Predictive value of the World falls guidelines algorithm within the AGELESS-MELoR cohort

Aim: The World Falls Guidelines (WFG) Task Force published a falls risk stratification algorithm in 2022. However, its adaptability is uncertain in low- and middle-income settings such as Malaysia due to different risk factors and limited resources. We evaluated the effectiveness of the WFG risk str...

Full description

Bibliographic Details
Published in:Archives of Gerontology and Geriatrics
Main Author: Lee S.J.S.; Tan M.P.; Mat S.; Singh D.K.A.; Saedon N.I.; Aravindhan K.; Xu X.J.; Ramasamy K.; Majeed A.B.A.; Khor H.M.
Format: Article
Language:English
Published: Elsevier Ireland Ltd 2024
Online Access:https://www.scopus.com/inward/record.uri?eid=2-s2.0-85195869383&doi=10.1016%2fj.archger.2024.105523&partnerID=40&md5=59af096fa3ef68b7e6141fffca1d589d
Description
Summary:Aim: The World Falls Guidelines (WFG) Task Force published a falls risk stratification algorithm in 2022. However, its adaptability is uncertain in low- and middle-income settings such as Malaysia due to different risk factors and limited resources. We evaluated the effectiveness of the WFG risk stratification algorithm in predicting falls among community-dwelling older adults in Malaysia. Methods: Data from the Malaysian Elders Longitudinal Research subset of the Transforming Cognitive Frailty into Later-Life Self-Sufficiency cohort study was utilized. From 2013–2015, participants aged ≥55 years were selected from the electoral rolls of three parliamentary constituencies in Klang Valley. Risk categorisation was performed using baseline data. Falls prediction values were determined using follow-up data from wave 2 (2015–2016), wave 3 (2019) and wave 4 (2020–2022). Results: Of 1,548 individuals recruited, 737 were interviewed at wave 2, 858 at wave 3, and 742 at wave 4. Falls were reported by 13.4 %, 29.8 % and 42.9 % of the low-, intermediate- and high-risk groups at wave 2, 19.4 %, 25.5 % and 32.8 % at wave 3, and 25.8 %, 27.7 % and 27.0 % at wave 4, respectively. At wave 2, the algorithm generated a sensitivity of 51.3 % (95 %CI, 43.1–59.2) and specificity of 80.1 % (95 %CI, 76.6–83.2). At wave 3, sensitivity was 29.4 % (95 %CI, 23.1–36.6) and specificity was 81.6 % (95 %CI, 78.5–84.5). At wave 4, sensitivity was 26.0 % (95 %CI, 20.2–32.8) and specificity was 78.4 % (95 %CI, 74.7–81.8). Conclusion: The algorithm has high specificity and low sensitivity in predicting falls, with decreasing sensitivity over time. Therefore, regular reassessments should be made to identify individuals at risk of falling. © 2024
ISSN:1674943
DOI:10.1016/j.archger.2024.105523