Improved Prediction of Femoral Shortening in Total Hip Arthroplasty for Developmental Dysplasia of the Hip Using the Etfal Classification: A 7-Center Validation Study

Introduction

Developmental dysplasia of the hip (DDH), a common congenital disorder that frequently progresses to symptomatic degeneration and ultimately necessitates total hip arthroplasty (THA), is a major cause of secondary osteoarthritis in adults [1–5]. Unlike THA performed for primary osteoarthritis, arthroplasty in dysplastic hips is considered one of the most challenging procedures in adult reconstructive surgery because of distorted bone morphology, increased soft-tissue tension with heightened neurological risk, hypoplastic acetabula with deficient walls, and high-riding femoral heads, each contributing to substantial anatomic variability on both the acetabular and femoral sides [1,3,4]. In high-riding hips, achieving stable reconstruction requires technically demanding maneuvers, including accurate implant positioning, correction of limb-length discrepancy, restoration of physiologic femoral version, and, when necessary, subtrochanteric shortening osteotomy to prevent sciatic nerve injury during reduction [4,6–8].

Among the conventional systems used to grade DDH severity in adults, the Crowe classification, introduced in 1979 and widely adopted to guide expectations for THA, is the most frequently used [5,9]. Similarly, the Hartofilakidis system primarily functions as a descriptive radiographic tool. Despite their popularity and acceptable interobserver reliability, the major limitation of these systems is their failure to incorporate the key biomechanical determinants of reduction difficulty. The Crowe system, in particular, quantifies only the vertical displacement of the femoral head relative to the Köhler line; it offers no assessment of periarticular soft-tissue tension, nor does it provide insight into acetabular or proximal femoral morphology [2,6,9]. This shortcoming is especially problematic in Crowe type 3 dysplasia, where moderate dislocation may mask substantial mechanical difficulty and leave the surgeon uncertain as to whether subtrochanteric shortening osteotomy is required to achieve a safe neurological and mechanical reduction [1,3,4].

Since the early descriptions of trochanteric osteotomy in the 1970s by Charnley, techniques have evolved into modern subtrochanteric shortening and derotational osteotomies, and recent literature has increasingly focused on more physiologic, ratio-based assessments to determine the true need for osteotomy in various patterns of DDH [2,3,6,7,10]. Many authors agree that the primary determinant of whether subtrochanteric shortening osteotomy is necessary is intraoperative reduction tension, an element not captured by displacement-based systems and therefore not reliably predicted preoperatively using Crowe or other traditional classifications [1,4,8].

This collective limitation has driven the need for more functional radiographic classifications that reflect biomechanical load and soft tissue tension, rather than merely vertical displacement. Ratio-based femoral pelvic indices have been shown to outperform traditional descriptive systems in estimating surgical difficulty [3,6,10]. Building on this rationale, Demirkale et al proposed the Etfal classification, a proportional system based on reproducible femoral and pelvic landmarks, designed to estimate the need for femoral shortening osteotomy before THA in DDH [11]. Although promising, this system was originally derived from single-center data, and its ability to bridge the gap between descriptive classification and real surgical planning has not yet been validated in a broader population sample.

Therefore, this study aimed to determine whether the Etfal classification can more accurately predict the intraoperative requirement for subtrochanteric shortening osteotomy in adults with DDH undergoing THA compared with the traditional Crowe system. Using a multicenter cohort of patients treated with THA for DDH, we evaluated the ability of the Etfal system, relative to the Crowe classification system, to identify hips requiring subtrochanteric shortening osteotomy. We hypothesized that a proportional femoral-pelvic ratio–based classification would provide superior predictive accuracy for the subtrochanteric shortening osteotomy requirement, particularly in moderate dysplasia, and thereby serve as a more reliable tool for preoperative surgical planning.

Material and Methods

STUDY DESIGN AND PATIENT POPULATION:

After obtaining approval from the ınstitutional ethics committee, a retrospective multicenter validity study was conducted. The aim of this study was to evaluate the reliability and clinical utility of the Etfal classification, originally proposed by Demirkale et al, based on 79 hips from 76 patients undergoing THA for high-riding DDH [11].

For this purpose, the Etfal system was applied to adult patients who underwent THA for dysplastic hip-related osteoarthritis between 2018 and 2024 by 7 high-volume arthroplasty surgeons across 7 tertiary referral centers.

The final cohort consisted of 152 patients with 157 hips. Five patients underwent bilateral THA during the study period, and both hips were included in the analysis. Because the classification evaluates radiographic hip morphology, the hip was the unit of analysis.

Owing to the exploratory classification-validation design, no a priori sample size estimation was conducted. All methodological steps adhered to the STROBE guidelines.

INCLUSION AND EXCLUSION CRITERIA:

Demographic data were collected from all patients.

The inclusion criteria were as follows: age >18 years; Crowe type 1 to 4 dysplastic hips with secondary osteoarthritis; primary THA performed; availability of preoperative standing anteroposterior pelvis radiographs; and complete operative documentation including osteotomy details, shortening length, and complication records.

