Расовая неравноправие, множественный склероз и внедрение нового лечения алгоритма на уровне системы здравоохранения.

Introduction

Decades of advances in multiple sclerosis (MS) have not translated into equitable improvements in MS treatment or outcomes among minoritized people residing in the United States. Modestly effective disease-modifying therapies (meDMTs) have been available since 1993, and by 2010, the evidence basis for 4 highly efficacious treatments (HETs) was established (natalizumab, 1 , 2 alemtuzumab, 3 fingolimod, 4 , 5 and rituximab 6 , 7 ). Yet, studies continue to report that Black and Hispanic people with MS (pwMS) have higher levels of disability and acquire disability at younger ages than White pwMS. 8 , 9 Young Black people are also more likely to die from MS than their White peers. 9 Although there are currently 12 available HETs, 10 they continue to be underprescribed, particularly among Black and Hispanic pwMS compared with their White counterparts. 11

In 2012, Kaiser Permanente Southern California (KPSC)—one of the largest health maintenance organizations in the United States with over 4.7 million members—launched the Multiple Sclerosis Treatment Optimization Program (MSTOP). 12 MSTOP is a novel, multifaceted health systems–level intervention designed to increase the use of HETs, mostly rituximab, while simultaneously improving access to and affordability of MS specialty care. Through the implementation of a treatment algorithm, MSTOP increased the use of HETs to 62.5% among DMT-treated pwMS in 2018, which corresponded to a 69% decrease in the annual relapse rate. 13 The treatment algorithm 14 was designed to increase the use of HETs because internal analyses during the preimplementation phase showed that disability outcomes 15 in newly diagnosed pwMS were no different compared with historical untreated controls 16 despite prompt diagnosis and initiation of meDMTs. Even among the 41.3% who developed moderate or worse MS-related disability, only 5.6% were escalated to an HET before the onset of disability. 15

Implementation of the treatment algorithm was accomplished by (1) achieving consensus around MSTOP's programmatic goals from key interested, affected or relevant parties in the health system; (2) providing neurology practice leaders with quarterly audit and feedback of the proportion of DMT-treated pwMS on an HET stratified by each medical center; (3) negotiating an initial target that 60% of DMT-treated patients should be on an HET (the most conservative estimate of relapsing-onset pwMS with poor long-term outcomes); (4) educating general neurologists about MS prognosis, the relative efficacy of MS disease-modifying therapies (DMTs), and health systems science (e.g., how expensive, low-value treatments get paid for and the ensuing societal ramifications); and (5) creating a multidisciplinary MS care team. 12 , 13

Although the treatment algorithm is intended to increase HET use such that there is equity in utilization and outcomes, a person's race and ethnicity are not considered in the algorithm or in the quarterly audit and feedback of the proportion of DMT-treated pwMS on an HET. The treatment algorithm does not consider race or ethnicity explicitly because: there is no evidence that race and ethnicity are predictors of poor prognosis independent of health care quality; 9 such algorithms can increase risk of unconscious bias; and there is growing evidence for the harm of racialized medicine. 17 Instead, the algorithm incorporates attention to barriers to care that might be driven by structural racism and more frequently affect Black and Hispanic patients. These include unaffordable out-of-pocket costs, wage reductions, transportation barriers, inadequate childcare, psychosocial stressors, and work schedules that make frequent infusions, daily pills, or regular injections unrealistic, as well as patient preference. 14 There is evidence that nonrace-conscious interventions can result in unequal harms on the basis of race among minoritized people. 18 However, we hypothesized that the MSTOP algorithmic approach would benefit Black, Hispanic, and White pwMS equally (i.e., avoid unintended racialized harm) for several reasons: (1) algorithmic approaches to treatment have been shown to reduce racial and ethnic disparities in other conditions 19 , 20 ; (2) focusing on an individual's disease characteristics should lead to reduced unconscious bias; and (3) incorporating attention to care barriers that arise from structural racism and often emerge during shared decision-making discussions (e.g., unaffordable co-pays, challenges with transportation, work, psychosocial stress, and caregiving) would result in real-world effectiveness that approximates the efficacy of DMTs demonstrated in highly selected clinical trial populations.

