Longitudinal kidney function trajectories predict major bleeding, hospitalization and death in patients with atrial fibrillation and chronic kidney disease

Background: Chronic kidney disease (CKD), commonly described by estimated glomerular filtration rate (eGFR), is a frequent comorbidity in patients with atrial fibrillation (AF) and associated with thromboembolic and bleeding complications. Instead of single eGFR measurements, kidney function decline over time may better predict clinical outcomes but this has not been studied so far. Methods: Patients with AF and stage 3/4 CKD were prospectively followed within a primary care electronic database from the United Kingdom (IMS-THIN). The associations between the longitudinal eGFR trajectory of these patients and stroke/systemic embolism, major bleeding, first hospitalization-for-any-cause, and death-from-any-cause were estimated with joint models of longitudinal and time-to-event data. Results: 18,240 patients were included (median age 80.4 years, median CHA ₂ DS ₂ -VASc score 4). In 133,676 eGFR measurements (mean: 6 per patient) median “baseline” eGFR was 49 ml/min/1.73m ² [41–55] and mean eGFR decline was 0.54 ml/min/1.73m ² /year (95%CI: 0.47–0.62). During follow-up (median 3.2 years; 50,841 patient-years at risk), 5-year cumulative incidence estimates were 9%, 3%, 32% and 76% for stroke/systemic embolism, major bleeding, hospitalization and death, respectively. In joint modeling, an accelerated decline in kidney function strongly predicted for a higher risk of major bleeding (hazard ratio [HR] 1.09 per ml/min/1.73m ² /year increase in eGFR decline), hospitalization (HR 1.06), and death-from-any-cause (HR 1.11; all p < 0.05), but not for stroke/systemic embolism (HR 0.97; p = 0.239). Conclusions: Declining kidney function is a critical determinant of unfavourable outcomes in patients with AF and CKD. Longitudinal kidney function trajectories may enable a much more individualized prediction of adverse outcomes in this vulnerable patient population.