Background
Controversy exists about any differences in longer-term safety across different intravenous iron formulations routinely used in hemodialysis (HD) patients. We exploited a natural experiment to compare outcomes of patients initiating HD in facilities that predominantly (in ≥90% of their patients) used iron sucrose (IS) versus sodium ferric gluconate complex (FG).
Study Design
Retrospective cohort study of incident HD patients.
Setting & Participants
Using the US Renal Data System, we hard-matched on geographic region and center characteristics HD facilities predominantly using FG with similar ones using IS. Subsequently incident HD patients were assigned their facility iron formulation exposure.
Intervention
Facility-level use of IS versus FG.
Outcomes
Patients were followed up for mortality from any, cardiovascular, or infectious cause. Medicare-insured patients were followed for infectious and cardiovascular (stroke, myocardial infarction) hospitalizations and for composite outcomes with the corresponding cause-specific deaths.
Measurements
Hazard ratios.
Results
We matched 2,015 IS facilities with 2,015 FG facilities, in which 51,603 patients (IS: 24,911; FG: 26,692) subsequently initiated HD. All recorded patient characteristics were balanced between groups. Over 49,989 person-years, 10,381 deaths (3908 cardiovascular; 1209 infectious) occurred. Adjusted all-cause (HR, 0.98; 95% CI, 0.93–1.03), cardiovascular (HR,0.96; 95% CI, 0.89–1.03), and infectious mortality (HR, 0.98; 95% CI, 0.86–1.13) did not differ between IS and FG facilities. Among Medicare beneficiaries, no differences between FG and IS facilities were observed in fatal or non-fatal cardiovascular events (HR, 1.01; 95% CI, 0.93–1.09). The composite infectious end point occurred less frequently in IS versus FG facilities (HR, 0.92; 95% CI, 0.88–0.96).
Limitations
Unobserved selection bias from non-random treatment assignment.
Conclusions
Patients initiating HD in facilities almost exclusively using IS versus FG had similar longer-term outcomes. However, there was a small decrease in infectious hospitalizations and deaths in patients dialyzing in facilities predominantly using IS. This difference may be due to residual confounding, random chance, or a causal effect.