Data-driven identification of co-morbidities associated with rheumatoid arthritis in a large US health plan claims database

BACKGROUND: In drug development, it is important to have an understanding of the full spectrum of co-morbidities to be expected in the group of patients with the disease of interest. It is usually a challenge to identify the less common events associated with the target disease, even if these events are severe. The purpose of this study is to identify co-morbidities associated with rheumatoid arthritis (RA) as compared with a control group, using a large health care database. METHODS: Marketscan US claims database was used for this retrospective cohort study. Selected were records of persons aged at least 16 Y with at least two claims for RA, and with active insurance status on June 30,2007. The control group had at least two claims for eczema/dermatitis. Controls were matched by age, gender and insurance status (Medicare or not). All co-morbidities with an ICD9 diagnostic code were identified in the RA and control groups, during a one-year window. Relative risks (RRs) were calculated. Diagnoses were rank-ordered by magnitude of RR. Codes covering RA and arthropathy were excluded. In order to get stable estimates, rank-ordering was performed for diagnoses occurring in at least 20 persons in the control group. RESULTS: Records were selected of 62,681 persons with RA (mean age was 59.0 Y, with 73.8% female, Medicare-covered 35%). A total of 6897 different ICD9 diagnostic codes were recorded, with 2220 codes in at least 20 persons of the control group [listed with Relative Risk]. Apart from joint/bone related conditions, strong associations with RA (RR > 3) were found for Adverse effect medicinal and biological substance not elsewhere classified, Unspecified adverse effect drug properly administered, Idiopathic fibrosing alveolitis, Osteomyelitis, Immune deficiency, Elevated sedimentation rate, Tuberculin test reaction abnormal or positive, Anemia and Cushing syndrome. CONCLUSIONS: Data on a large number (> 60,000) of patients with a diagnosis of RA were used to analyze and to list a large number (> 2,000) of co-morbidities. Rank-ordering of RRs of diagnostic codes is a tool to identify quickly many conditions associated with RA.