Digital interventions for improving quality of life and physical function in adults aged 50 and over living with rheumatic and musculoskeletal diseases

Objectives

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To evaluate the effectiveness of digital health interventions, compared with usual care, non‐digital structured intervention, or no intervention, in improving quality of life and physical function among community‐dwelling adults (i.e. people living in the community) aged 50 years or over, diagnosed with osteoarthritis, osteoporosis, low‐back pain, rheumatoid arthritis, or polymyalgia rheumatica.

Background

Description of the condition

Rheumatic and musculoskeletal diseases (RMDs) are among the leading causes of disability and reduced quality of life worldwide, particularly in middle‐aged and older adults [1, 2]. These conditions, such as osteoporosis, osteoarthritis, low back pain, and rheumatoid arthritis, are often chronic, progressive, and closely associated with ageing [3, 4, 5, 6, 7].

Non‐inflammatory musculoskeletal diseases are strongly influenced by age‐related mechanisms such as bone and cartilage degeneration, hormonal decline, chronic low‐grade inflammation, and oxidative stress [6, 7], while age‐related inflammatory diseases reflect immunosenescence and immune system dysregulation in ageing [8]. Since these conditions share determinants related to ageing and have similar impacts on mobility, function, frailty and independence, analysing them within a unified framework offers insights that disease‐specific reviews fail to capture.

According to the World Health Organization (WHO) International Classification of Functioning, Disability, and Health, health changes in older adults significantly impact essential activities such as mobility, social interactions, and self‐care tasks like feeding and grooming [9]. This decline in activity often leads to what many older adults fear the most: the inability to participate in society, social isolation, and loss of independence. These factors significantly affect the quality of life of older adults [9].

Description of the intervention and how it might work

RMDs require consistent follow‐up and active self‐management to achieve optimal outcomes. However, these demands are becoming increasingly challenging due to rising patient numbers and limited healthcare resources [10, 11, 12].

Digital health interventions (DHIs) are health‐related programmes or services delivered through digital technologies, such as mobile applications, websites, wearable devices, or telehealth platforms. They offer scalable, patient‐centred alternatives to traditional care [13]. According to the WHO’s definition, digital health is a broad umbrella term encompassing eHealth (which includes telemedicine, clinical information systems, wearables and sensors, mobile apps, integrated networks), as well as emerging areas, such as the use of advanced computing sciences in big data, genomics, and artificial intelligence (AI) [14].

These technologies can support disease management and behavioural change through mechanisms such as delivering personalized education, facilitating remote monitoring (i.e. collecting health‐related information while people are at home, which may be reviewed by healthcare professionals or used to support self‐management), promoting treatment adherence, and encouraging healthier lifestyles. By improving access and fostering engagement, DHIs may empower individuals to manage their condition more effectively, helping to maintain physical function and quality of life [12, 15].

Moreover, integrating DHIs throughout the patient care pathway has the potential to support the quadruple aim of healthcare: 1) improving patient health, 2) improving patients’ experience of care, 3) improving the work experience of healthcare providers, and 4) reducing healthcare costs. This comprehensive approach not only addresses the needs of individuals but also contributes to more sustainable and efficient healthcare systems [12].

Despite the potential of DHIs for older adults with RMDs, their adoption presents several challenges. Common barriers include difficulties with digital literacy, sensory or motor limitations, unintuitive interfaces, privacy concerns, and the lack of adequate technical support. Limited digital literacy among family members – who are often essential in providing support – as well as the perception that technology may replace human interaction can also hinder engagement [16, 17]. On the other hand, simple tools, continuous support, and personalized content can facilitate acceptance and sustained use [16, 17]. These usability and accessibility considerations are particularly important to ensure that DHIs effectively benefit this population.

Why it is important to do this review

Over the past decade, wearable sensors, smartphone apps, teleconsultation platforms, and AI‐driven decision‐support tools have proliferated in rheumatology [12, 18, 19]. Observational studies and early trials suggest that these technologies can enhance treatment adherence, expand access to underserved populations, and enable real‐time monitoring [12, 18, 19]. However, the impact of these interventions on clinical practice and patients’ everyday lives remains uncertain due to several challenges.

• At the patient level, the digital health divide, encompassing differences in access to digital technologies and the internet, digital health literacy, and the availability of accessible and easy‐to‐use tools, may blunt the effectiveness of these interventions, particularly among older adults [12, 17, 19, 20].

• At the clinical level, integration into workflows is patchy; the absence of standardized guidance often leaves clinicians unsure how to act on remote data streams [12, 19, 21].

• At the organizational and technological level, limited interoperability between digital systems, insufficient IT infrastructure, and lack of technical support can undermine implementation and scalability in real‐world settings [12, 19].

• At the policy level, robust evidence of effectiveness and cost‐effectiveness is needed to justify large‐scale implementation and reimbursement [12, 19].

Importantly, DHIs have potential not only for improving disease outcomes but also for supporting autonomy, fostering active ageing, and transforming care delivery systems by addressing the quadruple aim of healthcare. The integration of digital health technologies throughout the entire patient pathway – from prevention and self‐management to monitoring and follow‐up – offers opportunities to align care with patients’ preferences and priorities, particularly in RMDs [12, 15].

