This study explored the associations between US-MT of VL at hospital admission and levels and trajectories of physical performance over time, as well as selected short-term clinical outcomes, in acutely hospitalized older patients with admission-related walking limitation. We found that US-MT was associated with GS in women, but not in men, and with 30 s STS performance across sexes. Importantly, US-MT was not associated with recovery trajectories over time but instead reflected differences in physical performance level that persisted across the recovery period. No associations were identified between US-MT and clinical outcomes including LOS, 3-month falls incidence, and 3-month mortality.

The association between US-MT and GS requires careful interpretation. Among women, greater US-MT was associated with faster GS, but the effect was modest: each 1-mm increase in US-MT corresponded to a 0.037 m/s higher GS. In men, the effect was smaller (0.011 m/s per 1 mm) and not statistically significant. Both estimates fall below commonly suggested thresholds for clinically meaningful changes in GS (≈0.1 m/s)[29]. This indicates that although US-MT may reflect aspects of lower-limb muscle size relevant to walking, its standalone association with GS is limited in magnitude. A similar pattern was observed for 30 s STS performance, where greater US-MT was associated with a 24% increase in the number of repetitions, with no evidence of sex-specific differences. However, the wide confidence intervals indicate limited precision, suggesting that the strength of this association should be interpreted with caution. Importantly, no interaction between US-MT and time was observed for either GS or 30 s STS. This suggests that baseline US-MT was associated with physical performance level, but not with the trajectory of recovery. The observed improvement in GS and 30 s STS from admission to follow-up is clinically plausible, as admission assessments were performed during the acute phase of illness, when mobility may have been transiently reduced by symptoms, inflammation, fatigue, bed rest, or treatment-related factors. Later assessments may therefore reflect partial recovery toward pre-hospital functional capacity. These findings should be interpreted in the context of the inherently multifactorial nature of GS and STS. Both outcomes depend on a complex interplay of physiological and contextual factors, including balance, coordination, motivation, fear of falling, pain, cognitive function, chronic disease burden, and the severity of acute illness. In acutely hospitalized older adults, these factors may overshadow the contribution of muscle morphology to objectively measured physical performance. This may explain why US-MT demonstrated only modest associations with GS and STS, despite its biomechanical relevance. This interpretation is consistent with previous literature showing that US-MT is more strongly associated with direct measures of muscle strength and power than with performance-based outcomes. Several studies have demonstrated consistent associations between quadriceps muscle thickness and knee extension strength in healthy populations [17, 18, 35]. Similarly, Yuan et al. reported that US-MT correlates more strongly with strength and power than with GS and STS and found weak associations with these performance-based outcomes, closely aligning with the present findings [36]. Nevertheless, GS and STS remain clinically relevant, ecologically valid measures of functional capacity, and even modest associations with US-MT may therefore hold clinical value.

The observed sex-specific association between US-MT and GS warrants further consideration. In the present study, US-MT was associated with GS in women but not in men. This finding aligns with recent work by Yuan et al., who also reported sex-specific associations between US-MT and measures of muscle strength and physical performance in older adults [25]. While biological explanations may contribute, including differences in muscle mass distribution and muscle quality, the findings may also reflect differences in baseline functional status between men and women in the current study cohort.

Functional performance in acutely hospitalized older adults is influenced not only by muscle morphology but also by broader factors such as frailty, comorbidity, and functional independence. Differences in these domains between sexes may influence the extent to which muscle thickness contributes to observable performance. In this context, muscle thickness may play a relatively larger role in determining performance in individuals with lower overall functional reserve, which could partly explain the stronger association between US-MT and GS observed in women. These findings indicate that it could be important to consider sex-specific mechanisms when interpreting muscle morphology-function relationships in heterogeneous clinical populations.

The finding that admission US-MT was not associated with any of the clinical outcomes suggests that a one-dimensional measure of local muscle size (e.g., muscle thickness) does not adequately capture the multifactorial nature of these outcomes. Importantly, these null findings should not be interpreted as evidence that muscle size more broadly is unrelated to clinical outcomes. Muscle thickness was selected as a clinically feasible one-dimensional measure, whereas cross-sectional area and muscle volume may better capture muscle size but are more technically demanding and may be limited by the transducer field of view. Furthermore, LOS is influenced by the acute medical course, discharge planning, social circumstances, and organizational factors, which may attenuate associations. Although rectus femoris muscle thickness has previously been associated with LOS in more homogeneous cohorts [37], such relationships may be less apparent in heterogeneous acutely ill older patients. Falls and mortality are driven by factors beyond local muscle morphology, and the clinical outcome analyses were limited by a few observations and events, particularly for falls and 3-month mortality. The absence of statistically significant associations should therefore be interpreted as inconclusive rather than as evidence of no prognostic relevance. Overall, our findings suggest that admission to US-MT alone has limited prognostic utility for clinical outcomes in the acute care of older patients.

