The collection, classification, and reporting of adverse events in Multiple Sclerosis rehabilitation randomised trials: A systematic review short report

2.1 Patient and public involvement

Two people with MS (pwMS) provided insights and guidance on our review aims, objectives, and questions. Both PPI members made suggestions for disseminating research findings into MS communities, and examples of how future reviews could involve pwMS as members of the research team.

2.2 Systematic review methods

This systematic review followed PRISMA 2020 guidance⁹ and was prospectively registered on the Open Science Framework.¹ Eligible studies included RCTs, feasibility or pilot RCTs, and their associated protocols or registration records involving adults with MS. Interventions were required to align with the Cochrane Rehabilitation definition, indicating active participation in a therapeutic process.¹⁰ Publications were limited to 1 January 2022–16 January 2025 to reflect alignment with emerging AE reporting standards. Fifteen databases, trial registries, and grey literature platforms were searched 16th January 2025 using an information specialist supported search strategy. After deduplication, two reviewers independently screened titles, abstracts, and full texts, resolving disagreements through discussion with a third reviewer. Non-English papers were translated where necessary. Where more than one record was identified (e.g. trial registration, protocol and results paper), the data was collated into a single unique study record. In line with recent AE reporting guidance^7,8 items for data extraction focused on AE collection (approach, mode, personnel, timing, frameworks), classification (coding, causality, severity), and reporting (terminology, definitions, counts, group comparisons, statistical methods, display methods). Additional descriptive study characteristics were extracted to aid contextual interpretation of the included studies. A narrative synthesis was conducted. Meta analysis was not appropriate due to the qualitative nature of extracted data.

3.1 Study selection and characteristics

From 1224 records identified, 1029 were screened at title/abstract stage and 151 at full text. Fifty three unique studies met the inclusion criteria (See Figure 1, and Supplementary File for full reference list).

Figure 1. PRISMA flow diagram.

These studies represented at least 16 countries and involved a total of 4,903 participants of varying MS phenotypes (relapse remitting and progressive), ages (range of mean 27–62 years), and disability levels (range of mean Expanded Disability Status Scale scores 2.3 to 6.4). Interventions spanned multiple rehabilitation domains, including physical activity/exercise, virtual reality, behavioural change, psychological therapies, cognitive rehabilitation, hippotherapy and voice rehabilitation (See Supplementary File Table 1, with referencing of relevant studies).

3.2 Adverse event methodology

3.2.1 Collection of AEs (See Supplementary File Table 2, with referencing of relevant studies)

Across the 53 studies, reporting of AE collection was limited and varied considerably. Of the 24 (45%) studies describing their method for AE data collection, 11 different modes of AE data capture were identified—ranging from telephone calls, in person assessments, and diaries, to email, video consultations, and online platforms. These methods were described inconsistently. Only seven (13%) studies specified whether they used an open or closed approach to collecting AEs, and the majority provided no indication of the rationale that guided their choice of method(s). This lack of clarity made it difficult to determine whether AE data were gathered systematically or whether methods were appropriate to the characteristics of rehabilitation interventions.

Responsibility for collecting AE data also varied. Four (8%) studies relied primarily on participant -self-report, while seven (13%) used a shared approach between participants and study personnel. Many papers did not indicate who held primary responsibility, leaving uncertainty about the reliability of AE ascertainment, particularly in a population where fluctuating MS symptoms and cognitive challenges may affect recall. Timing of AE capture was similarly under-reported. A minority of studies specified whether AEs were documented at the point of occurrence or retrospectively (n = 14, 26%), with a mixed approach most often used (n = 7, 13%); for most studies, this information was absent. Only 2 (4%) studies referenced the use of structured AE collection frameworks, such as the Food and Drug Administration (FDA) FDA Adverse Event Reporting System (FAERS) or institutional harms policies. Of the 40 studies that included a follow-up- period, only five (9%) explicitly stated that AE monitoring would be continued during those follow-up phases.

3.2.2 Classification of AEs (See Supplementary File Table 3, with referencing of relevant studies)

Classification of AEs was reported infrequently; only one (2%) study used a recognised coding framework (Common Terminology Criteria for Adverse Events), with seven (13%) others using a study-specific approach to their classification. Structured approaches to causality assessment were uncommon, appearing in only three (6%) studies; typically drawing on internal institutional checklists, the National Institute of Health (NIH) scheme, or semi structured qualitative judgement. Severity assessments were similarly inconsistent, with only seven (13%) studies reporting methods for investigating the severity of AEs. Of these, some studies referenced institutional criteria, or bespoke rating scales, while others used descriptive categories such as “minor” or “resolved” without clear methodological grounding. These inconsistent and often absent classification procedures limit the interpretability and comparability of AE findings across MS rehabilitation randomised trials.

3.2.3 Reporting of AEs (See Supplementary File Table 4, with referencing of relevant studies)

The reporting of AEs demonstrated substantial variability. Across the included studies, 17 different terms were used to refer to adverse events, and most studies (n = 45, 85%) did not provide a definition of the term used. “Adverse event” was the most frequently used term (by 41 studies, 77%); 22 studies (42%) applied multiple terms interchangeably when referring to AEs. The lack of definitional clarity created ambiguity regarding how study teams conceptualised and categorised safety related information.

Only a minority of studies reported individual AE counts (n = 8, 15%), wherein the specific instances of AEs are recorded and reported. Separate reporting for intervention and control groups was undertaken by only 13 (25%) studies making it difficult to interpret any potential between group differences in safety outcomes. Statistical analysis methods were described explicitly by only a few studies, with most relying implicitly on descriptive summaries without specifying how these were generated. Only three studies (6%) provided tabular or graphical summaries of the collected AE data. No studies presented time to event data (the length of time from a defined start point, such as treatment initiation), duration of AEs, or AE selection rules wherein criteria are used to determine which unfavourable, unintended, or harmful consequences will be reported. No study provided a lay summary of either the trial results or the AEs.

3.3 Limitations and suggestions identified by trialists

Only three studies (6%) explicitly reflected on limitations in their AE methodology. Two (4%) highlighted restricted AE collection windows, one including an inability to capture events occurring outside scheduled contact periods.^11,12 One (2%) reported not including a data safety monitoring board.^13,14 Suggested improvements included incorporating remote or outside of session AE collection methods,¹¹ extending monitoring periods,¹² and integrating qualitative enquiry to capture contextual information around adverse experiences.¹¹ These reflections underline the need for more comprehensive AE methodology in MS rehabilitation trials.