Stroke systems of care: evaluating disparities after stroke legislation implementation in Rhode Island Authors: Abigail A Teshome, Joshua R Feler, Mazen Taman, Carl Michael Porto, Elizabeth Perelstein, Mark Cort, Elias Shaaya, Krisztina Moldovan, Radmehr Torabi, Dylan N Wolman, Mahesh V Jayaraman, Jeffrey Proulx Published: 2026 ABSTRACT AbstractIntroductionThe Rhode Island Stroke Prevention and Treatment Act aimed to establish a standardised system of care for stroke patients through designated stroke centres, revise emergency medical services (EMS) protocols, and establish a centralised stroke database. While Stroke Systems of Care (SSOC) laws have proven effective in reducing overall stroke mortality, their benefits among racial minorities and economically disadvantaged groups remain unclear. This study investigates the impact of the Act and its amendment on stroke mortality rates across different racial groups.MethodsData from the Rhode Island Department of Labor and Training, Centers for Disease Control Atlas of Heart Disease and Stroke, and US Census Bureau for 2006–2019 were analysed. A linear regression model analysed mortality rates among White, Black and Hispanic residents before and after the Act’s implementation. Trends were further validated using a segmented regression and mixed-effects model.ResultsDifference-in-differences analysis showed SSOC implementation reduced statewide all-stroke mortality by 8.56 deaths per 100 000 (95% CI −10.56 to −6.56; p<0.001). White and Hispanic groups declined, but the Black population diverged significantly from White (+20.80; 95% CI 4.83 to 36.77; p=0.011). Interrupted time series confirmed higher Black baseline mortality (β=9.70; 95% CI 2.00 to 17.40; p=0.014) and an abrupt post-implementation increase (β=29.14; 95% CI 16.64 to 41.63; p<0.001), followed by annual decline (β=−6.98; 95% CI −10.24 to −3.61; p<0.001). After amendment, Black mortality rose immediately (+19.80; 95% CI 2.15 to 37.44; p=0.028) and annually (+8.40; 95% CI 1.33 to 15.47; p=0.020). White and Hispanic groups showed no significant amendment-period changes.ConclusionWhile SSOC significantly reduced overall stroke mortality in Rhode Island, the benefits were not equitably distributed across racial groups. These findings underscore the need for future stroke legislation to incorporate culturally competent, evidence-based strategies addressing social determinants of health to reduce disparities and ensure equitable care. WHAT IS ALREADY KNOWN ON THIS TOPIC The number of people afflicted with stroke in the USA is steadily increasing, with a large proportion of stroke patients identifying as Black and/or Hispanic. Stroke Systems of Care (SSoC) aim to optimise acute stroke management by standardising protocols for emergency triage, patient assessment, stroke centre designation, statewide stroke data collection, among other factors. SSoC implementation has been associated with decreases in both in-hospital and crude stroke mortality. The number of people afflicted with stroke in the USA is steadily increasing, with a large proportion of stroke patients identifying as Black and/or Hispanic. Stroke Systems of Care (SSoC) aim to optimise acute stroke management by standardising protocols for emergency triage, patient assessment, stroke centre designation, statewide stroke data collection, among other factors. SSoC implementation has been associated with decreases in both in-hospital and crude stroke mortality. WHAT THIS STUDY ADDS While SSoC has been linked to lower state-level stroke mortality, there is sparse literature about how the degree to which this benefit is experienced by various demographic groups. This study analyses the association between SSoC and stroke mortality among White, Black and Hispanic groups. While SSoC has been linked to lower state-level stroke mortality, there is sparse literature about how the degree to which this benefit is experienced by various demographic groups. This study analyses the association between SSoC and stroke mortality among White, Black and Hispanic groups. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY This study emphasises the fact that while SSoC addresses acute stroke care, chronic factors may play a role in why high-risk groups demonstrated a different trend in mortality rate. Social determinants of health may play contribute to inequities in stroke prevalence between demographic groups and may not be addressed by directly by SSoC. Initiatives aimed at closing gaps in primary care access may act synergistically with SSoC to address the high rates of stroke prevalence among high-risk groups. This study emphasises the fact that while SSoC addresses acute stroke care, chronic factors may play a role in why high-risk groups demonstrated a different trend in mortality rate. Social determinants of health may play contribute to inequities in stroke prevalence between demographic groups and may not be addressed by directly by SSoC. Initiatives aimed at closing gaps in primary care access may act synergistically with SSoC to address the high rates of stroke prevalence among high-risk groups. Introduction Stroke Systems of Care (SSOC) laws are intended to optimise stroke treatment by coordinating the four major components of stroke care as defined by the Centers for Disease Control: prehospital patient assessment, emergency transport, acute treatment and rehabilitation.1 Currently, 41 states in the USA have stroke legislation, though their contents are heterogenous; nearly half of all states lack efficient emergency medical services (EMS) protocols, delaying access to medical attention and increasing the risk of major disability and mortality.2 3 Current literature suggests that states with any stroke legislation demonstrate lower age-adjusted stroke mortality, a greater number of certified stroke centres, better access to primary stroke centres (PSCs) and lower hospital costs.3 4 Despite these findings, there is limited research on the effectiveness of stroke policies among racial minorities. Communities of colour, among whom there is often greater prevalence of stroke risk factors, also experience barriers to acute stroke care, including lower EMS utilisation, delays in treatment and lower likelihood of transfer to endovascular-capable or other centres.5 6 The Rhode Island Stroke Prevention and Treatment Act of 2009 sought to establish a standardised and comprehensive system of care for stroke patients in the state of Rhode Island.7 Key components of the bill include the establishment of a process for PSC designation, designation of seven PSCs, designation of one Acute Ready Stroke Hospital (ARSH), review of EMS protocols, development of a statewide stroke database and initiation of an EMS data collection system for quality improvement (figure 1A). This Act was amended in 2015 to include protocol for field-triage directly to a comprehensive stroke centre (CSC), bypassing PSCs, following the evidence in support of mechanical thrombectomy as standard of care for large vessel occlusion (LVO).8 Other amendments to the bill included the development of a CSC certification programme, participation with nationally recognised stroke data platforms, establishment of standardised reporting requirements for data in the statewide stroke database and standardised protocols for transporting patients receiving intravenous alteplase who are candidates for thrombectomy (figure 1B). Previous studies have demonstrated within-hospital racial differences after stroke centre certification, with increased likelihood of stroke treatment being demonstrated in White patients but not Black patients.9 Black patients may have decreased odds of EMS prehospital notification as compared with White patients, which may contribute to the gap in treatment likelihood.5 While studies evaluating SSOC suggest decreased overall stroke mortality in a given population, it is unclear whether these policies have an equitable impact across races.3 This study aims to evaluate the extent to which SSOC law implementation was associated with decreased stroke mortality rates among racial and ethnic subgroup populations in Rhode Island. Overview of changes enacted by Rhode Island Stroke Legislation. (A) Components of the Stroke Prevention and Treatment Act of 2009. (B) Components of the Act’s Amendment in 2015. CSC, comprehensive stroke centre; EMS, emergency medical services; LVO, large vessel occlusion; PSC, primary stroke centres. Methods Publicly available data from the Rhode Island Department of Labor and Training10 were collected to capture total population counts stratified by race for each county in Rhode Island between 2005 and 2019, reflecting 4 years before the implementation of the SSOC until 4 years after passage of the amendment and field triage implementation. Epidemiological data from the Centers for Disease Control (CDC) Atlas of Heart Disease and Stroke11 were collected, including county-level stroke prevalence and mortality rates for all strokes, measured per 100 000 individuals aged 35 and older (35+) across White, Black and