Effectiveness of a digital health application (levidex) on quality of life in people with multiple sclerosis: A pragmatic, randomized controlled trial (LAMONT)

Abstract

Background

People with multiple sclerosis (pwMS) often experience impaired quality of life (QoL) despite receiving standard care. Digital therapeutics (DTx) may offer support, but prior trials yielded mixed results, possibly due to active controls and high baseline QoL.

Objectives

We evaluated a DTx (levidex) as an adjunct to treatment as usual (TAU) in pwMS with impaired QoL.

Methods

In this pragmatic, online randomized controlled trial (LAMONT; NCT06090305), n = 470 pwMS with a score ≥ 2 on the Hamburg Quality of Life Questionnaire in Multiple Sclerosis (HAQUAMS) were randomized to levidex + TAU or TAU alone. The primary endpoint was HAQUAMS total score at 6 months, analysed by intention-to-treat analysis of covariance.

Results

Compared with TAU, levidex + TAU improved multiple sclerosis-specific QoL at 6 months (baseline-adjusted mean difference −0.10; 95% CI −0.18 to −0.03; p = 0.008; Cohen's d = 0.26). Clinically relevant HAQUAMS improvement (≥0.22) occurred more often with levidex (39.5% vs. 27.8%; number needed to treat = 9). Benefits also emerged for depressive symptoms and social/work functioning but not for anxiety. No serious adverse events occurred, and user satisfaction was high.

Conclusions

In pwMS with impaired QoL, adding the scalable DTx levidex to TAU yields meaningful improvements in QoL and functioning.

Introduction

Multiple sclerosis (MS) is a chronic, inflammatory, neurodegenerative disease affecting an estimated 2.8 million people globally. ¹ Beyond sensory and motor symptoms, fatigue, cognitive difficulties, and emotional distress profoundly impact health-related quality of life (QoL). ² While disease-modifying drugs (DMDs) can reduce inflammation and relapses, they do not directly address these psychosocial and emotional aspects, which are sometimes termed ‘invisible symptoms’. ³ Accordingly, guidelines increasingly recommend holistic approaches, including non-pharmacological approaches such as cognitive behavioural therapy (CBT) and health-behaviour interventions (e.g. physical activity, diet, and stress management) to improve QoL.^4,5

Access to psychological and behavioural support is limited for many people with MS (pwMS) due to therapist shortages, geographical barriers, and disability-related constraints such as reduced mobility and fatigue. ⁶ Digital therapeutics (DTx) may help by delivering evidence-based interventions remotely, although efficacy and safety need to be established for each program. ⁷ Despite a growing DTx landscape, few interventions are designed to address the multi-domain needs of pwMS.

levidex is a DTx tailored for pwMS and is evaluated in this randomized controlled trial (RCT). The program's development and initial pilot testing confirmed its relevance, trustworthiness, and perceived utility for inspiring healthy lifestyle changes, as reported by pwMS and clinicians. ⁸ Two previous RCTs investigated the program's efficacy, but with notable limitations. The first compared levidex to web-adapted informational material, finding significant but small improvements in QoL, sick leave, and daily living activities. ⁹ However, the trial suffered from high drop-out and, crucially, its active control group (CG) precluded comparison to standard treatment as usual (TAU). The second RCT compared levidex to a non-personalized active psychoeducational program (‘dexilev’) in recently diagnosed pwMS and failed to meet its ambitious primary clinical endpoint (time to relapse or new lesion on magnetic resonance imaging), and no differential QoL effects emerged, possibly due to the sample's only minor baseline QoL impairments. ¹⁰

Both RCTs were limited by the inclusion of participants with high baseline QoL, creating potential ceiling effects. Furthermore, neither trial could determine the effect of levidex as an add-on to TAU, which reflects its intended real-world use. Despite the ambiguous QoL findings in the second trial, related quantitative and qualitative data suggest levidex users report greater health-behaviour changes (e.g. diet and stress management) than controls.^11,12

Previous research, therefore, left two critical questions unanswered: (1) the efficacy of levidex as an adjunct to true TAU (instead of an active control), and (2) its impact on an MS population with pre-existing QoL impairment, thus avoiding ceiling effects.

