Muscle mass in critically ill patients: analysis using ultrasound, traditional bioelectrical impedance analysis, and wearable devices: a prospective observational study

Abstract

Background

Accurate muscle mass assessment is essential in critical care, but no consensus exists on the optimal method because of invasiveness and limited feasibility in immobile patients. We evaluated the effectiveness of non-invasive and easily applicable ultrasound, traditional bioelectrical impedance analysis (BIA), and wearable technology in this context.

Methods

We recruited patients who stayed in the intensive care unit (ICU) for at least 7 days. Muscle mass was assessed at three time points: within 48 h of admission and on days 7 and 14 after admission.

Results

We analyzed the data of 53 patients (age: 71.8 ± 11.3 years; length of ICU stay: 10 days). Their muscle mass significantly declined during their ICU stay. The rectus femoris (RF) muscle thickness declined by day 7 and declined further by day 14 (89.7% of initial value). The total anterior thigh muscle thickness and cross-sectional area (CSA) of the RF showed a similar trend. The traditional BIA device reflected a decrease in skeletal muscle mass to 96.2% and 91.8% on day 7 and 14, respectively, while wearable device assessment showed a modest decline to 95% and 89.2% on day 7 and 14, respectively. The groups experiencing weaning failure, death, or discharge to a facility other than home exhibited a higher percentage decline in muscle mass compared with the groups experiencing weaning success, survival, or discharge to home, respectively.

Conclusions

Ultrasound is a reliable method for monitoring muscle atrophy in critically ill patients, demonstrating a consistent decrease in muscle mass. The outcomes of traditional BIA and wearable devices suggest that careful consideration of their interpretations is necessary. Utilizing these tools may be valuable for assessing patient prognosis and facilitating more proactive interventions for patients experiencing muscle loss.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-026-03888-x.

Background

Intensive care unit (ICU)-acquired weakness, arising not only from neuromuscular dysfunction but also from the progressive loss of skeletal muscle mass and function [1], is emerging as a common and serious complication in critically ill patients and is strongly associated with poor clinical outcomes [2–4], including prolonged hospital stays and increased morbidity and mortality. The rapid detection and quantification of muscle wasting in critical care settings are essential to stratify risk, tailor nutritional and rehabilitation interventions, and ultimately improve patient outcomes [5]. Traditional methods for assessing muscle mass and strength, such as computed tomography (CT), magnetic resonance imaging (MRI), dual-energy X-ray absorptiometry (DXA), muscle biopsies [6, 7] and various force measurements tools [8], are accurate and valuable but often challenging in ICU patients owing to cost, immobility, invasiveness, and vital sign instability.

These limitations have led to growing interest in alternative bedside methods for assessing muscle mass in critically ill patients. Ultrasound [9–11], the traditional bioelectrical impedance analysis (BIA) device [12], and other less invasive options [13] offer promising continuous and dynamic insights into muscle status with minimal patient burden. More recently, wearable devices [14, 15] have been proposed to measure body mass index and physical activity. Despite a variety of available methods, the relative reliability and clinical correlations of these techniques in critical care settings are poorly understood.

In this study, we investigated changes in muscle mass among critically ill patients using ultrasound, a traditional BIA device, and a wearable device. Our aim was to address a gap in the literature by systematically comparing the performance of these methods and assessing their clinical applicability.

Methods

Study design and population

We conducted a prospective single-center study in the ICU of a university-affiliated hospital between February 2022 and November 2023. The study protocol was approved by the Clinical Research Ethics Committee of the Chungnam National University Hospital (Approval Number: 2021-11-062). Patients were eligible if they were ≥ 18 years of age and admitted to the ICU with informed consent obtained from themselves or a legal representative. We excluded patients who were expected to die within 72 h of ICU admission, those with conditions preventing reliable muscle assessment (e.g., limb amputation or severe skin injury), or those without available consent. The research director and co-researcher performed ultrasound at day 3 (within 48 h), 7, and 14 after admission to the ICU; at the same time, muscle mass was measured using the traditional BIA device and a wearable device. Almost all patients admitted to our ICU received rehabilitation therapy. The 53 patients enrolled in the study received bedside physical therapy, primarily limited to passive or active range-of-motion exercises.

Ultrasonographic measurement

Ultrasound is a reliable tool for measuring muscle thickness, but it is influenced by the operator’s skill, probe pressure, and patient positioning. To minimize this variability and errors, two operators underwent training for muscle mass measurements three times a week for 1 year before the study. All measurements were conducted using the same ultrasound machine (VIVID GE E UltraEdition with B-mode imaging connected to a 6.5 MHz 3.8 cm linear transducer [GE L3-12-D]; GE Healthcare, Chicago, IL, USA). In addition, to minimize time-related variables, measurements were conducted between 10 AM and 12 PM.

All scans were performed with patients in the supine position, with their elbows and knees in passive extension. The operators were instructed to perform the ultrasound with minimal hand pressure, and the probe was positioned perpendicular (90°) to the skin surface to avoid obliqueness or angulation. Generous amounts of contact gel were applied to prevent muscle compression caused by the transducer. To minimize measurement errors, the ultrasound was conducted at precisely defined anatomical locations on the anterior surface of the quadriceps in both legs from: (1) the midpoint between the anterior superior iliac spine (ASIS) and the upper pole of the patella, and (2) the boundary between the lower third and the upper two-thirds of the distance between the ASIS and the upper pole of the patella. Four points on both quadriceps were identified and marked with an indelible pen.

