Assessment of antiseizure medication adherence based on postmortem toxicology and claimed prescriptions in cases of sudden unexpected death in epilepsy

Abstract

Objective

Individuals with epilepsy are at a higher risk of dying suddenly and without explanation, known as sudden unexpected death in epilepsy (SUDEP). Seizures are a risk factor for SUDEP, and nonadherence to antiseizure medication (ASM) increases this risk. We aimed to assess adherence to ASMs at the time of death among young SUDEP cases by comparing prescription claims with postmortem toxicology at an individual level.

Methods

We identified all forensically autopsied SUDEP cases by assessing autopsy reports, toxicology reports, and Danish health registries from a previously identified nationwide population of sudden unexplained deaths in Denmark. We included cases aged 1–35 and 36–49 years between 2000–2019 and 2007–2019, respectively. We assessed adherence to ASMs by examining discrepancies or consistencies between any register‐based claimed prescriptions and postmortem toxicology findings, resulting in four possible outcomes: evidence of adherence, nonadherence, nontreatment or undertreatment, and nonprescribed medication.

Results

Of 477 sudden unexplained deaths, 84 (18%) were identified as SUDEP. Among the SUDEP cases, 73 (87%) claimed ASMs before death and 67 (80%) had ingested ASM according to postmortem findings. Evidence of nonadherence was found in 16 (19%) SUDEP cases, constituting 22% (n = 16/73) of SUDEP cases who claimed ASM before death. Adherence was observed in 53 SUDEP cases (63%), comprising 73% (n = 53/73) of SUDEP cases who claimed ASM before death. Nonadherent SUDEP cases tended toward a higher prevalence of psychiatric diseases when compared with the remaining SUDEP cases.

Significance

By uniquely matching register‐based prescription data with postmortem toxicology findings, we ascertained that 73% of SUDEP cases adhered to their claimed ASM, underscoring the necessity for continued vigilance in seizure management. However, improving medication adherence may possess preventive potential for SUDEP, as 22% exhibited nonadherence to their claimed ASM.

Key points.

• Sudden unexpected death in epilepsy (SUDEP) cases comprised 18% of all autopsy‐verified sudden unexplained deaths, suggesting epilepsy as a major risk factor for sudden unexplained death.

• At the time of death, 73% of SUDEP cases adhered to their claimed antiseizure medication.

• Optimizing antiseizure medication adherence may help prevent SUDEP, as 22% demonstrated nonadherence to their claimed medication.

• Nonadherent SUDEP cases tended to have a higher prevalence of psychiatric diseases than the remaining SUDEP cases.

1. INTRODUCTION

Epilepsy affects more than 45 million individuals globally. ¹ Individuals with epilepsy have an increased mortality compared to the general population, ² , ³ particularly before the age of 50, where most deaths are epilepsy‐related. ⁴ , ⁵ , ⁶ Directly, epilepsy‐related deaths include sudden unexpected death in epilepsy (SUDEP), status epilepticus, drowning, and other seizure‐related accidents, ⁷ with SUDEP being the most common among individuals younger than 50 years. ³ SUDEP is a sudden, unexpected death in individuals with epilepsy, without trauma, drowning, or status epilepticus, and where the cause of death remains unexplained after autopsy, including toxicological screening. ⁸ Thus, SUDEP is comparable with sudden unexplained death.

The pathophysiological mechanism of SUDEP is not fully established. Studies propose seizure‐induced cardiorespiratory dysfunctions, ⁹ whereas others have identified arrhythmia‐related genetic variants and prolonged ventricular repolarizations among SUDEP cases, suggesting a contribution of cardiac arrhythmias. ⁹ , ¹⁰ , ¹¹ Risk factors include the presence and frequency of generalized tonic–clonic seizures. ¹² , ¹³ Pharmacological treatment with an antiseizure medication (ASM) is typically the first choice of treatment ¹⁴ and achieves seizure freedom in two thirds of individuals with epilepsy. ¹⁵ Nonadherence, drug discontinuation, and lack of treatment with ASMs increase the risk of seizures ⁹ , ¹⁶ and SUDEP. ¹⁷ , ¹⁸

