Clinical validation of a frailty management mHealth tool in a cohort of community-dwelling older adults: the Geras Fit-Frailty App Authors: Courtney Clare Kennedy, George Ioannidis, Kenneth Rockwood, Aastha Relan, Jonathan Adachi, Alexandra Papaioannou, Pauline Fisher Published: 2025 ABSTRACT AbstractObjectivesThis study describes the prototype testing and clinical validation of the Fit-Frailty App, a fully guided, interactive mobile health (mHealth) app to assess frailty and sarcopenia. This multi-dimensional tool is freely available on the App Store and considers medical history, physical performance, cognition, nutrition, daily function and psychosocial domains. To guide management, a total frailty score and clinical summary of underlying “risk flags” are provided. Our objectives were to examine usability, feasibility, criterion and construct validity.DesignCross-sectionalSettingOutpatient geriatric medicine clinicParticipantsCommunity-dwelling older adults, age 65 years or olderMethodsThe primary outcome of the clinical validation study was criterion validity. A research nurse administered the Fit-Frailty App during a routine clinic appointment. Clinicians simultaneously completed a paper-based frailty index (FI) tool with similar items from a comprehensive geriatric assessment (FI-CGA). Total scores for both assessments were computed using the cumulative deficits frailty index scoring method. Intraclass and Pearson correlation coefficients and 95% CIs were calculated to examine criterion validity. Secondary outcomes were construct validity, feasibility (eg, completion rates, safety occurrences, resources) and usability (eg, ratings on ease of use, time to complete the app).ResultsIn the clinical validation study (n=75, mean age 79.2, SD=7.0, 53% female), the mean total Fit-Frailty App score was 0.33 (SD=0.13) with 73% of our sample considered frail or severely frail. The app presented comparable results to FI-CGA (moderate to good validity; ICC=0.65, 95%CI=0.50–0.76) with a strong association between the measures (r=0.74, 95%CI=0.62–0.83). In our prototype and clinical cohorts, the app had a 100% completion rate with no safety occurrences and had high usability ratings.ConclusionsThe Fit-Frailty App is a feasible and valid tool that can be used in research and clinical settings to comprehensively assess frailty and sarcopenia by non-geriatricians and could assist with developing targeted interventions. STRENGTHS AND LIMITATIONS OF THIS STUDY The Fit-Frailty App is a multidimensional frailty assessment which also includes muscle strength and physical performance to assess for the presence of sarcopenia, a key component of frailty. The Fit-Frailty App is a multidimensional frailty assessment which also includes muscle strength and physical performance to assess for the presence of sarcopenia, a key component of frailty. This m-Health tool utilises both objective and self-reported items and guides non-expert assessors via standardised script, instructional videos and automated scoring and summarises underlying “risk flags” to assist with targeted treatment plans. This m-Health tool utilises both objective and self-reported items and guides non-expert assessors via standardised script, instructional videos and automated scoring and summarises underlying “risk flags” to assist with targeted treatment plans. The utilisation of the cumulative deficits frailty index methodology provides an estimate of the severity of frailty along a continuum. The utilisation of the cumulative deficits frailty index methodology provides an estimate of the severity of frailty along a continuum. The sample was drawn from community-dwelling older adults and hence may not represent patients in other settings such as acute care or rehabilitation. The sample was drawn from community-dwelling older adults and hence may not represent patients in other settings such as acute care or rehabilitation. The tool requires further validation such as test–retest reliability and whether it can measure response to treatment. The tool requires further validation such as test–retest reliability and whether it can measure response to treatment. Introduction Frailty has been described as “the most problematic expression of population ageing”.1 An age-related, cumulative decline in multiple organ systems leads to a loss of physiologic “reserve”, decreased function and increased risk of adverse outcomes including falls, fractures, emergency room visits, hospitalisation and mortality.13 Sarcopenia, a progressive and generalised loss of skeletal muscle mass and function,4 is considered a key component of frailty which causes or exacerbates the decline in physical function.1 5 The presence of sarcopenia additionally contributes to increased risk of falls, muscle weakness and reduced mobility (eg, slow walking, difficulty transferring from bed or a chair).6 7 Internationally, there is a shortage of geriatricians and specialist geriatric services and alternative approaches to comprehensive assessment and care planning for older adults must be explored.8 Given increasing evidence that frailty and sarcopenia can be improved,912 intervening to effectively prevent or manage frailty and sarcopenia