Full title: Wearable telemonitoring systems for fall-risk assessment and fall detection in adults aged 65 years and older
Authors: León Salas B, Manzano Armas KJ, Rada Ramírez IC, Herrera Ramos E, Capafons Sosa JL, García Hernández M, González Hernández Y, Pérez García L, Vallejo Torres L, Ramallo Fariña Y, Guirado Fuentes C, Carmona Rodríguez M, Rivas Ruiz F, Trujillo Martín MM
Contact person: Beatriz León Salas (beatriz.leonsalas@sescs.es)
SUMMARY
Introduction
Falls are one of the leading causes of injury, disability and mortality in people aged 65 and over, and constitute a major public health problem. Between 28% and 35% of older adults experience at least one fall per year, and this problem is expected to increase due to the progressive aging of the population. In 2020, there were approximately 727 million people aged 65 or older in the world, accounting for 9.3% of the total population.
Detecting when a fall occurs and assessing the risk of its occurrence is crucial to prevent future problems and improve quality of life. Beyond immediate physical injuries, which can range from minor bruises to severe fractures or traumatic brain injuries, falls can lead to prolonged periods of immobility, with serious health consequences, or even death.
In clinical practice, fall detection and risk assessment are carried out via self-reports, medical history, and physical tests based on standardized scales and instruments. However, the incorporation of new technologies is being considered, such as telemonitoring systems based on wearable devices, for example bracelets, watches or sensor-equipped belts, which allow real-time fall detection or help identify individuals at higher risk of falling.
Aims
The main objective of this Health Technology Assessment report is to evaluate the safety, clinical effectiveness, and cost-effectiveness of wearable devices used for fall risk assessment and fall detection in people aged 65 and over. In addition, ethical, legal, organizational and social aspects related to their use are analyzed. The aim is to inform the decision on whether this technology should be incorporated into the common service portfolio of the Spanish National Health System (SNS).
Method
Effectiveness and Safety
A systematic review (SR) of the scientific literature published up to June 16, 2025, was conducted in the following electronic databases: MEDLINE (OVID), Embase (Elsevier), CENTRAL (Cochrane Library-Wiley), CINAHL (EBSCOhost), Web of Science Core Collection (Clarivate Analytics), and PEDro (Institute for Musculoskeletal Health, University of Sydney). A comprehensive, broad search was conducted, with no filters or restrictions by date or language.
Original studies published in English or Spanish evaluating wearable devices intended to detect falls and/or assess fall risk were selected. Priority was given to randomized clinical trials (RCTs), and, in their absence, non-randomized trials (NRTs) or observational studies with control groups were considered. In the absence of this evidence, the diagnostic validity of the devices was assessed, including diagnostic performance studies and prospective and retrospective longitudinal studies developing and/or validating predictive or prognostic models.
The key outcome measures considered were: number of falls, response time (detection and reporting of the fall), severity of falls (emergency visits, hospitalizations, etc.), health-related quality of life (HRQoL), adherence, and acceptability. Sensitivity, specificity, accuracy, and area under the ROC curve were considered for diagnostic performance studies.
Risk of bias was assessed using specific tools according to study type: RoB-2 for RCTs, ROBINS-I for non-randomized studies, and QUADAS-2, PROBAST, or QUIPS for diagnostic performance studies, predictive models or prognostic factor studies.
Quantitative synthesis of the results was performed with meta-analysis using Review Manager version 5.4.
The quality of evidence and the strength of recommendations were assessed following the GRADE methodology.
Cost-effectiveness and Economic analysis
The same literature search used for effectiveness and safety allowed the identification of economic evaluations, whether conducted alongside primary studies or through economic models, reporting incremental cost-effectiveness ratio (ICER), costs in monetary units and benefits expressed in quality-adjusted life years (QALYs), life-years gained (LYG), monetary units or any of the effectiveness outcome measures. Methodological quality was assessed using Drummond et al.’s criteria and/or the OSTEBA Critical Reading Form FLC 3.0, as well as data extraction and a narrative synthesis of results.
In addition, the immediate healthcare cost of a fall in the SNS was estimated using a decision tree with four main branches according to the healthcare received after the fall (two in the outpatient setting and two in the hospital).
A comparative economic evaluation of the wearable G-STRIDE device versus two standard clinical tests—the Timed Up & Go (TUG) and Gait Speed (GS) tests—was also conducted. A second decision tree was designed based on a community-dwelling individual aged 65 or older with an initially uncertain risk of falling. The alternatives compared were: the G-STRIDE wearable device, which analyzes gait parameters using an inertial sensor and a machine learning algorithm; the TUG test, assessing balance and functional capacity over a short course; and the GS test, measuring walking speed as an indicator of frailty and fall risk. The cost-effectiveness analysis aimed to estimate the incremental cost per additional correctly identified case provided by the wearable device compared to traditional tests.
Ethical, Legal, Organisational, Social and Environmental Aspects
The analysis of these aspects was based on the same population, intervention and comparison used in the effectiveness and cost-effectiveness assessment. The same literature search was also used. The synthesis was carried out narratively, selecting and organizing results according to their relevance and coherence for decision-making.
Results
Effectiveness and Safety
No RCTs, NRTs, or observational studies with control groups meeting the pre-established selection criteria were found.