Nonadherence involves refraining or deviating from an assigned medication regimen. ¹⁹ This is not solely the patient's responsibility and can be intentional, due to adverse effects or patient expectations, or unintentional, because of forgetfulness or misunderstandings. ²⁰ Hence, clinicians should thoroughly educate patients on the importance of medication adherence. ²⁰ , ²¹

Previous studies have employed various methods to assess nonadherence in individuals with epilepsy and SUDEP cases, including next‐of‐kin information, missed prescription claims, treating physician assessments, and antemortem ASM levels, estimating nonadherence rates at 26%–49%. ⁶ , ¹⁷ , ²² , ²³ , ²⁴ However, these assessments provide information on adherence over a period, rather than at the time of death, where nonadherence may heighten seizure susceptibility and potential SUDEP. ⁹ To our knowledge, no research has assessed both ASM prescriptions and postmortem ASM detections to determine adherence at the time of death.

This study compared register‐based prescription claims with postmortem toxicological findings for individual SUDEP cases, focusing on ASMs, to determine adherence at the time of death and explore if adherence optimization holds preventive potential against SUDEP.

2. MATERIALS AND METHODS

2.1. Study design and population

We identified forensically autopsied SUDEP cases by reviewing autopsy reports, toxicology reports, and Danish health registries from existing nationwide populations of sudden deaths and sudden unexplained deaths in Denmark. Sudden deaths were previously identified through death certificates and autopsy reports. ²⁵ , ²⁶ Sudden unexplained deaths (under the term SADS) were previously identified through forensic autopsy reports and toxicology reports. ²⁵ , ²⁶ , ²⁷ The methodologies applied are briefly outlined below. Cases from 2000–2019 and 2007–2019 were included for deceased aged 1–35 and 36–49 years, respectively. The article was prepared according to Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines.

2.2. Definitions

Sudden death is defined as a sudden, natural, and unexpected death within 1 h of an acute change in cardiovascular status in witnessed cases, or when found dead for individuals last seen alive and functioning normally within 24 h before being found. ²⁵ , ²⁶

Sudden unexplained death is defined as a sudden death of unknown or inconclusive cause after forensic autopsy, including comprehensive toxicological screening with negative or non‐lethal findings. ²⁷

SUDEP is defined as a sudden unexplained death in individuals with epilepsy, meeting also the unified SUDEP definitions by Nashef et al., ²⁸ including “definite SUDEP” and “definite SUDEP plus,” defined as a nontraumatic or non–drowning‐related death in individuals with epilepsy, excluding status epilepticus and including any concomitant conditions not deemed the cause of death at autopsy. Cases of probable or possible SUDEP were not included.

2.3. Inclusion process

2.3.1. Sudden deaths

Sudden deaths were previously identified through death certificates. ²⁵ , ²⁶ The Danish Health Act requires police involvement in all sudden deaths in Denmark, who may then request a forensic external examination, performed together with a medical officer of health or forensic pathologist. Circumstantial information, police‐obtained health data, and relevant interviews are accessible at forensic external examinations. If the cause and manner of death remain uncertain, the police may request an autopsy. ²⁹

2.3.2. Sudden unexplained deaths

Sudden unexplained deaths were previously identified by examining forensic autopsy and toxicology reports of sudden deaths. ²⁵ , ²⁶ , ²⁷ Clinically autopsied cases were excluded due to the general lack of toxicological screenings. All procedures complied with ISO 17020:2012 and ISO/IEC 17025 requirements. Standard guidelines on gross cardiac examination were followed at the time of autopsy. ³⁰ Histological investigations of the heart were performed as a minimum requirement. The intracranial space was examined in all cases through standardized macroscopic examination of the brain or following fixation and neuropathological investigation.

Subdiagnostic, concomitant pathologies not deemed as cause of death, such as cardiac issues (myocardial fibrosis, hypertrophy in the absence of myocardial disarray, or coronary artery atherosclerosis ≤50%) and/or pulmonary findings (edema or pulmonary congestion), were included. ³⁰ , ³¹ , ³² Only cases with comprehensive toxicological screening for medicinal and abuse drugs at the time of autopsy were included. The drug library contained between 700 and 1500 identifiable drugs at the start of inclusion. Cases were excluded if a drug was detected at a lethal concentration, as evaluated by a forensic toxicologist and/or forensic pathologist during the time of autopsy. ³³ , ³⁴ Deaths of unknown or inconclusive cause after autopsy and comprehensive toxicological screening were classified as sudden unexplained deaths. ²⁷

2.3.3. SUDEP

We identified SUDEP cases by extracting cases with prevalent epilepsy from the sudden unexplained deaths using data from two Danish health registries.

