Session Type: Poster Session B
Session Time: 9:00AM-11:00AM
Background/Purpose: Identifying and accurately classifying comorbid conditions in large, real-world data sources is crucial for cohort establishment and confounder adjustment. However, the ability to conduct proper confounding control may be challenging when data for diagnosis codes and lab tests are incomplete, such as in a single-specialty EHR system where comorbidities are incompletely recorded by specialists. Thus, we updated a method to allow researchers to identify comorbid conditions using only medication information in circumstances where diagnosis codes are under-captured.
Using the RxNorm application program interface (API) and its web-based clients, RxMix and RxClass, we mapped Drug Concept Unique Identifiers (RxCUIs) to the Rx-Risk. In an established RA cohort in the ACR RISE registry, we compared Rxrisk with other comorbidity indices. These included the Charlson comorbidity index, Rheumatic Disease Comorbidity Index (RDCI) and Elixhauser.
Methods: For each of the 46 Rx-Risk categories, we identified corresponding RxNorm ingredients via RxClass and expanded them to all RxCUI term types (TTYs) using RxMix (RxNorm May 2020 release) based on FDA indications. Glucocorticoid products were confined to the comorbidity categories: Allergies, Chronic airway disease, and steroid-responsive disease by administration route. After finalizing the Rxrisk categories, we conducted descriptive analyses and compared the distribution of Rxrisk with more traditional diagnosis-based comorbidity scores among RA patients who were 18 years of age with ³2 consecutive visits with ICD-10 codes for RA using ACR RISE data, a rheumatologist-based RA registry. Medications and comorbidity diagnoses were assessed using all available data prior to the 2nd RA diagnosis code.
Results: We identified 913 ingredient RxCUIs representing the 46 Rx-Risk comorbidity categories. 65,895 RxCUIs for all TTYs were returned from our initial query. After excluding Dosage Form related RxCUIs, a total of 56,217 RxCUIs were included in our analysis. The most common TTYs were Semantic Clinical Drug (19.8%), Demantic Brand Drug (13.0%), and Semantic Branded Drug Component (11.9%), while pain (17.7%), allergy (11.7%), and malignancy (6.2%) products were most frequent among the 46 comorbidity categories. The median score (25th/75th percentile) for Rxrisk was much greater: 8 (5,12) than for charlson index: 0 (0,0); Elixhauser: 1 (1,2); RDCI: 0 (0,1). For patients with Charlson score of 0 (85% of total), both the RDCI and Elixhauser were close to 1, but the Rxrisk score ranges from 0 to 20 (Figure).
Conclusion: The misclassification and under ascertainment of comorbidities in a single specialty EHR can largely be overcome by using a medication-focused comorbidity index that we have recently updated.
To cite this abstract in AMA style:Vanderbleek J, Owensby J, Mccannaly A, Chen L, England B, Curtis J, Yun H. Mapping Multimorbidity Using Drug Concept Unique Identifiers (RxCUIs) via the Rx-Risk Comorbidity Index [abstract]. Arthritis Rheumatol. 2020; 72 (suppl 10). https://acrabstracts.org/abstract/mapping-multimorbidity-using-drug-concept-unique-identifiers-rxcuis-via-the-rx-risk-comorbidity-index/. Accessed September 27, 2021.
« Back to ACR Convergence 2020
ACR Meeting Abstracts - https://acrabstracts.org/abstract/mapping-multimorbidity-using-drug-concept-unique-identifiers-rxcuis-via-the-rx-risk-comorbidity-index/