Serial ultrasound measurements for each patient were performed by the same operator to ensure intra-rater consistency. Two trained operators participated in the study, with each operator conducting all measurements for their assigned patients throughout the study period. Both operators completed standardized training and joint practice sessions to minimize inter-operator variability and ensure consistent measurement techniques. In addition, during the procedure, another trained operator verified the positioning and measurement to ensure accuracy. Furthermore, also to ensure accuracy, ultrasound images were independently re-evaluated by two additional co-researchers who were blinded to the patients' information. After these procedures, the following measurements of muscle size and echogenicity were obtained from both legs at the same locations: rectus femoris (RF) thickness, RF echogenicity, RF cross-sectional area (CSA), vastus intermedius thickness, and total anterior thigh muscle thickness. Anterior thigh muscle thickness was defined as the distance between the anterior fascia of the RF muscle and the posterior fascia of the vastus intermedius muscle at the axial aspect of the image. The mean values for each parameter were calculated based on the measurement of four points (Fig. 1).

Fig. 1.

Ultrasound measurement points and view. (A) Measurement sites at mid-thigh (1/2) and distal third (2/3) between the anterior superior iliac spine (ASIS) and patella. (B) Representative transverse ultrasound image showing rectus femoris (a), vastus intermedius (b), total anterior thigh muscle thickness (c), and rectus femoris cross-sectional area (d, yellow outline).

Bioelectrical impedance analysis (BIA)

BIA is a non-invasive technique that is used to quantitatively measure body composition based on impedance, which is generated when a small electrical current passes through the body. This principle relies on the fact that the human body, composed predominantly of water that conducts electricity well, exhibits varying levels of resistance to electrical current based on the amount of water present. By applying a small alternating current to the body and measuring the resulting impedance index, BIA is used to estimate body composition parameters such as fat, water, bone mass, lean body mass, and muscle.

Impedance refers to the total resistance encountered by the alternating current as it passes through the body, encompassing both resistance and reactance. Resistance indicates the extent to which the current flow is impeded; for instance, fat, which contains minimal water, offers higher resistance, while muscle, which contains a significant amount of water, allows current to pass more easily. Thus, individuals with higher water or muscle content exhibit lower resistance, while those with lower water content show higher resistance. Reactance is a component of impedance that arises when the alternating current encounters the cell membranes. To traverse the cell membrane, the current must pass through it, creating reactance, which is distinct from the resistance encountered by the current flowing through extracellular and intracellular water.

The InBody S10 analyzer (Biospace, Seoul, Korea)—as a traditional BIA device—and a wearable device (Galaxy Watch 4, Samsung, Seoul, Korea) utilize these principles of BIA to measure skeletal muscle mass.

The InBody S10, which requires the placement of electrodes on both the upper and lower limbs, has the advantage of measuring muscle mass in specific body regions (right arm, left arm, torso, right leg, and left leg) as well as overall skeletal muscle mass. The InBody S10 utilizes multi-frequency BIA technology and employs a 4-pole, 8-point detachable electrode method to measure impedance at five locations (right arm, left arm, torso, right leg, and left leg) across six frequencies (1, 5, 50, 250, 500, 1000 kHz). In this study, we focused primarily on the overall skeletal muscle mass. According to the device’s manual, measurements can vary slightly depending on factors such as posture, meal intake, and the time of measurement. To enhance reliability, we standardized measurement conditions for all patients as closely as possible. Measurements were taken between 10 AM and 12 PM, with patients resting in a supine position for at least 10 min before the assessment. Electrodes were then placed on both thumbs, the third fingers, and just below the ankles of both feet.

The Galaxy Watch 4 automatically measures and displays skeletal muscle mass by having the participant place the middle and fourth fingers of the right hand on two buttons after wearing the watch on the left wrist, as per the device's configuration and manufacturer’s recommendations. To ensure accurate measurements, patients were instructed to use only their right middle and ring fingers for the assessment. The wrist strap was tightened to ensure complete electrode contact with the skin.

BIA measurements using the Galaxy Watch 4 were performed at the same time (10 AM to 12 PM) and in the supine position after 10 min of rest as were those conducted with the InBody S10. The Galaxy Watch 4 employs a single 50 kHz electrical current that passes through the fat-free mass (FFM) in the upper body from the left to the right arm. It utilizes four electrodes: two (made of chromium silicone carbon nitride) located on the back of the watch in contact with the wrist, and two (made of stainless steel) positioned on the side buttons for contact with the right fingers. This setup creates a circuit for conducting the 50 kHz electrical current to measure impedance in the upper body. The impedance measurements are based on the electrical conductivity of the FFM within the measurement loop. The tetrapolar electrode configuration, with separate source electrodes placed outside the detector electrodes, minimizes the effects of higher current density at the electrodes and ensures a uniform current distribution.

Outcomes

The primary outcome was the time course of muscle wasting as assessed using ultrasound, the traditional BIA device, and the wearable device. To assess changes in muscle mass, we calculated the rate of change in muscle mass from baseline and the rate of atrophy, defined as a percentage of the baseline. The secondary outcome was the comparison of muscle mass assessments between ultrasound, BIA, and the wearable device, which were performed by determining the correlation between each test. In addition, we aimed to determine the clinical significance of various measurement values in critically ill patients, specifically concerning ventilator weaning, survival, and discharge to home. For this study, ventilator weaning failure was defined as failure of the first extubation attempt requiring reintubation or tracheostomy, patients whose clinical condition precluded extubation, or those who remained intubated until death. The death group included patients who were extubated in the ICU but subsequently died either in the ICU or on the general ward. We also compared the degree of muscle loss between patients who required mechanical ventilation and those who did not, as well as between patients who received mechanical ventilation for ≥ 7 days and those who required shorter durations (< 7 days).

Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics, version 22, for Windows (IBM Corp., Armonk, NY, USA). All values are expressed as mean ± standard deviation or median and interquartile range (IQR) for continuous variables. Categorical variables are expressed as numbers and percentages. Paired t-tests for continuous data and Wilcoxon signed-rank tests for nonparametric data were used to examine the changes over time. A two-tailed p value ≤ 0.05 was considered statistically significant. Depending on the normality of the data, the Pearson r or Spearman rho was used to assess the relationship between the ultrasound and physical function measures. Coefficients were interpreted as weak (0.00–0.25), mild (0.25–0.50), moderate (0.50–0.75), and strong (≥ 0.75) associations. Figures were generated using Prism 9.0 (GraphPad software, Boston, USA) and SPSS.

Results

Patients’ baseline characteristics

We enrolled 53 patients and analyzed their data. The baseline characteristics of the enrolled patients are shown in Table 1. Their mean age was 71.8 ± 11.3 years, 35 (66.0%) were male, and their mean body mass index was 22.7 ± 4.6 kg/m². Regarding the severity and frailty of the patients, the mean Acute Physiologic and Chronic Health Evaluation (APACHE) II score was 20.7 ± 7.2, Clinical Frailty Scale (CFS) score was 3.9 ± 1.4, and Sequential Organ Failure Assessment score was 5.9 ± 3.1. The most common comorbidities were hypertension, diabetes, solid tumors, and chronic obstructive pulmonary disease in 31 (58.5%), 21 (39.6%), 12 (22.6%), and 12 (22.6%) patients, respectively. The laboratory results of the enrolled patients are shown in Table 2. At ICU admission, most patients had hypoalbuminemia (mean 3.1 ± 0.6 g/dL, 84.9% below the normal range) and elevated inflammatory markers (median CRP 7.25 mg/dL, 77.4% above normal; median procalcitonin 0.32 ng/mL, 71.7% above normal). In addition, anemia was common, with 75.5% of patients having hemoglobin levels below the normal range.

Table 1.

Baseline characteristics of enrolled patients

Characteristics	All patients
Age (years)	71.8 ± 11.3
Male	35 (66.0)
Body mass index (kg/m2)	22.7 ± 4.6
APACHE II score	20.7 ± 7.2
Clinical frailty scale	3.9 ± 1.4
SOFA score	5.9 ± 3.1
mNUTRIC score	5.3 ± 1.9
SARC-F score	3.1 ± 2.9
Katz ADL score	3.6 ± 2.4
Comorbidities
Hypertension	31 (58.5)
Diabetes	21 (39.6)
Cerebral infarction	2 (3.8)
Heart failure	11 (20.8)
Liver cirrhosis	5 (9.4)
Solid tumor	12 (22.6)
Hematologic malignancy	1 (1.9)
COPD	12 (22.6)
Dementia	3 (5.7)

Data are presented as mean ± standard deviation or n (%) unless otherwise noted

APACHE: Acute Physiologic and Chronic Health Evaluation; SOFA: Sequential Organ Failure Assessment; mNUTRIC: Modified Nutrition Risk in Critically Ill; SARC-F: Strength, assistance with walking, rising from a chair, climbing stairs, and falls; Katz ADL: Katz Index of Independence in Activities of Daily Living; COPD: Chronic obstructive pulmonary disease

Table 2.

Laboratory findings of enrolled patients

Characteristics	All patients (N = 53)	Normal value
Hemoglobin, g/dL	11.9 (9.7–13.5)	13.5–17.5
Albumin, g/dL	3.13 ± 0.60	4.0–5.0
BUN, mg/dL	21.4 (16.0–40.0)	8–20
Creatinine, mg/dL	0.82 (0.62–1.56)	0.6–1.1
Na, mEq/L	137.1 ± 5.9	135–150
CRP, mg/dL	7.25 (1.38–8.00)	0–0.5
Procalcitonin, ng/mL	0.32 (0.05–2.36)	0–0.05
Lactic acid, mEq/L	1.90 (1.30–3.05)	0.7–2.1

Data are presented as mean ± standard deviation or n (%) unless otherwise noted

BUN: Blood urea nitrogen; NA: Sodium; CRP: C-reactive protein,

The patient prognoses are shown in Table 3. The median ICU length of stay (LOS) was 10.0 days (IQR 7.0–16.0 days), while the median hospital LOS was 19.0 days (IQR 12.5–34.5 days). Among the patients, 42 (77.8%) required mechanical ventilation (MV). The median duration of MV for these patients was 7 days (IQR 4.0–10.0 days). Among all the patients, ventilator weaning was attempted in 32 (60.4%); however, it was unsuccessful in 15 (28.3%) of these patients. At discharge, 30 patients (56.6%) survived and went home, 8 (15.1%) were transferred to a nursing facility, and 15 (28.3%) died.

Table 3.

