Predicting Hospice Use Among American Indian/Alaska Native Persons with End-Stage Kidney Disease

Area of Science:

• Nephrology
• Public Health
• Health Disparities

Background:

• American Indian/Alaska Native populations experience disproportionately high rates of end-stage kidney disease (ESKD).
• Treatment options for ESKD, including kidney transplantation and dialysis, are subject to social determinants of health, with rural populations often favoring dialysis due to limited transplant access.
• Patients in these groups are less likely to receive hospice care, potentially leading to reduced quality of life.

Purpose of the Study:

• To develop a predictive model estimating hospice use likelihood among kidney transplant recipients.
• To identify key factors influencing hospice care utilization in this population.

Main Methods:

• Retrospective analysis of the 2022 USRDS Standard Analysis Files.
• Utilized Area Deprivation Index (ADI) data, encompassing socioeconomic and healthcare access variables.
• Employed machine learning techniques including regression trees, random forest, boosting, and support vector machines to predict hospice use.

Main Results:

• Random forest and boosting models demonstrated superior performance in predicting hospice use.
• Key predictors identified include IHS region, age, and Area Deprivation Index (ADI).
• Results suggest a nonlinear relationship between covariates and hospice use, potentially better captured by machine learning models than traditional logistic regression.

Conclusions:

• Predictive models incorporating measures like ADI can reveal geographic disparities in healthcare access.
• Interventions should target underserved communities to improve access to hospice and palliative care.
• Healthcare systems must prioritize equitable access to palliative care for all ESKD patients.