Introduction
Hip fractures in older adults represent a significant consequence of osteoporosis, characterised by high morbidity, mortality and disability rates.1–3 These fractures manifest as two primary types: femoral neck fractures (FNF) and trochanteric fractures (TRF), each necessitating distinct treatments and associated with varying clinical outcomes.4 For instance, FNFs are linked to higher incidences of femoral head necrosis and non-union compared with TRFs while TRFs may carry greater mortality risks.5 6 Therefore, understanding the differences between these fracture types is crucial. Previous studies have identified factors such as bone structure, spatial distribution and femoral bone mineral density (BMD) as associated with fracture types.7–9 However, conclusive evidence regarding disparities between FNFs and TRFs remains insufficient.
With advancing age, the progressive loss of muscle composition and function significantly impairs balance in older adults, thereby increasing the risk of falls. Despite this, only a limited number of studies have explored differences in muscle parameters between these two types of hip fractures. Importantly, these studies did not account for hip BMD in their comparisons, despite BMD reduction being widely recognised as a key contributor to hip fractures.10 11 Therefore, further investigation into the relationship between muscle properties and hip fracture types is warranted.
In this cross-sectional study, using a cohort of older women with hip fractures who underwent hip CT scans immediately after injury, we aimed to examine differences in hip muscle area and density between patients with femoral neck (FN) and TRF. We hypothesised that CT-based measurements of gluteal muscle density and area could contribute to the classification of hip fracture types in older women, independent of BMD considerations.
Materials and methods
Study design and participants
From January 2012 to December 2019, a total of 1134 consecutive patients aged over 65 years with diagnosed hip fractures were enrolled in this study (figure 1). At our institution, CT scans are standard practice for individuals presenting with suspected or confirmed hip fractures in the emergency department. Fractures were categorised as either FNF or TRF based on CT images interpreted by an experienced musculoskeletal radiologist. Following the CT examination, patients or their relatives completed a one-page questionnaire capturing demographic data (eg, age, gender, height and weight), details of the fall (timing, location and mechanism), fracture history and medical background.
Schematic flow diagram illustrating the stepwise exclusion of subjects with hip injuries. FNF, femoral neck fractures; TRF, trochanteric fractures.
Inclusion criteria for hip fracture patients mirrored those outlined by Wang et al,12 specifically women who sustained hip fractures due to low-energy injuries and underwent hip CT scans within 48 hours. 48 hours was chosen as the cut-off to minimise the influence of disuse atrophy on the measures of muscle size, and this project focused on women as there were not enough men in the sample to do a meaningful between-sex comparison to account for muscle and bone density differences between the sexes. Exclusion criteria encompassed individuals with prior hip fractures, conditions preventing standing or walking and metabolic or inflammatory diseases affecting muscle quality and bone density.
The study followed Strengthening the Reporting of Observational Studies in Epidemiology guidelines for reporting observational studies.
CT acquisition and quantitative CT analysis
CT scans of both hips for all study participants were performed using the Toshiba Aquilion spiral CT scanner (Toshiba Medical Systems Division, Tokyo, Japan). Subjects were scanned in a supine position, with a solid calibration body model (Mindways Software, Austin, Texas, USA) positioned just below the hips. Scans encompassed from the top of the acetabulum to 3 cm or more below the lesser trochanter (TR), covering the proximal femur. Scan parameters included 120 kVp, 125 mAs, 50 cm field of view, 512×512 matrix, 1 mm reconstructed slice thickness and a standard reconstruction kernel with filtered back projection. Following the CT scan, images were automatically uploaded to the Mindways quantitative CT (QCT) workstation.
CT X-ray absorptiometry technique (CTXA V.4.2.3, Mindways, Austin, Texas, USA) is a QCTPro scan analysis module for the hip that generates a two-dimensional image from three-dimensional CT images of the proximal femur. The measurement procedure has been previously described in detail.13 In summary, it divides the proximal femur into three regions of interest (ROIs): the FN, TR and intertrochanter (IT), which correspond to standard ROIs commonly used in DXA hip scans. This allows for the calculation of areal BMD (aBMD, g/cm2) results for each ROI, as well as a combined measurement of all three, equivalent to the total hip (TH) ROI. The aBMD of the FN and TH were calculated from the hip CT scans using CTXA. Hip BMD measurements were conducted on the healthy side for all patients.
