Abstract
Background:
Fatty atrophy of hip abductors is negatively associated with surgical outcomes in abductor repairs; however, there is no standardized system for its classification.
Purpose:
To (1) create a simple, reliable magnetic resonance imaging (MRI) classification system for fatty infiltration specific to hip abductors and (2) examine the correlation between fatty infiltration grades and hip pathology.
Study Design:
Cross-sectional study; Level of evidence, 3.
Methods:
We retrospectively identified 200 hips from 100 patients who underwent bony pelvic MRI with no infection, fracture, tumor, or history of hip surgery. Fatty infiltration of the gluteus medius (GMed) and gluteus minimus (GMin) of both hips was graded on MRI. Grades consisted of a whole number (–2 to +2) corresponding to muscle thickness compared with mean values; a decimal number (0 to 0.4) representing various ratios of fat-to-muscle area; and a letter A (anterior), P (posterior), B (both), or N (none) indicating fat localization. Demographic information and details of MRI-identified hip pathology were collected for each patient. Kappa coefficients were calculated to determine intra- and interrater reliability. Descriptive data were compiled, and Wilcoxon rank sums were used to determine significant differences in grades based on pathology.
Results:
For both GMed and GMin, median muscle thickness was ≤1 SD of the reference value. In most cases, fat content fell between 0% and 50% in GMin and 0% and 25% in GMed. Fat was predominantly localized to the anterior portion of both muscles. Intra- and interrater reliability was moderate to substantial in all grading categories. Higher grades for fat content in the GMin and GMed correlated with tendon tear, tendinosis, and intra-articular cartilage loss (GMin: P < .001, P = .002, P < .001; GMed: P = .03, P = .001, P = .005, respectively). Higher fat content grades in the GMin were also associated with labral tear (P = .03), and in the GMed with greater trochanteric bursitis (P = .02). There was a positive association between width of the GMed and incidence of tendon tear, tendinosis, and intra-articular cartilage loss (P = .006, P = .006, P = .047, respectively).
Conclusion:
This newly described classification system offers a straightforward method of grading fatty atrophy in hip abductor muscles and demonstrates correlation with various hip pathologies.
The hip abductors, made up of the gluteus medius (GMed) and minimus (GMin), serve a crucial role in pelvic stability during stance and gait.1,14 Pathology of these muscles—ranging from abductor tendinosis to partial- or full-thickness tears (with or without retraction)—is increasingly recognized as the etiology of greater trochanteric pain syndrome and dysfunction of the hip joint.4,8,9,21,24,29,34 Patients who suffer from tendinosis or tears often present with chronic pain and tenderness over the greater trochanter that can be exacerbated with walking, climbing stairs, or lying on the affected hip. Additionally, patients may experience a limp or weakness with hip abduction.24,26,28,29 Magnetic resonance imaging (MRI) is an effective modality for confirming diagnosis of abductor pathology.10,20,21,29
Treatment of abductor pathology begins with conservative measures including activity modification, formal physical therapy, and use of oral anti-inflammatory medications. The next step in management is often a greater trochanteric bursal corticosteroid injection. If nonsurgical options fail to resolve pain, surgical repair may be necessary.7,24,29 Abductor repair surgeries improve patient range of motion and coordination of activities of daily living, increase abduction strength, reduce pain, and overall increase patient satisfaction.7,16,19,28,33 While patient-reported satisfaction is generally high postoperatively, reduced preoperative quality of the muscle, graded by degree of fatty infiltration, is associated with poor surgical outcomes.5,27 Therefore, to determine which patients will benefit from surgery, it is important to understand the level of fatty infiltration that exists in the muscle, and a hip abductor muscle–specific classification system is needed.
