Related Experiment Video
Updated: Jun 21, 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
Assessing frailty in older adults: discriminative properties of physical activity questionnaires
Mert Doğan1, Özge Erol Doğan2, Fatma Ayvat3
1Department of Physiotherapy and Rehabilitation, Faculty of Health Sciences, Akdeniz University, Antalya, Türkiye.
Introduction:
Frailty is prevalent among older adults and is a growing public health concern. Physical activity questionnaires are pragmatic tools for screening; however, their discriminative capacity remains uncertain. This study aimed to determine the discriminative performance and optimal cut-offs of the International Physical Activity Questionnaire-Short Form (IPAQ-SF) and the Physical Activity Scale for the Elderly (PASE) for identifying frailty states (non-frail, pre-frail, and frail) in a clinically derived cohort of older adults.
Methods:
The study enrolled 289 older adults who were assessed using a demographic form, the FRAIL Scale, the IPAQ-SF, and the PASE. Frailty discrimination was evaluated via receiver operating characteristic analysis (ROC); optimal thresholds were identified using the Youden index, diagnostic indices (sensitivity, specificity, accuracy, predictive values) were computed, micro-/macro- area under the curves (AUC) were reported, and scale AUCs were compared using the DeLong test.
Results:
Participants had a mean age of 72.5 ± 7.0 years, and 56.7% were women. Frailty prevalence was 31.1% non-frail, 29.4% pre-frail, and 39.4% frail. For IPAQ-SF, AUCs were 0.45 (non-frail), 0.40 (pre-frail), and 0.67 (frail); for PASE, AUCs were 0.39 (non-frail), 0.50 (pre-frail), and 0.65 (frail). Macro-/micro-AUCs were 0.51/0.53 for IPAQ-SF and 0.51/0.52 for PASE. No significant differences were found between the two instruments across the frailty categories (DeLong test, p > 0.05). At optimal cut-offs (IPAQ-SF ≤ 322.5; PASE ≤ 63.6), both questionnaires showed moderate agreement with the FRAIL Scale (κ = 0.46-0.48), whereas agreement between the instruments was only low-to-moderate (κ = 0.32, McNemar p = 0.015).
Conclusion:
IPAQ-SF and PASE demonstrated limited ability to discriminate frailty status in older adults, with moderate accuracy in identifying frail individuals and poor discrimination between the non-frail and pre-frail groups. The IPAQ-SF showed slightly higher specificity, whereas the PASE demonstrated higher sensitivity, indicating that they capture distinct aspects of physical activity behavior but cannot be used interchangeably.
