EFFECTS OF MEETING EXERCISE GUIDELINES ON DEPRESSION AND ANXIETY IN MULTIPLE SCLEROSIS – A SYSTEMATIC REVIEW AND META-ANALYSIS

Abstract

Background:

Depression and anxiety are commonin multiple sclerosis (MS), significantly affecting quality of life. Previous studies on exercise interventions for mental health in persons with MS (PwMS) have shown mixed results, partly due to inconsistent adherence to the Physical Activity Guidelines for Multiple Sclerosis (PAG-MS).

Objective:

This study aimed to evaluate the effects of PAG-MS-compliant exercise on depression and anxiety outcomes in PwMS.

Methods:

A systematic search was performed in EMBASE, PsycINFO, Web of Science, SPORTDiscuss, PubMed, Scopus, and CINAHL through October 2024 to identify randomized controlled trials (RCTs). Eligible studies involved adults with MS and focused on exercise regimens compliant with PAG-MS. Comparisons were made with non-PAG-MS exercise regimens or no exercise at all.Two independent reviewers screened studies, extracted data, and assessed the risk of bias. Depression and anxiety outcomes were analyzed using meta-analyses, incorporating minimal clinically important difference (MCID) thresholds to determine clinical significance.

Results:

Twelve RCTs involving 458 participants we included. Depression outcomes showed significant improvement (MD: −4.46 [95% CI: −6.90, −2.01], P < 0.01), exceeding the MCID threshold of 3.00points on the Beck Depression Inventory scale. Anxiety outcomes, analyzed from two RCTs, showed non-significant improvement (SMD: −0.87 [95% CI: −2.46, 0.72], P = 0.29). Overall, 58.33% of studies had a high risk of bias.

Conclusion:

Interventions that meet PAG-MS demonstrate clinically meaningful improvements in depression, supporting their use as a non-pharmacological treatment strategy. However, given the limited data on anxiety outcomes, further research is needed to clarify the potential benefits in this domain.

PROSPERO Registration:

CRD42023387305

INTRODUCTION

Multiple Sclerosis (MS) is a chronic and progressive neurological condition affecting approximately 2.8 million individuals worldwide (Walton et al, 2020). It is the leading cause of non-traumatic neurological disability in young adults and is characterized by demyelination and neurodegeneration in the central nervous system, leading to a broad spectrum of physical and cognitive impairments (Dutta and Trapp, 2011; Hauser and Cree, 2020). Typically manifesting in the second or third decade of life, MS significantly reduces quality of life and imposes substantial socioeconomic and healthcare burdens (Dutta and Trapp, 2011). Common symptoms include fatigue, balance and gait impairments, spasticity, visual disturbances, neuropathic pain, and cognitive decline, all of which contribute to long-term disability and diminished quality of life (Saguil, Farnell Iv, and Jordan, 2022).

While MS is predominantly recognized for its physical symptoms, its impact on mental health is equally profound (Mustač et al, 2021). Depression and anxiety are among the most prevalent psychological comorbidities in persons with MS (PwMS), with up to 30.5% of individuals experiencing depression and 34% experiencing anxiety—rates significantly higher than those observed in the general population (Boeschoten et al, 2017; Fatheddine et al, 2024). These mental health conditions arise from a combination of psychosocial and biological factors, including inadequate coping mechanisms, insufficient social support, and MS-related changes in brain structure, immune function, and inflammatory pathways (Hanna and Strober, 2020; Mustač et al, 2021). Beyond their mental health consequences, depression and anxiety exacerbate physical symptoms, reduce daily functioning, hinder adherence to disease-modifying treatments, and contribute to heightened healthcare costs, poorer disease outcomes, and increased risks of suicidal ideation and attempts (Nelson and Bourdette, 2020; Sartorious, 2013). These challenges underscore the urgent need for effective interventions that address both depression and anxiety in PwMS.

Pharmacotherapy and psychotherapy are the most commonly used interventions for managing depression and anxiety in PwMS, but both approaches have significant limitations (Jones, Motl, and Sandroff, 2021; Šilić, Motl, and Duffecy, 2023). Medications such as selective serotonin reuptake inhibitors and serotonin and noradrenaline reuptake inhibitors, while widely prescribed, lack robust MS-specific evidence and often cause side effects such as nausea, weight gain, dependency, and withdrawal symptoms (Carvalho et al, 2016). Similarly, benzodiazepines are associated with dependency and reduced long-term efficacy, making them less suitable for long-term use in MS (Bandelow, Michaelis, and Wedekind, 2017; Garakani et al, 2020). Psychotherapy, particularly cognitive-behavioral therapy, shows promise and can be delivered remotely via telephone or internet-based platforms, which may improve accessibility for some individuals (Mohr et al, 2000; Ratajska, Zurawski, Healy, and Glanz, 2019). However, despite these advancements, barriers such as cost, digital literacy requirements, privacy concerns, and the need for MS-specific adaptations may still limit its accessibility and effectiveness for PwMS(Ratajska, Zurawski, Healy, and Glanz, 2019).