The purposes of the analyses presented herein were to determine whether implementation of this treatment algorithm led to similar increases in HET use and decreases in relapse rates in Black, Hispanic, and White DMT-treated pwMS.

Methods

Study Design

We conducted a trend study of KPSC members with MS identifying as Black, Hispanic, or White who were treated with a DMT and compared annual DMT utilizations and average annual relapse rates (ARRs) before (2009–2011) and during (2012–2023) implementation of the treatment algorithm across these groups (n = 6,119).

Standard Protocol Approvals, Registrations, and Patient Consents

The study protocol received previous approval by the KPSC Institutional Review Board (#5707). Informed consent was waived because there was no direct patient contact.

Setting

KPSC is a prepaid health care organization that provides comprehensive health care to >4.7 million members across 7 counties in Southern CA. KPSC uses a comprehensive electronic health record (EHR) system that includes all inpatient and outpatient encounters, laboratory and imaging tests, diagnoses, medications, and demographic characteristics. Approximately 25% of the general population in the geographic areas served belongs to health plan. The sociodemographic characteristics of KPSC members are representative of the underlying population including those with low income. 21 KPSC subcontracts with Medicaid and other state programs for indigent persons, and 19% have some form of subsidized insurance. Neurologic care is provided at all 15 medical centers within KPSC by >200 neurologists including rural areas with scarce health care resources.

Intervention

The treatment algorithm aligns a patient's underlying risk of disability with an appropriately efficacious DMT. 14 It calls for escalating to an HET in pwMS who have breakthrough disease activity on a modestly efficacious DMT (meDMT) and initiating an HET in pwMS who have spinal cord involvement or other poor prognostic indicators early in the disease course. 22 Social determinants of health (SDOH) that can affect adherence, potential for pregnancy, and patient preference are also incorporated. To aid implementation, a multidisciplinary MS specialist group that provides care at all 15 medical centers was created. This group includes physicians, nurses, and pharmacists and meets monthly to review progress and update protocols. Providing audit and feedback on relapse rates or differences in care across racial or ethnic groups has not been part of MSTOP because of constrained analytical resources.

The treatment algorithm was updated in early 2019. 14 Criteria for use of HETs as first-line agents were loosened as clinicians became familiar with how to minimize risk of infections on the most used HETs (natalizumab, rituximab, and fingolimod), and the option of induction with affordable B-cell–depleting HETs (e.g., rituximab) for intermediate-risk and low-risk strata was added as patients increasingly reported unaffordable out-of-pocket costs as a major barrier to adherence. The HET use target was reset to 70% of DMT-treated patients to account for these changes. During the COVID-19 pandemic (2020–2022), rituximab-dosing protocols were updated to minimize risk of severe COVID-19 23 but the overall treatment algorithm was not altered. Additional details of the design and implementation of the treatment algorithm and the 4-part MSTOP intervention are provided elsewhere. 12 - 14

Race and Ethnicity

Race and ethnicity were obtained from health plan records, which capture self-reported race or ethnicity, and, if missing, EHR review. 24 In measuring race and ethnicity, we assume that they are both an individual sociocultural identity and a marker of a matrix of societal, historical, and political experiences that converge to create what we observe empirically as racial and ethnic inequities, including structural racism. 25 We categorized race and ethnicity as follows: (1) White, defined as non-Hispanic White people including persons from the Middle East; (2) Hispanic, defined as persons from Spanish-speaking countries, regardless of race; (3) Black, persons who identify as Black, non-Hispanic; (4) Asian or Pacific Islander (Asian/PI); (5) other (Native American or Alaskan Native) or multiple races (e.g., Black and Asian); and (6) unknown because of missing information. For the purposes of these analyses, we excluded patients who identified as Asian/PI (n = 165), other or multiracial (n = 107), Native American or Alaskan Native (n = 13), or unknown (n = 173), in total, 6.96% of DMT-treated pwMS, during the study period because these small sample sizes preclude meaningful interpretation.

