Seizure clusters, treatment patterns, and healthcare resource utilization in patients with epilepsy: A Wisconsin-based claims analysis

Background: Seizure clusters are underresearched and associated with adverse outcomes in patients with epilepsy. This study was a noninterventional, retrospective claims-based analysis using the Wisconsin Health Information Organization (WHIO) All-Payer Claims Database to characterize the epilepsy population in Wisconsin, with a focus on prevalence, treatment patterns, and healthcare resource utilization (HCRU) in patients with seizure clusters prior to the introduction of nasal spray rescue medications. This timeframe allows characterization of a historical baseline for future comparisons with newer treatments. Methods: Four cohorts were defined: (1) all-epilepsy (all patients with epilepsy); and subcohorts of: (2) patients receiving a monotherapy antiseizure medication (ASM); (3) patients receiving ASM polytherapy; and (4) patients treated for seizure clusters (ie, those taking rescue medications and ≥ 1 ASM). Primary outcomes were HCRU over a 12-month follow-up period, which were descriptively analyzed. Results: Between 2017 and 2019, 16,384 patients were included in the all-epilepsy cohort; 11,688 (71.3 %) were on monotherapy, 3,849 (23.5 %

Background: Seizure clusters are underresearched and associated with adverse outcomes in patients with epilepsy.This study was a noninterventional, retrospective claims-based analysis using the Wisconsin Health Information Organization (WHIO) All-Payer Claims Database to characterize the epilepsy population in Wisconsin, with a focus on prevalence, treatment patterns, and healthcare resource utilization (HCRU) in patients with seizure clusters prior to the introduction of nasal spray rescue medications.This timeframe allows characterization of a historical baseline for future comparisons with newer treatments.Methods: Four cohorts were defined: (1) all-epilepsy (all patients with epilepsy); and subcohorts of: (2) patients receiving a monotherapy antiseizure medication (ASM); (3) patients receiving ASM polytherapy; and (4) patients treated for seizure clusters (ie, those taking rescue medications and ≥ 1 ASM).Primary outcomes were HCRU over a 12-month follow-up period, which were descriptively analyzed.Results: Between 2017 and 2019, 16,384 patients were included in the all-epilepsy cohort; 11,688 (71.3 %) were on monotherapy, 3,849 (23.5 %) were on polytherapy, and 526 (3.2 %) were treated for seizure clusters.Twelvemonth retentions to the ASM treatments were 46.7 % (7,895/16,904) in the all-epilepsy cohort, and 40.0 % (4,679/11,688) and 40.1 % (1,544/3,849) in the monotherapy and polytherapy subcohorts, respectively.Rescue medication prescriptions were obtained 1,029 times by the 526 patients in the treated seizure cluster subcohort, with infrequent refill rates (mean 1.6-1.9times/year).A higher proportion of patients in the treated seizure cluster subcohort had epilepsy-related outpatient visits (89.7 %), other visits (71.3 %), and hospitalizations (25.3 %) than patients in the monotherapy (72.2 %, 50.2 %, 19.3 %, respectively) and polytherapy (83.3 %, 63.3 %, 22.8 %, respectively) subcohorts.Mean (standard deviation) all-cause ($114,717 [$231,667]) and epilepsyrelated ($76,134 [$204,930]) costs over 12 months were higher in the treated seizure cluster subcohort than the monotherapy ($89,324 [$220,181] and $30,745 [$145,977], respectively) and polytherapy ($101,506 [$152,931] and $49,383 [$96,285], respectively) subcohorts.Conclusions: Patients treated for seizure clusters incurred higher all-cause and epilepsy-related costs and epilepsyrelated HCRU than other subcohorts and had infrequent rescue medication refills.The findings of this analysis highlight the need for appropriate treatment for those patients with epilepsy experiencing seizure clusters.The effect of newer rescue medications to alter these findings will be explored in a follow-up study.Regardless, specialist providers with expertise in treating refractory epilepsy and seizure cluster patients may help to reduce the burden of seizure clusters.
Abbreviations: ASM, antiseizure medication; CDC, Centers for Disease Control and Prevention; ED, emergency department; HCRU, healthcare resource use; HIPAA, Health Insurance Portability and Accountability Act; ICD-10, International Statistical Classification of Diseases and Related Health Problems 10th Revision; OP, outpatient; SD, standard deviation; US, United States; WHIO, Wisconsin Health Information Organization.

