Standardized Interoperable Data Collection for Myositis Research: Developing Expert Consensus on Common Data Elements for Myositis Outcome Measures

Abstract

Background.

Recent progress has been made in developing validated myositis outcome measures. However, critical deficiencies remain for data standardization across myositis registries. While the National Institute of Health (NIH) Common Data Elements (CDE) Repository has been developed to facilitate standardized data collection and sharing, few myositis-specific CDEs currently exist. We developed CDEs for myositis outcome measures using novel data science strategies.

Methods.

Data dictionaries of myositis registries were examined to understand how outcome measures are currently captured. We used the Linked data Modeling Language (LinkML), an open-source data modeling framework, to develop computable CDEs. After drafting CDEs for myositis core set activity measures (CSMs), an international conference was held with an expert myositis panel to reach consensus on the coding of CDEs and prioritize additional measures for CDE creation, using Delphi and modified nominal group techniques. This workflow was repeated for the prioritized measures in the second phase.

Results.

A workflow was established for CDE creation. CDEs for ten myositis CSMs were drafted. After receiving comments to improve their coding, universal agreement among participants was reached for CSM CDEs. The prioritized measures for future CDEs included myositis response and classification criteria, damage measures, physical function measures, and PROMIS instruments. CDEs for 18 additional measures were discussed at a second consensus conference. Similarly high agreement rates were achieved, except for flare criteria. Altogether 852 new CDEs were created for 27 myositis forms and achieved consensus, readying their deposit in the NIH CDE Repository.

Conclusion.

Leveraging multispecialty expertise in myositis and its patient communities and data science expertise of the National Library of Medicine, the first myositis-specific CDEs have been developed to accelerate the ability to conduct interoperable myositis clinical studies and therapeutic trials. The workflow established here should also benefit creation of CDEs and data sharing for other autoimmune diseases.

Introduction

Idiopathic inflammatory myopathies (IIM) are rare systemic autoimmune diseases characterized by chronic inflammation of muscle and are associated with high morbidity.¹ To develop international multidisciplinary consensus on the conduct and reporting of clinical studies in the IIM, the International Myositis Assessment and Clinical Studies Group (IMACS) was established more than 20 years ago.² IMACS has achieved many milestones, including development and validation of core set measures (CSMs) of disease activity and damage of muscle, skin and other target organs in adult and juvenile IIM.^3,4 Additionally, the Paediatric Rheumatology International Trials Organization (PRINTO) developed a set of CSMs for juvenile dermatomyositis (JDM), with a number of measures in common with the IMACS CSMs⁵ (Table 1). These CSMs have been widely used in IIM clinical studies and therapeutic trials.^6–8

Table 1.

Myositis Core Set Activity Measures Reaching Consensus at First Myositis Common Data Elements Consensus Conference.

Myositis Measure (reference)	Characteristics	CDE Features and Key Revisions
Common Variables	Study ID number, visit number, visit date, assessor, certification	Data elements that are reused in every measure 7 CDEs created
Physician and Patient/Parent Global Disease Activity*4	Overall disease activity on 10 cm VAS	Encode line length and adjustment to 100 mm VAS scale in CDEs Added 21-point numeric rating scale on form Ordinal scale made optional on form 7 CDEs created and 1 reused for PGA and 8 CDEs created for Patient/Parent Global Activity
HAQ and CHAQ*4	Assessing activities of daily living in 8 domains,	CDEs included scoring of domains, total score and adjusted scoring to allow for 1-2 missing domains Clarified use of assistive devices in CDEs Made Pain VAS optional in CDEs 49 and 55 CDEs created for HAQ and CHAQ, and 1 reused in each
MMT-8*4	Assesses isometric strength in eight proximal, distal and axial muscle groups on the Kendall scale	Bundled muscle strength measure, muscle group, dominant side, side of testing in CDEs Added questions to specify the dominant side and laterality of test on form Recommendation to create persistent URL of IMACS website with detailed instructions, link to it in CDE repository 9 CDEs created
Myositis Disease Activity Assessment Tool, Extramuscular Global Disease Activity4	Assessment of disease activity in 7 organ systems, primarily assessing extramuscular activity	Recommended making separate form for Extramuscular Global Activity by VAS and include the 21-point numeric rating scale 90 CDEs created and 1 reused
Muscle Enzymes and Metabolites*4	Laboratory results of serum CK, aldolase, AST, ALT, LDH, and creatinine	Bundled enzyme or metabolite, value, unit of measure, and upper/lower limit of normal values in CDEs SGOT and SGPT added to naming of CDEs Results of enzyme tests made optional in CDEs Use both IU/l and microcat/dl, and for creatinine mg/dl and umol/l, to include European units on form 11 CDEs created
Childhood Myositis Assessment Scale*4	Performance based tool assessing muscle strength, endurance, and physical function in children	Items and scoring coded in CDEs, including adjusted score to allow for 1-2 missing items 28 CDEs created
Disease Activity Score4	Overall disease activity in JDM, assessing muscle strength/function and skin activity.	Items and scoring coded in CDEs, including adjusted score to allow for 1-2 missing items 39 CDEs created