Main Quality Metric

The main process metric intended to reflect quality of care was the annual proportion of Black, Hispanic, or White pwMS treated with an HET among those treated with any DMT. 12 HETs were defined as natalizumab, rituximab, or other B-cell–depleting therapies, and fingolimod or other sphingosine-1 phosphate receptor modulators.

Main MS outcome measure was defined as ARR among members treated with any DMT. Relapse episodes were identified from the EHR as previously described. 13 A relapse was defined as an inpatient admission with MS as the primary discharge diagnosis or receiving 3–5 days of IV methylprednisolone. We did not include outpatient oral prednisone use because chart reviews throughout the study period showed that these were prescribed primarily for other indications (e.g., allergies) or pseudorelapses. We conducted sensitivity analyses that included at least 3 consecutive days of 1000 mg of oral prednisone or higher to account for a possible shift away from IV to high-dose oral corticosteroids during the COVID-19 pandemic and considered repeated corticosteroid treatments within 30 days as 1 relapse.

Statistical analyses are as previously described. 13 We included all KPSC members who were treated with any MS DMT and had at least 6 consecutive months of membership after starting a DMT during the study period. DMT use was defined based on the onset of action to eliminate transient DMT users in whom no effect on ARR would be expected (eMethods). Person-years (PY) contributions began at the start of the first DMT during the study period and ended at membership termination, death, or DMT discontinuation.

ARR was calculated by dividing annual relapse counts by annual PY. We used linear spline regression, crude and adjusted for age and sex, during implementation (2011–2023) compared with zero percent change to determine whether increased HET use corresponded to decreased ARR. All analyses were stratified by race and ethnicity. Overlapping 95% CIs (Poisson distribution) and similar ARRs that did not exceed a clinically meaningful difference of 30 per 1,000 PY across groups were interpreted as similar effectiveness. For the outcome percent HET, we interpreted a difference not exceeding 5% and overlapping 95% CI as an indicator of MSTOP's similar effectiveness across racial and ethnic groups. Analyses were conducted using SAS statistical software, version 9.4 (Cary, NC).

Data Availability

Anonymized data are available on reasonable request.

Results

Patient Characteristics

We identified 978 Black, 1741 Hispanic, and 3,400 White pwMS treated with a DMT between 2009 and 2023. The characteristics of patients treated during each study year are presented in the Table . The average age was highest among White patients and lowest among Hispanic DMT-treated patients throughout the study period. Black DMT-treated pwMS were more likely to be female compared with Hispanic and White pwMS.

Table Characteristics of Black, Hispanic, and White Patients With Multiple Sclerosis Treated With Disease-Modifying Therapies per Year 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 Black  PY 288.1 334.9 375.8 414.7 455.1 471.0 504.1 531.4 548.3 567.5 612.3 638.3 662.4 681.2 685.4  Age, mean (SD), y 48.1 (10.8) 48.2 (11.0) 48.7 (11.4) 48.7 (12.1) 49.4 (12.0) 49.8 (12.4) 50.2 (12.4) 50.4 (12.8) 50.7 (13.0) 51.3 (13.1) 51.1 (13.3) 51.3 (13.5) 51.7 (13.6) 52.2 (13.6) 52.4 (13.7)  Female, % 84.2 82.7 83.5 82.4 82.6 81.8 81.4 80.8 80.8 80.4 80.1 80.3 79.9 80.3 80.8 Hispanic  PY 343.5 400.5 452.8 522.0 574.9 644.5 695.8 748.4 804.0 856.3 972.6 1,075.2 1,157.0 1,217.4 1,266.1  Age, mean (SD), y 42.0 (11.8) 42.7 (11.9) 42.4 (12.3) 42.6 (12.5) 42.5 (12.6) 42.8 (12.6) 43.2 (12.7) 43.1 (12.9) 43.3 (12.9) 43.4 (13.0) 43.4 (12.9) 43.4 (12.9) 43.4 (12.9) 43.8 (13) 44.1 (13)  Female, % 75.4 76.6 76.2 73.8 74.5 74.3 74.4 74.0 73.6 73.7 73.4 73.6 73.1 72.9 73.1 White  PY 1,204.7 1,367.2 1,490.9 1,606.0 1715.1 1794.0 1862.7 1913.9 1960.1 1983.6 2060.2 2,114.6 2,130.1 2,130.0 2096.9  Age, mean (SD), y 50.1 (11.2) 50.7 (11.3) 51.3 (11.5) 51.8 (11.7) 52.2 (11.9) 52.6 (12.0) 53.2 (12.1) 53.5 (12.5) 53.9 (12.6) 54.2 (12.8) 54.5 (12.9) 54.6 (13.2) 54.9 (13.3) 55.3 (13.5) 55.8 (13.7)  Female, % 76.7 76.2 75.5 75.1 75.3 75.4 75.2 74.0 73.7 73.6 74.1 74.2 74.3 74.0 73.7 Abbreviations: PY = person-years; y = years.

