Frailty and In-Hospital Mortality Risk Using EHR Nursing Data

Area of Science:

• Gerontology
• Health Informatics
• Clinical Medicine

Background:

• Frailty is a significant predictor of adverse health outcomes, including mortality.
• Accurate identification of frailty in hospitalized older adults is crucial for timely intervention.
• Electronic Health Record (EHR) data offers potential for developing robust frailty assessment tools.

Purpose of the Study:

• To evaluate four definitions of a Frailty Risk Score (FRS) using EHR data.
• To assess the predictive accuracy of these FRS definitions for in-hospital mortality in adults aged 50 and above.
• To compare FRS performance based on nursing flowsheet data, ICD-10 codes, and blood biomarkers.

Main Methods:

• Retrospective observational study analyzing EHR data from medical-surgical units.
• Development and evaluation of four distinct Frailty Risk Score (FRS) definitions.
• Application of survival analysis and Cox regression models, including integrated area under the ROC curve (iAUC) for accuracy assessment.

Main Results:

• The best performing FRS (FRS-NF-26-LABS), incorporating nursing flowsheet data and blood biomarkers, showed significant predictive power for in-hospital mortality (Adj. HR = 1.30).
• This FRS achieved good predictive accuracy (iAUC = .794), comparable to a reduced model with age, sex, and FRS alone.
• An FRS based solely on ICD-10 codes demonstrated the poorest predictive performance.

Conclusions:

• The Frailty Risk Score (FRS) effectively captures patient-specific information that increases the risk of in-hospital mortality.
• Identifying frailty through FRS allows for enhanced patient care, including closer monitoring and individualized resource allocation.
• The FRS supports more informed clinical decisions regarding treatment and discharge planning for at-risk older adults.