Introduction
In the United States (US), an estimated 3.4 million people or 1.2 % of the population have active epilepsy, a neurologic disorder characterized by recurring seizures [1].Optimal use of antiseizure medications (ASMs) can lead to higher rates of seizure remission; however, seizure control remains incomplete in approximately one-third of patients with epilepsy [2].
Patients with epilepsy and continued seizures may experience acute repetitive seizures or seizure clusters (SCs) [3], which may differ from their usual seizure pattern [4].The prevalence of acute repetitive seizures varies considerably across studies in the literature, primarily due to lack of consensus on the definition of seizure clusters [5].Seizure clusters are broadly described as multiple seizures during a short time interval with short interictal periods [5,6].Specific clinical definitions include: ≥ 2 seizures in 6 h [7]; ≥ 3 seizures in 24 h [8][9][10][11]; ≥ 2 seizures in 24 h [3]; and ≥ 2 seizures in 48 h [12].In addition, there may be differences in seizure cluster rates according to study type and population: database analyses and chart reviews often report lower rates (eg, 3 % [10], 14.9 % [10]) than prospective studies capturing patients at hospitals or epilepsy centers (29 % [9], 32.3 % [13]); studies conducted at tertiary epilepsy centers may overestimate the prevalence of seizure clusters because refractory epilepsy is more common in this setting [5,14].
Seizure clusters are associated with progression to status epilepticus [15], higher mortality rates [11], reduced quality of life [3], and hospital admissions [9].There has been limited research on seizure clusters and their outcomes despite the association with these adverse consequences [16].Clinical trials investigating seizure clusters have addressed important aspects of rescue treatment, but practical realworld clinical issues remain unresolved [16].For example, the underutilization of rescue medications was observed in a large real-world survey study that characterized patient and caregiver burden of seizure clusters [3].The study found that only 20 % of patients choose rescue medication as their first option for seizure cluster management, whereas 24 % choose emergency department (ED) services [3].
This analysis aimed to characterize the epilepsy population in Wisconsin, including prevalence of epilepsy, treatment patterns, and healthcare utilization, from 2017 through 2019, with a particular focus on patients with seizure clusters.By using the 2017 to 2019 time period that preceded the introduction of nasal spray rescue medications, the study provides a historical baseline, which can be used to assess the impact of newer medications for seizure clusters.

Data sources
This was a noninterventional, retrospective claims-based analysis using the Wisconsin Health Information Organization (WHIO) All-Payer Claims Database, which includes all care settings, all services, and all professionals that are paid for by insurance companies.WHIO data are provided voluntarily by health insurance companies, self-funded employers, and Medicaid in Wisconsin, US.The WHIO database is Health Insurance Portability and Accountability Act (HIPAA) compliant and all patient data were anonymized.This database does not capture the Medicare fee-for-service population, which represents 56 % of the Medicare-insured population in Wisconsin [17].

Prevalence
Annual epilepsy prevalence was assessed from 2017 through 2019, with separate cohorts for each calendar year.To be eligible for inclusion in the annual cohorts, patients must have had 365 days of continuous enrollment in the specified calendar year, ≥ 1 epilepsy medical claim (International Statistical Classification of Diseases and Related Health Problems 10th Revision [ICD-10]) by any physician type (index date) in the specified calendar year, and ≥ 1 prescription claim for chronic ASM in the specified calendar year.A patient with a diagnosis of epilepsy in year was counted in all subsequent years ("carried forward"), irrespective of whether they had a claim for that same diagnosis, as long as the patient remained enrolled and alive during the entire calendar year.The patient count by calendar year in the WHIO dataset (ie, patients enrolled and alive for the entire calendar year and having a charge during the calendar year,) served as the "at-risk" population for each calendar year and was the denominator in prevalence calculations.