Hispanic populations. State-level mortality rates were collected, as well as county-level rates across the five counties in Rhode Island: Bristol, Kent, Newport, Providence and Washington County. The CDC reports stroke mortality as 3-year smoothed averages. Data from the period 2005–2007 to 2018–2020 were collected. A difference-in-difference (DiD) analysis was conducted to compare stroke mortality rates among White, Black and Hispanic groups before and after the Act’s implementation in 2009, as well as following the passage of the amendment in 2015. To account for secular trends in mortality, an interrupted time series (ITS) regression with county-level random effects was used to compare stroke mortality rates among racial groups across three intervals: before the Act’s implementation (pre-2009), after implementation (2010–2015) and after the passage of amendments (post-2016). The first intervention (t1) was set at 2010, corresponding to the 2009–2011 3-year range when the initial Act was implemented. The second intervention was marked as 2016, corresponding to the 2015–2017 range when the amendment took effect. This was done to account for the expected lag in policy implementation after passage. The year was centred in 2006, meaning the intercept represents predicted mortality in the starting period (2005–2007 average). Fixed effects included a linear time trend, piecewise linear time trends following each intervention, race and the interaction between race and policy implementation (online supplemental figure 1). Statistical analysis was performed in RStudio version 4.4.1. Results Epidemiological landscape of Rhode Island In the beginning of the study period, Rhode Island had 1 060 196 residents, with over half residing in Providence County (630 186). When isolating people greater than 35 years of age, there were 574 626 residents in the state with 341 560 in Providence County. This county houses the state’s only CSC as well as four of the seven designated PSCs. Providence County also had the highest proportion of Black and Hispanic residents (9.2% and 16.5% respectively). In the same period, the least racially diverse county that year was Bristol County, which had a 35+ population of 27 175; 1.1% of the population was Black and 1.6% was Hispanic. This county did not have a certified stroke centre designated by the Act. At the end of the study period, the total population in Rhode Island increased by 3.2%, and the 35+ population grew by 3.9% with minimal change to county-level racial makeup. Total population mortality decreased from 70.9, 59.9 and 55.1 per 100 000 at the beginning of the study in 2006, implementation of the initial policy in 2010 and implementation of the amendment in 2016, respectively (figure 2A–C). Across the state, the White group had mortality rates of 70.6, 58.7 and 55.0 per 100 000 in 2006, 2010 and 2016, respectively (figure 2D–F). The Black group had the highest average rate throughout the study period. Statewide mortality started at 88.3, 92.2 and 70.8 per 100 000 at the same time points (figure 2G–I). Hispanic mortality was the lowest throughout the study period with mortality rates of 57.9, 52.7 and 41.4 in 2006, 2010 and 2016 respectively (figure 2J–L). When analysing individual counties, Providence County had the highest overall stroke mortality in the total population throughout the entire study period at 59.4 per 100 000. Conversely, Washington County had the lowest stroke mortality in the total population at 57.4 per 100 000. Heat maps depicting the change in all stroke deaths in the total population (A) before vs after initial policy implementation, (B) after amendments calling for field-triage in the post-policy group only, and (C) before and after field-triage implementation in the whole sample. (D) Changes in stroke mortality were then stratified across White (D–F), Black (G–I) and Hispanic (J–L) subgroups. Overlayed symbols represent the designation of stroke centres by level of certification. All rates are reported per 100 000 people. ASRH, Acute Stroke Ready Hospital; CSC, comprehensive stroke centre; PSC, primary stroke centre. The effect of stroke legislation implementation DiD analysis revealed that after SSOC implementation, all-stroke mortality decreased statewide by 8.56 deaths per 100 000 (95% CI −10.56 to −6.56; p<0.001). Though White and Hispanic subgroups demonstrated a decline, the Black population was not identified to have changed and diverged significantly from the White group (+20.80 deaths per 100 000 (95% CI 4.83 to 