A total of sixteen studies were selected: one prognostic study, two diagnostic performance studies, and thirteen predictive model validation studies. Together, they included 4,323 participants (analyzed sample n=3,645), of whom 48.8% were women, with a mean age of 75 years (SD=6.3). Thirty-one percent (1,338) had experienced at least one fall in the previous 12 months.
In thirteen studies, the performance of wearable devices was analyzed for classifying people who had fallen or predicting fall risk, compared with clinical assessments, functional tests used in consultations, or other non-wearable devices. These studies reported sensitivity, specificity, accuracy and/or area under the ROC curve.
No studies were identified that evaluated the use of wearable devices for real-time fall detection, or studies that assessed some of the clinical measures of interest: number of falls, response time (speed of detection and reporting), fall severity (emergency visits, hospitalizations, and fractures), HRQoL, acceptability or adherence.
Only one diagnostic study provided the necessary data (2×2 table) to calculate sensitivity and specificity. In this study, the G-STRIDE wearable device achieved a sensitivity of 77% (95% CI: 55, 85) and specificity of 83% (95% CI: 73, 91), compared to the TUG test, which had a sensitivity of 51% (95% CI: 39, 62) and specificity of 83% (95% CI: 73, 91), and the GS test, with a sensitivity of 75% (95% CI: 64, 84) and a specificity of 79% (95% CI: 68, 87).
The remaining studies only reported results as percentages, allowing only meta-analysis of proportions. In studies analyzing wearable devices for identification/classification of subjects who had already experienced falls (retrospective studies), devices showed 18.8% higher sensitivity than comparators (95% CI: 7.95, 29.74; p=0.0007; I²=100%; k=5), 2.5% lower specificity (not significant; 95% CI: -12.21, 7.28; p=0.62; I²=100%; k=5), and 10.81% higher accuracy (95% CI: 8.45, 13.16; p<0.0001; I²=99%; k=4).
In studies evaluating devices’ ability to predict future falls (prospective studies, mostly with 12-month follow-up), sensitivity was 5.87% lower than comparators (95% CI: -11.18%, -0.57; p=0.03; I²=100%; k=6), specificity was 13.75% higher (95% CI: 10.47, 17.03; p<0.0001; I²=100%; k=5), and accuracy showed a non-significant 1.33% lower difference (95% CI: -4.87, 2.21; p=0.46; I²=99%; k=5).
Overall evidence quality was considered very low.
Cost-Effectiveness and Economic Analysis
No published economic evaluations meeting the selection criteria were identified.
The economic analysis provides a first approximation of the immediate healthcare cost of falls and the diagnostic efficiency of the G-STRIDE wearable device. Results suggest that G-STRIDE improves fall risk detection compared to the TUG test, although its performance is very similar to the GS test. The incremental cost per additional correctly identified case varies depending on the implementation context, being more favorable in large-scale acquisition scenarios and lower software costs.
Ethical, Legal, Organizational, Social and Environmental Aspects
The systematic review identified 22 studies meeting the inclusion criteria.
Findings indicate that the acceptability of these devices depends on perceived usefulness, the sense of safety they provide and the support offered within the care system. Adherence depends on the user’s functional conditions, ease of use, integration into daily routines and technical reliability. Failures in alerts and operational complexity are identified as relevant barriers, whereas intuitively designed and easy-to-use devices facilitate adoption.
Regarding usability and comfort, users value aspects such as manageable size and weight, aesthetic integration with common accessories and adaptation to specific health needs. These features facilitate incorporation into daily life. However, concerns arise related to users’ digital competence, as well as ethical and legal issues regarding data protection and potential unauthorized use of information. Challenges related to preserving individual autonomy and privacy are also noted, requiring a careful balance between safety and personal control.
Conclusions
Based on the SR, economic evaluation and analysis of ethical, legal, organizational, and social aspects, the following conclusions can be drawn regarding the incorporation of wearable devices for fall risk assessment in people aged 65 and over:
- Effectiveness and Safety: The best available evidence comes from 16 studies (N=4,323), but the quality is very low.
- In one diagnostic study, the G-STRIDE device showed 77% sensitivity and 87% specificity for identifying people who had fallen, with results superior to the TUG test and similar or slightly better than the GS test.
- In five retrospective studies, wearable devices showed higher sensitivity and accuracy than comparators for identifying/classifying people who had fallen, with no significant difference in specificity.
- In seven prospective studies, devices were less sensitive but more specific than comparators for predicting future falls.
- Cost-Effectiveness: No published economic evaluations were identified. The average healthcare cost of a fall treated in the Spanish SNS was estimated at €1 489,87. The economic efficiency of wearable devices depends on the implementation context and the availability of precise cost information.
- Ethical, Legal, Organizational and Social Aspects: Adoption of these devices depends on perceived usefulness, technical reliability and adaptation to the characteristics and routines of each older adult. Barriers related to digital skills, economic resources and functional capacity may create inequalities in access. Therefore, planning, training and continuous support for all stakeholders are required. Concerns persist regarding autonomy, privacy and appropriate data use, requiring clear regulatory frameworks and support for both users and professionals.
Key words: Falls, fall detection, fall risk assessment, older adults, telemonitoring, wearable devices.
DOCUMENTS:
- Full report:
