Предсказатели высоко- и низкоинтенсивного лечения склероза первой линии.

Introduction

Selecting an appropriate disease-modifying therapy (DMT) for patients newly diagnosed with multiple sclerosis (MS) is a nuanced decision. Current guidelines recommend two initial visits with the newly diagnosed patient, one to discuss the disease itself and how it will affect the patient, and a second to discuss the various medication options and issues considering delivery, efficacy, risks, cost, and nuances. 1 Since the introduction DMTs, the number of these agents has increased substantially to more than 20 options, further adding to the complexity of this process, 2 but factors predictive of DMT selection have not been comprehensively examined.

At present, there are two general treatment approaches. The first approach is a “low-intensity” or “escalation” approach of starting with a safer and generally lower efficacy drug (e.g., glatiramer acetate, interferon beta, teriflunomide, dimethyl fumarate), then escalate to a high-intensity medication if disease progresses. The second approach is a “high-intensity” or “induction” approach of starting with the highest efficacy drug (e.g., ocrelizumab, natalizumab, ofatumumab, alemtuzumab) early in the disease course. “Reconstitution” refers to a treatment regimen wherein a patient takes a high-intensity medication for a short period of time, with the goal of “restarting” or “rebooting” the immune system. 3 , 4

In theory, DMT selection should be most directly influenced by the individual MS case. Those with high number of T2 lesions, high T2 lesion volume and the presence of infratentorial and spinal cord lesions may be at higher risk of symptomatic progression or accumulation of disability in the future, potentially justifying an immediate jump to high-intensity therapies. 5 Likewise, younger patients have a longer time until immunosenescence, and these patients may be more likely to be started on high-intensity treatment. 6 This is further related to the patient's personal tolerance for risk, either due to the disease or the side-effects of its treatment. Some patients may simply prefer the peace of mind afforded by a low risk of relapse, while others may prefer a wait and watch approach to their care. 7 The treatment of primary progressive multiple sclerosis (PPMS) is a separate discussion, as only ocrelizumab has been approved in this population. 8

Logistical factors are also considered. Some medications, such as cladribine only need to be taken for two monthly cycles for two years, while other medications, such as dimethyl fumarate need to be taken twice per day orally. Some patients find it easier to adhere to a daily oral medication, while others find it easier to go to an infusion center several times per year. 9 Patients living in rural areas or with limited transportation may find that driving to an infusion center (in addition to appointments with their neurologist, pharmacist, physical therapist, urologist, dermatologist, psychiatrist, and more) places an excessive appointment burden that limits their quality of life. 9

The question of whether an escalation or induction approach leads to better long-term outcomes in MS remains unanswered. This critical issue is being addressed by two large volume, ongoing randomized controlled clinical trials: TREAT-MS (TRaditional versus Early Aggressive Therapy for MS) and DELIVER-MS (Determining the Effectiveness of earLy Intensive Versus Escalation approaches for the treatment of Relapsing-remitting MS). 10 TREAT-MS is evaluating time to sustained disability progression as its primary outcome, while DELIVER-MS focused on normalized whole brain volume loss as the main outcome. Findings from these trials are expected to provide important guidance on sequencing disease-modifying therapy (DMT) from the time of diagnosis and throughout the disease course.

In summary, the choice of DMT is a nuanced decision influenced by a wide variety of factors. But little is known about which factors are most important or the relative magnitude of these effects. This study aims to examine the association of patient- and provider-level factors with first-line high- versus low-intensity MS treatment strategies using a large commercial claims database.

Methods

Data source

We used the Merative MarketScan commercial and Medicare supplemental databases, nationally representative US claims databases, from 2012 to 2021 (date range available at our institution) that include privately insured individuals up to 64 years and older individuals enrolled in Medicare with an employer-sponsored Medigap plan and have very little missing data. All variables were defined using International Classification of Diseases, 9th and 10th revision (ICD-9 and ICD-10) codes, and National Drug Codes (NDCs).

