Work

Pos-536 Predicting the Risk of Bleeding in Hemodialysis Patients in Dopps (bleed-hd)

Year: 2021

Type: article

Abstract: Individuals with kidney disease experience a large number of bleeding events and as such, tools that aid in identifying those at high risk may aid mitigation strategies. Among 29,148 intermittent hemodialysis patients captured in the Dialysis Outcomes and Practice Patterns (DOPPS) from 15 countries (2002-2015), BLEED-HD was derived to distinguish the risk of an incident bleeding event. Validation was internal via bootstrap resampling and external using ICES linked databases in Ontario, Canada. Main outcomes and measures: A bleeding event requiring hospitalization. Among the DOPPS cohort (mean age, 65.2 years; women, 40.6%), a bleeding event occurred in 2,770 patients (9.5%, crude incidence rate 61.67 per 1000 person-years) over 0.79 (IQR 0.35-1.41) median years of follow-up. BLEED-HD included age, sex, country, DOPPS phase, previous gastrointestinal bleed, cancer, stroke, diabetes, atrial fibrillation, number of hospitalizations, anticoagulant and proton pump inhibitor use. Overall, bleeding events occurred between 3.77 to 22.31% of patients across deciles of risk. Model accuracy with BLEED-HD was significantly improved compared to existing bleeding score from the non-dialysis population in the DOPPS cohort (AUC: Bleed-HD 0.66, HEMORRHAGE 0.55, HAS-BLED 0.53 and ATRIA 0.53, improved calibration) with underprediction among those of highest risk (see Figure) In chronic hemodialysis patients, BLEED-HD improves on existing risk tools in predicting the risk of hemorrhage.

Source: Kidney International Reports

DOI: 10.1016/j.ekir.2021.03.564