Ultrasound assessment of muscle atrophy and its association with functional outcomes in critically ill patients: a systematic review and meta-analysis

Abstract

Background

Critically ill patients experience rapid skeletal muscle wasting due to immobility and systemic inflammation. Traditional imaging modalities—including computed tomography, magnetic resonance imaging, and dual-energy X-ray absorptiometry—can accurately quantify muscle mass but are impractical for monitoring critically ill patients because of cost, radiation, and logistical constraints. Ultrasound offers a bedside, radiation-free, and repeatable alternative capable of assessing both quantitative (cross-sectional area, muscle thickness) and qualitative (echointensity, pennation angle) muscle changes. This systematic review and meta-analysis synthesizes existing evidence regarding ultrasound-derived measures of muscle atrophy monitoring in critically ill adults.

Main body

A systematic search identified 69 eligible studies evaluating lower-limb muscle ultrasound in the ICU. Meta-analysis demonstrated substantial, progressive atrophy. By Day 7 of ICU admission, the rectus femoris cross-sectional area showed a decline of 15.76% (95% CI − 18.06 to − 13.47, I² = 88.72%), while quadriceps muscle thickness exhibited a parallel, yet smaller, reduction of 11.00% (95%CI − 14.38 to − 7.62, I² = 90.63%). Echointensity tended to increase, while pennation angle decreased, reflecting compositional and architectural deterioration. Considerable heterogeneity may reflect differences in patient characteristics, anatomical landmarks, and ultrasound technical factors. Limited data linked early ultrasound changes with ICU-acquired weakness. However, after excluding an outlier, pooled analysis of Day 1–3 rectus femoris cross-sectional area changes showed a statistically significant greater muscle loss in patients who developed ICU-acquired weakness to those who did not (mean difference: − 3.58%; 95% CI − 5.95 to − 1.22, I² = 0%).

Conclusions

Ultrasound detects early, rapid muscle wasting and concurrent quality degradation in critically ill adults. However, very low certainty of evidence due to heterogeneity and methodological limitations hinders clinical translation. Standardized protocols and prospective validation studies are needed to clarify ultrasound’s potential clinical utility in ICU muscle monitoring.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13054-025-05825-6.

Background

Rapid skeletal muscle wasting is common in critically ill patients because of their immobility and systemic illness. This loss of muscle mass impairs functional recovery and increases the risk of infection, prolonged mechanical ventilation, and mortality [1]. Although clinical assessment remains essential in identifying patients at risk of poor outcomes, bedside ultrasound is a valuable adjunct because it enables direct, serial evaluation of muscle atrophy and its progression. Early detection of muscle loss through ultrasound can help guide nutritional and rehabilitative interventions, although additional evidence is required to determine its influence on clinical decision-making.

Conventional imaging techniques used for muscle assessment—such as computed tomography (CT), magnetic resonance imaging (MRI), and dual-energy X-ray absorptiometry (DEXA) —are highly accurate but impractical for frequent monitoring [2]. Additionally, these methods are limited by their high cost, associated radiation exposure, and the difficulties of transporting critically ill patients to imaging sites. By contrast, ultrasound can be performed at the bedside, is repeatable, and involves no radiation risk, making it an attractive alternative for routine muscle monitoring in the intensive care unit (ICU).

Ultrasound assessments of muscle can be broadly divided into quantitative and qualitative assessments. Quantitative ultrasound involves a focus on muscle size metrics such as muscle thickness (MT) and cross-sectional area (CSA) and enables the tracking of muscle size over time [1, 3–6]. Qualitative ultrasound involves evaluating muscle quality through parameters such as echointensity (EI) (or echogenicity) and pennation angle (PA) [1, 7–10]. These qualitative measures provide insight into muscle’s composition and architecture (e.g. fat infiltration, fibrosis, and muscle fibre orientation); such information has traditionally required invasive biopsies. However, substantial variability exists among studies regarding measurement techniques, muscle sites, and timing, which limits clinical translation. This systematic review and meta-analysis therefore aimed to: [1] synthesize current evidence on ultrasound-based quantitative and qualitative assessment of skeletal muscle in critically ill adults [2], summarize its measurement properties and temporal changes [3], evaluate associations with clinically relevant outcomes, particularly ICU-acquired weakness (ICU-AW) and functional recovery, and [4] quantitatively pool muscle parameters using random-effect models to estimate the magnitude of muscle atrophy over the first week of ICU stay and to assess heterogeneity and potential publication bias.