is an area of increasing focus across the continuum of care. Systematic assessment to understand the severity of frailty and identify which underlying elements require further intervention ultimately equips clinicians to make better targeted recommendations and develop individualised care plans. Conceptually, defining frailty has traditionally been separated into two “schools of thought”: (1) frailty as a syndrome with observable physical properties13 or (2) frailty as a nonspecific state of increasing risk due to an accumulation of “deficits” (eg, symptoms, signs, disabilities, diseases and laboratory measurements).14 15 Both of these concepts have associated measurement tools (Fried phenotype,13 cumulative deficits frailty index14). These and several other screening tools have varied properties16 including: (1) scoring methodology (ordinal “not frail/pre-frail/frail” vs interval/continuous score), (2) domains included (physical only vs multidimensional), (3) objectivity (patient self-report only or the inclusion of additional objective measurements), (4) comprehensiveness (quick screen vs multiple items or clinical examination/history), (5) administration modality (paper-based vs electronic application) and (6) level of training required (basic training vs expert clinical judgement). The clinical diagnosis of sarcopenia typically involves a graded approach of muscle strength (eg, grip strength) and performance measures (Short Physical Performance Battery) and testing for low muscle mass as needed using Dual-energy X-ray Absorptiometry or Bioelectrical Impedance Analysis.6 7 Objectives Mobile health (mHealth) apps (ie, using smartphones or tablets) that help with assessment, decision-making and management at the point of care are increasingly used in both clinical and research settings.17 Our aim was to create a user-friendly application that provides a multidimensional assessment of frailty and sarcopenia, which also captures level of severity, measures clinical change and informs care planning by summarising underlying “risk flags”. Below, we provide an overview of the creation of the Fit-Frailty mHealth App from conception to prototype and clinical validation in a cohort of community-dwelling older adults in Ontario, Canada. Our primary objectives were (1) to examine usability and feasibility of administering this tool with patients and family caregivers and (2) to examine criterion and construct validity by comparing results with comprehensive geriatric assessment (CGA) and other geriatric assessment tools. Methods Phase 1: App development Ideation and research The Fit-Frailty App was developed at the Geras Centre for Aging Research at Hamilton Health Sciences (affiliated with McMaster University) in consultation with the Geriatric Medicine Research Group at Dalhousie University. We invited professionals involved in geriatric care (physicians, occupational and physical therapists, pharmacists, nurses, kinesiologists) to be expert evaluators regarding the composition of the tool. The self-reported sections are adapted from the Canadian Multicentre Osteoporosis Study (CaMOS) Frailty Index, which was developed using the cumulative deficits approach14 18 and validated in CaMOS (n=9423 participants aged 25 years and older).19 Additional physical performance and muscle strength measures were chosen as well-validated indicators of physical frailty and sarcopenia.2022 The interactive cognitive screening was adapted from the Trail Making Test23 (TMT-A: Rote memory and TMT-B: executive function) and intended as a broad screen of whether cognitive impairment may be present. Prototype development As described in figure 1, the research team partnered with digital health experts (Mohawk College mHealth and eHealth Development and Innovation Centre and the Centre for Data and Digital Health (CREATE)). Early usability and wireframe testing was completed in a “live lab” with students, researchers and older adult volunteers. Across phases, an iterative process of seeking end-user feedback from clinical collaborators and research participants was utilised to design the user interface (UI; interaction with screens, buttons, toggles and other visual elements) and user experience (UX).24 Based on mHealth methodology25 and user feedback, several interactive elements were incorporated to enhance usability including the ability to access instructional videos and “help” buttons, automated timers and scoring and separating the medical history section, so it can be abstracted from the electronic medical record, as appropriate. The overall Fit-Frailty score and clinical summary of underlying frailty “risk flags” (ie, deficits) are provided on the final screen; results can be downloaded as PDF or CSV files. Fit-Frailty App Prototype Development and Clinical Validation Phases. Phase 2: prototype testing An initial prototype, with self-report questions only, was implemented for pilot testing at baseline in two cohorts of older adults participating in other research studies (Figure 1).26 These sub-studies were approved by the Hamilton Integrated Research Ethics Board (HiREB #1565 and #2557). Cohort 1 (n=52, mean age=74.0 years, SD=7.6) was undergoing joint replacement surgery