Eleven myositis forms relating to CSMs were voted for consensus at the first Myositis CDE Consensus Conference. CDASI, Derm Life Quality Index, and CHQ-PF50 were also discussed; however, not pursued for the CDE repository due to licensing requirements.

Abbreviations: CDE: Common data element, CSM: Core set measure, HAQ: Health Assessment Questionnaire, CHAQ: Childhood Health Assessment Questionnaire, MMT-8: Manual muscle testing of 8 muscle groups, MDAAT: Myositis Disease Activity Assessment Tool, DAS: Disease Activity Score, CHQ-PF50: Child Health Questionnaire Parent Form 50; VAS, Visual analog scale; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase; LDH: Lactate dehydrogenase; SGOT: Serum glutamic-oxaloacetic transaminase; SGPT: Serum Glutamic Pyruvic Transaminase; IU: International Unit, URL: Uniform Resource Locator.

^*Could be relevant to other neuromuscular and/or autoimmune rheumatic diseases

Several other milestones towards standardized data collection have been achieved, including development of patient-reported outcome measures (PROMs) in collaboration with an OMERACT working group⁹; and developing consensus standards for conducting IIM therapeutic trials and their design.¹⁰ IIM Classification Criteria endorsed by the European League Against Rheumatism/American College of Rheumatology (EULAR/ACR) perform better than previous classification criteria, capturing a broader range of clinical subgroups, and are now preferentially used in IIM clinical studies.¹¹ ACR-EULAR Response Criteria for DM, PM, and juvenile DM (JDM) that are hybrid criteria also provide gradations of response (minimal, moderate, and major clinical improvement).¹¹ These criteria have been further validated, performing consistently across multiple studies, and are now used as the primary efficacy endpoint for many myositis therapeutic trials. While significant progress has been made over the last two decades in the development of standardized outcome assessment, there are critical deficiencies for data standardization which hamper the comparability of studies. For example, a common source of discordance between databases is assessment of Patient Global Disease Activity. This assessment is captured using visual analog scales at some centers, while others use numeric rating scales with different levels of precision and units. To work on addressing these challenges, a Data Harmonization Scientific Interest Group has recently been established within IMACS to bring together experts with interest and/or experience in data collection and data sharing across myositis datasets.

The National Institute of Health (NIH) Common Data Elements (CDE) repository aims to promote research data interoperability, helping researchers share and combine datasets. CDEs are “standardized, precisely defined questions paired with a set of specific allowable responses, used systematically across different sites, studies, or clinical trials to ensure consistent data collection”. The NIH CDE Repository (https://cde.nlm.nih.gov/home) contains 1,603 standard CDEs managed by the NIH CDE Working Group and 22,743 CDEs (as of August 9, 2025) compiled into 19 collections. Three of the CDE collections containing 177 CDEs are “NIH-Endorsed” by the NIH CDE Governance Committee. The criteria for NIH endorsement of CDEs includes i) clear definition of the variable with a prompt and a response, ii) evidence of reliability and validity, iii) human and machine-readable format, iv) recommended or designated by a recognized NIH body, and v) clear licensing and intellectual property status, with preference for open source or for creative commons attribution license. Currently, there are no myositis-specific CDEs and few for autoimmune disease in the NIH CDE repository, although the development of CDEs for autoimmune disease is a high priority¹². Development and adoption of CDEs representing the CSMs and other validated outcome measures, as well as IIM classification and response criteria would help address data interoperability in the myositis field, aiding researchers in sharing and combining datasets and further standardizing data collection. This is particularly important for rare and heterogenous diseases such as IIM where single center cohorts are often small and multi-center collaboration is critical for research. The CDEs developed would provide a standard terminology of concepts, standardized data structures, and independent semantics that would be reusable across physical data models, forms, and datasets, and support the FAIR data principles.¹³

