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

Abstract

Purpose

Health insurance claims databases provide an opportunity to study uncommon events, such as venous thromboembolism (VTE), in large patient populations. This study evaluated case definitions for identifying VTE among patients treated for rheumatoid arthritis (RA) using International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes in claims data.

Patients and Methods

Study participants were insured adults who received treatment for and had a diagnosis of RA between 2016 and 2020. After a 6-month covariate assessment window, patients were observed for ≥1 month until health plan disenrollment, occurrence of a presumptive VTE, or end of the study (12/31/2020). Presumptive VTEs were identified using predefined algorithms based on ICD-10-CM diagnosis codes, anticoagulant use, and care setting. Medical charts were abstracted to confirm the VTE diagnosis. Performance of primary and secondary (less stringent) algorithms was assessed by calculating the positive predictive value (PPV; primary and secondary objectives). Additionally, a linked electronic health record (EHR) claims database and abstracted provider notes were used as a novel alternative source to validate claims-based outcome definitions (exploratory objective).

Results

A total of 155 charts identified with the primary VTE algorithm were abstracted. The majority of patients were female (73.5%), with mean (standard deviation) age 66.4 (10.7) years and Medicare insurance (80.6%). Obesity (46.8%), ever smoking (55.8%), and prior evidence of VTE (28.4%) were commonly reported in medical charts. The PPV for the primary VTE algorithm was 75.5% (117/155; 95% confidence interval [CI], 68.7%, 82.3%). A less stringent secondary algorithm had a PPV of 52.6% (40/76; 95% CI, 41.4%, 63.9%). Using an alternative EHR-linked claims database, the primary VTE algorithm PPV was lower, potentially due to the unavailability of relevant records for validation.

Conclusion

Administrative claims data can be used to identify VTE among patients with RA in observational studies.

Keywords:

• algorithm validation
• rheumatoid arthritis
• venous thromboembolism
• positive predictive value
• administrative claims data
• ICD-10-CM

Abbreviations

bDMARD, Biologic disease-modifying antirheumatic drug; CAW, Covariate assessment window; CI, confidence interval; COPD, Chronic obstructive pulmonary disease; DVT, Deep vein thrombosis; EHR, Electronic health record; ER, Emergency room; FP, False positive; ICD-9, International Classification of Diseases, Ninth Revision; ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification; ICD-10, International Classification of Diseases, Tenth Revision; ICD-10-CM, International Classification of Diseases, Tenth Revision, Clinical Modification; JAKi, Janus kinase inhibitor; MC, Market Clarity; ORD, Optum Research Database; PE, Pulmonary embolism; PPV, Positive predictive value; RA, Rheumatoid arthritis; SD, Standard deviation; TNFi, Tumor necrosis factor inhibitor; TP, True positive; US, United States; VT, Venous thrombosis; VTE, Venous thromboembolism.

Data Sharing Statement

The data contained in Optum’s database contains proprietary elements owned by Optum and therefore cannot be broadly disclosed or made publicly available at this time. The disclosure of this data to third-party clients assumes certain data security and privacy protocols are in place and that the third-party client has executed our standard license agreement, which includes restrictive covenants governing the use of the data.

Ethics Approval and Informed Consent

All data (claims based and chart/notes based) obtained in this study were accessed in compliance with the Health Insurance Portability and Accountability Act of 1996. This study was reviewed and approved by an institutional review board (WCG IRB).

Acknowledgments

The authors thank Jerry Seare, MD (Medical Director, Optum HEOR), Marcelo Gomes, MD (Cleveland Clinic), and Kim Brown, RN (Cleveland Clinic) for their clinical review.

Medical writing support was provided by professional medical writer Catherine Champagne, PhD, CMPP, employee of Kay Square Scientific (Newtown Square, PA, USA), which received funding from Optum.

Disclosure

Sangmi Kim, Shiyao Gao, and Claudia A Salinas are employees and shareholders of Eli Lilly and Company. Carolyn Martin, John White, Maureen Carlyle, and Bonnie Bui were employees of Optum Life Sciences at the time of the study. Eli Lilly and Company contracted with Optum Life Sciences for the purposes of this research. The authors report no other conflicts of interest in this work.

Additional information

Funding

This study was funded by Eli Lilly and Company. Optum received funding from Eli Lilly to conduct this study. The funding agreement did not impact the authors’ independence in designing the study, collecting the data, interpreting the data, writing the manuscript, and submitting the manuscript for publication.