and had early frailty.27 Cohort 2 (n=23, mean age=76.9 years, SD=5.7) had mild cognitive impairment (MCI) or early frailty.28 Overall, 68% were women, and 41% had post-secondary education. Basic usability elements in the self-report sections were measured (eg, ratings on ease of use, time to complete the app, unclear questions or item responses). Results indicated the mean time to complete the self-report sections was 3.9 min (SD=2.2) in the early frailty group and 4.3 min (SD=2.3) in the group with MCI. Twice as many participant queries were related to the meaning or phrasing of the question rather than problems navigating or interacting with the tablet screens. Overall, 70% of participants were able to complete all the screens without assistance from the research assistant. Based on a 5-point scale, 87% of participants rated the app as “Very Easy” or “Easy” to use. Final version Following the initial prototype, the self-report section was updated for clarity, and optional interactive components were added: (1) an interactive cognitive screen and (2) physical performance and muscle strength measures. The Fit-Frailty App V1.4 is currently available in English on the App Store at https://apps.apple.com/au/app/fit-frailty/id1540125260 and is functional on iPad and iPhone. A user manual is available for download. An Android version and additional languages are anticipated in the future. figure 2 provides an overview of the domains included in the Fit-Frailty App. The test is administered in four consecutive sections, as follows: Section 1: Physical performance and muscle strength tests including gait speed, balance, chair stands and grip strength. Physical performance items can be skipped, if necessary, and the self-report items are used instead in final scoring (online supplemental file 1). Fit-Frailty App Domains. Section 2: Cognitive screening whereby verbal instructions are given by the administrator and the patient interacts with the device. The two screens are adapted from the TMT-A and TMT-B tests.23 The self-report Ottawa 3DY can be used instead for individuals with dexterity or visual impairments.29 Section 3: Self-reported health including daily function, nutrition and psychosocial aspects is completed by the patient. In our clinical protocol, caregivers were asked to complete this section together with participants who did not pass the cognitive screen. Section 4: Medical history can be abstracted from the health record by the test administrator. If the health record is not available, the patient or caregiver can provide information. The complete table of all Fit-Frailty App items and scoring criteria is provided in online supplemental table 1 of the supplementary materials. Phase 3: clinical validation Study design and participants Phase 3 (Figure 1) was a non-experimental, cross-sectional clinical validation study whereby our app was compared with CGA, which is considered the gold standard clinical assessment of frailty and function in older adults. We recruited a convenience sample of participants over age 65 from a regional geriatric medicine outpatient clinic in Ontario, Canada, during 2021. Initially 87 patients were approached, of which 75 patients agreed to participate. Patients had been referred to the clinic for CGA, typically related to cognition, falls, medication review and multisystem morbidity. The only inclusion criteria were age 65 years or older and the patient or accompanying caregiver spoke English. A research assistant explained the study and obtained written informed consent. If the patient lacked decisional capacity (ie, impaired cognition or reasoning ability evaluated based on clinical judgement of the geriatrician or nurse), a legally designated caregiver provided written consent given the patient indicated a willingness to participate. This study was approved by the Hamilton Integrated Research Ethics Board (HiREB #11288). Once patients were consented, the Fit-Frailty App assessment was administered by a study nurse utilising an iPad tablet during the routine clinic appointment. De-identified Fit-Frailty App data were exported and stored in REDCap (Research Electronic Data Capture), a secure web-based software platform designed to support data capture for research studies.30 31 Results from the Fit-Frailty App administration were for research purposes only and were not shared with any member of the clinical team. The geriatrician or resident physician independently completed a hard-copy assessment flowsheet from their CGA, and an overall FI-CGA score was later computed from this flowsheet.32 The FI-CGA and Fit-Frailty App had the same domains (figure 2); however, individual items were not identical. Additional measures (with licensed use) completed during the routine clinic appointment were: 9-point Clinical Frailty Scale (CFS),33 the Standardised Mini-Mental State Examination (SMMSE),34 the Montreal Cognitive Assessment (MoCA),35 and the Geriatric Depression Scale-15 (GDS-15).36 Data to compute Fried Frailty Phenotype scores13 were also collected. Blinding: The principal investigator and App co-creators were not involved with the collection, data entry or analysis of the data. A research nurse administered