A collection of CDEs for myositis outcome measures and criteria was developed by a working group of myositis researchers, data scientists and informaticians. These were presented to a group of myositis experts including rheumatologists, neurologists, dermatologists, and patients/patient advocates from the IMACS community. Their accuracy and utility in representing myositis CSMs, other outcome measures and validated criteria were discussed and vetted in two consensus conferences. These consensus-approved myositis CDEs have now been submitted to the NIH CDE Governance Committee for endorsement and placement in the NIH CDE Repository. The purpose of this report is to inform about the development of these CDEs for myositis as a model for CDE development in other autoimmune diseases.

Methods of CDE development and mapping for myositis outcome measures and consensus conference organization

CDE Development for Myositis CSMs.

In addition to the IMACS Outcomes Repository, which includes a database of the myositis outcome measures, additional data dictionaries from several myositis databases, including the MYONET registry¹⁴, JDM Cohort Biomarker Study and Repository (JDCBS)¹⁵, PRINTO, Childhood Arthritis and Rheumatology Research Alliance (CARRA), INSPIRE-IBM registry, and several university myositis research centers were examined to understand how the measures are coded and captured across different databases (Figure 1). Copyright status and appropriate permissions for all measures were obtained from copyright holders and publishers to deposit the CDEs in the NIH CDE Repository.

Figure 1.

Flow diagram of Myositis Common Data Element Project.

Abbreviations: CDEs: Common Data Elements; OMOP: Observational Medical Outcomes Partnership;

UMLS: Unified Medical Language System; NIH: National Institute of Health

The working group developed CDEs for CSMs from 10 IMACS and PRINTO CSMs as well as other myositis outcome measures and EULAR-ACR classification and response criteria. Generally, CDEs are constructed using human-readable labels and descriptions; however, we developed CDEs that are computable, meaning that the CDEs’ information and metadata are explicitly captured in a machine-readable format that leverages existing data standards for reuse.¹⁶ The goal of creating computable CDEs is to maximize their findability, accessibility interoperability, and reusability at scale in accordance with FAIR (findability, accessibility, interoperability, and reusability) principles for data sharing, which aligns with the main purpose for creating CDEs.^13,17

To develop computable CDEs, we used the Linked data Modeling Language (LinkML), which is an open-source data modeling framework for creating schemas for structured, linked data¹⁷. We selected this framework because of its compatibility with numerous Semantic Web standards and data formats; the ability to create both machine-readable and machine-actionable schemas; support for reuse and extension of data elements; its ability to support mapping to existing data standards (e.g., terminologies and ontologies, including Unified Medical Language System [UMLS] and Observational Medical Outcomes Partnership [OMOP]); and extensive tooling for data validation, conversion, and management. Components of a LinkML schema include classes, slots, and enumerations. A class is a template for organizing data based on shared characteristics, while slots and enumerations are templates for representing data elements and value lists, respectively. Computable CDEs were developed using a protocol developed in-house.

In this myositis CDE project, where possible, CSMs and other fields were bundled based on identified design patterns that are common to each question to reduce the number of CDEs needed for a CSM or other measures to provide a flexible way to combine them for any expanded or short-form versions of the assessment. CDE bundling is a step in the direction of design patterns that supports automation, scalability, reusability, uniform standardized representation, and validation of CDEs. For example, we did not create 8 different CDEs for testing the strength of each muscle group for the MMT8 tool; instead, we create one CDE for strength and another for the muscle group assessed. We then re-used and bundled this pair of CDEs for each muscle group. This approach reduces redundancy and duplication in the NIH CDE repository. Examples of myositis forms that utilize bundled CDEs included Manual Muscle Testing 8, Serum Muscle Enzymes and Metabolites, and Timed Up and Go.

Mapping Myositis CDEs to Standardized Terminologies.

