This is a prospective cross-over study including participants with hip or knee osteoarthritis. Recruitment is planned from September 2025 to March 2026 in the Netherlands. Each participant will be followed for 12 months. Assessments occur every 2 weeks (Figure 1). Participants will complete 3 self-care programs (physical activity promotion, sleep optimization, and weight management) and 1 control period in random order. Programs and control will start in weeks 3, 15, 27, and 39, each lasting 12 weeks. No washout period is included. This study has a cross-over design that allows the evaluation of behavioral interventions in a real-world context. Rather than assuming a return to baseline, the design captures the sustained and cumulative nature of behavior change [17]. This aligns with the objective of identifying an optimal sequencing of self-care programs over time. Recent methodological work by Kulnik et al [17] and Shi et al [18] supports that such designs remain valid in the presence of behavioral carryover effects when appropriately addressed analytically. This study has been reviewed and approved by the NedMec Medical Ethical Committee (NL87119.041.24) and is registered at the International Clinical Trials Registry (registered on May 21, 2025, NL-OMON57573) and ClinicalTrials.gov (registered February 20, 2026, NCT07423858). This protocol was developed in accordance with the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) checklist [19] and the TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis) + AI checklist.

We aim to include 600 participants through a broad recruitment strategy across multiple channels to reach Dutch individuals with osteoarthritis. Eligibility is based on the criteria listed in Textbox 1. Participants must be adults with self-administered osteoarthritis according to the National Institute for Health and Care Excellence criteria [10]; key exclusion criteria include systemic arthritis or recent or upcoming lower limb surgery.

Recruitment is coordinated by the University Medical Center Utrecht and supported by different health care professionals (ie, physiotherapists, general practitioners, and an orthopedic surgeon) partnered with the research consortium. Outreach methods include online and offline leaflets, local newspaper advertisements, engagement with local health care professionals, letters from general practitioners, social media, and promotion via the Dutch Arthritis Society. Individuals who are interested in participating will be able to register via a Castor web-based study portal, phone, or email.

For this study, the e-cOAch version 1 web app was developed in a co-design process with a patient panel and health care professionals to ensure technical functionality, usability, and relevance. The e-cOAch version 1 is a Class IIa medical device under the European Union Medical Device Regulation (2017/745) and is used in this study under the Medical Device Regulation article 82 [20]. The e-cOAch version 1 was primarily designed for data collection, with additional self-care and osteoarthritis knowledge content based on national and international hip and knee osteoarthritis guidelines [5,6,21] (Table 1).

The e-cOAch offers education on osteoarthritis, self-care advice (ie, physical activity promotion, weight management, and sleep optimization), and guidance on seeking professional care. Content is delivered through text, videos, and interactive assignments (Table 1). Participants who start supervised treatment with a health care professional may continue with the app or pause participation. To minimize dropout and maintain engagement, the study will combine monitoring, timely tailored reminders, and optional personal phone follow-ups. Harms and adverse events will be followed up by phone after they are reported by the participant.

All outcome measures are self-reported via questionnaires in the e-cOAch. To improve inclusivity, all questionnaires are adapted to language level B1. This is in line with the Pharos guideline for understandable questionnaires [33]. Pharos is a Dutch national expertise center on health equity.

The primary outcome is self-reported physical functioning, measured using the Hip Disability and Osteoarthritis Outcome Score (HOOS) [34] and the Knee Injury and Osteoarthritis Outcome Score (KOOS) [35]. Secondary outcomes are pain (Numeric Pain Rating Scale) and participation (Patient-Reported Outcomes Measurement Information System Experience, Short-Form version 8a) [36]. All outcome measures are presented in Table 2. All selected instruments have demonstrated acceptable-to-good reliability and validity. Outcome measures are based on 2 systematic reviews and include known factors that influence physical functioning, weight management, and sleep behavior in osteoarthritis [37,38]. Baseline characteristics (Table 3) will be collected at the start of using the e-cOAch, including digital health literacy. Ethnicity (Asian or non-Asian) is recorded because BMI calculations differ in Asian populations [39].

