Subchondral stress fractures of the femoral head associated with military training: Clinical features and analysis

Abstract

BACKGROUND

Subchondral stress fracture of the femoral head (SSFFH) is an uncommon femoral-head injury associated with repetitive mechanical loading and may mimic osteonecrosis of the femoral head (ONFH), particularly when collapse is present. Military personnel represent a clinically relevant population in whom early differentiation is important.

AIM

To characterize the clinical and imaging features of SSFFH associated with military training and identify features that help distinguish SSFFH from ONFH, with particular attention to the relationship between structural collapse and clinical status.

METHODS

We retrospectively reviewed 242 inpatient records of military patients with hip pain or femoral-head injury admitted between June 2013 and June 2019. After exclusion of 68 records, 174 patients (220 hips) were included: 127 patients with SSFFH (139 hips) and 47 patients with ONFH (81 hips). Clinical characteristics, exposure history, laboratory data, radiographs, magnetic resonance imaging (MRI) findings, and SSFFH stage were analyzed. Two experienced musculoskeletal radiologists independently reviewed imaging while blinded to clinical information; disagreements were resolved by consensus. All primary comparative analyses were performed at the patient level, with one index hip per patient for stage-based analyses. Statistical analyses were performed using SPSS version 27.0.

RESULTS

Compared with ONFH, patients with SSFFH were younger [median interquartile range: 20.0 (19.0-21.0) years vs 23.0 (20.5-29.0) years; P < 0.001], had lower body mass index [21.28 (20.07-22.54) kg/m² vs 23.17 (21.20-25.79) kg/m²; P < 0.001], and more often had unilateral involvement (90.6% vs 27.7%; P < 0.001). Creatine kinase was higher in SSFFH [95.0 (79.0-136.5) U/L vs 87.0 (54.0-109.5) U/L; P = 0.002], whereas total cholesterol was lower (3.86 ± 0.72 mmol/L vs 4.33 ± 1.01 mmol/L; P = 0.004). In multivariable binary logistic regression, bilateral involvement was associated with higher odds of ONFH [adjusted odds ratio (OR) = 21.188, 95%CI: 6.168-72.784; P < 0.001], whereas each increase in training-intensity category was associated with lower odds of ONFH (adjusted OR = 0.120, 95%CI: 0.045-0.319; P < 0.001); alcohol exposure contributed to the model overall (P = 0.019). Across SSFFH Stages I-III, symptom duration, Visual Analog Scale score, Harris Hip Score, range of motion, bone marrow edema, joint effusion, and Tönnis grade differed significantly. These stage comparisons were cross-sectional and do not demonstrate individual longitudinal improvement.

CONCLUSION

In young military personnel with unilateral femoral-head subchondral abnormalities occurring in the setting of repetitive training and with fewer conventional ONFH risk factors, SSFFH should be considered. MRI is central to recognition and differentiation. In this cohort, femoral head collapse was not consistently associated with greater pain or worse function; therefore, collapse alone may not be sufficient to determine the need for arthroplasty. Treatment should incorporate symptoms, functional impairment, structural damage, treatment response, and longitudinal clinical and imaging findings. All stage-associated differences were cross-sectional observations and do not establish spontaneous lower current symptom burden or a fixed natural history.

Key Words: Subchondral stress fracture; Femoral head; Osteonecrosis; Military personnel; Magnetic resonance imaging; Retrospective study

Core Tip: This retrospective study compared subchondral stress fracture of the femoral head (SSFFH) with osteonecrosis of the femoral head in military personnel. SSFFH was associated with younger age, lower body mass index, higher training intensity, predominantly unilateral involvement, and less corticosteroid/alcohol exposure. Within SSFFH, Stage II had the greatest pain and functional limitation, whereas Stage III showed more advanced structural changes but lower pain and better function. Because these comparisons were cross-sectional, collapse alone should not automatically determine the need for arthroplasty; treatment should integrate symptoms, function, structural damage, treatment response, and longitudinal findings.

INTRODUCTION

Subchondral stress fracture of the femoral head (SSFFH) is a rare disease that has been mentioned in many reports over the past few decades[1-3]. This fracture type predominantly occurs in the elderly population, where it is termed subchondral insufficiency fracture. In contrast, reports in young individuals with a history of excessive physical activity—termed SSFFH—remain limited. SSFFH shares certain overlapping characteristics with osteonecrosis of the femoral head (ONFH), particularly in imaging findings[4,5]. In both conditions, early-stage hip pain presents with normal plain radiographs, while later stages may show subchondral fracture collapse and demarcation sclerosis—features classically associated with ONFH[6,7]. However, SSFFH with femoral head collapse can exhibit similar radiographic manifestations[3].

Histopathological examination can provide definitive confirmation of subchondral fracture pathology, but biopsy of femoral-head lesions is not routinely feasible in this population. Accordingly, the present study used predefined magnetic resonance imaging (MRI)-based criteria, blinded independent image review, and clinical correlation for retrospective classification. The lack of histopathological confirmation is acknowledged as a limitation.

Currently, awareness of SSFFH remains insufficient, particularly regarding this distinct hip condition resulting from military training injuries. When femoral head collapse occurs, SSFFH is frequently overlooked, with clinicians more commonly attributing the findings to ONFH, femoral head trauma, or osteochondral injury.