Methods

This systematic review and meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020 guidelines and prospectively registered in PROSPERO (CRD420251052372). The review followed the registered protocol with methodological enhancements made during conduct: we performed meta-analysis (in addition to planned narrative synthesis) after recognizing sufficient outcome homogeneity, expanded database coverage to four databases including PubMed, Embase, Cochrane Library, and Web of Science and removed language restrictions.

Eligibility criteria

Studies were eligible if they [1] enrolled critically ill adults (≥ 18 years) admitted to an ICU and [2] used ultrasound to assess skeletal muscle parameters such as MT, CSA, EI and PA. Because the majority of published studies have focused on lower-limb muscles—particularly the quadriceps and its components—we restricted inclusion to lower-limb assessments to ensure methodological consistency and comparability. We excluded review articles, case reports, pediatric studies, interventional trials, and studies evaluating only upper-limb muscles.

Literature search

A systematic search was conducted in PubMed, EMBASE, Cochrane Library, and Web of Science from inception to October 11, 2025 (searches conducted October 10–11, 2025). The search strategy combined controlled vocabulary (MeSH and Emtree) with free-text keywords across three domains: [1] patient population (“critically ill patients”, “intensive care unit”) [2], imaging technique (“ultrasound”, “ultrasonography”, “sonography”), and [3] muscle parameters (“muscle atrophy”, “muscle thickness”, “cross-sectional area”, “pennation angle”, “echointensity”, “ICU-acquired weakness”). Additionally, reference lists of included studies were hand-searched to identify additional relevant publications.

Study selection and data extraction

Two reviewers independently screened titles, abstracts, and full texts. Disagreements were resolved through discussion or consultation with a third reviewer when necessary. Data were extracted using a standardized template capturing study design, patient characteristics, muscle site, ultrasound protocol, timing of assessments, and outcomes. Study quality was assessed using a modified Newcastle–Ottawa Scale (NOS) adapted for cohort studies (Supplementary Table 1).

Statistical analysis

Quantitative outcomes were pooled using random-effects models implemented in Comprehensive Meta-Analysis version 3.7. For longitudinal changes in muscle parameters, results were expressed as pooled mean differences between time points with 95% confidence intervals (CIs). For comparisons between groups (ICU-AW vs. non-ICU-AW), standardized mean differences were calculated when appropriate. Heterogeneity was quantified by I², with > 50% indicating substantial heterogeneity, and > 75% indicating considerable heterogeneity. Between-study variance (τ²) was estimated using the Restricted Maximum-Likelihood (REML) method, which provides less biased estimates than the DerSimonian-Laird method. For meta-analyses with small numbers of studies (n ≤ 5), 95% CIs were calculated using the Hartung-Knapp/Sidik-Jonkman (HKSJ) adjustment method to provide more robust coverage. Publication bias was evaluated using Egger’s regression test. Subgroup analyses were conducted to explore the potential sources of heterogeneity—specifically, differences in ultrasound measurement sites. Detailed search strategies, data extraction templates, risk-of-bias assessment criteria, and additional methodological details are provided in the Supplementary Methods.

Results

Study selection and characteristics

A total of 961 records were identified through database searching—PubMed (n = 187), Embase (n = 356), Cochrane Library (n = 142), Web of Science (n = 272) and reference lists (n = 4). After removing 556 duplicates, 405 unique records were screened, and 106 full texts were retrieved. Ultimately, 70 studies initially met the inclusion criteria; however, one was subsequently excluded due to duplicate data reporting with another included study, resulting in 69 unique studies incorporated into qualitative and/or quantitative synthesis (Fig. 1, PRISMA 2020 flow diagram).

Fig. 1.