Exercise, a form of physical activity, has emerged as a promising non-pharmacological intervention for managing both depression and anxiety in PwMS (Motl and Sandroff, 2020; Šilić, Motl, and Duffecy, 2023). It is relatively low-cost, widely accessible, and associated with minimal side effects, making them an appealing alternative or complement to traditional treatments (Phillips, Kiernan, and King, 2003). Exercise is purposeful, structured, and repeated over time to improve fitness, performance, or health unlike lifestyle activity, which is the other form of physical activity that includes occupational, household or leisure activities which may be planned or unplanned (Caspersen, Powell, and Christenson, 1985; Dasso, 2019; Kalb et al, 2020). In the general population, exercise has been shown to be as effective as antidepressant medication and psychotherapy for treating depression (Netz, 2017). However, evidence specific to MS remains limited. Meta-analyses by Herring et al (2017) and Dalgas, Stenager, Sloth and Stenager (2015)) reported small-to-moderate effects of exercise on depressive symptoms in PwMS, but many of the studies included in these reviews did not adhere to the standardized exercise recommendation for PwMS (Kalb et al, 2020; Kim et al, 2019; Latimer-Cheung et al, 2013). Evidence for the effects of exercise on anxiety in PwMS is even more inconclusive. A meta-analysis by Gascoyne, Karahalios, Demaneuf and Marck (2019), found no significant improvements in anxiety outcomes following exercise intervention, potentially due to small sample sizes and methodological inconsistencies.

Moreover, previous studies and meta-analyses have focused primarily on statistical significance without assessing whether observed changes are clinically meaningful to patients. Statistical significance reflects the likelihood that results are not due to chance but does not address the practical importance or real-world impact of an intervention. Assessing the minimal clinically important difference (MCID) is essential for determining whether changes in depression and anxiety symptoms are substantial enough to be perceived as beneficial by PwMS (Bartels et al, 2017; Prasad, 2024). Without incorporating MCID, the practical implications of exercise interventions remain uncertain, limiting their relevance to clinical practice and decision-making (Sharma, 2021).

Until recently, there was no consensus on exercise recommendations for PwMS. However, emerging Physical Activity Guidelines for Multiple Sclerosis (PAG-MS) now recommend engaging in a minimum of 150 minutes per week of exercise and/or lifestyle physical activity (Kalb et al, 2020), OR at least 10–30 minutes of moderate intensity aerobic activity 2 times per week and strength training exercises for major muscle groups 2 times per week (Kim et al, 2019; Latimer-Cheung et al, 2013). These guidelines are based on safety, feasibility, and effectiveness, but it remains unclear whether adherence to these recommendations will result in meaningful improvements in depression and anxiety outcomes in PwMS. Discrepancies in findings from previous reviews may stem from the inclusion of studies with insufficient exercise dosages, as dose-response relationships between exercise and mental health are well-documented (Hamer, Stamatakis, and Steptoe, 2009).

The aim of this study was to evaluate the effects of meeting PAG-MS on depression and anxiety outcomes in PwMS. Specifically, this systematic review and meta-analysis aims to compare exercise regimens that meet PAG-MS criteria to those that do not or to non-exercise control groups, with a focus on determining the clinical significance of these interventions using MCID thresholds. This study seeks to provide evidence on the potential of structured exercise to improve mental health outcomes in PwMS and guide clinicians and researchers in optimizing interventions for this population.

METHODS

Data sources and search strategy

This systematic review was conducted in accordance with the guidelines outlined in the Cochrane Handbook (Higgins et al, 2024) and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al, 2021). This systematic review is part of a larger initiative aimed at determining the effects of physical activity guidelines for multiple sclerosis (PAG-MS) on MS symptoms. As part of this initiative, a related systematic review examining the effects of PAG-MS on fatigue has been published (Abou, Murphy, Truong, & Peters, 2025). The review protocol was prospectively registered on PROSPERO, the International Prospective Register of Systematic Reviews (CRD42023387305). A comprehensive search was performed across EMBASE, PsycINFO, Web of Science, SPORTDiscuss, PubMed, Scopus, and CINAHL, covering all available records from the inception of each database to October 2024. The search strategy utilized a combination of MeSH terms, Emtree terms, keywords, and synonyms as appropriate. The database-specific Boolean search string was structured as follows: (“Physical Activity”[MeSH] OR “Exercise”) AND (“Depression”[MeSH] OR “Anxiety”) AND (“Multiple Sclerosis”[MeSH] OR “MS”). Detailed search strategies tailored for each database can be found in Supplementary Appendix A.

Screening studies and selection criteria

Two independent reviewers (JP) and (ET) screened all identified articles to assess eligibility, with disagreements resolved through mediation by a third reviewer (LA) until consensus was reached. Titles and abstracts were first screened in Zotero (version 6.0.30) using broad keywords derived from the search strategy. Discrepancies were mediated by (LA) until an agreement was reached. Full texts of potentially relevant studies were then compiled into an Excel spreadsheet and independently reviewed by (JP) and (ET) to determine final eligibility based on the inclusion and exclusion criteria outlined below. To ensure thoroughness, both forward and backward manual searches of the references were conducted, including citations of the included studies.

The eligibility of studies was determined using the Population, Intervention or Exposure, Comparators, Outcomes, and Study Designs (PICOS) framework: (1) Population: adults (18+ years old) diagnosed with MS; (2) Intervention: exercise adhering to PAG-MS criteria (outlined below). We define exercise a form of physical activity, which is purposeful, structured, and repeated activity aimed at improving fitness, performance, or health (Caspersen, Powell, and Christenson, 1985; Dasso, 2019; Kalb et al, 2020); (3) Comparators: either no intervention or an active control group not meeting PAG-MS; (4) Outcomes: measures of depression or anxiety; (5) Study design: only randomized controlled trials (RCTs) were considered.

To be eligible, studies needed to measure anxiety or depression using validated tools. Anxiety measures included the Beck Anxiety Inventory (BAI), Spielberg Anxiety Inventory, anxiety subscale of the Hospital Anxiety and Depression Scale (HADS), the state anxiety subscale of the State-Trait Anxiety Inventory (STAI), the tension-anxiety subscale of the Profile of Mood States (POMS), or similar validated instruments. Depression measures included the Beck Depression Inventory (BDI), or the depression subscale of the HADS. These validated tools ensured consistency and reliability in assessing mental health outcomes.