This study retrospectively analyzed electronic medical records of patients with femoral head injuries admitted to our hospital from June 2013 to June 2019. By thoroughly reviewing MRI characteristics and clinical features, we aimed to improve recognition of SSFFH as a distinct stress fracture caused by military training. A central question we sought to address was whether femoral head collapse in SSFFH carries the same ominous prognosis as in ONFH—a question with direct implications for clinical decision-making regarding arthroplasty vs hip-preserving management.

MATERIALS AND METHODS

Study population and eligibility criteria

Inclusion criteria: Hospitalized patients with femoral head injuries from June 2013 to June 2019 who had complete electronic medical records.

Exclusion criteria: Incomplete electronic medical records (n = 5), traumatic femoral head osteochondral fractures (n = 8), transient osteoporosis (n = 6), rapidly destructive arthropathy (n = 2), femoroacetabular impingement syndrome (n = 41), and other non-stress-fracture/non-ONFH conditions (n = 6).

Study design and data collection

A retrospective analysis was conducted on military patients with femoral head injuries admitted from June 2013 to June 2019. A total of 242 inpatient records were screened; 68 were excluded, leaving 174 patients (220 hips) for analysis: 127 patients with SSFFH (139 hips) and 47 patients with ONFH (81 hips) (Figure 1).

Figure 1 Participant flow diagram. A total of 242 inpatient records were screened. Sixty-eight records were excluded: Incomplete electronic medical records (n = 5), traumatic femoral head osteochondral fractures (n = 8), transient osteoporosis (n = 6), rapidly destructive arthropathy (n = 2), femoroacetabular impingement syndrome (n = 41), and other non-stress-fracture/non-osteonecrosis of the femoral head (ONFH) conditions (n = 6). The final analysis included 174 patients (220 hips): 127 patients with subchondral stress fracture of the femoral head (139 hips) and 47 patients with ONFH (81 hips). Diagnostic-group classification was based on blinded magnetic resonance imaging review. SSFFH: Subchondral stress fracture of the femoral head; ONFH: Osteonecrosis of the femoral head.

Treatment exposure was heterogeneous and was not assigned according to a standardized study protocol. Patients received extracorporeal shock wave therapy and/or arthroscopic femoral head drilling decompression according to clinical judgment. In the ONFH group, 3 patients underwent total hip arthroplasty. In the SSFFH group, 6 patients with severe osteoarthritis underwent total hip arthroplasty, 2 underwent intertrochanteric rotational osteotomy, and 1 underwent iliac bone grafting. Clinical and laboratory variables were abstracted from inpatient electronic medical records, and the analyzed dataset contained complete observations for all listed variables. Laboratory values were treated as cross-sectional measurements because their timing relative to training and symptom onset was not standardized retrospectively. Treatment exposure was not used to infer causal effects on stage, pain, or function, and the study was not designed to evaluate the efficacy of a specific treatment strategy.

Imaging acquisition

All patients underwent X-ray and MRI examinations (Figures 2, 3, 4, and 5). X-ray examinations included anteroposterior and frog-leg views of both hips. MRI was performed using either a 1.5T or 3.0T system with the following parameters: T1-weighted images, repetition time/echo time = 400-675/8-19 ms, field of view = 35-68 cm, matrix = 220-512 × 256-512, and slice thickness = 3-5 mm. All patients were examined in the coronal plane with the hip joint in a neutral position.

Figure 2 Representative radiographs and T1-weighted magnetic resonance images of subchondral stress fracture of the femoral head and osteonecrosis of the femoral head. A-F: Imaging findings in a 20-year-old male recruit with subchondral stress fracture of the femoral head (SSFFH) (A-C) and a 45-year-old military officer with osteonecrosis of the femoral head (ONFH) (D-F). A and D: Radiographs show similar articular-surface collapse in both SSFFH and ONFH; the white triangles indicate subchondral fracture; B and C: Magnetic resonance imaging (MRI) of SSFFH reveals an irregular subchondral low-signal band (white arrow) with surrounding bone marrow edema and an irregular, thickened sclerotic margin; E and F: MRI of ONFH demonstrates a smoother, serpiginous low-signal band demarcating the necrotic segment, with a subchondral fracture indicated by the white triangle. These findings illustrate the potential overlap on radiographs and the added value of MRI for differential classification.

Figure 3 Representative X-ray and magnetic resonance imaging findings of subchondral stress fracture of the femoral head across the predefined imaging stages. A and E: Stage I: Mild articular-margin depression with localized magnetic resonance imaging bone marrow edema; B and F: Stage I: Mild articular-margin depression with an indistinct fracture line and joint effusion; C and G: Stage II: Obvious articular depression with a clear subchondral fracture line, marked effusion, and extensive bone marrow edema; D and H: Stage III: Femoral-head flattening and advanced degenerative change with less prominent bone marrow edema. The staging framework was adapted from previously described imaging features[6,9].

Figure 4 Illustrative longitudinal case of subchondral stress fracture of the femoral head in a 20-year-old recruit. A-D: Initial Stage II imaging demonstrates a right femoral-head subchondral fracture with marked joint effusion and bone marrow edema; E-H: At 6-month follow-up, the femoral head shows collapse and degenerative change, while joint effusion is markedly reduced. This case illustrates that structural progression may coexist with reduction of selected magnetic resonance imaging fluid-related abnormalities; it is illustrative and does not establish the natural history of the cohort.