The current trial (the LAMONT trial) addresses these gaps. We aimed to evaluate whether adding levidex to TAU improves MS-specific health-related QoL at 6 months compared with TAU alone in a population screened for impaired QoL. We hypothesized that levidex + TAU would be superior to TAU alone for the primary QoL outcome and for secondary outcomes, including mood, anxiety, functioning, and health utilization.

Methods

Trial design

We conducted a randomized, parallel-group superiority trial (LAMONT; NCT06090305) comparing levidex + TAU versus TAU alone. The primary endpoint was MS-specific QoL at 6 months (Hamburg Quality of Life Questionnaire in MS (HAQUAMS)). ¹³ Secondary endpoints included depressive symptoms, anxiety, social/work functioning, MS-specific QoL (alternate instrument), and instrumental activities of daily living.

Eligibility criteria

Eligible participants were ≥18 years with physician-confirmed MS (ICD-10-GM G35.x), impaired QoL (HAQUAMS total ≥2; a cut-off chosen to capture patients with at least moderate disability, approximating an Expanded Disability Status Scale score ≥ 3), specialist MS care within 3 months, and ability to use an online programme. Additional criteria were German language proficiency, internet access, and written informed consent. An exclusion criterion was severe dependence (Pflegegrad, §15 SGB XI ≥ 3).

Study setting and data collection

The trial was conducted fully remotely (9 November 2023–6 December 2024). Participants were recruited via online campaigns/newsletters and physician referrals to a study website. Data were collected via LimeSurvey ¹⁴ using range/validity checks; study procedures and data integrity were monitored regularly with routine backups.

Interventions

CG

CG participants received TAU, reflecting routine outpatient care (e.g. primary/specialist care, DMDs/other medication, psychotherapy, or no active treatment). All CG participants also received a brief digital information sheet with publicly available MS resources (e.g. German MS Society). TAU was not standardized, allowing for changes in treatment during follow-up in line with real-world practice.

Intervention group (IG): levidex

Following randomization, participants in the IG received access to levidex adjunctively to TAU. This is a web-based DTx for pwMS, developed, owned, and operated by GAIA, a small-to-medium enterprise specializing in DTx. As described previously, it was developed in collaboration with a multidisciplinary team of neurologists, clinical psychologists, health scientists, and nutritionists. ⁸ The intervention is delivered via standard internet browsers on common devices (computer, tablet, and smartphone) and is implemented on GAIA's proprietary broca^® platform, which uses rule-based algorithms to generate personalized ‘simulated dialogues’. Users progress through modules by selecting predefined responses; subsequent content is algorithmically tailored to user characteristics and preferences. The platform's dialogue-based CBT approach has been shown to be effective in previous RCTs of other DTx, including for MS-related fatigue and depression.^15,16

levidex comprises 16 sequential sessions (‘conversations’; ∼30–45 min each) covering: (1) MS self-management/education, (2) psychological strategies for emotional well-being, (3) dietary approaches, and (4) physical activity and sleep. Sessions are delivered over several months, with new content released successively after predefined waiting periods to encourage reflection and completion of between-session activities. The programme follows a structured progression, with early sessions focusing on foundational knowledge and skills, followed by modules supporting integration into daily routines and later sessions consolidating and reinforcing behaviour change.

Each session typically addresses a specific thematic focus rather than combining multiple domains, and includes both in-programme exercises (e.g. planning, mindfulness, and relaxation or imagery-based audio exercises) and between-session activities (e.g. activity plans and preparing specific foods). Progression is guided by user responses within each session, allowing content to be tailored to individual needs and preferences. Optional message reminders support engagement and adherence. The programme is primarily CBT-based and integrates common behaviour change techniques, such as cognitive restructuring, behavioural activation, goal-setting, action planning, self-monitoring, provision of information about health consequences, and prompts to support health-promoting behaviours. ¹⁷ In addition, levidex incorporates mindfulness- and acceptance-based components, mental imagery exercises, and elements of motivational interviewing, and is aligned with the concept of patient empowerment. ¹⁸ Overall, levidex was designed with the aim of supporting pwMS in optimizing health-related behaviours (e.g. physical activity, dietary behaviours, stress management, and sleep) and improving health-related QoL. Representative screenshots illustrating the structure, content, and interactive dialogue format of the intervention are shown in Figure 1.

Figure 1.

