Kidney Week

Abstract: TH-OR18

Prognostication of 90-Day Kidney Function Recovery in Patients with AKI Receiving Outpatient Dialysis

Session Information

• AKI: New Frontiers in Prognostication and Management
  October 24, 2024 | Location: Room 6C, Convention Center
  Abstract Time: 05:50 PM - 06:00 PM

Category: Acute Kidney Injury

• 101 AKI: Epidemiology, Risk Factors, and Prevention

Authors

• Babroudi, Seda, Tufts Medical Center, Boston, Massachusetts, United States

Seda Babroudi, MD

• Maas, Carolien C.h.m., Erasmus MC, Rotterdam, Zuid-Holland, Netherlands

Carolien C.h.m. Maas

• Weiner, Daniel E., Tufts Medical Center, Boston, Massachusetts, United States

Daniel E. Weiner, MD, MS, FASN

• Neyra, Javier A., UAB Hospital, Birmingham, Alabama, United States

Javier A. Neyra, MD, MS, FASN

• Kent, David M., Tufts Medical Center Predictive Analytics and Comparative Effectiveness (PACE) Center, Boston, Massachusetts, United States

David M. Kent, MD

• Sanders, Ronald, Dialysis Clinic Inc, Nashville, Tennessee, United States

Ronald Sanders

• Manley, Harold, Dialysis Clinic Inc, Nashville, Tennessee, United States

Harold Manley, PharmD, FASN

• Lacson, Eduardo K., Dialysis Clinic Inc, Nashville, Tennessee, United States

Eduardo K. Lacson, MD, MPH, FASN

• Drew, David A., Tufts Medical Center, Boston, Massachusetts, United States

David A. Drew, MD, MS

Background

Prognostic uncertainty regarding kidney function recovery for patients with AKI-D may result in suboptimal dialysis care. We developed and internally validated a prognostic model for 90-day kidney function recovery to dialysis independence in patients with AKI-D.

Methods

Candidate predictors included demographic, clinical, laboratory, and dialysis characteristics readily available as part of usual care, ascertained in the first week of outpatient dialysis from a national dialysis provider. We derived a parsimonious model by applying backward selection guided by the Bayesian Information Criterion to a multivariable Cox proportional hazards model. Model calibration and discrimination were assessed. Internal-external cross-validation was performed using three geographic regions.

Results

Of 2,544 individuals with AKI-D, 771 patients recovered kidney function within 90-days of outpatient dialysis initiation (cumulative incidence 33.8%, 95% CI 31.8-35.7). Thirteen predictor variables were retained in the final model. Internal-external cross-validation suggested good external calibration and discrimination (Uno’s C-index 0.70 (95% CI 0.67-0.73) Figure 1).

Conclusion

We developed a prediction model to prognosticate 90-day kidney function recovery among patients with AKI-D within the first week of outpatient dialysis initiation. The model demonstrated good performance, supporting its use to inform shared decision-making between dialysis clinicians and patients.

Figure 1. (A) Predictors and Calibration Plots of Final Model after Uniform Shrinkage using (B) Region 1; (C) Region 2; or (D) Region 3 as an external validation set

Funding

• Other NIH Support