Prognosis of patients

Characteristics	All patients (N = 53)
ICU LOS (days)	10.0 (7.0–16.0)
In-hospital LOS (days)	19.0 (12.5–34.5)
Mechanical ventilator use	42 (77.8)
Mechanical ventilation (days)	7.0 (4.0–10.0)
Ventilator weaning try	32 (60.4)
Ventilator weaning fail	15 (28.3)
State of discharge
Home	30 (56.6)
Nursing care center	8 (15.1)
Death	15 (28.3)

Data are presented as median and interquartile range or n (%) unless otherwise noted

ICU: Intensive care unit, LOS: Length of stay

Serial change in muscle mass in critically ill patients

We observed serial changes in muscle mass over a 14-day period (Table 4). Initially, RF thickness decreased slightly from 1.11 cm² to 1.03 cm² at day 7 (p = 0.003) and further to 1.01 cm² at day 14 (p = 0.020). Similarly, the total anterior thigh muscle thickness showed a significant reduction from 2.99 cm to 2.78 cm at day 7 (p = 0.006), with no additional change at day 14 (p = 0.009). The RF CSA decreased from 4.93 cm² to 4.36 cm² at day 7 (p = 0.006) and showed a similar pattern of reduction, reaching 4.48 cm² at day 14 (p = 0.002). However, the changes in echogenicity were not statistically significant at any time point. Skeletal muscle mass as measured using traditional BIA decreased significantly from 24.20 kg to 23.00 kg at day 7 (p = 0.012), with a non-significant rebound to 23.40 kg at day 14 (p = 0.298). The muscle mass measurements obtained from the wearable devices showed a non-significant decrease from 24.35 kg at baseline to 23.60 kg at day 7 (p = 0.217) and to 23.25 kg at day 14 (p = 0.119).

Table 4.

Serial change of muscle quantification

	Initial	7 days	p value	14 days	p value
Rectus femoris muscle thickness (cm)	1.11 (0.87–1.44)	1.03 (0.82–1.31)	0.003*	1.01 (0.72–1.32)	0.020*
Total anterior thigh muscle thickness (cm)	2.99 (2.33–3.53)	2.78 (2.14–3.37)	0.006*	2.78 (2.04–3.79)	0.009*
Cross sectional area (rectus femoris, cm2)	4.93 (3.95–6.19)	4.36 (3.53–5.04)	0.006*	4.48 (3.35–5.60)	0.002*
Echogenicity, dB	45.98 (41.13–49.65)	44.21 (40.14–48.69)	0.249	44.68 (42.15–47.40)	0.155
Inbody–skeletal muscle mass (kg)	24.20 (20.05–28.95)	23.00 (19.30–25.80)	0.012*	23.40 (21.63–25.83)	0.298
Wearable devices–muscle mass (kg)	24.35 (19.48–27.83)	23.60 (19.30–26.10)	0.217	23.25 (20.50–25.35)	0.119

Data are presented as median and interquartile range unless otherwise noted

Table 5 (Fig. 2) shows the decrease in muscle percentage over a 14-day period, with the initial values set at 100%. By day 7, the percentage of RF muscle mass had decreased by 12.1% (p = 0.015). However, by day 14, the percentage decrease was 9.7% (p = 0.022), which was less pronounced compared to that by day 7. The total anterior thigh muscle thickness decreased by 8.1% by day 7 (p = 0.035) and continued to decrease to 10.39% by day 14 (p = 0.005). The CSA of the RF showed a similar pattern, with a 9.9% decrease by day 7 (p < 0.001) and a 16.04% decrease by day 14 (p = 0.001). Changes in echogenicity were not significant on day 7 (p = 0.377), with a slight but non-significant increase on day 14 (p = 0.154). The BIA-measured skeletal muscle mass decreased by 3.8% by day 7 (p = 0.043) and continued to decrease to 8.2% by day 14 (p = 0.239). Muscle mass measured using the wearable device showed a non-significant decrease of 4.5% by day 7 (p = 0.684) and a trend toward a decrease of 10.8% by day 14 (p = 0.070).

Table 5.

Serial change of muscle percentage

	Initial	7 days	%change (D7–D3)	p value	14 days	%change (D14–D3)	p value
Rectus femoris muscle thickness (%)	100	87.9 (83.0–102.7)	− 12.1 (− 17.0–2.7)	0.015*	89.7 (75.3–98.4)	− 9.7 (− 23.2–− 0.5)	0.022*
Total anterior thigh muscle thickness (%)	100	91.9 (85.4–102.1)	− 8.1 (− 14.6–2.1)	0.035*	89.2 (80.9–100.3)	− 10.39 (− 20.6–0.6)	0.005*
Cross sectional area (rectus femoris, %)	100	90.1 (82.7–98.5)	− 9.9 (− 17.3–− 1.5)	< 0.001*	85.2 (75.2–95.4)	− 16.04 (− 24.9–− 6.3)	0.001*
Echogenicity (%)	100	98.9 (95.2–101.6)	− 1.1 (− 4.8–1.6)	0.377	102.0 (99.1–103.8)	2.3 (− 1.2–4.8)	0.154
Inbody–skeletal muscle mass (%)	100	96.2 (93.8–102.2)	− 3.8 (− 6.3–2.4)	0.043*	91.8 (85.9–97.9)	− 8.2 (− 14.0–− 2.1)	0.239
Wearable devices–muscle mass (%)	100	95.5 (92.4–102.9)	− 4.5 (− 7.6–2.9)	0.684	89.2 (84.3–94.2)	− 10.8 (− 15.7–− 5.7)	0.070

Data are presented as median and interquartile range unless otherwise noted

Fig. 2.