Figure 5 Illustrative longitudinal case of subchondral stress fracture of the femoral head in a 21-year-old recruit. A-D: Initial Stage II imaging demonstrates a left femoral-head subchondral fracture with marked joint effusion and bone marrow edema; E-H: At 17-month follow-up, the femoral head shows collapse and degenerative change, while joint effusion is markedly reduced. This case illustrates a longer interval between examinations and should not be interpreted as evidence of a uniform disease trajectory.

Clinical and imaging assessment

Demographic and clinical data included age, sex, symptom duration, military training history, trauma history, alcohol consumption, and corticosteroid exposure. Training intensity was recorded from training logs as hours per week and categorized as low (≤ 30 hour/week), moderate (31-45 hour/week), or high (> 45 hour/week), coded as 1, 2, and 3, respectively. This variable was not used for imaging diagnosis; MRI-based SSFFH/ONFH classification was performed while readers were blinded to training data, which were analyzed only as group associations. Clinical status was assessed using the Visual Analog Scale (VAS), Harris Hip Score, hip flexion-extension range of motion (ROM), and internal-external rotation ROM.

Radiographic evaluation: The Tönnis scoring system was used to assess X-ray images for subchondral fractures, femoral head collapse, loss of spherical contour, and osteoarthritis severity[8]. MRI evaluation focused on bone marrow edema, joint effusion, and articular cartilage collapse, based on previously reported criteria[6,9-11].

Diagnostic workflow: SSFFH and ONFH were classified primarily on the basis of predefined MRI features, independently of clinical risk factors. Two senior musculoskeletal radiologists independently reviewed all MRI images while blinded to age, occupation, military rank, training intensity, corticosteroid exposure, alcohol history, and other clinical information. Disagreements were resolved by consensus. Interobserver agreement was assessed using Cohen’s kappa statistic, yielding κ = 0.86 (95%CI: 0.78-0.94), indicating excellent agreement. After imaging classification, clinical information was reviewed for contextual confirmation and comparative analysis. Because histopathological confirmation was unavailable for the retrospective cohort, the imaging-based classification should be regarded as an operational clinical research definition rather than a histologically validated gold standard.

Diagnostic criteria for SSFFH[6,8-10]: On T1-weighted imaging, an irregular hypointense band parallel to or convex toward the articular surface and surrounded by bone marrow edema was considered characteristic in the early stage; incomplete irregular discontinuity of the cartilaginous surface was considered a mid-stage feature; and an irregular articular surface with reduced bone marrow edema signal represented advanced disease. On T2-weighted imaging, hyperintensity below the articular cartilage surface, heterogeneity, and hypointensity were considered early-, mid-, and late-stage features, respectively (Figures 2, 3, 4, and 5). The SSFFH staging criteria were adapted from the classification system originally described by Song et al[6] and modified according to Iwasaki et al[9]. Stage 0 was retained as an imaging category for early MRI abnormalities; no patient in the present cohort met the predefined Stage 0 criteria. The unified staging criteria are shown in Table 1 and Figure 3.

Table 1 Staging and imaging features of subchondral stress fracture of the femoral head.

	Plain radiographs	Magnetic resonance images
Stage 0	No definite findings	Bony edema limited to the femoral head area
Stage I	Mild depression of articular margin (Crescent sign)	Low signal intensity lesion on T1 image and joint effusion
Stage II	Severe depression of articular margin	Subchondral fracture and a large joint effusion with edema extending to below the trochanter
Stage III	Arthritic changes (bone spurs or cysts)	Osteoarthritis with receding bone marrow edema

Diagnostic criteria for ONFH: On T1-weighted imaging, a smooth, serpiginous low-signal band that is continuous and concave toward the articular surface (map-like appearance) was considered characteristic (Figure 2). Clinical risk factors, including corticosteroid exposure and alcohol history, were reviewed only after imaging-based classification and were not used as the primary imaging criterion.

Alcohol consumption classification: Heavy drinking was defined as > 400 mL ethanol per week; moderate as 100-400 mL per week; light as ≤ 100 mL per week[12].

Corticosteroid intake classification: High dose (e.g., prednisone > 1 mg/kg/day); moderate dose (prednisone 0.5-1 mg/kg/day); low dose (prednisone < 0.5 mg/kg/day). While no precise standard exists for corticosteroid-induced osteonecrosis risk stratification, underlying diseases and corticosteroid administration are recognized to increase ONFH risk[13].

Trauma energy classification: High-energy trauma (e.g., motor vehicle accidents, crush injuries); medium-energy trauma (e.g., contact sports, bicycle accidents); low-energy trauma (e.g., falls from standing height or tripping).

Joint effusion grading[14]: Grade 0: No obvious effusion; Grade I: Mild fluid accumulation, limited to one side of the femoral head and neck; Grade II: Moderate fluid accumulation on both sides of the femoral head and neck, without significant joint capsule expansion; Grade III: Severe fluid accumulation around the femoral head or neck, with significant joint capsule and crypt expansion.

Grade of bone marrow edema[15]: Grade 0: No edema in the proximal femur; Grade 1: Edema limited to the femoral head; Grade 2: Edema involving the femoral head and neck; Grade 3: Edema extending from the femoral head to below the trochanter.

Tönnis classification[8]: Grade 0: Normal hip; Grade 1: Sclerosis of the femoral head and acetabulum, slight joint space narrowing, mild osteophyte formation; Grade 2: Femoral head or acetabular cysts, moderate joint space narrowing, moderate loss of femoral head sphericity; Grade 3: Large cysts within the femoral head or acetabulum, severe joint space narrowing or obliteration, severe femoral head deformity.