Existing systematic reviews have been narrow in scope, typically focusing on a single disease or outcome (e.g. telerehabilitation for knee osteoarthritis, digital exercise programmes for chronic pain, or low‐back‐pain self‐management apps) [12, 19, 22, 23, 24, 25,26]. Although previous reviews sometimes include measures of quality of life or physical function, these outcomes are rarely the primary focus, and the existing literature does not offer an integrated synthesis of these outcomes across the most prevalent age‐related RMDs: osteoarthritis, osteoporosis, low‐back pain, rheumatoid arthritis and polymyalgia rheumatica. As the world’s population continues to age and the prevalence of RMDs is projected to rise accordingly, there is an urgent need for research approaches that are inclusive, feasible, and adapted to the realities of older adults [27]. Furthermore, these reviews seldom assess evidence certainty using GRADE and rarely involve patients in setting review priorities.

A recent systematic review evaluated the adoption, effectiveness and cost‐effectiveness of DHIs focused on inflammatory rheumatic diseases, including rheumatoid arthritis and systemic lupus erythematosus [28]. The review found that several telehealth, wearable technology and educational interventions significantly improved disease control, patient adherence, knowledge and self‐efficacy, with some cost‐effectiveness analyses also demonstrating favourable economic outcomes [28]. Our review differs from this work in terms of the target population and disease scope: we focus on the RMDs that are most prevalent in middle‐aged and older adults, include both inflammatory and non‐inflammatory conditions, and restrict the population to adults aged 50 years or older living in the community.

We performed a preliminary search on ClinicalTrials.gov and identified 17 ongoing studies related to this topic. Most of them are expected to be completed between late 2025 and 2026, with only two scheduled to end between 2028 and 2030. We will continue to monitor their progress throughout our work. No Cochrane review or protocol currently overlaps with this topic.

DHIs change quickly as technologies advance, software is updated, and clinical and regulatory contexts evolve. As a result, specific intervention features, modes of delivery, or effectiveness estimates evaluated in primary studies may change over time [29, 30]. The findings of this review should therefore be interpreted within the timeframe of the available evidence. Nevertheless, synthesising current evidence remains important to inform decision‐making and to guide future research in this rapidly evolving field.

The research team includes the president of the Portuguese League Against Rheumatic Diseases and a clinical expert. Additional League members will be engaged throughout the review process to support a Patient and Public Involvement and Engagement (PPIE) approach.

This review is therefore timely and necessary. It will accomplish the following.

• Synthesize evidence on the effects of digital health technologies on quality of life and physical function in middle‐aged and older adults with the most common age‐related RMDs.

• Fill methodological gaps by evaluating evidence certainty with GRADE and involving patients in defining priorities.

• Inform clinical guidelines and policy decisions on scaling digital care for older adults with RMDs, reflecting priorities identified by patients, clinicians, and the European Alliance of Associations for Rheumatology [18].

By explicitly addressing patient needs and bridging gaps in prior reviews, the findings of this review may offer decision‐makers up‐to‐date, patient‐centred evidence to guide the responsible deployment of digital health solutions in routine rheumatology care.

Objectives

To evaluate the effectiveness of digital health interventions, compared with usual care, non‐digital structured intervention, or no intervention, in improving quality of life and physical function among community‐dwelling adults (i.e. people living in the community) aged 50 years or over, diagnosed with osteoarthritis, osteoporosis, low‐back pain, rheumatoid arthritis, or polymyalgia rheumatica.

Methods

Criteria for considering studies for this review

Types of studies

We will include quantitative studies that assess the effectiveness of DHIs on quality of life or physical function in adults aged 50 and over living with a confirmed diagnosis of at least one of the selected rheumatic conditions.

Given that DHIs typically undergo a phased development process, in which feasibility and pilot studies are used to evaluate usability, acceptability, and preliminary signals of effectiveness prior to large‐scale trials, and that digital health technologies evolve rapidly, we will include non‐randomized quantitative study designs (non‐randomized studies of interventions; NRSIs) to capture complementary early‐phase and real‐world evidence that may not be fully reflected in randomized controlled trials (RCTs) [29, 30].

In accordance with methodological guidance, we will use evidence from NRSIs to complement and contextualise findings from RCTs, while restricting meta‐analysis to RCTs.

Eligible studies must have a clearly defined intervention and comparison group, with outcome data collected at baseline and at least one postintervention time point.

The following study designs will be eligible.

• RCTs, including cluster‐randomized trials

• Quasi‐randomized trials – for example, allocation by date of birth (even versus odd dates), day of the week, order of admission, or predetermined allocation by centre

• Controlled before‐and‐after studies (i.e. studies that compare outcomes in two groups before and after an intervention, where one group receives the intervention and the other serves as a comparison, to help assess whether observed changes are attributable to the intervention)

• Feasibility or pilot studies reporting data on quality of life or physical function (i.e. small, early‐stage studies conducted to assess whether an intervention or research approach is practical and workable before proceeding to a larger study)

We will exclude cross‐over trials, qualitative studies, case reports, protocols, conference abstracts, systematic reviews, and unpublished studies or those without accessible full‐text reports. Cross‐over trials will be excluded due to the risk of carry‐over effects and difficulties in interpreting outcomes that may persist beyond intervention periods.

Types of participants

We will include studies involving community‐dwelling adults aged 50 years or older with a confirmed diagnosis of at least one of the following RMDs.