the App, a research assistant entered all data into REDCap and an independent statistician was hired to analyse the data. Fit-Frailty App scoring Total Fit-Frailty App scores are computed using the cumulative deficits frailty index method.18 19 32 Each item (termed a “deficit”) is scored as 0, 0.5 or 1, and the total frailty score is based on the sum of individual deficits divided by the total number of deficits considered. Total Fit-Frailty App scores range from 0 (representing full health) to 0.7 (representing maximal frailty). A score of 1.0 is theoretically possible; however, >99% of people have scores below 0.7. We also categorised total scores into non-frail (<0.25), frail (0.25–0.4) and more severely frail (>0.4) for clinical utility on the results screen. This was based on previous studies where scores above 0.25 are indicative of the “frail zone” and scores above 0.4 may indicate more severe frailty with an elevated chance of poor outcomes including mortality.37 38 Sample size calculation The PASS v.20.0.3 software package (NCCS Statistical Software, Kaysville, UT, USA) was used to perform the sample size calculation. Assuming an intraclass correlation coefficient (ICC) of 0.75 and a CI width of±0.10, we estimated a sample size of n=75 was needed to examine concurrent validity (ie, a type of criterion validity). Statistical analysis Baseline characteristic variables were reported as mean (SD) for continuous variables and as counts (percent) for categorical variables. To assess concurrent validity (ie, agreement between the Fit-Frailty App and FI-CGA), intraclass and Pearson correlation coefficients (r), and 95% CIs were calculated. ICCs was calculated using a one-way random effect model with absolute agreement and single measurement. Our interpretation of the ICCs was >0.65 indicates good validity, 0.5–0.65 indicates moderate validity, and <0.5 indicates poor validity.39 40 We examined convergent validity (ie, a type of construct validity) by comparing the Fit-Frailty App mood (self-report questions related to loneliness and sadness) and cognitive (TMT-A, TMT-B and self-reported problem-solving/memory) sub-sections with scores on the SMMSE, MoCA and GDS-15. As part of construct validation, we examined how the Fit-Frailty App compared with physical frailty (Fried Frailty Phenotype) and the CFS, a judgement-based tool used to quantify the degree of disability from frailty. This included examining Fit-Frailty App score distributions according to frailty categories on the Fried Frailty Phenotype and 9-item CFS. Pairwise deletion was used to handle any missing data for secondary outcomes. Any missing outcome data are reported in table 1. All statistical analyses were two-sided with the criterion for statistical significance set at α≤0.05, performed using SPSS v.20 and the SAS/STAT v.9.4 software package (SAS Institute Inc., Cary, NC, USA). Table 1Demographic and clinical characteristics of the study sampleCharacteristicBaseline (n=75)Age (years), mean±SD79.2±7.0Female, n (%)41 (53.4)Body Mass Index, mean±SD27.45±5.71Lives alone, n (%)23 (30.7)Dwelling, n (%) House/Apartment70 (93.3) Retirement Home/Assisted Living5 (6.7)Cognitive Status, n (%) Within Normal Limits20 (26.7) Mild Cognitive Impairment29 (38.7) Dementia26 (34.7)MoCA,* mean±SD18.89±6.48 Missing, n (%)9 (12.0)SMMSE,† mean±SD23.82±5.52 Missing, n (%)2 (2.7)GDS-15,‡ mean±SD3.48±3.07Clinical Frailty Scale,§ mean±SD4.19±1.50Fried Frailty Phenotype, n (%) Robust (0)5 (6.7) Pre-frail (1-2)32 (42.7) Frail (3-5)28 (37.3) Missing10 (13.3)FI-CGA,¶ mean±SD0.28±0.11Fit-Frailty App Score, mean±SD0.33±0.13 Not frail (<0.25)20 (26.7) Frail (0.25–0.40)30 (40.0) Severely frail (>0.40)25 (33.3)*Montreal Cognitive Assessment range: 0–30; lower scores indicate worsening cognition.†Standardised Mini-Mental State Examination range: 0–30; lower scores indicate worsening cognition.‡Geriatric Depression Scale (short form) range: 0–15; higher levels indicate worsening depression.§Clinical Frailty Scale range: 1–9; higher levels indicate worsening frailty.¶Frailty Index based on comprehensive geriatric assessment range: 0–1; higher scores indicate worsening frailty. Demographic and clinical characteristics of the study sample Montreal Cognitive Assessment range: 0–30; lower scores indicate worsening cognition. Standardised Mini-Mental State Examination range: 0–30; lower scores indicate worsening cognition. Geriatric Depression Scale (short form) range: 0–15; higher levels indicate worsening depression. Clinical Frailty Scale range: 1–9; higher levels indicate worsening frailty. Frailty Index based on comprehensive geriatric assessment range: 0–1; higher scores indicate worsening frailty. Patient and public involvement Volunteers (eg, older adults and students) contributed to UI and UX design for the initial Fit-Frailty App prototype. Patients and the public were not involved in the research design, conduct, reporting or dissemination plans of this research. Results Table 1 presents the baseline characteristics of our clinical validation sample (n=75). Overall, the mean age was 79.2 (SD=7.0), 53% were female, and 73% of the cohort had cognitive impairment (MCI or dementia). The mean total Fit-Frailty App score