Muscle cross-sectional area and density assessments
OsiriX software (Lite V.12.0.2, Pixmeo, Geneva, Switzerland) was used for analysis. The muscle measurement procedure and precision have been previously documented.14 Two investigators, trained by an expert radiologist in CT muscle imaging, conducted all muscle measurements, and their respective averages were obtained. The muscle measurement results demonstrated high intraobserver agreement (intraclass correlation coefficients, ICC: 0.932–0.998, p<0.001) and inter-bserver consistency (ICC: 0.913–0.961, p<0.001), with investigators blinded to each other’s analyses during the imaging analysis.
Figure 2 illustrates the measurement of cross-sectional area and density of the gluteus maximus (G.MaxM) at the level of the greater TR, and the gluteus medius and minimus muscles (G.Med/MinM) at the level of the third sacral vertebra (S3). Due to potential muscle oedema and bleeding on the fractured side, which could influence the cross-sectional area and CT value measurements of the muscles, thus not accurately reflecting their prefracture state, muscle parameters were measured exclusively on the non-fractured side.
(A) Measurement of the gluteus medius and minimus muscles at the third sacral (S3) level. (B) Measurement of cross-sectional area and mean CT values of the gluteus maximus at the level of the greater trochanter of the femur. Muscle regions are highlighted in red.
Statistical analysis
Data were presented as means and SDs for parametric variables while categorical variables were described using frequencies and percentages. The χ2 test assessed differences between groups for categorical variables, and Student’s t-test was used for continuous variables. Age was stratified using a cut-off of 80, the mean and median age of the sample, to explore age-specific relationships between muscle parameters and fracture type. Logistic regression models were employed, both with and without adjustments for age, body mass index (BMI) and THaBMD. Generalised additive models were also used to further explore dose–response relationships between muscle densities, areas and probabilities of TRF, adjusting for the aforementioned covariates. All analyses were conducted by using R V.4.1.1 (The R Foundation, http://www.R-project.org). A two-tailed test was applied, and significance was set at p<0.05.
Patient and public involvement
Patients and the public did not participate in the design or conduct of this study.
Results
Characteristics of subjects
Figure 1 illustrates the recruitment of study participants. Out of 1134 low-trauma hip fracture patients, 580 cases were excluded. Notably, 215 subjects imaged more than 48 hours after hip fracture were excluded due to prolonged immobilisation. A total of 554 hip fracture subjects were eligible for further analysis, comprising 314 FNF cases and 240 TRF cases. Table 1 presents the distribution of relevant demographic data for these subjects. The FNF group was significantly younger and taller, with higher gluteus muscle area and density, as well as higher aBMD in the TH and FN regions. Participants were then stratified into two subgroups using an age cut-off of 80, yielding largely similar results (table 1).
Characteristics of subjects who sustained femoral neck or trochanteric fractures grouped by age
Associations of muscle size and density variables with TRF
All area and density measurements, except for G.Med/MinM density, were significantly associated with TRF after adjusting for age and BMI (table 2, column 4). These associations remained significant after further adjustment for THaBMD (table 2, column 5). G.Med/MinM density (adj. OR 0.98, 95% CI 0.95 to 1.01) showed a marginal association with TRF after adjustments for age, BMI and THaBMD.
ORs of having a TRF per 1 SD of variables
Relationship between muscle variables and age
Additionally, a stronger relationship between gluteus muscles and TRF was observed in the younger group (age <80) compared with the older group (age >80) (table 2). After adjustment, all associations of gluteus muscles remained statistically significant in the younger group (age <80) (G.Med/MinM area, OR 0.96, 95% CI 0.92 to 0.99; G.Med/MinM density, OR 0.95, 95% CI 0.91 to 0.98; G.MaxM area, OR 0.94, 95% CI 0.91 to 0.98; G.MaxM density, OR 0.95, 95% CI 0.92 to 0.99) (p<0.01, table 2). Figure 3A–D visualises the relationship between muscle parameters and the risk of TRF. It shows a clear decreasing trend in the risk of TRF as the area or density of the gluteal muscles (both G.maxM and G.Med/MinM) increases. However, it should be noted that with increasing values on the x-axis, the number of samples for individuals over 80 years old decreases significantly. Therefore, the trends described in this figure should be interpreted with caution.
Relationship between the density and area of the gluteal muscles and the risk of trochanteric fractures(adjusted for age and body mass index). Bar chart includes a dividing line above which represents the sample size of two subgroups.
Discussion
In this cross-sectional study, CT images were used to collect data on the density and area of hip muscles in acute low-energy hip fracture women. Our findings highlight that in older women, particularly those under 80 years of age, both the area and density of the gluteus muscles were significantly associated with TRF. Even after adjusting for THaBMD, these associations persisted, although attenuated for most muscle parameters.