Example screenshots of the levidex digital therapeutic. (A) Interactive self-assessment in the domain of nutrition. (B) Psychoeducational sequence introducing physical activity concepts with selectable response options. Screens are representative excerpts from the intervention and have been translated into English for illustration purposes. MS: multiple sclerosis.

Outcomes

Outcome data were collected at 3 months (T1) and 6 months (T2) after randomization, with the 6-month assessment pre-specified as the primary evaluation time point to meet regulatory requirements for permanent inclusion in the German Digital Health Applications (DiGA) directory. The 3-month assessment was included to explore early effects. After completion of the T2 assessment, participants in the CG were offered access to levidex.

Primary outcome

The primary endpoint was MS-specific health-related QoL at 6 months (HAQUAMS total score). ¹³

Secondary outcomes

Secondary endpoints included depressive symptoms (Patient Health Questionnaire-9 (PHQ-9)), ¹⁹ social and work-related functioning (Work and Social Adjustment Scale (WSAS)), ²⁰ MS-specific health-related quality of life (global index score of the MS International Quality of Life questionnaire (MusiQoL)), ²¹ anxiety symptoms (Generalized Anxiety Disorder Scale-7 (GAD-7)), ²² and instrumental activities of daily living (Frenchay Activities Index (FAI)). ²³

Exploratory outcomes

Exploratory endpoints included HAQUAMS subscales, overall DMD intake in the previous 3 months, and DMD intake classified according to efficacy (categories 1–3 per current German guideline). Additional exploratory endpoints were health care utilization indicators (number of days on sick leave/sick pay and the number of days in inpatient treatment during the previous 3 months), and user satisfaction (assessed by the Net Promoter Score (NPS) ²⁴ and the ZUF-8 patient satisfaction questionnaire ²⁵ ), a validated 8-item measure with total scores ranging from 8 to 32, with higher scores indicating greater satisfaction).

Harms

Adverse events (AEs) were assessed via active queries at each assessment time point (3 and 6 months) and through spontaneous reporting. AEs were defined as unplanned or emergency outpatient or inpatient treatments or patient-reported deterioration in MS-specific quality of life. Participants could report events via email, telephone, or a platform-based contact form. All reports were collected and monitored centrally by the study management team.

Sample size

As pre-specified in the clinical investigation plan and trial registration, the trial aimed to include 470 pwMS, randomized in a 1:1 allocation ratio to receive levidex plus TAU or TAU alone. Reanalysis of data from a previous levidex RCT ⁹ restricted to patients with impaired QoL (HAQUAMS ≥ 2) yielded an effect size of Cohen's d = 0.28 for the HAQUAMS total score; assuming a slightly larger expected effect of d = 0.30, a total of 352 participants (176 per group) would provide 80% power at α = 0.05. Allowing for 25% attrition, we therefore aimed to recruit N = 470 participants.

Randomization

Participants were allocated 1:1 to the IG or the CG using simple randomization via an external, web-based computerized tool. The allocation sequence was generated using computer-produced random numbers and was concealed from study staff; group assignment was released only after completion of baseline assessments. No blocking or stratification factors were used.

Blinding

Due to the nature of the web-based intervention, neither participants nor investigators were blinded to group allocation. All outcomes were assessed via self-administered online questionnaires.

Statistical methods

Details of the statistical analyses were pre-specified in the clinical investigation plan. The primary analysis followed the intention-to-treat (ITT) principle. The primary endpoint (HAQUAMS at 6 months) was analysed using analysis of covariance (ANCOVA) with group as factor and baseline HAQUAMS as covariate; adjusted mean differences (95% CI) and p values are reported. Secondary continuous outcomes were analysed analogously; between-group effect sizes (Cohen's d) were calculated from baseline-adjusted mean differences. Intermediate treatment effects at 3 months were examined using the same ANCOVA framework. Exploratory categorical outcomes were analysed using χ² tests.

Missing outcomes were handled under a missing-at-random assumption using bootstrapped maximum-likelihood multiple imputation (R packages bootImpute and mice), ²⁶ with imputation models including the respective baseline outcome, treatment group, and relevant sociodemographic and clinical variables (age, sex, MS type, concomitant psychotherapy, and antidepressant use). A conservative reference-based sensitivity analysis (jump-to-reference imputation) was additionally performed for primary and secondary outcomes. ²⁷ Analyses were conducted using R (version 4.4.1), ²⁸ and two-sided p values <0.05 were considered statistically significant. To control for multiplicity across secondary outcomes, a hierarchical gatekeeping strategy was applied; if a test in the pre-specified sequence (PHQ-9, WSAS, MusiQoL, GAD-7, and FAI) was not significant, subsequent endpoints were considered exploratory.

