This was a descriptive cross-sectional study that evaluated clinical, physiological, body composition, and psychometric variables in a sample of older adult participants. Recruitment and data collection took place between March and June of 2023. Participants were recruited through advertisements from the public Andalusian Health Service system, Spain. Participants were recruited from primary care centers through referrals by health care professionals. Recruitment was conducted in collaboration with general practitioners, who identified potential participants meeting the inclusion criteria. The study adhered to the ethical principles of the Declaration of Helsinki for medical research involving human subjects. All participants provided written informed consent. This study was approved by the Costa del Sol Institutional Ethics Committee, under protocol number BON22, on December 23, 2022. All participants provided written informed consent prior to their inclusion in the study. Participants’ privacy and confidentiality were ensured throughout the study: all data were anonymized prior to analysis, and identifying information was stored securely and separately from the study data. No identifiable images or information of participants are included in this publication. Participants did not receive compensation for their participation..

Inclusion criteria for the study were as follows:

Exclusion criteria were the following:

The aim of this study was to identify differences in measured parameters to classify individuals into different states of frailty within primary care. To identify different states of frailty, the sit-to-stand test was used, and relative power was calculated using the equations validated by Losa-Reyna et al [16]. Using the cutoff points determined by Losa-Reyna et al [22], participants were classified into 3 groups: frail, prefrail, and robust. This test was performed in the health center’s office.

After identifying the study subjects, a detailed evaluation of the remaining variables was carried out in the Functional Testing Laboratory of the Physical Education and Sports Area of the University of Málaga. Participants came to the laboratory fasting (a minimum of 3 hours) and without having done any previous exercise, wearing comfortable and light clothing without metal objects. The tests were carried out between 9:30 AM and 2:00 PM. Once in the laboratory, 2 qualified researchers administered the different tests and assessments; body composition determinations were performed fasting and in the early morning, and the rest of the scheduled evaluations were carried out after breakfast.

We collected the following clinical and demographic variables:

We collected the following body composition variables:

To assess cognitive and nutritional status, the following questionnaires were administered digitally using the Google Forms application.

To evaluate the physical function of the participants, the Short Physical Performance Battery (SPPB) was used, score 0 to 12 [26]. This battery includes three tests:

In addition to the tests included in the SPPB, two tests were performed:

The sample size was estimated using an ANOVA model for 3 independent groups (robust, prefrail, and frail). The calculation was performed using the G*Power software. For the sample size calculation, a power of 90% (1–β=.90), a significance level of 5% (α=.05), and an effect size of 0.4 were used. The result was a total sample size of 84 participants. This size was obtained with a noncentrality parameter (λ) of 13.44 and a critical F of 3.11, with 2 degrees of freedom in the numerator and 81 in the denominator. To avoid losses in the initial calculation, enough participants were recruited to maintain the robustness of the study. Finally, 94 subjects participated in the study, distributed across 3 groups: robust, prefrail, and frail.

To compare the differences on all variables between robust participants and those with frailty and prefrailty, a 1-way ANOVA was performed on the total sample. Logistic regression analysis was performed to investigate the relationship between body composition and frailty. WBPhA, thickness US, and DXA lean muscle from the lower limbs were entered into the regression model as independent variables, as they were found to have significant differences between the studied groups. Considering that age and sex may influence the relationship between body composition and frailty, these factors were introduced as a confounding variable into an adjusted regression model, treating age as a continuous covariate and sex as a categorical factor. The dependent variable was a binary indicator of frailty, coded as 1 for participants with frailty and 0 for robust participants and those with prefrailty.

To extract values for body composition to identify the presence of prefrailty or frailty, we conducted an analysis using the receiver operating characteristic (ROC) curve. This analysis was applied only to body composition that was significant in the logistic regression analysis. In the ROC analysis, the outcome variable was the presence or absence of frailty. The test variable was the body composition that was significantly associated with frailty. Youden index [29] was calculated with the following formula: Youden index = sensitivity + specificity – 1. The area under the curve (AUC), sensitivity, and specificity were calculated to evaluate the accuracy of the identified predictive models. The AUC could distinguish between nonpredictive (AUC<0.5), less predictive (0.5<AUC<0.7), moderately predictive (0.7<AUC<0.9), and highly predictive (0.9<AUC<1) values, as well as perfect prediction (AUC=1) [30].

This study provides significant evidence on the predictive capacity of body composition variables and their relationship with frailty syndrome in older adults. Technology tools such as BIA, WBPhA, US of the anterior rectus quadriceps muscle thickness, and DXA, were used in this study. The older adult frailty status was calculated using the PowerFrail mobile app [15] (Figure 1). This app is the first scientifically based app that allows the assessment of the muscle power and frailty of older adults in a simple way. Our results showed that these parameters are useful predictors for identifying frailty, in line with previous findings.