Characteristics of Black, Hispanic, and White Patients With Multiple Sclerosis Treated With Disease-Modifying Therapies per Year

Abbreviations: PY = person-years; y = years.

Inequities in HET Use Before Treatment Algorithm Implementation (2009–2011)

Before treatment algorithm implementation, HET use was equally rare and rose only slightly across all groups ( Figure 1 ). The crude ARR increased significantly among Hispanic DMT-treated pwMS, rising from 142.6 per 1000 PY (95% CI 105.7–179.6) in 2009 to 184.7 (95% CI 146.7–222.8) in 2010, and peaked at 245.1 (95% CI 205.5–284.8) in 2011 while White DMT-treated pwMS experienced a clinically significant decline between 2009 (180.1, 95% CI 158.4–201.8 per 1000 PY) and 2011 (156.3, 95% CI 137.8–174.7 per 1000 PY; Figure 2A ). Yet, despite the significantly higher ARR among Hispanic compared with White DMT-treated pwMS ( Figure 2A ), HET use increased equally in these groups ( Figure 1A ). Among Black DMT-treated pwMS, the ARR was similarly high between 2009 (190.9, 95% CI 145.5–236.3) and 2011 (183.6, 95% CI 144.4–222.7 per 1000 PY). By 2011, the difference in ARR exceeded 30 per 1000 PY when comparing Hispanic with Black pwMS ( Figure 2C ) and in Black compared with White pwMS ( Figure 2B ), although neither finding reached statistical significance. Despite this higher ARR in Black compared with White DMT-treated pwMS, HET use increased equally across these groups ( Figure 1B ).

Treatment Algorithm Implementation and Highly Effective Disease-Modifying Therapy (DMT) Use in Black, Hispanic, and White DMT-Treated People With MS

Depicted is the annual proportion of patients with MS treated with highly effective DMTs (HETs) among people treated with any DMT with 95% CIs stratified by Black (black line), White (red line), and Hispanic (green line) people with MS (pwMS). Before implementation of the MS treatment algorithm (preimplementation, 2009–2011), HET use was similarly rare across groups. HET use increased markedly after initiation of MSTOP in 2012 in all 3 groups (D). The increase in HET use was most marked in Hispanic DMT-treated pwMS. Even during the early implementation period (2013–2016), HET treatment was significantly higher both clinically (greater than 5%) and statistically denoted by the nonoverlapping 95% CI among Hispanic compared with White pwMS, a trend that continued through 2023 (A). The increase in HET use was higher among Black compared with White DMT-treated pwMS exceeding 5% in 2014–2017 after which HET use was similar between both groups (B). HET use after implementation was slightly higher among Hispanic compared with Black pwMS, but this difference was not statistically significant and only exceeded 5% in 2016 and 2019 (C). HET = highly efficacious treatment; MS = multiple sclerosis.