Baseline and follow-up assessments are conducted remotely and distributed over 52 weeks to minimize participant burden. Pain, physical activity, and stiffness are assessed biweekly, while slower-changing outcomes are measured every 10 weeks (Table 3).

In total, 18 sequences of the self-care programs are possible due to the restriction that participants cannot end the study with the control period. Participants are assigned to one of the 18 sequences (Table 1, Figure 1). Each sequence consists of three 12-week self-care programs: physical activity promotion, weight management, and sleep optimization, as well as a 12-week control period. Programs start at weeks 3, 15, 27, and 39. Each program is delivered once per participant consecutively, without intermissions.

Randomization is performed at baseline using a computer-generated allocation sequence embedded within the e-cOAch web app, assigning participants with equal probability to one of the 18 intervention sequences. Allocation was computer generated and concealed until disclosure by an independent staff member before baseline assessment.

For safety reasons, extremely underweight participants (BMI<17.5) are not assigned to the weight management program; these participants instead follow a 12-week no-intervention period and are advised to consult a dietitian. Additionally, underweight participants (BMI<18.5), overweight participants (BMI>30), or those requiring medical clearance for physical activity (Physical Activity Readiness Questionnaire >0), are advised to consult a health care professional but remain eligible for all programs.

The study is conducted in an open-label design, as blinding participants and researchers is not feasible given the nature of the self-care programs.

We aim to recruit 600 participants within 6 months. Data will be split into training and test sets for the development and validation of predictive algorithms. A random subset of the data will be reserved for testing to evaluate model performance on unseen data [50].

Combining multiple predictors and allowing for nonlinear relationships is expected to improve the prediction of pain, physical functioning, and participation [50]. A large sample is therefore required to capture diverse combinations of participant characteristics and support model validation. Following standard practice, 20% to 30% of the data will be used for testing and 70% to 80% for training [50], resulting in an estimated 100 to 150 participants in the test set.

To account for an anticipated 15% dropout [14], we aim to recruit 600 participants, yielding an expected final sample size of approximately 510 participants. Given the high prevalence of osteoarthritis and the broad reach of project partners, this recruitment target is considered feasible.

This study has been reviewed and approved by the NedMec Medical Ethical Committee under the following reference: NL87119.041.24. The study will be conducted according to the principles of the Declaration of Helsinki (version 13) and following the Medical Research Involving Human Subjects Act. Furthermore, the study will adhere to the General Data Protection Regulation to ensure the protection of personal data and the Medical Device Regulation where applicable. The University Medical Center Utrecht is the sponsor of this study and can be contacted through the corresponding author. Any protocol deviations will be communicated to relevant parties. Informed consent will be obtained digitally via Castor eConsent. After registration, individuals will receive detailed study information. After at least 1 week, they will be contacted by telephone to confirm eligibility and understanding before providing consent. Eligible participants will be asked to sign the consent form digitally. Paper consent will be available upon request. Participants can opt out of the study at any moment by contacting the research staff through phone or email.

Reflexivity enhances transparency and rigor in research by prompting researchers to acknowledge how their assumptions and positionalities influence design, analysis, and interpretation [57]. This study is conducted within a multidisciplinary consortium including specialists in behavior change, psychology, vulnerable groups, computer science, and physiotherapy, whose diverse professional lenses shape our design choices and interpretations. Guided by a pragmatic epistemological stance, we view knowledge as generated through the combined insights of researchers, participants, and their real-world context [58]. To counterbalance disciplinary assumptions, individuals with lived experience of osteoarthritis are acting as consultative contributors: they reviewed the protocol during its development and may collaborate with the research team during the interpretation of the findings. Their involvement and ongoing reflexive team discussions can improve the research quality by challenging researcher assumptions and strengthening interpretation [59].