Though researchers have graded the severity of hip abductor tears for surgical technique planning, a system to quantify the quality of their abductor muscles does not exist, requiring the use of systems created for other joints (such as the Goutallier/Fuchs system) to provide a general idea. 17 In the shoulder, the Goutallier/Fuchs classification system qualifies the rotator cuff musculature in the setting of rotator cuff tears. It utilizes MRI to grade fatty infiltration and has been proven to consistently correlate with surgical repair outcomes.6,15 However, the lack of a standardized parallel targeted to hip abductors makes it difficult to generate informed decisions, plans, and predicted outcomes for abductor repairs. While researchers have attempted to apply the Goutallier/Fuchs and similar systems for fatty infiltration to hip musculature, limitations exist in their reliability, subjectivity, and specificity to hip pathology.13,18,22,36 As such, the purpose of this study is to (1) create a simple, reliable, objective classification system specific for describing hip abductors based on fatty infiltration and (2) examine the correlation between fatty infiltration grades and hip pathology to validate its accuracy in describing overall quality of the joint and surrounding structures.
Methods
For an overview of study steps, see Figure 1.
Flowchart of study methodology. GMed, gluteus medius; GMin, gluteus minimus; MRI, magnetic resonance imaging.
Patient Population
Our institution's electronic medical record was queried for imaging studies titled “MRI bony pelvis w/o contrast” from the year 2018 (to allow for adequate follow-up, including subsequent imaging), yielding 346 results. Patient names, medical record numbers, and MRI dates were extracted. Studies were excluded in patients who had previously undergone hip surgery or had existing fracture, tumor, or infection. Images were randomly reviewed until we obtained 100 bony pelvic MRIs that did not meet any exclusion criteria and which contained an axial, T1-weighted image with bilateral gluteal muscles in view.
Image Analysis
Pelvic MRI scans, capturing bilateral hips, were performed on a 1.5-T magnetic resonance scanner with 5.0-mm slice thickness. The first author (M.N.B.), trained by a musculoskeletal radiologist (D.G.B.), performed the measurement analysis using axial T1-weighted images. Six different measurements of the GMed and GMin bilaterally were taken using picture archiving and communication system (PACS) annotation, totaling 24 measurements per patient (Table 1). For each muscle, (1) thickness of muscle (Figure 2A) and (2) thickness of fat was measured in the slice where the muscle was the thickest. Then, the slice with the most fat was used to measure (3) thickness of fat (Figure 2B), (4) depth of fat (Figure 2B), (5) thickness of muscle (Figure 2C), and (6) depth of muscle (Figure 2C). Thickness was measured by drawing a line across the width of the region of interest, perpendicular to the medial edge. Depth was measured by drawing a line from the anterior-most tip to the posterior-most tip of the region of interest (Figure 2). For muscles with no focal areas of fat, but only fatty streaks, the number of fatty streaks was recorded.
Measurements Taken on Each Muscle (GMin and GMed Bilaterally) to Grade Its Respective Fatty Atrophy a
GMed, gluteus medius; GMin, gluteus minimus.
Axial T1-weighted pelvic magnetic resonance imaging. Example measurements for (A) muscle thickness in the slice where muscle is thickest (M1), (B) fat thickness and depth in the slice where fat is the thickest (M3 and M4, respectively), and (C) muscle thickness and depth in the slice where fat is the thickest (M5 and M6, respectively) in the right gluteus minimus (GMin). Each of these measurements was repeated for both the right and the left GMin and gluteus medius.
After recording the above measurements, separate grades for GMed and GMin were assigned based on our newly described WHAb (Wisconsin Hip Abductor) classification system. This grading system is defined by 3 categories: muscle thickness, fat content, and fat localization of the GMed and GMin (Table 2). First, the values of muscle thickness at the thickest slice (M1) were standardized to mean values for GMed and GMin in women (41.62 ± 8.57 mm and 20.77 ± 3.85 mm, respectively) and men (51.17 ± 6.00 mm and 24.39 ± 2.84 mm, respectively). The means were obtained from a reference, healthy population described in a study by Belzunce et al. 3 Using the standardized values, thickness was graded on a scale from −2 to +2 based on standard deviations above or below the mean as outlined in Table 2. Fat content was represented by decimal numbers, ranging from 0.0 (no fatty atrophy or <6 fatty streaks) to 0.4 (76%-100% fat). The degree of fatty infiltration was determined using the ratio of fat-to-muscle area ([M3*M4]/[M5*M6]) in the slice where fat was the thickest. Measurements of fat thickness in the slice where muscle was thickest (M2) were not used in fatty atrophy calculations. Last, fat localization was graded with a letter; A represented fat in the anterior portion of the muscle, P the posterior portion of the muscle, B both, and N neither.
