Practice Points
- Although psychological therapies demonstrate benefits in people with multiple sclerosis (MS) across several domains, this study identifies potential mechanisms of action and characteristics of those likely to benefit.
- People with MS with an undergraduate degree or higher demonstrate greater benefit from psychological therapies.
- Addressing fatigue can improve the outcomes of psychological therapies among people with MS.
Multiple sclerosis (MS) is a chronic neurodegenerative disease characterized by inflammation and demyelination of the central nervous system.1 In addition to a diverse range of physical symptoms, psychological and psychosomatic symptoms are highly prevalent among people with MS. Current estimates suggest that 36.5% to 78.0% of all people with MS have fatigue, 26.3% to 35.1% have depression, and 15.2% to 31.0% have anxiety.2,3 These symptoms are commonly intertwined with pain, stress, sleep dysfunction, and other quality-of-life outcomes, which can have downstream effects on overall symptom burden and functional status.4,5 In this regard, the burden of these symptoms and the relationships between symptoms and outcomes highlight the importance of psychological therapies in the treatment and symptomatic management of MS.
Current treatment guidelines on mental health in people with MS suggest that evidence on individual and group-based psychological therapies is insufficient.6 Moreover, these guidelines avoid making targeted demographic-specific recommendations as to which people with MS may benefit or on the components of therapies that may be removed or enhanced to improve outcomes.6 Nevertheless, results from qualitative studies on the use of cognitive behavioral therapies (CBTs) among people with MS suggest a range of outcomes and optimization factors.7 Outcomes include changes in psychological status, social connection or sense of belonging, and lifestyle and behavior; intervention features such as instructor characteristics and group processes were suggested to be beneficial.7 Given the growing body of evidence supporting the use of psychological therapies among people with MS, further quantitative investigation into the mechanisms of action and moderating factors is warranted.
Current research on factors that explain the mechanism of action, or mediating factors, and on factors that affect the direction and strength of the relationship, or moderating factors, specific to psychological therapies for people with MS is limited. However, numerous mediating and moderating factors explaining the relationship between psychological therapies and change in outcomes in the general population have been identified. In a recent review, the most supported change mechanisms for individual psychological therapy involved insight, affective awareness, and reflective functioning.8 In the context of mindfulness-based cognitive therapy and depression, cognitive factors such as mindfulness, rumination, and meta-awareness have been linked to treatment outcomes.9 Psychological factors, such as self-efficacy and anxiety, have also been shown to predict changes in panic symptoms.10 In terms of moderating factors, commonly supported modifiers include pretreatment symptom level, readiness for change, and assertiveness.8
The objective of this study is to examine mediating and moderating factors on the effect of psychological therapies in people with MS. Understanding these factors would allow for the identification of intervention components and participant demographics that may enhance the benefits of psychological therapies. In addition, this analysis will provide insight into the mechanisms of action among commonly used psychological therapies. Taken together, these results could be used to produce targeted treatment recommendations and inform the design of future interventions for people with MS.
Methods
Protocol and Registration
The data used in this study were collected under a protocol that was prospectively registered with PROSPERO, Centre for Reviews and Dissemination, University of York: CRD42023435206.
Information Sources and Search Strategy
A comprehensive search strategy was applied across 5 major electronic databases (Cumulative Index to Nursing and Allied Health Literature [EBSCO], Cochrane Central Register of Controlled Trials [Ovid], EMBASE [Ovid], MEDLINE ALL [Ovid], and the American Psychological Association’s PsycInfo [Ovid]). The search was performed in June 2023, with filtering criteria to select studies published after 1983, in English, involving human participants, and with a randomized controlled trial (RCT) study design. The full search strategy is outlined in Scandiffio et al.11
Selection Process
Search results were imported into Covidence for screening and data collection. Initial title and abstract screening was performed, followed by full-text screening to assess eligibility. Five independent reviewers contributed to the screening process (G.F., J.S., A.S., V.S., M. Bromberg). Each study was to be screened by any 2 of the 5 reviewers. Discrepancies were resolved through group discussion to reach a consensus.
RCTs examining the use of a psychological therapy (defined as activities aimed at changing behaviors, feelings, and emotional states) among people with MS were selected.12 All forms of psychological interventions were included (eg, CBT, mindfulness-based therapy), and all modes of delivery (ie, individual or group, in person or online) were included. Studies that did not involve live contemporaneous interactions with a trained therapist or facilitator and studies that did not report quantitative outcomes were excluded.
Data Extraction and Variables of Interest
Five independent reviewers engaged in data extraction (G.F., J.S., A.S., V.S., M. Bromberg). Data from each study was extracted by any 2 of the 5 reviewers using an extraction template guided by the Consolidated Standards of Reporting Trials and Template for Intervention Description and Replication checklists.13,14 Variables of interest pertained to the study design, participants, intervention, results, and limitations of studies. Authors were contacted via email when data were missing or not reported, and follow-ups were sent in nonresponse cases.
