Association of race and ethnicity to incident epilepsy, or epileptogenesis, after subdural hematoma

Abstract

Objective

To determine whether race is associated with the development of epilepsy after subdural hematoma (SDH), we identified adult survivors of SDH in a statewide administrative dataset and followed them up for at least 1 year for revisits associated with epilepsy.

Methods

We performed a retrospective cohort study using claims data on all discharges from emergency departments (EDs) and hospitals in California. We identified adults (age ≥18 years) admitted from 2005 to 2011 with first-time traumatic and nontraumatic SDH. We used validated diagnosis codes to identify a primary outcome of ED or inpatient revisit for epilepsy. We used multivariable Cox regression for survival analysis to identify demographic and medical risk factors for epilepsy.

Results

We identified 29,342 survivors of SDH (mean age 71.2 [SD 16.4] years, female sex 11,954 [41.1%]). Three thousand two hundred thirty (11.0%) patients had revisits to EDs or hospitals with a diagnosis of epilepsy during the study period. Black patients (n = 1,684 [5.7%]) had significantly increased risk compared to White patients (n = 16,945 [57.7%]; hazard ratio [HR] 1.45, 95% confidence interval [CI] 1.28–1.64, p < 0.001). Status epilepticus during the index SDH admission, although infrequent (n = 94 [0.3%]), was associated with a nearly 4-fold risk of epilepsy (HR 3.75, 95% CI 2.80–5.03, p < 0.001). Alcohol use, drug use, smoking, renal disease, and markers of injury severity (i.e., intubation, surgical intervention, length of stay, disposition other than home) were also associated with epilepsy (all p < 0.05).

Conclusions

We found an association between Black race and ED and hospital revisits for epilepsy after SDH, establishing the presence of a racial subgroup that is particularly vulnerable to post-SDH epileptogenesis.

A particularly increased risk of seizures has been seen in traumatic brain injury (TBI) complicated by subdural hematoma (SDH).^1,2 Prior studies have estimated cumulative 2-year incidences of posttraumatic epilepsy (PTE) after SDH to be upward of 40%.^1,3

Posttraumatic seizures are an independent marker of poor functional and social outcome^4–7; therefore, identification of vulnerable populations is important for efforts to improve the treatment and long-term management of TBI. Delayed posttraumatic seizures have a higher rate of recurrence and development of epilepsy,^8,9 along with a significant associated burden of disease.^10,11 In addition, antiepileptogenesis trials have been limited by the large patient numbers needed to test interventions and the potential unnecessary exposure to adverse effects in low-risk patients. Identifying subpopulations at increased risk for PTE would thus contribute to optimized patient identification for preventive studies.²

Among epidemiologic studies, few have examined racial differences in the development and outcome of epilepsy specifically after SDH. Prior studies have identified Black race to be a risk factor for undifferentiated PTE¹² and for status epilepticus in patients admitted with SDH,⁷ but the existing literature is inconsistent,¹³ and the reasons behind such observations are poorly understood. We used administrative claims data from a statewide population to describe the epidemiology of epilepsy specifically in survivors of SDH and to test the hypothesis that Black race is associated with emergency department (ED) and hospital revisits for epilepsy after SDH.

Methods

Standard protocol approvals, registrations, and patient consents

Because the databases used comprised publicly available data and include only deidentified data, our study was exempt from review by our institutional review board.

Study design

We performed a retrospective cohort study using administrative claims data from the California State Inpatient Database¹⁴ (SID) and California State Emergency Department Database.¹⁵ The databases include standardized, deidentified, and quality-checked inpatient and ED records and is part of a federal-state-industry partnership called the Healthcare Cost and Utilization Project (HCUP), sponsored by the Agency of Healthcare Research and Quality.¹⁶ Patients in these databases are assigned an anonymous record number that allows them to be followed up between encounters across years.