Treatment Algorithm Implementation and MS Relapses in Black, Hispanic, and White People With MS Treated With Disease-Modifying Therapies (DMTs)

Depicted is the average annual relapse (ARR) rate per 1,000 person-years (PY) among people treated with DMTs with 95% CIs stratified by Black (black line), White (red line), and Hispanic (green line) people with MS (pwMS). This figure shows that ARR decreased markedly after the implementation of the MS treatment algorithm in 2012 among all 3 racial and ethnic groups (D) except for 2015, when there was transient but clinically (exceeding 30 per 1000 PY) and statistically significant (as indicated by nonoverlapping 95% CIs) increase in ARR among Black compared with White DMT-treated pwMS (B). After implementation, there was no significant difference in ARR between Hispanic and Black pwMS, except for 2016 (C). By contrast, before implementation (preimplementation, 2009–2011), Hispanic DMT-treated pwMS experienced a clinically significant increase in ARR (greater than 30 per 1000 PY) each year, a finding that also reached statistical significance when comparing 2009 with 2011 among Hispanic DMT-treated pwMS and also when comparing Hispanic with White pwMS (A). Among Black DMT-treated pwMS, the ARR remained high but stable before implementation (B) while among White DMT-treated pwMS, the ARR declined, a finding that was clinically but not statistically significant when comparing 2009 with 2011. The age-adjusted and sex-adjusted ARR declined most among Hispanic DMT-treated pwMS between 2011 and 2023 (90%, 95% CI 89%–91%), followed by 86% (95% CI 85%–87%) among White pwMS and 82% (95% CI 80%–84%) among Black DMT-treated pwMS (D). MS = multiple sclerosis.

Increased Utilization of HETs, Primarily Rituximab, After Treatment Algorithm Implementation

After initiation of treatment algorithm implementation in 2012, the increase in HET use among DMT-treated patients with MS was similar among Black and Hispanic pwMS and slightly lower in White pwMS ( Figure 1 ). HET use increased from 9.7% to 89.3% of all DMT use among Hispanic pwMS, 8.2%–87.4% in Black pwMS, and 8.5%–82.9% in White pwMS between 2011 and 2023. Rituximab and its biosimilars became the most used DMT in our health system and, by 2023, accounted for 95.2% of HET use among Hispanic and White pwMS and 95.8% among Black HET-treated pwMS. By 2023, 84.5%, 83.2%, and 78.1% of Hispanic, Black, and White DMT-treated pwMS were treated with rituximab or its biosimilars, respectively ( Figure 3 ). The use of US Food and Drug Administration (FDA)-approved HETs (i.e., not rituximab or its biosimilars) was low and did not differ between Hispanic, Black, or White HET-treated pwMS (4.8%, 4.2%, and 4.8% in 2023, respectively).

Rituximab Use in Black, Hispanic, and White People Treated With MS Disease-Modifying Therapies (DMTs)

Depicted is the annual proportion of patients with MS treated with the highly effective DMTs, rituximab or its biosimilar, among persons treated with any DMT with 95% CIs stratified by Black (black line), White (red line), and Hispanic (green line) people with MS. This figure shows that rituximab use increased markedly after implementation of the MS treatment algorithm across all groups. The increase in rituximab was most marked in Hispanic DMT-treated pwMS and similarly high among Black and lowest among White DMT-treated pwMS. Rituximab use accounted for most of the HET use across all 3 racial and ethnic groups by 2015 ( Figure 1D ), predating the availability of the first FDA-approved B-cell–depleting therapy in 2018. HET = highly efficacious treatment; MS = multiple sclerosis.

Decline in ARRs After Treatment Algorithm Implementation

After initiation of treatment algorithm implementation, the ARR declined in all racial and ethnic groups, except from 2014 to 2015 where there was a transient but significant increase in ARRs among Black pwMS from 129.5 (95% CI 99.2–159.8) to 170.6 (95% CI 137.8–203.4) per 1000 PY. During 2015, the crude ARR was significantly higher among Black DMT-treated pwMS compared with that of White pwMS (110.6, 95% CI 96.3–124.8) by both criteria and exceeded a 30-point difference compared with that of Hispanic pwMS (145.1, 95% CI 119.0–171.3), although the 95% CIs overlapped. This change example and other years where the differences in ARR exceeded 30 points (2015 Hispanic higher than White pwMS; 2016 White higher than Hispanic pwMS; 2016 Black higher than Hispanic pwMS) were not sustained, and overall, the ARR continued to decline.