Changes in muscle mass (%), as assessed by ultrasound, traditional bioelectrical impedance analysis (BIA) (InBody), and a wearable device, over time. Ultrasound measurements include rectus femoris thickness, total anterior thigh thickness, and cross-sectional area. Data are presented as the mean ± standard deviation. The reduction in sample size at Day 14 reflects patient discharge or death

Relationship between ultrasound, BIA, and wearable device in muscle mass evaluation

In Fig. 3, we present a comparative analysis, represented by scatter plots, showing the relationship between muscle mass assessed using ultrasound (CSA), the traditional BIA device, and the wearable device on days 3, 7, and 14. On day 3, we observed a moderate positive correlation between CSA and BIA measurements (r = 0.515, p < 0.001), a weaker but positive correlation between CSA and the wearable device measurements (r = 0.417, p = 0.004), and a strong positive correlation between BIA and the wearable device measurements (r = 0.827, p < 0.001). As day 7 progressed, the correlation between the CSA and BIA measurements became more pronounced (r = 0.645, p < 0.001), as did the association between the CSA and the wearable device measurements (r = 0.569, p < 0.001). Notably, the correlation between the BIA and wearable device measurements was strong (r = 0.906, p < 0.001). By day 14, these relationships maintained their positive trends, with CSA and BIA measurements showing a correlation of r = 0.631 (p = 0.028), and the wearable device measurements showing a slightly stronger correlation with CSA measurements at r = 0.653 (p = 0.021). Notably, the correlation between BIA and the wearable device muscle mass measurements peaked at an almost perfectly positive relationship by day 14 (r = 0.987, p < 0.001), highlighting the high degree of agreement between the two methods.

Fig. 3.

Relationship between ultrasound, BIA, and the wearable device in muscle mass measurement

Figure 4 shows a comparison of the change in muscle mass measured over a 2-week period using ultrasound (CSA), the traditional BIA device, and the wearable device. On day 7, the CSA and traditional BIA measurements showed non-significant correlations with the baseline values (r = 0.121, p = 0.501). In addition, the wearable device measurements showed no significant correlation (r = 0.187, p = 0.307). Conversely, traditional BIA and the wearable device measurements showed a significant correlation (r = 0.410, p = 0.018). By day 14, the CSA and traditional BIA measurements showed non-significant correlations with the baseline values (r = 0.388, p = 0.212). The wearable device measurements also showed no significant correlation (r = 0.541, p = 0.069). However, traditional BIA and the wearable device measurements showed the strongest correlation (r = 0.962, p < 0.001).

Fig. 4.

Relationship between ultrasound, BIA, and the wearable device in determining muscle mass change

Associations between ultrasound, BIA, the wearable device, and ventilator weaning

Additional Table 1–1 displays the average measurements of ultrasound parameters (RF muscle thickness, total anterior thigh muscle thickness, CSA of the RF, echogenicity), BIA, and wearable device measurements at days 3, 7, and 14 for patients who were successfully and unsuccessfully weaned off the ventilator. For patients who were successfully weaned, there were no statistically significant differences between the average values at day 14 and those at day 3. In contrast, for patients who were unsuccessfully weaned, almost all measurements at day 14, except for echogenicity, showed statistically significant differences compared to those at day 3. Specifically, significant p values were observed for RF muscle thickness (p = 0.020), total anterior thigh muscle thickness (p = 0.022), CSA of the RF (p = 0.005), BIA measurements (p = 0.027), and the wearable device measurements (p = 0.009).

Additional Table 1–2 shows the differences in muscle changes between days 7 and 14 relative to day 3 for the two groups. Statistically, there were no significant differences in the percentage change from day 3 to day 7 or from day 3 to day 14 between the groups who were successfully and unsuccessfully weaned off the ventilator. However, a trend was observed indicating that patients who were unsuccessfully weaned exhibited a greater overall reduction in muscle mass at day 14 across ultrasound, BIA, and wearable device measurements.

Associations between ultrasound, BIA, the wearable device, and survival outcomes

Additional Table 2–1 presents the average measurements of ultrasound parameters (RF muscle thickness, total anterior thigh muscle thickness, CSA of the RF, echogenicity), BIA, and the wearable device measurements at days 3, 7, and 14 for patients who died during their hospital stay versus those who survived. In the survive group, there were no statistically significant differences between the averages at day 14 and those at day 3. Conversely, in the deceased group, almost all measurements at day 14 were lower than were those at day 3. Except for echogenicity, almost all measurements at day 14 exhibited statistically significant differences compared to those at day 3. Specifically, significant p values were observed for RF muscle thickness (p = 0.009), total anterior thigh muscle thickness (p = 0.017), CSA of the RF (p = 0.003), BIA measurements (p = 0.026), and the wearable device measurements (p = 0.003).

Additional Table 2–2 illustrates the differences in muscle changes from day 3 to days 7 and 14 between the deceased and surviving groups. Statistically, a significant difference was observed in the percentage change in RF muscle thickness and CSA from day 3 to day 14 (p = 0.049 and p = 0.018 respectively). Although other measures did not show significant differences, a noticeable trend emerged, indicating that the deceased group experienced a greater overall reduction in muscle mass at day 14 than did the survival group across ultrasound, BIA, and the wearable device measurements.

