Protocol for DESCIDE-RD: Determining Effectiveness of Shared Decision Making in Drug Therapy Selection for Rheumatic Diseases

Abstract

The availability of biologic agents and Janus kinase (JAK) inhibitors has expanded therapeutic options for rheumatic diseases, and international guidelines now recommend shared decision-making (SDM) to guide treatment selection. This study will evaluate whether a structured SDM model enhances patients’ perceptions of SDM when selecting biologic agents or JAK inhibitors for rheumatoid arthritis (RA) and ankylosing spondylitis (AS), compared to standard care. This open-label, multicentre, stratified cluster-randomised trial will enrol 300 patients with RA or AS. Participants will be assigned in a 1:1:1 ratio to one of three groups: SDM Group I, SDM Group II, or standard care. In the SDM groups, the Three-Talk Model (Team Talk, Option Talk, and Decision Talk) will be implemented. SDM Group I will receive decision aids during Option Talk, while SDM Group II will receive both decision aids and an artificial intelligence-based chatbot. In the standard care group, treatment recommendations will be made at the physician’s discretion. The primary outcome will be the Shared Decision-Making Questionnaire score. Secondary outcomes will include changes in the Decisional Conflict Scale, the Health Empowerment Scale, the Leeds Satisfaction Questionnaire, and the EuroQol-5 Dimension. Data will be collected using iCReaT (http://icreat.kr). This study will provide evidence on the effectiveness of structured SDM in selecting biologic agents and JAK inhibitors for RA and AS, with potential implications for improving treatment decisions and patient outcomes.

Trial Registration

Clinical Research Information Service Identifier: KCT0011310

Graphical Abstract

INTRODUCTION

Rheumatoid arthritis (RA) and ankylosing spondylitis (AS) are the two major inflammatory arthropathies, primarily affecting the peripheral joints and axial skeleton, respectively.1,2 Patients with RA or AS typically experience joint pain and stiffness, which, without effective treatment, can lead to substantial joint damage and disability. Advances in understanding the pathogenesis of these diseases have led to the rapid development of targeted immunomodulatory therapies, resulting in marked improvements in remission rates and long-term outcomes.3,4 Multiple therapeutic options with distinct mechanisms of action and routes of administration are now available, and evidence on their comparative effectiveness and safety continues to evolve. As therapeutic choices expand, integrating the best available evidence with patient preferences and values becomes increasingly complex. The objective is to reach treatment decisions that the clinician and patient both agree upon, which offer benefits that outweigh the risks. This underscores the necessity of shared decision-making (SDM) when adjusting treatment regimens.

SDM is a collaborative process in which patients and healthcare providers jointly make care decisions, drawing on the best available scientific evidence, clinical expertise, and the patient’s values and preferences.5 With the increasing range of therapeutic options in rheumatology, SDM can assist patients and clinicians in reaching individualised decisions that reflect a thorough understanding of the benefits and risks of each option, as well as patient priorities. Major organisations, including the American College of Rheumatology (ACR) and the European Alliance of Associations for Rheumatology (EULAR), endorse SDM as a foundational principle in their guidelines for managing RA6,7 and AS.8,9 These bodies identify SDM between patient and rheumatologist as the primary element in formulating a treatment plan that aims to deliver optimal care.

Despite growing interest in SDM within rheumatology, there are no official guidelines or standardised tools for implementing SDM in clinical practice. Essential components of SDM include fostering awareness of choices, describing available options, deliberating, eliciting patient preferences, and making a final decision.10 Several frameworks have been proposed to integrate these elements, such as the Three-Talk Model,11 the five-step SHARE approach (Seek participation, Help comparison, Assess values, Reach decision, Evaluate decision),12 and the six-step model by Makoul and Clayman.13,14 The Three-Talk Model by Elwyn et al.11 is a straightforward and intuitive framework comprising three steps: introducing choice, describing options, and supporting patients in exploring preferences and making decisions. SDM can be facilitated by decision support tools, such as patient decision aids, which provide evidence-based information on available therapies to structure and support SDM discussions.15

This study will apply a structured SDM model to patients with RA and AS when choosing treatment options and will assess whether this approach leads to significant improvements in patient-perceived SDM compared to standard care. SDM models based on the Three-Talk Model, incorporating systematically developed decision aids and an artificial intelligence (AI)-based chatbot, will be compared with standard care. Additional outcomes will include levels of decisional conflict, patient empowerment, health-related quality of life, disease activity, and medication adherence.