The Danish National Patient Registry contains information on all physical hospital patient contacts in Denmark, including somatic inpatient admissions since 1977, psychiatric inpatient admissions since 1995, and outpatient visits since 1995, all of which are linked by the Danish Civil Registration System. Each contact is assigned at least one International Classification of Diseases (ICD) code.

The Register of Medicinal Products Statistics compiles data on pharmaceutical sales in Denmark. Reporting pharmaceutical sales for the register is statutory. Thus the register holds information on all prescribed drugs claimed by Danish residents. Prescription sales are linked by the Danish Civil Registration System. Each prescription sale includes information on the Anatomical Therapeutic Chemical (ATC) code and dispensing date. The register lacks details on prescribed dosages and indications.

Following the standards for epidemiological studies, described by the International League Against Epilepsy, prevalent epilepsy required at least one hospital contact with ICD codes G40.x (ICD‐10) or 345.xx (ICD‐8; except 345.2x; status epilepticus) and at least one ASM claim with ATC code N03xxxx or N05BA09. ³⁵ To exclude individuals with resolved epilepsy, we only included individuals with a hospital contact related to epilepsy or seizure within 10 years before death (G40.x, G41.x, F80.3, R25.2, R56.x [ICD‐10], 345.xx, 331.29, or 780.2x [ICD‐8]) and/or an ASM claim within 5 years before death. ³⁶

We obtained comorbidity data from the Danish National Patient Registry, comprising all hospital contacts within 10 years of death. Substance use disorder did not include acute substance intoxication. Relevant ICD codes are available in Table S1.

2.3.4. Postmortem antiseizure medication findings

This study focused exclusively on detected ASMs (N03xxxx or N05BA09) from toxicology reports, excluding other drugs such as rescue benzodiazepines. Postmortem drug concentrations were categorized into therapeutic ranges (subtherapeutic, supratherapeutic, or therapeutic levels) using validated reference tables. ³³ , ³⁴ All therapeutic ranges were included and considered adherent if they matched a corresponding prescription.

2.3.5. Claimed antiseizure medication prescriptions

Data on all ASMs claimed within 180 days of death were obtained in all SUDEP cases through the Register of Medicinal Products Statistics. For sensitivity analysis, data on claimed ASMs within 90 days before death were also gathered.

2.3.6. Adherence assessment method

Adherence was assessed by matching the ATC codes of claimed ASMs to identical ATC codes of ASMs detected postmortem in each SUDEP case. This resulted in four adherence assessments, each assigned a specific title and description, as presented in the matching matrix in Table 1.

TABLE 1.

Matching matrix presenting the four possible adherence assessments assigned to cases of sudden unexpected death in epilepsy (SUDEP).

	+ Postmortem antiseizure medication	– Postmortem antiseizure medication
+ Claimed anti seizure medication	“Evidence of adherence”	“Evidence of nonadherence”
	A claimed prescription matching with a postmortem finding	A claimed prescription not matching with a postmortem finding
– Claimed antiseizure medication	“Evidence of nonprescribed medication”	“Evidence of nontreatment or undertreatment”
	A postmortem finding not matching with a claimed prescription	No claimed prescriptions matching with no postmortem findings

Note: Each assessment is provided with a representative title.

The “evidence of adherence” and “evidence of nontreatment or undertreatment” assessments required consensus regarding the entire treatment regimen or lack of treatment, respectively, in each SUDEP case. Thus, for polytherapy cases, a SUDEP case was assessed as “evidence of adherence” only if all claimed ASMs matched the corresponding postmortem ASMs. A case was only assessed as “evidence of nontreatment or undertreatment” if no ASMs were detected and none were claimed. Therefore, cases within “evidence of adherence” and “evidence of nontreatment or undertreatment” were uniquely represented in the matching matrix. Discrepancies led to classification as “evidence of nonadherence” and/or “evidence of nonprescribed medication,” with the same case potentially appearing in both categories.

