Related Experiment Video
Updated: May 9, 2026
Measuring Frailty in HIV-infected Individuals. Identification of Frail Patients is the First Step to Amelioration and Reversal of Frailty
Published on: July 24, 2013
Using Discriminant Analysis to Predict Frailty in Community-Dwelling Older Adults in Taiwan
Meei-Horng Yang1,2,3, Hung-Ru Lin4, Chieh-Yu Liu5
1Department of Nursing, Landseed International Hospital, Taoyuan, Taiwan, ROC.
Background:
The definition of frailty is still debated, resulting in the development of various measurement tools. Having a convenient and accurate frailty screening instrument is essential to providing appropriate care to community-dwelling older adults in terms of facilitating the delayed onset of frailty and preventing disability.
Purpose:
This study was conducted to develop a simple, convenient, and rapid screening method for frailty classification in community-dwelling older adults that incorporates the most significant predictive factors from the Study of Osteoporotic Fractures index components and the Kihon Checklist tool domains.
Methods:
Convenience sampling was used to gather longitudinal data from 110 community-dwelling older adults at baseline (T0), 6 months (T1), and 1 year (T2) using three different frailty screening tools. The Fried frailty phenotype tool was used as the gold standard. Linear discriminant analysis was conducted to create an effective model for accurately classifying frailty states.
Results:
The discriminant analysis generated three statistical significant functions, which respectively explained 33.6% ( Rc =.58; df =3; p <.0001), 26.0% ( Rc =.51; df =2; p <.0001), and 29.2% ( Rc =.54; df =2; p <.0001) of the predictive power of prefrail/frail risk. The discriminant functions demonstrated sensitivities of 64.6%-69.4% for identifying the prefrail/frail group and specificities of 77.1%-90.9% for identifying the robust group. The developed method successfully classified the correct robust and prefrail/frail states for 71.6%-79.1% of participants. The findings of this longitudinal study show weight loss, reduced energy levels, physical function, activities of daily living (IADL lifestyle), and eating function to be the most significant factors at baseline for accurately classifying community-dwelling older adults into robust and prefrail/frail states over a 1-year follow-up period.
Conclusions/Implications For Practice:
Eating function was identified as the strongest factor of influence on the correct prediction of frailty status. Nurses may use the five questionnaire-based domains in initial assessments to classify frailty in community-dwelling older adults with a 1-year accuracy of at least 70%. Those identified as at-risk should be referred to physicians, nutritionists, rehabilitation specialists, and/or long-term care services to optimally delay or prevent the onset of frailty in this population.
