Sensitivity and specificity of three screening tools for frailty in hospitalized older adults

Abstract

Aim: To determine the test accuracy, including sensitivity, specificity, positive predictive value, negative predictive value and area under curve, of three frailty screening tools in identifying the risk of frail outcomes among hospitalized older patients.

Design: Prospective longitudinal study.

Methods: The screening tools [Frail-PPS (Frail-Physical, Psychological and Social), Frailty Assessment Measure (FAM), and Identification of seniors at-risk hospitalized patients (ISAR-HP)] were administered by ward nurses to patients aged 65 years and older within 24 h of admission to an acute hospital. Sensitivity, specificity, positive predictive value, negative predictive value and area under curve analysis of the three tools in the context of three frail outcomes, (a) functional decline at three months after discharge-defined as a decline of at least one point on the Katz Index, (ii) requiring a full-time caregiver upon discharge, and (iii) death by three months after discharge, was assessed.

Results: Of 366 patients enrolled in the study, 78 (21.3%) experienced one or more frail outcomes, with 65 (17.76%) experiencing functional decline, 61 (16.67%) requiring a full-time caregiver upon discharge and 8 (2.19%) dying by three months. Frail-PPS had sensitivity 12.5% to 31.4% and specificity 91.2% to 94.8%, varying by the considered frail outcome. Similarly, FAM had sensitivity 12.5% to 29.4% and specificity 90.9% to 94.1%, and ISAR-HP had sensitivity 2.9% to 19.2% and specificity 92.2% to 99.1%. positive predictive value of the FAM, Frail-PPS and ISAR-HP ranged from 3.0 to 45.5%, 3.1 to 50.0% and 3.9 to 23.6% respectively, while their negative predictive value ranged from 87.1% to 97.9%, 87.7% to 97.9% and 92.2% to 99.4% respectively. The area under curve values were moderate for the Frail-PPS (0.56 to 0.75), FAM (0.58 to 0.70) and ISAR-HP (0.71 to 0.77) for the three outcomes.

Conclusions: With high specificity and negative predictive values, as well as low sensitivity, FAM and Frail-PPS may be beneficial in identifying older individuals who are not frail, minimizing unnecessary further assessment and intervention.

Keywords: Frailty; Hospitalized; Older adults; Screening; Validity.