Ab1246 identification of giant cell arteritis in real-world data using an administrative claims-based algorithm

Background:Giant cell arteritis (GCA), the most common systemic vasculitis in adults, is often associated with significant morbidity and mortality. A claims-based algorithm that accurately identifies GCA patients in large real-world data can offer new opportunities for future epidemiological studies.Objectives:We aimed to develop and validate a claims-based algorithm for GCA.Methods:We developed and tested 5 claims-based GCA algorithms using U.S. Medicare claims (Parts A/B/D) linked to electronic medical record data from a large academic medical center, 2006-2014:Algorithm 1) ≥1 International Classification of Diseases, Ninth Revision (ICD-9) code for GCA (446.5x) by any physician, high dose steroid dispensing (i.e., prednisone equivalent ≥40 mg/day for ≥14 days), and ≥1 Current Procedural Terminology (CPT) code for ESR/CRP;2) ≥1 ICD-9 for GCA by a rheumatologist, high dose steroid dispensing, and ESR/CRP;3) ≥2 ICD-9 for GCA by a rheumatologist separated by 7-30 days and high dose steroid dispensing;4) ≥1 ICD-9 for GCA by a rheumatologist, high dose steroid dispensing, ESR/CRP, and CPT code for temporal artery biopsy; and5) ≥1 ICD-9 for GCA or its subtypes (447.6, 437.4, 417.8) by a rheumatologist, high dose steroid dispensing, ESR/CRP, and CPT code for chest imaging. For all algorithms, the index date was defined as the date of first steroid dispensing. Two physicians reviewed medical records for the gold standard definition of GCA: documentation of GCA by the treating physician or a temporal biopsy result consistent with GCA. Probable GCA cases based on treating physician’s records were included as well. Positive predictive value (PPV) and 95% confidence intervals (CI) of the algorithms were calculated.Results:We identified 1,930 patients with Medicare claims data linked to electronic medical records. Among these, 799 unique records with physician notes documenting GCA or biopsy results were included in the PPV calculations. Algorithm 1, which identified the greatest number of patients (n=896), yielded the lowest PPV of 60.7%. Algorithms 4 and 5, which required disease-specific workups (temporal artery biopsy or chest imaging), mildly improved the PPV to 76.2% and 68.2%, respectively. The PPV was highest in algorithm 3 (84.8%), which required 2 or more diagnoses of GCA by a rheumatologist in addition to high dose steroid dispensing.Table.PPV of Claims-based AlgorithmsNo. of patientsAlgorithm #Records IdentifiedRecords ReviewedAdequate RecordsPPV* (95% CI)1896446†206 (46.2%)60.7% (53.7-67.4)2471471270 (57.3%)78.6% (73.2-83.3)3220220125 (57.4%)84.8% (77.3-90.6)4296296172 (58.1%)76.2% (69.1-82.3)5474726 (55.3%)68.2% (48.2-85.7)* True positive cases included both definitive and probable GCA patients†446 records were randomly selected for chart review.Conclusion:A claims-based algorithm including two or more diagnosis codes for GCA by a rheumatologist separated by 7-30 days and high dose steroid dispensing (prednisone equivalent ≥40 mg/day for ≥14 days) can be a useful tool for identifying patients with GCA, allowing for future large real-world data studies.References:[1]Crow RW et al. Giant cell arteritis and mortality. J Gerontol A Biol Sci Med Sci. 2009 Mar;64(3):365-9.Acknowledgments:This study was supported by an investigator-initiated research grant from Genentech/Roche. The sponsor was given the opportunity to make non-binding comments on a draft of the abstract, but the authors retained the right of publication and to determine the final wording.Disclosure of Interests:Hemin Lee: None declared, Sarah Chen Employee of: After finishing the work for this abstract, she has moved to work for Gilead., Sara Tedeschi: None declared, Paul Monach: None declared, Jun Liu: None declared, Attila Pethoe-Schramm Shareholder of: Current employee of F. Hoffmann-La Roche and own company stocks/stock options, Employee of: Current employee of F. Hoffmann-La Roche, Vincent Yau Shareholder of: Current employee of F. Hoffmann-La Roche/Genetech and own company stocks/stock options, Employee of: Current employee of F. Hoffmann-La Roche/Genetech, Seoyoung Kim Grant/research support from: Seoyoung C Kim has received research grants from AbbVie, Roche, Bristol-Myers Squibb and Pfizer.