Methodologies for Validation of Diagnoses in Real-World Data: BONES-A Case Study

The accuracy of any claim-based study is dependent on the quality of real-world coding of the condition of interest. This retrospective, administrative claims analysis presents a method for using a real-world data source to evaluate the accuracy of coding for nonunion of a fifth metatarsal fracture. Patients 21-80 years old with a diagnosis of a fifth metatarsal fracture between January 1, 2016, and October 31, 2016, and a nonunion of the fifth metatarsal fracture within the next 9 months were identified in the MarketScan Databases. Patient health care claims in the 12 months after the nonunion diagnosis were examined for health care encounters and pharmaceutical treatments considered indicative of treatment for nonunion, such as claims for bone growth stimulation or a second claim with a diagnosis of nonunion. Of the 230 patients who had at least one health care encounter attributable to a nonunion of the fifth metatarsal, 95.2% had at least one subsequent health care encounter confirming nonunion diagnosis. The mean number of supporting health care claims was 5.8, and the mean time between nonunion and first confirmatory claim was 33 days. This analysis demonstrated a method for evaluating the quality of coding for a specific condition when a traditional medical chart comparison is not feasible.