Highly Accessed Open Badges Letter

Further evaluation of a claims-based algorithm to determine the effectiveness of biologics for rheumatoid arthritis using commercial claims data

Jeffrey R Curtis1*, Benjamin Chastek2, Laura Becker2, David J Harrison3, David Collier3, Huifeng Yun4 and George J Joseph3

Author affiliations

1 Division of Clinical Immunology and Rheumatology, University of Alabama at Birmingham, 510 20th Street South, FOT 802D, Birmingham, AL 35294, USA

2 OptumInsight, 12125 Technology Drive, Eden Prairie, MN 55344, USA

3 Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320, USA

4 Department of Epidemiology, School of Public Health, University of Alabama at Birmingham, 1700 University Boulevard, Birmingham, AL 35294-0013, USA

For all author emails, please log on.

Citation and License

Arthritis Research & Therapy 2013, 15:404  doi:10.1186/ar4161

See related research by Curtis et al., http://arthritis-research.com/content/13/5/R155, and related editorial by Kim and Solomon, http://arthritis-research.com/content/13/5/129

Published: 8 March 2013

First paragraph (this article has no abstract)

As more biologics are approved, there is increasing interest in comparative effectiveness research (CER). Health insurance claims databases contain information about outpatient visits, hospital discharges, procedures, and outpatient pharmacy dispensing but seldom contain clinical outcomes. In a previous issue of Arthritis Research & Therapy, we presented an algorithm that assessed the clinical effectiveness of rheumatoid arthritis (RA) biologics which used Veterans Affairs (VA) claims data and which was validated against the DAS28-ESR (Disease Activity Score 28 using erythrocyte sedimentation rate) [1]. The algorithm had a sensitivity of 72% (95% confidence interval (CI) = 67% to 77%) and a specificity of 91% (95% CI = 89% to 93%). In an editorial in the same issue, Kim and Solomon [2] commented the following: 'a claims-based effectiveness algorithm with acceptable performance characteristics across different data settings will be a powerful and desired tool for CER of RA. Such an algorithm will enable large-scale, population-based studies comparing the effectiveness of different DMARD [disease-modifying antirheumatic drug] regimens. Such studies will facilitate head-to-head comparisons, supplementing typical randomized controlled trials and prospective registries that usually include disease activity. Whether the algorithm will have a similar performance in other claims databases therefore needs to be further examined'. We performed an independent analysis to evaluate the algorithm's positive predictive value (PPV) in a commercial claims data source compared with a clinical gold standard.