Abstract Body: Introduction: Identification of modifiable drivers of the disparity between neighborhood socioeconomic status (nSES) and stroke outcomes remains elusive. Using a causal inference framework, we evaluate the degree of mediation by modifiable individual- and neighborhood-level characteristics in the association between worse nSES and worse 90-day stroke outcomes.
Methods: Our analytic sample was composed of 90-day community-dwelling stroke survivors with non-missing data between 2010-2020 from the Brain Attack Surveillance in Corpus Christi (BASIC) cohort. Variables were derived from BASIC, Redfin and the National Neighborhood Data Archive (NaNDA). The census-derived nSES index was our exposure. 90-day outcomes were functional status via the average of the 22-item ADL/IADL Scale, post-stroke depression symptoms via the PHQ-8, and quality-of-life via the average of the Stroke Specific QoL Scale (SS-QOL). Confounder-adjusted multi-level Bayesian G-computation causal mediation analysis was used to determine to what extent hypothetical interventions on individual level (receipt of thrombolysis or thrombectomy) and neighborhood level (density of home health services, walkability, transportation access, density of recreation centers, and density of rehabilitation providers) characteristics would reduce the disparity of 90-day outcomes by nSES. We repeated the analyses stratified by stroke severity (NIHSS ≤ 5 or >5).
Results: Among the 2,118 stroke survivors in 68 census tracts, 47% were female and 59% were Mexican-American. The respective median (IQR) values of nSES, PHQ-8, IADL/ADL score, and SS-QOL were: -0.7 (-1.2 – -0.2), 5.0 (1.0 – 11.0), 2 (1.3 – 3.0), and 3.5 (2.6 – 4.3). Figure 1 displays the total, direct and decomposed mediated indirect effects by groupings of correlated mediators. Figure 2 displays the results stratified by stroke severity. Table 1 displays the indirect effects mediated by each specific characteristic in our total and stratified samples.
Discussion: Lower nSES portends worse stroke outcomes. However, this disparity was not significantly driven by receipt of hyperacute treatment, neighborhood walkability, transportation access, density of rehabilitation providers, density of home health services, or density of recreation centers irrespective of stroke severity. Future work should explore other potential intervenable drivers of the disparity such as time-to-treatment and discharge destination using a similarly rigorous framework.