Criteria of the WHAb Classification System for Fatty Atrophy of Hip Abductors a
WHAb, Wisconsin Hip Abductor.
Putting it together in an example, a WHAb GMed grade −2.1P translated to fatty atrophy localized in the posterior portion of the muscle, with muscle volume >2.5 SD below normal, and 1% to 25% fat content or ≥6 fatty streaks. To increase ease of use of this system, we created an open access, online tool (https://ortho.wisc.edu/abductor-calculator/) (Figure 3) that automatically calculated a grade based on measurement inputs (M1, M3, M4, M5, and M6). 37
Classification system open access online calculator. Standardized muscle thickness and fat percentage are automatically calculated based on measurement inputs, and grades are assigned by the program using these calculations. This example uses the measurements from Figure 1 for a female gluteus minimus (GMin). Because standardized muscle thickness is ≤1 SD of the mean (–0.85 for female), fat content is between 26% and 50% (31.334%), and fat was in the anterior portion of the muscle, this GMin received a grade of 0.2A.
In a subset of 20 patients, the image analysis and grading steps were repeated at a separate time point, both by the first author and by a second author (M.R.T.) trained in the system. These grades were subsequently used to determine the intra- and interrater reliability of the system.
Data Collection
Following the classification grade assignments for all patients, electronic medical records were reviewed for further information on demographics and hip pathology. Included in the demographic data were sex, age, and body mass index (BMI). Then, indication for pelvic MRI, whether a unique hip MRI sequence was obtained, and positive or negative existence of a labral tear, intra-articular cartilage loss, tendon tear (including partial- or full-thickness tears, with or without retraction), greater trochanteric bursitis, or tendinosis were recorded. These conditions were chosen to represent common pathologies in and around the hip joint that may cause, or be affected by, degradation in muscle quality.31,38 They have been well-documented as having a high prevalence on MRI in adults and thus serve as common findings on otherwise “normal” MRIs.
Statistical Analysis
Demographics were compiled in the form of frequency (%), mean ± SD, and median (IQR). A combination of t tests for continuous variables and chi-square or exact Fisher tests for categorical variables were used to analyze differences between female and male patients included in the study. Correlation tables were created to examine the strength of relationships between different variables of interest, and mixed model regression was used to determine if age or sex had a significant effect on the outcomes of interest. The effect of fatty streaks on labral tears was evaluated with mixed logistic regression. Intra- and interrater reliability was assessed using weighted Cohen kappa coefficients. Finally, Wilcoxon rank sums were used to determine if grade values were significantly different for those who did and did not have specified hip pathologies. All analysis was done using SAS, Version 9.4 (SAS Institute). A P value <.05 was considered statistically significant in all instances.
Source of Funding
A portion of the funding for this project was from the Herman and Gwendolyn Shapiro Foundation to support medical student research opportunities.
Results
Patient Population
The final patient population is described in Table 3 and included 68 female and 32 male patients. The mean age of the cohort was 49 years, and mean BMI was 27.99 kg/m2. Selected hip pathologies for analysis were labral tear, intra-articular cartilage loss, tendon tear, greater trochanteric bursitis, and gluteal tendinosis. Among 100 study patients, 79% (79/100) had ≥1 of these pathologies. More specifically, 32% of hips had labral tears, 24% had cartilage loss, 30.5% had tendinosis, 24.5% had bursitis, and 11.5% had tendon tears.
Demographic and Hip Pathology Data for Patients Included in Study (N = 100 patients) a
Data are presented as n (%) or mean ± SD. Bold values were considered statistically significant at P < .05. BMI, body mass index.