Adherence assessments were presented using the matching matrix. The presentation of identified ASMs within the matching matrix were independent of individual linkages; hence, ASMs from the same SUDEP case could be assigned to the “evidence of adherence,” “evidence of nonadherence,” and “evidence of nonprescribed medication” categories. Table S2 presents an example of adherence assessments in a SUDEP case undergoing polytherapy.

2.4. Statistical analysis

Statistical analysis was conducted using the R software (R Core Team, 2023, Vienna, Austria). Continuous variables are presented as medians with interquartile ranges, with group differences tested by the Mann–Whitney U test. Categorical variables are presented as proportions, with differences tested by the Pearson chi‐square test or Fisher's exact test, as appropriate. Statistical significance was set at p < .05.

2.5. Ethical considerations

This study was approved by the Faculty of Health and Medical Sciences, University of Copenhagen (514‐0348/19‐3000), and the Data Protection Agency in Copenhagen (P‐2019‐813; by data transfer agreement). Ethical approval is not required for retrospective register‐based studies in Denmark.

3. RESULTS

3.1. SUDEP population

Between 2000–2019 and 2007–2019, a total of 33 913 individuals aged 1–35 and 36–49 years, respectively, died, of which 4106 (12%) were sudden deaths. Among the sudden deaths, 2321 (57%) were autopsied (Figure 1). Non‐autopsied sudden deaths had a higher prevalence of prevalent epilepsy compared to autopsied sudden deaths (n = 186, 10% vs n = 172, 7.4%; p < .001). Non‐autopsied sudden deaths with epilepsy occurred less often during sleep, and exhibited a tendency toward a higher cardiovascular disease prevalence than autopsied sudden deaths with epilepsy (Table S3). Forensic autopsies constituted 1960 (84%) of the autopsied sudden deaths. The cause of death was known in 1440 (73%) of the forensically autopsied cases, and 43 (2%) cases with an unknown or inconclusive cause of death lacked (comprehensive) toxicological screening, resulting in 477 (24%) sudden unexplained deaths. SUDEP accounted for 84 (18%) of the sudden unexplained deaths (Figure 1).

FIGURE 1.

Flowchart of the inclusion process of cases of sudden unexpected death in epilepsy (SUDEP) in Denmark, aged 1–35 years and 36–49 years in the periods 2000–2019 and 2007–2019, respectively.

The majority of SUDEP cases were male (n = 53, 63%), with a median age of 36 (28–42) years at death (Table 2). They were diagnosed with epilepsy at the age of 20 (12–30) years, marked by their first epilepsy‐related hospital contact. Most deaths were unwitnessed (n = 78, 93%) and half occurred during sleep (n = 42, 50%).

TABLE 2.

Characteristics of sudden unexpected death in epilepsy (SUDEP) cases with a complete toxicological screening.

Characteristics a	SUDEP (n = 84)
Demographics and clinical characteristics
Age at death (years)	36 (28–42)
Sex, male	53 (63%)
Age at first diagnosis of epilepsy	20 (12–30)
Median duration of epilepsy	12 (5–22)
Unwitnessed death	78 (93%)
Death in sleep	42 (50%)
Comorbidities
Charlson comorbidity index
0	69 (82%)
1	11 (13%)
≥2	4 (4.8%)
Drug and/or alcohol use disorder	19 (23%)
Chronic obstructive pulmonary disease	5 (6.0%)
Cardiovascular disease	5 (6.0%)
Diabetes	5 (6.0%)
Psychiatric disease	30 (36%)
Schizophrenia, schizotypal, and delusional disorders	7 (8.3%)
Mood affective disorders	5 (6.0%)
Neurotic, stress‐related, and somatoform disorders	7 (8.3%)
Autopsy findings
Body mass index (kg/m2)	27 (23–32)
Cardiac weight (g)	364 (304–420)
Left ventricular dimension (mm)	12 (11–14)
Postmortem toxicology
Any positive postmortem toxicology	75 (89%)
Antiseizure medication detected	67 (80%)
Claimed antiseizure medication
Claimed antiseizure medication within 180 days of death	73 (87%)
Number of different antiseizure medications claimed
0	11 (13%)
1	35 (42%)
2	26 (31%)
≥3	12 (14%)

Abbreviation: SUDEP, sudden unexpected death in epilepsy.