Of note, the radiographic appearances of SSFFH and ONFH on plain X-ray are often similar and difficult to distinguish (Figure 2).

Statistical analysis

Statistical analyses were performed using SPSS version 27.0. Normality was assessed with the Shapiro-Wilk test and homogeneity of variance with Levene’s test. Approximately normally distributed continuous variables were summarized as mean ± SD and compared using Student’s independent-samples t test when variances were homogeneous or Welch’s t test when they were not. Nonnormally distributed continuous variables and ordinal clinical outcomes were summarized as median (interquartile range) and compared using the Mann-Whitney U test. Categorical variables were summarized as n (%) and compared using Fisher’s exact test for 2 × 2 tables or the Fisher-Freeman-Halton exact test for larger tables because several cells were sparse. All primary SSFFH-vs-ONFH comparisons were conducted at the patient level, and there were no missing observations in the analyzed variables.

For comparisons among SSFFH Stages I-III, one index hip per patient was used; in patients with bilateral disease, the more severely affected side was selected. Continuous variables were analyzed using one-way ANOVA for normally distributed data or the Kruskal-Wallis H test for non-normally distributed data; categorical variables were analyzed using the χ² or Fisher’s exact test; and ordinal variables, including BME grade, effusion grade, and Tönnis grade, were analyzed using the Kruskal-Wallis H test. When an overall ANOVA was significant, Bonferroni correction was applied for post-hoc pairwise comparisons, with P < 0.017 considered statistically significant.

A multivariable binary logistic regression model was fitted with diagnostic group as the dependent variable (ONFH = 1, SSFFH = 0). Variables with P < 0.10 in univariable analyses were considered for model entry. Corticosteroid exposure was excluded from conventional maximum-likelihood regression because moderate/high exposure showed quasi-complete separation (0 SSFFH vs 38 ONFH patients), which produced unstable estimates. Forward likelihood-ratio selection was used (entry P < 0.05; removal P > 0.10). Laterality was coded as unilateral = 0 and bilateral = 1; training intensity was entered as an ordinal variable (1 = low, 2 = moderate, 3 = high), so its odds ratio represents a one-category increase; alcohol exposure was entered categorically with none as the reference category. Model fit was assessed with the omnibus likelihood-ratio and Hosmer-Lemeshow tests. All tests were two-sided, and P < 0.05 was considered statistically significant.

RESULTS

Patient characteristics and SSFFH stage distribution

A total of 174 patients (220 hips) with hip pain were included in this retrospective study. Based on imaging analysis and diagnostic criteria, 127 patients (139 hips) were diagnosed with SSFFH, while 47 patients (81 hips) were diagnosed with ONFH. Among the 127 SSFFH patients, 12 (9.4%) had bilateral involvement. For patients with bilateral involvement, the more severely affected side was selected for patient-level between-group comparisons.

Among the 127 SSFFH patients, imaging-based staging identified 13 patients in Stage I (10.2%), 52 in Stage II (40.9%), and 62 in Stage III (48.8%). The corresponding numbers of affected hips were 23, 47, and 69, respectively, because 12 patients had bilateral disease. For all cross-stage comparisons in Table 2, one index hip per patient was selected; therefore, the analytical sample sizes were Stage I n = 13, Stage II n = 52, and Stage III n = 62. No patient met the predefined Stage 0 criteria. Bilateral involvement rates in the stage groups were 5/13 (38.5%), 5/52 (9.6%), and 2/62 (3.2%), respectively.

Table 2 Comparison of clinical and imaging characteristics across subchondral stress fracture of the femoral head stages at the patient level (one index hip per patient; n = 127), mean ± SD/n (%).

Parameter	Stage I (n = 13 patients)	Stage II (n = 52 patients)	Stage III (n = 62 patients)	P value
Age (year)	21.15 ± 2.76	19.96 ± 1.96	20.56 ± 1.92	0.105
Gender (male/female)	8/5	51/1	61/1	0.058
BMI (kg/m2)	21.13 ± 1.55	21.34 ± 1.94	21.78 ± 2.72	0.455
Time interval (month)	5.74 ± 3.19	7.71 ± 4.78	10.13 ± 6.41	0.005a
Bilateral involvement	5/13 (38.5)	5/52 (9.6)	2/62 (3.2)	0.001a
VAS score	5.84 ± 1.42	7.63 ± 1.42	6.57 ± 1.49	0.001a
Harris score	73.37 ± 7.21	64.15 ± 8.03	70.34 ± 8.26	0.001a
Hip flexion-extension ROM (º)	137.84 ± 9.76	117.78 ± 18.78	130.13 ± 15.98	0.001a
Hip inner-external rotation ROM (º)	68.47 ± 16.69	38.34 ± 21.65	56.43 ± 22.82	0.001a
BME
Grade 0	0	0	0
Grade 1	9	6	26
Grade2	2	17	14
Grade 3	2	29	22	0.001a
Effusion
Grade 0	9	0	0
Grade 1	3	2	30
Grade 2	1	14	26
Grade 3	0	36	6	0.001a
Tönnis Classification
Grade 0	11	0	0
Grade 1	2	42	20
Grade 2	0	10	41
Grade 3	0	0	1	0.001a
Laboratory findings
ALP (U/L)	82.38	77.31	71.84	0.225
CK (U/L)	122.23	107.77	115.39	0.608
TC (mmol/L)	3.9877	3.9071	3.7860	0.528
UA (μmol/L)	424.92	411.81	390.23	0.244

^aP < 0.05.