Study population

The criteria for the study population were designed to capture incident MS patients and their first-line MS prescription. The study population criteria were: (1) outpatient MS diagnosis (ICD-9/-10: 340.xx, G35.xx) 11 ; (2) evidence of a DMT fill (natalizumab, alemtuzumab, ocrelizumab, rituximab, ofatumumab, cladribine, interferon beta, glatiramer acetate, teriflunomide, dimethyl fumarate, diroximel fumarate, fingolimod, ponesimod, siponimod, or ozanimod) fill following the first MS diagnosis; (3) no DMTs during 1 year prior to first MS diagnosis through first DMT fill (i.e., medication washout period); and (4) 6 months of continuous database enrollment prior to first DMT fill. The criteria for the study population were designed to capture newly diagnosed MS patients and their first-line MS prescription, and no subsequent initiation episodes.

Outcome

The first DMT filled during the study period was deemed first-line MS treatment, and its first fill date was designated as the index date. The first-line MS treatment was classified as either high-intensity (natalizumab, alemtuzumab, ocrelizumab, rituximab, ofatumumab, or cladribine) or low-intensity (interferon beta, glatiramer acetate, teriflunomide, dimethyl fumarate, diroximel fumarate, fingolimod, ponesimod, siponimod, or ozanimod); as designated by the ongoing TREAT-MS trial. 10 As many of the MS treatments were introduced during the study period, the percent of patients with high-intensity first-line MS treatment and each individual treatment was summarized yearly.

Patient and provider characteristics

Patient and provider variables were summarized for the total population and stratified by high- versus low-intensity first-line MS treatment. Patient characteristics included: age at index date, sex, U.S. region, rurality, high deductible health plan (HDHP), Medicare Advantage plan, baseline (6 months prior to index date): Charlson Comorbidity Index (CCI) score, transverse myelitis, back pain, inpatient visits, emergency department visits, outpatient visits, outpatient visits with a neurologist, number of brain MRIs (Appendix 1), index year, and number of steroid prescription fills. Additionally, to estimate baseline MS severity, we calculated a severity score described by Toliver et al. 12 Briefly, the following weights were assigned during the 6-month baseline period: MS hospitalization [3 points]; bladder/bowel/sexual dysfunction, cerebral/cognitive impairment, cerebellar symptoms (movement), pyramidal symptoms [2 points each]; brainstem symptoms, eye symptoms, sensory problems, cerebellar symptoms (walking/gait problems), fatigue/malaise, speech symptoms, durable medical equipment (DME) [cane, walker, hospital bed, or wheelchair] use [1 point each]. Patients with 0‒4 points were low severity, 5‒10 points were moderate severity, and >10 points were high-risk. The MS severity score was summarized continuously and categorically.

The nearest outpatient encounter with an MS diagnosis prior to the medication index date was used to identify the provider for the first-line DMT. Provider type was categorized as MS specialist (providers with ≥25 claims in the database with >50% having an MS diagnosis), neurologist (non-MS specialist) (Provider Type = Neurologist), or other/missing provider type. Because the data source does not contain provider information specific to MS fellowship/training completion, and NPIs are de-identified and therefore non-linkable to training or board certifications; we used a proxy definition to identify providers with a high proportion of visits with an MS diagnosis. We scanned the database internally to identify whether providers were “high-volume MS providers,” defined as providers that diagnosed MS in ≥80% of their patient visits, and that had ≥25 visits in the database. This definition is similar to the methods described by Terman et al. 13 and Gusovsky Chevalier et al. 14 to capture providers of other neurologic specialties.

Statistical analysis

We used logistic regression to examine patient and provider factors associated with high- versus low-intensity first-line MS treatments. Variables were selected by reviewing relevant literature 15 to identify factors that could be associated with the outcome, as well as exploratory hypothesized factors. 16 To account for multiple comparisons, p-values presented were adjusted using the Benjamini-Hochberg procedure to control the False Discovery Rate (FDR) at the 0.05 level. 17

To detect how much variance in the outcome was attributed to individual providers, we used a multilevel model with a provider-level random intercept using a restricted subject-level pseudo-likelihood approximation. 18 The intraclass correlation coefficient (ICC) for the predicted outcome probability was used to understand the provider-level variance in the outcome. To adjust for other patient characteristics, literature-derived and exploratory variables were selected as covariates. All available variables in the Merative MarketScan databases were considered for selection. For the multilevel modelling, we conducted a complete case analyses (patients with non-missing provider ID = 39%), as such patients that had missing data for provider of the first-line DMT were dropped for the models.

For both models, we assessed multicollinearity using a variance inflation factor (VIF) and considered VIF>5.0 a violation. To assess linearity of continuous variables, we created scatterplots of each continuous predictor against binned logits and visually examined linearity. Patients with unknown/missing U.S. region information (<2%) were dropped from all models. Odds ratios > 1.0 indicate association with high-intensity first-line DMT, and odds ratios < 1.0 indicate association with low-intensity first-line DMT. We defined statistical significance as p < .05 (two-tailed) and conducted analyses using SAS 9.4 (Cary, NC, USA).

Sensitivity analysis

There is no standardized classification of what constitutes high-intensity MS treatment, thus we opted to include a second definition of the outcome that classifies additional DMTs as high-intensity. In this sensitivity analysis definition of high-intensity first-line MS treatment, medications include those in the primary definition (natalizumab, alemtuzumab, ocrelizumab, rituximab, ofatumumab, and cladribine) with the addition of fingolimod, ponesimod, siponimod, and ozanimod (sphingosine-1-phosphate receptor modulators [S1Ps]). Classification of S1Ps remains heterogeneous across studies; while sometimes considered moderate- to high-efficacy therapies, they are often categorized as intermediate-efficacy agents relative to monoclonal antibody therapies, which consistently demonstrate greater reductions in relapse rates and disability progression. 19 , 20 The remaining MS treatments were deemed low-intensity. All analyses were conducted using this secondary definition of the outcome.

Standard protocol approvals, registrations, and patient consents

The Ohio State University institutional review board deemed this study exempt; therefore, patient consent was not needed.

Data availability

Merative™ MarketScan ® databases are commercially available and deidentified prior to analysis.

Results

After applying our study criteria (266,140 [100%] patients had an MS diagnosis, 101,145 [38%] of them had a DMT prescribed, 56,155 [21%] met the washout period criteria, and 16,405 [6%] met the continuous enrollment criteria), a total of 16,405 adult patients were included in the study population ( Figure 1 ), and they were 48 years old on average (SD = 12). The population was 75% female, 43% resided in the South U.S. region, 7% had a HDHP plan type, and on average had a CCI of 1 (SD = 2) ( Table 1 ).

Among the study population, 486 (3%) had a high-intensity and 15,919 (97%) had a low-intensity first-line DMT. Compared to low-intensity, the high-intensity first-line MS treatment group had a larger proportion of males (33% vs. 25%), Northeast-residing patients (27% vs. 18%), and paraplegia/hemiplegia (12% vs. 6%). MS severity score was 1.6 on average (SD = 1.8), with a range of 0–13, and 93% had low severity (0–4 points). Provider type of the first-line DMT prescription was most commonly a non-MS specialist (49%) or an MS specialist (36%), which did not vary between high- versus low-intensity first-line treatment groups ( Table 1 ). The most common first-line DMT was glatiramer acetate (37%), followed by dimethyl fumarate (25%), and interferon beta (18%) ( Table 2 ). The proportion of patients with a high-intensity first-line DMT rose from <1% in 2012 to 13% in 2021 ( Figure 2 ). As ocrelizumab, ponesimod, monomethyl fumarate, ozanimod, diroximel fumarate, cladribine, and siponimod were introduced in later years of the study period, their rates of use do not emerge until after the time of their introduction. The percent of patients with interferon as their first-line MS treatment began as the highest (47%) and dropped throughout the study period ending at only 4% in 2021. Fingolimod, teriflunomide, and dimethyl fumarate (after 2012) remained relatively stable throughout the study period ( Figure 3 ).

No covariates were excluded due to violation of the VIF threshold, and continuous variables were kept untransformed. Variables significantly associated with higher odds of high-intensity first-line MS therapy were (OR [95% CI], p -value): being male (1.36 [1.12, 1.67], p = . 0094), index year (1.53 [1.47, 1.59], p < . 0001). Variables significantly associated with lower odds of high intensity first-line MS therapy were (OR [95% CI], p -value): North Central versus Northeast region (0.31 [0.23, 0.43], p < . 0001), South versus Northeast region (0.63 [0.50, 0.79], p = . 0004), and West versus Northeast region (0.31 [0.22, 0.45], p < . 0001), rural versus urban (0.59 [0.42, 0.85], p = . 0121, unknown versus urban (0.55 [0.37, 0.83], p = . 01265), baseline emergency department visits (0.95 [0.92, 0.98], p = . 0077) ( Table 4 ). The model c-statistic was 0.81.

Among the study population, 61% had a missing value for provider ID, thus, the multilevel model included a sample of 6334. There were 3544 distinct providers, with a mean of 1.8 (SD = 10.0, median = 1.0) patients per provider. Baseline characteristics between high- versus low-intensity first-line MS therapy on this subgroup were similar to the larger study population. The multilevel model revealed an ICC for the random provider-level intercept was 0.17, indicating 17% of the total variance in high- versus low-intensity first-line MS therapy was attributable to a provider effect, with a c-statistic > 0.9. Characteristics in the multilevel model significantly associated with high- versus low-intensity first-line MS therapy were North Central versus Northeast region (0.33 [0.18, 0.6], p < . 001), West versus Northeast region (0.43 [0.24, 0.77], p < . 01) and index year (1.62 [1.50, 1.74], p < . 0001) ( Table 4 ).

Sensitivity analysis

When applying the sensitivity analysis definition of high-intensity DMTs to the study population, 1993 (12%) had a high-intensity and 14,412 (88%) had a low-intensity first-line medication, and patient characteristics were well-balanced (Appendix 2). The percent of patients with high-intensity first-line treatment using the sensitivity analysis definition was 6% in 2012 and rose to 28% in 2021 ( Figure 2 ). In the logistic regression model, index year was associated with higher odds of high-intensity fist-line therapy (1.20 [1.18, 1.23], p < . 0001). Variables associated with lower-odds of high intensity first-line therapy in the sensitivity analysis definition were: age (0.99 [0.99, >0.99], p = . 0025), and West versus Northeast region (0.65 [0.54, 0.77], p < . 0001), and rural versus urban (0.80 [0.68, 0.93], p = . 0249) (Appendix 3). In the multilevel model, index year (1.22 [1.18, 1.26], p < . 0001), West versus Northeast region (0.58 [0.44, 0.76], p < . 0001), and rural versus urban residence (0.75 [0.57, 0.98], p = . 036) were associated with the outcome, and the ICC was 0.09, indicating 9% total variance in the outcome attributable to a provider effect (Appendix 4).

Discussion

In this large, nationally representative cohort of patients with MS from 2012 to 2021, we observed that the vast majority initiated low-intensity DMT, with only 3% receiving a high-intensity agent as first-line treatment. Overall, the trend of high-intensity DMTs on first-line increased throughout the study period, and the pattern of this trend was aligned when using varying definitions of high-intensity DMTs. When excluding the S1Ps (Fingolimod, Ponesimod, Siponimod, and Ozanimod) as high-intensity (primary definition from TREAT-MS), the rate of first-line high-intensity DMTs was lower throughout the study period, beginning at 1% in 2012 and ending at 13% in 2021. When including the S1Ps in the definition of high-intensity DMTs (sensitivity analysis), the trend was similar but higher overall, beginning at 6% in 2012, and ending at 28% in 2021. Thus, the addition of S1Ps in the definition increases the rate of high-intensity DMT prescriptions by 14% in 2021. These findings reflect introduction of more high-intensity medications over time and evolving treatment practices in MS and suggest a gradual shift toward earlier use of higher efficacy therapies, possibly reflecting changes in provider behavior based on the increasingly accepted idea that inflammatory activity is a major driver of long-term disability. 21 This period also coincided with expansion of the therapeutic landscape, with the introduction of multiple high-efficacy DMTs beyond natalizumab, including alemtuzumab (2014), ocrelizumab (2017), cladribine (2019), and ofatumumab (2020). 22 The availability of these newer agents likely contributed to the observed increase in high-intensity therapy use. Despite this trend, the relatively low overall adoption of high-intensity therapies indicates that the escalation model continues to dominate routine U.S. clinical care, reflecting ongoing uncertainty about the long-term benefits and risks of early high-intensity therapy versus traditional escalation and likely uncertainty about which patients are most likely to benefit from high-intensity treatment. The association with more recent index year, observed uniformly in all models and sensitivity analyses, likely reflects both the availability of additional high-intensity agents and updated guidelines that increasingly support induction-based strategies. 23