The exclusion criteria were as follows: prior surgery on the ipsilateral hip; history of septic arthritis; incomplete radiographic data; traumatic etiologies; and neuromuscular disorders.

RADIOGRAPHIC MEASUREMENTS AND ETFAL CLASSIFICATION:

The index test was the Etfal classification system. Preoperative radiographs were assessed according to the Etfal system, which quantifies hip displacement based on the ratio of the distance between the tuberculum minus and the tuber ischiadicum to pelvic height (TM–TI divided by P).

All radiographic measurements were performed on standardized standing anteroposterior pelvic radiographs using a calibrated digital measurement software. Anatomical landmarks were defined as the most inferior point of the ischial tuberosity and the most prominent point of the lesser trochanter.

Measurements were performed by 2 experienced orthopedic surgeons at each center and repeated 3 weeks later. Interobserver and intraobserver reliability were assessed using intraclass correlation coefficients (ICC) based on a 2-way random-effects model with absolute agreement. If the interobserver discrepancy exceeded 2 mm, the measurements were repeated, and the final classification was assigned only after consensus on the third measurement.

REFERENCE STANDARD:

The primary endpoint was strictly defined as the intraoperative requirement for subtrochanteric shortening osteotomy, treated as a binary variable (yes/no). Other intraoperative maneuvers, including soft tissue release and controlled femoral cracking, were recorded descriptively but were not included in the primary endpoint and were not used in any diagnostic accuracy analyses. For comparison, the Crowe classification (types 1–4) was assigned using preoperative radiographs. All postoperative complications were documented.

BIAS MITIGATION AND QUALITY CONTROL:

The evaluators were blinded to the surgical findings. Interobserver reliability was quantified using the ICC. Consecutive sampling minimized selection bias, and all participating centers were anonymized to reduce center-specific bias.

STATISTICAL ANALYSIS:

Statistical analyses were performed using the Python programming language. Continuous variables are summarized as mean±SD and categorical variables as n (%). Because 5 patients contributed bilateral hips, the hip was the unit of analysis for all statistical evaluations.

Frequency and proportion analyses were used to assess the relationship between the Etfal and Crowe types and osteotomy requirements.

For the receiver operating characteristic (ROC) analysis, the Etfal and Crowe grades (types 1–4) were encoded as ordinal numerical variables. ROC curves were generated for the Etfal and Crowe classifications using the subtrochanteric shortening osteotomy requirement as the binary reference standard, and area under the ROC (AUC) curve values with 95% CIs were calculated.

Diagnostic thresholds (≥2, ≥3, and 4) were evaluated for sensitivity, specificity, positive predictive value, and negative predictive value. Contingency counts (true positive [TP], false positive [FP], true negative [TN], and false negative [FN]) were used to derive these metrics. For each evaluated diagnostic threshold, full contingency data (TP, FP, TN, and FN) were explicitly reported to ensure reproducibility and allow independent verification. Comparisons between the ROC curves were performed using the DeLong method. Optimal thresholds were determined using the Youden index.

Complication analysis was conducted using a 2×2 association approach, as all events occurred in high-grade dislocations (types 3 and 4). Associations between the presence of complications and each classification threshold were analyzed using contingency tables with the Haldane-Anscombe correction for zero cell values.

Odds ratios (OR) with 95% CIs were calculated to compare the predictive performance between the systems. Because complications were clustered within specific severity groups, logistic regression was not used; instead, effect size differences were illustrated using a forest plot.

Statistical significance was set at P<0.05.

Results

Data from 152 patients with 157 hips across 7 centers met the eligibility criteria. Five patients underwent bilateral THA during the study period. The mean patient age was 52.55±9.32 years (range, 20–74 years), and most patients were women.

According to the Etfal classification, 87 hips (55.4%) were type 1; 28 (17.8%) type 2; 28 (17.8%) type 3; and 14 (8.9%) type 4. The interobserver reliability for the Etfal measurements was excellent (ICC=0.90; 95% CI, 0.84–0.94), and the intraobserver reliability was similarly high (ICC=0.93; 95% CI, 0.88–0.96). According to the Crowe system, 31 hips (19.7%) were type 1; 33 (21.0%) type 2; 33 (21.0%) type 3; and 60 (38.2%) were type 4.

Subtrochanteric shortening osteotomy was performed in 50 of the 157 hips (31.8%). The primary outcome, intraoperative requirement for subtrochanteric shortening osteotomy, increased in a clear and monotonic fashion across Etfal categories (Table 1). No type 1 hip required osteotomy, whereas all type 4 hips required osteotomy. Osteotomy was performed in approximately 33% of type 2 hips and in 96.4% of type 3 hips.

In contrast, the Crowe classification demonstrated weaker alignment with actual surgical need: only 6.1% of Crowe type 3 hips required osteotomy despite being anatomically comparable to Etfal type 3, whereas 80% of Crowe type 4 hips required subtrochanteric shortening osteotomy.