Muscle and Fat Measurements
Initially, the raw muscle and fat measurements were analyzed for trends between age groups and differences between sexes (Table 4). For muscle thickness, raw width measurements were significantly greater in men and women, with a difference of 2.25 mm in the GMin (P = .003) and 5.20 mm in the GMed (P < .0001). This relationship was mirrored in the normalized width measurements (GMed, P < .001; GMin, P = .008). There were no significant differences in GMed or GMin thickness between age groups. When looking at fat content, female GMin had a 9% greater fat proportion than that of male patients (P < .001), but there was no significant difference in GMed. It was additionally determined that fat increased with age. In the GMin, fat width increased by 0.10 mm per year and depth increased by 0.25 mm per year (P < .001), while in the GMed, fat width increased by 0.25 mm per year and depth increased by 0.48 mm per year (P < .001).
Raw Muscle Measurement Data Stratified by Sex (N = 200 hips) a
Data are presented as median (IQR). Bold values were considered statistically significant at P < .05. GMed, gluteus medius; GMin, gluteus minimus.
To validate the categorization of fatty streaks, the association between number of streaks and hip pathology was assessed using labral tear as the primary outcome, as labral tears are well-documented to be highly prevalent in hip MRIs and have shown association with gluteal muscle atrophy.25,31 The odds of the patient having a labral tear increased as the number of fatty streaks increased (odds ratio [OR], 1.4). At 6 fatty streaks, the probability of having a labral tear became greater than the probability of not having a labral tear.
Grade Analysis
The overall median grade for muscle thickness was zero in the GMin (IQR, −1 to 0]) as well as in the GMed (IQR, −1 to 0). Fat content showed an overall median grade of 0.2 in the GMin (IQR, 0.1-0.2) and zero in the GMed (IQR, 0 to 0.1). Grade distributions broken down by sex are displayed in Figure 4. Women had increased odds of higher grades in each muscle for both muscle thickness (GMin: OR, 7.84; 95% CI, 3.03 to 20.31; versus GMed: OR, 11.07; 95% CI, 4.21 to 29.12) and fat content (GMin: OR, 6.55; 95% CI, 2.64 to 16.22; versus GMed: OR, 3.14; 95% CI, 1.19 to 8.26) categories. Last, fat was most often localized to the anterior portion of the muscle for both the GMin and the GMed in male and female patients.
Distribution of grades differentiated by patient sex for muscle thickness in (A) gluteus minimus (GMin) and (B) gluteus medius (GMed) and fat content in (C) GMin and (D) GMed.
In nearly all categories, grades showed moderate to substantial agreement between and within raters. Comparing grades between the 2 authors, weighted kappa coefficients were 0.79 (95% CI, 0.69-0.88) for muscle thickness, 0.78 (95% CI, 0.64-0.92) for fat content, and 0.47 (95% CI, 0.17-0.76) for localization. Comparing grades assigned by the same rater at separate time points, weighted kappa coefficients were 0.66 (95% CI, 0.51-0.81) for muscle thickness, 0.71 (95% CI, 0.60-0.83) for fat content, and 0.38 (95% CI, 0.12-0.63) for localization.
The overall distribution of grades in GMin and GMed for hips with and without pathology is displayed in Figure 5. The association between grade and hip pathology was separately assessed for each grade category and hip condition to determine the validity of the classification system. Analysis of muscle thickness showed that a thicker GMed was positively associated with incidence of tendon tear, tendinosis, and cartilage loss (P = .006, P = .006, P = .05). Thickness of GMin did not show any significant associations. Higher fat content grades correlated with tendon tear, tendinosis, and cartilage loss in both the GMin and the GMed (GMin: P < .001, P = .002, P < .001; GMed: P = .048, P < .001, P = .005). Higher fat content grades in the GMin were additionally associated with labral tear (P = .03), while the same relationship was found between the GMed and bursitis (P = .02). Hips with fat localized in the posterior or both portions of the GMin muscle were more likely to be pathologic (all P values < .02). Cartilage loss displayed an additional association with fat localized in the posterior or both portions of the GMed (P = .001).
Distribution of grades for muscle thickness, fat content, and localization between hips without and with pathology in the gluteus medius (GMed) and gluteus minimus (GMin). A, anterior; B, both; N, none; P, posterior.
