Efficacy of a cognitive-behavioral digital therapeutic on psychosocial outcomes in rheumatoid arthritis: randomized controlled trial

Rheumatoid arthritis (RA) is a chronic inflammatory disease affecting about 0.6% globally and up to 1.6% in Germany, leading to joint damage and functional limitations. Psychosocial distress is common: approximately 25–40% of patients experience depression and up to 77% experience anxiety, with substantial overlap. Despite effective disease activity control with DMARDs, psychosocial outcomes often remain suboptimal and are associated with increased disease activity, reduced quality of life, work impairment, and higher mortality. Guidelines from EULAR, ACR, and the German Society for Rheumatology recommend multidisciplinary nonpharmacologic strategies, including CBT, which has shown benefits for depression, anxiety, and fatigue in RA. However, access to CBT is limited, patients report unmet needs and low self-management capabilities, and specialist shortages further restrict access. Digital therapeutics are proposed to address these gaps. The study developed and evaluated reclarit, a fully automated CBT- and lifestyle-based digital therapeutic, to test whether it improves psychosocial outcomes in adults with RA and impaired health-related quality of life compared to an active control.

The paper reviews evidence that psychosocial distress (depression, anxiety, fatigue) is prevalent in RA and predicts worse clinical and quality-of-life outcomes. Prior pharmacologic control of RA disease activity does not reliably improve psychosocial outcomes. International guidelines (EULAR, ACR) endorse nonpharmacologic strategies and self-management, with CBT demonstrating positive effects on depression, anxiety, and fatigue in RA in randomized trials and meta-analyses. Surveys highlight unmet patient needs in education and coping, limited self-management, and barriers due to time, training, and workforce shortages. Digital health and telehealth are recommended by EULAR to support self-management, but approved digital therapeutics for RA are lacking in Germany. This context motivates evaluating a CBT-based digital therapeutic to address psychosocial needs at scale.