Associations between ultrasound, BIA, the wearable device, and discharge

Additional Table 3–1 presents the average measurements of ultrasound parameters (RF muscle thickness, total anterior thigh muscle thickness, CSA of the RF, echogenicity), BIA, and the wearable device measurements at days 3, 7, and 14 for patients who were discharged to home versus those who were discharged to other locations (another hospital or deceased). For patients who were discharged to home, there were no statistically significant differences between the averages at day 14 and day 3. In contrast, for patients who were not discharged to home, almost all measurements at day 14 were significantly lower compared to those at day 3. Specifically, significant p values were observed for RF muscle thickness (p = 0.007), total anterior thigh muscle thickness (p = 0.017), CSA of the RF (p = 0.001), BIA measurements (p = 0.047), wearable device measurements (p = 0.026), and echogenicity (p = 0.003).

Additional Table 3–2 shows the differences in muscle changes from day 3 to days 7 and 14 between the two groups. Statistically, a significant difference was observed in the percentage change in CSA from day 3 to day 14 (p = 0.035). Although other measures did not show significant differences, a trend was observed indicating that patients who were discharged to other locations experienced a greater overall reduction in muscle mass at day 14 across ultrasound, BIA, and the wearable device measurements than did those who were discharged to home.

Associations between ventilator application, duration, and muscle wasting assessed by ultrasound, BIA, and a wearable device

When comparing muscle changes at day 7 between patients with and without mechanical ventilation (Additional Table 4–1, 4–2), no significant changes were observed in the non-ventilated group, whereas the ventilated group showed significant decreases in RF thickness (p = 0.003), total anterior thigh muscle thickness (p = 0.010), CSA of RF (p < 0.001), BIA measurements (p = 0.015), and wearable device measurements (p = 0.028). Similarly, from day 3 to day 7, muscle mass did not change significantly in the non-ventilated group, but the ventilated group exhibited significant declines across all parameters except echogenicity, including BIA (p = 0.033) and wearable device assessments (p = 0.043). In subgroup analyses of ventilated patients (Additional Table 5–1, 5–2), those with prolonged ventilation (≥ 7 days) demonstrated more pronounced muscle wasting at day 7 compared with those ventilated for < 7 days, with significant reductions in BIA (p = 0.025), wearable device values (p = 0.015), and all ultrasound parameters except echogenicity. Likewise, between day 3 and day 7, patients ventilated for ≥ 7 days showed significant muscle loss in nearly all measures, including wearable device assessments (p = 0.018), while no significant changes were detected in patients ventilated for < 7 days.

Discussion

This study assessed muscle mass changes in critically ill patients using three different measurement techniques: ultrasound, traditional BIA, and a wearable device. Although these methods did not show statistically significant agreement, all three demonstrated consistent patterns of muscle mass reduction in ICU patients. To our knowledge, this represents the first evaluation of muscle mass using a wearable device in critically ill patients and the first examination of concordance between traditional BIA and wearable devices in this population. These findings provide insights into the feasibility and comparative performance of different bedside monitoring approaches for muscle assessment in critical care settings.

There are several limitations to using BIA in critically ill patients. First, the accuracy of body composition measurements using BIA devices such as the InBody or Galaxy Watch has not been precisely established in this population. BIA devices employ regression models to estimate body composition, and although they incorporate multiple variables beyond body weight, individual variability in hydration and body composition may affect accuracy. Previous studies have shown that BIA measurements correlate reasonably well with CT, MRI, and DXA, which are considered accurate and reproducible methods for body composition analysis [16–18]. However, only limited studies have been conducted in critically ill patients. Kim et al. found a significant correlation between skeletal muscle mass measured by BIA and CT scans (p < 0.0001) [19], and Looijaard et al. reported that although absolute values differed, both BIA and CT were effective in identifying patients with low muscle mass [20]. Second, BIA accuracy can be influenced by fluid balance, including hydration status, edema, and patient positioning [13]. Since muscle tissue contains a high percentage of water, both ultrasound and BIA are affected by fluid shifts. To minimize this issue, we excluded patients with significant overhydration and monitored daily fluid input/output during ICU stay. The median net fluid balance was approximately – 500 mL, and only a few patients experienced substantial overhydration (> + 2000 mL). Nevertheless, residual confounding from fluid status cannot be fully excluded. Third, the Galaxy Watch primarily measures upper body muscle mass, whereas the InBody reflects whole-body composition and ultrasound focuses on lower limbs, limiting direct comparability between devices. These factors suggest that while BIA and wearable devices may provide supportive information in ICU patients, their results should be interpreted with caution and require further validation in this specific population.