PRISMA 2020 flow diagram illustrating the process of study selection for the review

Included studies were predominantly conducted in mixed, medical or surgical ICUs. Most investigations targeted the quadriceps muscle group—particularly the rectus femoris (RF) and vastus intermedius (VI)—using linear array transducers. Ultrasound parameters commonly assessed included MT, CSA, PA and EI. Ultrasound assessments were performed at various anatomical landmarks, representing a major source of methodological variability across studies. However, measurements were predominantly concentrated at two anatomically distinct sites: [1] the midpoint between the anterior superior iliac spine (ASIS) and the superior border of the patella (SP), which corresponds to the site of maximal RF thickness and demonstrates high sensitivity to muscle changes; and [2] the two-thirds of the distance from the ASIS to the SP, which is less susceptible to edema-related artifacts and offers superior measurement reproducibility [11]. Most studies specified that measurements were performed with minimal probe compression to avoid altering muscle morphology during scanning (Supplementary Table 2).

Quantitative analysis of muscle mass changes

Meta-analysis demonstrated a progressive decline in RF cross-sectional area (RFCSA) during the early ICU course. RFCSA decreased by 7.20% (95% CI − 8.66 to − 5.75; I² = 83.13%) within the first three days, and by 11.29% (95% CI − 15.14 to − 7.43; I² = 96.62%) at Day 5, and by 15.76% (95% CI − 18.06 to − 13.47; I² = 88.72%) at Day 7, confirming substantial early muscle wasting (Fig. 2A).

Fig. 2.

Percentage Changes in Ultrasound-Measured Muscle Parameters from Day 1 to Day 7 of ICU Stay

The pooled data for quadriceps muscle thickness (QMT) showed a comparable but less pronounced reduction. During the first three days, QMT decreased by 5.51% (95% CI − 7.62 to − 3.40; I² = 91.67%), and by Day 5, the reduction reached 5.72% (95% CI − 6.86 to − 4.58; I² = 0%). By Day 7, cumulative loss was 11.00% (95% CI − 14.38 to − 7.62; I² = 90.63%) (Fig. 2B).

Analysis of rectus femoris muscle thickness (RFMT) revealed a more pronounced decline compared with total quadriceps thickness. Within the first three days, RFMT decreased by 7.95% (95% CI − 12.41 to − 3.49; I² = 75.55%), and by the end of the first week, the cumulative loss reached 13.31% (95% CI − 16.46 to − 10.16; I² = 69.12%).

Substantial heterogeneity was observed across most analyses. When RFCSA analysis was restricted to studies using the two-thirds distance from the ASIS to the SP (ASIS-SP), the pooled reduction remained significant at 16.10% (95% CI − 18.67 to − 13.54; I² = 85.72%) at Day 7, with a modest decrease in heterogeneity compared with the overall analysis. QMT analysis restricted to the same landmark showed a reduction of 10.99% (95% CI − 15.06 to − 6.92; I² = 91.04%) at Day 7.

Qualitative analysis of muscle quality parameters

Muscle quality parameters demonstrated consistent deterioration during the first week of ICU admission. Four studies provided serial RF echointensity (RFEI) data, showing a pooled mean increase of 6.95% (95% CI − 2.17 to 16.06; I² = 87.95%) from baseline to Day 7 (Fig. 2C). Although the confidence interval crossed zero, the upward trajectory indicates progressive muscle echogenicity, commonly reflecting intramuscular fat infiltration and fibrosis.

Three studies reported serial changes in RF pennation angle (RFPA), demonstrating a pooled mean decrease of − 20.81% (95% CI − 27.47 to − 14.14; I² = 88.36%) (Fig. 2D).

Together, a structured summary of all relevant quantitative details—namely the number of studies, pooled mean change estimates with 95% confidence intervals, I² values, and Egger’s test results for both the main analysis and the standardized-site subgroup analysis indicated that these qualitative changes parallel the quantitative atrophy observed during early critical illness, indicating concurrent structural remodeling and compositional degradation (Table 1).

Table 1.