• Inflammatory RMDs

  • Rheumatoid arthritis

  • Polymyalgia rheumatica

• Non‐inflammatory RMDs

  • Osteoarthritis

  • Osteoporosis, including fragility fractures

  • Low back pain

Community‐dwelling refers to individuals living in the community (e.g. in their own homes or other independent living settings), rather than in nursing homes, long‐term care facilities, or hospital settings. Individuals living in nursing homes, long‐term care facilities, or hospital settings are outside the scope of this review, and the findings may therefore not be applicable to these populations or to individuals with other RMD diagnoses not listed above.

We selected 50 years as the age threshold to capture midlife and older adults, consistent with international definitions of early older age and the increased prevalence of RMDs and digital exclusion in this age group [15]. To ensure that included samples adequately represent this target population, studies will only be eligible when all participants are aged 50 years or over, or when stratified data for participants aged 50 years or over are available, or when the overall mean or median age of the sample is 50 years or over.

We will exclude the following.

• Studies in which participants are institutionalized.

• Studies in which some participants are under 50 years of age and stratified data are not available for participants aged 50 years or over, or where the mean or median age is not over 50 years old.

• Studies including participants without a diagnosis of one of the specified RMDs.

• Studies focusing exclusively on participants undergoing or following major orthopaedic surgery for these conditions (e.g. total joint arthroplasty, spinal surgery such as discectomy or fusion), due to the distinct care pathways and rehabilitation strategies compared with non‐surgical management of chronic RMDs.

Types of interventions

Eligible DHIs will be those aimed at improving quality of life or physical function in adults aged 50 years or older with osteoarthritis, osteoporosis, low back pain, rheumatoid arthritis, or polymyalgia rheumatica, as specified in the eligibility criteria.

DHIs will be eligible when the core intervention content is delivered primarily through a digital platform, such as a mobile application, web‐based programme, telemonitoring system, wearable sensor interface, or other digital health tool. Interventions may vary in format, intensity, mode of interaction, and degree of human involvement, but the defining criterion for eligibility is that the main intervention content and interaction occur digitally. Information on intervention frequency, intensity/dose, and duration will be extracted and described, when reported, as part of the intervention characteristics.

Included interventions may be fully self‐delivered or may incorporate occasional support from a healthcare professional or another appropriately trained technician (e.g. nurses, psychologists, or research assistants). This support may occur at enrolment, for example, when introducing the digital health tool, providing initial training, or addressing questions, or it may take place during follow‐up. However, any face‐to‐face contact must be complementary and clearly secondary to the digital component. For example, sporadic in‐person visits to monitor progress, provide technical assistance, or replace materials. Interventions in which face‐to‐face contact represents more than a complementary and clearly secondary component will not be considered to be DHIs and will be excluded.

To facilitate comparison across studies, we will group eligible DHIs into the following five predefined intervention functions, reflecting the primary mechanism by which each intervention aims to influence outcomes.

1. Medication optimization (e.g. adherence reminders, medication tracking)

2. Remote patient monitoring (e.g. sensor‐based activity tracking, symptom logging, teleconsultations/telehealth)

3. Disease management (e.g. digital care plans, self‐management apps)

4. Health promotion and behavioural change (e.g. digital coaching for exercise, diet, fall prevention)

5. Social support (e.g. virtual support groups, peer forums)

Interventions that incorporate elements from more than one intervention function will be coded as multi‐component interventions (i.e. interventions that combine two or more distinct digital functions, such as education, monitoring, and behavioural support, within the same programme). Because it is often not possible to identify a dominant component within such interventions, we will treat multicomponent interventions as indivisible units and include them as such in the synthesis. We will explore differences between single‐component and multicomponent interventions through predefined subgroup analyses.

We will exclude telephone‐only interventions, defined as interventions delivered exclusively through voice calls without digital content, and isolated SMS messages used solely for reminders or administrative purposes.

We will only include teleconsultations when they form part of a structured DHI delivered through a digital platform (e.g. video consultations integrated into telemedicine systems, remote monitoring programmes, or app‐based interventions). Teleconsultations provided solely as part of usual care will not be considered an intervention. We will include mobile health (mHealth) applications, provided they deliver the core content of the intervention digitally.

Comparators will include the following.

• Usual care, defined as standard, non‐digital clinical and supportive care, which may include both non‐pharmacological and pharmacological components (e.g. routine clinical follow‐up, physiotherapy, patient education, and pharmacological management according to standard clinical practice). Participants in both the intervention and comparator groups may therefore receive anti‐rheumatic medication. Pharmacological treatment may coexist with the interventions under study; however, it is not the focus of this review and is not considered a primary comparator.

• Non‐digital interventions, defined as structured health promotion interventions delivered without the use of any digital technology. These interventions go beyond usual care and involve planned content aimed at improving health or health‐related behaviours. Examples include face‐to‐face educational programmes (e.g. scheduled group sessions or workshops), printed materials (e.g. manuals, exercise guides, or informational pamphlets), supervised exercise or physiotherapy classes not part of routine care, and in‐person behavioural or psychological interventions, in line with the WHO definition of health promotion [31].

• No intervention. This category may include attention control or sham interventions, provided that these do not include active content intended to influence the review outcomes.

Outcome measures

Critical outcomes

This review will focus on two critical outcomes: quality of life and physical function. For this review, we will adopt the WHO definition of quality of life as “an individual's perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards and concerns” [32]. We will define physical function according to the WHO's International Classification of Functioning, Disability and Health (ICF) as “the individual's physical capacities, such as mobility, strength, endurance, and coordination, that enable the performance of basic actions (e.g. walking, bathing) and participation in daily life activities” [9]. These outcomes are highly relevant to patients, clinicians, and policymakers, particularly in the context of RMDs in older adults.