The Observational Health Data Science and Informatics (OHDSI) web application Athena-OHDSI Vocabularies Repository (Odysseus Data Services, Inc.) was utilized to facilitate mapping from the candidate myositis CDEs to OMOP vocabulary, where they existed (https://athena.ohdsi.org/search-terms/start). Both phrase and exact search parameters were used to identify equivalent OMOP concepts. CDE measures were also mapped to UMLS concepts that were identified as close or exact matches via the UMLS Metathesaurus Browser (https://uts.nlm.nih.gov/uts/umls/home).

Consensus Conferences with Myositis Experts.

In the first phase of the project, CDEs were developed for myositis CSMs, and an in-person consensus meeting was held. The goals of the meeting were to discuss the coding of CDEs for myositis CSMs, receive critical feedback from myositis experts, and prioritize additional myositis study forms for creation of new CDEs in the next phase of the project. Key stakeholders were identified and invited to participate, including myositis health care professionals/investigators and patients/patient advocates from IMACS (Supplementary Table 1). This included 13 voting panelists, 11 discussants, and 10 observers in the first consensus meeting. The health professional participants were identified as recognized experts in the myositis field from a variety of specialties and geographic locations with expertise in outcome measures, including those who host, manage, and contribute to local or global myositis databases, and those who have led and/or participated in the development of myositis CSMs and/or in clinical trials. The patient/advocate participants were identified through The Myositis Association, Cure JM Foundation, and Myositis Support and Understanding based on their experience in participating in myositis research studies. In the second phase, CDEs were developed for the outcome measures prioritized by the expert panel in the first consensus conference, and a second consensus meeting was held virtually 6 months later. This study did not involve human subjects and therefore did not require institutional review board approval.

Pre-meeting surveys.

All participants were provided with study forms and instructions for each outcome measure, relevant literature supporting validation, and the CDEs for each of the myositis study forms as .csv or .yml files to review before the conference. Participants received an electronic REDCap-based survey to provide feedback on the coding of the CDEs for each myositis measure. The results of the survey were reviewed by the study team prior to the conference.

Consensus conference methodologies.

Consensus sessions involved brief presentations about each myositis measure or criteria set, followed by review of the CDE survey results, and voting on the corresponding CDEs for that measure. Two consensus methodologies (Delphi method, modified nominal group technique (mNGT)) were utilized in tandem to discuss the CDEs for each myositis measure. After each CDE was shared, panelists and discussants were asked to submit any final comments using the REDCap-based survey; survey responses were shared with participants to inform discussions. Any suggested changes to the CDEs agreed to during the discussions were recorded and subsequently incorporated into edited CDE versions following the conference.

This initial step of comments and group discussion was followed by mNGT, which involved formal electronic voting. Consensus was defined as ≥80% agreement among the panelists. If consensus was achieved, then the process proceeded to the next set of CDEs for the next myositis measure without further discussion. If consensus was not achieved (<80% agreement), then a round robin discussion was held, and each panelist was asked to share their opinion and any comments about the CDEs for a measure, which continued until there were no new comments. The round robin discussion was followed by a second round of electronic voting to aim to reach consensus.

Consensus on Myositis Core Set Disease Activity Measure CDEs at First Conference.

CSMs were discussed at the first conference and all achieved consensus, with the panelists endorsing the CDEs as presented and discussed. Altogether 301 new CDEs were created for 11 forms. The majority of comments related to clarifications of the content of the forms, rather than the coding of the CDEs. Highlights of the discussions are presented here and in Table 1.

A Core Set of Variables common to every myositis measure were encoded in a single common variables set of CDEs. This included patient identification number, study patient number, assessment number, assessment date, reporter or assessor, certified by (person checking that the data are complete and within range), and certification date. Based on feedback from the participants, some of these variables were recommended to be made optional.

Physician, Patient and Parent Global Disease Activity.

The Physician, Patient and Parent Global Disease Activity CDE sets were reused based on overlapping components and similar coding between forms. The CDEs were coded in millimeters (mm) to accommodate the integer requirement for electronic capture in the application to a REDCap database. Both vertical marks and line lengths of the visual analog scales (VAS) were coded. Feedback received from the Delphi survey stated that numeric rating scales (NRS) are currently more frequently used than VAS and have potential to be more informative (Table 1). One panelist pointed out the infrequent use of the ordinal scale. Two participants recommended adopting only one scale instead of having both ordinal and VAS for the same measure. The group discussion led to a collective decision to make the VAS a required element and the ordinal scale optional and to develop a 21-point NRS for these forms¹⁸.