36.77, p=0.011) (table 1). In the interrupted time series regressions, the Hispanic group demonstrated a significantly lower base mortality rate compared with the White population (β=−11.18, 95% CI −19.09 to −3.26; p=0.006) while the Black group was observed to have a higher baseline mortality rate (β=9.70, 95% CI 2.00 to 17.40; p=0.014). The period immediately after policy implementation was associated with an instantaneous change in mortality among the Black group (β=29.14, 95% CI 16.64 to 41.63; p<0.001) followed by an additional decrease in mortality rate of 6.98 per 100 000 people per year (95% CI −10.24 to −3.61; p<0.001) over the base rate of change. When analysing individual counties, Kent County had the largest decrease in this group (β=−13.30, 95% CI −18.70 to −7.90; p<0.001). This period was not associated with significant changes in stroke mortality among the White or Hispanic groups when accounting for secular trends (table 2). Model-predicted mortality trajectories demonstrate persistently higher mortality burden in the Black population, along with race-specific shifts in level and slope after each policy intervention (figure 3). Model-predicted all-stroke mortality rates by race/ethnicity across Rhode Island (2006–2019), derived from an ITS with mixed effects model with the White group as the reference group. Vertical dashed lines indicate the two policy interventions: T1 (Statewide Stroke Management Policy, 2010) and T2 (Field Triage Amendment, 2016). Predicted values reflect fitted regression estimates accounting for baseline secular trends, immediate level shifts and post-intervention slopes for each racial/ethnic group. ITS, interrupted time series. Table 1Difference-in-difference analysis of crude mortality rate changes across the study periodPopulation groupAll-stroke mortality change (95% CI)P valueDifference in difference from non-Hispanic White (95% CI)P valueA. Differences in crude mortality rate changes pre-policy and post-policy Total population−8.56 (−10.56 to −6.56) <0.001 —— Non-Hispanic White−9.01 (−10.94 to −7.08) <0.001 Reference— Non-Hispanic Black-11.79 (−27.65 to +4.07)0.139+20.80 (+4.83 to +36.77) 0.011 Hispanic−6.39 (−10.66 to −2.12) 0.005 +2.62 (−2.06 to +7.30) 0.273 B. Differences in crude mortality rates pre-field and post-field triage—whole cohort (2006–2019) Total population−7.41 (−9.37 to −5.44) <0.001 —— Non-Hispanic White−6.73 (−8.71 to −4.75) <0.001 Reference— Non-Hispanic Black−11.07 (−20.97 to −1.17) 0.029 −4.34 (−14.43 to +5.75)0.399 Hispanic−10.26 (−14.59 to −5.94) <0.001 −3.53 (−8.29 to +1.23)0.146C. Differences in crude mortality rates pre-field and post-field triage—post-policy cohort only (2010–2019) Total population−3.13 (−4.48 to −1.77) <0.001 —— Non-Hispanic White−2.23 (−3.46 to −1.00) <0.001 Reference— Non-Hispanic Black−15.86 (−30.52 to −1.19) 0.035 −13.63 (−28.31 to +1.05)0.069 Hispanic−7.07 (−11.81 to −2.32) 0.005 −4.84 (−9.74 to +0.06)0.053Difference-in-difference analysis of stroke mortality rates among the total population and stratified by race. (A) After policy implementation, all stroke mortality only decreased in the total, White and Hispanic population. The Black population demonstrated an increase in all stroke mortality during this period. (B) When analysing the entire sample and categorising pre-field and post-field triage eras, all groups demonstrated a decline in all stroke mortality. (C) In the post-policy cohort only (4 years before and after field triage implementation in 2015), all groups demonstrate a decline in all-stroke mortality.Bold values indicate statistical significance (p<0.05). Difference-in-difference analysis of crude mortality rate changes across the study period Difference-in-difference analysis of stroke mortality rates among the total population and stratified by race. (A) After policy implementation, all stroke mortality only decreased in the total, White and Hispanic population. The Black population demonstrated an increase in all stroke mortality during this period. (B) When analysing the entire sample and categorising pre-field and post-field triage eras, all groups demonstrated a decline in all stroke mortality. (C) In the post-policy cohort only (4 years before and after field triage implementation in 2015), all groups demonstrate a decline in all-stroke mortality. Bold values indicate statistical significance (p<0.05). Table 2Interrupted time series with mixed effects model across Rhode Island stateTermEstimateStd. ErrorStatisticDfP value95% CI low95% CI highWhite reference (fixed