Association of computed tomography-based radiodensity of the peri-hip muscles with hip fracture type and severity: a retrospective study in middle-aged and elderly patients

Abstract

Background

Sarcopenia and myosteatosis—hallmarks of aging associated with reduced muscle quality and function—can be assessed using computed tomography (CT)-based measurements of muscle cross-sectional area and radiodensity. This study quantitatively compared CT-measured cross-sectional area and radiodensity of peri-hip muscles among patients with non-displaced femoral neck, displaced femoral neck, and intertrochanteric fractures.

Methods

In total, 279 patients were retrospectively analyzed and categorized into three groups: non-displaced femoral neck fractures (n = 89; 63 women, 26 men), displaced femoral neck fractures (n = 101; 65 women, 36 men), and intertrochanteric fractures (n = 89; 52 women, 37 men). Cross-sectional area and radiodensity were measured on cross-sectional CT images: the gluteus medius and minimus at the level inferior to the sacroiliac joint, the gluteus maximus at the acetabular roof, and the quadriceps femoris and medial thigh muscles inferior to the lesser trochanter.

Results

No significant differences in sex, age, or cross-sectional area of peri-hip muscles were observed among the three groups. Significant differences were found in radiodensity of the gluteus maximus, medius, minimus, and medial thigh muscles but not in the quadriceps femoris (P = 0.001, P < 0.001, P < 0.001, P = 0.010, and P = 0.212, respectively). Post-hoc analyses showed significantly lower radiodensity in the gluteus medius, minimus, and medial thigh muscles in both displaced femoral neck and intertrochanteric fracture groups than in the non-displaced femoral neck fracture group (P = 0.008, P = 0.006, P = 0.020; P < 0.001, P < 0.001, P = 0.024; respectively), with the gluteus maximus also lower in the intertrochanteric group (P = 0.001).

Conclusion

Our study demonstrated that CT-measured peri-hip muscle radiodensity differs significantly among hip fracture types and severity. Specifically, higher radiodensity (reduced muscle fatty infiltration) was associated with non-displaced femoral neck fractures compared with displaced femoral neck or intertrochanteric fractures.

Background

Hip fracture remains a major public health challenge and economic burden worldwide, particularly with a rapidly aging population. The total number of hip fractures in individuals aged 50 years and older is projected to nearly double between 2018 and 2050 [1]. Patients with hip fractures often experience long-term disabilities, reduced quality of life, and increased mortality rates [1–3].

The main types of hip fractures include femoral neck and intertrochanteric fractures. According to the Garden classification, femoral neck fractures are further categorized into two groups: non-displaced (Garden Stages 1 and 2) and displaced (Garden Stages 3 and 4) [4, 5]. These fracture types differ in their mechanisms of injury, relationships with the joint capsule and surrounding musculature, and corresponding management strategies.

Falls are the leading direct cause of hip fractures [6]. Skeletal muscles play a fundamental role in movement and stability. Increased fatty infiltration, which indicates poor muscle quality, is associated with reduced strength of the thigh muscles [7]. Age-related loss of skeletal muscle mass and strength, as seen in sarcopenia, impairs physical stability and increases the risk of falls [8].

Computed tomography (CT) is considered the gold standard for non-invasive assessment of muscle quality and mass. Skeletal muscle mass and fatty infiltration can be quantified on CT by measuring the muscle cross-sectional area (CSA) and radiodensity, respectively. Kim et al. [9] reported that reduced CSA and radiodensity of the psoas and spinal extensor muscles may predispose individuals to hip fractures by impairing trunk stability and increasing fall risk. Furthermore, Yerli et al. [10] found that higher psoas muscle radiodensity (indicating less fatty infiltration) was associated with a greater likelihood of femoral neck fractures rather than intertrochanteric fractures, suggesting a potential role of muscle composition in determining fracture type .