ASM cohorts and outcomes
Four cohorts were defined: (1) an all-epilepsy cohort; and subcohorts of patients with: (2) monotherapy ASM treatment; (3) polytherapy ASM treatment; and (4) treated seizure clusters.The identification period was from January 1, 2017, through December 31, 2018; the follow-up period was 12 months starting on the index date (day 0 to day 365).Accordingly, the study period was from January 1, 2017, through December 31, 2019.This time period allowed assessment of treatment patterns prior to the introduction of nasal spray options for rescue medication, thereby providing a baseline for future research.
The all-epilepsy cohort included patients aged ≥ 12 years, with a diagnosis of epilepsy (ICD-10: G40.*, R56.9, G25.3), and ≥ 1 prescription claim for a chronic ASM (see list in the Supplementary Methods) during the identification period.Patients who did not have continuous medical and pharmacy benefit coverage for ≥ 365 days after the index date were excluded.The monotherapy ASM treatment subcohort was defined as having ≥ 1 prescription claim for one type of chronic ASM during the identification period, or 1 prescription claim for one type of chronic ASM during the identification period plus 1 prescription claim for another type of chronic ASM during the identification period with a supply overlap of ≤ 60 days.Patients with a prescription claim for rescue medication during the identification period were excluded.Rescue medication was defined as a branded or generic form of rescue medication indicated for the treatment of seizure clusters (ie, branded or generic diazepam rectal gel, branded intranasal diazepam spray, and branded intranasal midazolam spray; see Supplementary Methods).It should be noted that branded intranasal midazolam spray was not commercially available until December 2019.The polytherapy ASM treatment subcohort was defined as having prescription claims for ≥ different types of chronic ASMs during the identification period with a supply overlap of > 60 days.Patients who had a prescription claim for rescue medication during the identification period were excluded.The treated seizure cluster subcohort was defined as having a prescription claim for ≥ 1 chronic ASM during the identification period and ≥ prescription claim for rescue medication during the identification period.The monotherapy, polytherapy, and seizure clusters subcohorts were mutually exclusive.
The index date (Day 0) was the first date when the inclusion criteria were met.For the all-epilepsy cohort and monotherapy subcohort, this was either the first claim for an ASM after a diagnostic claim for epilepsy or the first diagnostic epilepsy claim after a claim for an ASM during the identification period.The index date for the polytherapy subcohort was the first date of polytherapy initiation.The index date for the seizure cluster subcohort was the date of the first prescription claim for rescue medication.
The primary outcomes were healthcare resource utilization (HCRUmedical and pharmacy) and healthcare costs (all-cause and epilepsyspecific) over the 12-month follow-up period.HCRU included hospital admissions (overnight), ED visits, and outpatient (OP) visits."Other visit" was defined as a medical claim that could not be clearly classified as hospitalization (overnight), ED, or OP.Healthcare costs included medical, pharmacy, and diagnostic claims.All-cause costs were defined as the sum of all allowed (patient-paid and payer-paid) pharmacy and medical costs over the 12-month post-index period.Epilepsy-related costs were defined as the sum of pharmacy costs associated with a claim for an ASM and medical costs associated with a primary diagnosis code for epilepsy.Costs were adjusted to the 2021 Consumer Price Index.
Rescue medication frequency distribution for the index ASM, and ASM retention measure for the index ASM were assessed over a 12month follow-up.ASM retention was measured using pharmacy claims for the index claim for an ASM until death, disenrollment, or the study period end date.The number of days of treatment for each patient was calculated.Discontinuation of index ASM was defined as a gap in treatment of > 60 days where the patient had no supply.The discontinuation date was defined as the date of the last prescription claim prior to the refill gap being exceeded, plus the days of supply for that claim.
Definitions and measurements for exposure variables are provided in Supplementary Material Table S1.

Statistical analyses
As this study is descriptive, no formal statistical testing or comparisons were performed.No imputation was undertaken for missing data.Categorical variables are presented as counts and frequencies ( %).Continuous variables are presented as mean and standard deviation (SD) or median and interquartile range.

Population and prevalence of epilepsy
The overall WHIO population currently includes 4.5 million people of all ages [18].From 2017 through 2019, 1.6 to 2.0 million people had a charge entered onto the WHIO database (Table 1).Thus, using the current population as the denominator, between 36 % and 46 % of people in the WHIO database had a charge between 2017 and 2019.
As shown by the all-epilepsy cohort, the prevalence of epilepsy decreased slightly between 2017 and 2019 (Table 1).Prevalence rates for two of the three subcohorts also decreased, while a small increase in the prevalence of ASM monotherapy was observed.The number of people with a charge captured by WHIO increased during the study period.