One index hip per patient (the more severely affected side in bilateral cases) was used for cross-stage comparisons, yielding n = 13, 52, and 62 patients in Stages I, II, and III, respectively. Bilateral involvement is reported at the patient level. ALP: Alkaline phosphatase; BMI: Body mass index; CK: Creatine kinase; ROM: Range of motion; TC: Total cholesterol; UA: Uric acid; VAS: Visual Analog Scale.

Comparison between SSFFH and ONFH

Univariable comparisons between SSFFH and ONFH are shown in Table 3. Patients with SSFFH were younger [median interquartile range (IQR), 20.0 (19.0-21.0) years vs 23.0 (20.5-29.0) years; P < 0.001] and had lower body mass index (BMI) [21.28 (20.07-22.54) kg/m² vs 23.17 (21.20-25.79) kg/m²; P < 0.001]. Training-intensity, alcohol, corticosteroid-exposure, laterality, and Tönnis-grade distributions differed between groups (all P < 0.001). These variables were compared only after blinded imaging-based diagnostic classification.

Table 3 Clinical and laboratory characteristics of subchondral stress fracture of the femoral head and osteonecrosis of the femoral head.

Parameter	SSFFH (n = 127)	ONFH (n = 47)	P value
Age (years)	20.0 (19.0-21.0)	23.0 (20.5-29.0)	< 0.001a
BMI (kg/m2)	21.28 (20.07-22.53)	23.17 (21.20-25.79)	< 0.001a
Sex			0.116
Male	120 (94.5)	41 (87.2)
Female	7 (5.5)	6 (12.8)
Time from symptom onset to hospitalization (months)	6.0 (3.0-10.5)	6.0 (4.0-8.0)	0.736
Single-trauma history			0.703
None	29 (22.8)	10 (21.3)
Low-energy	50 (39.4)	22 (46.8)
Moderate-energy	44 (34.6)	15 (31.9)
High-energy	4 (3.1)	0 (0.0)
Alcohol exposure			< 0.001a
None	74 (58.3)	7 (14.9)
Light	34 (26.8)	12 (25.5)
Moderate	18 (14.2)	16 (34.0)
Heavy	1 (0.8)	12 (25.5)
Corticosteroid exposure			< 0.001a
None	125 (98.4)	6 (12.8)
Low	2 (1.6)	3 (6.4)
Moderate	0 (0.0)	12 (25.5)
High	0 (0.0)	26 (55.3)
Training intensity			< 0.001a
Low	2 (1.6)	18 (38.3)
Moderate	10 (7.9)	19 (40.4)
High	115 (90.6)	10 (21.3)
Laterality			< 0.001a
Unilateral	115 (90.6)	13 (27.7)
Bilateral	12 (9.4)	34 (72.3)
Tönnis grade			< 0.001a
Grade 0	2 (1.6)	3 (6.4)
Grade 1	19 (15.0)	17 (36.2)
Grade 2	44 (34.6)	20 (42.6)
Grade 3	62 (48.8)	7 (14.9)
VAS score	7.0 (6.0-8.0)	7.0 (6.0-8.0)	0.863
Harris Hip Score	68.90 ± 8.82	68.38 ± 7.68	0.724
Hip flexion-extension ROM (°)	133.0 (120.0-140.0)	130.0 (121.0-136.0)	0.457
Hip internal-external rotation ROM (°)	50.0 (30.5-70.5)	50.0 (40.0-70.0)	0.560
ALP (U/L)	71.0 (63.0-85.0)	69.0 (56.5-79.0)	0.225
CK (U/L)	95.0 (79.0-136.5)	87.0 (54.0-109.5)	0.002a
TC (mmol/L)	3.86 ± 0.72	4.33 ± 1.01	0.004a
UA (μmol/L)	402.61 ± 84.83	385.13 ± 81.77	0.225

^aP < 0.05. Data are presented as n (%), mean ± SD, or median (IQR). Median (IQR) and Mann-Whitney U tests were used for age, BMI, time interval, Visual Analog Scale, range of motion, alkaline phosphatase, and creatine kinase. Harris Hip Score and uric acid were compared with Student’s t tests; total cholesterol was compared with Welch’s t test. Categorical variables were compared with Fisher’s exact or Fisher-Freeman-Halton exact tests. Training intensity: Low ≤ 30 hour/week, moderate 31-45 hour/week, high > 45 hour/week. All analyses were patient-level. ALP: Alkaline phosphatase; BMI: Body mass index; CK: Creatine kinase; IQR: Interquartile range; ONFH: Osteonecrosis of the femoral head; ROM: Range of motion; SSFFH: Subchondral stress fracture of the femoral head; TC: Total cholesterol; UA: Uric acid; VAS: Visual Analog Scale.

Unilateral involvement predominated in SSFFH (115/127, 90.6%), whereas bilateral involvement was more common in ONFH (34/47, 72.3%; P < 0.001). Heavy alcohol exposure occurred in 1 SSFFH patient and 12 ONFH patients, and moderate/high corticosteroid exposure occurred in 0 SSFFH patients and 38 ONFH patients (both overall exact-test P < 0.001). Sex and single-trauma history did not differ significantly between groups.