An a priori list of potentially mediating and moderating variables of interest was developed based on domain knowledge and a review of the current literature (Table S1, all supplemental materials are available in a PDF at the end of the online article). Potential mediators were mainly focused on patient-level factors related to psychology, physical symptoms, and quality of life. The final mediation analyses were performed examining affect, pain, stress, sleep, fatigue, cognition, and social functioning, as these variables were reported by a sufficient number of studies.
Potential moderators were focused on sociodemographic factors such as age, sex (or gender, if reported), and employment status. After screening for appropriate functional form, moderator analyses were performed examining age, sex, phenotype, disability score (eg, Expanded Disability Status Scale [EDSS]), and education level (ie, university degree or higher).
Mediator and Moderator Analyses
The data used in these analyses are a subset of the data used in a larger systematic review and network meta-analysis on the effects of psychological therapies in people with MS.11 Statistics were performed using Review Manager 5.4 (Cochrane), SAS 9.4 (SAS), and R version 4.3 (R Core Team) with the mediation package.15 Causal mediation analysis of the variables selected a priori was performed. The pooled mean change across multiple domains (physical, cognitive, psychological, quality of life, feasibility, and health economics) was used as the outcome in mediation models. Average causal mediation effect (ACME) estimates were used to examine the indirect effect of psychological therapies on outcomes through potential mediators, whereas average direct effect (ADE) estimates were used to examine the direct effect of psychological therapies on outcomes. The proportion mediated represented the percentage of the total effect on the outcome due to the mediator.
In moderator analyses, continuous variables (eg, years of age, EDSS score) were used as is, and the proportion of participants in levels of categorical variables was used (eg, proportion of female participants, proportion with an undergraduate degree). Continuous moderator variables were initially assessed graphically via functional form to determine any potential interactions. Interactions were then assessed by comparing slopes of pre-post outcomes to their slopes in the presence of a potential moderator. Moderator variables that were reported in less than 10% of all included studies were not assessed due to insufficient power and potential bias.
Results
Study Characteristics
Of 111 studies included in the full systematic review and network meta-analysis, 68 studies (61%) contained sufficient data for inclusion in mediation analyses and 65 studies (59%) for inclusion in moderator analyses. Among selected mediators, affect was reported in the largest number of studies (27 of 68 studies; 40%). Among selected moderators, EDSS was reported in the largest number of studies (38 of 65 studies; 58%). The full systematic review and network meta-analysis publication presents further details on study characteristics (eg, Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram) and risk of bias assessments.11 Details on included mediators and moderators are outlined in Tables 1 and 2, respectively.
Mediator and Moderator Analyses
Of the 29 potential mediators identified a priori, only 7 were reported sufficiently in studies. In the mediation analyses, the direct effects of psychological interventions on mean change across pooled domains were significant (P < .0001) for all mediators except for fatigue (Table 3). This suggests that psychological therapies have a strong and direct impact on outcomes, regardless of mediators. Direct positive effects on outcomes were particularly visible for depression and mood or affect (ADE, 8.21; 95% CI, 4.96-11.59; P < .0001), stress or anxiety (ADE, –11.51; 95% CI, –18.73 to –5.20; P < .0001), pain (ADE, 9.54; 95% CI, 4.73-14.18; P < .0001), and sleep (ADE, 13.72; 95% CI, 2.75-23.77; P < .0001), suggesting that these factors, although deriving positive effects, do not substantially mediate the effect of psychological therapies on outcomes.
Most of the factors included in this analysis displayed little to no evidence of mediation, as measured by indirect effects. Fatigue showed a borderline significant (P = .06) mediation effect (ACME, 4.44; 95% CI, 0.02-10.03), suggesting that fatigue may partially explain the impact of psychological therapies on outcomes (Figure 1). Cognition, stress or anxiety, and social functioning may have roles as mediators, as demonstrated by negative proportions mediated, although these proportions did not reach statistical significance. These results are further displayed in Figure S1.
Of the 9 potential moderators identified a priori, 4 variables (ethnicity, number of comorbidities, employment, medications) had an insufficient number of studies to plot an expected value line for inspection. In 4 included variables (age, sex, phenotype, disability score), the functional form of the variable and outcome was sufficiently flat to exclude interaction. There was a significant influence (P < .0001; 95% CI, 0.05-0.57) of the percentage of participants with an undergraduate education or higher on the expected value for the trial effect size of 0.27, indicating a positive interaction between education level and standardized mean difference. This suggests that those people with MS with a higher level of education may experience greater benefits from psychological therapies compared with those with a lower level of education. These results are further displayed in Figure S2.
Taken together, these results suggest that fatigue may have a mediating role in the mechanism between psychological therapies and outcomes. Moreover, participants’ level of education may have a moderating role, as people with MS with an undergraduate degree demonstrate greater benefit compared with those with a lower level of education. The relationship between mediator and moderator included in these analyses is depicted in Figure 2.