However, evidence linking CT-based muscle CSA or radiodensity to specific types of hip fractures remains limited. To date, only one study has examined this association, focusing solely on muscles at the lumbar level [10]. By contrast, our study investigates peri-hip muscles using conventional hip CT scans obtained directly from patients with hip fractures, thereby improving clinical relevance and applicability. Additionally, we further subdivide femoral neck fractures into displaced and non-displaced types to provide a more detailed perspective on how CT-based muscle characteristics relate to fracture severity. Therefore, our study aims to compared CT-measured CSA and radiodensity of peri-hip muscles among patients with non-displaced femoral neck, displaced femoral neck, and intertrochanteric fractures.

Methods

Patient selection and grouping

Patients diagnosed with hip fractures (either femoral neck or intertrochanteric fractures) at our institution between January 2020 and April 2025 were retrospectively reviewed. The exclusion criteria were an age of < 55 years, pathological hip fractures, prior hip arthroplasty, neuromuscular disorders, hemiplegia secondary to cerebrovascular disease, and advanced cancer with metastasis. Based on these criteria, 279 patients with hip fractures were included in the study. The cohort comprised 180 women and 99 men aged 55–94 years. According to the Garden classification of femoral neck fractures [4, 5], the patients were categorized into two groups: non-displaced (Garden Stages 1 and 2) and displaced (Garden Stages 3 and 4). Accordingly, the 279 patients were divided into three groups: non-displaced femoral neck fractures (n = 89; 63 women, 26 men), displaced femoral neck fractures (n = 101; 65 women, 36 men), and intertrochanteric fractures (n = 89; 52 women, 37 men). The above procedures are illustrated as Fig. 1.

Fig. 1.

Flowchart of patient enrollment and exclusion

Patient demographics

Data on sex, age, and the prevalence of hypertension and type 2 diabetes were collected from medical records and are summarized in Table 1.

Table 1.

Patient demographics

	ND-FNF (n = 89)	D-FNF(n = 101)	ITF(n = 89)	P-value
Age, yr (Mean, SD)	74.01 ± 9.46	75.04 ± 9.99	77.30 ± 9.67	0.078a
Sex (n,%)				0.226b
Female	63(70.8%)	65(64.4%)	52(58.4%)
Male	26(29.2%)	36(35.6%)	37(41.6%)
Hypertension (n,%)	43 (48.3%)	63 (62.3%)	56 (62.9%)	0.078b
Type 2 Diabetes (n,%)	29 (32.6%)	40 (39.6%)	28 (31.5%)	0.437b

ND-FNF Non-displaced femoral neck fracture, D-FNF Displaced femoral neck fracture, ITF Intertrochanteric fracture, SD Standard deviation

^aKruskal–Wallis test

^bPearson chi-square test

CT imaging protocol

Preoperative bilateral hip CT scans were obtained for all patients during hospitalization following a fall, using a multidetector spiral CT scanner from GE HealthCare (Discovery 750HD or LightSpeed VCT). Daily air calibration was automatically performed after tube warm-up in accordance with the manufacturer’s specifications to ensure HU measurement accuracy. The scanning range extended from the anterior superior iliac spine to the inferior margin of the lesser trochanter. Scanning parameters were as follows: tube voltage 120 kVp, automated dose modulation (smartmA; GE HealthCare), matrix 512 × 512, detector collimation: 64 × 0.625 mm, pitch: 0.984:1, rotation time: 0.5–0.6 s. Quantitative measurements were performed on cross-sectional CT images reconstructed using a standard soft tissue algorithm with a 5 mm slice thickness. A uniform soft tissue window (width 350 HU; level 40 HU) was applied to facilitate muscle contour delineation.