Creatine kinase (CK) was higher in SSFFH [median (IQR), 95.0 (79.0-136.5) U/L vs 87.0 (54.0-109.5) U/L; P = 0.002], whereas total cholesterol (TC) was lower (3.86 ± 0.72 mmol/L vs 4.33 ± 1.01 mmol/L; P = 0.004). Time from symptom onset to hospitalization, VAS score, Harris Hip Score, both ROM measures, alkaline phosphatase (ALP), and uric acid (UA) did not differ significantly (all P > 0.05).

Multivariable analysis differentiating SSFFH from ONFH

In the final forward likelihood-ratio multivariable logistic regression model (Table 4), bilateral involvement was associated with higher odds of ONFH [adjusted odds ratio (OR) = 21.188, 95%CI: 6.168-72.784; P < 0.001], whereas each one-category increase in training intensity was associated with lower odds of ONFH (adjusted OR = 0.120, 95%CI: 0.045-0.319; P < 0.001). Alcohol exposure contributed to the model overall (likelihood-ratio P = 0.019): Compared with no exposure, moderate exposure (adjusted OR = 6.336, 95%CI: 1.334-30.100; P = 0.020) and heavy exposure (adjusted OR = 20.353, 95%CI: 1.675-247.233; P = 0.018) were associated with ONFH, whereas light exposure was not statistically significant. The omnibus model test was significant [likelihood-ratio χ²(5) = 124.18, P < 0.001], and the Hosmer-Lemeshow test did not indicate poor fit [χ²(4) = 3.91, P = 0.418]. Corticosteroid exposure was not included because of quasi-complete separation.

Table 4 Multivariable logistic regression analysis of factors associated with diagnostic group.

Variable	B	SE	Adjusted OR	95%CI	P value
Training intensity (per category increase)	-2.117	0.498	0.120	0.045-0.319	< 0.001a
Bilateral vs unilateral	3.053	0.630	21.188	6.168-72.784	< 0.001a
Alcohol exposure (overall)					0.019a
Alcohol exposure: Light vs none	1.181	0.749	3.259	0.751-14.137	0.115
Alcohol exposure: Moderate vs none	1.846	0.795	6.336	1.334-30.100	0.020a
Alcohol exposure: Heavy vs none	3.013	1.274	20.353	1.675-247.233	0.018a

^aP < 0.05. Outcome coding: ONFH = 1 and SSFFH = 0. Forward likelihood-ratio binary logistic regression was used (entry P < 0.05; removal P > 0.10). Training intensity was coded 1-3 and its odds ratio represents a one-category increase. Alcohol exposure used none as the reference. Corticosteroid exposure was excluded because quasi-complete separation produced unstable maximum-likelihood estimates. Overall alcohol P value is from the likelihood-ratio test; contrast P values are Wald tests. Model n = 174; omnibus likelihood-ratio χ²(5) = 124.18, P < 0.001; Hosmer-Lemeshow χ²(4) = 3.91, P = 0.418. OR: Odds ratio; SE: Standard error.

Clinical and imaging characteristics across SSFFH stages

Significant differences were observed among the three SSFFH stages in multiple parameters (Table 2). The time from symptom onset to hospitalization increased progressively with stage (Stage I: 5.74 ± 3.19 months; Stage II: 7.71 ± 4.78 months; Stage III: 10.13 ± 6.41 months; P = 0.005). Of note, the association between stage and time interval should be interpreted as a correlation rather than evidence that stage is simply a function of time—higher stages may reflect more severe injury or slower recovery. Pain severity (VAS score) was highest in Stage II (7.63 ± 1.42), followed by Stage III (6.57 ± 1.49) and Stage I (5.84 ± 1.42) (P = 0.001). Functional status (Harris Hip Score) was poorest in Stage II (64.15 ± 8.03), with improvement in Stage III (70.34 ± 8.26) compared to Stage II (P = 0.001). ROM in both flexion-extension and internal-external rotation followed a similar pattern: Significantly reduced in Stage II and partially recovered in Stage III (P = 0.001 for both).

Imaging characteristics also varied significantly. BME severity differed across stages (P = 0.001), with Grade 3 BME most frequent in Stage II (29/52) and less frequent in Stage III (22/62). Joint effusion was absent in most Stage I patients, whereas Stage II patients predominantly exhibited moderate to severe effusion (Grades 2-3 in 50/52), and Stage III patients showed mostly mild to moderate effusion (Grades 1-2 in 56/62) (P = 0.001). Tönnis grade also worsened across stages (P = 0.001).

The bilateral involvement rate decreased across the stage groups (Stage I: 5/13, 38.5%; Stage II: 5/52, 9.6%; Stage III: 2/62, 3.2%; P = 0.001). No significant differences were found among stages in serum ALP, CK, TC, or UA (all P > 0.05).

DISCUSSION

SSFFH is an underrecognized stress-type injury that may be encountered in military personnel undergoing intense physical training. This study compared SSFFH with ONFH and examined stage-associated clinical and MRI findings. The principal observations were that SSFFH occurred in a younger, more intensively training population with predominantly unilateral involvement and less corticosteroid/alcohol exposure, and that within SSFFH, the highest pain burden and greatest functional limitation were observed in Stage II rather than Stage III. The latter observation is clinically relevant because it suggests that structural collapse and current symptom severity may not be tightly coupled in SSFFH.