Quality of life will be assessed using validated instruments, including the following.

• EuroQol five‐dimensional questionnaire (EQ‐5D) [33, 34]

• Short Form‐36 Health Survey (SF‐36) [35], and Short Form‐36 Health Survey (SF‐12) [36]

• World Health Organization Quality of Life –100 items (WHOQOL–100) and brief version (WHOQOL–BREF) [32, 37]

• Control, Autonomy, Self‐Realization and Pleasure (CASP) 19‐item and 12‐item scales: CASP‐19, CASP‐12 [38, 39]

• McGill Quality of Life Questionnaire (MQoL) [40]

• Rheumatoid Arthritis Quality of Life Questionnaire (RAQoL) [41]

• Other validated tools where applicable

Physical function will be assessed through the following.

• Health Assessment Questionnaire and Health Assessment Questionnaire Disability Index (HAQ and HAQ‐DI) [42]

• Barthel Index [43]

• Morton Mobility Index [44]

• Katz Index [45]

• Knee injury and Osteoarthritis Outcome Score (KOOS) [46]

• Hip disability and Osteoarthritis Outcome Score (HOOS) [47]

• The Quebec Back Pain Disability Scale [48]

• TUG test (Timed up‐and‐go test) [49]

• Single chair stand [50]

• Gait speed (400 m) [51]

• Short physical performance battery [52]

• Other validated disability or function scales

If studies report more than one outcome measure per outcome, we will prioritize instruments commonly used in RMD research and recommended in core outcome sets [53]. We will not consider composite outcomes unless their components are clearly defined and relevant to the review’s objectives.

If outcomes are reported at multiple time points, we will extract all time points but will use the latest available for the primary analysis, as meaningful changes in quality of life and physical function generally require longer intervention periods [26, 54].

Important outcomes

We will consider the following important outcomes.

• Acceptability of the DHI, assessed through validated measures or quantitative indicators such as attrition rates, frequency of use, or satisfaction with the intervention.

• Adverse events (e.g. pain, misdiagnosis) [28]

Search methods for identification of studies

Electronic searches

We will perform a comprehensive search of the following electronic bibliographic databases and trial registers.

• Databases

  • Cochrane Central Register of Controlled Trials (CENTRAL) (via Cochrane Library; from inception to date of search)

  • MEDLINE (via PubMed; from inception to date of search)

  • Embase (via Elsevier; from inception to date of search)

  • Scopus (via Elsevier; from inception to date of search)

  • Web of Science (via Clarivate; from inception to date of search)

  • Institute of Electrical and Electronics Engineers (IEEE) Xplore (IEEE; from inception to date of search)

  • ACM Digital Library (ACM; from inception to date of search)

• Trial registers

  • US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; from inception to date of search)

  • ISRCTN Registry (ISRCTN) (www.isrctn.com; from inception to date of search)

  • EU Clinical Trials Register (EUCTR) (www.clinicaltrialsregister.eu; from inception to date of search)

  • Clinical Trials in the European Union – EMA (euclinicaltrials.eu; from inception to date of search)

  • World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch; from inception to date of search)

• Preprint and grey literature sources

  • OSF (osf.io; from inception to date of search)

  • MedRxiv (www.medrxiv.org; from inception to date of search)

  • Preprints.org (www.preprints.org; from inception to date of search)

  • Europe PMC (europepmc.org; from inception to date of search)

An initial search strategy was developed for MEDLINE (via PubMed) at the proposal stage, using a combination of MeSH terms and free‐text keywords, and subsequently adapted to the syntax and indexing terms of each database, trial register, and grey literature source in collaboration with information specialists (SS, TC, and CA). No date or language restrictions were applied.

The study design component of the MEDLINE search strategy includes RCT‐related terms based on the Cochrane highly sensitive search strategy for identifying randomized controlled trials and controlled clinical trials, as described in Chapter 4 of the Cochrane Handbook for Systematic Reviews of Interventions [55]. As the eligibility criteria were not restricted to RCTs, terms for quasi‐experimental and other non‐randomized study designs were also included. The complete line‐by‐line search strategy for MEDLINE, including all applied filters and limits, is provided in Supplementary material 1.

For trial registers, the strategies differ slightly from those used in other sources, focusing on the condition and intervention, as recommended, as suggested in the work of Hunter and colleagues [56].

The searches will be rerun prior to publication to ensure currency. If more than 12 months pass between the initial search and publication, we will update the searches and include any newly identified studies.

Searching other resources

To ensure a comprehensive identification of eligible studies, we will apply additional search strategies.

• Reference list screening: we will examine the reference lists of all included studies and of relevant systematic reviews on similar topics to identify additional eligible studies not captured through database and register searches.

• Citation tracking: we will conduct backward and forward citation tracking of all included studies using Web of Science and Scopus to identify relevant studies that cited, or were cited by, the included records.

• Contact with experts: if necessary, we will contact the corresponding authors of included studies and other experts in the field of DHIs and RMDs to enquire about ongoing or unpublished studies.

• Identification of postpublication amendments: close to the publication of the review, we will check MEDLINE and Embase, as well as the Retraction Watch Database (retractionwatch.com), to identify any postpublication amendments (including retractions, corrections, errata, or expressions of concern) related to included or eligible studies.