Assessment of CSA and radiodensity of peri-hip muscles

To minimize measurement errors caused by structural distortion, swelling, or hematoma on the fractured side, all measurements were performed on the contralateral side. The gluteus medius and gluteus minimus muscles were assessed at the level inferior to the sacroiliac joint, the gluteus maximus at the level of the acetabular roof, and the quadriceps femoris and medial thigh muscles at the level inferior to the lesser trochanter. These muscles were manually outlined on cross-sectional CT images (window width 350 HU; level 40 HU) with a slice thickness of 5 mm using a commercially available workstation (Advantage Windows Workstation 4.4; GE Healthcare, Milwaukee, WI, USA), as illustrated in Fig. 2. The skeletal muscle CSA were measured using Hounsfield unit thresholds of − 29 to 150 HU, as previously described [11, 12]. The radiodensity of each muscle was defined as the average Hounsfield unit value within the outlined region. All measurements were independently performed by two trained physicians, and inter-observer reliability was evaluated using the intraclass correlation coefficient (ICC).

Fig. 2.

A 60-year-old woman with a left femoral neck fracture, in whom the contralateral (right) uninjured side was used for muscle cross-sectional area (CSA) and radiodensity measurements. Axial computed tomography measurement planes. A Measurement of the gluteus medius and gluteus minimus muscles at the level inferior to the sacroiliac joint. B Measurement of the gluteus maximus muscle at the level of the acetabular roof. C Measurement of the quadriceps femoris and medial thigh muscles at the level inferior to the lesser trochanter

Statistical analysis

An a priori power analysis using G*Power 3.1 [13] indicated that a sample size of 159 (53 per group) was required to detect a medium effect (f = 0.25) for one-way ANOVA with α = 0.05 and power = 0.80. Our final sample of 279 patients (89, 101, and 89 in the three groups respectively) exceeded this threshold, confirming adequate power to detect medium-sized differences among the three groups.

Measurement variables are expressed as mean ± standard deviation. The Kolmogorov–Smirnov test was used to assess normality, and the Levene test was used to examine homogeneity of variance. One-way analysis of variance was applied for group comparisons when variables met assumptions of normality and homogeneity of variance, whereas the Kruskal–Wallis H test was used for variables that did not meet these assumptions. When statistically significant differences were identified, post-hoc pairwise comparisons were conducted using Tukey’s HSD test following one-way analysis of variance and Dunn’s procedure with Bonferroni adjustment for non-parametric data. Categorical variables are presented as number (percentage), and the chi-square test was used to compare categorical variables such as sex and the prevalence of hypertension and type 2 diabetes among groups. A two-tailed p value of < 0.05 was considered statistically significant. Inter-observer reliability was evaluated using the ICC. ICC values were interpreted as poor (< 0.40), fair (0.41–0.60), good (0.61–0.75), or excellent (> 0.75). All statistical analyses were conducted using SPSS software version 27.0 (IBM Corp., Armonk, NY, USA).

Results

There were no significant differences in age (mean ± SD), sex distribution, or the prevalence of hypertension and type 2 diabetes among patients with non-displaced femoral neck fractures, displaced femoral neck fractures, and intertrochanteric fractures (all P > 0.05) (Table 1).

The ICCs for CSA, and radiodensity of the peri-hip muscles are summarized in Table 2. The ICC for the radiodensity of the gluteus minimus muscle was 0.734, indicating good reliability, while all other ICC values exceeded 0.75, indicating excellent reliability. Therefore, all data were included in subsequent analyses.

Table 2.

Interobserver reliability analysis

	ICC(95% CI)	P-value
Gluteus maximus muscle
CSA	0.979	˂0.001
Radiodensity	0.967	˂0.001
Gluteus medius muscle
CSA	0.860	˂0.001
Radiodensity	0.948	˂0.001
Gluteus minimus muscle
CSA	0.872	˂0.001
Radiodensity	0.734	˂0.001
Quadriceps femoris muscle
CSA	0.905	˂0.001
Radiodensity	0.902	˂0.001
Medial thigh muscles
CSA	0.884	˂0.001
Radiodensity	0.892	˂0.001

ICCs were classified as poor (˂ 0.40), fair (0.41–0.60), good (0.61–0.75), or excellent (> 0.75)

ICC Intraclass correlation coefficient, CI Confidence interval, CSA Cross-sectional area

The mean values of CSA and radiodensity for each peri-hip muscle according to fracture type are detailed in Table 3. Significant differences were observed among the three groups in radiodensity of the gluteus maximus, gluteus medius, gluteus minimus, and medial thigh muscles, but not in the quadriceps femoris muscle (P = 0.001, P < 0.001, P < 0.001, P = 0.010, and P = 0.212, respectively). By contrast, there were no significant differences in CSA of any of the muscles across the groups (all P > 0.05).