Pooled quantitative and qualitative ultrasound estimates of skeletal muscle changes within the first week of ICU admission

Muscle parameter	Time interval	Main analysis			Egger’s test (p=)	Subgroup analysis (standardized site)*			Egger’s test (p=)
		No. of studies	Pooled mean change (%) (95%CI)	I2(%)		No. of studies	Pooled mean change (%) (95% CI)	I2(%)
RFCSA	Day 1–3	13 Ref. [5, 10, 12, 16, 21, 37–43]	–7.20% (− 8.66 to − 5.75)	83.13	0.19	9 [5, 12, 16, 21, 37–39, 42, 43]	–6.20 (–7.60 to − 4.79)	78.13	-
	Day 1–5	8 [16, 26, 32, 38, 40–42, 44]	–11.29% (− 15.14 to − 7.43)	96.62	-	5 [16, 26, 32, 38, 42]	–8.95 (–12.55 to − 5.35)	93.09	-
	Day 1–7	21 [1, 5, 7, 10, 16, 21, 27, 38–42, 45–53]	–15.76% (− 18.06 to − 13.47)	88.72	< 0.01	16 [1, 5, 7, 16, 21, 27, 38,39, 42, 45–50, 52]	–16.10 (–18.67 to − 13.54)	85.72	< 0.01
QMT	Day 1–3	5 [5, 6, 44, 54, 55]	–5.51% (–7.62 to − 3.40)	91.67	-	4 [5, 6, 54, 55, 55]	–5.18 (–7.41 to − 2.94)	93.21	-
	Day 1–5	3 [6, 26, 44]	–5.72% (–6.86 to − 4.58)	0	-	-	-	-	-
	Day 1–7	8 [5, 6, 27, 48, 54–56]	–11.00% (–14.38 to–7.62)	90.63	-	6 [5, 6, 27, 48, 54, 55]	–10.99 (–15.06 to − 6.92)	91.04	-
RFMT	Day 1–3	3 [12, 37, 42]	–7.95% (–12.41 to -3.49)	75.55	-	-	-	-	-
	Day 1–7	7 [1, 2, 42, 49, 53, 57, 58]	–13.31% (–16.46 to − 10.16)	69.12	-	4 [1, 42, 49, 52]	–16.55 (–20.16 to − 12.99)	29.2	-
RFEI	Day 1–7	4 [1, 7, 42, 49]	6.95(–2.17 to 16.06)	87.95	-	-	-	-
RFPA	Day 1–7	3 [8, 10, 48]	-20.81% (–27.47 to − 14.14)	88.36	-	-	-	-

RFCSA, rectus femoris cross-sectional area; QMT, quadriceps muscle thickness; RFMT, rectus femoris muscle thickness; RFEI, rectus femoris echointensity, RFPA, rectus femoris pennation angle, CI, confidence interval; ASIS, anterior superior iliac spine; SP, superior border of patella

*Standardized site refers to measurements performed at two-thirds of the distance from the anterior superior iliac spine (ASIS) to the superior border of the patella (SP)

Association between muscle mass changes and ICU-Acquired weakness

To evaluate the relationship between early muscle mass changes and the development of ICU-AW, a random-effects meta-analysis was conducted using studies that reported Day 1–3 changes in muscle size, as this was the only time frame with sufficient data (≥ 3 studies) for quantitative synthesis. Initial pooling of all four eligible studies yielded a pooled mean difference of − 13.80% (95% CI − 63.60 to 36.01, I² = 92%) (Fig. 3A). One study was identified as an outlier [12], reporting physiologically implausible findings: the interquartile range for RFCSA change exceeded 100%, indicating that at least 25% of patients without ICU-AW exhibited > 100% increases in RFCSA within three days—incompatible with expected muscle wasting in critical illness. After excluding this outlier, the pooled mean difference narrowed substantially to − 3.58% (95% CI − 5.95 to − 1.22; I² = 0%), demonstrating that patients who developed ICU-AW experienced significantly greater muscle loss during the first three days compared with those who did not (Fig. 3B). Detailed forest plots for all time points and subgroup analyses are provided in Supplementary Figures.