The age-adjusted and sex-adjusted ARR declined by 90% (95% CI 89%–91%) among Hispanic DMT-treated pwMS, 86% (95% CI 85%–87%) among White pwMS, and 82% (95% CI 80%–84%) among Black pwMS between 2011 and 2023 ( Figure 2 ). While markedly lower than before MSTOP initiation, the crude ARR was still slightly higher among Hispanic DMT-treated pwMS at 35.5 (95% CI 25.2–45.9) per 1000 PY compared with 19.0 (95% CI 8.7–29.3) among Black pwMS and 18.1 (95% CI 12.4–23.9) among White DMT-treated pwMS in 2023, although this did not meet our threshold for a clinically significant difference. Sensitivity analyses that included high-dose oral corticosteroids into the definition of relapse did not significantly alter these findings (data not shown).

Discussion

The past decade has brought about a seismic shift in thinking about racial and ethnic disparities in health outcomes from attributable to individual behaviors or immutable factors to recognizing the profound impact racism has on SDOH including socioeconomic status (SES), health care access, and quality. 26 , 27 The field of implementation science, which concerns itself with how to translate evidence-based interventions into real-world settings, is no exception. It has long been lamented how slowly scientific progress is translated into clinical practice. 28 Now, there is also an unequivocal recognition that even those scientific advances that are translated into clinical practice, such as tissue plasminogen activator (tPA) for stroke or DMTs for MS, do not lead to equitable improvements in health. 29

That evidence-based interventions, like new, highly efficacious treatments, are implemented inequitably is not surprising because they are usually extremely costly (e.g., MS HETs) and/or require resource-rich clinical settings (e.g., tPA for stroke). Clinics and hospitals in minoritized communities are notoriously under-resourced, and minoritized people in the United States are less likely to have adequate health insurance and be able to afford high out-of-pocket costs while more likely to experience racism and discrimination in medicine. 30 , 31 For these reasons, there is increasing concern that new, costly treatments—like the FDA-approved B-cell–depleting DMTs for MS (e.g., ocrelizumab, ofatumumab, and ublituximab)—could lead to widening health disparities. 32

Yet, within this context, we were able to achieve not only rapid but also equitable implementation of HETs for MS. Implementation of our novel treatment algorithm led to large and sustained increases in the use of HETs and similarly marked and sustained declines in relapse rates among Hispanic, Black, and White DMT-treated pwMS. Twelve years after implementation began, HET use accounts for 83%–90% of all DMT use and relapses among DMT-treated patients have become rare, even among Black and Hispanic pwMS.

These findings contrast with a commercially insured US population 11 and our preimplementation period where Black and Hispanic pwMS received lower quality of care compared with White pwMS. In the commercially insured population, overall use of HETs still remains under 60% and Hispanic and Black people were less likely to be treated with an HET than their White peers (20%, 30%, and 39% of DMT-treated pwMS, respectively). 11 During our preimplementation period, there was a mismatch in the relapse rates and HET use across groups. Although the relapse rates increased in Hispanic pwMS and remained stubbornly high among Black pwMS, these groups were no more likely to be treated with HETs compared with White pwMS in whom relapse rates declined. While care was equal, it was inequitable.

Our intervention may have been successful in increasing HET use because the treatment algorithm is concrete and relatively simple, relying on clinical and MRI characteristics that are readily available to clinicians. 14 We believe that it was successful in achieving not only equality but also equity because it incorporates SDOH that adversely affect adherence, including SES, and are rooted in structural racism, 32 yet it does not mention race or ethnicity, thereby reducing unconscious bias as has been shown with other algorithmic approaches. 19 , 20 Examples of other MS treatment algorithms are scarce, 33 - 35 lack detail, 34 , 35 or are overly complex, 33 and none that we are aware of incorporates SDOH or have demonstrated improvements in outcomes.