Currently CT, MRI, DXA, and muscle biopsy are considered to be the most accurate methods for assessing skeletal muscle mass [6, 21–24]. However, these methods are time-consuming, expensive, and invasive. Although the reliability, validity, and accuracy of muscle measurement using ultrasound have not been conclusively established, studies have shown that in patients with cerebral palsy, ultrasound measurements of muscle thickness and CSA correlate moderately to well with MRI-derived muscle volume measurements [21]. Moreover, in young individuals, ultrasound has demonstrated effectiveness in the measurement of muscle size, comparable to that obtained using MRI or CT [22, 25]. In healthy individuals, ultrasound is also a reliable method for muscle evaluation [26]. In addition, in older patients, ultrasound has been found to have good validity when compared to CT, MRI, and DXA for assessing muscle size [27]. Furthermore, in critically ill patients, ultrasound has proven useful in detecting muscle necrosis and inflammation, which indicates its significance in predicting muscle changes without the need for muscle biopsy [28]. Although no statistical significance was observed, a previous study revealed that, in critically ill patients, the group with muscle atrophy identified using sonography had lower Medical Research Council muscle strength scores and more severe handgrip weakness than did the group without muscle atrophy [29]. In the present study, ultrasound was used as the reference standard for assessing muscle mass, a choice supported by its well-documented accuracy in numerous studies. The precision of ultrasound in depicting muscle architecture provides a unique advantage, especially in critical care settings where patients undergo rapid physiological changes. Its non-invasiveness, repeatability, and sensitivity in detecting even small changes in muscle size make it an invaluable tool for tracking the catabolic effects of critical illnesses on muscle tissue.

In this study, we observed progressive decreases in RF and total anterior thigh muscle thickness and CSA using ultrasound. These findings are consistent with other studies demonstrating muscle mass reduction after ICU admission. Formenti et al. [30] comprehensively reviewed ultrasound assessment of peripheral muscle changes in ICU patients, demonstrating consistent muscle wasting across studies. Hernández-Socorro et al. [31] reported significant reductions in RF muscle CSA and thickness (p < 0.001), while Lee et al. [10] found quadriceps muscle layer thickness reductions of 8.61 ± 19.44% at day 7, 15.63 ± 23.75% at day 14, and 13.03 ± 25.21% at ICU discharge. Hrdy et al. [32] reported that ≥ 10% reductions in rectus femoris CSA were frequent within the first week and associated with subsequent muscle weakness and poor outcomes. In our cohort, muscle wasting was most pronounced during the first week, followed by relative stabilization through Day 14. This pattern differs from previous reports of progressive decline over several weeks [33, 34] and findings in the Fazzini et al. meta-analysis [35]. Possible explanations include routine rehabilitation provided to almost all patients and differences in patient characteristics, suggesting that muscle loss may occur most rapidly in the early phase. Recent studies demonstrate that muscle cross-sectional area changes correlate with functional outcomes. Mayer et al. [36] found early RF size decreases significantly associated with in-hospital physical function and strength at discharge. Quadros et al. [37] observed that ICU discharge CSA was independently associated with post-discharge activities of daily living, while Ong et al. [38] demonstrated RF-CSA changes correlated with motor function in pediatric patients. Although our study did not measure strength directly, the decline in muscle dimensions mirrors functional deterioration, reinforcing the need for interventions targeting both muscle quantity and quality.

Ultrasound showed prognostic value in our study. Patients who failed ventilator weaning had greater muscle loss from day 3 to day 14, with CSA changes showing more significant reduction than successful weaning patients (− 19.89% vs. − 10.78%) (Additional Table 1–2). Although not all changes reached statistical significance, these results suggest ultrasound can detect greater muscle loss in patients with poorer outcomes. Similarly, patients who died had significantly greater reductions in RF thickness and CSA from day 3 to day 14 than survivors (p = 0.049 and p = 0.018, respectively) (Additional Table 2–2), and those unable to be discharged home had significant CSA reductions (p = 0.035) (Additional Table 3–2). While no clear correlation was observed between ultrasound and BIA or wearable device measurements, both modalities demonstrated similar trends of greater muscle wasting in patients with poorer outcomes. Subgroup analyses according to ventilator application and duration (≥ 7 vs. < 7 days) revealed that ultrasound, BIA, and wearable devices consistently detected significant muscle mass reductions (Additional Tables 4 and 5). Therefore, while BIA and wearable devices may have potential as supportive prognostic tools in critically ill patients, their clinical utility remains uncertain and requires further validation in larger cohorts.

Previous studies suggest that muscle wasting may be more pronounced in women, with insulin sensitivity and muscle metabolism potentially contributing to this sex-specific difference [39, 40]. In our study, we explored muscle loss by sex and found that while both men and women experienced muscle wasting following ICU admission, women showed greater reductions in RF thickness, total anterior thigh muscle thickness, CSA, and wearable device-measured muscle mass by day 14. This observation aligns with prior reports and suggests that female patients may be particularly vulnerable to ICU-acquired weakness [41]. ICU-acquired weakness, defined as neuromuscular dysfunction in the absence of other causes beyond critical illness and its treatment [42], often results from prolonged immobilization and systemic inflammation. This condition is associated with delayed recovery and poor outcomes [43–45], highlighting the importance of early diagnosis. Proactive strategies such as tailored nutritional support [46, 47] and early mobilization [48, 49] should be implemented to mitigate these effects. As demonstrated in our study, routine muscle mass assessment in the ICU could offer a practical means to identify at-risk patients and promote better long-term recovery should be implemented to mitigate these effects. As demonstrated in our study, routine muscle mass assessment in the ICU could offer a practical means to identify at-risk patients and promote better long-term recovery.