^a Median (25%–75%); n (%).

Nineteen SUDEP cases (23%) had a history of drug and/or alcohol use disorder, and 30 cases (36%) had a psychiatric diagnosis. A few SUDEP cases had other comorbidities (Table 2). At autopsy, body mass indices were slightly elevated (mean 27 kg/m²; 23–32 kg/m²).

Postmortem toxicology revealed the presence of drugs at non‐lethal concentrations in 75 SUDEP cases (89%), with ASMs identified in 67 SUDEP cases (80%) (Table 2). A total of 114 ASMs were identified.

Seventy‐three SUDEP cases (87%) claimed at least one ASM prescription within 180 days before death, whereas 26 (31%) and 12 (14%) SUDEP cases claimed two and three or more prescriptions, respectively (Table 2). A total of 126 claimed ASMs were registered.

3.2. Adherence assessment

Evidence of adherence was observed in 53 SUDEP cases (63%) (Table 3), constituting 73% (n = 53/73) of the SUDEP cases that claimed ASM before death. Individually claimed ASMs matched the postmortem findings 107 times (85% of all prescribed ASMs).

TABLE 3.

Matching matrix of adherence assessments in cases of sudden unexpected death in epilepsy (SUDEP).

	+ Postmortem antiseizure medication 114 antiseizure medications	– Postmortem antiseizure medication
+ Claimed antiseizure medication 126 antiseizure medications	“Evidence of adherence” a	“Evidence of nonadherence”
	53 SUDEP	16 SUDEP
	107 antiseizure medications	19 antiseizure medications
– Claimed antiseizure medication	“Evidence of nonprescribed medication”	“Evidence of nontreatment or undertreatment” a
	6 SUDEP	10 SUDEP
	7 antiseizure medications

Abbreviation: SUDEP, sudden unexpected death in epilepsy.

^a SUDEP cases represented within the categories of “evidence of adherence” and “evidence of nontreatment or undertreatment” are unique.

Evidence of complete or partial nonadherence was found in 16 SUDEP cases (19%), constituting 22% (n = 16/73) of the SUDEP cases that claimed ASM before death. In this regard, four SUDEP cases claimed ASM before their final epilepsy‐ or seizure‐related hospital contact, with no subsequent claims for new ASMs. Claimed ASMs did not match postmortem ASMs in 19 instances (15% of all prescribed ASMs), with clonazepam (n = 4, 21%), lamotrigine (n = 3, 16%), and valproate (n = 3, 16%), making the highest numerical contribution to nonadherence. Among the 16 nonadherent SUDEP cases, 11 (69%) received polytherapy, compared with 27 (40%) of the remaining 68 SUDEP cases (p = .062; Table 4). There was a trend toward a higher prevalence of psychiatric diseases among nonadherent SUDEP cases compared to the remaining cases (n = 9, 56% vs n = 21, 31%, p = .057).

TABLE 4.

Characteristics of cases of sudden unexpected death in epilepsy (SUDEP), stratified by adherence assessment.

Characteristics a	Nonadherent SUDEP (n = 16)	Remaining SUDEP (n = 68)	p‐Value
Age (years)	35 (32–39)	36 (27–43)	.7
Sex, male	11 (69%)	42 (62%)	.6
Age at first diagnosis of epilepsy	28 (12–32)	19 (12–28)	.5
Median duration of epilepsy	10 (4–17)	14 (6–22)	.3
Comorbidities
Drug and/or alcohol use disorder	5 (31%)	14 (21%)	.3
Psychiatric disease	9 (56%)	21 (31%)	.057
Claimed antiseizure medication
Number of different antiseizure medications claimed
0	0 (0%)	11 (16%)	.062
1	5 (31%)	30 (44%)
2	6 (38%)	20 (29%)
3+	5 (31%)	7 (10%)

Abbreviation: SUDEP, sudden unexpected death in epilepsy.

^a Median (25%–75%); n (%).

Evidence of nonprescribed medication use was observed in six SUDEP cases. Three cases had recent epilepsy‐ and seizure‐related hospital contact before death. Postmortem ASM findings lacked a corresponding ASM claim in seven instances (6.1% of all ASMs detected).