The Physical Activity Guidelines for PwMS (PAG-MS) were established based on recommendations from multiple reviews and guidelines endorsed by the National MS Society. These guidelines define meeting PAG-MS as engaging in a minimum of 150 minutes per week of exercise and/or lifestyle physical activity (Kalb et al, 2020), OR at least 10–30 minutes of moderate intensity aerobic activity 2 times per week and strength training exercises for major muscle groups 2 times per week (Kim et al, 2019; Latimer-Cheung et al, 2013).

Data extraction and data synthesis

Two reviewers (JP) and (ET) conducted data extraction independently, with discrepancies reviewed and resolved by a third reviewer (LA). Extracted data were compiled into a finalized table after verification for accuracy. The following details were collected from the included studies: author(s), year of publication, population characteristics (e.g., sample size, average age, % females in the study, disease duration, type of MS, and disability measures), exercise intervention details (e.g., type, duration, intensity, and frequency of sessions), and the mental health assessment tools used for depression and anxiety. Changes in depression and anxiety outcomes from baseline to post-intervention were summarized for both intervention and control groups. Key numerical findings were reported as percentage changes in mean depression and anxiety scores from baseline to post-intervention.

Risk of bias assessment

The included studies were independently evaluated by two reviewers (TM) and (JP) using the Cochrane Risk of Bias Tool for Randomized Trials (RoB 2) (Higgins et al, 2024; Sterne et al, 2019), with discrepancies resolved through consultation with a third reviewer (LA). The RoB 2 tool provides a comprehensive and systematic approach to assessing potential bias in randomized trials across five key domains: (1) Bias arising from the randomization process, which evaluates the adequacy of random sequence generation, allocation concealment, and baseline comparability; (2) Bias due to deviations from intended interventions, which assesses adherence to interventions and whether deviations were related to the trial context; (3) Bias due to missing outcome data, considering the extent, reasons, and impact of missing data on study results; (4) Bias in measurement of the outcome, which examines the risk of measurement errors and the blinding of outcome assessors; and (5) Bias in selection of the reported result, which checks for selective reporting of pre-specified outcomes. Each domain was assessed using targeted signaling questions, and studies were rated as having low risk of bias, some concerns, or high risk of bias. The overall risk of bias for each study was assigned based on the highest level of concern identified across the domains, ensuring a consistent and transparent evaluation of methodological quality.

Statistical analysis

A quantitative synthesis, in addition to a qualitative summary, was conducted to analyze the extracted data. Meta-analyses were performed to evaluate the overall effects of the included studies on depression and anxiety outcomes. Data were pooled when two or more studies assessed a specific outcome, providing baseline and post-intervention mean values along with standard deviations (SD). All analyses were performed using Review Manager (Version 5.3.6, Cochrane, UK). The mean difference (post-intervention - baseline) and SD change for intervention and control groups were computed for depression and anxiety outcomes in the meta-analyses.

The choice of statistical model was based on the level of heterogeneity. A fixed-effect model was applied when heterogeneity was low (I² = 0%–49%), while a random-effects model was used for high heterogeneity (I² > 50%)(Higgins et al, 2024). For outcomes measured consistently with a single assessment tool, the mean difference (MD) was calculated to reflect the absolute change in scores between intervention and control groups. However, when multiple tools were used to assess the same outcome domain, the standardized mean difference (SMD) was calculated. SMD standardizes the effect sizes and allows for comparability across studies that utilize different measurement scales for the same outcome. When lower scores indicated more favorable outcomes, data were transformed to maintain consistent directionality across all assessment tools.

Clinical significance was assessed by comparing meta-analyses effect sizes to the Minimal Clinically Important Difference (MCID) for each assessment tool (Kluger, Garimella, and Garvan, 2017). For tools without an established MCID, an estimated MCID was calculated as 0.5 times the baseline SD of the experimental group (Watt, Veroniki, Tricco, and Straus, 2021). To compare MCID with MD scores, a pooled SD from all included studies was used to ensure consistent calculations. This approach enabled a comprehensive and standardized evaluation of the clinical impact of exercise interventions on depression and anxiety outcomes.

RESULTS

Study selection

The search strategy retrieved 6911 records, with 4877 duplicates removed, leaving 2034 articles for title and abstract screening. After the initial screening process, 1858 articles were excluded, and 176 full-text reports were assessed for eligibility. Following a detailed review, 164 reports were excluded, resulting in the inclusion of 12 studies in this review. A PRISMA flow diagram outlining the selection process is presented in Figure 1.

Figure 1:

A PRISMA flow diagram outlining the studies selection process

Characteristics of included studies

Characteristics of the included studies are depicted in Table 1. A total of 458 individuals with MS participated in the included RCTs, with 233 in experimental groups and 225 in control groups. The mean age of participants varied from 28.31 to 48.30 years. The majority of participants were female, ranging from 55% to 100% of participants per study. Eight of the twelve studies reported the type of MS examined (Bilek et al, 2022; Cakt et al, 2010; Correale et al, 2021; Hortobagyi et al, 2022; Ozgen, Karapolat, Akkoc, and Yuceyar, 2016; Ozkul et al, 2020; Razazian et al, 2016; Tollár et al, 2020), with relapsing-remitting MS (RRMS) being the most common type, though secondary progressive MS (SPMS) and primary progressive MS (PPMS) were also reported. Nine studies reported Expanded Disability Status Scale (EDSS) scores (Ahmadi et al, 2013; Bilek et al, 2022; Hejazi et al, 2012; Hortobagyi et al, 2022; Ozgen, Karapolat, Akkoc, and Yuceyar, 2016; Ozkul et al, 2020; Razazian et al, 2016; Sadeghi Bahmani et al, 2020; Tollár et al, 2020), ranging from 1.00 to 6.50, reflecting participants with mild to moderate disability levels. Mental health outcomes were unevenly reported across the studies. Eleven studies assessed depression (Ahmadi et al, 2013; Bilek et al, 2022; Cakt et al, 2010; Correale et al, 2021; Hejazi et al, 2012; Hortobagyi et al, 2022; Ozgen, Karapolat, Akkoc, and Yuceyar, 2016; Ozkul et al, 2020; Razazian et al, 2016; Sadeghi Bahmani et al, 2020; Tollár et al, 2020), while only two studies reported anxiety outcomes (Ahmadi et al, 2013; Hassanpour Dehkordi, 2016).

