Does the risk of readmission and revision events after cervical spine fusion differ in hospital billing versus claims data? Example of evaluating quality of hospital billing databases for real-world evidence (RWE) of medical devices

Hospital billing data can provide safety and effectiveness insights on medical devices. However, hospital data follow patients longitudinally in the same hospital only. Comparison of outcome risks (e.g., readmission and revision) in hospital versus claims data is important to understand any potential loss of data capture when evaluating outcomes for devices using hospital data. The objective of this study was to evaluate accuracy of hospital databases for long-term patient follow-up by type of endpoint using cervical fusion surgery as an example. The study included 150,459 and 62,752 patients in the claims and hospital databases, respectively. Cohort balancing was performed using propensity score methods. Risks of readmission at 90 and revision at 365-days post-surgery were 11.9% and 2.8% in the claims database and 7.5% and 2.5% in the hospital database. The difference in 90-day readmission risk (95% CI) (hospital – claims) was − 4.3% (-4.6%, -4.0%), while for revision, the difference in 365-day revision risk was < 1% (-0.3% ( -0.5%, < 0.1%)) between the databases. As time increased to ≥ 2 years, the difference in risk increased, with the cumulative 1,825-day difference in revision risk at -2.0% (-2.9%, -1.0%).