Muscle density, measured by CT as the mean attenuation of skeletal muscle in Hounsfield units (HU), has been extensively employed in research15–18 to assess muscle quality. Low tissue HU (indicating low muscle density) may signify lipid or fluid infiltration in skeletal muscles, potentially accompanied by functional changes.19 Wang et al demonstrated that muscle density outperforms aBMD derived from hip CTXA and muscle size in distinguishing between individuals with and without hip fractures.12 Lang et al observed trends towards lower hip muscle CSA and reduced lean tissue muscle HU (indicative of greater fatty infiltration) in subjects with hip fractures compared with controls.20 Subsequently, Lang et al reported that decreased thigh muscle HU is associated with an elevated risk of hip fracture.21 These studies collectively underscore the critical role of muscle density in evaluating physical function and fracture risk.22
We hypothesised that gluteal muscle density and area play a role in classifying hip fracture types in older women. The G.MaxM, situated superficially in the gluteal muscle, primarily functions in hip extension and external rotation, with its upper part also contributing to hip abduction.23 24 The anterior upper portion of the G.Med muscle lies beneath the skin while its posterior lower part lies deep to the G.MaxM. Its main function involves hip abduction, with the anterior bundle rotating the hip joint internally and the posterior bundle externally.24 25 The gluteus MinM, located deep to the G.Med muscle, function similarly to the G.Med in hip abduction. Therefore, this study analysed these two muscles collectively. Erinç et al reported that the areas of the G.Med and MinM were higher in the FNF group compared with the TRF group, although there was no significant difference in atrophy scores between subjects with TRF versus FNF.10 Our study found that women older than 65 years in the TRF group exhibited smaller G.Med/MinM areas than those in the FNF group, consistent with the findings of the aforementioned study. Importantly, this difference remained statistically significant after adjusting for age, BMI and THaBMD. Furthermore, G.MaxM density and size were independently associated with the risk of TRF in women older than 65 years, regardless of hip aBMD. Similarly, Wang et al demonstrated that G.MaxM density significantly correlates with physical performance in older women, even after adjusting for age, height and weight.14 This study underscores the significant role of the G.MaxM muscle in hip fracture risk assessment.
Interestingly, after we grouped patients by age 80, the difference in muscle parameters between the two fracture types in the over 80 years old group was no longer statistically significant after adjustment of covariates. However, in the 65–80 age group, muscle parameters, especially G.MaxM, were more strongly related to TRF. The explanations for the age effect on muscle parameters with the risk of TRF were unclear. Hip fracture women aged over 80 years seem to be especially frail with low BMD, low cortical thickness and low muscle quality, thus, we speculated that the incidence of hip fracture type might be a random event.
Strengths and limitations
To our knowledge, this study is the first to utilise an age cut-off of 80 to stratify and investigate the age-specific relationship between G.MaxM and G.Med/MinM area and density with hip fracture type. Additionally, we rigorously excluded subjects imaged more than 48 hours after hip fracture, enhancing the reliability of our bone and muscle measurements. Prolonged immobility or reduced activity following a fracture can exacerbate muscle atrophy, rendering muscle area or CT values measured post-48 hours less reflective of the muscle state at or before the fracture. Furthermore, our study calibrated several factors in binary logistic regression, including BMD, an essential factor that had been overlooked in previous relevant studies.
This study possesses several notable limitations. First, its cross-sectional design warrants future longitudinal cohort studies to further explore the relationship between gluteal muscles and fracture types over time. Second, our decision to measure the healthy side instead of the fractured side introduces potential bias. However, this approach was taken to mitigate the impact of factors like fracture, bleeding and oedema on muscle parameter accuracy. Future advancements in technology, as suggested by Cheng et al,26 may offer improved symmetry assessment of the hip joint sides. Lastly, our study is inherently limited by its exclusive focus on older female patients with fractures, which limits generalisability to older males. This gender-specific focus was driven by a predominance of female cases in our dataset. Given known sex differences in muscle characteristics, combining datasets into a unified cohort for analysis was deemed inappropriate, necessitating our concentration on the larger female patient sample. Future research should strive to address this limitation by recruiting a more balanced cohort encompassing both genders, thereby broadening the applicability and robustness of findings concerning skeletal muscle health in older patients following fractures.
Conclusions
In conclusion, our study demonstrates that in older women, particularly those under 80 years of age, gluteus muscle parameters are associated with TRF. Age-related loss of muscle mass is a well-known risk factor for hip fractures. Therefore, preserving muscle mass and minimising fat infiltration in muscles may be crucial in preventing TRF in this demographic, especially those under 80 years.
This post was originally published on https://bmjopen.bmj.com