was 0.33 (SD=0.13) with 73% of our sample considered frail or severely frail. Feasibility Overall, 75 of 87 clinic patients (86%) who were approached agreed to participate in the study. Reasons for not participating in the study included: aphasic, agitated, drowsy, patient or spouse not interested or reason not given. The Fit-Frailty App was successfully administered to all study participants (100% completion rate) with no safety occurrences. Administration time was not formally measured in this study as the administration was embedded into the flow of the clinic; the research nurses reported that the total time to complete the full assessment was 15 min or less. The past medical history section was completed separately by a research nurse using the electronic medical record (<2 min). Other than the tablet with the app, additional resources included: a Jamar Hydraulic Hand Dynamometer, a sturdy chair and enough space to conduct a 4 m walk test. Criterion validation When we compared Fit-Frailty App scores with the FI-CGA (n=75), the app presented comparable results with moderate to good validity (ICC=0.65, 95%CI=0.50–0.76), and there was strong association between the two measures (r=0.74, 95%CI=0.62–0.83). Construct validation As displayed in table 2, there was a linear trend between scores on the GDS-15 and the Fit-Frailty App mood sub-section. A similar linear trend was observed for cognitive scores, measured by the MoCA and SMMSE and deficits on the Fit-Frailty App cognitive sub-section (table 2). Fit-Frailty App means increased (i.e., increasing frailty deficits) with each level on the Fried Frailty Phenotype (robust/pre-frail/frail). Mean Fit-Frailty App scores were 0.13 (SD=0.11) for the robust phenotype (0 criteria present), 0.28 (SD=0.04) for the pre-frail phenotype (1 or 2 criteria) and 0.42 (SD=0.09) for the frail phenotype (3 or more criteria present). Table 2Comparison of cognition and mood Fit-Frailty App sub-sections with standardised toolsSub-section deficit on the Fit-Frailty App*CognitionMood MoCA † mean±SD SMMSE‡ mean±SD GDS-15§ mean±SD None20.82±5.6926.09±1.412.16±1.78Mild18.81±5.6923.61±5.013.58±2.79Major15.81±8.0321.00±6.7010.00±2.74*Means calculated for each standardised test according to the deficit level (none/mild/major deficit) on the corresponding Fit-Frailty App sub-section (cognition or mood).†Montreal Cognitive Assessment range: 0–30; lower scores indicate worsening cognition.‡Standardised Mini-Mental State Examination range: 0–30; lower scores indicate worsening cognition.§Geriatric Depression Scale (short form) range: 0–15; higher scores indicate worsening depression. Comparison of cognition and mood Fit-Frailty App sub-sections with standardised tools Means calculated for each standardised test according to the deficit level (none/mild/major deficit) on the corresponding Fit-Frailty App sub-section (cognition or mood). Montreal Cognitive Assessment range: 0–30; lower scores indicate worsening cognition. Standardised Mini-Mental State Examination range: 0–30; lower scores indicate worsening cognition. Geriatric Depression Scale (short form) range: 0–15; higher scores indicate worsening depression. Figure 3 demonstrates Fit-Frailty App distributions by CFS level. The Fit-Frailty App non-frail range (<0.25) corresponded with CFS levels 1–3 (very fit/well/managing well), with mean scores 0.16 (SD=0.09), 0.26 (SD=0.09), 0.25 (SD=0.10) per CFS level, respectively. The Fit-Frailty App frail range (0.25–0.40) corresponded with CFS levels 4–5 (living with very mild frailty/living with mild frailty), with mean scores 0.32 (SD=0.11); 0.36 (SD=0.13) per CFS level, respectively. CFS level 6 (living with moderate frailty) was aligned with Fit-Frailty App scores in the more severe frailty range (>0.4) with mean score 0.43 (SD=0.08). There was only one participant with a CFS score of 7 (living with severe frailty), and no participants had a CFS score of 8–9 (approaching end of life/terminally ill). Distribution of Fit-Frailty App Scores by Clinical Frailty Scale Levels. Discussion Based on a well-validated method,14 18 19 an interactive mHealth app to assess frailty and sarcopenia was developed and evaluated. Initial prototype testing of the Geras Fit-Frailty App in a research setting (ie, two cohorts of older adults participating in other research studies, Figure 1-Phase 2) demonstrated high usability ratings for self-report sections. In a clinical setting (Figure 1-Phase 3), the full interactive version was feasible to administer in 15 min or less and was successfully completed for all participants without any safety occurrences. Our criterion and construct validation demonstrated that the Fit-Frailty App was a valid measure of frailty and sarcopenia at the point of care. It exhibited strong association with clinician ratings of a frailty index based on CGA and correlated with other common geriatric medicine assessment tools41). As we expected, a logical dose-response trend was observed with worsening depression, cognition, physical frailty and disability scores across increasing Fit-Frailty App deficit levels. When we compared Fit-Frailty App scores with the FI-CGA, the app presented comparable results with the ICC indicating