We identified several factors associated with initiation of high-intensity first-line therapy. Male sex, residence in the Northeast region, and more recent index year were positively associated with initiation of high-intensity therapy. Conversely, patients residing in the North Central, South, or Western regions; those residing in rural areas; and those with higher emergency department utilization were less likely to receive high-intensity therapy. The stronger likelihood of males receiving high-intensity therapy is notable and may reflect clinician perceptions of a more aggressive disease course in males. 16 , 24 Significant regional differences were observed in both models, indicating region as an important predictor of first-line treatments irrespective of provider variance. Geographic disparities in DMT utilization and access to MS care have been described in prior work. 25 , 26 Further, there are known differences in access to neurologists by region. In areas with higher rates of Hispanic residents 26 (such as the South and West U.S. Regions), it is known that there is lower access to neurologist care; which may indicate our findings potentially could point to elucidation of ethnic disparities in MS care. However, this data source does not contain race or ethnicity information, and therefore more research should be done in databases containing this information to robustly explore this relationship. These findings underscore that system-level context plays a significant role in treatment selection. Notably, MS severity was not significantly associated with high-intensity first-line therapy, which may be attributable to low severity scores being dominant at the time of treatment initiation. We also found that higher numbers of ED visits were associated with lower odds of initiating high-intensity therapy, which is aligned with a prior case control study that found a lower rate of high-intensity DMTs in MS patients who visited the ED versus those who did not. 27 While ED visits may be a reflection of underlying disease severity and thus be anticipated to increase the use of high intensity DMTs, they may also be a reflection of access limitations or other health problems that lead to lower use of high-intensity DMTs. Notably, we did not find a significant association between health plan differences, or comorbidities and first-line MS treatment.

Provider-level effects also played an important role. Multilevel modeling showed that 17% of the variance in initiation of high-intensity therapy was attributable to the prescribing provider, underscoring that individual practice style, institutional culture, and comfort with risk-benefit trade-offs strongly influence prescribing patterns. 28 , 29 Additionally, when accounting for the provider-level random effect, ED visits was no longer statistically significant; potentially indicating that adjustment for the provider-level effect encompassed baseline utilization of healthcare. Such variability in providers may contribute to inequities in access to high-efficacy therapy and ultimately affect patient outcomes. Interestingly, while the provider-level effect we found was significant, the type of provider (neurologist vs. MS specialist vs. other) was not significantly associated with first-line DMT, potentially indicating individual provider behavior is more informative to DMT choice compared to provider specialty/training. More research should be done to understand provider beliefs and knowledge, as well as organizational factors that guide this therapeutic decision-making, as well as associated outcomes with differing treatment strategies. Although differences between general neurologists and MS specialists have been reported in prior studies, 30 our study did not find a significant difference. This may be partially explained by missing provider ID data in 61% of the sample, limiting precision in provider-level analyses.

Our findings align with prior observational and registry-based studies that have reported relatively low early adoption of high-efficacy therapies despite increasing evidence in their favor. 20 , 31 The persistence of an escalation-dominant approach underscores ongoing uncertainty about optimal DMT sequencing, which remains the focus of ongoing randomized controlled trials including TREAT-MS and DELIVER-MS.

Taken together, these findings emphasize that treatment decisions are shaped not only by clinical characteristics but also by provider practice style and health system context. At the same time, patient-level considerations such as comorbidities, contraindications, and patient preferences remain critical and may justify use of escalation in selected cases.