Study population
A total of 16,384 patients met the criteria for the all-epilepsy cohort.Of these, 71.3 % were on monotherapy, 23.5 % were on polytherapy, and 3.2 % were treated for seizure clusters (Table 2).
The majority of patients in the all-epilepsy cohort and the monotherapy and polytherapy subcohorts fell into the 18-64 years age range (84.3 %, 84.0 %, and 89.3 %, respectively).However, the majority of patients in the treated seizure cluster subcohort were in the 12-39 years age range (95.4 %).The mean index age of this subcohort was approximately half that of the other subcohorts (21.6 versus 40.3-41.6years).
Most patients had Medicaid insurance coverage (75.3 % of the allepilepsy cohort, 73.3 %, 77.9 %, and 91.8 % of the monotherapy, polytherapy, and treated seizure cluster subcohorts, respectively).Commercial insurance was the second-largest insurance provider in the allepilepsy cohort, accounting for 17.4 % of patients.

ASM prescriptions and retention
In the all-epilepsy cohort, levetiracetam was the most filled ASM over the 12-month follow-up period, with 28.4 % (53,338/187,530) of the prescriptions, followed by lamotrigine ( ).ASM refill rates during 12month follow-up period were generally higher in the treated seizure cluster and polytherapy subcohorts than the monotherapy subcohort.For example, the mean number of levetiracetam prescriptions was 10.1 (SD 5.5) in the treated seizure cluster subcohort, 9.6 (5.2) in the polytherapy subcohort, and 7.2 (5.3) in the monotherapy subcohort.A wide range of ASMs were used in the monotherapy subcohort; the top three filled ASMs over 12-month follow-up were carbamazepine (mean [SD] 8.5 [5.8]), lamotrigine (7.7 [5.9]), and valproic acid (7.5 [5.6]).

Rescue medication prescriptions
Over 12-month follow-up, there were 940 rescue medication prescriptions in the treated seizure cluster subcohort (n = 526) for Diastat (branded diazepam gel) and 89 for generic diazepam gel.No other medications approved for seizure clusters were identified.Refill rates during the 12-month follow-up period were low for both branded diazepam gel (mean [SD] and median [interquartile range] 1.9 [2.05] and 1.0 [1.0]) and generic diazepam gel (1.6 [1.3] and 1.0 [1.0]).

HCRU and costs
Over 12-month follow-up, the percentages of patients with ≥ 1 allcause OP visit, all-cause other visit, and all-cause hospitalization were similar in the all-epilepsy cohort and all three subcohorts (Table 3).The percentage of patients with ≥ 1 all-cause ED visit was higher in the polytherapy and monotherapy subcohorts (64.4 % and 63.1 %, respectively) than the treated seizure cluster subcohort (54.4 %).
The percentages of patients with ≥ 1 epilepsy-related OP visit and ≥ 1 epilepsy-related other visit were higher in the treated seizure cluster Note: Race is not available in the WHIO data.
Abbreviations: ASM = antiseizure medication; SD = standard deviation.* The sum of the patients from the three subcohorts is smaller than the number of patients in the all-epilepsy cohort because some patients were excluded from the monotherapy and polytherapy subcohorts due to the number of index ASMs being above the defined limits, and some patients were excluded because they did not meet a subcohort criterion of 12-month continuous enrollment during follow-up from the subcohort index date.Note: Claims that were denied by the insurer for payment were excluded.
Abbreviations: ASM = antiseizure medication; ED = emergency department; HCRU = healthcare resource utilization; OP = outpatient.* Other visit was defined as a medical claim that could not be clearly classified as ED, hospitalization, or OP.For example: a medical claim with hospitalization code but without overnight stay; a medical claim with and OP code that lasted more than one day; services provided in hospice, home, school, homeless shelter, military treatment facility, alcohol detoxication facility, etc.
subcohort versus the other subcohorts.Among the subcohorts, the rate of epilepsy-related hospitalizations was highest in the treated seizure cluster subcohort, with 25.3 % of patients experiencing one or more events.The rate was 22.8 % in the polytherapy subcohort and 19.3 % in the monotherapy subcohort.The rate of epilepsy-related ED visits was higher in the polytherapy subcohort (42.9 %) than the treated seizure cluster (40.1 %) and monotherapy (38.4 %) subcohorts.Over the 12-month follow-up, a higher percentage of OP visits in the all-epilepsy cohort involved primary care (general) services (34.0 %) than epilepsy/neurology specialists (11.2 %) (Table 4).The treated seizure cluster subcohort had the highest percentage of OP visits involving hospital (24.6 %), epilepsy/neurology specialists (15.3 %), and pediatric care (11.6 %).
The majority of patients in the all-epilepsy cohort had OP visits with primary care (general) services (85.1 % [13,364/15,705]).Among the subcohorts, a higher percentage of treated seizure cluster and polytherapy patients sought care from epilepsy/neurology specialists (69.Epilepsy-related costs over 12-month follow-up were highest in the treated seizure cluster subcohort, followed by the polytherapy and monotherapy subcohorts (Table 5).The same pattern was observed with epilepsy-related medical and pharmacy costs.All-cause healthcare costs showed similar differences between the cohorts as those related to epilepsy, and again the pattern was the same for medical and pharmacy costs.