moderate to good validity.39 40 This was not surprising, as the FI-CGA had different items, did not include sarcopenia measures and was completed by a trainee or staff physician in the context of a routine clinical encounter. While there are many frailty assessment tools, a key strength of the Fit-Frailty App is the detailed clinical summary that identifies the underlying contributors to a frailty state to assist with individualized, tailored interventions. Another key strength is its guided approach: standardised scripts, access to instructional videos and “help” buttons, automated timing and scoring all enhance standardisation and accessibility for “non-expert” assessors. In a busy clinical setting, patients and family caregivers can complete the self-report section on their own. To our knowledge, the Fit-Frailty App is the only multidimensional tool which includes objective measures of cognitive and physical performance (including components of sarcopenia diagnostic criteria), in addition to other frailty markers including medical history, nutrition, daily function and psychosocial aspects. Further, our utilisation of the cumulative deficits frailty index methodology for scoring yields a continuous total score that quantifies the degree of frailty on a continuum rather than characterising frailty as either an “all or none” construct.19 42 For research purposes, a CSV file is generated that enables results to be easily uploaded into a database. The systematic identification of frailty and sarcopenia across the care spectrum (eg, primary care, homecare, acute care, rehabilitation) is important for several reasons including: (1) to identify patients who are at risk and could benefit from prevention to delay or prevent frailty and sarcopenia, (2) triage patients most in need of specialised geriatric services, (3) identify patients in need of prehabilitation prior to elective surgery (eg, hip or knee replacement), (4) assist with identifying risks in elective/emergent surgical patients for informed, shared decision making given high risk for poor postoperative outcomes in frail patients, (5) identify the urgency and level of intensity for rehabilitation services after hospitalisation, (6) assist with prognostication to realistically plan for future transitions and better anticipate needs and (7) support discharge planning to identify patients who require additional medical care, homecare and professional services to prevent readmission or institutionalisation. For clinical research settings, the cumulative deficits frailty index approach offers a powerful predictive ability (eg, mortality, falls, fractures, disability) based on its track record in population-based studies around the world.194346 Several limitations should be acknowledged. The Fit-Frailty App self-report items were based on a validated frailty index from a Canadian population-based sample19 ; however, this version did not include performance-based cognitive and physical measures. Our study methodology did not allow us to directly compare our adapted interactive cognitive screen with the paper-based version (eg, TMT A/B), although our physical performance measures were identical to well-validated measures of muscle strength and performance.6 7 Further, our study was situated within a “real world” clinical setting; as such, there were several different FI-CGA raters over time including staff geriatricians, internal, geriatrics or family practice residents. Given our clinical validation cohort was a convenience sample of patients who were referred to an outpatient geriatric medicine clinic, generalisability to other settings including rehabilitation, acute or primary care may be limited. Further validation studies need to be conducted in these settings. We were not able to examine inter-rater or test-retest reliability, as it would not have been feasible in this clinical setting. Our clinical validation study was based on the full interactive assessment; in some settings with more resource constraints (eg, lacking time, space or equipment), it may be necessary to skip some or all of the physical performance items, and the built-in proxy measures would be used instead. To conclude, the Fit-Frailty App is a feasible and valid tool that can be used in research and clinical settings to comprehensively assess frailty and sarcopenia by non-geriatricians. The use of systematic data collection to monitor patients’ health status and directly inform care decisions (referred to in the literature as “measurement-based care”) can not only guide practice but also result in enhanced patient outcomes.47 Our future studies include cross-validation of our tool in other settings and examining its ability to detect treatment response after rehabilitation interventions (currently underway by our group). Linking assessment results to evidence-based pathways for frailty interventions will be our next step in utilising this tool to its fullest capability. Given the lack of geriatricians available to manage frail patients in health systems across the world, accessible frailty management tools and care pathways that inform clinical decision-making for frail patients are critical.48 49 Supplementary material