Table 3.

Mean values of CSA and radiodensity of the studied muscles in patients, categorized by fracture group

	ND-FNF (n = 89)	D-FNF (n = 101)	ITF (n = 89)	P-value
Gluteus maximus
Radiodensity	30.22 ± 11.43	26.85 ± 11.15	23.21 ± 13.48	0.001a
CSA	3081.63 ± 693.16	3065.62 ± 718.39	3018.76 ± 698.68	0.864a
Gluteus medius
Radiodensity	41.05 ± 10.63	36.67 ± 10.10	34.36 ± 12.83	˂0.001a
CSA	2209.40 ± 439.66	2257.25 ± 469.23	2227.11 ± 460.57	0.814a
Gluteus minimus
Radiodensity	34.96 ± 15.55	28.33 ± 16.40	23.99 ± 22.46	˂0.001a
CSA	748.11 ± 250.11	765.43 ± 250.72	811.44 ± 352.24	0.313b
Quadriceps femoris
Radiodensity	43.73 ± 9.47	41.99 ± 8.24	41.48 ± 9.25	0.212b
CSA	2090.54 ± 544.04	2004.91 ± 547.53	2165.48 ± 692.15	0.326a
Medial thigh muscles
Radiodensity	38.73 ± 8.83	35.42 ± 7.67	35.40 ± 8.72	0.010b
CSA	3105.35 ± 711.01	3086.34 ± 715.41	3222.84 ± 808.37	0.595a

ND-FNF Non-displaced femoral neck fracture, D-FNF Displaced femoral neck fracture, ITF Intertrochanteric fracture, CSA Cross-sectional area

Data are presented as mean ± standard deviation

^aKruskal–Wallis test

^bAnalysis of variance

Post-hoc pairwise comparisons (Table 4) revealed that radiodensity of the gluteus maximus, gluteus medius, gluteus minimus, and medial thigh muscles was significantly lower in patients with intertrochanteric fractures than in those with non-displaced femoral neck fractures (P = 0.001, P < 0.001, P < 0.001, and P = 0.024, respectively). Similarly, patients with displaced femoral neck fractures showed significantly lower radiodensity in the gluteus medius, gluteus minimus, and medial thigh muscles than did patients with non-displaced fractures (P = 0.008, P = 0.006, and P = 0.020, respectively). However, there were no significant differences in radiodensity between those with displaced femoral neck fractures and intertrochanteric fractures (all P > 0.05).

Table 4.

Post-hoc pairwise comparisons of radiodensity among groups (P < 0.05)

Radiodensity	ND-FNF vs. D-FNF (P value)	ND-FNF vs. ITF (P value)	D-FNF vs. ITF (P value)
Gluteus maximusa	0.115	0.001	0.253
Gluteus mediusa	0.008	0.000	1.000
Gluteus minimusa	0.006	0.000	1.000
Medial thigh musclesb	0.020	0.024	1.000

ND-FNF Non-displaced femoral neck fracture, D-FNF Displaced femoral neck fracture, ITF Intertrochanteric fracture

^aDunn’s procedure with Bonferroni adjustment

^bTukey’s HSD

Discussion

This study examined the association between CT-based characteristics of peri-hip muscles and the type and severity of hip fractures. Specifically, we compared CSA and radiodensity of the gluteus maximus, gluteus medius, gluteus minimus, quadriceps femoris, and medial thigh muscles among patients with non-displaced femoral neck fractures, displaced femoral neck fractures, and intertrochanteric fractures. Our findings showed that radiodensity of the gluteus maximus, gluteus medius, gluteus minimus, and medial thigh muscles was significantly higher in patients with non-displaced fractures than in both patients with displaced femoral neck fractures and those with intertrochanteric fractures. These findings demonstrate an association between fatty infiltration of these muscles and the type and severity of hip fractures.