Participants

We included adult (age ≥18) patients admitted between 2005 and 2010 with first-time SDH, identified with ICD-9-CM codes 852.2x, 852.3x, and 432.1 in the primary diagnosis code position. These codes have previously been validated in an institutional cohort and found to have 96% sensitivity and 89% to 92% specificity for traumatic and nontraumatic SDH.¹⁷ The years 2005 to 2010 were chosen because patient-specific longitudinal identifiers were introduced to the California SID in 2005 and are available up to 2011; therefore, to allow at least 1 year of follow-up, we included patients with index visits through 2010. Patients without longitudinal identifiers and nonresidents of California were excluded to maximize patient follow-up. Transfers to another acute care hospital were considered part of the same index hospitalization. Patients who did not survive to discharge were excluded because we were interested in delayed outcomes among survivors of SDH.

In sensitivity analyses, we excluded records of patients who had traumatic SDH and excluded patients with SDH identified to have a diagnosis of epilepsy before or during their index SDH hospitalization.

Outcome

We followed up patients with first-time SDH from their index hospitalization for any revisit, including hospital admissions in SID and revisits to EDs in the State Emergency Department Database, with a diagnosis of epilepsy. Epilepsy was identified with ICD-9-CM codes 345.x, which have previously been shown to have a positive predictive value of 84% to 98% in adult patients.^18–20

Covariates

To account for potential confounders, we evaluated the following epidemiologically and biologically relevant variables: demographic factors (age, sex, race, and insurance payer) and baseline comorbid conditions (dementia, alcohol use, drug use, acquired coagulopathy, platelet dysfunction, hypertension, hyperlipidemia, smoking, myocardial infarction, renal disease, liver disease), as well as the Elixhauser comorbidity index, which comprises a set of 28 comorbidity measures that can be used with large administrative datasets to control for the confounding effect of these comorbid conditions.²¹ We also adjusted for likely indications for anticoagulant use (venous thromboembolism, atrial fibrillation, valvular disease). Finally, we noted markers of severity during the index hospitalization (sepsis, intubation, surgical intervention including craniectomy or craniotomy, status epilepticus, length of stay, discharge home).

Coding for race in the HCUP SID includes race and ethnicity and is categorized into White, Black, Hispanic, Asian or Pacific Islander, Native American, and other. Because our cohort had only 18 Native Americans, for the purpose of our study, we included Native Americans in the “other” category. HCUP notes that if data sources supplied race and ethnicity in separate data elements, ethnicity takes precedence over race in setting the HCUP value for race. Patients with unknown or missing race were excluded from the analysis.

Statistical analysis

Descriptive statistic

To report crude rates, we used means (SD) for continuous variables as appropriate and counts (percentages) for discrete variables.

Univariable analysis

We used χ² tests for discrete covariates and t tests or Kruskal-Wallis tests for continuous variables with normal and nonnormal distribution, respectively.

Survival analysis

We used Kaplan-Meier survival statistics to calculate cumulative rates of revisit for epilepsy. Patients entered observation at the time of their initial admission with SDH and were censored at the time of in-hospital death or last available follow-up data.

Multivariable analysis

We used multivariable Cox proportional hazards models to assess the associations between covariates and our outcome of interest. For Cox analysis, patients with missing data on race and those reported as “other” were excluded. We visually inspected log-log plots to verify the proportional hazards assumption.

Matching

We compared epilepsy revisit rates between Black patients and age- and sex-matched White patients at a ratio of 3:1 from the final sample of survivors of first-time SDH admission.

All p values were 2 sided, and statistical significance was set at p < 0.05 for the single association test related to the primary analysis of interest. R version 3.4.1 was used for all analyses.

Data availability

This study was performed with publicly available data, as described above.

Results

Patient characteristics

We identified 29,342 survivors of first-time admissions for SDH in California hospitals from 2005 to 2010. Inclusion and exclusion criteria are summarized in figure 1. Participant characteristics, stratified by race, are displayed in table 1.

Figure 1. Inclusion and exclusion criteria.

ED = emergency department; SDH = subdural hematoma.

Table 1.