Discussion
We studied an inclusive epilepsy population using a unique multipleinsurer statewide charge dataset covering all medical service-related charges.WHIO is the only statewide voluntary all-payer claims database in the US and includes more than 4.5 million individuals (75 % of the Wisconsin population), more than 320 million claims and $147 billion in billed charges from commercial and Medicaid (health maintenance organizations and fee-for-service) plans [18].Of 16,384 patients from the WHIO population included in the all-epilepsy cohort, 526 (3.2 %) were in the treated seizure cluster subcohort, representing a large population of this understudied patient group.This group was younger than the monotherapy and polytherapy subcohorts, consistent with prior findings of higher rates of seizure clusters with younger age [10].
From 2017 to 2019 data-a timeframe prior to the availability of nasal spray rescue medications-a historical baseline situation was characterized, which will be used to inform future studies of the impact of new treatment options.Overall, we found: (1) ASM retention rates just below 50 % for the all-epilepsy group and less than 50 % for the most prescribed medications; (2) the treated seizure cluster subcohort had low refill rates for rescue medications; and (3) the treated seizure cluster subcohort incurred considerably higher medical service costs than the monotherapy and polytherapy subcohorts.A follow-up study will use this dataset as a historical baseline to assess the effect of nasal spray rescue medications after their integration into routine use (ie, post-COVID era, from 2022).
Our population comes from a dataset not biased by single insurer sources, potentially reducing bias introduced into the sample by specific populations by their insurance source.The population size yielded a large dataset including many patients treated for seizure clusters, and can be used to help define the challenges facing the seizure cluster population.In the current study, the treated seizure cluster subcohort was identified based on patients having a rescue medication claim and 12 months of continuous medical and pharmacy coverage, resulting in an observed seizure cluster prevalence of 3.2 %.This rate is within the range reported by another database study (3.0 % [10]) and retrospective chart review study (14.9 % [8]).However, because seizure clusters were identified based upon claims for approved seizure cluster medications in our study, rather than diagnostic codes for seizure clusters, and our study excluded a pediatric population < 12 years of age, our reported  prevalence may be conservative.Important findings from the treated seizure cluster subcohort included: (1) medications typical for the treated seizure cluster subcohort were similar to other subcohorts; (2) prescriptions for the available rescue medications were low for this subcohort; and (3) higher all-cause and epilepsy-caused costs than other subcohorts.
The similarity of ASMs between our subcohorts was unexpected.Patients treated for seizure clusters experience greater challenges than patients without seizure clusters.Previous reports have shown the indirect cost of seizure clusters in terms of negative impact on patients' and their caregivers' work productivity.In an online survey, 69 % of patients with seizure clusters and 48 % of their caregivers reported that seizure clusters had negatively affected their job/career or ability to work [3].A retrospective online chart review of patients with seizure clusters found that only 36 % were employed [19].The treated seizure cluster subcohort, despite these additional burdens, shared similar ASM use with other subcohorts, suggesting they are otherwise being managed similarly.The frequency of ASM prescriptions during follow-up in the treated seizure cluster subcohort suggests the use of seemingly unsuccessful medications in the presence of continued seizure clusters, which is unexpected, considering the burden of ongoing seizure clusters.It would be of interest to assess retention of index ASM in this subgroup, which may suggest the potential for improvement by changing ASM, a capitulation to ASM ineffectiveness, or exhaustion of ASM options.
The treated seizure cluster patient group was sampled when existing rescue medications approved for seizure clusters was limited to branded and generic diazepam gel.Oral lorazepam was not included and Nayzilam (branded midazolam nasal spray) was not yet available.Refill rates over 12-month follow-up for rescue medications in the treated seizure cluster subcohort were low, suggesting significant underutilization of rescue medications.Patients with seizure clusters can experience more than ten seizure cluster episodes per year [20].If patients in our study used rescue medication consistently, a higher refill rate would be expected.Our findings corroborate data from a systematic review suggesting that rescue medications were underprescribed and underutilized: in adults with seizure clusters, prescription rates varied from 24.6 % to 61.4 %, and fewer than half of the patients had used rescue medications [20].