Fig. 3.

RFCSA Change form Day 1 to Day 3 in Patient With or Without ICU-AW

Publication bias

Publication bias was assessed using Egger’s regression test for meta-analyses including ≥ 10 studies. Among RFCSA outcomes, no significant asymmetry was detected at Day 3 (p = 0.19, n = 13), but potential publication bias was suggested at Day 7 (p < 0.01, n = 21). For meta-analyses with fewer than 10 studies—including Day 5 RFCSA (n = 8), QMT at Day 3 (n = 5) and Day 7 (n = 8), RFMT (n = 7), RFEI (n = 4), and RFPA (n = 3)—the number of studies was insufficient for reliable publication bias assessment (Table 1).

Certainty of evidence (GRADE)

GRADE assessment revealed very low certainty for 11 of 12 outcomes and low certainty for one outcome (QMT Day 1–5, I² = 0%), primarily due to observational study design, substantial heterogeneity, imprecision, and suspected publication bias for select outcomes. Detailed GRADE evaluations are provided in Supplementary Table 3. The very low certainty indicates that true effects may differ substantially from current estimates.

Discussion

This systematic review and meta-analysis demonstrated a consistent pattern of early skeletal muscle wasting in critically ill adults during the first week of ICU admission, although the magnitude of decline varied substantially across studies. By approximately 7 days, pooled estimates showed reductions of 15.76% in RFCSA, 11.00% in QMT, and 13.31% in RFMT. Muscle quality parameters changed in parallel, with RFEI increasing by 6.95% and RFPA decreasing by 20.81%. These findings align with prior evidence indicating rapid catabolic loss in the early ICU phase; however, the high heterogeneity observed (I² often > 75%), variability in measurement landmarks and ultrasound protocols, and very low certainty of evidence (GRADE) necessitate cautious interpretation of the absolute values.

Meta-analysis examining the association between early muscle loss (Day 1–3) and subsequent ICU-AW development revealed findings highly sensitive to study quality. Initial pooling of all four available studies showed no significant association (mean difference − 13.80%, 95% CI − 63.60 to 36.01; I² = 92%). After excluding one outlier study with physiologically implausible findings, heterogeneity resolved completely (I² = 0%) and a statistically significant association emerged: patients who developed ICU-AW experienced an additional 3.58% points of muscle loss during the first three days (95% CI − 5.95 to − 1.22). This dramatic shift from null to significant findings upon outlier exclusion underscores the fragility of evidence in this area. Additional methodological constraints limit definitive interpretation: only three studies remained after exclusion, all relied exclusively on Medical Research Council (MRC) sum scores for weakness assessment, timing of ICU-AW evaluation varied across studies, and only the Day 1–3 interval had sufficient data for quantitative synthesis. While the sensitivity analysis suggests that early muscle loss may predict ICU-AW development, the evidence base remains insufficient to establish this relationship with confidence or determine clinically meaningful thresholds for risk stratification.

Temporal pattern of muscle loss

Observational studies have reported that the rate of muscle atrophy is most rapid during the first week of ICU admission, with subsequent deceleration after 1–2 weeks [13–16]. Our meta-analysis confirmed the rapidity of early muscle loss, revealing significant RFCSA decline within the first three days (7.20%) and continued loss through the first week (15.76%). However, we were unable to quantitatively assess the deceleration phase due to insufficient numbers of studies with extended follow-up beyond the first week.

During the early ICU period when critical illness is most severe, systemic inflammation, immobilization, and elevated catabolic hormones drive accelerated proteolysis and suppressed protein synthesis [17, 18]. Muscles rich in fast-twitch type II fibers—such as the RF—are preferentially affected due to their higher metabolic activity and greater vulnerability to catabolic stress, resulting in disproportionate early atrophy [1, 3, 8, 19–21]. As systemic inflammation stabilizes, hemodynamic stability improves, nutritional support is initiated, and mobilization opportunities increase, the rate of muscle loss muscle loss would be expected to decelerate [17, 18]. This temporal pattern underscores the importance of early prevention strategies during the first week, when muscle atrophy is most rapid.