Table 1.

Participants characteristics

Study	Sample size	Age (y)	Female (%)	Adherence (%)	Type of MS	Disease duration (y)	EDSS/PDDS	Total Duration of Exercise per Week (minutes)	Total Entire Duration of Exercise (minutes)
Ahmadi et al (2013)	Yoga: 11 Control (Waitlist): 10	Yoga: 32.27±8.68 Control (Waitlist): 36.70±9.32	Yoga: 100% Control (Waitlist): 100%	Yoga: 100% Control (Waitlist): 100%	NP	Yoga: 4.72±5.62 Control (Waitlist): 5.00±3.05	EDSS Yoga: 2.00±1.09 Control (Waitlist): 2.25±1.25	180–210	1440–1680
Bilek et al (2022)	Combined (FKE + aerobic): 16 Control (FKE): 16	Combined (FKE + aerobic): 28.31±5.89 Control (FKE): 32.50±8.75	Combined (FKE + aerobic): 94% Control (FKE): 75%	Combined (FKE + aerobic): 100% Control (FKE): 100%	RRMS	Combined (FKE + aerobic): 5.44±3.92 Control (FKE): 7.66±5.05	EDSS Combined (FKE + aerobic): 1.69±0.85 Control (FKE): 1.97±0.89	180–210	1080–1260
Cakt et al (2010)	Combined (HIIT + balance): 14 Control: 9	Combined (HIIT + balance): 36.40±10.5 Control: 35.50±10.9	Combined (HIIT + balance): 64% Control: 67%	Combined (HIIT + balance): 93.33% Control: 60%	RRMS or SPMS	Combined (HIIT + balance): 9.2±5.00 Control: 6.6±2.4 0	NP	170–190	1360–1520
Correale et al (2021)	Combined (Aerobic + resistance): 14 Control: 9	Combined (Aerobic + resistance): 45.40±7.2 Control: 48.30±6.1	Combined (Aerobic + resistance): 100% Control: 100%	Combined (Aerobic + resistance): 100% Control: 69.23%	RRMS	NP	NP	90–120	1080–1440
Hassanpour Dehkordi (2016)	Yoga: 30 Control: 30	Yoga: 30.00 Control: 30.00	NP	Yoga: 100% Control: 100%	NP	NP	NP	180–210	2160–2520
Hejazi et al (2012)	Aquatic: 20 Control: 20	Aquatic: 35.45 Control: 30.40	Aquatic: 100% Control: 100%	Aquatic: 100% Control: 100%	NP	Aquatic: 7.00 Control: 5.45	EDSS Aquatic: 1–4 Control: 1–4	150–210	1200–1680
Hortobagyi et al (2022)	Intervention: Exergaming: 14 Balance: 14 Cycling: 14 PNF: 14 Control (Waitlist): 12	Intervention: Exergaming: 48.20±5.48 Balance: 46.90±6.46 Cycling: 48.10±5.65 PNF: 46.90±5.57 Control: 44.40±6.76	Intervention: Exergaming: 86% Balance: 86% Cycling: 93% PNF: 93% Control (Waitlist): 92%	Exergaming: 100% Balance: 100% Cycling: 100% PNF: 100% Control (Waitlist): 100%	Intervention: Exergaming: RRMS 50%; PPMS 50% Balance or Cycling or PNF: RRMS 64%; PPMS 36% Control: RRMS 66%; PPMS: 34%	Intervention: Exergaming: 12.10±2.68 Balance: 13.60±4.07 Cycling: 13.20±4.42 PNF: 12.70±4.25 Control: 14.00±4.11	EDSS Intervention and Control: 5.00 (5.00–6.00)	180–300	20,220
Ozgen, Karapolat, Akkoc and Yuceyar (2016)	Home-based exercise: 20 Control (Waitlist): 20	Home-based exercise: 42.5 (22–60) Control (Waitlist): 39.5 (24–56)	Home-based exercise: 80% Control (Waitlist): 60%	Home-based exercise: 100% Control (Waitlist): 100%	Home-based exercise: RRMS – 40%; PPMS – 15%; SPMS – 45% Control (Waitlist): RRMS – 45%; PPMS – 20%; SPMS – 35%	Home-based exercise: 10 (1–24) Control (Waitlist): 5.50 (1–20)	EDSS Home-based exercise: 3.50 (2.00–6.50) Control (Waitlist): 3.50 (2.00–6.00)	210–280	2940–3920
Ozkul et al (2020)	Combined (Aerobics + Pilates): 17 Control (Relaxation exercises): 17	Combined (Aerobics + Pilates): 35.88±9.74 Control (Relaxation exercises): 36.76±9.02	Combined (Aerobics + Pilates): 76% Control (Relaxation exercises): 76%	Aerobics + Pilates: 100% Relaxation exercises: 100%	RRMS	Combined (Aerobics + Pilates): 7.18±6.12 Control (Relaxation exercises): 5.71±4.90	EDSS Combined (Aerobics + Pilates): 1.50±0.77 Control (Relaxation exercises): 1.71±0.94	315	2520
Razazian et al (2016)	Aquatic: 18 Yoga: 18 Control (Peer-support): 18	Aquatic: 35.39±6.89 Yoga: 33.33±7.40 Control (Peer-support): 33.11±6.60	Aquatic: 100% Yoga: 100% Control (Peer-support): 100%	Aquatic: 100% Yoga: 100% Control (Peer-support): 100%	Aquatic: RRMS – 72%; SPMS – 6%; PRMS – 22% Yoga: RRMS – 61%; SPMS – 11%; PRMS – 28% Control (Peer-support): RRMS – 67%; SPMS – 11%; PRMS – 22%	Aquatic: 7.11±0.95 Yoga: 6.90±0.90 Control (Peer-support): 6.78±0.65	EDSS Intervention: Aquatic: 3.44±0.95 Yoga: 3.89±1.02 Control (Peer-support): 3.25±1.24	180	1440
Sadeghi Bahmani et al (2020)	Aquatic: 18 Control: 22	Aquatic: 40.61±8.97 Control: 33.77±6.56	Aquatic: 100% Control: 100%	Aquatic: 100% Control: 100%	NP	Aquatic 3x: 6.58±3.92 Control: 6.34±4.34	Aquatic 3x: 1.50 (4.00) Control: 1.50 (6.00)	180	1440
Tollár et al (2020)	Exercise: 14 Balance: 14 Cycling: 14 PNF: 14 Control (Waitlist): 12	Exergaming: 48.20±5.48 Balance: 46.90±6.46 Cycling: 48.10±5.65 PNF: 46.90±5.57 Control: 44.40±6.76	Exergaming: 86% Balance: 86% Cycling: 93% PNF: 93% Control (Waitlist): 92%	Exergaming: 100% Balance: 100% Cycling: 100% PNF: 100% Control (Waitlist): 100%	Exergaming: RRMS 50%; PPMS 50% Balance or Cycling or PNF: RRMS 64%; PPMS 36% Control (Waitlist): RRMS 66%; PPMS: 34%	Exergaming: 12.10±2.68 Balance: 13.60±4.07 Cycling: 13.20±4.42 PNF: 12.70±4.25 Control (Waitlist): 14.00±4.11	EDSS Intervention and Control: 5.00 (5.00–6.00)	300	1500

