OHDSI Home | Forums | Wiki | Github

Network Study: Seeking Data Partners in Rheumatology

Good afternoon, we are happy to announce an upcoming network study lead by the Johns Hopkins Myositis Center in Baltimore, MD, USA. The goal of the study is to evaluate several phenotypes for a rare rheumatic disease known as ‘dermatomyositis’. We intend to characterize the performance of each phenotype using PheValuator on a variety of data sources along with comparison to a gold standard chart review of our own patient population.

Corresponding study lead: Christopher Mecoli, MD, MPH; cmecoli1@jhmi.edu @Christopher_Mecoli @benwmar @wkelly19

GitHub: https://github.com/ohdsi-studies/DermatomyositisPhenotypeCharacterization

Protocol: In development

Infrastructure and Data: The ability to execute PheValuator and DatabaseDiagnostics on a CDM likely to contain adult patients with dermatomyositis. The CDM can include inpatient, outpatient, EHR, claims, registry, or any combination thereof.

Participant deadline: If you are interested in participating in the study, please contact the study team by May 15th.

Primary and secondary objectives:

This study’s main objective is to evaluate and validate different dermatomyositis phenotypes across a range of OMOP databases. Secondary objectives include (a) Validating PheValuator’s probabilistic gold standard against a manual chart review gold standard in an EMR-based OMOP CDM; (b) To characterize index misclassification errors in the EMR based CDM of a tertiary referral center; (c) Raise awareness of OHDSI and the OMOP CDM in the clinical rheumatology community; and (d) Provide proof of concept of potential to perform large-scale network studies in rare diseases.

Background and rationale:

Rare diseases like dermatomyositis (DM) present a greater challenge in observational research studies, where the execution of large-scale studies is even more important in generating evidence due to the sparse cases observed in real-word data. This study presents the development and evaluation of a computable phenotype for DM to support coordinated analysis of this rare disease across manifold data sources while addressing reproducibility and generalizability challenges in the generation of real-world evidence supporting care for DM.

Additionally, within the rheumatology clinical community, publications addressing OHDSI methods have historically been limited to informatics journals not typically followed by clinical investigators. By presenting the PheValuator probabilistic gold standard alongside classification via manual chart review conducted at the Johns Hopkins Myositis Center, we aim to publish in a clinical journal and introduce OHDSI methodologies to a wider clinical audience in the field of rheumatology.

1 Like


Looks like you dropped the “Dermato” prefix in the Github name and called it “network study”, rather than “phenotype characterization”. Therefore, clicking on the link gets you a 404 error.

Thanks for pointing this out @Christian_Reich!

Revised link: ohdsi-studies/MyositisNetworkStudy: OHDSI Network Study lead by John Hopkins Myositis Group (github.com)

Ajou University Medical Center in South Korea has interests in participating in your study.

AUMC database currently has a cohort of 113 patients diagnosed with this condition. Could you please inform us if this number meets your criteria for participation, or if it is considered too small?

Thank you very much for your interest @Jinchoi! Unfortunately the cohort count is too small for this particular study. Given we are running PheValuator, we need a minimum number of patients with the condition of interest to run a valid analysis. We just updated our protocol to reflect this inclusion criteria requirement: https://github.com/OHDSI-JHU/NetworkStudy_Myositis/blob/master/Documents/protocol_20240506.pdf

Regardless, it would be great to work together in the future with other network studies we have planned, as these would not have any minimum count required.

Hello. We are Dr. You’s Research Group at Severance Hospital in Sinchon, South Korea. We are interested in your research. Please include our institution as well. Thank you

Hi @changjun0711,

We have just finished optimizing the script internally at Hopkins and have posted it on Github:

We have annotated the script as much as possible with contingencies to make it environment-agnostic (although I’m sure there will be some tweaking involved).

Our team at Hopkins (@bmarti86 @wkelly19) can help troubleshoot as issues arise.

Very interested to hear your experience in trying to run this on your CDM. Let us know if you have any questions or issues.