Differences between CSA and thickness measurements

An intriguing finding of this meta-analysis is the discrepancy in loss rates between CSA and MT measurements. By Day 7, RFCSA decreased by 15.76%, substantially exceeding the decline observed in RFMT (13.31%) and QMT (11.00%). This disparity reflects not only methodological differences but also the impact of pathophysiological changes in critically ill patients on ultrasound imaging. Muscle thickness measurements are susceptible to confounding by tissue edema, which is nearly universal during early critical illness due to capillary leak, fluid resuscitation, and hypoalbuminemia. Several studies have reported paradoxical early increases in MT despite ongoing catabolism, suggesting that interstitial fluid accumulation artificially preserves or increases this linear dimension [22–26]. In contrast, CSA—as a two-dimensional planimetric measurement—remains relatively resistant to fluid-related confounding and more accurately reflects true muscle mass loss [22, 27]. Although CSA measurements are more technically demanding than thickness assessments, their superior resistance to edema-related artifacts makes them preferable for serial monitoring in critically ill patients. Thickness measurements may systematically underestimate early muscle loss, potentially delaying recognition of at-risk patients and missing critical intervention windows.

Changes in muscle quality parameters

Beyond quantitative atrophy, muscle quality parameters showed evidence of deterioration during early critical illness, though interpretation is substantially constrained by limited studies and high heterogeneity. For RFEI, the pooled increase of 6.95% had wide confidence intervals crossing zero, indicating no statistically significant change despite the upward trend. In contrast, RFPA demonstrated a statistically significant reduction.

Increased EI would indicate compositional alterations including intramuscular fat infiltration, connective tissue deposition, and edema [9]. Decreased PA reflects architectural disruption and loss of the oblique fiber arrangement that optimizes force generation [28]. However, these qualitative measurements are highly susceptible to technical and operator-dependent factors compared with quantitative size measurements.

Echointensity is highly sensitive to ultrasound gain settings, which are typically adjusted by operators to enhance image clarity rather than standardized according to protocol. Additionally, EI requires offline quantification using region-of-interest (ROI) analysis, and inconsistencies in ROI placement, size, and shape contribute to measurement variability [29]. Pennation angle measurements require precise probe alignment perpendicular to muscle fascicles; minor deviations in angulation or transducer pressure can substantially alter values. Furthermore, identification of aponeurotic landmarks may be challenging in edematous or disrupted muscle [30].

Given these methodological challenges and limited evidence, the clinical utility of muscle quality parameters for routine monitoring remains uncertain. While PA changes suggest that architectural degradation occurs early in critical illness, additional prospective studies employing rigorous protocol standardization and operator training are necessary to establish whether these parameters provide clinically meaningful information beyond quantitative muscle size measurements.

Association between ultrasound muscle parameters and functional outcomes

Multiple studies have investigated the relationship between ultrasound-detected muscle changes and functional outcomes in critically ill patients, but substantial methodological heterogeneity precluded comprehensive meta-analysis. Studies employed varied ultrasound parameters (different muscle groups, measurement sites, and timing) and diverse functional outcomes (ICU-AW, handgrip strength, mobility scales, performance tests), preventing data pooling across the full outcome spectrum.

Our meta-analysis was therefore restricted to the relationship between Day 1–3 RFCSA change and ICU-AW, the only outcome with sufficient consistency for synthesis. After excluding one outlier, pooled analysis of three studies showed that patients who developed ICU-AW experienced 3.58% greater muscle loss during the first three days. Although modest in magnitude, this early divergence suggests that ultrasound may serve as an early indicator of ICU-AW risk.

Unfortunately, insufficient studies with measurements at Day 5 or Day 7 precluded meta-analysis at these later timepoints, when cumulative muscle loss becomes more pronounced and effect sizes might be larger. This represents an important knowledge gap, as delayed assessment may offer stronger predictive utility for clinical risk stratification. Given the small number of studies (n = 3) and the modest effect size, these findings require validation in larger prospective cohorts before informing clinical practice.