Results

Participants and follow-up

Participants were enrolled between November 2023 and June 2024 and followed for 6 months, with the last patient's last visit on 6 December 2024. Numbers for screening, allocation, follow-up, and analysis are shown in Figure 2. All 470 randomized participants were analysed in their originally assigned groups according to the ITT principle. Reasons for drop-out were not systematically collected in this fully remote, pragmatic trial and are therefore not available.

Figure 2.

Participant flow.

MS: multiple sclerosis; IG: intervention group; TAU: treatment as usual; CG: control group; ITT: intention-to-treat.

Although allocation was planned at a 1:1 ratio, the final group sizes were 215 (intervention) and 255 (control), which is consistent with simple randomization without blocking. A two-sided binomial test indicated that this deviation from equal allocation did not differ significantly from chance (p = 0.072), indicating that the observed imbalance is compatible with random allocation. Baseline characteristics are summarized in Table 1 and indicate broadly comparable groups.

Table 1.

Baseline sociodemographic and clinical characteristics (N = 470).

Characteristic	Intervention (levidex + TAU) (n = 215)	Control (TAU) (n = 255)	Total (N = 470)
Demographics
Age, mean (SD)	45.58 (10.41)	47.51 (10.72)	46.63 (10.62)
Female, n (%)	188 (87.4%)	197 (77.3%)	385 (81.9%)
Education, n (%)
Secondary school or less	25 (11.6%)	39 (15.3%)	64 (13.6%)
High school/vocational	86 (40.0%)	122 (47.8%)	208 (44.2%)
University degree	104 (48.4%)	94 (36.9%)	198 (42.1%)
Clinical characteristics
MS type, n (%)
RRMS	138 (64.2%)	154 (60.4%)	292 (62.1%)
PPMS	25 (11.6%)	48 (18.8%)	73 (15.5%)
SPMS	42 (19.5%)	44 (17.3%)	86 (18.3%)
MS duration (years), mean (SD)	11.91 (9.49)	11.69 (10.05)	11.79 (9.79)
Stage of disability (PDDS), n (%)
Normal	12 (5.6%)	12 (4.7%)	24 (5.1%)
Mild disability	36 (16.7%)	36 (14.1%)	72 (15.3%)
Moderate disability	58 (27.0%)	67 (26.3%)	125 (26.6%)
Gait disability	46 (21.4%)	55 (21.6%)	101 (21.5%)
Cane use or higher	63 (29.3%)	85 (33.4%)	148 (31.4%)
Concomitant care
Currently in psychotherapy, n (%)	55 (25.6%)	55 (21.6%)	110 (23.4%)
Currently taking antidepressants, n (%)	35 (16.3%)	44 (17.3%)	79 (16.8%)
Currently taking any DMD, n (%)	117 (54.4%)	144 (56.5%)	261 (55.5%)
Baseline outcome scores
HAQUAMS total score, mean (SD)	2.72 (0.52)	2.68 (0.47)	2.70 (0.49)
PHQ-9 total score, mean (SD)	11.34 (4.44)	11.05 (4.38)	11.19 (4.41)
WSAS total score, mean (SD)	18.94 (8.12)	18.51 (8.34)	18.71 (8.23)
MusiQoL global index, mean (SD)	55.02 (13.83)	55.57 (12.81)	55.32 (13.27)
GAD-7 total score, mean (SD)	9.01 (4.96)	7.71 (4.17)	8.30 (4.59)
FAI total score, mean (SD)	29.01 (7.44)	28.47 (7.74)	28.72 (7.60)

Data are n (%) or mean (SD). All data are from baseline (T0). TAU: treatment as usual; MS: multiple sclerosis; RRMS: relapsing–remitting MS; PPMS: primary progressive MS; SPMS: secondary progressive MS; PDDS: patient-determined disease steps; DMD: disease-modifying drug; HAQUAMS: Hamburg Quality of Life Questionnaire for MS; PHQ-9: Patient Health Questionnaire-9; WSAS: Work and Social Adjustment Scale; MusiQoL: MS International Quality of Life; GAD-7: Generalized Anxiety Disorder-7; FAI: Frenchay Activities Index; SD: standard deviation.