Data collection and analysis

Selection of studies

All titles and abstracts retrieved from the searches will be imported into Covidence [57], where duplicates will be identified and removed.

Two reviewers will independently screen the titles and abstracts of all records in Covidence to assess their eligibility. Potentially relevant articles will then undergo full‐text screening, also conducted independently by the same reviewers. Reviewers will conduct both stages blinded to each other’s decisions. Disagreements at any stage will be resolved through discussion. If no consensus is reached, a third reviewer will be consulted.

In parallel, we will utilize ASReview to screen titles and abstracts [58]. We will compare its performance with the traditional screening methodology.

Reasons for exclusion of studies at the full‐text stage will be documented in a table of excluded studies, following a predefined hierarchy of exclusion reasons where appropriate. We will summarize the study selection process in a PRISMA 2020 flow diagram [59, 60].

If information essential to determining study eligibility or extracting data is missing or unclear, we will attempt to contact the study authors via email to request clarification or additional data.

Any reviewers with direct involvement in the conduct, analysis, or publication of an included study will not participate in eligibility decisions, data extraction, risk of bias assessment, or GRADE evaluation of that study, in line with Cochrane’s conflict of interest policy.

Data extraction and management

Two reviewers will independently extract data from the included studies using a predesigned data extraction form. Before full data extraction begins, they will pilot the form on a small sample of studies to ensure clarity and consistency across reviewers. Discrepancies in data extraction will be resolved by discussion or by consulting a third reviewer if needed.

We will extract the following information from each included study.

• Bibliographic details: authors, publication year, study date.

• Methods: study registration, study design, total duration, setting, country, informed consent, ethics approval, method of randomization (if applicable).

• Participants: total number, number randomized (if applicable), number lost to follow‐up, number analyzed for each outcome, age, sex, inclusion and exclusion criteria.

• Interventions: description of intervention and comparator, type of intervention. Intervention data will be extracted following the template for intervention description and replication (TIDieR) checklist [61].

• Outcomes: quality of life and physical function, including the instruments used and the time points reported.

Risk of bias assessment in included studies

Two review authors will independently assess the risk of bias in the included studies. Disagreements will be resolved through discussion or, if necessary, by consulting a third reviewer. Reviewers will be blinded to each other’s assessments during the initial evaluation.

We will use the appropriate risk of bias tool for each eligible study design, following the Cochrane Handbook for Systematic Reviews of Interventions [62].

• For RCTs, we will use the Cochrane Risk of Bias 2 (RoB 2) tool, a standardized tool used to assess the risk of bias in randomized trials, as implemented in the RoB 2 Excel tool. The effect of interest will be the effect of assignment to intervention [63].

• For cluster‐randomized trials, we will apply the cluster‐RoB 2 Excel tool. The effect of interest will be the effect of assignment to intervention [64].

• For quasi‐randomized trials, we will use the Risk Of Bias In Non‐randomized Studies of Interventions (ROBINS‐I) tool, which is designed to assess the risk of bias in studies where allocation is not fully random [65].

• For controlled before‐and‐after studies, we will use the ROBINS‐I tool [65].

• For feasibility or pilot studies, we will apply the risk of bias tool corresponding to the study design (RoB 2 if randomized; ROBINS‐I if non‐randomized) [63, 65].

We will assess risk of bias according to the following domains.

• For RCTs (including cluster‐RCTs), the five RoB 2 domains

  • Bias arising from the randomization process

  • Bias due to deviations from intended interventions

  • Bias due to missing outcome data

  • Bias in measurement of the outcome

  • Bias in selection of the reported results

• For NRSIs (quasi‐RCTs, controlled before‐and‐after, feasibility if non‐randomized), the seven ROBINS‐I domains

  • Bias due to confounding

  • Bias in selection of participants into the study

  • Bias in classification of interventions

  • Bias due to deviations from intended interventions

  • Bias due to missing data

  • Bias in measurement of outcomes

  • Bias in selection of the reported results

Risk of bias will be assessed at the outcome level for all outcomes, including both critical outcomes (quality of life and physical function) and important outcomes (adverse events and acceptability), at the longest time point available within the predefined categories.

In accordance with Cochrane’s conflict of interest policy, any reviewers with direct involvement in the conduct, analysis, or publication of an included study will abstain from assessing its risk of bias.

Overall risk of bias at the study level

We will consider an outcome to be at high risk of bias if at least one domain is rated as high risk. Conversely, we will consider an outcome to be at low risk of bias only if all domains are rated as low risk.

Analyses

In accordance with our predefined strategy, only RCTs will contribute to the quantitative synthesis. Our primary analysis will include all eligible RCTs, no matter their risk of bias rating. However, if there is substantial heterogeneity that could be explained by a high versus low risk of bias subgroup analysis, our conclusions will be based on the low risk of bias studies to avoid downgrading our certainty in the evidence. This is in accordance with GRADE recommendations [66, 67].

Assessing bias in the conduct of the review

We will conduct the review according to this published protocol and will document any deviations from it in the ‘Differences between protocol and review’ section of the full review.

We will also keep an updated Open Science Framework Project register where any changes will be transparently reported and made publicly available throughout the conduct of the review.