Beyond ICU-AW, individual studies examining relationships between ultrasound parameters and other functional outcomes report inconsistent findings. Some demonstrated significant associations between muscle mass measurements and MRC scores, handgrip strength, or mobility scales [27, 31–33], while muscle quality parameters (EI and PA) showed variable predictive value (Table 2) [1, 22, 33, 34]. These inconsistencies highlight the need for prospective studies with standardized protocols and adequate power to establish ultrasound’s predictive utility for functional recovery.

Table 2.

Summary of ultrasound muscle metrics and their correlation with functional assessments in published studies

Study (First Author)	n	Ultrasound Parameter	MRC Correlation (r)	Statistical Significance	Other Functional Measures
Borges [32]	37	RF CSA change	-0.40	p < 0.05	HGS: r=-0.51, p < 0.01
Dams [22]	30	RF CSA	-	-	HGS: r = 0.290, p = 0.010
		RF PA	-	-	HGS: p = 0.581
		RF EI	-	-	HGS: p = 0.852
Guzmán-David [37]	31	RF CSA loss ≥ 10%	NS	p = 0.877	-
Mayer [1]	41	RF EI	0.337	P = 0.048	Muscle power: r = − 0.48, p = 0.005; 5xSTS: r = 0.462, p = 0.013
Nakanishi [31]	64	BB CSA	0.47	p = 0.01	HGS: r = 0.50, p = 0.01; FSS-ICU: r = 0.56, p < 0.01
		RF CSA	NR	p < 0.01	IMS: r = 0.35, p = 0.07
Palakshappa [27]	29	RF CSA change	0.51	p = 0.03	PFIT-s: r = 0.40, p = 0.10
Parry [33]	22	VI MT	0.56	p = 0.05	PFIT-s: r = 0.82, p < 0.001; IMS: r = 0.84, p < 0.001
		VI EI	-0.57	p = 0.04	PFIT-s: r=-0.77, p = 0.001; IMS: r=-0.73, p = 0.003
		VL PA	NS	-	PFIT-s: r = 0.81, p = 0.008
Yin [34]	47	RF PA	0.70	< 0.001	-
		VI PA	0.57	< 0.001	-
		VI MT change	-0.68	< 0.01	-
		Q MT	0.365	< 0.001	-
		VI MT	0.132	0.215	-
Zhang [59]		RF CSA change	-0.71	< 0.001	-

MT = Muscle Thickness; CSA = Cross-Sectional Area; PA = Pennation Angle; EI = Echointensity; RF = rectus femoris; Q = quadriceps; VI = Vastus Intermedius; BB = Biceps Brachii; VL = Vastus Lateralis; MRC = Medical Research Council muscle strength score; BI = barthe index; PFIT-s = Physical Function in ICU Test-scored; IMS = ICU Mobility Score; HGS = Handgrip Strength; FSS-ICU = Functional Status Score for the ICU; NR = Not Reported; NS = Not Significant; OR = odd ratio; AUC = area under curve

Correlation coefficients (r) are Spearman or Pearson as reported; positive r indicates higher parameter values linked to better MRC scores, negative r indicates worse MRC scores

Sources of heterogeneity

Substantial heterogeneity was observed across nearly all pooled estimates, with I² values frequently exceeding 85%. These values indicate that the majority of variability in effect estimates arises from true between-study differences rather than sampling error. To explore potential sources, subgroup analysis was conducted restricting inclusion to studies employing a standardized anatomical landmark (two-thirds of the ASIS-SP distance). This yielded only modest reductions in heterogeneity for some parameters, indicating that anatomical landmark standardization alone cannot explain the observed variability.

Multiple additional factors likely contribute to measurement heterogeneity. Patient-level characteristics including age, sex, baseline muscle mass, illness severity, and underlying disease processes may influence the rate and pattern of muscle loss [4, 7, 15, 35, 36]. Technical factors—including differences in ultrasound equipment specifications, probe frequency, imaging depth, gain settings, and operator technique—introduce further variability. However, the absence of individual patient data in most included studies precluded meta-regression analysis to quantify the relative contribution of these factors. This represents a major limitation of the current evidence base. Future meta-analyses incorporating individual patient data are necessary to delineate how patient characteristics and technical factors interact to produce heterogeneity, and to identify subgroups with distinct muscle wasting phenotypes that may benefit from tailored interventions.