The incidence of hip fractures is known to be strongly influenced by both sex and age, with higher rates in female patients and a sharp increase among older adults [1]. In our study, there were no significant differences in the distribution of sex or age among the three groups, allowing us to reasonably exclude these factors as potential confounders.

Also, the three fracture groups were comparable in terms of prevalence of hypertension and type 2 diabetes, suggesting that both comorbidities do not appear to have systematically biased the observed differences in muscle radiodensity. However, the absence of medication use for these conditions—particularly antihypertensive and antidiabetic drugs, which might have effects on muscle metabolism—represents a potential source of residual confounding due to the retrospective design and incomplete medical records.

Myosteatosis, or fatty infiltration within skeletal muscle, is characterized by the pathological accumulation of fat within muscle tissue and is commonly observed in older adults. It consists of two primary depots: intramyocellular lipid, located within muscle fibers, and intermuscular adipose tissue, found between fibers and within the surrounding fascia [14]. As summarized by Kalinkovich and Livshits [15] and Ahn et al. [16], skeletal muscle fatty infiltration contributes to mitochondrial dysfunction, insulin resistance, and localized inflammation, all of which lead to muscle atrophy and functional decline. Additionally, fatty infiltration promotes a shift in muscle fiber composition from fast-twitch type II fibers to slower type I fibers, thereby reducing muscle contractile strength and response speed [16]. Consequently, muscle fatty infiltration—negatively correlated with muscle quality—acts as a key pathological driver of sarcopenia rather than merely a coexisting feature.

In elderly outpatients, muscle strength has been identified as a more critical determinant of standing balance than muscle mass [17]. Specifically, the gluteus medius and gluteus minimus muscles play essential roles in stabilizing the pelvis during walking [18]. From a biomechanical perspective, fracture risk during a fall is determined by both the point of impact and the velocity of impact, which influence the type and severity of injury, respectively [19]. Moreover, skeletal muscle functions as a natural shock absorber, dissipating mechanical energy and thereby protecting bones and joints from excessive stress [20, 21]. Therefore, better muscle quality around the hip not only supports postural stability but also enables protective adjustments during a fall, helping to prevent direct impact or alter the point of contact. Additionally, well-developed peri-hip muscles act as a cushioning layer, effectively absorbing and distributing impact forces, thereby reducing the mechanical load on the hip joint. These mechanisms may therefore contribute not only to differences in fracture type but also to less severe fracture outcomes in the event of a fall.

The gluteal and medial thigh muscles act as antagonistic partners across all planes of motion, and their dynamic opposition between abduction–adduction and internal–external rotation is crucial for maintaining hip control and stability [22]. Linhart et al. [23] found that fatty infiltration was significantly more pronounced in the gluteus medius and minimus than in the gluteus maximus among adults of advanced age, possibly because of the decreased demand for hip abduction during daily activities in this population. Building on these findings, we hypothesize that disproportionate fatty infiltration of the gluteal and medial thigh muscles may disturb the mechanical balance between hip abduction–adduction and rotation, which could be associated with an increased likelihood of displaced fractures following trauma.

Skeletal muscle fatty infiltration is inversely associated with CT-measured radiodensity. Goodpaster et al. [24] reported that for every 1 g/100 mL increase in lipid content, mean radiodensity decreased by approximately 1 Hounsfield unit .

Our results showed that the gluteus maximus, gluteus medius, gluteus minimus, and medial thigh muscles in patients with non-displaced fractures had significantly higher radiodensity—indicating less fatty infiltration and better muscle quality—than those in patients with either displaced or intertrochanteric fractures, between whom radiodensity did not differ significantly. Based on the mechanisms discussed above, mild fatty infiltration may reflect better preservation of gluteal and medial thigh muscle quality and coordinated function. Consistent with this, our findings demonstrate that mild fatty infiltration is associated with non-displaced femoral neck fractures rather than displaced femoral neck or intertrochanteric fractures.