SSFFH was compared with ONFH across clinical phenotype, exposure history, laboratory findings, and stage-associated clinical/imaging characteristics. ONFH was selected as the reference condition because SSFFH may be misclassified as ONFH when femoral-head collapse is present. Importantly, imaging classification was performed before clinical risk factors were reviewed, allowing the clinical comparisons to be interpreted as associations rather than as components of the imaging diagnosis.

Distinct clinical phenotypes

SSFFH and ONFH demonstrated distinct clinical profiles. SSFFH patients were significantly younger and had lower BMI, while bilateral involvement and corticosteroid/alcohol exposure were more common in ONFH. The association between intensive military training and SSFFH is biologically compatible with a stress-fracture mechanism. Importantly, diagnostic classification was based primarily on blinded MRI interpretation, with clinical risk factors incorporated only after imaging classification was completed. Thus, the subsequent comparisons and regression analyses quantified clinical associations rather than defined the imaging diagnosis itself.

Laterality proved to be an important discriminating feature. Unilateral involvement predominated in SSFFH (90.6%), whereas bilateral disease was more common in ONFH (72.3%). This pattern is consistent with localized mechanical loading in SSFFH and systemic risk factors in ONFH. The logistic regression analysis quantified the association between bilateral involvement and ONFH, but this finding represents an association, not a standalone diagnostic criterion.

Biomarker differentiation

Laboratory findings added another dimension to the differentiation. Elevated CK in SSFFH patients likely reflects recent intense muscular activity and muscle microtrauma accompanying bone injury—consistent with the stress fracture mechanism. In contrast, higher total cholesterol in ONFH patients may reflect underlying metabolic disturbances or prolonged corticosteroid exposure, both implicated in osteonecrosis pathogenesis. While not diagnostic in isolation, these laboratory patterns may support clinical decision-making when imaging findings are equivocal.

Pathophysiological mechanisms

The pathogenesis of SSFFH is compatible with established stress-fracture concepts. Repetitive loading and muscular fatigue may increase local mechanical stress, while cumulative microdamage may exceed the capacity for bone remodeling[16,17]. In the present study, the higher training intensity and higher CK levels observed in SSFFH are compatible with this mechanism, but the retrospective design does not permit causal inference.

Stage-associated clinical and imaging findings

The stage-associated findings are descriptive and should not be interpreted as a demonstrated natural history. Stage I patients had relatively mild symptoms and preserved function. Stage II had the highest VAS scores, poorest Harris Hip Scores, greatest ROM restriction, and most prominent joint effusion. Stage III had lower VAS scores and higher Harris Hip Scores than Stage II despite more advanced structural changes. Previous reports have described a generally benign natural history of SSFFH, with symptom improvement during long-term follow-up in some patients[18-21]; however, because the stage comparisons were cross-sectional, these differences do not prove that individual patients improved as disease stage advanced. Variations in disease duration, treatment exposure, patient adaptation, referral patterns, and other unmeasured factors may contribute.

The difference between structural severity and current symptom burden is clinically noteworthy. In ONFH, collapse is an established marker of structural progression and is often associated with worsening joint function. In this SSFFH cohort, however, Stage III patients did not have the highest pain or worst functional scores. This finding suggests that collapse alone may be insufficient to determine treatment intensity or the need for arthroplasty in SSFFH, but it does not establish that collapsed lesions have a favorable long-term prognosis.

The MRI pattern provides a plausible, but not proven, explanation for the cross-sectional clinical differences. BME and joint effusion were most prominent in Stage II and less prominent in Stage III. These findings may reflect reduced acute tissue response in more advanced lesions, but MRI edema and effusion are indirect imaging markers and should not be equated with histologically proven inflammation. Longitudinal studies with serial MRI and clinical assessment are needed to determine whether these changes represent a reproducible temporal sequence.

The lower proportion of bilateral involvement in later stages also requires caution. Because the stage groups were not longitudinally followed as a single cohort, the cross-sectional decrease in bilateral disease cannot be interpreted as spontaneous resolution of contralateral lesions. Only serial imaging of the same patients can determine whether contralateral abnormalities regress, remain stable, or progress.

The imaging findings were also stage-associated. BME Grade 3 and severe effusion were most frequent in Stage II, whereas Stage III was characterized by persistent deformity, sclerosis, more advanced Tönnis changes, and less prominent BME and effusion. Figures 3 and 4 provide illustrative longitudinal cases showing that collapse can coexist with reduced effusion over time; however, these two cases cannot establish the typical natural history of SSFFH for the entire cohort.

Clinical implications of femoral head collapse in SSFFH

A central clinical implication of this study is that femoral head collapse in SSFFH should not automatically be equated with the same prognosis as collapse caused by ONFH[22]. The present cohort showed that patients with Stage III collapse had lower pain scores and better functional scores than Stage II patients. However, because these comparisons were cross-sectional and treatment exposures were heterogeneous, the findings are hypothesis-generating rather than confirmatory. They support careful clinical assessment rather than an automatic arthroplasty decision based on collapse alone.

The evidence therefore supports a more nuanced interpretation: Collapse is an important structural finding, but its clinical significance should be considered together with pain severity, functional limitation, articular degeneration, treatment response, and longitudinal progression. This interpretation avoids directly extrapolating the ONFH treatment paradigm to SSFFH while recognizing that some collapsed lesions may still require surgery.

