VALIDATION OF RHEUMATOID ARTHRITIS DIAGNOSIS IN THE KOREAN NATIONAL HEALTH INSURANCE CLAIMS DATABASE

Abstract

Background Healthcare claims database is one of powerful sources of epidemiologic study for RA patients. It is crucial, however, to assess the validity of using diagnosis codes contained in this healthcare claims database to identify RA cases prior to utilizing them for epidemiological studies. Prior studies have examined the validity of RA diagnostic code and insisted that the addition of disease modifying anti-rheumatic drugs (DMARDs) with RA diagnostic code increase the PPV and accuracy for identifying RA patients in the claims database. However, this algorithm tends to identify only patients in use of DMARDs.

Objectives To validate various algorithms using information of medication and health utilization for identifying RA patients in the healthcare claims database in Korea.

Methods A total of 134,930 patients≥19 years old with seropositive RA code (M05) covering the period of January 2009 to June 2011 were included in this validation study. We regarded patients registered in the “individual copayment beneficiaries program for rare and intractable diseases (ICBP)” in Korea as gold standard RA. This registration requires an official report by doctors documenting that the patients fulfilled the 1987 ACR criteria for RA. We constructed algorithms using RA diagnostic code and incorporated claims of (1) RA treatment: DMARDs, NSAIDs, or corticosteroid, (2) Laboratory tests: ESR, CRP, or rheumatoid factor, and (3) Health utilization: outpatient visit or hospitalization. We calculated the sensitivity, positive predictive value (PPV), and accuracy of each algorithm and PPV and accuracy were mainly compared between them.

Results PPV for identification of RA with RA diagnostic code and any DMARDs or biologics use was 82% and it was higher than those of combination with NSAIDs (52%), corticosteroid (63%), outpatients visits≥3 (74%) or hospitalization (63%). PPV of algorithm for identifying RA with diagnostic code and any DMARDs or biologics or hospitalization was not increased (80%) but the sensitivity (94.9%) was higher than that of RA diagnostic code and any DMARDs or biologics (94.6%).

Conclusions We suggest that RA patients can be identified with the algorithm of RA diagnostic code of any DMARDs or biologics with high level of PPV in claims database. Addition of hospitalization increased the sensitivity of algorithm in identifying RA patients.