XX±XX: Mean ± SD; XX (XX-XX): Mean or Median (95% CI or Range)

EDSS: Expanded Disability Status Scale; PDDS: Patient-Determined Disease Steps; NP: Not provided;RRMS: Relapse-remitting multiple sclerosis; PPMS: primary progressive multiple sclerosis; SPMS: Secondary progressive multiple sclerosis; PNF: Proprioceptive Neuromuscular Facilitation; PRMS: Progressive relapsing multiple sclerosis; HIIT: High-intensity interval training: FKE: Frenkel Coordination Exercise; y: Years; %: Percentage; MS: Multiple sclerosis

All included studies implemented interventions that met PAG-MS, with modes of exercise varying across studies. Yoga was used as the intervention in three studies (Ahmadi et al, 2013; Hassanpour Dehkordi, 2016; Razazian et al, 2016), while aquatic exercise was implemented in three studies (Hejazi et al, 2012; Razazian et al, 2016; Sadeghi Bahmani et al, 2020). Other interventions combined different exercise modalities, such as aerobic exercise paired with resistance training (Correale et al, 2021), or Pilates (Ozkul et al, 2020), and aerobic exercise combined with Frenkel Coordination Exercise (FKE) (Bilek et al, 2022). High-intensity interval training (HIIT) combined with balance training was used in one study (Cakt et al, 2010), while exergaming, balance training, cycling, and proprioceptive neuromuscular facilitation (PNF) were examined in two studies (Hortobagyi et al, 2022; Tollár et al, 2020). One study utilized a home-based exercise program (Ozgen, Karapolat, Akkoc, and Yuceyar, 2016). The duration of interventions ranged widely, from 5 weeks (Tollár et al, 2020) to 109 weeks (Hortobagyi et al, 2022). Most interventions lasted between 8 and 12 weeks. Participants engaged in exercise sessions 2–5 times per week, depending on the study, with session durations varying from 15 minutes(Ozgen, Karapolat, Akkoc, and Yuceyar, 2016) to 105 minutes (Ozkul et al, 2020). Control groups also varied in their design. Some used waitlist controls (Ahmadi et al, 2013; Ozgen, Karapolat, Akkoc, and Yuceyar, 2016), while others included non-exercise protocols such as relaxation exercises (Ozkul et al, 2020) or discussion groups (Razazian et al, 2016). The most common tool used for depression is BDI, while BAI and Spielberg were used for anxiety. Comprehensive details of the interventions, control conditions and assessment tools can be found in Table 2.

Table 2.