This study has several strengths. It included a relatively large and well-characterized cohort of acutely hospitalized older patients. Ultrasound assessments were performed using a standardized and previously validated protocol by a single experienced operator, supported by excellent intra-rater reliability. Functional outcomes were assessed using consistent procedures by trained clinicians, thereby minimizing measurement error. The focus on the vastus lateralis (VL) muscle—a muscle central to functional tasks such as walking and sit-to-stand—enhances the clinical relevance of the findings. Furthermore, the prospective design with repeated assessments enabled longitudinal evaluation, and the use of mixed-effects modeling allowed inclusion of all available data while appropriately accounting for incomplete follow-up.

Several limitations should also be considered. First, US-MT represents a relatively narrow, one-dimensional measure of local muscle size and does not capture cross-sectional area, whole-muscle volume, or qualitative aspects of muscle composition. Although muscle thickness is clinically feasible, cross-sectional area and muscle volume may provide more robust indicators of muscle size. Moreover, US-MT was assessed only at admission and therefore does not describe changes during hospitalization or the following recovery period. Second, the assessment was limited to the VL muscle, which may restrict generalizability to other muscle groups. Third, attrition at later follow-up time points reduced the precision of the estimates, although sensitivity analyses suggested limited evidence of systematic bias. Regarding GS measurement, although the 4-m GS test was the standard protocol, a 3-m version was used when the standard course was not feasible. Despite strong correlations between short- and longer-distance GS tests, this pragmatic adaptation may have introduced measurement variability. In addition, assigning 0 m/s to participants unable to walk may have introduced a floor effect, although sensitivity analyses excluding these observations did not materially alter the findings. Fourth, the cohort represents a relatively frail and heterogeneous subgroup of older patients, which may limit generalizability to healthier and more homogeneous populations. Nevertheless, the cohort broadly resembles the Danish geriatric inpatient population described previously [38]. Fifth, residual confounding from unmeasured factors cannot be excluded. Sixth, the observational nature of the analyses precludes causal inference, and the modest effect sizes warrant cautious interpretation. In addition, the study was not specifically powered for the present exploratory clinical outcome analyses, and the limited number of available observations and events—particularly for falls and 3-month mortality—reduced statistical precision and increased the risk of type II error. The absence of a formal comorbidity index (e.g., Charlson Comorbidity Index) further limits the ability to fully account for illness burden. Several participants were admitted with conditions in which altered hydration may occur. This is particularly relevant because conditions associated with altered hydration or fluid shifts may have influenced both US-MT and the outcomes. Edema or fluid shifts may affect apparent muscle thickness independently of contractile tissue, whereas dehydration, cardiovascular or renal dysfunction, and electrolyte disturbances may affect physical performance and clinical outcomes. Thus, residual confounding and measurement influence related to fluid status cannot be ruled out. Finally, embedding the study within an RCT introduces potential selection bias, as patients with severe dementia, pre-hospital non-ambulatory status, and fully independent walking at inclusion were excluded. Selection bias may also have occurred because ultrasound assessment depended on the availability of a single trained operator, which was chosen to minimize between-operator variation but limited continuous assessment of all enrolled patients. Although primarily logistical, this may have contributed to the selection of participants who were easier to assess. Furthermore, patients unable to tolerate ultrasound acquisition or complete gait speed testing may represent a frailer subgroup with greater illness burden and poorer outcomes. The analytical cohort may therefore have been relatively healthier than the broader target population, limiting generalizability. Substantial attrition at 1- and 3-month follow-up further reduced precision and may limit the generalizability of the longitudinal estimates.

Future studies should adopt multidimensional ultrasound protocols including measures such as cross-sectional area, muscle volume, echo intensity, shear-wave elastography, and measures more closely related to muscle strength and power. These may provide a more complete characterization of muscle health and improve the prediction of functional and clinical outcomes. Integrating ultrasound data with assessments of strength, cognition, functional status, and inflammatory markers may improve characterization of functional status and recovery potential in acute care. Studies with extended follow-up could clarify the longer-term implications for clinical outcomes such as LOS, falls, and mortality.