We believe that our implementation approach was successful and sustained because (1) it was implemented at the health system rather than individual level; (2) the focus was on quality rather than expenditures; (3) we chose a process metric (proportion of DMT-treated pwMS receiving an HET) that is an excellent proxy measure for MS relapses, is easy to measure, and for which there is clear accountability (i.e., the prescribing clinician); (4) regular audit and feedback of this process metric were provided to accountable clinicians and practice leaders; (5) we created a multidisciplinary MS specialist group to implement and update treatment algorithms and protocols; (6) the program was expert-led; and (7) we successfully negotiated with the health plan 36 by 2013 to allow access to rituximab and natalizumab as first-line treatments without requiring previous approvals.

This contrasts with how MS care is implemented in the United States where (1) there is no clearly defined goal or metric intended to reflect improvement in quality of care; (2) health systems are generally focused on expenditures rather than quality; (3) clinicians, even within the same practice, generally have not agreed on protocols to ensure that patients are provided with similarly high quality care regardless of which provider they see; (4) treatments patients have access to are determined by health plans, whose formularies are often opaque and developed without expert input; (5) access to rituximab and its biosimilars is not provided; and (6) there is no accountability for the quality of MS care patients receive. In the United States, onerous previous authorization requests, even for rituximab biosimilars, add another barrier to providing minoritized communities with highly effective and affordable therapies.

That the increase in use of HETs, particularly rituximab and its biosimilars, was highest in Hispanic, followed by Black and then White pwMS, is consistent with 24 epidemiologic studies and the recommendations of the algorithm. 14 MS disease activity is higher in younger pwMS, 37 , 38 Hispanic pwMS have a younger age at onset compared with Black and White pwMS, 24 and Black and Hispanic persons in the United States are more likely to experience adherence barriers because of SDOH and justified mistrust in health care institutions than White persons. Thus, our findings that Hispanic DMT-treated pwMS were younger and, when treated primarily with modestly effective DMTs, had higher relapse rates compared with White pwMS are novel yet not surprising.

The risk-stratified treatment algorithm recommends HETs in patients with the following characteristics: (1) highly active disease and/or spinal cord involvement at onset; (2) breakthrough disease activity on modestly effective DMTs; and (3) active relapsing MS but without these risk factors who have adherence challenges or are planning pregnancy. Driving this higher increase among Hispanic and Black pwMS is most likely a combination of higher relapse rate on modestly effective DMTs among Hispanic pwMS, a novel yet not surprising finding; more spinal cord involvement in Black pwMS 39 ; and significantly lower household incomes, fewer childcare resources, and other SDOH that pose greater adherence challenges among Hispanic and Black compared with White pwMS. 40

This study also adds to the growing literature supporting the real-word effectiveness of rituximab in treating relapsing MS. 41 , 42 Before MSTOP when HETs were rarely used in any racial or ethnic group, approximately 1 in 5 DMT-treated pwMS experienced a clinically significant relapse each year. By 2023, when 84.5% of Hispanic, 83.2% of Black, and 78.1% of White DMT-treated pwMS were on rituximab, only an estimated 1 in 35 patients experienced a clinically significant relapse. Rituximab has superior staying power (a combined measure of compliance, tolerability, and efficacy) compared with natalizumab and fingolimod 41 , 42 and does not result in rebound disease activity like natalizumab 43 and fingolimod if a patient loses health insurance, gets pregnant, or, for other reasons, must abruptly stop treatment. Rituximab has also demonstrated superiority to the commonly used modestly effective DMT, dimethyl fumarate, in a randomized controlled trial. 44 In patients with MS who do not require a walker or worse, rituximab is associated with fewer hospitalizations for infections (0.8 per 100 PY) 45 than currently the most commonly used B-cell–depleting therapy in the United States—ocrelizumab. Ocrelizumab in healthy, mildly affected newly diagnosed patients with relapsing MS resulted in a 2.1 per 100 PY risk of hospitalization for infections. 46 Rituximab and its biosimilars are also far less expensive (∼$2–10K per year) than ocrelizumab or other FDA-approved B-cell–depleting therapies (∼$60K–114K per year). Generic cladribine is another affordable alternative in selected settings, but we did not use it because its safety profile is less favorable than rituximab. 14