Measures of treatment effect

Continuous outcomes

For continuous outcomes measured on the same scale across studies, we will report the mean difference with corresponding 95% confidence intervals (CIs). If studies use different validated instruments to assess the same outcome domain (e.g. different quality of life questionnaires), we will calculate and present the standardized mean difference (SMD) with 95% CIs to allow for synthesis.

If studies report continuous data as a median and interquartile range (IQR), and data pass the test of skewness, we will convert the median to mean, and estimate the standard deviation (SD) as IQR/1.35 [68].

Dichotomous data

For dichotomous outcomes (e.g. presence versus absence of physical limitations), we will extract or calculate risk ratios (RRs), or odds ratios (ORs), each with 95% CIs.

Unit of analysis issues

In individually randomized RCTs, the unit of analysis will be the individual participant, who will be considered only once per analysis.

For cluster‐randomized trials, the unit of randomization will be the cluster (e.g. hospital, rehabilitation unit, primary care practice, or community group). The unit of analysis will remain the individual participant, with appropriate statistical adjustment for clustering. We will extract information on the study design and unit of randomization for each study, indicating whether clustering of observations is present due to group‐level allocation. We will extract any available statistical information needed to account for clustering, such as design effects or intra‐cluster correlation coefficients (ICCs), and whether the study’s results were adjusted accordingly.

If a study does not account for clustering in its analysis, we will adjust the effective sample size using Cochrane‐recommended methods [69]. Where possible, we will obtain the ICC from the study itself or from a similar published study. If no appropriate ICC is available, we will conduct sensitivity analyses imputing a plausible range of ICC values to assess the impact of clustering.

For studies with more than two intervention arms contributing to the same comparison, we will avoid double‐counting by combining relevant intervention arms into a single group where appropriate, or by selecting the most relevant comparison based on the predefined criteria in the Types of interventions. Any potential bias introduced by this selection will be acknowledged in the discussion of the results.

For interventions with multiple components (e.g. app combined with coaching and reminders), we will treat the combined intervention as defined in the original study, and analyze its overall effect compared to the control group.

For before‐and‐after and feasibility studies, the unit of analysis will be the individual participant, with paired measurements over time where applicable.

For standard parallel‐group RCTs with two arms, no unit of analysis issues are expected.

Dealing with missing data

We will contact study authors to obtain any missing or unclear information related to study characteristics, risk of bias assessment, or outcome data. If we do not receive a response, we will analyze the available data and report the extent of missing data in the included studies.

For dichotomous outcomes, we will conduct analyses based on the number of participants with data available at the relevant time point. We will not assume that participants with missing outcomes experienced the event, unless explicitly stated in the study. In sensitivity analyses, we will explore the potential impact of missing data using approaches such as best‐case and worst‐case scenarios.

For continuous outcomes, we will use the mean and SD as reported. If SDs are not reported but standard errors, CIs, t values, or P values are available, we will calculate the SDs using standard formulas recommended in the Cochrane Handbook [68]. Where appropriate, we will explore the impact of missing continuous data in sensitivity analyses by excluding studies with high or unclear levels of missing data or by comparing studies that used different imputation approaches.

We will clearly report the amount and pattern of missing data for each study and discuss how it may affect the certainty of the evidence [67, 70].

Reporting bias assessment

Two review authors will independently assess the risk of bias due to missing results for each synthesis. Disagreements will be resolved through discussion or, if necessary, by consulting a third review author.

We will compare the stated primary and secondary outcomes in the study reports against those prespecified in study protocols or trial registrations, when available, to identify potential selective outcome reporting. Any studies that appear to meet the inclusion criteria but do not report any of the primary or secondary outcomes will be documented in the Characteristics of included studies table.

For each meta‐analysis that includes at least 10 studies reporting the same outcome, we will generate funnel plots and visually assess asymmetry, which may indicate publication bias or small‐study effects. If asymmetry is detected, we will explore potential causes and incorporate this consideration into the assessment of the certainty of the evidence [71]. If our review includes fewer than 10 studies eligible for meta‐analysis, the ability to detect publication bias will be largely diminished, and we will simply note our inability to rule out possible publication bias or small‐study effects [70].

We will contact study authors to obtain or confirm relevant information when reporting bias is suspected or when unpublished outcomes are identified in protocols or trial registrations.

We do not plan to use any automation tools to assess risk of bias due to missing results.

Synthesis methods

As described in the Types of interventions section, three comparison groups are planned.

1. Digital health intervention versus usual care

2. Digital health intervention versus non‐digital structured intervention

3. Digital health intervention versus no intervention

All eligible DHIs will be synthesized within each comparison group (digital versus usual care; digital versus non‐digital structured intervention; digital versus no intervention). We will not compare different types of DHIs directly against each other. Differences in intervention characteristics (e.g. single‐ versus multi‐component structure, delivery modality) will be examined only through predefined subgroup analyses within each comparison group.

Following Cochrane guidance (Section 9.3.2 of the Cochrane Handbook) [72], before undertaking any synthesis, we will first tabulate detailed study characteristics, including: population; disease type (inflammatory versus non‐inflammatory); intervention functions (medication optimisation, remote patient monitoring, disease management, health promotion and behavioural change, social support); delivery modality (e.g. mobile application, web‐based platform, wearable device); comparator type (usual care, non‐digital structured intervention, or no intervention); outcome domain; outcome measure; and length of follow‐up. Considering that usability and accessibility are particularly important for older adults, we will also extract descriptive information on features such as interface simplicity, availability of technical support, clarity of language, adaptability to users’ abilities, and any reported usability assessments.

