MedTech Epidemiology & Real-World Data Sciences, Johnson & Johnson, United States
Background: Hospital database tracks patients after surgery if they returned to the same hospitals but not if they go to different hospitals. Hospital system databases therefore may contain multiple treatment episodes for the same patients counted as different patients.
Objectives: To evaluate how many unique patients ID are repeated patient IDs and completeness of atrial fibrillation (AF) recurrence post-ablation within and across hospitals in the hospital database through tokenization.
Methods: Patients undergoing catheter ablation for AF from 2018 to 2020 were extracted from the Premier Healthcare Database. Tokenization was conducted to create a unique patient ID from patients’ characteristics and identify the same patient but with different identifiers outside the index hospital. The effect of de-duplication of patients’ IDs on recurrence of AF within 1-year post-ablation were assessed. AF recurrence was defined by components of repeat ablation, AF-related hospitalization, or direct cardioversion (DCCV, a treatment for AF). Sensitivity was calculated by dividing the number of cases identified within the index hospital by the number of cases across hospitals. Subgroup analyses were performed by patients' age, sex, insurance type, and hospital teaching status and bed size.
Results: Among 39741 patients in the study cohort, there were 25,939 unique patients with encounters within 1-year post-ablation among 31196 patients IDs. 7,703 (19.4%) patients had encounters outside of index hospital and 3457 (8.7%) patients had only encounter outside of index hospital during 1-year follow-up. There were 5.9% vs 6.1% of patients with repeat ablation, 5.7% vs 6.4% with AF hospitalization, and 10.3% vs 11.5% with DCCV if patient were followed up within index hospital and across hospitals respectively. Overall sensitivities were 0.96 for repeat ablation, 0.89 for AF hospitalization, and 0.90 for DCCV. The sensitivity of AF hospitalization and DCCV increased with age while sensitivity of repeat ablation was consistent across different age groups. There was no difference of sensitivity between males and females or between different insurance types. The sensitivity of repeat ablation and AF hospitalization increased with hospital bed size. Teaching hospitals (0.98) had slightly higher sensitivity than non-teaching hospitals (0.94) in repeat ablation.
Conclusions: Using hospital billing data linked with token allows the identification of different episodes from same patient but being treated as separate patients in the hospital billing database. We identified that 19.4% patients of AF visited the different hospitals post ablation within 1-year follow-up. Sensitivity of outcome depends on the outcome type, patients’ age and hospital types.