Limitations

Several limitations must be acknowledged. First, our search strategy did not include grey literature or conference proceedings, potentially missing unpublished studies. While we conducted manual reference list searching of all included studies, some relevant evidence may not have been captured. Second, many studies did not report percentage changes from baseline, limiting meta-analysis inclusion and preventing assessment of absolute versus relative changes. Third, substantial heterogeneity (I² >85%) indicates considerable between-study variability in muscle loss rates. The absence of individual patient data precluded meta-regression to explore effect modifiers including age, frailty, illness severity, and comorbidities. Fourth, functional outcomes were limited predominantly to ICU-AW via MRC scores; other functional measures—Physical Function ICU Test Score, ICU Mobility Scale, handgrip strength—were insufficiently reported. Heterogeneity in measurement protocols and timing contributed to unresolved variability. Finally, the small number of studies for most outcomes limited statistical power and precluded reliable publication bias assessment for many analyses.

Conclusions

This systematic review and meta-analysis identified a notable early skeletal muscle wasting in critically ill adults, although the magnitude of decline varied widely across studies. Limited evidence suggests that early muscle loss may be associated with subsequent ICU-AW, but this finding is based on only three studies and warrants cautious interpretation. Cross-sectional area measurements appears less affected by edema than muscle thickness, yet comparative evidence remains sparse. Overall certainty of evidence certainty was very low due to observational study designs, substantial heterogeneity (I² >85%) whch limits the precision of pooled estimates, and methodological variability in ultrasound protocols and timing. Despite these constraints, bedside ultrasound offers practical and feasible tool for monitoring muscle status in the ICU given its non-invasive, repeatable, and radiation-free natures. Future research should prioritize standardized measurement protocols, adequately powered prospective cohorts, and individual patient data meta-analyses to identify sources of heterogeneity. Prospective validation studies are also needed to establish whether early ultrasound-based muscle assessments can reliably predict functional outcomes and support clinical interventions in the ICU setting.

Forest plots showing percentage changes from baseline in (A) rectus femoris cross-sectional area (RFCSA), (B) quadriceps muscle thickness (QMT), (C) rectus femoris muscle thickness (RFMT), and (D) rectus femoris pennation angle (RFPA) during Day 1–7 of ICU admission. Data are presented as pooled mean difference with 95% confidence intervals.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ASIS

Anterior superior iliac spine

CSA

Cross-sectional area

CI

Confidence interval

CT

Computed tomography

DEXA

Dual-energy X-ray absorptiometry

EI

Echointensity

ICU

Intensive care unit

ICU-AW

ICU-acquired weakness

MRC

Medical Research Council

MRI

Magnetic resonance imaging

MT

Muscle thickness

PA

Pennation angle

PRISMA

Preferred Reporting Items for Systematic reviews and Meta-Analyses

QMT

Quadriceps muscle thickness

RF

Rectus femoris

RFCSA

Rectus femoris cross-sectional area

RFEI

Rectus femoris echointensity

RFMT

Rectus femoris muscle thickness

RFPA

Rectus femoris pennation angle

ROI

Region of interest

SP

Superior border of the patella

VI

Vastus intermedius

Author contributions

C.C.L.: conducted literature searches, screened articles, synthesised findings, drafted the manuscript, and contributed to data interpretationY.J.L.: conceptualised the study, designed the review framework, and provided critical guidance throughout the manuscript developmentC.T.C.: conducted literature searches, screened articles, conducted data extraction and statistical analysis, contributed to data interpretationH.M.C.: conducted literature searches, screened articles, conducted data extraction and statistical analysis, contributed to data interpretationW.C.H.: conceptualised the study, supervised the overall research process, critically revised the manuscript, ensured accuracy and coherence, and approved the final version for submission.

Funding

Not applicable.

Data availability

The datasets generated and analysed in the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.