In addition, and informed by the PROGRESS‐Plus framework (Place of residence, Race/ethnicity/culture/language, Occupation, Gender/sex, Religion, Education, Socioeconomic status, Social capital, plus additional factors such as age, disability, or other characteristics associated with disadvantage) [73], we will extract equity‐relevant contextual factors consistently reported across studies – specifically country, participant gender, and setting (urban versus rural), to support exploration of equity‐related differences.

If included studies are sufficiently similar in terms of participants, interventions, comparators, and outcomes, we will perform a meta‐analysis for RCTs only. We will synthesize NRSIs narratively.

Given the expected diversity of DHIs, we anticipate that meta‐analysis may only be possible within subgroups of RCTs. We will perform meta‐analysis using Review Manager (RevMan) [74]. Where additional or more advanced analyses are required, we will use RStudio [75], employing the meta [76] and metafor [77] packages as appropriate.

Our critical outcomes (quality of life and physical function) may be reported either as continuous data (e.g. scores on validated scales) or as dichotomous data (e.g. proportion of participants classified as having physical limitations versus no physical limitations, based on a predefined cut‐off).

For dichotomous outcomes, we will calculate RR or OR with 95% CIs. For continuous outcomes, we will calculate mean differences or standardized mean differences with 95% CIs. We will use a random‐effects model to account for expected heterogeneity [70]. Between‐study variance will be estimated using the Restricted Maximum Likelihood (REML) method. When there are at least three studies and heterogeneity is greater than zero, we will apply the Hartung–Knapp–Sidik–Jonkman (HKSJ) method to calculate confidence intervals for pooled effect estimates. When only two studies are available, or when heterogeneity is estimated as zero, we will use Wald‐type confidence intervals, in line with current Cochrane guidance [70].

We will assess heterogeneity visually (forest plots) and statistically (Chi² test, I² statistic, and Tau²). If substantial heterogeneity is identified, we will explore potential sources through subgroup analyses and narrative explanation.

Narrative synthesis of non‐randomized studies, and of RCTs when a meta‐analysis is judged inappropriate, will follow the guidance in Chapter 12 of the Cochrane Handbook [78] and the SWiM (Synthesis Without Meta‐analysis) [79] reporting guidelines to provide a structured narrative synthesis. We will create tables of individual study results ordered by risk of bias and include standardized effect estimates, modelled on Table 12.4.b of the Cochrane Handbook [78]. We will also use forest plots to visually present study data where applicable.

When multiple outcome measures or time points are reported for the same outcome, we will prioritize the most clinically relevant and comparable measure, or the latest time point where appropriate, to avoid double‐counting.

We will also describe any differences related to equity considerations (place of residence, country income level, and gender) narratively, when such information is available.

Investigation of heterogeneity and subgroup analysis

Before investigating heterogeneity, we will use the study characteristics table described in the Synthesis methods to assess the diversity of the studies.

We will assess statistical heterogeneity exclusively within meta‐analyses of RCTs. As we will synthesize NRSIs narratively, they will not undergo quantitative assessment of heterogeneity.

To assess statistical heterogeneity, we will visually inspect forest plots, describing the direction and magnitude of effects and the overlap of confidence intervals. We will quantify inconsistency using the I² statistic and report the Chi² test P value (considering P < 0.1 to be indicative of heterogeneity) [80].

We will interpret I² values according to the Cochrane Handbook (Chapter 10) [70], acknowledging these thresholds are approximate and should be considered in the context of effect sizes and strength of evidence.

• 0 to 40%: might not be important

• 30 to 60%: may represent moderate heterogeneity

• 50 to 90%: may represent substantial heterogeneity

• 75 to 100%: may represent considerable heterogeneity

We will treat I² and Tau² as approximate indicators, recognizing the inherent uncertainty of these estimates, particularly when few studies are available.

In line with guidance in the Cochrane Handbook, Section 10.11.2, we will interpret subgroup comparisons with caution, given the potential for confounding and the observational nature of such analyses [70]. In particular, we will not emphasize subgroup analyses with fewer than five studies per category. Where feasible, we will conduct stratified meta‐analyses and apply formal tests of interaction (e.g. Cochran’s Q test) or, where sufficient studies are available, meta‐regression to explore heterogeneity.

We plan to explore the following factors as potential effect modifiers.

• Type of intervention: single‐ versus multi‐component

• Disease type: inflammatory versus non‐inflammatory

• Timing of outcome assessment: according to follow‐up period

• Age: 50 to 60 years, 60 to 75 years and 75 years or above

Factors such as specific intervention functions (medication optimization; remote patient monitoring; disease management; health promotion and behavioural change; social support), country income level (high versus middle/low), setting (urban versus rural), and gender will be explored narratively.

For RCTs, if subgroup analyses are not possible, we will describe heterogeneity qualitatively. For NRSIs, which will not undergo statistical subgroup analysis, we will examine heterogeneity narratively by comparing variations in population, intervention characteristics, comparators, and outcomes across studies.

Equity‐related assessment

According to the PROGRESS‐Plus framework, documenting factors such as place of residence, race/ethnicity/culture/language, occupation, gender/sex, religion, education, socioeconomic status, and social capital can help identify and interpret potential equity issues [73, 81].