METHODS

Study design and setting

This study is an open-label, multicentre, stratified cluster-randomised controlled trial conducted in Korea. Participants will be allocated in a 1:1:1 ratio to SDM Group I, SDM Group II, or the standard care group. Recruitment will take place at 13 university hospitals across Korea: Asan Medical Center (Seoul), Ajou University Hospital (Suwon), Chonnam National University Hospital (Gwangju), Daegu Catholic University Medical Center (Daegu), Gangnam Severance Hospital (Seoul), Hanyang University Guri Hospital (Guri), Hanyang University Hospital for Rheumatic Diseases (Seoul), Kangbuk Samsung Hospital (Seoul), Kosin University Gospel Hospital (Busan), Kyung Hee University Medical Center (Seoul), Soonchunhyang University Bucheon Hospital (Bucheon), Soonchunhyang University Cheonan Hospital (Cheonan), and Uijeongbu Eulji Medical Center (Uijeongbu). A total of 30 rheumatologists will participate.

Cluster randomisation will be implemented at two levels to ensure unbiased participant allocation. At Hanyang University Hospital for Rheumatic Diseases, patients will be individually randomised by the treating physician to one of the three groups (SDM Group I, SDM Group II, or standard care). This process ensures a balanced distribution of each physician’s patients across the study arms, reducing potential physician-related bias. At the remaining 12 centres, randomisation will occur at the hospital level; each centre will be randomly assigned to one study group, ensuring comparable group representation across all sites. This two-tiered randomisation approach minimises biases at both physician and institutional levels, thereby strengthening the validity and generalisability of the study findings.

Eligibility criteria

The study will enrol patients diagnosed with RA or AS who have active disease despite treatment with conventional synthetic disease-modifying antirheumatic drugs (DMARDs) or nonsteroidal anti-inflammatory drugs (NSAIDs), and who are currently considering biologic or targeted synthetic DMARDs.

Inclusion criteria

• 1. Adults aged 19–64 years.

• 2. Diagnosis of RA according to the 2010 ACR/EULAR criteria or AS according to the modified New York criteria.

• 3. Fulfilment of Korean reimbursement criteria for initiation of biologic or targeted synthetic DMARDs:

  • • For RA: Disease Activity Score-28 (DAS28) > 5.1, or between 3.2 and 5.1 with radiographic progression after 6 months of conventional synthetic DMARD therapy.

  • • For AS: Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) ≥ 4 despite at least 3 months of NSAID therapy.

Exclusion criteria

• 1. Inability to provide voluntary informed consent.

• 2. Inability or unwillingness to use a compatible device for the chatbot intervention.

• 3. Inability to complete study assessments.

Participants who withdraw consent or fail to attend required visits (despite three documented contact attempts) will be considered lost to follow-up.

Intervention

Three-Talk Model framework

This study utilises an SDM model tailored for patients with RA or AS considering targeted therapies, based on the Three-Talk Model framework (Fig. 1). Decision aids were developed according to International Patient Decision Aid Standards (IPDAS) through a systematic process, including field testing with patients.16,17,18 An AI-based chatbot was also created to support patients in reflecting on treatment preferences and values through interactive conversations; this chatbot underwent additional field testing with patients not enrolled in the present study.

Fig. 1. Shared decision-making model used in DESCIDE-RD. The shared decision-making model in DESCIDE-RD is structured into three core steps, based on the Three-Talk Model described by Elwyn et al.28 .

The SDM model comprises three core steps. At each stage, either decision aids or the AI-based chatbot is incorporated, depending on group allocation, to promote patient engagement and enhance decision-making transparency.

Step 1: Team Talk

In the initial step, the clinician and patient form a partnership to guide the decision-making process. The principal aim is to introduce the SDM approach and encourage active patient participation. The clinician presents a clear explanation of the decision to be made, informs the patient that multiple treatment options are available, and emphasises that the final decision should be shared.

At this stage, the treating physician identifies the patient’s current medical issue and states that the purpose of the consultation is to make a decision together. The physician informs the patient that multiple treatment options are available and emphasises the patient’s right to participate in decision-making based on their values and preferences, using information provided by the healthcare team. Team Talk will be conducted during the consultation immediately following study enrolment and consent. A minimum of 10 minutes is recommended for this stage. Key messages and example phrases for this step are listed in Table 1.

Table 1. Key messages and example phrases at each step of the shared decision-making model in DESCIDE-RD.

Step	Key messages	Example phrases
Team Talk	Explain the patient’s condition and decision context	“You need to make a decision about [medical condition] because [reason].”
	Acknowledge that more than one option exists	“Several treatment options are available.”
		“You may also continue your current treatment without choosing a new one.”
	Emphasise joint decision-making	“We will decide together.”
		“We will reach a decision based on the information I provide and your values and preferences.”
Option Talk	Present the options	“Your options are A, B, and C.”
	Explain benefits and harms	“The advantages of option A are [pros], and the disadvantages are [cons].”
		“If you choose not to pursue any of the options, [expected prognosis].”
	Explore patient preferences	“The decision may depend on your lifestyle and values.”
		“What is your preference?”
		“What concerns do you have?”
Decision Talk	Indicate the decision stage	“Now, let us decide together which treatment option to pursue.”
	Support decision-making	“Since our last consultation, what thoughts have you had?”
		“Would you like further explanation to help you decide?”
	Confirm the decision	“Which treatment do you think is most appropriate?”
		“I support your decision.”

Step 2: Option Talk

Option Talk focuses on a detailed discussion of the available treatments. The physician or nurse educator outlines the benefits, risks, anticipated outcomes, and potential side effects of each option using pre-developed decision aids. This stage ensures a clear, balanced understanding of all reasonable choices—including taking no action—how each aligns with individual values and circumstances, and the probabilities of potential benefits and harms, expressed numerically where possible. The clinician actively explores the patient’s preferences, needs, and concerns, and invites the patient to comment on each option.

In SDM Group I, patients use decision aids to support their understanding. These aids provide structured, accessible information about each treatment, including pros, cons, and potential outcomes. In SDM Group II, patients use both decision aids and the AI-based chatbot. After receiving decision aids in the clinical setting, patients engage with the chatbot on their mobile devices, allowing them to further consider and express their preferences outside the clinic. The chatbot generates a preference summary, which is then shared directly with the physician for the Decision Talk stage.

Option Talk will be conducted immediately following Team Talk. The recommended duration is 20–30 minutes. Key messages and example phrases for this step are presented in Table 1.

Step 3: Decision Talk

In the Decision Talk stage, the patient and physician make the final treatment decision collaboratively, informed by the preceding Option Talk. At this stage, the clinician summarises the key information from Option Talk, confirms the patient’s understanding of benefits, risks, and likely outcomes, and signals the transition to making a final choice together. The clinician reviews the available options, and supports the patient in clarifying and expressing preferences, addressing any remaining questions. For SDM Group I, this step is based on the pre-discussed decision aids, without input from the AI chatbot. For SDM Group II, the physician also reviews the chatbot-generated summary of the patient’s values, preferences, and concerns.

Decision Talk will be conducted at the first follow-up outpatient visit after Team Talk. The recommended duration is approximately 10 minutes. Key messages and example phrases for this step are summarised in Table 1.

Decision aids and chatbot components

Decision aids in this trial were developed through a systematic process with field testing in compliance with IPDAS. They present balanced, evidence-based information in plain language with side-by-side comparison tables covering indications, expected benefits and adverse effects, mode and frequency of administration, and practical considerations such as cost and access. The decision aids are provided in two complementary formats: a brief version that directly compares available options in tabular form for quick reference during consultation, and an expanded booklet version containing a detailed explanation of treatment options. Materials are available in Korean in print and digital formats (Supplementary Fig. 1). Decision aids are introduced during Option Talk and may be taken home for further review.

The chatbot is designed to help patients further explore their preferences and values regarding available treatment options. Following Option Talk in SDM Group II, the chatbot briefly reviews key features of each option and prompts the patient to express preferences, values, and concerns through structured questions. Topics include views on differences in treatment effectiveness, potential side effects, the balance between benefits and risks, and practical features such as dosing intervals and administration methods. The chatbot then compiles a structured summary of the patient’s responses, including final preferences and reasons (Supplementary Fig. 2). This summary is provided to the treating physician before Decision Talk, enabling a more focused, personalised discussion.

Group-specific interventions

Participants will be allocated to one of three groups: SDM Group I, SDM Group II, or the Standard Care Group (Fig. 2).

Fig. 2. DESCIDE-RD study flow. Participants in the DESCIDE-RD study will be assigned to one of three groups: SDM Group I, SDM Group II, or the Standard Care Group. The shared decision-making model will be implemented in SDM Group I and SDM Group II. Patients in SDM Group I will use decision aids, whereas those in SDM Group II will use decision aids together with a chatbot during Option Talk. Patients in the Standard Care Group will receive routine clinical care without decision aids or chatbot support.

SDM = shared decision-making, SDM-Q-9 = Shared Decision-Making Questionnaire-9.

