Identifying frail older people using predictive modeling.

Area of Science:

• Geriatric Medicine
• Health Services Research
• Predictive Analytics

Background:

• Frailty is a significant concern in the elderly population, impacting health outcomes and quality of life.
• Accurate identification of frail individuals is crucial for targeted interventions and resource allocation.
• Existing frailty assessment tools vary in their clinical applicability and predictive power.

Purpose of the Study:

• To evaluate the ability of the Adjusted Clinical Groups-diagnoses based computerized predictive model (ACG Dx-PM) to identify frail elderly individuals.
• To compare the ACG Dx-PM's frailty identification with the validated Vulnerable Elders Survey (VES).
• To assess the predictive capability of both the ACG Dx-PM and VES for adverse health outcomes in older adults.

Main Methods:

• Secondary analysis of administrative and survey data from 195 participants (age > 65) in a geriatric assessment study.
• Screening for frailty using the ACG Dx-PM and the VES.
• Comparison of demographic, clinical characteristics, and adverse outcome prediction between frail and non-frail groups identified by ACG Dx-PM.

Main Results:

• Patients identified as frail by ACG Dx-PM exhibited clinical characteristics consistent with frailty, including older age, lower marriage rates, and higher incidence of falls and incontinence.
• The ACG Dx-PM demonstrated moderate success in identifying frail individuals compared to the VES (Area Under Curve = 0.62).
• Both the ACG Dx-PM and VES were able to predict adverse outcomes, with the ACG frailty tag predicting hospitalizations and VES predicting emergency department visits.

Conclusions:

• The ACG Dx-PM effectively identifies elderly individuals with frailty characteristics and shows moderate agreement with the VES.
• Both frailty assessment tools demonstrate utility in predicting adverse health events in older adults within a health maintenance organization.
• A combined approach utilizing predictive modeling and function-based surveys warrants further investigation for comprehensive frailty screening.