Health Assessment Questionnaire (HAQ)/Childhood Health Assessment Questionnaire (CHAQ).

Even though the current NIH CDE repository does not include scoring of forms, this was determined to be an important addition by the study team. Thus, the HAQ/CHAQ CDE sets included domain scores, a total score ranging from 0 to 3, as well as an adjusted score, which allows for 1 or 2 missing domains, along with instructions for scoring. The feedback received from the Delphi surveys included a comment to clarify that scoring should be adjusted when a patient uses assistive devices or assistance from another person. Another comment reported a possible difference between the HAQ used in Europe, which generally does not include a Pain VAS. Based on group discussions, the pain scale was made optional and the scoring adjusted when assistive devices/aids are used.

Manual Muscle Testing (MMT).

The MMT-8 CDE set was coded both as unilateral using right-sided proximal, distal and axial muscles with a range of 0 to 80, as originally developed¹⁹ and as a bilateral set with a range of 0 to 150, as is frequently used in therapeutic trials. The CDEs bundled muscle strength measure, muscle group, dominant side, and the side of testing. The coding included a potential score, adjusted score, and a percent score adjusted to a final score with the same denominator (80 or 150) and allowing for missing value(s). The suggestions received from the Delphi survey included adding instructions for performing the testing, specifying the testing side and dominant side, and adding an image or URL of the study form.

Myositis Disease Activity Assessment Tool (MDAAT) and Extramuscular Global Disease Activity.

The MYOACT CDE VAS scales were coded as 100 mm for all 7 organ systems. Twenty-three extramuscular items and 3 muscle items were categorized as not present, improving, same, worse and new, as used in the MDAAT.⁴ It was suggested that the Extramuscular Global VAS also be completed on a NRS. As the MDAAT, a copyrighted tool, cannot be altered, it was agreed that a separate form would be created for the Extramuscular Global Activity that includes both VAS and NRS scales, which is a core set measure for the ACR-EULAR Myositis Response Criteria.^20,21

Serum muscle enzymes.

A data collection form for serum levels of muscle enzymes specified the result and upper limit of normal for each, including creatine kinase, aldolase, aspartate transaminase (AST), alanine transaminase (ALT), and lactate dehydrogenase (LDH), as surrogate laboratory measures of disease activity. Serum creatinine is also included as a measure of muscle atrophy or damage.³

The serum muscle enzymes CDE set included coding of each blood laboratory value in IU/L and microkat/l, due to unit differences in the United States and Europe. Upper limit of normal was provided, and values were adjusted to a common upper limit of normal (252 IU/L for CK, 7 IU/L for aldolase, 226 IU/L for LDH, 34 IU/L for AST, and 41 IU/L for ALT) to be able to compare values between centers, regardless of demographic characteristics (age, gender, race). Minimum and maximum potential values were included. Serum creatinine was encoded in mg/dl and umol/L and adjusted to the lower limit of normal of the testing laboratory and a common lower limit of normal of 0.7 mg/dl. In the CDEs for this form, enzyme or metabolite, value, unit of measure, and upper/lower limit of normal values were bundled. Suggestions received from the Delphi survey pointed to the more common use of SGOT and SGPT terminology, instead of AST and ALT and the limited availability of aldolase in Europe. Based on this feedback, the enzyme names were updated to include SGOT and SGPT as well as AST and ALT, and the results of each of the laboratory tests were made optional, as the availability of the tests varies between institutions. The name of the form was suggested to be changed to Serum Muscle Enzymes and Muscle Metabolites, given that creatinine is not a muscle enzyme.

Other CDEs that reached consensus at the first conference included Childhood Myositis Assessment Scale (CMAS) and Disease Activity Scale (DAS) (Table 1).

At the first consensus conference, two myositis measures were discussed, the Child Health Questionnaire - Parent Form 50 (CHQ-PF50), part of the PRINTO CSMs for JDM, and the Cutaneous DM Disease Area and Severity Index (CDASI), which assesses skin activity and damage. Both tools are licensed with restrictive use, precluding their dissemination through the NIH CDE Repository. Therefore, neither tool was coded nor included for voting.