• SDM Group I: Patients receive decision aids during Option Talk to support understanding of available treatments and facilitate informed choices. These aids are structured to guide patients through treatment options clearly and systematically.

• SDM Group II: In addition to the decision aids provided in SDM Group I, patients interact with an AI-based chatbot during Option Talk. The chatbot offers personalised guidance, enabling patients to reflect on and express their treatment preferences outside the clinical context. This engagement helps patients clarify their values and preferences prior to making a final decision.

• Standard Care Group: Treatment options are discussed in routine clinical practice without decision aids or chatbot support. The physician provides information on available treatments, encourages patient questions, and following this discussion, the physician and patient make the final treatment decision together.

Outcomes

The primary outcome is the Shared Decision-Making Questionnaire-9 (SDM-Q-9) score, which assesses the degree of SDM as perceived by patients.19 The SDM-Q-9 is a 9-item instrument using a 6-point Likert scale (0 = completely disagree to 5 = completely agree). Scores are calculated by summing item scores, multiplying by 100, and dividing by 45, resulting in a total score from 0 (no SDM perceived) to 100 (maximum SDM perceived). Higher SDM-Q-9 scores indicate greater perceived SDM.

Secondary outcomes include patient-reported measures: decisional uncertainty (Decisional Conflict Scale [DCS])20; patient empowerment (Korean Health Empowerment Scale [K-HES])21; patient satisfaction (Korean Leeds Satisfaction Questionnaire [K-LSQ])22; and health-related quality of life (5-level EuroQol-5 dimension, EQ-5D-5L).23 Disease activity, assessed by physician global assessment, DAS28 for RA or BASDAI for AS, and medication adherence, measured by patient-reported adherence and medication possession ratio (MPR), will also be evaluated.

Assessment and follow-up schedule

Patients will be randomised to one of the three study groups according to institutional allocation. All participants will attend three scheduled visits: Week 0, Week 4, and Week 12 (Table 2). The study timeline is as follows:

Table 2. Procedures and items assessed at each follow-up visit.

Time point		Week 0	Week 4	Week 12
Corresponding date		Day 0	Day 28 ± 14	Day 84 ± 30
Informed consent		v
Demographic data		v
Socioeconomic status		v
Health literacya		v
RA or AS medical historyb		v
Comorbid conditions		v
Physical examinationc		v
Laboratory assessments
	Routine laboratory panel	v	v	v
	Latent tuberculosis screening	v
	Hepatitis serology	v
Imaging studies		v
Previous medication use		v
Current medication use		v	v	v
Disease activityd		v	v	v
Patient-reported outcomes
	Shared Decision-Making Questionnaire-9	v	v
	Decisional Conflict Scale	v	v	v
	Korean Health Empowerment Scale	v	v
	Korean Leeds Satisfaction Questionnaire			v
	EuroQol-5 Dimension, 5-Level	v		v
	Medication adherence			v
Consultation duration		v	v	v

RA = rheumatoid arthritis, AS = ankylosing spondylitis.

^aIncludes the Health Literacy Index for the Community and Digital Health Technology Literacy.

^bIncludes disease duration, prior use of conventional synthetic disease-modifying antirheumatic drugs or nonsteroidal anti-inflammatory drugs, and disease markers such as rheumatoid factor, anti-cyclic citrullinated peptide antibodies, and HLA-B27.

^cIncludes height, weight, and blood pressure.

^dIncludes the Physician Global Assessment, Disease Activity Score-28, and Bath Ankylosing Spondylitis Disease Activity Index.

Week 0 (baseline):

• • Obtain written informed consent, conduct screening, and allocate participants to groups.

• • Collect baseline data: demographics, disease history, comorbidities, socioeconomic status, disease activity, and laboratory results.

• • Administer baseline questionnaires: SDM-Q-9, DCS, EQ-5D-5L, K-HES, Health Literacy Index for the Community, and Digital Health Technology Literacy.

• • SDM Groups I and II: Complete Team Talk and Option Talk. Patients complete a Treatment Preference Checklist and, in SDM Group II, are instructed to engage with the chatbot.

• • Standard Care Group: Routine counselling without decision aids or chatbot.

Week 4 (follow-up 1):

• • SDM Groups I and II: Complete Decision Talk using the Treatment Preference Checklist and chatbot-generated summary.

• • All groups: Complete SDM-Q-9, DCS, and K-HES; assess disease activity; and collect medication use and laboratory data.