Characteristics of interventions, symptoms of depression and/or anxiety, and summary of findings

Study	Intervention	Control	Outcome	Findings
Ahmadi et al (2013)	Mode: Yoga Intensity: NP Intervention duration: 8 weeks Frequency: 3x/week Session duration: 60–70 min	Mode: Waitlist Intensity: NP Intervention duration: 8 weeks Frequency: NP Session duration: NP	Depression: BDI Anxiety: BAI	Yoga significantly improved depression (36% vs no change), and anxiety (49% vs no change) compared to waitlist control.
Bilek et al (2022)	Mode: Aerobic exercise + FKE Intensity: Moderate Intervention duration: 6 weeks Frequency: 3x/week Session duration: 60–70 min	Mode: FKE Intensity: NP Intervention duration: 6 weeks Frequency: 3x/week Session duration: 20–30 min	Depression: BDI	Aerobic exercise + FKE significantly improved depression (48% improvement) compared to FKE alone, which showed a smaller, less pronounced improvement (15%), with a significant between-group difference favoring the intervention group.
Cakt et al (2010)	Mode: Progressive Resistance training (HIIT) on a bike + balance training Intensity: Vigorous Intervention duration: 8 weeks Frequency:2x/week Session duration: 85–95 minutes	Mode: Control Intensity: NP Intervention duration: NP Frequency: NP Session duration: NP	Depression: BDI	Progressive Resistance training + balance significantly improved depression (24% vs no change) from baseline compared to control group. There was a significant between-group difference favoring the intervention group.
Correale et al (2021)	Mode: Aerobic + resistance training Intensity: Moderate-to-vigorous Intervention duration: 12 weeks Frequency: 2x/week Session duration: 45–60 minutes	Mode: Control Intensity: NP Intervention duration: NP Frequency: NP Session duration: NP	Depression: BDI-II	There were no significant differences between exercise and control group from baseline to post-test.
Hassanpour Dehkordi (2016)	Mode: Yoga Intensity: NP Intervention duration: 12 weeks Frequency:3x/week Session duration: 60–70 minutes	Mode: Control Intensity: NP Intervention duration: NP Frequency: NP Session duration: NP	Anxiety: Spielberg Anxiety Inventory	Anxiety significantly improved in the yoga group, but did not significantly change in the control group.
Hejazi et al (2012)	Mode: Aquatic exercise Intensity: NP Intervention duration: 8 weeks Frequency:3x/week Session duration: 50–70 minutes	Mode: Control Intensity: NP Intervention duration: NP Frequency: NP Session duration: NP	Depression: BDI	Aquatic exercise significantly improved depression (25%) from baseline to post-test. Control group did not significantly change. No between groups information was provided.
Hortobagyi et al (2022)	Mode: Exergaming or Balance or Cycling or PNF Intensity: Vigorous Intervention duration: 109 weeks Frequency:5x/week for 5 weeks then 3x/week for 104 weeks Session duration: 60 minutes	Mode: Control Intensity: NP Intervention duration: NP Frequency: NP Session duration: NP	Depression: BDI	No significant change or difference was observed
Ozgen, Karapolat, Akkoc and Yuceyar (2016)	Mode: Home-based exercise Intensity: NP Intervention duration: 8 weeks Frequency:14x/week Session duration: 15–20 minutes	Mode: Waitlist Intensity: NP Intervention duration: NP Frequency: NP Session duration: NP	Depression: BDI	The home-based exercise group improved depression (36% vs no change) significantly more than the waitlist control group from baseline to post-test.
Ozkul et al (2020)	Mode: Aerobics + Pilates Intensity: Moderate Intervention duration: 8 weeks Frequency:3x/week Session duration: 105 minutes	Mode: Relaxation exercises (non-physical activity) Intensity: NP Intervention duration: 8 weeks Frequency: 3x/week Session duration: 15–20	Depression: BDI	The Aerobic + Pilates group showed a non-significant improvement in depression (17%), while the relaxation group showed a non-significant increase in depression scores (-21%), with no significant difference between the groups.
Razazian et al (2016)	Mode 1: Yoga Intensity: NP Intervention duration: 8 weeks Frequency:3x/week Session duration: 60 minutes Mode 2: Aquatic exercise Intensity: NP Intervention duration: 8 weeks Frequency: 3x/week Session duration: 60 minutes	Mode: Control – discussion group Intensity: NP Intervention duration: 8 weeks Frequency: 2–3x/week Session duration: 60–90 minutes	Depression: BDI	There was no significant difference in depression between yoga (74%) and aquatic exercise (75%). Both groups improved depression significantly and were greater than the control group from baseline to post-test.
Sadeghi Bahmani et al (2020)	Mode: Aquatic exercise Intensity: NP Intervention duration: 8 weeks Frequency: 3x/week Session duration: 60 minutes	Mode: Active Control (social engagement) Intensity: NP Intervention duration: 8 weeks Frequency: 2–3x/week Session duration: 60 minutes	Depression: BDI-FS	No significant changes or differences were observed.
Tollár et al (2020)	Mode: Exergaming or Balance or Cycling or PNF Intensity: Vigorous Intervention duration: 5 weeks Frequency:5x/week Session duration: 60 minutes	Mode: Control Intensity: NP Intervention duration: NP Frequency: NP Session duration: NP	Depression: BDI	No significant change or difference was observed

BAI: Beck Anxiety Inventory; BDI: Beck Depression Inventory; BDI-FS: Beck Depression Inventory-Fast Screen; FKE: Frenkel Coordination Exercise; HIIT: High intensity interval training; NP: Not Provided; PNF: Proprioceptive neuromuscular facilitation; SDS: Self-rating depression scale.

Effect of exercise on depression

The overall effect of exercise meeting PAG-MS on depression, analyzed across 11 studies using BDI, was statistically significant, with MD of −4.46 (95% CI: −6.90, −2.01, P < 0.01; Figure 2a). However, substantial heterogeneity was observed (I² = 88%). The effect size of 4.46 points obtained for the BDI exceeds the estimated MCID of 3.00 points on the BDI scale. Thisunderscores a clinical significance of the effect of meeting PAG-MS on depression measured with the BDI.

