AB1200 IMPROVEMENT IN THE QUANTITY AND QUALITY OF OBSERVATIONAL DATA COLLECTED FOR US VETERANS ENROLLED IN THE VETERANS AFFAIRS RHEUMATOID ARTHRITIS REGISTRY USING AN ELECTRONIC AUDIT, FEEDBACK, AND DATA CORRECTION SYSTEM

Background The Veterans Affairs (VA) Rheumatoid Arthritis (RA) (VARA) registry is an observational cohort study of US Veterans with RA at 11 VA Medical Centers. VARA investigators capture clinical and laboratory disease activity measures (DAMs) during clinic visits via standardized templates in the electronic health record (EHR). Six clinical (tender/swollen joints, patient/provider global, MD-HAQ, pain) and 2 laboratory (ESR, CRP) DAMs are extracted post-visit using natural language processing (NLP). Objectives This analysis determined the impact of an audit, feedback and data correction system on the quantity and quality of DAMs collected in the VARA registry. Methods After September 2017, VARA site investigators were provided monthly feedback reports of incomplete/missing DAMs for the prior month to allow sites to correct data entry errors. Updated and/or corrected data were entered into the EHR via note addendums and then automatically re-extracted by NLP to complete the capture of DAM data. DAMs from October 1, 2016 to September 30, 2017 (pre-feedback implementation – Pre-IMP) was compared to October 1, 2017 to September 30, 2018 (post-feedback implementation – Post-IMP). Results During the pre-IMP period, there were 2,411 notes with DAMs collected on 1,116 unique patients compared to 2,873 notes on 1,208 unique patients in the post-IMP period - an increase of 92 (8.2%) unique patients and 460 (19.1%) notes. Enrollment in the VARA registry only increased by 121 (6.5%) during post-IMP period. During post-IMP period, there were 541 notes identified with deficiencies in clinical DAMs and monthly audit and feedback reports were provided to VARA sites to allow corrections. Individual site review resulted in 376 additional DAMs in 225 notes, with complete resolution of all error in 137 (25.3%) notes. The quantity of DAMs collected increased from 15,709 to 21,064, a 34.1% increase with the average number of DAMs collected per note rising from 4.9 to 5.6. The quality of data improved as demonstrated by the proportion of notes with all 6 clinical DAMs increasing from 52.5% to 81.1% and other improvements in quality/completeness as noted in table. Conclusion An audit, feedback, and efficient data collection system improved both the quantity and quality of DAMs collected. The improvement in the collection of DAMs in RA patients will further enhance epidemiologic and outcomes studies of RA and provide higher quality longitudinal data to enhance the care of RA patients. References [1] Fed Pract 2015; 32(5):24-29 Acknowledgement Work Support in Part by VA HSR&D and VA Specialty Care Centers of Innovation Disclosure of Interests Grant Cannon Grant/research support from: Amgen Inc., Jorge Rojas: None declared, Neill Bell: None declared, Namrata Singh: None declared, Ted Mikuls: None declared, Liron Caplan: None declared, Gail Kerr: None declared, Joshua Baker: None declared, Angelo Gaffo: None declared, Jennifer Barton: None declared, Deana Lazaro: None declared, J Steuart Richards: None declared, Brian Sauer Grant/research support from: Amgen Inc.