Limitations

Our study has several limitations. First, the MS severity score was derived from ICD diagnosis and procedure billing codes, and therefore is unlikely to fully capture clinical severity or disease activity and could lead to misclassification. Next, provider-level analyses were constrained by the high proportion of missing provider identifiers, which may have introduced bias or reduced precision. Next, we were unable to distinguish between MS phenotypes using ICD codes, potentially obscuring differences in treatment patterns across relapsing and progressive subtypes because our study was conducted prior to the introduction of MS phenotype ICD codes in October 2025. Next, our claims data source did not contain testing and results, and therefore presence of the JC virus was not detectable for adjustment. Next, Merative MarketScan databases do not contain any information on race, ethnicity, or socioeconomic status; and we were not able to explore these factors as a predictor of first-line MS treatment. More research should be carried out to examine race as a predictor for MS treatment selection, as racial disparities have been shown to influence treatment choice in MS. 32 , 33 Next, it is known that there are various treatment-related influences on DMT selection including route of administration, frequency of administration, DMT adherence rates, and cost. In the claims data source used, we do not have information on how these factors influenced treatment choice before the DMT prescription. Route, frequency, adherence rates, and cost of the first-line DMT are inherent characteristics of the outcome that we defined (high vs. low intensity DMT), and would require observation of these attributes after the occurrence of the outcome. Further, DMT route and frequency of administration are both highly aligned with the definition of our outcome which would result in perfect/near-perfect prediction which is known to lead to invalid model performance and data leakage issues. 34 For these reasons, we were unable to measure and include these pre-prescription, DMT-related influences on the outcome, which is a limitation of our analysis. There is an opportunity to examine how these pre-prescription, DMT-related influences are related to DMT selection in future research. Next, our study period ended in 2021, as this was the most recent year available at our institution, which does not permit a full examination of DMTs approved toward the end of our study period (ofatumumab, monomethyl fumarate, and ozanimod approved in 2020; ponesimod approved in 2021) or medications approved beyond 2021 (ublituximab). As more years of data become available, more research should be done to examine prescribing patterns of the newer medications. Next, the data source includes patients with private insurance plans, and the findings may not be generalizable to a publicly-insured population. The data source does not contain information on facility type (i.e., academic medical center vs. private practice), and therefore we were not able to examine whether facility type was a predictor of first-line MS treatment. Finally, claims-based data lack direct measures of clinical or radiologic disease activity, which limits granularity in assessing treatment drivers.

Strengths

This study also has notable strengths. We leveraged a large, nationally representative cohort spanning diverse U.S. practice settings, enhancing generalizability. The 10-year period enabled assessment of temporal trends and shifts in treatment paradigms. Use of multilevel modeling allowed us to quantify provider-level contributions to prescribing decisions, an underexplored dimension in prior large-scale MS research. Sensitivity analyses supported the robustness of results. Finally, by linking patient-, system-, and provider-level factors, we provide a comprehensive perspective on the drivers of therapeutic decision-making in MS care.

Conclusion

In this large, nationally representative cohort of patients with MS, initiation of high-intensity therapy as first-line treatment remained relatively uncommon, though use increased steadily over the past decade. Patient demographics, geographic region, and provider-level effects were major determinants of treatment choice, underscoring the influence of both clinical and non-clinical factors. These findings highlight the need for strategies to reduce unwarranted variation, promote equitable access to high-efficacy therapies, and support patient-centered, evidence-based decision-making. Results from ongoing randomized controlled trials will be critical in determining whether early initiation of high-intensity therapy improves long-term outcomes, and future work should examine how emerging evidence translates into real-world prescribing patterns.

Appendices

Appendix 1. CPT codes used to identify brain MRIs

Appendix 2. Sensitivity analysis characteristics

Appendix 3. Logistic regression model results predicting high- versus low-intensity first-line multiple sclerosis medication using sensitivity analysis definition

Appendix 4. Multilevel model results predicting high- versus low-intensity first-line disease-modifying therapy (DMT) with provider-level random effect using sensitivity analysis definition

Footnotes

ORCID iDs: Amanda V. Gusovsky Chevalier https://orcid.org/0000-0002-1380-4839

Tirisham V Gyang https://orcid.org/0000-0002-6936-9186

Funding: The authors received no financial support for the research, authorship, and/or publication of this article.

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: T. Gyang has received personal compensation for consulting with EMD Serono and Genentech.