Primary outcome

In the ITT analysis, participants in the IG showed better MS-specific health-related QoL at 6 months than those in the CG (Table 2). The baseline-adjusted between-group difference on the HAQUAMS total score at T2 was −0.10 points (95% CI −0.18 to −0.03; p = 0.008; Cohen's d = 0.26), with a similar effect in the sensitivity analysis (Supplemental Table 2). A higher proportion of pwMS in the IG achieved at least a minimal clinically important difference (MCID) of ≥0.22 points on the HAQUAMS than in the CG (39.5% vs. 27.8%; OR 1.69, 95% CI 1.15–2.50; p = 0.007; number needed to treat (NNT) = 9). Improvements were already apparent at 3 months (Supplemental Table 1).

Table 2.

Primary outcome (MS-specific QoL, HAQUAMS) at 6 months (ITT analysis).

Outcome	Group	N	Unadjusted M (SD)	Adjusted mean difference (95% CI)*	Cohen’s d	p-value
(A) Continuous outcome analysis
HAQUAMS total score	Control (TAU)	255	2.66 (0.58)	−0.10 (−0.18, −0.03)	0.26	0.008
	Intervention	215	2.60 (0.57)

Outcome	Group	N	Responders, n (%)	Odds ratio (95% CI)	NNT	p-value
(B) Responder analysis
HAQUAMS responder (≥0.22 improvement)	Control (TAU)	255	71 (27.8%)	1.69 (1.15, 2.50)	9	0.007
	Intervention	215	85 (39.5%)

Data are from the 6-month (T2) ITT analysis. MS: multiple sclerosis; QoL: quality of life; HAQUAMS: Hamburg Quality of Life Questionnaire for MS; ITT: intention-to-treat; M: mean; SD: standard deviation; 95% CI: 95% confidence interval; TAU: treatment as usual; NNT: number needed to treat.

*Adjusted for baseline HAQUAMS score. A negative difference favours the intervention group.

Secondary outcomes

At 6 months (Table 3), levidex plus TAU yielded greater improvements than TAU alone in depressive symptoms (PHQ-9; baseline-adjusted mean difference −0.8 points, 95% CI −1.4 to −0.1; p = 0.025; d = 0.21), social and work functioning (WSAS; −1.8 points, 95% CI −2.9 to −0.6; p = 0.003; d = 0.30), and MS-specific health-related QoL assessed with the MusiQoL global index (2.1 points, 95% CI 0.3 to 3.9; p = 0.020; d = 0.23). Anxiety symptoms were not significantly different (GAD-7; baseline-adjusted mean difference −0.5 points, 95% CI −1.2 to 0.2; p = 0.205; d = 0.13). In line with the gatekeeping strategy, subsequent endpoints were treated as exploratory; for these outcomes, including instrumental activities of daily living (FAI; 0.8 points, 95% CI −0.1 to 1.8; p = 0.097; d = 0.17), no consistent intervention effects were detected. Responder analyses for secondary outcomes are available in Supplemental Table 3; results for exploratory HAQUAMS subscales and health utilization metrics in Supplemental Tables 4 and 5.

Table 3.

Secondary outcomes at 6 months (ITT analysis).

Outcome	Adjusted mean difference (95% CI)*	p-value	Cohen's d
Depressive symptoms (PHQ-9)	−0.8 (−1.4, −0.1)	0.025	0.21
Social & work functioning (WSAS)	−1.8 (−2.9, −0.6)	0.003	0.30
MS-specific QoL (MusiQoL)	2.1 (0.3, 3.9)	0.02	0.23
Anxiety symptoms (GAD-7)	−0.5 (−1.2, 0.2)	0.205	0.13
Instrumental activities (FAI)†	0.8 (−0.1, 1.8)	0.097	0.17

Data are from the 6-month (T2) ITT analysis. All differences are adjusted for their respective baseline scores. ITT: intention-to-treat; PHQ-9: Patient Health Questionnaire-9; WSAS: Work and Social Adjustment Scale; MS: multiple sclerosis; MusiQoL: MS International Quality of Life; GAD-7: Generalized Anxiety Disorder-7; FAI: Frenchay Activities Index.

