Predictors of mortality risk in hemodialysis patients in the medium term: a case series

BACKGROUND: Identifying predictors of mortality risk in patients receiving hemodialysis in the medium term enables the development of measures aimed at reducing this risk. However, data obtained from different dialysis centers may be highly heterogeneous, which reduces the reliability of the resulting estimates. In such cases, statistical tools for aggregating results, traditionally used in meta-analyses, may be helpful. This approach can substantially increase confidence in the findings. AIM: To identify predictors of mortality risk in patients receiving maintenance renal replacement therapy in the medium term, with adjustment for data heterogeneity in a multicenter study. METHODS: The study included data from a retrospective, continuous November 1, 2017, to April 30, 2020, involving patients aged ≥18 years who had been receiving outpatient maintenance hemodialysis for at least 3 months and had a functioning vascular access. Censoring criteria included transition to peritoneal dialysis or kidney transplantation during the 30-month follow-up period. At the first stage, hierarchical Bayesian Cox proportional hazards models were estimated separately for four clusters of dialysis centers formed territorially. At the second stage, significant predictors of survival were selected, and generalizing models with fixed or random effects were estimated for these predictors (models were selected based on Cochran’s Q test). RESULTS: Retrospective data over 2.5 years were collected for 2120 patients who met the inclusion and exclusion criteria across 30 dialysis centers. Over the 30-month follow-up, 468 patients died (22.08%); 7 patients (0.33%) were transferred to peritoneal dialysis (all in Ufa); and 42 patients (1.98%) underwent kidney allotransplantation. The analysis demonstrated that only patient age was a risk factor for mortality (hazards ratio [HR], 1.02; 95% confidence interval [CI], 1.01–1.03). All other factors were associated with a reduced risk of death as their values increased: duration of dialysis therapy (HR, 0.95; 95% CI, 0.92–0.99), body mass index (HR, 0.93; 95% CI, 0.87–1), duration of anti-anemic therapy in months (HR, 0.91; 95% CI, 0.86–0.97), and duration of therapy aimed at correction of calcium–phosphate metabolism in months (HR, 0.94; 95% CI, 0.92–0.96). The mean single-pool (sp) ratio of dialyzer urea clearance (K) multiplied by dialysis time (t) to the volume of urea distribution in the patient’s body (V) (spKt/V) showed no significant association with mortality risk: although the HR was 0.35, the CI was very wide (0.08–1.57), indicating that an increase in spKt/V was associated with a reduced risk of death in some patients and an increased risk in others. CONCLUSION: Based on preliminary differences in survival among dialysis patients, four clusters belonging to different constituent entities of the Russian Federation were identified. An approach that aggregates modeling results across territorially defined clusters of dialysis centers resulted in greater confidence in the estimated survival outcomes of patients receiving outpatient maintenance hemodialysis.