References

1. Practice guideline recommendations summary: Disease-modifying therapies for adults with multiple sclerosis. Neurology. Accessed April 17, 2025. https://www.neurology.org/doi/10.1212/WNL.0000000000005347 .

2. Pardo G, Jones DE. The sequence of disease-modifying therapies in relapsing multiple sclerosis: safety and immunologic considerations. J Neurol 2017; 264: 2351–2374.

3. Karussis D, Petrou P. Immune reconstitution therapy (IRT) in multiple sclerosis: the rationale. Immunol Res 2018; 66: 642–648.

4. Multiple Sclerosis Therapy Consensus Group (MSTCG): position statement on disease-modifying therapies for multiple sclerosis (white paper) - Heinz Wiendl, Ralf Gold, Thomas Berger, Tobias Derfuss, Ralf Linker, Mathias Mäurer, Orhan Aktas, Karl Baum, Martin Berghoff, Stefan Bittner, Andrew Chan, Adam Czaplinski, Florian Deisenhammer, Franziska Di Pauli, Renaud Du Pasquier, Christian Enzinger, Elisabeth Fertl, Achim Gass, Klaus Gehring, Claudio Gobbi, Norbert Goebels, Michael Guger, Aiden Haghikia, Hans-Peter Hartung, Fedor Heidenreich, Olaf Hoffmann, Boris Kallmann, Christoph Kleinschnitz, Luisa Klotz, Verena I. Leussink, Fritz Leutmezer, Volker Limmroth, Jan D. Lünemann, Andreas Lutterotti, Sven G. Meuth, Uta Meyding-Lamadé, Michael Platten, Peter Rieckmann, Stephan Schmidt, Hayrettin Tumani, Frank Weber, Martin S. Weber, Uwe K. Zettl, Tjalf Ziemssen, Frauke Zipp, for the ‘Multiple Sclerosis Therapy Consensus Group’ (MSTCG), for the ‘Multiple Sclerosis Therapy Consensus Group’ (MSTCG), 2021. Accessed April 17, 2025. https://journals.sagepub.com/doi/full/10.1177/17562864211039648 .

5. Rotstein D, Montalban X. Reaching an evidence-based prognosis for personalized treatment of multiple sclerosis. Nat Rev Neurol 2019; 15: 287–300.

6. Tintore M, Rovira À, Río J, et al. Defining high, medium and low impact prognostic factors for developing multiple sclerosis. Brain 2015; 138: 1863–1874.

7. Wilson LS, Loucks A, Gipson G, et al. Patient preferences for attributes of multiple sclerosis disease-modifying therapies: development and results of a ratings-based conjoint analysis. Int J MS Care 2015; 17: 74–82.

8. OCREVUS® (ocrelizumab) Results for PPMS (Progressive MS) . ocrevus. Accessed April 17, 2025. https://www.ocrevus.com/patient/ocrevus-101/primary-progressive-multiple-sclerosis.html .

9. Mansfield C, Thomas N, Gebben D, et al. Preferences for multiple sclerosis treatments: using a discrete-choice experiment to examine differences across subgroups of US patients. Int J MS Care 2017; 19: 172–183.

10. Simpson A, Mowry EM, Newsome SD. Early aggressive treatment approaches for multiple sclerosis. Curr Treat Options Neurol 2021; 23: 19.

11. ICD-9 to ICD-10 Conversion Commonly Used Neurologic Diagnosis. American Academy of Neurology. Accessed April 29, 2024. https://www.aan.com/siteassets/home-page/tools-and-resources/practicing-neurologist–administrators/billing-and-coding/icd-10-cm/16icd10aancrosswalk_tr.pdf

12. Toliver JC, Barner JC, Lawson KA, et al. Use of a claims-based algorithm to estimate disease severity in the multiple sclerosis Medicare population. Mult Scler Relat Disord 2021; 49: 102741.

13. Terman SW, Niznik JD, Slinger G, et al. Incidence of and predictors for antiseizure medication gaps in Medicare beneficiaries with epilepsy: a retrospective cohort study. BMC Neurol 2022; 22: 328.

14. Gusovsky Chevalier AV, Lin CC, Kerber K, et al. Adherence to antiseizure medications around pregnancy. Neurol Open Access 2026; 2: e000070.