This study found several significant correlations between variables related to body composition, physical performance, and cognitive status in female participants, but these were not the same as the values observed in their male counterparts. This may be because women experience more functional limitations than men during the aging process [31]. Women tend to suffer a greater decline in physical function with age, which has been attributed to hormonal changes, particularly the decrease in estrogen levels during menopause [32]. Menopause is associated with changes in body composition characterized by an increase in body fat and a progressive decrease in muscle mass and strength [33]. These alterations can lead to a higher prevalence of sarcopenia and frailty in older women compared to men.

Based on these postmenopausal changes, muscle and fat composition might be more closely related to physical performance in women, which we observed in our correlation results from Figures 2-4. Specifically, the associations between body composition measures (evaluated through BIA, DXA, and US) and expressions of strength, such as lower limb power from the mobile app and handgrip strength, were more pronounced in women. This suggests that changes in muscle quality and quantity may have a greater impact on the physical function of older women.

Although previous research has indicated that muscle quality contributes to physical capacity in older adults, the effect of sex differences on this association has not been thoroughly investigated [34]. Some studies have suggested that muscle quality independently predicts physical function in older men but not in women [35]. However, our findings differ, as we observed significant correlations in women. Unlike previous studies, we included analyses of physical function status, which may explain the different results. The underlying mechanisms of these sex-specific differences still need to be investigated and clarified.

Contributing factors may include differences in muscle structure composition, hormonal influences, and neuromuscular activation patterns between men and women [36]. Additionally, the accuracy and sensitivity of measurement techniques, such as BIA and US, may vary between sexes due to differences in body fat distribution and hydration status [37]. Further research is needed to explore these factors and understand how they influence the relationship between body composition and physical function in older adults.

The connection between body composition and frailty indicators has been the focus of extensive research. However, to our knowledge, no studies have simultaneously compared DXA, BIA, and US, specifically for the identification of frailty syndrome in a primary care setting. Our study addresses this gap by evaluating the predictive capacities of these 3 modalities, offering a comprehensive analysis that can inform clinical practice regarding the most suitable and practical tools for frailty assessment in older adults. Although previous studies have compared two modalities, our inclusion of a third (US) provides a broader perspective on body composition assessment tools available for primary care settings. For instance, research has demonstrated that muscle US is an emerging tool for diagnosing sarcopenia, with studies summarizing its diagnostic accuracy [38]. Additionally, studies have evaluated the reliability and validity of sarcopenia diagnosis using BIA compared with the gold standard, DXA, assessing the predictive accuracy of BIA for diagnosis [39,40]. However, these studies did not include US in their comparisons. Rossini-Venturini et al [41] highlighted that the anthropometric prediction equations developed in their study provide a reliable, practical, and low-cost instrument to assess the components that change the most during the aging process, corroborating our findings. This perspective emphasizes the significance of considering diverse elements of body composition in the evaluation of health among older adults. Although some studies have recognized muscle mass, assessed via DXA or BIA, as a critical element in forecasting frailty [42,43], our analysis demonstrated that lower extremity lean mass (left left lean mass and LeanM RL) did reveal a direct relationship in the models. The present results are in line with findings that indicate muscle mass reduction alone does not suffice to predict frailty without factoring in physical performance [44]. In fact, physical performance, which can be measured through functional tests such as gait speed or grip strength, is vital for evaluating frailty status among older adults [45].

There has been a growing acknowledgment in recent literature regarding the significance of assessing both the quantity and quality of muscle. Xu et al [46] demonstrated that body composition, encompassing both muscle mass and quality, is significantly associated with frailty in older adult inpatients. The current literature suggests that it is not enough to analyze the quantity of muscle mass in absolute terms; quality is equally important. Quality can be assessed through methods such as US and BIA, which provide insights into muscle integrity and performance in terms of its functional capacity. In this context, analyzing muscle quality through US could effectively complement DXA, which has traditionally been considered the gold standard in the assessment of body composition. Although DXA provides valuable information on lean mass and fat mass, it does not provide details on muscle distribution and quality, critical for understanding frailty in older adults [47].

Moreover, our findings indicated that muscle strength and gait speed are important indicators of frailty, corroborating the work of Tsukasaki et al [48], who found a strong association between muscle strength, gait speed, and cross-sectional muscle area determined by midthigh computed tomography. These findings reinforce the idea that comprehensive evaluations of muscle function should be integrated into frailty assessments. Therefore, muscle composition emerges as a potential public health assessment by enabling the clinical quantification of muscle mass and an estimation of physical function in the older adult population [49].

