(156) How well does hospital billing data estimate readmissions and revision risks after spinal fusion surgery compared to hospital billing linked to claims data?

Background: Hospital databases are used to study medical devices but cannot follow patients’ healthcare if they don’t return to the same hospital. Claims databases can follow insured patients over time, but can’t detect types of medical devices used or study surgery characteristics. Linking hospital and claims databases may strengthen longitudinal follow-up of surgical patients.

Objectives: To assess the incremental detection of revision surgery and all-cause readmissions from hospital billing data linked to claims data versus hospital billing data alone among patients that had cervical, lumbar or thoracic spinal fusion surgeries.

Methods: A data quality assessment study was performed. Patients undergoing cervical, lumbar or thoracic fusion surgery between January 1, 2018 to September 30, 2021 were identified in a hospital billing database (Premier) and linked to an administrative claims database (Komodo) using tokenization. Komodo consists of an aggregate of claims from 150 healthcare providers. Medicare Advantage insurance and ≥90 days continuous insurance enrollment were required. Cumulative incidence was estimated for 90-day all-cause readmission and 90-, 180-, 365-, 366 to 730- and 730-day revision surgery outcomes. Analysis was conducted in: (1) hospital database only and (2) linked database.

Results: Overall, 978 patients met inclusion criteria (cervical, 30.4%; lumbar, 60.0%; thoracic, 9.7%). The risk of revision at 90 days was 1.9% in hospital and 2.1% in linked data and at 730 days was 3.3% versus 4.4% in hospital and linked data, respectively. The risk of all-cause readmission (90-day) was 20.2% overall in the hospital data and 29.2% in the linked data.

Conclusions: Preliminary results from this study suggest that the 90-day risks of revision and all-cause readmission were higher in linked data as compared to hospital data alone at 90-days post spinal surgery; however, the difference between risks in the hospital alone versus linked data was greater for readmissions than revisions, which is consistent with revision surgeries for an initial surgery more likely to return to the original hospital, whereas general admissions for possibly other medical reasons are less likely to return to the same hospital. This study is potentially limited by the completeness of the claims data.