Figure 2a:

Forest plot of the impact of meeting PAG-MS on depression (BDI only) outcomes in PwMS

A qualitative summary of the included studies shows that six of eleven studies found significant improvements in depression following exercise interventions meeting PAG-MS guidelines compared to controls (Ahmadi et al, 2013; Bilek et al, 2022; Cakt et al, 2010; Hejazi et al, 2012; Ozgen, Karapolat, Akkoc, and Yuceyar, 2016; Razazian et al, 2016). However, five studies reported no significant improvement (Correale et al, 2021; Hortobagyi et al, 2022; Ozkul et al, 2020; Sadeghi Bahmani et al, 2020; Tollár et al, 2020). Four studies favored the experimental group, with different interventions showing significant effects, including Bilek et al (2022) combining aerobic exercise and FKE, Cakt et al (2010) using progressive resistance training and balance, and Ozgen, Karapolat, Akkoc and Yuceyar (2016) employing a home-based program.

Effect of exercise on anxiety

A meta-analysis of two studies assessing anxiety outcomes found a non-significant improvement in anxiety symptoms following Exercise interventions meeting PAG-MS, with an SMD of −0.87 (95% CI: −2.46, 0.72, P = 0.29; Figure 2b). Substantial heterogeneity was observed (I² = 87%). However, a qualitative summary revealed that both studies reported significant improvement in anxiety levels following interventions (Ahmadi et al, 2013; Hassanpour Dehkordi, 2016). Ahmadi et al (2013) reported a 49% improvement in anxiety among participants in the yoga intervention group, compared to no change in the waitlist control group. Similarly, Hassanpour Dehkordi (2016) observed significant improvement in anxiety for participants in the yoga group compared to controls.

Figure 2b:

Forest plot of the impact of meeting PAG-MS on anxiety outcomes in PwMS

A qualitative summary of the included studies also shows that two studies observed significant improvements in anxiety with exercise (Ahmadi et al, 2013; Hassanpour Dehkordi, 2016), including a 49% improvement in a yoga group (Ahmadi et al, 2013). Razazian et al (2016) also found aquatic exercise significantly improved depression.

Risk of bias of included studies

Overall, three studies (25%) were judged as low risk (Ozgen, Karapolat, Akkoc, and Yuceyar, 2016; Razazian et al, 2016; Tollár et al, 2020), two studies (16.67%) raised some concerns (Hortobagyi et al, 2022; Sadeghi Bahmani et al, 2020), and seven studies (58.33%) were classified as high risk (Ahmadi et al, 2013; Bilek et al, 2022; Cakt et al, 2010; Correale et al, 2021; Hassanpour Dehkordi, 2016; Hejazi et al, 2012; Ozkul et al, 2020) as shown in Figure 3a and Figure 3b. This distribution highlights the variability in methodological quality across the included studies. In terms of the randomization process, seven studies (58.33%) were classified as low risk, while five studies (41.67%) were rated as high risk. For deviations from intended interventions, six studies (50%) were considered low risk, four studies (33.33%) raised some concerns, and two studies (16.67%) were classified as high risk. Bias in outcome measurement was generally low, with 11 studies (91.67%) rated as low risk and one study (8.33%) assessed as high risk. Bias due to missing outcome data was rated as low in seven studies (58.33%), as having some concerns in two studies (16.67%), and as high in three studies (25%). Finally, nine studies (75%) were classified as low risk for selective reporting, while two studies (16.67%) raised some concerns, and one study (8.33%) was assessed as high risk.

Figure 3a:

Risk of Bias Assessment Across Individual Studies in Key Domains

Figure 3b:

Summary of Risk of Bias Across All Studies for Each Domain

DISCUSSION

This systematic review and meta-analysis evaluated the impact of Exercise interventions meeting PAG-MS on depression and anxiety outcomes in PwMS, with a focus on clinical significance. The findings reveal that exercise interventions meeting PAG-MS not only lead to statistically significant improvement in depression but also achieve clinically meaningful improvements, as evidenced by the meta-analysis effect size exceeding the estimated MCID. This highlights the real-world relevance of such interventions in improving depression for PwMS. However, the evidence for anxiety remains less conclusive due to a limited number of studies and methodological inconsistencies. The risk of bias of the included studies varied; although few studies demonstrated low risk of bias (Ozgen, Karapolat, Akkoc, and Yuceyar, 2016; Razazian et al, 2016; Tollár et al, 2020), two were with some concerns (Hortobagyi et al, 2022; Sadeghi Bahmani et al, 2020), and seven with high risk of bias (Ahmadi et al, 2013; Bilek et al, 2022; Cakt et al, 2010; Correale et al, 2021; Hassanpour Dehkordi, 2016; Hejazi et al, 2012; Ozkul et al, 2020).

The findings of this study expand upon prior studies evaluating exercise interventions for mental health outcomes in PwMS. Consistent with earlier meta-analyses, such as those by Herring et al (2017), Dalgas, Stenager, Sloth and Stenager (2015), and Ensari, Motl and Pilutti (2014), this study confirms that exercise significantly improves depression. However, this systematic review advances the field by specifically analyzing interventions meeting PAG-MS and addressing inconsistencies in exercise modalities and intensities that limited prior works and reporting its clinical relevance. The need to assess clinical relevance alongside statistical significance in medical research has been emphasized in several studies (Bartels et al, 2017; Prasad, 2024). Notably, the statistical effect size for the BDI exceeded the MCID, demonstrating that these interventions achieved changes that are not only statistically significant but also meaningful to patients. This distinction is critical, as surpassing the MCID threshold enhances patients’ perceptions of improvement, potentially reinforcing engagement and adherence to exercise regimens, which may lead to sustained mental health benefits for PwMS.