Our findings demonstrated that all measures exhibited high specificity but varying sensitivity in detecting frailty (Table 3). Our ROC analysis showed moderate predictive ability for WBPhA, US thickness, and lean mass of legs from DXA, with an AUC between 0.678 and 0.749. This aligns with the results from previous studies [50,51], which emphasized that models combining body composition measurements with physical performance tests can improve the predictive ability of frailty. Unadjusted US showed perfect specificity but low sensitivity (10.7%), indicating it is highly effective at correctly identifying nonfrail individuals but less capable of detecting those who are frail. This aligns with previous studies suggesting that muscle US, while precise in measuring muscle thickness, may have limitations in sensitivity due to operator dependency and variability [52].

Adjusting the US measure for age and sex increased sensitivity to 32.1% but reduced specificity to 93.9%, and the adjusted model did not reach statistical significance (P=.24). This suggests that while adjustments improve sensitivity, they may not be sufficient to make US a standalone diagnostic tool for frailty in primary care settings. The WBPhA showed an unadjusted sensitivity of 24% and specificity of 95.2%, with an AUC of 0.704 (P=.003). After adjusting for age and sex, sensitivity improved to 48%, specificity slightly decreased to 93.7%, and AUC increased to 0.762, although the P value was nonsignificant (P=.09). These results suggest that WBPhA, particularly when adjusted for sex and age factors, has potential as a screening tool for frailty. Previous research has shown that lower phase angle values are associated with decreased muscle function, poor nutritional status, and higher frailty risk [12,53]. The noninvasive nature and ease of use of BIA make WBPhA a practical option for primary care, although standardization of measurement protocols is necessary.

LeanM RL emerged as a significant predictor when adjusted for age and sex (P=.045), with sensitivity and specificity of 35.7% and 93.9%, respectively, and an AUC of 0.749. This indicates that the adjusted LeanM RL model had a good balance between sensitivity and specificity and may be valuable in predicting frailty. The importance of lower limb muscle mass in frailty assessment is well-documented. For instance, [54] reported that decreased appendicular lean mass is associated with physical disability and increased risk of adverse outcomes in older adults. However, the use of DXA in primary care is limited due to cost, accessibility, and exposure to low-dose radiation.

In comparison, the unadjusted left leg lean mass showed similar specificity (98.5%) but low sensitivity (10.7%), and the adjusted model did not achieve statistical significance. This suggests that while lean mass measurements are informative, the right leg may provide more predictive value than the left in this context, possibly due to dominance or functional differences, although further research is needed to confirm this observation. Overall, our results indicated that LeanM RL, adjusted for age and sex, may be the most effective measure among those studied for predicting frailty. This is significant because identifying reliable, accessible markers for frailty is crucial for early intervention. Given the limitations of DXA in primary care, exploring alternative methods to estimate lean mass, such as predictive equations or portable devices, could enhance feasibility. The high specificity observed across all measures suggests they are effective in ruling out frailty in individuals without frailty. However, the variable sensitivity underscores the need for multicomponent assessment tools. The Comprehensive Geriatric Assessment remains the gold standard for frailty evaluation but is resource-intensive [55]. Incorporating measures like WBPhA and simplified lean mass assessments could enhance screening efficiency in primary care.

Limitations of the study include the sample size and homogeneity in some variables, such as BMI and arm lean mass (left arm and right arm), which could have affected the accuracy of the predictive results. Heterogeneity in these measurements has been reported in previous research [56], suggesting that variability in body composition could influence frailty prediction. There is potential for the integration of advanced body composition analysis tools and mobile technology into primary care services for the early identification and management of frailty syndrome in older adults. Methods such as WBPhA, US of the rectus femoris muscle, and DXA analysis proved to be effective predictors of frailty, offering high specificity in detection. The implementation of the PowerFrail mobile app enables rapid and personalized assessments, optimizing preventive interventions. These tools not only improve the accuracy of clinical evaluations but also reduce costs associated with frailty-related complications, promoting healthier aging and alleviating the burden on public health care systems. In addition, while this study incorporated a range of assessment tools, it is essential for future research to concentrate on validating these measures in larger and more diverse cohorts to ensure their wider applicability. The cross-sectional design of this study further restricts the ability to infer causal relationships between body composition variables and frailty. Longitudinal investigations could provide valuable information on how variations in body composition affect the progression of frailty over time. Understanding these dynamics could facilitate developing more effective and personalized interventions for the older adult population. Finally, future studies should explore the integration of these mHealth technologies with advanced body composition analysis systems to optimize early detection and management of frailty.

The findings of this study reinforce the utility of various body composition evaluations, such as WBPhA, LeanM RL measured by DXA, and quadriceps thickness assessed by US, as effective indicators for predicting frailty in older adults, aligning with previous research. However, our results highlight the necessity of not relying exclusively on muscle mass as a predictor of frailty. It is essential to incorporate assessments of muscle function and physical performance into clinical evaluations to enhance the accuracy of identifying individuals susceptible to frailty. The combination of tools such as WBPhA, bioimpedance, and US, along with the PowerFrail app, could provide a more complete and accurate assessment of the health status of older adults, allowing effective preventive interventions to be implemented in primary care services.