Discussion
This study aimed to explore mediators and moderators in the relationship between psychological therapies and outcomes among people with MS. Of the 111 studies included in the full systematic review and network meta-analysis, 68 studies contained sufficient data to perform mediation analyses on 7 variables, and 65 studies contained sufficient data to examine the moderating effects of 5 variables. Regardless of potential mediators, psychological therapies displayed strong and direct effects on outcomes. However, results from the mediation analyses suggest that fatigue may play a mediating role. Cognition, stress or anxiety, social functioning, and depression/mood or affect may also serve as mediators, although these coefficients showed lower evidence of statistical significance. Regarding moderators of outcome, level of education had a statistically significant effect, suggesting that those with an undergraduate degree or higher demonstrate greater benefit from psychological therapies.
These findings align with those of previous studies reporting on the mediating role of fatigue in MS-related outcomes. In a study examining a 12-week exercise intervention among people with MS, fatigue was shown to display a significant mediating role on outcomes such as depression, social activity, and quality of life. Additional research findings have also found fatigue to be a fully mediating variable in the relationship between nonsomatic aspects of depression (eg, mood, cognitions) and health-related quality of life among people with MS.97 Given that fatigue is one of the most prevalent symptoms among people with MS, with well-documented adverse effects on functioning, findings from this study further highlight the importance of fatigue management (eg, through CBT mechanisms) as a means to improve outcomes.98 Although findings from other studies on people with MS have found cognition and affect to be significant mediators in the context of CBT and physical activity interventions, our study was underpowered but still provides some support in favor of these findings.68,99,100
The role of education on outcomes has been well-documented in the literature. Among people with MS, a higher level of education is known to be a protective factor against cognitive decline and physical disability due to the development of a larger cognitive reserve.101-104 Nevertheless, people with MS with lower levels of cognitive impairment may be more likely to enter higher education in the first place. As people with MS with higher cognitive reserves tend to report greater levels of self-reported physical and mental health,104 they may have a greater capacity for change when engaging in psychological therapies. Similarly, some research data suggest that related concepts, such as lower socioeconomic status, can be associated with poorer psychological therapy outcomes.105 Taken together, existing psychological interventions may not be optimally designed to account for some of the documented social determinants of health in people with MS.106 Exploring a mix of formatting, componentry, and/or dose may be required to optimize outcomes among participants of diverse backgrounds.
To our knowledge, this is one of the first studies to synthesize data across the existing body of literature. Although various studies examining the impact of psychological therapies on people with MS have been conducted, few studies have examined the mediating and moderating factors of these interventions. This research begins to explore potential optimization factors that could be targeted to enhance outcomes. In addition, the potential mediators and moderators examined in this analysis were chosen a priori based on domain knowledge and the existing literature. This allowed relevant variables of interest to be examined, rather than erroneous variables identified based on data-driven processes. Nevertheless, this study also has limitations. In mediation analyses, pain, stress, fatigue, social functioning, and sleep had fewer than 10 included studies to draw from, likely leading to underpowered analyses. Moreover, 22 of the 29 mediators of interest were insufficiently reported in the existing literature and thus could not be examined. Future RCTs on psychological therapies in people with MS should consider the role of mediators known to drive beneficial outcomes, ultimately advancing the limited amount of literature on mechanisms of action in this area.
In the moderator analyses, ethnicity, number of comorbidities, employment status, and medication type were also insufficiently reported and could not be explored, thereby limiting the scope of our analyses. This highlights gaps in the existing body of literature on the social determinants of MS-related health outcomes (what works, for whom, under which circumstances), which matter when creating personalized care for people with MS. As a one-size-fits-all approach to psychological therapy may not be applicable for diverse people with MS, more systematic collection of these variables in future RCTs is warranted.7 In addition, heterogeneity in the use and reporting of measurement tools in the included studies may have affected the precision of our estimates (ie, due to differing psychometric properties of tools).
Further examination of the role of education as a moderator of psychological therapy outcomes among people with MS is needed, specifically in the ways in which psychological therapies may be designed to provide greater support for people with MS with a lower level of education. Further research into the mediating role of fatigue may also be warranted—for instance, ways in which fatigue management could be incorporated explicitly into psychological therapies for people with MS. Finally, greater reporting of sociodemographic, psychological, and clinical factors (eg, ethnicity, employment status, MS lesion burden, disease-modifying drug use, medication use, beliefs) is needed, as these may also play an important role in the outcomes of psychological interventions among people with MS.
Conclusions
Level of education moderates and addressing fatigue likely mediates the relationship between psychological therapies and outcomes among people with MS. Further research into how psychological interventions may be optimized to better support people with MS with a lower level of education is warranted. Greater reporting of patient sociodemographic, psychological, and clinical variables is needed to inform further investigation.