Importantly, medical services costs were higher in the seizure cluster subcohort.This subcohort had the highest epilepsy-related and all-cause medical, pharmacy, and total costs among the subcohorts.The seizure cluster subcohort had higher rates of epilepsy-related OP visits and epilepsy-related other visits (89.7 % and 71.3 %, respectively) than the all-epilepsy cohort (76.4 % and 55.5 %), suggesting the effect of this patient population on healthcare systems.The rate of epilepsy-related hospitalizations was higher in the seizure cluster subcohort (25.3 %) than in the ASM monotherapy and polytherapy subcohorts (19.3 % and 22.8 %, respectively).This finding is consistent with a cohort study which found that at baseline, a significantly higher percentage of patients with versus without seizure clusters reported ever having had a seizure-related hospitalization (73 % versus 59 %; adjusted odds ratio 5.3, 95 % confidence interval 1.5-17.6;P = 0.006) [9].The current study extends these results by providing direct costs for patients treated for seizure clusters.The treated seizure cluster subcohort had the highest epilepsy-related and all-cause medical, pharmacy, and total costs among the subcohorts.The mean epilepsy-related total cost was $76,134 for the seizure cluster subcohort and $30,745 and $49,383 for the monotherapy and polytherapy subcohorts, respectively, while mean all-cause total cost was $114,717 for the seizure cluster subcohort and $89,324 and $101,506 for the monotherapy and polytherapy subcohorts.
This study has limitations.The lack of consensus in defining seizure clusters leads to variability in reported rates in the literature.Because there was no diagnostic code for seizure clusters during the selected time period, identifying this patient population from claims data was challenging.In the current study, the treated seizure cluster subcohort was identified based on patients having a rescue medication claim and 12 months of continuous medical and pharmacy coverage.It is possible our all-epilepsy cohort included patients with seizure clusters who were not identified using our criteria.This study did not include oral ASMs as rescue medications.Lorazepam is commonly used for anxiety disorder and was excluded so use for behavioral health did not create a falsely elevated patient selection.The most commonly used rescue medication in adults with seizure clusters has been reported as oral lorazepam (28.9 %), followed by rectal diazepam (7.8 %) and oral diazepam (7.0 %) [8]; and in pediatric patients, rectal diazepam (56.4 %), followed by oral clonazepam (12.8 %) and oral lorazepam (7.7 %) [21].Our treated seizure cluster subcohort did not include patients in the all-epilepsy cohort who might rely on oral drugs as rescue medication, which could have led to underestimation of seizure cluster prevalence and the relatively low proportion of patients in this subcohort, which is another limitation.Exclusion of children aged < 12 years-a population with more frequent use of diazepam gel-limited the size of the treated seizure cluster subcohort.The lack of statistical comparison between the subcohorts also limits study interpretation.Finally, regarding prevalence estimates, the Centers for Disease Control and Prevention (CDC) estimated that 1.2 % of the US population (3.4 million people) were living with epilepsy in 2015, including 59,600 patients in Wisconsin [1].Our study captured approximately 27 % of these patients in Wisconsin.We showed that the prevalence of epilepsy decreased from 711.2 per 100,000 in 2017 to 685.7 per 100,000 in 2019.These figures are lower than those from the CDC, but our criteria excluded patients < 12 years of age.The exclusion of the Medicare fee-for-service population may contribute to the skew in prevalence rates as this group represents approximately 56 % of the Medicare population [17].
Over follow-up, the percentage of patients with all-cause OP visits, all-cause other visits, and all-cause hospitalizations were similar among the all-epilepsy cohort and subcohorts, whereas the percentage of patients with all-cause ED visits was higher in the ASM monotherapy and polytherapy subcohorts than in the seizure cluster subcohort.The latter finding may reflect the younger age of the treated seizure cluster subcohort, which had a lower percentage of patients with comorbidities.In a 1-year prospective study, a similar proportion of patients ≥ 12 years of age with active epilepsy who had seizure clusters and those who did not have seizure clusters had seizure-related ED visits (30.6 % versus 29.2 %, P = 1.000), and the association between seizure type (isolated seizure versus cluster) and ED visits was not significant (P = 0.272) [7].The discrepancies with our results may be explained by differences in the study design.
The use of insurance claims is subject to some common limitations (eg, missing or misclassification of data, no assertion that drugs filled were taken as prescribed, omitted data of interest [eg, seizure frequency]).Claims databases do offer value, however, when evaluating HCRU and cost estimates.Costs did not consider indirect costs (eg, lost productivity) or quality of life, which are not captured in claims data.Drugs used during hospital stays may have been underreported within the database.Drug information in the database came mostly from OP pharmacy claims, so this limitation can be considered negligible as ASMs are less likely to be prescribed in a hospital setting.Additionally, results may not be generalizable to the overall US epilepsy population or patients without health insurance, as only insured patients were assessed.