effects) Intercept74.1244.62716.02078.97 <0.001 64.91483.334 t (pre-policy slope)−3.2501.430−2.272148.43 0.025 −6.076−0.423 post_t1 (level shift at T1)0.3555.5530.064148.340.949−10.61911.328 time_after_t1 (slope after T1)2.4111.8381.311148.370.192−1.2226.043 post_t2 (level shift at T2)2.2336.0450.369148.280.712−9.71314.178 time_after_t2 (slope after T2)0.2692.4490.110148.280.913−4.5715.109Black vs white interactions raceBlack9.6983.8972.488150.84 0.014 1.99817.398 post_t1×Black29.1366.3234.608149.61 <0.001 16.64241.629 time_after_t1×Black−6.9751.703−4.095148.95 <0.001 -10.241−3.610 post_t2×Black19.7928.9292.217149.10 0.028 2.14937.436 time_after_t2×Black8.4003.5772.349148.34 0.020 1.33315.468Hispanic vs white interactions raceHispanic−11.1764.007−2.789151.43 0.006 −19.094−3.258 post_t1×Hispanic6.0956.9380.879148.280.381−7.61519.805 time_after_t1×Hispanic−2.1401.885−1.135148.280.258−5.8661.585 post_t2×Hispanic5.8569.8720.593148.280.554−13.65125.364 time_after_t2×Hispanic0.3003.9990.075148.280.940−7.6038.203Random effects SD (county intercept)3.735—————— SD (residual)10.800——————Interrupted time series with mixed effects model demonstrating trends in stroke mortality among White, Black and Hispanic populations in Rhode Island. Time periods captured include secular trends before policy implementation (t), immediate change after initial Act implementation (post_t1), annual post-policy changes (time_after_t1), immediate change after field triage implementation (post_t2) and annual change after field triage implementation (time_after_t2). Bold values indicate statistical significance (p<0.05)Df, degrees of freedom. Interrupted time series with mixed effects model across Rhode Island state Interrupted time series with mixed effects model demonstrating trends in stroke mortality among White, Black and Hispanic populations in Rhode Island. Time periods captured include secular trends before policy implementation (t), immediate change after initial Act implementation (post_t1), annual post-policy changes (time_after_t1), immediate change after field triage implementation (post_t2) and annual change after field triage implementation (time_after_t2). Bold values indicate statistical significance (p<0.05) Df, degrees of freedom. In DiD analysis of both the pre-field and post-field triage cohorts, all racial groups demonstrated significant reductions in all stroke mortality without significant differences in rate of change among groups (table 1). The same was true when analysing change in stroke mortality after amendment passage in the post-policy cohort only (table 1). ITS revealed that the post-amendment period in the immediate and long term was not associated with changes in mortality among the White or Hispanic groups (table 2). Immediately after amendment implementation, the Black group demonstrated a 19.80 per 100 000 increase in mortality (95% CI 2.15 to 37.44; p=0.028), followed by an annual increase of 8.40 per 100 000 (95% CI 1.33 to 15.47; p=0.020). When analysing individual counties, only Washington county demonstrated an annual decrease in stroke mortality in the Black group (β=−20.48, 95% CI −29.08 to −11.89; p<0.001). All other counties demonstrated no change in mortality rate or an increase after amendment implementation (online supplemental table 1). The ITS model assumed parallel trends in mortality rate across time among all groups; Analysis of Variance (ANOVA) between this model not accounting for race-dependent secular trends and another model with such a parameter showed no difference in fit (table 3). Table 3Model comparison ANOVA likelihood ratio testModelnparAICBIClogLikDevianceχ²DfP valueSecular181310.9341367.272−637.4671274.934———Race×secular201313.6151376.213−636.8081273.6151.31820.517Likelihood ratio test comparing two interrupted time series mixed effects models. The secular model includes a single shared secular trend, while the race×secular model adds race-specific secular trend interaction terms. The non-significant χ²(2) = 1.318, p=0.517 indicates no meaningful improvement in model fit with the addition of race-specific secular trends, supporting use of the more parsimonious secular model and suggesting that racial/ethnic groups did not differ significantly in their underlying pre-policy mortality trajectories. Df, degrees of freedom.AIC, Akaike Information Criterion; ANOVA, Analysis of Variance; BIC, Bayesian Information Criterion; Df, Degrees of freedom. Model comparison ANOVA likelihood ratio test Likelihood ratio test comparing two interrupted time series