Despite MSTOP's success in reducing racial and ethnic disparities, some differences across groups remain. Consistent with the goals of providing equitable, not just equal, care, DMT-treated Hispanic pwMS had the greatest increase in HET use because they had the highest relapse rates and, correspondingly, the greatest decline in relapse rates compared with White and Black pwMS. Among Black DMT-treated pwMS, the story is a bit different. The relapse rate was somewhat higher than in White pwMS before and during the early phase of treatment algorithm implementation (2012–2016), and the increased HET use was greater than in White pwMS, although not as high as among Hispanic pwMS; yet, the overall decline in ARRs, although still remarkable, was lower in Black compared with White or Hispanic pwMS after treatment algorithm implementation. The reasons for this are unclear. One possibility is that the proportion of undertreated pwMS due to nonadherence or other factors was greater among Black pwMS. This would also explain the transient increase in ARRs among Black pwMS in 2015, a year during which there was a relatively large influx of members who previously could not afford or were denied health insurance because of a preexisting condition through the Affordable Care Act. These findings indicate that the drivers of disparities in MS treatment and outcomes operate somewhat differently in Hispanic and Black communities and eliminating these disparities will require additional interventions tailored to each community.

This study has important limitations. It is a trend study rather than a randomized trial. We chose this because the poor outcomes and low quality of care pwMS were receiving needed to be urgently solved. Thus, outside trends to increase the use of B-cell–depleting DMTs as we have been doing since 2013 may have accelerated rituximab uptake after 2018 but do not account for equitable uptake. 11 We did not measure relapse rates among untreated pwMS. MSTOP focuses on HET use among DMT-treated pwMS because it is what clinicians can control—which DMT they prescribe in patients willing to take a DMT. This leaves open the possibility that more young Black or Hispanic pwMS are not receiving treatment and/or neurologic care altogether. We could not assess this because validated electronic algorithms to identify patients with MS rely heavily on DMT use and thus would miss patients who are not receiving DMTs and neurologic care. 47 Another limitation is that we cannot distinguish between relapsing and progressive forms of MS. It is possible that the lower relapse rates in White DMT-treated pwMS is explained by a higher proportion of patients with progressive MS receiving rituximab or other DMTs. This could explain the differences between Hispanic and White pwMS, as older age is a good proxy for progressive MS, but it would not explain the small differences in relapses rates between Black and White DMT-treated pwMS, where ages were similar. The precision of ARR estimates is another limitation, particularly in 2019 to 2023 when relapses were rare. However, the overall similarity in decline across racial/ethnic groups is unambiguous. The level of intervention (health system, not individuals) and time frame preclude the ability to report meaningful disability outcomes, which develop over decades. We also did not address the effects of neighborhood-level or individual-level SDOH including SES. Finally, it remains to be seen whether this program can be implemented in other US settings where MS care is incentivized through reimbursable value units, not quality or access. These important gaps in knowledge should be addressed in future studies.

This study underscores the importance of removing barriers to accessing rituximab biosimilars. Although rituximab has been listed as an essential medication for MS by the World Health Organization 48 and cost-effectiveness demonstrated by the Institute for Clinical and Economic Review (ICER), 10 access remains limited in most countries, with the laudable exception of Sweden. Barriers include government-sponsored and private health care companies' unwillingness to provide easy access, many clinicians unwillingness to prescribe, and the unusual silence of professional societies on rituximab for MS. 10 , 49 A US law (Section 351[a], Public Health Services Act) benefiting Roche/Genetech prevents others from applying for FDA approval for rituximab or its biosimilars in MS although the evidence to satisfy FDA approval criteria 50 has been met. These decisions are particularly troubling for minoritized groups that are less likely to be able to afford expensive medications and more likely to be priced out of the health care market altogether. To remove this critical barrier, clinicians should continue to advocate for rituximab biosimilars, Medicare should cover them, and professional societies should endorse their use for MS.

The findings presented herein that a health systems–level intervention led to a marked reduction in relapse rates in Black, White, and Hispanic pwMS by implementing an algorithm that increased HET use, primarily rituximab and its biosimilars, are unprecedented. It provides not only a roadmap for how to reduce inequities in MS treatment and outcomes but also a concrete example of how to translate scientific advances into large, real-world settings early and equitably.