At the end of the first meeting, a survey was conducted to prioritize 44 additional myositis forms as well as the criteria for creation of CDEs at the next phase. The participants were requested to consider the evidence of validation for each form while voting, as required by the NIH CDE Governance Committee (Supplementary Table 2). Fifteen of 25 forms (60%) received at least 30% agreement. Some prioritized forms, including inclusion body myositis-related and cancer screening forms were not pursued for CDE creation due to lack of or limited validation studies for these forms. Instead, some forms with <30% agreement, including the PROMIS forms, were prioritized due to presence of validation studies. Some forms that did not receive high endorsement, including skin global assessment, thigh MRI, and core elements forms, will be coded in REDCap only and CDEs would not be created for the NIH CDE Repository (see below).

Consensus on Myositis Outcome Measure and Criteria CDEs at the Second Conference.

A second Myositis CDE consensus conference was held six months later as a virtual conference with 23 panelists/voting members, almost all of whom participated in the first conference. CDEs for classification and response criteria, damage measures, patient reported outcome measures (PROMs), and functional tests were discussed, representing 18 measures. Altogether, 551 new CDEs were created from 17 myositis study forms which achieved consensus, and 43 CDEs were reused (Table 2).

Table 2.

Myositis Criteria and Outcome Measures Reaching Consensus at Second Myositis Common Data Elements Consensus Conference

Myositis Measure (reference)	Characteristics	CDE Features and Key Revisions
EULAR-ACR IIM Classification Criteria11	Combination of 16 variables, including age of onset, pattern of weakness, characteristic rashes, laboratory results, and optional muscle biopsy findings. A classification tree defines the myositis subgroup	CDEs for items and subclassification criteria developed. Clarified that features are present at any time during disease course on form 47 CDEs created
ACR-EULAR Response Criteria for DM and PM; ACR-EULAR Response Criteria for JDM20,21	Hybrid measure derived from IMACS and PRINTO CSMs for DM, PM and JDM, with a continuous score (Total Improvement Score) and improvement categories	Reused CDEs from CSMs Clarified two visits used (baseline and follow-up) in the CDEs 34 CDEs created and 11 reused
Physician and Parent/Patient Global Damage 4	Overall disease damage on 10 cm VAS, with optional ordinal scale	Coded similarly to Disease Activity Globals Added numeric rating scale on form Made ordinal scale optional on form Clarified definition of damage for patients/parents on form 5 CDEs for Physician and 2 CDEs for Patient/Parent Global Damage; 2 CDEs reused in each
Myositis Damage Index4	Physician reported; extent and severity of damage of 11 organ systems	VAS scales and items in 11 organ systems coded in CDEs, along with Severity and Extent of Damage and Extended Damage scores. Clarified damage is captured regardless of etiology, and completed every 6 months on form 159 CDEs created and 5 CDEs reused
PROMIS Short Forms*: Fatigue Short Form 7a, Pain Interference Short Form 6a, Physical Function Short Form 8a9	PROMs to assess fatigue, pain interference and physical function over past 7 days, with t score normalized for general population.	Mapped question IDs to CADSR for cross-referencing in CDE repository 7 CDEs for PROMIS Fatigue, 6 for Pain and 8 for Physical function forms created
PROMIS Forms*: PROMIS 57 v 2.0 ; PROMIS 49 Parent Proxy Profile, PROMIS 49 Pediatric Profile v2.0 22	PROMs with t score normalized for general population. Adult form includes assessment of several domains over past 7 days: Physical Function, Pain Interference, Fatigue, Social Role, Anxiety, Depression, and Sleep Disturbance. Pediatric version includes Peer Relations instead of Social Role	Mapped question IDs to CADSR for cross-referencing in CDE repository 57 CDEs for PROMIS 57 and 49 CDEs each for PROMIS 49 created
Functional Index 34	Performance based tool to assess endurance in proximal upper and lower extremities and axial muscle groups in adult DM/PM patients	Number of repetitions for each muscle group, percentage of repetitions relative to maximum possible, reasons for limitations and total score included in CDEs Clarified to score dominant side, but if there is a functional limitation, then score on non-dominant side on form 54 CDEs created and 2 reused
30-sec Sit to Stand Test*24	Maximum frequency of arising from chair and returning to seated position within 30 sec	Number of full repetitions, arm or device use, testing completion included in CDEs Clarified chair height, with thigh and calf at 90-degree angle and feet touching floor on form 10 CDEs created and 1 reused
Timed Up and Go Test*25	Time to stand up from chair, walk 3 meters, and sit down	Time to complete the test, dominant side, status of test completion, and side of gait aide used included in CDEs Bundled aide type and side of aide in CDEs 15 CDEs created and 6 CDEs reused
Six/Two-minute Walk Tests*25	Distance walked in two and six minutes	Walk distance, laps, heart rate, blood pressure, oxygen saturation, aide use and constraints in performing the test included in CDEs. Walk speed and gait velocity as calculated variables provided. Clarified lap distance and length of track as 30 meters on form Clarified oximetry to be measured immediately at end of test on form 64 CDEs created

