(135) Validation of an Algorithm to Identify Venous Thromboembolism (VTE) in Health Insurance Claims Data Among Patients with Rheumatoid Arthritis

Publication Number: 466

Background: Patients with rheumatoid arthritis (RA) are at increased risk of systemic complications which may be related to the disease process, treatment with disease-modifying antirheumatic drugs (DMARDs), the prevalence of risk factors in the population, or a combination. Health insurance claims databases provide an opportunity to study uncommon complications, such as VTE, in a large RA population. To date, however, no prior studies have validated an algorithm based on ICD-10-CM codes.

Objectives: To evaluate the performance of health insurance claims data for identifying VTE among patients treated for RA.

Methods: The study included adults aged ≥18 years with commercial or Medicare Advantage insurance with Part D (MAPD) who received treatment for RA between 01 May 2016 – 30 November 2020 and had an ICD-10-CM diagnosis code for RA. After a 6-month covariate assessment window and evidence of a biologic or targeted synthetic DMARD, patients were observed for ≥1 month. Observation time ended on disenrollment from the health plan, occurrence of a presumptive VTE, or the end of the study period (31 December 2020). Patients with a presumptive VTE were identified based on an algorithm using ICD-10-CM diagnosis codes, drug dispensing records, and type of care setting, adapted from a prior publication. Clinical data from the patient’s medical chart was abstracted to confirm the VTE diagnosis using pre-specified clinical confirmation rules. The validity of the algorithm to identify VTE was assessed by calculating the positive predictive value (PPV).

Results: Patient charts were procured from treating providers for 155 of 254 patients (61%) identified with the pre-specified VTE algorithm. The majority of patients were female (74%), with mean age 66 years (standard deviation [SD] 11), and MAPD insurance (81%). Chronic obstructive pulmonary disease (27%), ever smoking (56%), hypertension (72%), obesity (47%), and prior evidence of VTE (28%) were common. Follow-up ranged from 51 - 1,704 days (mean 912; SD 525 days). The PPV for the VTE algorithm was 75% (117/155; 95% CI 69-82%) using information in available medical charts as the reference standard.

Conclusions: An algorithm based on administrative claims data can effectively identify VTE among patients with RA for use in comparative observational studies.