In this review, we will investigate potential equity‐related differences in effects by country income level (high‐, middle‐, and low‐income countries), gender (male, female) and setting (urban versus rural), as the effectiveness and feasibility of DHIs may vary according to these contextual factors. We will extract and report the country, participant gender and setting of each study in the Characteristics of included studies table and, if data permit, explore these factors narratively.

Sensitivity analysis

We will conduct sensitivity analyses to assess the robustness of the results to key methodological assumptions and study characteristics. Specifically, we will:

• explore the effect of excluding studies judged at high risk of bias in at least two domains, to examine whether such studies overestimate intervention effects; and

• assess the impact of missing data by applying best‐case and worst‐case scenarios for dichotomous outcomes, and by excluding studies with high or unclear levels of missing continuous data or comparing different imputation approaches for continuous outcomes (see Dealing with missing data).

We will interpret the findings of sensitivity analyses qualitatively, considering whether the direction and magnitude of effect estimates remain consistent across different assumptions.

Certainty of the evidence assessment

Using GRADEpro GDT software [82], we will prepare summary of findings tables for each of the three planned comparisons.

1. Digital health intervention versus usual care

2. Digital health intervention versus non‐digital structured intervention

3. Digital health intervention versus no intervention

Each summary of findings table will include the two critical outcomes (quality of life and physical function) and two important outcomes (adherence and adverse events). For each outcome included in the summary of findings table, we will present the effect estimate derived from the longest comparable follow‐up time point and based on the validated instruments used in the included RCTs.

Evidence from non‐randomized studies will not contribute to the summary of findings tables. We will synthesize such evidence narratively to support interpretation of RCT findings.

Two review authors will independently assess the certainty of the evidence for each outcome, with disagreements resolved by discussion or by consulting a third reviewer. We will initially consider evidence from RCTs to be of high certainty and may downgrade it by one level for serious (or two levels for very serious) limitations in the following domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias.

We will categorize the certainty of the evidence as follows.

• High: very confident that the true effect lies close to the estimate.

• Moderate: moderately confident; the true effect is likely close but may be substantially different.

• Low: limited confidence; the true effect may be substantially different.

• Very low: very little confidence; the true effect is likely substantially different from the estimate.

Consumer involvement

We have discussed the relevance of this review with healthcare professionals (rheumatologists and physiotherapists), who emphasized the importance of digital health tools for the care of people with rheumatic diseases and the need to tailor such tools to patients’ needs. Our author team also includes clinical experts.

A public contributor living with an RMD has revised our review proposal – the President of the Portuguese League Against Rheumatic Diseases. She endorsed the relevance of the topic, provided input, and is a co‑author of this protocol (EM).

In addition, we will involve the League in gathering input from a broader group of people living with RMDs by establishing a Patient and Public Involvement and Engagement (PPIE) advisory group. Their contribution will help ensure that the review remains aligned with patients’ perspectives and real‑world priorities.

We will consider using the GRIPP2 (Guidance for Reporting Involvement of Patients and the Public) checklist to transparently report PPIE activities [83].

Supporting Information

Supplementary materials are available with the online version of this article: 10.1002/14651858.CD016341.

Supplementary materials are published alongside the article and contain additional data and information that support or enhance the article. Supplementary materials may not be subject to the same editorial scrutiny as the content of the article and Cochrane has not copyedited, typeset or proofread these materials. The material in these sections has been supplied by the author(s) for publication under a Licence for Publication and the author(s) are solely responsible for the material. Cochrane accordingly gives no representations or warranties of any kind in relation to, and accepts no liability for any reliance on or use of, such material.

Supplementary material 1 Search strategies

New

Additional information

Contributions of authors

• ARH identified the topic of the review, conducted the literature review to inform the protocol, drafted the search strategies, and wrote the first draft of the protocol.

• AR, NM, and DC identified the topic of the review, contributed to writing the draft, and reviewed and approved the final version of the protocol.

• EM validated the topic of the review and reviewed and approved the final version of the protocol.

• MC contributed to writing the draft and reviewed and approved the final version of the protocol.

• SS, TC, and CA reviewed and refined the search strategy, and reviewed and approved the final version of the protocol.

Declarations of interest

ARH, CA, SS, TC, NM, and DDC have no interests to declare.

MC has contributed as an advisor on topics related to physical rehabilitation in the past 36 months for the WHO and the Portuguese Professional Board of Physiotherapists.

EFM – Member of the jury (consultancy) for the BI Award for Innovation in Healthcare (Boehringer Ingelheim) in 2023; participant in panel discussion (speaker fees) ‘Shaping the Future of Digital Health’ (Daiichi Sankyo) in 2024; member of the OA Patient Voice Panel (advisory board) in 2024 (Grünenthal); member of CEIC – National Ethics Committee for Clinical Research.

AMR received honoraria from Amgen, ABBvie and Novartis; received grants from Amgen, AstraZeneca, Novartis, Phizer, MSD, Lilly, and Boehringer Ingeheim.

Sources of support

Internal sources

• No sources of support provided

External sources

• FCT, Portugal

  The first author received a PHD grant from Fundação Ciência e Tecnologia, IP under the PhD grant PRT/BD/154695/2023 within the scope of the EUTOPIA (https://doi.org/10.54499/PRT/BD/154695/2023).

Registration and protocol

Cochrane approved the proposal for this review in May 2025.

Data, code and other materials

Data sharing is not applicable to this article as it is a protocol, so no datasets were generated or analysed.