Baseline characteristics of SDH survivors, stratified by race and ethnicity

During a mean follow-up period of 3.2 years, 3,230 (11.0%) patients had revisit for epilepsy. Three thousand twenty-six (10.3%) patients had in-hospital deaths during the follow-up period. More than half of the revisits for epilepsy occurred in the first year after index SDH hospitalization. The cumulative rate of revisits with epilepsy among the population of interest was 6.5% at 1 year and 13.7% at 5 years (table 2). The median time to revisit for epilepsy was 0.79 years (9.47 months) (interquartile range 1.73 years).

Table 2.

Cumulative rates of epilepsy revisits after SDH^a

Primary analysis

Compared to other races, Black patients (n = 1,684 [6.1%]) had significantly higher rates of epilepsy after hospitalization for SDH. Cumulative 5-year rates of revisit for epilepsy were 24.2% for Black patients vs 15.7% for Hispanics, 12.7% for Whites, and 11% for Asians. Yearly cumulative rates are displayed in table 2 and plotted in figure 2. In multivariable Cox regression analysis, Black race was associated with a higher hazard of revisit for epilepsy. Alcohol use, drug use, smoking, and renal disease were also independently associated with epilepsy. Status epilepticus during SDH admission, although infrequent, was associated with a nearly 4-fold risk of epilepsy. These results are displayed in table 3.

Figure 2. Epilepsy revisit rate after subdural hematoma, stratified by race and ethnicity.

SDH = subdural hematoma.

Table 3.

Univariable and multivariable Cox regression analysis

Sensitivity analyses

Our findings were unchanged in subgroup analyses in which patients with traumatic SDH were excluded from the analysis and when patients with a diagnosis of epilepsy during or before SDH admission were excluded (table 4). Findings also persisted in a direct comparison of epilepsy revisit rates between Blacks and age- and sex-matched Whites (figure 3).

Table 4.

Multivariable Cox regression analyses for subgroups

Figure 3. Epilepsy revisit rate after subdural hematoma among Black participants and age- and sex-matched White participants.

Discussion

We observed an increased risk of ED and hospital revisits for epilepsy after SDH in Black patients compared to other races. This association persisted in subgroup analyses of specifically patients with nontraumatic SDH, patients with no prior visits for epilepsy, and age- and sex-matched comparison groups. These results establish the presence of a racial subgroup that is particularly vulnerable to post-SDH epileptogenesis and may inform future studies that evaluate antiepileptogenesis interventions and that more deeply investigate the elements of race that may be driving this association.

Limitations of our study have to do with those inherent to the use of large administrative datasets. While claims data offer large sample sizes and the power to detect even subtle epidemiologic patterns, the granularity of data is limited. Our data did not include radiographic information on the size of SDH, related edema and shift, or severity of clinical examinations. Our data also did not include information on EEG abnormalities thought to be predictive of PTE.² These and other unmeasured differences between Black and non-Black participants could represent sources of unmeasured confounding and partially underlie the observed associations.

Social factors that could mediate racial disparities in epilepsy risk might include disparities in socioeconomic status, postinjury resource availability, and preinjury exposures and risk factors.^22–24 We found that markers of illness severity during hospitalization (intubation, surgical intervention, length of stay, discharge disposition other than home) were independently associated with epilepsy revisit. Nonetheless, it is of note that these markers of illness severity trended toward higher frequency in Black participants and may still signal a related source of confounding. Similarly, we found an association between substance use (alcohol, smoking, and other drugs) and risk of epilepsy after SDH, which was independent of race after adjusted analysis but nonetheless involves variables that were seen at higher rates in Black participants. Because the variables available in the administrative dataset used cannot encompass the many complicated interactions between psychosocial risk factors, injury, and long-term health outcomes, ample suspicion should remain that unmeasured influences underlie the observed associations for which race may serve as a surrogate.

