Background/Purpose: To improve efficiency and transportability of observational research, common data models have been developed and maintained to translate data from source vocabularies into standardized vocabularies using vocabulary mapping.  The accuracy of mapping International Classification Codes to Systematized Nomenclature of Medicine — Clinical Terms (SNOMED-CT) of rheumatoid arthritis (RA) has been examined at the cohort level, however, its accuracy, in combination with standardized medication coding system RxNorm, has never been examined at the patient level.  The objective of this study is to examine the positive predictive value (PPV) of identifying patients with rheumatoid arthritis treated with biologics using Observational Medical Outcomes Partnership (OMOP) common data model.

Methods: NewYork-Presbyterian/Columbia University Irving Medical Center (NYP/CUIMC) EHR has been translated and maintained in OMOP common data model.  We searched NYP/CUIMC clinical data warehouse for prevalent cases of patients with rheumatoid arthritis treated with biologics using standardized codes for RA and biologics from 01/01/2013 to 12/31/2018, and excluded patients with other potential indications for biologics.  We randomly selected 100 patients from the search result for chart review.  We used documentation of a diagnosis of rheumatoid arthritis and a biologics at the index date (+/-4 days) as the gold standard.  For patients who had rheumatology visit notes, we examined whether they fulfilled 2010 Rheumatoid Arthritis Classification Criteria. 

Results: One hundred and two medical records were identified, and 11 were not accessible.  Among the 91 records that were retrieved and examined, 77 patients (84.6%) had a documented RA diagnosis and prescription of biologics, by either rheumatology providers or non-rheumatology providers.  Four patients did not have a documented diagnosis of RA, and ten did not have a documented use of biologics.  Of the 47 patients who had rheumatology visits, 41 (87.2%) fulfilled the 2010 RA classification criteria.

Conclusion: Using OMOP common data model and standardized vocabularies, we can identify RA patients treated with biologics with a PPV of 84.6% in a single center EHR system.

« Back to 2019 ACR/ARP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/identification-of-patients-with-rheumatoid-arthritis-treated-with-biologics-using-standardized-vocabularies/