*For PHQ-9, WSAS, and GAD-7, a negative difference favours the intervention. For MusiQoL and FAI, a positive difference favours the intervention.

†Considered exploratory per the pre-specified gatekeeping strategy, as the analysis for anxiety symptoms (GAD-7) was not statistically significant.

Intervention adherence

All participants randomized to IG registered to use levidex. Over 6 months, participants completed a mean of 10.0 (standard deviation (SD) = 5.0) of the 16 available modules and had 25.1 (SD = 23.7) days of active use. Completion rates declined gradually across sessions, from approximately 97% for the first session to 25% for the final session (Supplemental Table 12).

User satisfaction

User satisfaction was assessed in the IG at 3 and 6 months. The NPS was 35.6 at T1 and 31.5 at T2, indicating very high user satisfaction (on a scale of −100 to 100). This was supported by the ZUF-8 patient satisfaction questionnaire, where the mean total score was 25.5 (SD = 4.1) at T1 and 25.3 (SD = 5.0) at T2, indicating high satisfaction. This corresponds to a mean item score of 3.2 at both time points (item range 1–4, higher scores indicating greater satisfaction).

Ancillary analyses

Exploratory subgroup analyses of the primary outcome (HAQUAMS total score at 6 months) were conducted by sex, age, MS type, psychotherapy status, antidepressant use, DMD use, and educational level at baseline. Overall, effects of levidex were in the same direction across most subgroups, with statistically significant improvements observed in women but not men and in participants taking antidepressants or not receiving DMD treatment at baseline, whereas effects were smaller and non-significant in those on DMDs or not using antidepressants (Supplemental Table 6 for full subgroup data). Exploratory analyses of DMD use, including subgroup comparisons (any DMD vs. none) and categorization by efficacy level, did not show consistent differences between groups (Supplemental Table 7). When adjusting for age, sex, and education level, the intervention effect on the primary outcome also remained significant (Supplemental Table 8). Further exploratory subgroup analyses suggested larger effects among participants without a university degree, among women, and among older (>40 years) participants (Supplemental Tables 9–11).

Harms

No AEs, adverse device effects, device deficiencies or serious AEs were observed during the trial.

Discussion

This pragmatic, online RCT found that adding the DTx levidex to TAU improved QoL in adults with MS. levidex produced small but statistically significant and clinically relevant MS-specific QoL improvements at 3 and 6 months compared with TAU alone (NNT = 9 for HAQUAMS improvement). Additional benefits were observed for depressive symptoms, social/work functioning, and MS-specific QoL on MusiQoL; no significant effects were found for anxiety or instrumental activities of daily living. The observed improvements in social and work functioning further suggest that these effects extend beyond symptom change to aspects of everyday functioning. The intervention was well tolerated, with high user satisfaction (NPS at 6 months = 31.5). These findings suggest that this automated programme can yield meaningful patient-reported gains as an adjunct to routine care.

A key implication is that this automated, multi-domain programme can achieve clinically relevant improvements despite the challenges of sustained behaviour change in MS. The small-to-moderate effect sizes are comparable to those from more resource-intensive psychosocial interventions, while scalability and ease of implementation are substantially higher.^29,30 While the mean between-group difference was modest, the responder analysis provides a more clinically intuitive perspective: nearly 40% of participants in the IG achieved a clinically meaningful improvement compared with 28% in the CG (NNT = 9). This suggests that a substantial proportion of patients experienced benefits likely to be noticeable in everyday life. In the context of a fully automated, low-threshold intervention, such effects may be particularly relevant at the population level, where scalability can translate small individual gains into meaningful public health impact. Future work should examine whether enhanced tailoring and optional stepped support (including human guidance for those with greater needs) can increase benefit while preserving digital-first efficiency.