Discussion
Currently, a standardized system for grading fatty infiltration of hip abductors does not exist, and those that have been tried are subjective, complex, and lack specificity to sex and hip anatomy. This makes it difficult to reliably predict surgical outcomes in the setting of abductor muscle atrophy and tendon tears. The objective of this study was to describe such a system that is objective, straightforward, reliable, and accounts for sex-specific hip abductor anatomy and pathology. The classification system outlined uses simple linear measurements that translate easily to a grade based on muscle thickness, fat content, and fat localization. These grades showed moderate-to-substantial intra- and interrater reliability and proved to correlate with various hip pathologies, including labral tear, cartilage loss, tendon tears, tendinosis, and bursitis.
Studies have shown that a higher degree of fatty infiltration in preoperative abductor muscle negatively affects a variety of surgical outcomes including pain, modified Harris Hip Score, Hip Outcome Score, and patient satisfaction.5,27 This degree of fatty infiltration is inherently related to baseline muscle volume, which varies from male to female. 3 Researchers have also demonstrated that the location of fatty infiltration influences the presentation of hip pain. The GMin has a higher degree of fatty infiltration at baseline compared with the GMed,3,30 which explains why one study found fatty infiltration of the GMed to be more strongly associated with symptomatic abductor pathology. 35 Additionally, another study demonstrated that within these 2 muscles, the most severe fatty infiltration grades were localized to the anterior and middle portions of the muscle. 12 Therefore, when discussing the potential benefits of surgery, it is important to consider not only the amount of fatty infiltration, but also the muscle(s) involved, their baseline volume, and the location of fat, as well as consideration of the normative values for female versus male patients. A standardized grading system, as described here, is necessary to better characterize the quality of the abductor muscles. This can then be used to consider norms for a patient based on sex and delineate whether a patient is a suitable candidate for surgical intervention.
Previous attempts have been made to describe fatty infiltration in hip abductors, though each had significant limitations. Many researchers have directly applied the Goutallier system to analyze fatty infiltration as a variable in their hip studies.2,5,11,23,32,35,39 However, this system is based solely on subjective observations, and translation from the shoulder to the hip anatomy leaves room for discrepancy. As such, the inter- and intrarater reliability of the Goutallier system in hips demonstrate significant variability. Among existing literature, kappa coefficients for interrater reliability range from as low as 0.51 to as high as 0.92, and those for intrarater reliability similarly range from 0.61 to 1.00.5,11,13,18,22,32,36,39 In addition to the issue of reliability, there is also a question of its lack of specificity to the hip musculature, the patterns of infiltration in this region, and how the anatomy may vary from female to male patients.
Because of the imperfections of applying the Goutallier system to the hip, there have been other systems proposed, such as the quartile system, Bal and Lowe, radiodensity measurements, and fat-to-muscle thickness ratios.13,18,22,36 The quartile system uses a 0 to 4 scale correlating with normal muscle, 1% to 25% fat, 25% to 50% fat, 50% to 75% fat, and 75% to 100% fat, respectively.13,22 The Bal and Lowe system uses a 0 to 3 scale representing normal muscle, <30% decrease in muscle mass, 30% to 70% fatty infiltration, and >70% fatty infiltration, respectively. 22 While these systems attempted to better quantify fatty infiltration measurements, they still relied on subjective personal estimates of percentage fat infiltration. In contrast, the radiodensity and fat-to-muscle thickness systems worked to limit this subjectivity. The radiodensity system used imaging software to make automatic measurements of radiodensity across the muscle, which were converted to categories of normal muscle, low-attenuation muscle, and adipose tissue and used to grade percentage of fatty infiltration. While it produced better reliability than Goutallier, this system is complicated, time-consuming, and requires additional software. 36 Conversely, the system using fat-to-muscle thickness ratios used a simple measurement of width of fat divided by width of muscle to describe fatty infiltration, which again showed better reliability than the Goutallier system. However, this system still lacks the anatomical and sex specificity important to determine if the fatty infiltration is physiologic or demonstrating injury that can be resolved with surgery. 18
The system described here addresses most limitations of existing systems. It uses simple measurements to objectively quantify fatty infiltration and muscle thickness and is the only system that accounts for sex-specific variations in muscle volume and physiologic versus pathologic localization of fat in the hip. We also ensured ease of use with our open-access classification calculator website (https://ortho.wisc.edu/abductor-calculator/) (Figure 3) that requires only MRI measurements to provide a grade. However, reliability testing showed variable results. While there was generally substantial agreement across muscle-thickness and fat content categories, weighted kappa coefficients for localization were much lower on average, likely due to the very low prevalence of B, P, or N grades.