In the present cohort, Stage III patients had VAS scores of 6.57 ± 1.49 compared with 7.63 ± 1.42 in Stage II (P = 0.001), and Harris Hip Scores of 70.34 ± 8.26 compared with 64.15 ± 8.03 (P = 0.001). These differences indicate that current symptom severity was not directly proportional to radiographic stage. They do not demonstrate spontaneous lower current symptom burden within individuals.

The MRI pattern was also different between stages. Grade 3 BME was present in 29/52 Stage II patients (55.8%) compared with 22/62 Stage III patients (35.5%), while Grade 3 effusion was present in 36/52 Stage II patients (69.2%) compared with 6/62 Stage III patients (9.7%). These findings suggest that the most conspicuous edema and fluid abnormalities were concentrated in Stage II. Because these are indirect imaging markers, they should be interpreted as correlates of tissue response rather than direct measurements of inflammation.

The apparent dissociation between structural deterioration and current symptom burden is clinically relevant. Reduced BME and effusion may accompany lower pain in some patients, while persistent deformity and sclerosis remain structural abnormalities. This pattern differs from the assumption that every structural collapse must be accompanied by progressively worsening symptoms. Nevertheless, long-term joint survival and arthroplasty-free survival were not evaluated in this study.

Accordingly, the present data do not justify a universal recommendation for conservative treatment of collapsed SSFFH. Instead, they support individualized decision-making. In a young patient with collapse but tolerable pain and preserved function, close follow-up, activity modification, protected weight-bearing when appropriate, and hip-preserving strategies may be considered. Total hip arthroplasty may remain appropriate for patients with severe persistent pain, advanced joint destruction, or substantial functional limitation after appropriate nonoperative or hip-preserving treatment.

The therapeutic implications should therefore be framed as hypothesis-generating. Treatment exposure was heterogeneous in this retrospective cohort, and the study was not designed to compare conservative treatment, hip-preserving procedures, and arthroplasty. Future prospective studies should evaluate whether symptom severity and serial MRI findings can better identify patients who can safely avoid arthroplasty after collapse.

Therapeutic implications and applicability of existing guidelines

Existing clinical guidance for femoral head collapse is predominantly derived from ONFH populations and may not be directly applicable to SSFFH[23]. The present findings suggest that SSFFH should be considered a distinct stress-fracture entity and that treatment decisions should integrate symptoms, function, structural damage, and longitudinal progression rather than rely on collapse alone. This is a clinical hypothesis requiring prospective validation, not a replacement for established arthroplasty indications.

Non-surgical management—including protected weight-bearing, activity modification, and appropriate analgesia—remains a reasonable component of care for selected non-collapsed or minimally symptomatic SSFFH[24]. For collapsed lesions, management should be individualized and guided by symptoms, function, articular degeneration, treatment response, and serial assessment. Hip arthroscopy may also assist in the classification and treatment of precollapse subchondral insufficiency fracture of the femoral head associated with intra-articular pathology[25].

Prevention and early detection

Given the young, healthy population at risk, prevention deserves equal emphasis with treatment. Biomechanical studies have identified deep hip flexion under load—as occurs during weighted squats, lunges, and kneeling—as a risk position for anterior femoral head overloading[26,27]. Training modifications that limit such maneuvers, particularly during the early weeks of basic training when skeletal adaptation is incomplete, may reduce SSFFH incidence. Gradual progression of training intensity, adequate recovery periods, and attention to early pain symptoms are equally important.

Early diagnosis requires maintaining a high index of suspicion. Hip pain in a young recruit should prompt consideration of SSFFH, particularly when pain is activity-related and accompanied by antalgic gait. MRI with dedicated hip protocols is the preferred imaging modality, capable of detecting prefracture bone marrow edema before radiographic changes appear.

Limitations and future directions

This study has several limitations. First, its retrospective single-center design introduces potential selection and information bias, and the military population may limit generalizability to civilian athletes. Second, the diagnostic classification was based on predefined MRI criteria rather than histopathological confirmation, although blinded independent review showed excellent agreement. Third, the stage comparisons were cross-sectional and therefore do not establish a longitudinal natural history or spontaneous lower current symptom burden. Fourth, treatment exposures were heterogeneous and were not analyzed as causal determinants of stage or outcome. Fifth, some regression categories were sparse, resulting in wide confidence intervals and limited precision.

Future prospective studies should use predefined imaging-based diagnostic criteria, blinded independent image review, standardized staging, and patient-level longitudinal follow-up. Such studies should prospectively record training exposure, corticosteroid and alcohol exposure, treatment received, pain and functional outcomes, and serial MRI findings. Validation of potential laboratory correlates and development of evidence-based criteria for conservative vs surgical management of collapsed lesions are additional priorities.

CONCLUSION

SSFFH associated with military training predominantly occurred in young personnel and was more often unilateral than ONFH in this cohort. Bilateral involvement and substantial corticosteroid exposure were more common in ONFH. Within SSFFH, Stage II was associated with the greatest pain, functional limitation, ROM restriction, and joint effusion, whereas Stage III showed more advanced structural changes but lower pain and better function. These findings indicate that structural collapse and current symptom severity may be dissociated in SSFFH. Therefore, femoral head collapse alone should not necessarily be regarded as an automatic indication for arthroplasty; decisions should incorporate symptoms, functional impairment, structural damage, treatment response, and longitudinal clinical and imaging findings. Because the stage comparisons were cross-sectional and treatment exposures varied, the study does not prove spontaneous lower current symptom burden or a fixed natural history.