Evidence of nontreatment or undertreatment was found in 10 SUDEP cases. For these cases, the last epilepsy‐ or seizure‐related hospital contact occurred on average 1161 (746–1577) days before death, and the last ASM claim occurred on average 1004 (768–2366) days before death.

A sensitivity analysis, with a 90‐day instead of a 180‐day interval to obtain data on ASM claims, reduced SUDEP cases claiming ASM by 10% (n = 66, 79% of all SUDEP). We observed downward vertical shifts in the matching matrix, increasing the “evidence of nonprescribed medication” and “evidence of nontreatment or undertreatment” assessments by seven and four SUDEP cases, respectively (Table S4).

3.3. Antiseizure medication concentrations and frequencies

Lamotrigine, oxcarbazepine, and valproate were the most frequently claimed prescriptions among the 126 claims of ASMs (n = 30, 24%; n = 18, 14%; and n = 18, 14%, respectively) and the most frequently detected postmortem drugs among the 114 detected ASMs (n = 28, 25%; n = 17, 15%; and n = 15, 13%, respectively; Figure 2). ASM concentrations were therapeutic in 58 instances (51%), subtherapeutic in 46 instances (40%), and supratherapeutic in 10 instances (9%).

FIGURE 2.

Bar chart illustrating the antiseizure medication prescriptions claimed before death and the antiseizure medications detected postmortem by toxicological screening. Frequencies of n < 3 are grouped as “other.” The numerical differences between claimed prescriptions and postmortem findings do not accurately reflect the nonadherent drug frequency due to the possibility of a claimed prescription not matching with a postmortem finding for each case and vice versa.

4. DISCUSSION

In this nationwide study among deceased individuals younger than age 50 with forensic autopsy upon sudden death, 18% of the cases of sudden unexplained death had epilepsy. If compared to the general Danish population, where only 0.8% younger than age 50 have a diagnosis of epilepsy, ⁴ our findings are consistent with other studies identifying epilepsy as a risk factor for sudden unexplained death. ³ , ³⁷

We compared register‐based prescription claims to postmortem findings of ASM and discovered that 73% of the SUDEP cases who claimed ASM before death demonstrated adherence, whereas 22% exhibited some degree of nonadherence.

4.1. Proportions of claimed prescriptions and postmortem toxicology

Most SUDEP cases (87%) claimed at least one ASM within 180 days of death, whereas only slightly fewer (80%) had at least one ASM detected in their body postmortem. Both proportions exceed those reported previously in the literature: Proportions of SUDEP cases with a claimed ASM have been reported to range from 67% to 86%, ²⁴ , ³⁸ , ³⁹ , ⁴⁰ with potential underestimation attributed to missing data and non–register‐based sources. Nevertheless, our study potentially overestimated the proportion owing to both the presence of an ASM prescription claim in the inclusion criteria and the extensive prescription claim interval. However, none of the aforementioned studies provided information on time frames for comparison. ²⁴ , ³⁸ , ³⁹ , ⁴⁰ Proportions of SUDEP cases with an ASM detected postmortem have been reported to range from 50% to 76%, ⁶ , ³⁸ , ³⁹ , ⁴⁰ indicating that our SUDEP population ingested more ASMs. This observation aligns with the higher proportion of claimed ASMs in our study, given a high adherence rate.

4.2. Nonadherence

Our findings indicate that 22% of the SUDEP cases who claimed an ASM before death showed evidence of nonadherence. Previous studies with access to both prescription and postmortem data indicate nonadherence rates with substantial variation. Lear‐Kaul et al. ³⁸ conducted an autopsy‐based study in Denver and Arapahoe by reviewing death certificates, autopsy reports, and scene investigations. Among 64 SUDEP cases, 84% (n = 54) had at least one ASM prescription, and only 7.4% (n = 4/54) exhibited nonadherence with their prescriptions with no corresponding ASM detected postmortem, which is much lower than our assessment of 22%. However, cases with missing prescription data were included, and comprehensive drug findings were not fully accessible, potentially underestimating the extent of nonadherence. Lathers et al. ³⁹ conducted an autopsy‐based study in Pennsylvania and identified 12 SUDEP cases. Of these cases, 67% (n = 8) had prescriptions for ASM according to scene investigations, whereas 38% (n = 3/8) demonstrated nonadherence to their prescriptions through the absence of corresponding ASM findings. This nonadherence rate exceeds our estimates, potentially due to an underestimated prescription rate derived from the utilized information source.