In addition to muscle quantity, we evaluated muscle echogenicity as an indicator of muscle quality. Overall, no significant change in echogenicity was observed between Day 3 and Day 14, with no difference among patients discharged home. However, patients discharged to other facilities demonstrated a significant increase in echogenicity by Day 14 (p = 0.047, Additional Table 3–1). Although not statistically significant, there was a trend toward increased echogenicity in patients with weaning failure (p = 0.075, Additional Table 1–1) and non-survivors (p = 0.076, Additional Table 2–1). These findings suggest that early deterioration of muscle quality, reflected by rising echogenicity, may be more evident in vulnerable subgroups and could be linked to poor outcomes. Because echogenicity correlates with fatty and fibrotic infiltration of skeletal muscle [50, 51] and is associated with ICU-acquired weakness and reduced strength [52], our results underscore its potential prognostic relevance. However, given our relatively short observation window and limited sample size, further research with larger cohorts and longer follow-up is required to establish the predictive value of echogenicity in critically ill patients.

Our study has several strengths. First, the enrolled patients had APACHE II scores ranging from 8 to 47, with a mean score of 20.7 ± 7.2, which indicates a moderate mortality risk (25–40%), and a mean CFS score of 3.9 ± 1.4, which reflects moderate frailty. The inclusion of patients at a moderate risk of death, combined with the assessment of changes in muscle mass to evaluate prognosis, adds significant value to our research. Moreover, this study is the first to apply a wearable device for muscle assessment in critically ill ICU patients and to compare its performance directly with that of traditional BIA and ultrasound, representing a novel approach in this population. While the wearable device did not yield statistically significant results in detecting muscle mass changes, the attempt to evaluate its feasibility in the ICU is valuable. Wearable devices, if further validated, may offer advantages such as convenient, operator-independent, and potentially continuous monitoring of muscle status, which could broaden the feasibility of muscle monitoring in critically ill patients.

Our study has several limitations. First, this single-center study with a small convenience sample may limit generalizability, as patient characteristics and ICU practices vary across institutions. The sample size was not determined by power calculation, making this exploratory in nature. Larger multicenter studies are needed to validate these findings; second, we did not formally assess inter-rater reliability or calculate technical error of measurement despite standardized operator training. Future studies should incorporate formal reliability assessments and consider automated analysis tools to improve reproducibility. Third, the number of patients available for analysis varied across measurement modalities and time points. Ultrasound measurements were obtained from 41 patients at Day 7 and 15 patients at Day 14, whereas BIA and wearable device data were available from 39 patients at Day 7 and 12 patients at Day 14. This attrition resulted from early ICU discharge, death, or technical measurement failures. Because complete-case analysis was employed without data imputation, potential selection bias may have been introduced, and statistical power was reduced at later assessment points. Fourth, we did not assess functional outcomes, limiting our ability to link structural muscle changes with patient-centered outcomes. Future studies should incorporate functional measures alongside imaging assessments. Fifth, BIA devices have inherent accuracy limitations. The Galaxy Watch measures upper body muscle mass while InBody reflects whole-body composition and ultrasound focuses on lower limbs, limiting direct comparability. BIA accuracy can be affected by fluid shifts despite our attempts to control for hydration status. Sixth, we lacked gold standard validation with MRI, DXA, or muscle biopsy. While prior studies show correlations between ultrasound and these methods, direct validation in our cohort remains a limitation. Seventh, potential operator-dependent variability may exist despite blinding protocols and standardized techniques. Finally, our 14-day follow-up may not capture the complete trajectory of muscle changes, reflecting primarily the early phase of critical illness. We focused on quantitative changes without directly assessing muscle quality such as intramuscular adipose tissue infiltration, which are important determinants of function and outcomes [53]. Future studies should consider longer follow-up periods and incorporate muscle quality assessments.

Conclusions

In this study, we compared three non-invasive methods—ultrasound, BIA, and a wearable device—for assessing muscle wasting in critically ill patients. Among these, ultrasound appeared to be the most consistent method for tracking changes in muscle mass and may provide useful information regarding patient prognosis. BIA and the wearable device showed some trends but did not consistently reach statistical significance. Our findings suggest that incorporating these kinds of non-invasive muscle monitoring methods into routine ICU care could help identify patients at risk of poor outcomes and guide timely interventions, including nutritional support and early rehabilitation. Future studies should validate these findings in larger multicenter cohorts, standardize assessment techniques, and investigate the combined role of muscle quantity, quality, and functional measures in predicting prognosis and tailoring treatment strategies in critically ill patients.

Abbreviations

BIA

Bioelectrical impedance analysis

ICU

Intensive care unit

RF

Rectus femoris

CSA

Cross-sectional area

CT

Computed tomography

MRI

Magnetic resonance imaging

DXA

Dual-energy X-ray absorptiometry

ASIS

Anterior superior iliac spine

FFM

Fat-free mass

APACHE

Acute Physiologic and Chronic Health Evaluation

CFS

Clinical Frailty Scale

LOS

Length of stay

MV

Mechanical ventilation

I/O

Input/output

IQR

Interquartile range

Author contributions

Conceptualization; SYK, SIL Data curation, Formal analysis, Resources; SYK, YRJ, SIL Investigation; All authors Methodology; SYK, DKK, SIL Project administration; SYK, SIL Supervision; DHK, SIL Validation; DHK, GH, DKK Visualization; DKK, SIL Writing—original draft; SYK, SIL Writing—review and editing; All authors.

Funding

This work was supported by Chungnam National University Hospital Research Fund [2021-CF-038]. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Availability of data and materials

All data generated or analyzed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

This study was approved by the Clinical Research Ethics Committee of the Chungnam National University Hospital (approval number: 2021-11-062) and was therefore performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. At enrollment, written informed consent was obtained from each patient or from an authorized surrogate.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.