The grades from this system followed similar trends to what has been previously described. We found that overall, fatty infiltration increased with age, a greater degree of fatty infiltration was noted in female compared with male patients, and higher grades for fatty infiltration in both the GMin and GMed correlated with existence of hip pathology. Fatty infiltration was more common in the GMin and most often localized to the anterior portion of the muscle, varying from a few streaks to large, localized masses. While prior literature has not looked specifically at number of fatty streaks, we were able to determine a cutoff number of 6 fatty streaks that correlated with a greater likelihood of pathology, hence the classification of ≥6 fatty streaks in a higher grading category. We additionally found that the small sample of patients who had fat in the posterior portion or throughout the entirety of the muscle (68 of 400 muscles) had a higher frequency of hip pathology. This is likely due primarily to the larger volume of fat, and therefore higher fatty infiltration grade, in these patients. Last, we noted that a thicker GMed was associated with a greater incidence of hip pathology, specifically tendon tears and tendinosis in either the GMed or the GMin and cartilage loss in the joint, that has not been reported previously. This correlation could potentially be the result of GMed compensation in the case of pathology of the surrounding muscles. While these sex trends and pathology correlations do not represent the overall goal of the system, they were used as a marker to validate its functionality in describing the overall quality of the joint and surrounding structures.
Strengths and Limitations
The most important strengths of this study were its large sample size and robust data collection. With 100 MRIs and 200 hips, this is the highest-powered study on classification of hip fatty infiltration in the current literature. Additionally, by using a cohort of patients that varied in age, sex, and hip health, we were able to grade a wide variety of fatty infiltration for the initial description of this system. Finally, we collected a diverse assortment of measurements to describe hip muscle quality in a way that has not been done before. However, this study is not without limitations. First, intra- and interrater reliability were assessed on a small subset of patients with only 2 reviewers, showing variable results. We anticipate that the use of PACS system tools for straightforward measurement would limit error and increase precision of the measurements used to grade images, and a large-scale reliability study will be carried out at the conclusion of this initial proposal to better determine the reliability of the system. Second, there was no assessment of the correlation between preoperative grades and postoperative outcomes, which is the ultimate goal of creating such a system. Our patient population was purposely selected to represent a general control group for the initial description of this classification system. Therefore, while we were not able to enroll a cohort of entirely asymptomatic individuals to undergo pelvic MRI, and many patients were indicated for imaging based on hip, gluteal, or back pain, we intentionally excluded patients who were evaluated as surgical candidates or had other major pathological or anatomical disturbances such as infection or tumor.
Future Directions
Future research should complete a larger-scale analysis of intra- and interrater reliability of the system, using more evaluators and sample images. An additional study should evaluate a population of patients who underwent abductor tendon repair to examine the relationship between preoperative grades and postoperative outcomes.
Conclusion
This newly described classification system offers a straightforward method of grading fatty atrophy in hip abductor muscles and demonstrates correlation with various hip pathologies.
Footnotes
Final revision submitted December 2, 2025; accepted December 7, 2025.
One or more of the authors has declared the following potential conflict of interest or source of funding: A.M.S. is a paid consultant for Stryker. AOSSM checks author disclosures against the Open Payments Database (OPD). AOSSM has not conducted an independent investigation on the OPD and disclaims any liability or responsibility relating thereto.
Ethical approval for this study was obtained from the University of Wisconsin–Madison Internal Review Board (No. 2023-0683).