Week 12 (follow-up 2):

• • All groups: Complete secondary outcome questionnaires (DCS, K-LSQ, EQ-5D-5L) and disease activity assessments.

• • Collect medication adherence data and laboratory results.

Sample size

The planned sample size is 300 participants. Participants will be allocated 1:1:1 to SDM Group I, SDM Group II, and standard care (100 per group). The study is designed to test whether the pooled intervention groups (SDM Group I and SDM Group II) are superior to standard care for change in SDM-Q-9 from baseline to week 4. The study will test the following hypothesis.

Hypothesis:

H₀: µ_t − µ_c = 0; H₁: µ_t − µ_c ≠ 0

µ_t: Mean change in SDM-Q-9 from baseline to week 4 in the intervention group

µ_c: Mean change in SDM-Q-9 from baseline to week 4 in the standard care group

In a previous study with a similar design,24 the mean ± standard deviation of SDM-Q-9 scores before and after decision-making were 63.7 ± 21.6 and 71.2 ± 19.2 in the intervention group and 67.9 ± 21.1 and 67.7 ± 22.5 in the control group. Based on these results, the mean changes (µ_t and µ_c) were estimated as 7.5 for the intervention group and −0.2 for standard care, assuming a common standard deviation of 22.5. With a total sample of 300 (200 in the intervention group and 100 in the standard care group), power to test the hypothesis using a t-test at α = 0.05 was approximately 80% (79.5%). Sample size calculations were performed with PASS 2016.

To address potential deviations from the assumed effect size or variability, an interim review of the sample size assumptions will be undertaken during the trial. If required, the total sample size may be adjusted according to prespecified adaptive design principles to maintain adequate statistical power.

Statistical analysis

Analyses will follow the intention-to-treat principle. Descriptive statistics will summarise baseline characteristics. Between-group differences in primary and secondary outcomes will be evaluated using appropriate statistical tests (e.g., analysis of variance, χ²). Changes over time will be examined using repeated-measures analyses.

Ethics statement

This study will be conducted in accordance with the Declaration of Helsinki. The protocol has received approval from the Institutional Review Board of Hanyang University Hospital (approval No. 2024-12-022). Written informed consent will be obtained from all participants, and a copy of the signed consent form will be provided. No study-related procedures or examinations will be performed prior to obtaining informed consent. The protocol is registered at the Clinical Research Information Service (http://cris.nih.go.kr, trial registration number KCT0011310).

DISCUSSION

This protocol describes a nationwide, multicentre, stratified cluster-randomised controlled trial evaluating the effectiveness of a structured SDM process in enhancing patient participation in treatment decisions for individuals with active RA or AS. Although SDM is advocated in rheumatology, its widespread adoption in routine clinical practice remains limited. Most previous studies have concentrated on developing patient decision aids to reduce decisional conflict.25,26 However, little attention has been devoted to structuring the SDM process or assessing its large-scale implementation in this context.27

This study seeks to address this gap by developing and validating a comprehensive SDM model that empowers patients and clinicians as equal partners in treatment decisions. The SDM process is based on the Three-Talk Model, which offers a clear framework for patient–clinician interaction at each stage of the decision-making process.28 The primary outcome is the level of SDM, assessed using the SDM-Q-9, a widely used patient-reported measure of involvement in decision-making.29 By comparing intervention groups with standard care, this study will determine whether a structured SDM approach increases patient engagement. Secondary outcomes—including decisional conflict, patient empowerment, satisfaction, quality of life, disease activity, and medication adherence—will be measured to evaluate the broader impact of SDM on patient outcomes.

An integral component of the intervention is the AI-powered chatbot, designed to assist patients in SDM Group II in reflecting on and clarifying their preferences outside the clinical environment. This study will compare outcomes between SDM Group I (decision aids only) and SDM Group II (decision aids plus chatbot) to assess the added benefit of the chatbot in promoting patient engagement in decision-making. Furthermore, subgroup analyses will determine whether the chatbot’s effectiveness differs according to digital health literacy or other patient characteristics.

By implementing a structured SDM model with integrated tools, this study will strengthen the evidence base in rheumatology and offer insights into how such interventions can enhance patient-centred decision-making. The findings will also guide the development of personalised SDM strategies, allowing for tailored decision-making support based on individual patient profiles.

SUPPLEMENTARY MATERIALS

Supplementary Fig. 1

Brief version of the decision aids.

Supplementary Fig. 2

Chatbot application for patients and summary interface for physicians. (A) Patient-facing interface of the chatbot application. (B) Artificial intelligence-generated summary displayed on the web platform for physicians.