Development of a Mortality Prediction Model for Incarcerated Adults to Identify Palliative Care Needs

Area of Science:

• Public Health
• Gerontology
• Criminology

Background:

• Accurate mortality risk estimation is crucial for incarcerated adults.
• Identifying individuals for palliative care and compassionate release is a public health priority.

Purpose of the Study:

• To develop and internally validate a 2-year mortality prediction model for incarcerated adults.
• Enhance clinical decision-making for end-of-life care in correctional facilities.

Main Methods:

• A cohort study was conducted from February 1, 2018, to February 1, 2020.
• Incarcerated adults aged 18 years or older, residing in a California Department of Corrections and Rehabilitation (CDCR) prison for at least 1 year, were included.
• Cox proportional hazards regression with LASSO variable selection was employed, assessing demographics, housing, functional status, healthcare utilization, and chronic conditions.

Main Results:

• The study included 89,430 adults, with 506 (0.6%) experiencing natural death within 2 years.
• The model achieved an optimism-corrected area under the receiver operating characteristic curve (AUC) of 0.926, indicating strong predictive discrimination.
• Calibration plots demonstrated good agreement between predicted and observed mortality, with specific performance metrics evaluated at clinically relevant thresholds.

Conclusions:

• The developed mortality risk model provides valuable estimates for clinicians.
• Facilitates identification of incarcerated individuals who could benefit from advance care planning.
• Supports referrals for palliative care services and compassionate release programs.