Practice Patterns and Outcomes of Online Hemodiafiltration: A Real-World Evidence Study in a Russian Dialysis Network

Abstract

Background: Evidence suggests that online hemodiafiltration (OL-HDF) is associated with improved survival. Whether the dose-response relationship between convective volume and mortality may be confounded by selection bias or descends from practice patterns is not clear. We sought to evaluate the role of patients’ characteristics and practice patterns on OL-HDF dose and mortality in a large private dialysis network in the Republic of Russia. Methods: In this multicenter, historical cohort study, we included adult incident patients on OL-HDF with at least 90 days of survival on renal replacement therapy in centers belonging to the Russian Federation Fresenius Medical Care network (January 1, 2011, to December 31, 2016). We evaluated predictors and outcomes (survival) of substitution volume target achievement (Q_sub > 21 L/session). Results: Among 1,081 enrolled patients, the average Q_sub was 22.9 (±3.2) L/session; the mean ultrafiltration volume was 1.6 (±0.8) L/session. The mean age was 55.8 ± 13.2; 42% were woman. Most common comorbidities were congestive heart failure (39.7%) and peripheral vascular disease (21.7%). The average hemoglobin was 9.3 ± 1.3. The case-mix adjusted center effect accounted for 20% of variance in Q_sub. The top 10 most important variables associated with higher Q_sub were effective Q_b, serum protein, Charlson’s comorbidity index, hemoglobin, year of dialysis initiation (proxy of high Q_sub treatment policy in the clinic network), predialysis heart rate, serum bicarbonate, serum phosphate, age, serum sodium, and dry body weight. In addition, we found that the association of Q_b with Q_sub is moderated by year of enrollment, intradialytic weight gain, and coronary artery disease, whereas higher hemoglobin concentration moderated the relationship between treatment time and Q_sub. Finally, Q_sub between 21 and 25 L/session was associated with longer 5-year survival. Conclusions: Both center-dependent clinical practice standards and patient clinical conditions substantially contributed to the risk of low Q_sub. We confirmed previous evidence indicating better survival among patients with Q_sub ≥ 21 L/session.