One fourth (n = 4) of the nonadherent SUDEP cases of this study had an epilepsy‐related hospital contact after claiming their final ASM, potentially leading to a misclassification as nonadherent if pharmacotherapies were discontinued upon hospital contact. None showed evidence of medication change with new medication claims succeeding the hospital contact.

4.3. The influence of polytherapy and comorbidity on nonadherence

Most of our SUDEP population claimed an ASM, with more than half claiming multiple prescriptions. The literature is inconsistent regarding the association between polytherapy with ASMs and the risk of SUDEP. Certain studies have reported that polytherapy reduces the risk of death and SUDEP, ¹⁷ , ¹⁸ whereas others have shown the opposite. ⁴¹ , ⁴² However, these associations disappeared when adjustments were made for the frequency of generalized tonic–clonic seizures, ¹² suggesting that polytherapy may be indicative of severe epilepsy.

In the nonadherent SUDEP cases, 69% received polytherapy compared with 40% of the remaining SUDEP cases, which suggests that nonadherence is more common among patients receiving polytherapy. We speculate whether nonadherence may serve as a mediator in the association between polytherapy and increased risk of SUDEP.

A Swedish nationwide population study linked SUDEP with psychiatric comorbidity, revealing a 2‐fold higher SUDEP incidence in individuals with psychiatric disorders compared to those without, and a 5‐fold increase among female participants. The study identified 69 SUDEP cases (including probable SUDEP), with 46% (n = 32) having psychiatric comorbidity (including acute intoxication), ⁴³ closely aligning with the 36% (n = 30) found in our study, considering the methodological differences.

Depression is recognized to cause poor medical adherence. ⁴⁴ We observed nonadherent SUDEP cases showing a higher, although statistically insignificant, prevalence of psychiatric diseases compared to other SUDEP cases. Higher mortality in individuals with epilepsy and co‐occurring psychiatric disorders may be attributed to elevated nonadherence rates.

4.4. Definition of adherence

Our study assessed adherence only at the time of death. However, the overall adherent behavior can be more accurately measured over a certain period, along with assessments of deviations in dose, timing, and/or frequency. ¹⁹ Individuals with epilepsy can experience seizures even after just one dose omission. ¹⁶ Moreover, they may become more susceptible to seizures through sleep deprivation, stress, or alcohol withdrawal, ⁹ despite adherence to optimal pharmacotherapy. We observed nearly half of all postmortem ASMs to be outside the therapeutic reference range. Subtherapeutic levels may indicate minor nonadherent behavior, wherein a patient omits a single dose prior to death, and the inclusion of subtherapeutic levels potentially overestimated the adherence rate. However, subtherapeutic levels could also result from intentionally adjusted doses to achieve seizure control in the patient, ⁴⁵ , ⁴⁶ or from postmortem changes. Although postmortem concentrations of every ASM have not been validated against premortem serum concentrations, some studies and reference works have demonstrated decreased concentrations postmortem. ³⁴ Furthermore, ASM levels outside the reference range are not necessarily associated with an increased risk of seizures. Randomized controlled trials have reported that patients with subtherapeutic ranges experienced the same frequency of seizures as those who received medication within the reference range ⁴⁸ but had higher incidence rates of SUDEP; however, with wide‐ranging low confidence intervals. ¹⁸ There remains a knowledge gap regarding individual seizure risks and the mechanisms underlying SUDEP occurrence in certain individuals after only a few seizures. ⁹

4.5. Assessment of cases with no claimed prescriptions

Only 10 SUDEP cases were classified as exhibiting “evidence of nontreatment or undertreatment,” with both their final epilepsy‐related hospital contact and ASM claim occurring at a median of 3 years before death, suggesting a less severe but unresolved epilepsy. ³⁶ However, even in cases of resolution, prior requirement for ASM has been associated with an elevated risk of SUDEP compared with sudden unexplained deaths without any antecedent epilepsy diagnosis and ASM prescriptions. ³ In addition, we acknowledge the possibility that SUDEP cases categorized under “evidence of nontreatment or undertreatment” may have had unclaimed prescriptions, thus exhibiting nonadherent behavior that was undetectable within the utilized registries of this study.