Seventeen myositis measures were voted for consensus at the second conference. Additionally, CDEs for the IMACS Preliminary Criteria for Disease Flare were discussed, but not pursued for the CDE repository due to <80% agreement²⁶

Abbreviations: CDE: Common data element, CSM: Core set measure, PM: polymyositis, DM: dermatomyositis, JDM: juvenile dermatomyositis; VAS: Visual analog scale.

^*Could be relevant to other neuromuscular and/or autoimmune rheumatic diseases.

2017 EULAR-ACR IIM Classification Criteria.

CDEs were created for each subclassification category including PM/IMNM, DM, IBM, amyopathic DM, JDM, and other juvenile myositis. Scoring was not included. Given the active efforts to update these criteria, some panelists expressed concern for including CDEs for this form; however, the majority felt it was important to include the currently used criteria as it will take a while for new criteria to become available.

IMACS Preliminary Criteria for Disease Flare.

IMACS CSM CDEs were used as input variables. Absolute and relative percent change between baseline and follow-up values were included. Consensus was not achieved at 50% (10 out of 20); therefore, a round-robin discussion took place followed by a second vote. Concerns about the form included lack of validation, not being a data driven criteria but consensus criteria, missing important elements of flare such as pain, fatigue, and skin disease, and not capturing the range and extent of patient symptoms; while other panel members thought that the criterion is a good starting point and CDEs should be published. Consensus remained at 52.4% (11 out of 21 participants); therefore, CDEs for disease flare will not be submitted to the NIH CDE Governance Committee.

ACR-EULAR IIM Response Criteria for DM, PM and JDM.

IMACS CSM CDEs were used as input variables, and all coding elements were initially re-used from the disease flare criteria form. Panelists recommended revising the naming of the baseline visit to indicate that it is a prior visit and not necessarily the first patient visit. As CDEs for the disease flare criteria will not be submitted, the response criteria CDEs were re-developed and the CDEs were instead reused from CSM CDEs where possible. New CDEs were developed for absolute change, absolute percentage change, and the distinction between IMACS and PRINTO variables.

Other CDEs that reached consensus at the second conference included Physician and Parent/Patient Global Damage, Myositis Damage Index, PROMIS tools^9,22, Functional Index-3,²³ 30-second Sit-to-Stand (STS) Test, Timed Up and Go test and Six- and Two-minute Walk Tests²⁴ (Table 2).

Summary and Future Directions.

Leveraging the broad multispecialty international expertise in myositis and patient communities through the IMACS myositis consortium and the data science expertise of the National Library of Medicine, the first series of myositis-specific CDEs have been developed to accelerate the ability to conduct interoperable myositis clinical studies and therapeutic trials. The data science strategies and methodology utilized in this study will serve as a paradigm for all autoimmune diseases to develop their disease-specific CDEs. Several of the CDEs, including Physician and Patient/Parent Global Activity and Global Damage, as well as HAQ/CHAQ and PROMIS forms, will have utility for other autoimmune diseases, while MMT-8, muscle enzymes and metabolites, and functional tests will have broad applicability for other neuromuscular diseases. Additionally, the identification of critical CDEs will help advance database research approaches.