Racial disparities in the outpatient management of epilepsy have previously been noted in areas including specialty referrals, epilepsy surgery, medication adherence, and management of related psychosocial outcomes.²⁵ There is a possibility that Black patients seek care in the ED setting more often than patients of other races. In a study of ED visits for seizure in the United States from the National Hospital Ambulatory Medical Care Survey, Black patients formed a significantly higher proportion of visits than White patients.²⁶ Because the dataset we used was limited to inpatient and ED visits in California, our study would not be sensitive to identify new seizures managed entirely in the outpatient setting. Thus, our study is limited in the ability to precisely discern disparities in ED and inpatient seizure presentation from disparities in outpatient management of seizures. Structural racism may affect access to medications, counseling, and outpatient follow-up. This limitation notwithstanding, a minority of ED seizure visits are due to a known seizure disorder^26,27; thus, racial disparity in the development of new epilepsy must be considered. Regardless of the mechanism underlying presentation to the ED vs another outpatient setting, it would appear to signify a degree of seizure severity, failure of ambulatory support, and thus functional impact on post-SDH morbidity. Identifying risk factors for such events and correcting disparities in their prevention and treatment are therefore of particular interest to medical and public health efforts to improve long-term SDH outcomes.

Our subgroup analysis of epilepsy revisits among the nontraumatic SDH population found that, even in nontraumatic SDH, Black race was associated with increased risk of revisit to the ED or hospital for epilepsy (table 3). The presence of SDH in TBI has been found to be strongly associated with epileptogenesis²; our findings suggest that even outside the traumatic etiologies of SDH, having blood in the subdural space is by itself a risk factor for epileptogenesis. Patients with subdural blood may face a variety of predilections for epileptogenesis separate from those relatable to trauma, and our results raise the possibility that race is one of such predilections. Of course, interpretation of these results should be tempered by the possibility of clinical misclassification of SDH as traumatic vs nontraumatic, as in cases in which there might be historical uncertainty. However, the clinical codes used in this study have previously been validated in an institutional cohort and found to have 96% sensitivity and 89% to 92% specificity for traumatic and nontraumatic SDH.¹⁷

Prior epidemiologic studies have shown a higher prevalence of epilepsy both in Black populations in the United States compared to their White neighbors²⁸ and in populations in Africa relative to Europe and North America.²⁹ A survey-based study conducted in the District of Columbia found significantly higher rates of lifetime and active epilepsy among Blacks compared to Whites and among individuals with lower education levels.³⁰ Both genetic and socioeconomic theories have been described to explain racial differences in epilepsy prevalence¹²; however, whether similar explanations can be extrapolated to the development of post-SDH epilepsy is unknown. To remove the possible contribution of racial differences in baseline epilepsy to the results of this study, we performed a subgroup analysis in which patients identified as having a diagnosis of epilepsy before or during their index SDH hospitalization were excluded. The observed racial differences in outcome persisted in this analysis (table 3).

Age appeared to be inversely associated with epilepsy risk. Because Black patients in our cohort were on average younger, we evaluated the possibility of confounding due to dropout of disproportionately older White patients as a result of mortality by comparing epilepsy revisit rates for Blacks with those of age- and sex-matched White patients. In this analysis, the elevated risk of epilepsy among Black patients persisted (figure 3).

We found an association between Black race and ED and hospital revisits for epilepsy after SDH, which is unrelated to traumatic etiology or racial differences in the prevalence of baseline epilepsy. Our results should prompt additional investigation into socioeconomic and health service–related factors that may underlie this observation.

Glossary

ED

emergency department

HCUP

Healthcare Cost and Utilization Project

ICD-9-CM

International Classification of Diseases, 9th revision, clinical modification

PTE

posttraumatic epilepsy

SDH

subdural hematoma

SID

State Inpatient Database

TBI

traumatic brain injury

Appendix. Authors

Study funding

G.J. Falcone is supported by the NIH (K76AG059992 and R03NS112859), the American Heart Association (18IDDG34280056), the Yale Pepper Scholar Award (P30AG021342), and the Neurocritical Care Society Research Fellowship.

Disclosure

The authors report no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.