Design: Prospective, randomized, single-blind, active-controlled clinical trial (German Clinical Trials Register DRKS00025256) in Germany. Participants were randomized 1:1 to reclarit plus treatment as usual (TAU) or TAU plus educational material (active control). The trial compared psychosocial and related outcomes at 3 and 6 months. Participants: Adults (≥18 years) with RA (ICD-10 M05–M06) verified by physician documentation, German language proficiency, internet access, and online informed consent. Exclusions: suicide risk, psychoses, or non-RA chronic inflammatory diseases (self-report). An ethics-approved protocol amendment on April 12, 2022 required impaired QoL at baseline (SF-36 MCS <50 and PCS <50) to ensure potential for improvement. Recruitment and timeline: Online recruitment via Google Ads and health insurance companies from April 29, 2021 to April 20, 2022; final data collection on November 10, 2022. The study adhered to the Declaration of Helsinki and was approved by the Ärztekammer Hamburg ethics committee (PV7382). Randomization and blinding: Simple 1:1 randomization via online sequence generation based on participant IDs. Allocation was concealed to all staff except the study manager (not involved in data collection/analysis). Participants were aware of their assignment; other staff were blinded. Interventions: reclarit (GAIA, Hamburg) is a fully automated, internet-based expert-system digital therapeutic delivering CBT and lifestyle counseling through interactive, tailored dialogues across six modules (education on chronic pain; stress-pain interaction; pain coping; healthy sleep; healthy lifestyle/diet; mental well-being). Content is presented via text, audio, and images; users make choices guiding personalized content, receive tasks (e.g., diet changes), worksheets, summaries, and tailored reminders via email/SMS. No software installation; accessible via responsive web design on desktop, tablet, and smartphone. Usable for 180 days. Control: PDF educational material on lifestyle in RA from the German Rheumatism League via email, plus TAU as per routine care (medications, psychotherapy, physiotherapy, or none). Data collection and outcomes: Patient-reported outcomes (PROs) collected online (www.easyfeedback.de) at baseline, 3 months, and 6 months. Primary endpoint: SF-36 (RAND) mental component summary (MCS) and physical component summary (PCS) at 3 months (intention-to-treat, ITT). Secondary endpoints: SF-36 at 6 months; depression (PHQ-9; 0–27); anxiety (GAD-7; 0–21); fatigue (BRAF-MDQ; 0–70); social/work functioning (WSAS; lower scores better); pain (0–10 NRS); physical function (HAQ-DI; lower scores better) at 3 and 6 months. Post-hoc: user satisfaction via Net Promoter Score (NPS). Populations: ITT included all participants providing data; per-protocol (PP) included reclarit users completing ≥2 modules plus all control participants. Sample size: A priori power analysis (G*Power v3.2.9.4) indicated 352 participants for 0.8 power for the primary outcome; no attrition correction due to planned multiple imputation. Statistical analysis: Primary ITT analyses used multiple imputation under missing-at-random, with ANCOVA at 3 months (outcome at 3 months as dependent variable; baseline as covariate). Treatment effects reported as baseline-adjusted mean group differences with 95% CI; two-sided p values used. Bonferroni correction applied for MCS and PCS (p<0.025 significant). Other outcomes used p<0.05. Cohen’s d based on observed group mean differences. Similar analyses performed for 6 months and PP. Group differences in concomitant therapies at 3 and 6 months tested via chi-square. Post-hoc responder analyses: SF-36 MCS responders via reliable change index (RCI) with z≥1.96 at 3 months; depression responders defined as ≥5-point decrease in PHQ-9; anxiety responders defined as ≥4-point decrease in GAD-7. Subgroup exploratory ANCOVA by age, gender, DMARD use, physiotherapy, psychotherapy. Analyses conducted in R 4.2.1. Data security: Data pseudonymized, stored encrypted on secure servers; personal identifiers accessible only to the study manager. Participants received €10 Amazon vouchers after completing 3- and 6-month questionnaires; control participants offered reclarit after study completion.

- Enrollment and sample: 2221 expressed interest; 1484 consented/screened; 354 randomized (reclarit n=170; control n=184). At 3 months, loss to follow-up was 31% (reclarit) vs 21% (control), with stable group sizes to 6 months. PP included 149/170 (87.6%) reclarit users (completed ≥2 modules) and all controls. - Primary outcome (SF-36 MCS): At 3 months (ITT), reclarit improved MCS vs control with an adjusted mean difference of 3.3 points (95% CI 0.7, 5.9; p=0.014; Cohen’s d=0.23). Effects were sustained at 6 months (3.3 [0.5, 6.2]; p=0.020; d=0.23) and similar in PP (3 months: 3.6 [0.8, 6.5]; p=0.013; d=0.29; 6 months: 3.3 [0.4, 6.3]; p=0.025; d=0.26). - Primary outcome (SF-36 PCS): No significant between-group differences at 3 or 6 months in ITT or PP (e.g., ITT 3 months: 1.1 [−0.9, 3.1]; p=0.269; d=0.04). - Responder analyses: SF-36 MCS responders at 3 months were higher with reclarit vs control (36.7% vs 27.5%; p=0.019). - Secondary outcomes (3 months, ITT): Depression (PHQ-9) improved (−1.7 [−2.6, −0.8]; p<0.001; d=0.30); Anxiety (GAD-7) improved (−1.3 [−2.1, −0.4]; p=0.003; d=0.23); Fatigue (BRAF-MDQ) improved (−4.0 [−6.6, −1.4]; p=0.003; d=0.21); Social/work functioning (WSAS) improved (−1.8 [−3.4, −0.2]; p=0.028). Pain (NRS) and physical function (HAQ-DI) showed no significant differences. - Secondary outcomes (6 months, ITT): Improvements maintained for depression (−1.1 [−2.0, −0.1]; p=0.027), fatigue (−4.2 [−6.9, −1.5]; p=0.002), and WSAS (−1.9 [−3.6, −0.2]; p=0.029). Anxiety trended but was not statistically significant (−0.8 [−1.7, 0.02]; p=0.054). Pain and HAQ-DI remained non-significant. - Post-hoc responder outcomes: Depression responders (PHQ-9 ≥5-point decrease) at 3 months were 25.6% vs 12.9% (p=0.002); Anxiety responders (GAD-7 ≥4-point decrease) were 25.3% vs 16.4% (p=0.048). - Safety: No adverse events or adverse device effects observed. Among completers, fewer reclarit participants had lower QoL at 3 months vs baseline compared with control (MCS: 34.1% vs 43.8%, p=0.143; PCS: 24.7% vs 36.3%, p=0.159). - User satisfaction: NPS at 3 months was +19.8 (47.4% promoters; 27.6% detractors), indicating good satisfaction. - Concomitant treatments: No significant between-group differences in anti-rheumatic medications, physiotherapy, or psychotherapy during follow-up (chi-square tests), supporting that effects were attributable to reclarit. - Subgroups: No significant interactions by age, gender, DMARD use, physiotherapy, or psychotherapy for SF-36 MCS or PCS effects.