An earlier levidex trial in a recently diagnosed MS cohort with only mild QoL impairment (baseline HAQUAMS about 1.7–1.8) and therefore limited room for improvement, focused on inflammatory activity and showed no advantage over an active control and no QoL difference. ¹⁰ In contrast, another trial that recruited pwMS with more pronounced QoL impairment (baseline HAQUAMS about 2.5) reported a small but significant effect on QoL. ⁹ The present study likewise enrolled pwMS with impaired QoL (baseline HAQUAMS 2.6) and found a comparable effect size. Although this mean difference is below the HAQUAMS MCID of 0.22 at the group level, almost 40% of pwMS in the levidex group achieved at least an MCID-level improvement compared with 28% under TAU (NNT = 9). Taken together, the trials suggest that levidex can serve as a useful adjunctive tool for pwMS experiencing reduced QoL, including when used alongside pharmacological treatment.

In Germany, the programme has been permanently included in the national DiGA directory based on this pivotal trial, making it prescribable and reimbursable. ³¹ However, reimbursement alone may not ensure uptake, and implementation efforts may be needed. This context underscores the relevance of these findings, as many pwMS report impaired QoL, residual symptoms, and difficulty maintaining healthy behaviours despite DMDs. ³ This self-guided, low-threshold programme may help to narrow the gap between pharmacological treatment and the daily demands of living with MS. levidex could serve as an early, scalable adjunctive option, with more intensive interventions reserved for non-responders.

The mechanism of levidex remains an important question. The multitarget approach (mood, coping, physical activity, diet, and stress) suggests combined pathways rather than a single ‘active ingredient’. With self-reported outcomes and no objective behavioural indicators, mediation by behaviour change cannot be determined; future work should incorporate behavioural measures and mediation analyses.

Despite notable strengths, such as low attrition, recruitment of patients with impaired QoL, and the use of a true TAU comparison, several limitations must be acknowledged. First, recruitment may have attracted motivated, digitally literate pwMS. While this limits generalizability to the broader MS population, it reflects the subgroup most likely to engage with and benefit from DTx in routine care. Although the IG included a higher proportion of participants with a university degree, exploratory subgroup analyses did not suggest that higher educational level conferred greater benefit. If anything, treatment effects were numerically larger among participants without a university degree, although these exploratory analyses were not powered for definitive conclusions. A second limitation was that all outcomes were self-reported, lacking objective verification (e.g. accelerometry). While this introduces a potential for reporting bias, patient perception is arguably the most valid and direct measure for subjective constructs such as QoL and depressive symptoms. Moreover, recent evidence suggests that self-report instruments in behavioural intervention trials do not necessarily overestimate effects and may even be more conservative than clinician assessments. ³² Consistent with this, a process evaluation of a levidex trial showed that meaningful behaviour changes were often reported qualitatively but not fully reflected in standard questionnaires, indicating that commonly used outcome measures may lack sensitivity for lifestyle-related changes. ¹² Third, the 6-month follow-up limits conclusions regarding sustainability and longer-term clinical outcomes. Objective clinical endpoints such as relapse activity were not assessed because relapse ascertainment is challenging in a fully remote pragmatic trial and cannot be reliably based on self-report alone. Longer-term follow-up, including extension studies, will be important to determine whether the observed benefits persist, increase, or attenuate over time. In this context, usage declined over the course of the programme, as is typical for digital interventions. The present data do not allow us to determine whether declining use reflected reduced motivation, competing demands, sufficient perceived benefit, or other reasons; future studies should examine determinants of sustained engagement more directly. Fourth, the unblinded, TAU-alone control design cannot separate specific intervention effects from non-specific factors such as expectancy, engagement, or study participation. As all outcomes were self-reported, it is possible that part of the observed effect reflects expectations associated with receiving an active intervention. However, such influences are difficult to avoid in behavioural intervention trials and may themselves contribute to the overall effectiveness of low-intensity, real-world interventions. ³³ In addition, TAU was heterogeneous and could change over time, which reflects real-world clinical practice but may introduce variability that cannot be fully controlled. Fifth, while exploratory analyses did not indicate consistent effects of DMD use, the potential influence of specific pharmacological treatments cannot be fully excluded. Finally, the German-speaking cohort requires replication in other settings, and subgroup analyses were not powered for definitive conclusions.

In conclusion, adding levidex to TAU yielded small but meaningful improvements in MS-specific QoL, mood, and functioning in adults with MS and impaired QoL. While effects are modest, scalability and low demands on clinical resources suggest population-level value. Future research should refine implementation in stepped-care models, test longer-term and objective outcomes, and evaluate whether enhanced tailoring or added support improves effectiveness.