15. Prognosis of multiple sclerosis: clinical factors predicting the late evolution for an early treatment decision - PubMed. Accessed April 29, 2026. https://pubmed.ncbi.nlm.nih.gov/16533140/

16. Heeroma JH, Roelandse M, Wierda K, et al. Trophic support delays but does not prevent cell-intrinsic degeneration of neurons deficient for munc18-1. Eur J Neurosci 2004; 20: 623–634.

17. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B Methodol 1995; 57: 289–300.

18. DavidGarson G, Garson GD. Preparing to analyze multilevel data. In: Hierarchical linear modeling: guide and applications. Sage Publications, Inc., 2013, pp.27–54.

19. Samjoo IA, Drudge C, Walsh S, et al. Comparative efficacy of therapies for relapsing multiple sclerosis: a systematic review and network meta-analysis. J Comp Eff Res 2023; 12: e230016.

20. He A, Merkel B, Brown JWL, et al. Timing of high-efficacy therapy for multiple sclerosis: a retrospective observational cohort study. Lancet Neurol 2020; 19: 307–316.

21. Buron MD, Chalmer TA, Sellebjerg F, et al. Initial high-efficacy disease-modifying therapy in multiple sclerosis: a nationwide cohort study. Neurology 2020; 95: e1041–e1051.

22. Goldschmidt CH, Hua LH. Re-evaluating the use of IFN-β and relapsing multiple sclerosis: safety, efficacy and place in therapy. Degener Neurol Neuromuscul Dis 2020; 10: 29–38.

23. Hartung HP, Meuth SG, Thompson AJ. Paradigm shifts: early initiation of high-efficacy disease-modifying treatment in multiple sclerosis. Mult Scler J 2021; 27: 1473–1476.

24. Voskuhl RR, Patel K, Paul F, et al. Sex differences in brain atrophy in multiple sclerosis. Biol Sex Differ 2020; 11: 49.

25. N’Dri LA, Waters DD, Walsh K, et al. System-level variation in multiple sclerosis disease-modifying therapy utilization: Findings from the multiple sclerosis continuous quality improvement research collaborative. Perm J 2021; 25: 39–45.

26. McGinley MP, Harvey T, Lopez R, et al. Geographic disparities in access to neurologists and multiple sclerosis care in the United States. Neurology 2024; 102: e207916.

27. Satish S, Patel A, Mastick ML, et al. Multiple sclerosis in the emergency department. Neurol Neuroimmunol Neuroinflammation 2025; 12: e200445.

28. Kumar J, Cambron-Mellott MJ, Tencer T, et al. Patient and neurologist preferences in the United States for relapsing-remitting multiple sclerosis treatments: findings from a discrete choice experiment. Patient Prefer Adherence 2021; 15: 1515–1527.

29. Didierjean J, Kremer L, Lanotte L, et al. Evolution of high-efficacy treatment strategy for relapsing-remitting multiple sclerosis: a real-life observational study. Mult Scler Relat Disord 2025; 104: 106690.

30. Kürtüncü M, Tuncer A, Uygunoğlu U, et al. Differences between general neurologists and multiple sclerosis specialists in the management of multiple sclerosis patients: a national survey. Arch Neuropsychiatry 2017; 56: 269–272.

31. Prosperini L, Mancinelli CR, Solaro CM, et al. Induction versus escalation in multiple sclerosis: a 10-year real world study. Neurother J Am Soc Exp Neurother 2020; 17: 994–1004.

32. Langer-Gould A, Li BH, Smith JB, et al. Racial inequities, multiple sclerosis, and implementation of a novel treatment algorithm at the health system level. Neurology 2025; 104: e213607.

33. Geiger CK, Sheinson D, To TM, et al. Treatment patterns by race and ethnicity in newly diagnosed persons with multiple sclerosis. Drugs Real World Outcomes 2023; 10: 565–575.

34. Kaufman S, Rosset S, Perlich C, et al. Leakage in data mining: formulation, detection, and avoidance. ACM Trans Knowl Discov Data 2012; 6: 15:1–15:21.

Associated Data

Data Availability Statement

Merative™ MarketScan ® databases are commercially available and deidentified prior to analysis.