MAKING THE MOST OF EXTERNAL COMPARATORS: A STUDY OF FRACTURE HEALING IN PATIENTS AT RISK OF NON-UNION

• Published in: Value in health 2017-05, Vol.20 (5), p.A342
• Main Authors: Mack, C, Pavesio, A, Kelly, K, Wester, T, Jones, J, Maislin, G, Irwin, D, Brinkley, E, Zura, RD
• Format: Article
• Language: English
• Subjects: Bone healing; Chart reviews; Diagnosis; Fractures; Healing; Health care; Medical diagnosis; Methodological problems; Nonunion; Propensity; Questionnaires; Research methodology; Selection bias; Therapy; Tibia; Trade; Ultrasound
• ISSN: 1098-3015; 1524-4733
• DOI: 10.1016/j.jval.2017.05.005

OBJECTIVES: Fracture non-unions result in significant physical, emotional, and economic burden on patients. Low-intensity pulsed ultrasound (LIPUS) is an FDA-approved device to treat established fracture non-unions and facilitate the healing of acute radius and tibia fractures. However, commercial insurers routinely cover use in patients with fractures beyond approved indications in an attempt to avoid surgical non-union repair costs later. To align LIPUS's real-world use with its labeling, a novel real-world study to establish effectiveness of LIPUS for mitigating fracture nonunion is proposed. METHODS: The fracture non-union rate between a prospectively enrolled LIPUS-treated cohort will be compared to untreated historical controls selected from MarketScan™ Claims. Fracture status will be obtained from presence of non-union diagnosis codes 12 months post-fracture date. The potential for selection bias will be addressed using propensity score methodology. Given underlying differences in data collection, key methodological issues include creation of equivalent operational definitions for patient selection, study time points, patient characteristics, and outcomes to maximize the probability of equivalent ascertainment across different modes of data capture. RESULTS: Variables were denned for controls using diagnosis and procedure codes; patient questionnaires and chart abstraction will be used for the LIPUS cohort. For example, to minimize recall bias, the LIPUS cohort will use a 3-month pre-fracture recall period, while controls will use 6 months pre-fracture claims to account for 90-day supplies. Non-unions in the LIPUS cohort will be assessed by interview and chart review, while algorithms will be developed for claims analysis to capture non-unions reported within a clinically meaningful period. Methodological solutions will be presented. CONCLUSIONS: This novel approach demonstrates methodology for measuring device effectiveness in the real-world comparing cohorts enrolled both prospectively and within "Big Data".