The trial demonstrates that adding the fully automated digital therapeutic reclarit to usual care produces clinically relevant improvements in psychosocial outcomes for adults with RA and impaired QoL. The primary endpoint showed a significant and sustained benefit on mental health-related quality of life (SF-36 MCS), with responder analyses corroborating clinical meaningfulness. Secondary outcomes showed consistent improvements in depression, anxiety (short-term), fatigue, and social/work functioning, aligning with effect sizes from meta-analyses of conventional psychological interventions in RA. Benefits occurred without concurrent improvements in physical QoL, pain, or disability, indicating reclarit specifically targets psychosocial domains, as expected for a CBT-based intervention. Exploratory analyses suggest efficacy across demographic and treatment subgroups, supporting broad applicability among RA patients engaged in standard care. The absence of adverse events and favorable user satisfaction underscore safety and acceptability. Given limited access to face-to-face CBT and workforce constraints, these results support digital therapeutics as a scalable, evidence-based means to integrate mental health support into holistic RA management. The possibility that earlier use (closer to diagnosis) might yield larger benefits is raised by prior literature indicating greater psychological intervention effects in shorter disease duration.

In adults with RA and impaired health-related QoL, reclarit plus TAU significantly improved mental health-related QoL at 3 months with sustained effects at 6 months and yielded meaningful gains in depression, fatigue, anxiety (short-term), and social/work functioning versus an active control. No differences were observed for physical QoL, pain, or disability, and no safety concerns emerged. Because concomitant treatments did not differ between groups, benefits can be attributed to reclarit. The findings support integrating reclarit within holistic, hybrid RA care to address unmet psychosocial needs and optimize outcomes. Future work should evaluate performance in routine care, preventive applications (including earlier disease stages), include clinician-assessed disease activity and patient global assessment, assess self-efficacy and pain catastrophizing, and conduct economic evaluations.

- Generalizability: Inclusion required impaired QoL (SF-36 MCS and PCS <50), so findings may not generalize to less-impaired or preventive populations. - Subgroup sizes: Small numbers in some subgroups (e.g., men, participants >65 years) limit power for interaction analyses. - Outcomes scope: Some relevant measures (self-efficacy, pain catastrophizing, patient global assessment) and physician-derived outcomes (e.g., DAS28) were not collected due to the online design and efforts to minimize burden. - Attrition: Higher dropout in the intervention arm; expected attrition was not incorporated into initial sample size assumptions, which may affect precision, though multiple imputation was used. - Digital access: Individuals without reliable internet access or unwilling to use digital tools may be underserved, potentially limiting reach. - Data availability: Datasets are proprietary and not publicly available, which may limit external verification.