While some studies suggest exercise interventions may improve anxiety in the general population (Conn, 2010; McDowell, Dishman, Gordon, and Herring, 2019), evidence for anxiety outcomes was less conclusive in this study, aligning with previous findings by Gascoyne, Karahalios, Demaneuf and Marck (2019), which noted limited and inconsistent effects of exercise on anxiety in PwMS. While individual studies reported significant improvement in anxiety (Ahmadi et al, 2013; Hassanpour Dehkordi, 2016), the overall meta-analysis revealed non-significant effects. These mixed results likely stem from challenges identified in prior research, including variability in participant characteristics, baseline anxiety severity, and intervention designs (Gascoyne, Karahalios, Demaneuf, and Marck, 2019). None of the included studies specifically targeted anxiety as a primary outcome or prescreened participants for elevated anxiety levels, potentially diluting observed effects. AsButler, Matcham and Chalder (2016) emphasized, anxiety in MS remains under-researched and often overshadowed by a greater focus on depression and physical health outcomes. These findings underscore the need for targeted studies that prioritize anxiety as a primary outcome in PwMS.

Yoga and aquatic exercises significantly improved depression, as demonstrated in studies such as Ahmadi et al (2013) for yoga and Razazian et al (2016) and Hejazi et al (2012) for aquatic exercises, these interventions consistently involved three sessions per week over an eight-week period. Regular engagement in these activities, especially three times a week for periods of 8 weeks or more, may be particularly beneficial in addressing depression symptoms. The mechanisms behind these benefits likely include enhanced neuroplasticity (Cooney et al, 2013), improvements in self-control, self-esteem, and coping skills, which have been associated with both yoga and aquatic exercise (Razazian et al, 2016). Additionally, yoga’s focus on self-awareness and stress reduction through psychophysiological processes, similar to relaxation techniques, may further contribute to improvements in depression (Pascoe and Bauer, 2015). However, the evidence for anxiety was less conclusive. Although Ahmadi et al (2013) reported significant improvements in anxiety with yoga, Hassanpour Dehkordi (2016), which employed a similar yoga intervention with the same frequency and duration, did not observe significant effects. This variability may be influenced by differences in MS types, as anxiety levels have been shown to vary between MS subtypes, with women with relapsing-remitting MS reporting higher anxiety than men with the same type or women with other MS types (Jones et al, 2012). These findings highlight the need for future research to explore how MS-specific factors, such as disease type and progression, impact the effectiveness of exercise protocols for anxiety and to consider prescreening participants for elevated anxiety levels to optimize outcomes.

Limitations

This review has several limitations that should be considered when interpreting the findings. First, only studies published in English were included, potentially excluding relevant research in other languages and introducing language bias. Furthermore, the control groups varied across studies; while some participants engaged in active interventions, such as alternative exercise routines or relaxation activities, others participated in no exercise at all. Although this allowed comparisons between PAG-MS-compliant and non-PAG-MS-compliant interventions, it limits the ability to determine the true effectiveness of PAG-MS independently. Future studies should adopt standardized non-PAG-MS control groups to provide clearer insights.

Generalizability is another limitation, as most studies primarily recruited participants with RRMS, with a limited representation of SPMS and PPMS. These subtypes, with their distinct symptom profiles and psychological burdens, may respond differently to exercise interventions, and their underrepresentation may affect the applicability of findings. Furthermore, while the use of calculated MCID thresholds is a strength of this review, the lack of established MCIDs for some tools necessitated the use of estimated thresholds. This introduces some uncertainty into the evaluation of clinical significance. Stronger methods of calculating MCIDs, such as anchor-based or distribution-based approaches, or a combination of these methods with adequately powered sample sizes, may provide more accurate and generalizable estimates for mental health measures.

While a few studies demonstrated low risk of bias, two studies raised some concerns, and seven were assessed as having high risk of bias, particularly in domains such as randomization and missing outcome data. These methodological limitations likely contributed to the observed heterogeneity and reduced the reliability of pooled results. Although sensitivity analyses based on study quality can offer additional insights, conducting such analyses in the present review was limited by the small number of available studies meeting PAG-MS criteria. Subgrouping studies based on risk of bias would have further reduced the statistical power and reliability of the findings. As more RCTs that meet PAG-MS are conducted, future reviews should consider analyzing the data based on specific cut-off including risk-of-bias, type of interventions, and methodological qualities.

Another limitation is the small number of studies assessing anxiety outcomes. The lack of prescreening for elevated anxiety levels or targeting anxiety as a primary outcome likely diluted observed effects and contributed to the overall non-significant meta-analysis findings for anxiety. Additionally, many included studies had small sample sizes, which further reduced the robustness of the findings. Future research should prioritize larger, high-quality randomized controlled trials with standardized control groups, comprehensive prescreening for mental health outcomes, and consistent reporting of both statistical and clinical significance to strengthen the evidence base for PAG-MS-compliant exercise interventions in improving mental health outcomes in people with MS. Finally, it is important to note that exercise guidelines for PwMS may vary across regions, potentially influencing the observed effects. While this review followed the PAG-MS guidelines developed by Kim et al (2019) and Latimer-Cheung et al (2013), exercise recommendations in other parts of the world may differ in intensity, which could impact the outcomes of exercise interventions.

CONCLUSION

This systematic review and meta-analysis suggests that exercise interventions meeting PAG-MS significantly improve depression in PwMS, with statistically and clinically meaningful effects, as BDI scores exceeded the MCID. However, evidence for anxiety remains inconclusive due to the limited number of studies, lack of targeted designs, and variability in protocols. These findings support structured exercise as an effective non-pharmacological approach for improving depression in PwMS, while also highlighting gaps such as underrepresentation of non-RRMS subtypes, inconsistent outcome reporting, and methodological limitations.

Statement on Data Availability:

All data and search strategies can be found in supplementary appendix