mixed effects models. The secular model includes a single shared secular trend, while the race×secular model adds race-specific secular trend interaction terms. The non-significant χ²(2) = 1.318, p=0.517 indicates no meaningful improvement in model fit with the addition of race-specific secular trends, supporting use of the more parsimonious secular model and suggesting that racial/ethnic groups did not differ significantly in their underlying pre-policy mortality trajectories. Df, degrees of freedom. AIC, Akaike Information Criterion; ANOVA, Analysis of Variance; BIC, Bayesian Information Criterion; Df, Degrees of freedom. Discussion This study evaluated the impact of the Rhode Island Stroke Prevention and Treatment Act of 2009, the rise of mechanical thrombectomy, and the amendment of the Act in 2015 to reflect changes in the stroke care landscape on racial disparities in stroke mortality. Policy interventions associated with improved stroke outcomes include the creation of an SSOC task force, a statewide continuous quality improvement database, tiered stroke centre designation, EMS stroke assessment tools and standardised transport protocols.1 Difference-in-difference analysis suggested that, within the context of decreasing stroke mortality across the entire population, there was significant divergence in mortality trends between the White and Black groups, with the latter not observed to change significantly from across the study period. Subsequent interrupted time series regression including random-intercepts for county-level effects suggested that improvements resulted from progression of secular trends among the whole population; in contrast, policy implementation was not associated with significant changes among White or Hispanic groups but was associated with instanteous increases in stroke mortality and—summatively across both interventions—accelerations of stroke mortality among the Black population. The immediate increase in Black mortality after Act passage may be representative of improved data collection and reporting after the creation of the stroke registry. The gradual decrease in mortality in the period thereafter may speak to the benefits of EMS protocol revision and PSC designation. Despite this decline, the Black group remained the highest in all-stroke mortality. Current evidence suggests that while certification of various stroke centre types is associated with increased admission rates to certified centres, there may be no change in thrombolysis or thrombectomy rates for Black patients living near newly certified centres.12 In the post-field triage period, the Black group experienced increases in mortality. Direct transfer to thrombectomy-capable centres when LVO is suspected has been linked to shorter length of hospital stay and reduced mortality in both urban and rural settings.13 14 Despite this, studies have shown differences in EMS activation trends and prehospital stroke recognition between racial groups, possibly explaining the persistently high mortality rate among this group.15–17 The Hispanic group had significantly lower baseline mortality rates but did not demonstrate a significant trend throughout the study period. There is evidence that Hispanic mortality rates may be unreliably reported due to inconsistencies in origin reporting on documents such as death certificates.18 Since Hispanic heritage is not listed under the race category, there may be inconsistencies in reporting of ethnic background, leading to inaccurate mortality data. The lack of change in mortality over the study period may speak to factors faced by this community that are not directly addressed by policy. The literature demonstrates that primary language, socioeconomic status and health literacy strongly influence time to presentation, eligibility for reperfusion therapies, and functional recovery.19–21 Non-English-speaking patients may present with greater stroke severity and worse outcomes, likely due to barriers to effective communication in acute emergencies and reduced preventive care access.19 22 According to Migration Policy Institute, among the estimated 116 674 Spanish speakers above 18 years old in the state, 50.4% were proficient in English.23 Moreover, there are over 20 000 Rhode Island residents who speak African and/or Caribbean languages with upwards of 36% of this population identifying as having a non-proficient understanding of English. In Rhode Island, programmes such as Progreso Latino and Dorcas International enhance access to language education, employment