Abstracts

Rationale: The recent national focus on systemic racial inequities has highlighted the longstanding structural disparities in US healthcare^1–3. Racial and ethnic disparities are unfortunately seen in epilepsy care in access, diagnosis, and treatment^4–7. We sought to analyze these disparities in healthcare access, based on presentation to phase II monitoring, and treatment, based on time to surgery.

Methods: We analyzed demographic information of patients enrolled in the DARPA Restoring Active Memory (RAM) trial, which consists of 386 subjects across 9 institutions in the United States. Demographics included: age, sex, race, ethnicity, handedness, education, age at first seizure, age at surgery, seizure etiology, and institution. We calculated the time from diagnosis to surgery and used census data of the surrounding county of each institution to calculate a Shannon’s diversity index (H), given by the formula H=-Σ^N_i pi lnpi where i represents each racial and ethnic category^9,10. An H ratio was calculated by normalizing to county, with a score of 1 indicating equivalent diversity. Racial identity was categorized as Asian-American and Pacific Islander (AA), Black or African American (Black), White or Caucasian (White), or Other. Ethnic categories were Hispanic and non-Hispanic. County proportions of race and ethnicity were multiplied by 50 (average institutional sample size) for purposes of statistical comparison. A linear regression model with 5-fold cross-validation and LASSO variable selection/regularization was applied with time from diagnosis to surgery as the outcome measure.

Results: There were statistically significant differences in racial distribution at 5 institutions, with a larger proportion of White patients relative to their counties. Two institutions demonstrated significantly higher proportions of non-Hispanic patients. H ratios for both race and ethnicity demonstrated overall lower diversity at seven institutions. In our subjects, the average time from diagnosis to surgery was 18.6 years. College education, right-handedness, other etiology, and Black race were associated with an earlier time from diagnosis to surgery; left-handedness, White race, older age, and male sex were associated with delayed surgery. The average time from diagnosis to surgery was 15.6 years for Black subjects (n = 37) and 19.3 years for White subjects (n = 213), with p-value of 0.072; 17.96 years for college education (n = 56) and 18.2 for high school education (n = 158), with p-value of 0.88; and 17.8 years for right-handed patients (n = 227) and 23.8 for left-handed patients (n = 26), with p-value of 0.044 (2-sided, 2-sample Welch’s t-test).

Conclusions: The racial and ethnic distributions of patients undergoing Phase II monitoring are significantly less diverse compared to those of their respective counties. Patients who are college-educated or Black have shorter time to surgery compared to those who are male, older, or White. Further investigation is warranted into possible selection biases and the significance of disparities in earlier versus delayed surgeries, with the hope of equalizing access and treatment.

Funding: Please list any funding that was received in support of this abstract.: Defense Advanced Research Projects Agency (DARPA) Restoring Active Memory (RAM).