Data source
The healthcare database used for this study was provided by Medical Data Vision Co., Ltd. The database contains Japanese diagnosis-procedure combination data and health insurance claims data from hospitals located throughout Japan, anonymously managed to protect patient’s personal information; some hospitals also provide laboratory data. This study used clinical data from 71 hospitals collected from April 2011 to March 2014, including laboratory data from 14 hospitals.
Patients
Data were extracted for patients aged 18 years or older with RA or OA diagnosed for the first time during the study period, from April 2011 to March 2014, and with data available for a follow-up period of at least 90 days (Fig. 1). The following codes from International Classification of Diseases 10th revision (ICD-10) were used to identify cases for extraction: M05, M06.0, M06.2, M06.3, M06.8, and M06.9 for RA, and M179 for OA. Each patient was prescribed or had a treatment history of anti-RA drugs (DMARDs, oral corticosteroids, biologics, or oral NSAIDs) or OA treatments (oral or topical NSAIDs, hyaluronic acid injections, or corticosteroids injections) and was followed up until the end of March 2014 or until leaving the database. RA and OA patients in the sub-cohort were defined as patients with laboratory data for risk factor analyses.
Outcomes
Outcomes are defined as events that occurred after first diagnosis of RA or OA and were identified by a combination of diagnoses (ICD-10) leading to hospitalization and medical tests. Diagnoses of CVD in this study were defined as ischemic heart disease [angina pectoris (I20) or myocardial infarction (I21, I22)], heart failure (I50), and stroke [cerebral infarction (I63), intracerebral hemorrhage (I61), or subarachnoid hemorrhage (I60)]. Medical tests used to identify ischemic heart disease and heart failure were electrocardiogram and echocardiograph. Medical tests used to identify stroke were computed tomography and magnetic resonance imaging.
Statistical analysis
Incidence rates in each cohort were calculated as the number of cases per total period at risk for CVD. Crude and age–sex adjusted incidence rate ratios (IRRs) with 95 % confidence intervals (CIs) versus OA were also calculated on the assumption of a Poisson distribution.
Univariate and multivariate Cox-regression analyses were performed, including hazard ratios (HRs), 95 % CIs, and Wald test P values to evaluate risk factors for CVD in RA patients. The candidate covariates were age, sex, CRP at baseline, total cholesterol at baseline, and diabetes complications. CRP was selected as an index of systemic inflammation associated with RA. Total cholesterol and diabetes complications were selected as traditional risk factors for CVD. CRP and total cholesterol at baseline were measured on the date of or before first diagnosis of RA or OA. No variable selection techniques were used in multivariate Cox-regression analyses. Statistical analyses were performed using SAS version 9.3.
Ethics approval
Our study was approved by the institutional review board of the Public Health Research Foundation (http://www.phrf.jp/) on 5 November 2014 and was conducted according to the local ethical guidelines for epidemiological research. Informed consent was waived by the institutional review board because the database used in this study is anonymously managed and includes no personal information.