opportunities and healthcare navigation, factors linked to improved health and stroke outcomes.24 25 Additionally, integrating community health workers and embedding culturally relevant education and support into care protocols may effectively bridge gaps, ensuring that healthcare is inclusive, accessible and tailored to the specific needs of all populations.26 27 When analysing individual counties, Providence County did not demonstrate a correlation between policy implementation and stroke mortality in any group. This county is the most racially diverse in the state and houses the state’s only CSC as well as 4 PSCs. Despite the increased proportion of racial minority representation in urban tracts with higher stroke centre density, there is evidence that time to treatment may be increased among racial minorities.6 17 28 It has been suggested that a reason for this could be lower utilisation of EMS, less accurate prehospital recognition of stroke by EMS and differences in the perception of stroke severity on evaluation by medical providers among this population.5 16 28 29 These findings underscore that improvements in system infrastructure alone are insufficient. Although increasing the number of designated stroke centres improves access, many racially diverse communities already live near such centres, suggesting that proximity may not equate to accessibility. Our results emphasise that stroke legislation may not address all social determinants of health. Policymakers may consider prioritising diversity in healthcare leadership, equity-focused EMS training, collaboration with community health workers and funding for outreach initiatives that build trust and reduce barriers to care. Embedding equity metrics within SSOC evaluation frameworks may help ensure that legislative successes translate into equitable health outcomes across all racial and ethnic groups. Limitations A key limitation of this study is the absence of data for Asian and Native American populations, as the CDC Atlas of Heart Disease and Stroke does not report mortality rates for these groups in Rhode Island due to small population sizes within the state. The failure to observe associations between policy implementation and stroke mortality among the White and Hispanic subgroups might reflect a true lack of effect, inadequate power or imperfect correlation between changes in policy and changes in stroke practice. Stroke policy might lag and reflect changes perceived to be effective in some counties and lead those same changes in other counties. This omission underscores the need for more comprehensive data collection to understand how legislation impacts all racial and ethnic communities. In addition, county-level analysis may obscure within-county variation, warranting future analyses at the Zone Improvement Plan (ZIP) code or census tract level. Expanding metrics to include quality-of-life and functional recovery outcomes, as well as exploring interventional strategies such as mobile stroke units and telemedicine, could provide deeper insights into creating an equitable and effective stroke system of care. Future research on evaluating stroke interventions, such as thrombolysis and thrombectomy, while also examining biological, behavioural and social determinants of outcomes, may be indicated. The extent to which our findings are due to the designation of and updated triage protocols to CSCs independent of thrombectomy use remains unclear. Conclusions While the stroke mortality rates of the general population of Rhode Island decreased in the period between 2006 and 2019, this change was disproportionate among races, with the White group demonstrating greater decreases than the Black or Hispanic group. The introduction of CSC designation and field triage, the distribution of pre-hospital stroke protocols and the introduction of mechanical thrombectomy were associated with reductions in mortality across racial groups. However, when accounting for secular trends in mortality, policy did not have an independent significant effect on mortality among White and Hispanic groups. Persistently high mortality among patients of colour underscores the need for targeted interventions. Addressing structural inequities will require a multifaceted approach that incorporates culturally competent, evidence-based strategies tailored to the needs of diverse communities. This study provides a critical foundation for future research and policy development to promote equity in stroke care and outcomes. Supplementary material