Overall, the panelists achieved consensus on the coding of all but one of the 28 myositis study forms, creating 852 CDEs in this project. License requirements create barriers to use of forms, such as CHQ-PF50 and CDASI, and their distribution through the CDE repository. For all the other forms, high rates of agreement could be attributed to the panelists’ familiarity with the measures. In fact, most of the panelists had participated in the development of myositis CSMs that served as the foundation for the CDEs, had prior experience with consensus methodologies, and had a track record of participation in other myositis consensus projects, including the development of response criteria and clinical trial design elements. A notable weakness was the absence of CDEs for IBM measures, which largely lacked validation and/or agreement on specific measures, and this should be addressed in the future. Additionally, the CDEs created for myositis study forms are in English, which could significantly limit the global accessibility and adoption of these tools in non-English speaking countries. Nevertheless, the lessons learned and the methodology developed in this study could serve as a helpful model for development of CDEs in other languages.

Following similar approaches in other autoimmune diseases should expand the capacity for robust generalizable data sharing for these conditions. Further, as new myositis tools become available in the future, the workflow developed here should serve as a helpful guide. As guidance for groups who plan on conducting similar work in other conditions and myositis, we would like to highlight the importance of several elements that made this project successful. Importantly, this project required substantial time commitment from three teams with complementary expertise, including those with data science expertise, myositis subject matter experts and patient representatives, and a core team who organized and coordinated the efforts. Due to differences in collection of the same forms across different databases, we suggest the subject matter expert team have a wide international representation and include individuals with expertise in developing and/or using these forms. Lastly, we strongly recommend future groups take time at project commencement to understand the licensing and copyright requirements of each instrument and review the literature on validation of the instruments before proceeding with CDE development.

As the next steps of this study, the CDEs achieving consensus in this project are in process of being submitted for endorsement by the NIH CDE Governance Committee and uploaded to the NIH CDE repository, which is a publicly available website (https://cde.nlm.nih.gov/home). A second method to facilitate the adoption of the myositis CDEs developed in this project will be to develop REDcap data dictionaries and electronic case report form collection instruments for each of the 27 myositis study forms achieving consensus. REDCap is a secure web-based electronic data capture platform designed for building and managing databases for clinical and translational research studies. Each myositis study form created in REDCap will include individual components of the form as well as the built-in automated scoring. For two forms, CHAQ and MDAAT, we were unable to make the REDCap forms available due to licensure restrictions but will only share their data dictionaries. REDCap codebooks of the CDEs will be available on the IMACS website following deposit in the NIH CDE Repository. The link between the NIH CDE repository and REDCap will facilitate the dissemination of the myositis CDEs and use by myositis researchers, further promoting data compatibility and data sharing. Additionally, our group is working to include the myositis CDEs in the REDCap Shared Library repository for downloading and use by researchers. Using CDEs and REDCap forms for disease activity and damage, PROMs, other key assessment tools and criteria developed in this study will facilitate development of new myositis registries for investigators and advance the design and conduct of future therapeutic trials for these rare autoimmune diseases.

Significance & Innovation.

• This project fills a major gap in myositis research by creating the first comprehensive, computable myositis-specific CDEs using novel data science tools.

• This project establishes a rigorous and scalable approach for harmonizing outcome measures that can be leveraged in other autoimmune diseases.

Appendix

The participants who contributed to this project:

Myositis CDE Study Group:

Lisa G. Rider (co-principal investigator, NIEHS, NIH), Richard H. Scheuermann (co-principal investigator, NLM, NIH), Didem Saygin, Matthew Diller, Varsha Surampudi, Mark Bodkin, Payam Noroozi Farhadi, Christopher Mecoli, Audrey Kessel, Adam Schiffenbauer.

Participants in consensus conferences, Myositis Expert Panel:

Rohit Aggarwal, Helene Alexanderson, Christie Bartels, Olivier Benveniste, Hector Chinoy, Brian Feldman, Adam M. Huber, Hanna Kim, Susan Kim, Valerie Leclair, Andrew Mammen, Liza McCann, Tahseen Mozaffar, Chester Oddis, Julie Paik, Angelo Ravelli, Nicola Ruperto, Jens Schmidt, Victoria Werth, Pedro Machado.

Participants in consensus conference, Patient/Patient advocate Representatives:

Michelle Best, Ingrid de Groot, Linda Kobert, Manuel Lubinus, and Ellen Werner Shaller.