associated with epilepsy was explained by comorbidity. However, even after adjusting for this, the HR for death was 5.4 and the HR for SUD 16.3 in subjects with epilepsy compared with subjects without. These findings highlight the potential gains of risk factor modification in epilepsy patients for the prevention of premature death and SUDEP.

Acknowledgments

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

This work was supported by The Danish National Research Foundation Centre for Cardiac Arrhythmia (DARC), The John and Birthe Meyer Foundation, The Danish Heart Foundation (grant number 07–10-R60-A1751-B743–22412 and 12-04-R91-A3790-22689), The Research Foundation at the Heart Centre, Rigshospitalet, and the Villadsen Family Foundation.

Disclosure

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

References

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Method
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
epi12328-sup-0001-TableS1.docxWord document16KTable S1. Charlson comorbidity index according to international classification of diseases, version 10 (ICD-10).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.