Whereas previous research has identified a relationship between psoas muscle fatty infiltration and hip fracture types [10], our study provides a more comprehensive evaluation of key peri-hip functional muscle groups. By incorporating a detailed analysis of femoral neck fracture displacement status, we identified specific associations between muscle degeneration and both fracture type and severity. These findings offer new biomechanical insight into fracture mechanisms and may inform future research on the development of predictive tools for hip fracture type and severity. Furthermore, by using conventional hip CT scans with a narrowed field of view that excludes the lumbar spine, we achieved lower radiation exposure while enhancing the clinical relevance and applicability of our results.

Our study has several limitations that should be acknowledged. First, as a retrospective, single-center analysis, causal relationships cannot be inferred due to the inherent limitations of this design. Thus, the findings demonstrate only statistical associations between peri-hip muscle radiodensity and fracture type/severity. Second, despite adjusting for available covariates, residual confounding from unmeasured factors—such as lifestyle habits (e.g., smoking, physical activity), medication use, and nutritional status—cannot be excluded. Third, accurate height and weight measurements could not be consistently recorded in medical records for all patients because of their limited ability to stand or communicate effectively at admission. So body mass index (BMI) data were not available, precluding adjustment for overall adiposity. Specifically, interpreting CSA without adjustment for BMI may be limited because these two parameters are closely interrelated. Although our findings revealed no significant differences in the CSA of peri-hip muscles among the three fracture groups, future studies that normalize CSA by body mass index are needed to confirm this result. Fourth, bone mineral density (BMD) measurements were not routinely performed in the clinical setting and could not be obtained; therefore, it remains possible that the observed associations are confounded by underlying bone fragility. Fifth, although all CT scans were acquired using standardized protocols on GE Healthcare scanners (Discovery 750HD and LightSpeed VCT), the use of two different systems may have introduced subtle interscanner variability. Sixth, this study faced challenges in selecting an appropriate control group. Ideally, controls would consist of individuals who experienced similar fall mechanisms but did not sustain a fracture; however, such a cohort is difficult to systematically identify and recruit in clinical settings. Seventh, unlike femoral neck fractures, intertrochanteric fractures were not subclassified, which may have obscured potential associations between muscle fatty infiltration and specific intertrochanteric fracture subtypes. Collectively, these limitations underscore the need for cautious interpretation. Future prospective, multicenter studies would be valuable to provide a more definitive analysis. Such a study should ideally include a control group, detailed fracture classification, and systematic collection of anthropometric data.

Conclusions

In conclusion, CT-measured radiodensity of the gluteus maximus, gluteus medius, gluteus minimus, and medial thigh muscles differs significantly among hip fracture types and severity. Specifically, higher radiodensity (reduced muscle fatty infiltration) was associated with non-displaced femoral neck fractures compared with displaced femoral neck or intertrochanteric fractures.

Abbreviations

CT

Computed tomography

CSA

Cross-sectional area

ICC

Intraclass correlation coefficient

Authors’ contributions

Conceptualization: RBH, CYZ, HJM. Data curation: HJM, YYZ, XJY, ZTP, MPL, XQX. Formal analysis: HJM, RBH. Investigation: all authors. Methodology: RBH, CYZ, HJM. Project administration: RBH, CYZ. Software: HJM, YYZ, XJY. Supervision: RBH, CYZ. Validation: HJM, RBH. Visualization: HJM. Writing-original draft: HJM. Writing-review & editing: all authors.

Funding

Not applicable.

Data availability

The de-identified datasets generated or analyzed during the study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the ethical standards in the 1964 Declaration of Helsinki. It was approved by the Ethics Committee of Shantou University Medical College (No.B-2025-214) and the requirement for informed consent was waived.

Consent for publication

All the authors approved the final manuscript and agreed to publish it.

Competing interests

The authors declare no competing interests.