Comparative Effectiveness of Intravenous Iron Treatment Protocols in Hemodialysis Patients: Causal Inference with Dynamic Treatment Regimes Public Deposited

Downloadable Content

Download PDF
Last Modified
  • March 21, 2019
  • Li, Xiaojuan
    • Affiliation: Gillings School of Global Public Health, Department of Epidemiology
  • Decisions regarding intravenous iron treatment follow dosing protocols for anemia management of hemodialysis patients. These protocols are a type of dynamic treatment regimes, consisted of a set of decision rules with iron status values - transferrin saturation and ferritin - and corresponding iron dosing patterns. Multiple protocols exist in clinical practice, but their comparative safety is unknown. Using clinical data from a large US dialysis provider linked to healthcare utilization data from United States Renal Data System (2004-2012), our objectives were to (1) develop an approach to identify intravenous iron dosing protocols that were commonly used, and (2) evaluate the comparative safety of continuous exposure to commonly used protocols. The identification approach classified intravenous iron dosing protocols at measurements of iron status tests, where decisions regarding iron treatment occur in clinical practice. Using current test levels and iron treatment experience in a two-week assessment window, candidate protocols were assigned to a patient if they were consistent with treatment experience in the assessment window. Among 43,166 patients who initiated hemodialysis in 2004-2012, 79.1% of them were matched with candidate protocols. The prevalence of protocol matches increased from 75.0% in 2004 to 90.7% in 2012. Higher prevalence of knowingly implemented protocols confirmed the performance of this identification approach. In the comparative safety analyses, we estimated the effect of continuous exposure to the five most commonly initiated protocols in 2009-2012 on risks of mortality and infection-related events. Two less commonly initiated protocols were more aggressive, recommending a large amount of iron at higher iron status levels; their initiators were sicker at baseline. Compared with one commonly initiated and less intensive protocol, these two protocols were at elevated mortality risk (120-day risk differences (95% confidence interval): 1.5% (0.1, 3.1%), 3.1% (1.0, 5.6%)). The magnitude of elevated risk increased with the aggressiveness of the protocols. We observed similar trends in elevated risks for infection-related events among more aggressive protocols. Protocols that recommend less intensive use of iron at high levels of iron status tests may lower risks of mortality and infection-related events, but further exploration is needed to address potential residual confounding and selection bias.
Date of publication
Resource type
Rights statement
  • In Copyright
  • Brookhart, M. Alan
  • Cole, Stephen
  • Kshirsagar, Abhijit
  • Stürmer, Til
  • Fine, Jason
  • Doctor of Philosophy
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2017

This work has no parents.