Remote Observational Research for Multiple Sclerosis: A Natural Experiment

Abstract

Background and objectives: Prospective, deeply phenotyped research cohorts monitoring individuals with chronic neurologic conditions, such as multiple sclerosis (MS), depend on continued participant engagement. The COVID-19 pandemic restricted in-clinic research activities, threatening this longitudinal engagement, but also forced adoption of televideo-enabled care. This offered a natural experiment in which to analyze key dimensions of remote research: (1) comparison of remote vs in-clinic visit costs from multiple perspectives and (2) comparison of the remote with in-clinic measures in cross-sectional and longitudinal disability evaluations.

Methods: Between March 2020 and December 2021, 207 MS cohort participants underwent hybrid in-clinic and virtual research visits; 96 contributed 100 "matched visits," that is, in-clinic (Neurostatus-Expanded Disability Status Scale [NS-EDSS]) and remote (televideo-enabled EDSS [tele-EDSS]; electronic patient-reported EDSS [ePR-EDSS]) evaluations. Clinical, demographic, and socioeconomic characteristics of participants were collected.

Results: The costs of remote visits were lower than in-clinic visits for research investigators (facilities, personnel, parking, participant compensation) but also for participants (travel, caregiver time) and carbon footprint (p < 0.05 for each). Median cohort EDSS was similar between the 3 modalities (NS-EDSS: 2, tele-EDSS: 1.5, ePR-EDSS: 2, range 0.6.5); the remote evaluations were each noninferior to the NS-EDSS within ±0.5 EDSS point (TOST for noninferiority, p < 0.01 for each). Furthermore, year to year, the % of participants with worsening/stable/improved EDSS scores was similar, whether each annual evaluation used NS-EDSS or whether it switched from NS-EDSS to tele-EDSS.

Discussion: Altogether, the current findings suggest that remote evaluations can reduce the costs of research participation for patients, while providing a reasonable evaluation of disability trajectory longitudinally. This could inform the design of remote research that is more inclusive of diverse participants.

Figure 1. Selected Sociodemographic Characteristics of the 207 Participants Seen Remotely During the Study Period

(A) One-way driving distance between the participant's zip code and the clinic. (B) Median household income in participants' zip code, according to 2021 California Housing and Community Development. (C) Distribution of demographic characteristics.

Figure 2. Overlay of California Resident Study Participants' Residence and COVID-19 Infection Rates in Their Communities (Infections per 100,000 People Ever, as of December 2021)

Panel A represents a map of all (n = 174) participants who were California residents. Panel B represents participants from the San Francisco Bay Area (n = 120). The purple circles represent individual participants in a given area code (larger circles = more participants). The hue of the areas corresponding to participants' zip codes represents the cumulative COVID-19 cases as of December 2021 in that specific zip code.

Figure 3. Mean and Median Values for All 3 Disability Evaluation Modalities (Neurostatus, ePR-EDSS and Tele-EDSS) Obtained From 100 Visits (96 Participants) With Multiple Sclerosis

Panel A describes the summary statistics, and Panel B visualizes these in a box plot. tele-EDSS = televideo-enabled Expanded Disability Status Scale.

Figure 4. Correlation Between Disability Scores Across the 3 Examination Modalities

As the disability level increases, the correlation becomes higher. tele-EDSS = televideo-enabled Expanded Disability Status Scale.

Figure 5. Distribution of Changes in Participants' Scores Between Annual Visits

Panel A represents annual changes in NS-EDSS for all visits from these participants, 2004–2018, and Panel B between 2019-2022. Panel C represents annual changes during the 2019–2022 epoch when switching from NS-EDSS at 1 time point to tele-EDSS at the next annual evaluation. Panel D represents annual changes during the 2019–2022 epoch when switching from NS-EDSS at 1 time point to ePR-EDSS at the next annual evaluation. For each comparison, the bar chart depicts absolute change and the pie chart represents the % of participants with EDSS stability, worsening or improvement between the annual visits as defined above, and using the 1-year NS-NS EDSS change from participants during the 2004–2018 timeframe as reference. Statistical equivalence was valid for 1-year NS-NS, NS-tele, and NS-PR 2018–2022 (p = 0.000006, 0.000017, 0.0196, respectively). ePR-EDSS = electronic patient-reported EDSS.