Conclusions
This study characterized a large group of patients with epilepsy who were treated for seizure clusters with diazepam gel, prior to the availability of nasal spray rescue medications.Patients in the treated seizure cluster subcohort incurred higher all-cause and epilepsy-related costs, and higher epilepsy-related HCRU in terms of OP visits and other visits than the all-epilepsy group and the monotherapy and polytherapy subcohorts, demonstrating the impact of these patients on healthcare systems.The findings of this large retrospective database analysis highlight the need for rescue treatment for patients with epilepsy experiencing seizure clusters.Seeking care from specialist providers with expertise in treating refractory epilepsy and appropriate treatment for patients with seizure clusters may help to reduce the burden of seizure clusters on people with epilepsy.It will be of interest to assess if the introduction of nasal spray rescue medications has significantly shifted the care of patients experiencing seizure clusters.

Declaration of competing interest
GM is a consultant for UCB Pharma.PE, DE, and MT are employees of UCB Pharma.PE holds stock options in UCB Pharma.

cohort aged ≥ 12 years Subcohorts ASM monotherapy aged ≥ 12 years ASM polytherapy aged ≥ 12 years Seizure clusters aged ≥ 12 years
Abbreviations: ASM = antiseizure medication; WHIO = Wisconsin Health Information Organization.*Individualswith a charge during the calendar year; cohorts were not matched for age.G.L. Morris III et al.

Table 2
Patient characteristics at index date.

Table 4
Number of OP visits by type of provider over 12-month follow-up (patient numbers in column headings are those who had ≥ 1 OP visit over 12-month follow-up).
* Primary care (general) servicesincludes primary care/internal medicine/ family medicine/nurse practitioner.‡Hospitalsettingcarebya provider in a hospital setting.§Epilepsy/neurologyanycareassociated with an epilepsy or neurology provider.¶Otherthiscategory includes healthcare professionals that do not fall into the other categories (eg, pharmacy, psychology, other mental health, laboratory, imaging center, clinical social worker, durable medical equipment, midwife, dietician/nutrition, physical therapy, rehabilitation, speech therapy, etc.).‖Pediatric careany care by a pediatric provider.

Table 5
Costs over 12-month follow-up, adjusted to 2021 Consumer Price Index.