We detected only six SUDEP cases with “evidence of nonprescribed medication.” Half could be attributed to the observed, recent hospitalizations (with possible nonregistered dispensing), whereas others may stem from stockpiling, where prescription claims surpassed the 180 days, possibly underestimating adherence in some cases. The lack of dosage data prevented us from calculating a treatment duration for each claimed ASM; therefore, a fixed 180‐day claim interval was established. Sensitivity analysis using a 90‐day interval reduced adherence by seven cases, with most of these cases appearing adherent by claiming prescriptions within 90–180 days before death. Nonadherence decreased by four cases, unrelated to ASM changes within 90–180 days before death. Thus the 180‐day claim interval was deemed most valid.

4.6. Strengths and limitations

Our study included only forensically autopsied SUDEP cases, resulting in a homogeneous and definitive SUDEP population, although biased by the selection criteria for forensic autopsies. A diagnosis of epilepsy in unwitnessed sudden cardiac deaths decreases the likelihood of autopsy ⁴⁹ ⸺a trend we also observed among sudden deaths with epilepsy. Non‐autopsied sudden deaths with epilepsy tended toward a higher cardiovascular disease prevalence than autopsied sudden deaths with epilepsy, potentially mitigating but not eliminating the risk of missing potential SUDEP cases due to the absence of autopsy. A prospective study with an increased autopsy rate would be resource intensive but optimal to avoid this selection bias.

We identified SUDEP cases among sudden unexplained deaths, applying a time criterion of 1 h for witnessed deaths and 24 h for unwitnessed deaths. Although the unified SUDEP definition lacks a specific time frame, a 1 h window from a witnessed terminal event until death is recommended to reduce the probability of alternative causes of death. ²⁸ However, given that most SUDEP cases are unwitnessed, this time criterion may exclude some cases. A Danish register‐based study found that 12% (n = 5) of definite SUDEP cases were excluded when applying a 24‐h time criterion since the individual was last seen alive. ³

This register‐based study utilized highly accurate registries to identify individuals with epilepsy and their ASM claims. The National Patient Registry's positive predictive value for identifying epilepsy is 81%, ⁵⁰ which increases when ASM claims are included.

5. CONCLUSION

This study confirms epilepsy as a significant risk factor for sudden unexplained death, with SUDEP accounting for 18% of cases. The findings demonstrate that SUDEP occurs even among those adhering to ASM, with 73% of SUDEP cases with claimed medication before death exhibiting adherence. This underscores the importance of maintaining seizure control, also for adherent patients. Nonadherence was observed in 22% of the SUDEP cases with medication claims prior to death, and enhanced emphasis on medication adherence may particularly benefit individuals with epilepsy and a concurrent psychiatric condition.

AUTHOR CONTRIBUTIONS

M.K.P., C.J.H., M.K., J.T., and J.B. contributed to the study concept and design. M.K.P., C.J.H., and C.T.P. contributed to the data acquisition and analysis. M.K.P. drafted the manuscript.

CONFLICT OF INTEREST STATEMENT

J.T.H. received support from the John and Birthe Meyer Foundation, and reported relationships with Johnson & Johnson, Microport, Boston Medical Products, Cytokinetics, and Leo Pharma. C.T.P. has received grants for studies from Bayer and Novo Nordisk. The remaining authors have no conflicts of interest.

ETHICAL APPROVAL

Ethical approval is not required in Denmark for register‐based studies. 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.

Supporting information

Table S1.–S4.

Palsøe MK, Hansen CJ, Torp‐Pedersen C, Linnet K, Kløvgaard M, Tfelt‐Hansen J, et al. Assessment of antiseizure medication adherence based on postmortem toxicology and claimed prescriptions in cases of sudden unexpected death in epilepsy. Epilepsia. 2025;66:1933–1944. 10.1111/epi.18354

DATA AVAILABILITY STATEMENT

The availability of data is restricted under the license of this study, assigned to the Department of Forensic Medicine, University of Copenhagen, and is therefore not publicly available.