All ROC analyses, AUC calculations, and diagnostic threshold evaluations were performed using the predefined binary primary endpoint. Diagnostic performance analysis showed that the Etfal system substantially outperformed the Crowe system (Figure 1). ROC analysis demonstrated superior discrimination for the Etfal classification (AUC 0.98; 95% CI, 0.96–1.00) compared with the Crowe classification (AUC 0.94; 95% CI, 0.90–0.98).

Examination of the diagnostic thresholds (Table 2) demonstrated that Etfal type 2 or greater had excellent sensitivity and specificity (1.00 and 0.81, respectively). The optimal threshold identified by the Youden index was Etfal type 3 or greater, which demonstrated near-perfect specificity (0.99) and high sensitivity (0.82), representing a strong rule-in threshold. Etfal type 4, although perfectly specific (1.00), was clinically overly strict. The contingency counts used to calculate the diagnostic performance metrics are presented in Table 3. These data allowed for the independent verification of all reported diagnostic performance metrics.

In comparison, Crowe type 2 or greater, type 3 or greater, and type 4 yielded sensitivity and specificity values of 1.00 and 0.29, 1.00 and 0.60, and 0.96 and 0.89, respectively, demonstrating weaker discriminative sharpness relative to the Etfal classification.

Twelve complications were identified, representing 10 distinct complication types. The most frequent events were sciatic nerve palsy (n=2) and trochanteric fracture/fissure (n=3), which were both observed in patients with high-grade dysplasia.

The complication distribution showed marked clustering in Etfal types 3 and 4, whereas the Crowe categories exhibited a more diffuse pattern. In the 2×2 comparison, Etfal type 3 or greater demonstrated a significantly stronger association with complications (OR=14.2; 95% CI, 2.1–96.8), while Crowe type 3 or greater showed a weaker and statistically imprecise association (OR=3.1; 95% CI, 0.4–22.9). These findings demonstrate that the Etfal classification more effectively captures tension-related risks underlying complications, such as nerve palsy, intraoperative fractures, and instability (Figure 2).

Discussion

The most important finding of this study was that the Etfal classification demonstrated markedly superior accuracy in predicting the need for subtrochanteric shortening osteotomy compared with the traditional Crowe classification. This observation aligns with previous reports indicating that the typical reduction challenges and risk of neurological injury associated with THA in DDH cannot be adequately captured by vertical femoral head displacement alone [1,3,6]. Although the Crowe system has historically been the most widely used descriptive staging method for classifying adult DDH, it does not evaluate the femoral pelvic relationship, which becomes critical in high-riding dislocations, in which safe reduction often requires shortening [12,13]. The sharper stratification observed with the Etfal system suggests that proportional femoral pelvic ratios have far greater physiological relevance than displacement measurements, a concept consistent with recent calls for more functional radiological classification models [10,14].

The second major finding of this multicenter study was the clear discrepancy between radiographic severity and actual surgical need observed in Crowe type 3 hips. Despite being categorized as high-riding, only 6.1% of Crowe type 3 hips in this cohort required femoral shortening, whereas nearly all Etfal type 3 and 4 hips (96%–100%) required subtrochanteric shortening osteotomy. This discrepancy highlights a fundamental limitation of the Crowe classification, which fails to distinguish between cases requiring substantial soft tissue release and those amenable to standard reduction techniques [2,6]. Consistent with our findings, Salman et al previously reported poor concordance between the Crowe type 3 classification and the actual need for femoral shortening, emphasizing that migration height alone is insufficient for surgical planning [15]. Prior studies by Atilla, Tozun et al, Altay et al, and others have repeatedly underscored that periarticular muscle ligament tension, rather than displacement severity, determines the necessity of subtrochanteric shortening osteotomy; the Etfal system, which incorporates the lesser trochanter–ischial tuberosity to pelvic height ratio, appears to reflect this biomechanical reality with high fidelity [3,8,16]. Consequently, the Etfal system represents a classification that captures both osseous morphology and soft tissue tension.

Although both systems yielded AUC values greater than 0.90, the Etfal classification demonstrated superior diagnostic discrimination (Etfal, 0.98; Crowe, 0.94). Etfal type 2 or greater displayed excellent sensitivity, while Etfal type 3 or greater offered near-perfect specificity, consistent with the modern trend toward composite or proportional indices rather than displacement-only staging [10,17]. In Crowe type 3 hips, in which the operative strategy is often most difficult, the ability of a classification system to correctly “rule in” the need for subtrochanteric shortening osteotomy is critical, given the morbidity associated with excessive limb lengthening, prolonged operative time, or non-union related to unnecessary osteotomy [3,8]. Conversely, strong sensitivity is vital to avoid neurological injury caused by inadequate soft tissue accommodation during reduction [4,18]. With balanced sensitivity and specificity, the Etfal system directly addresses this critical unmet need by reliably distinguishing cases in which subtrochanteric shortening osteotomy is biomechanically required from those in which it is not.

However, the very high AUC values observed in this study should be interpreted cautiously. Near-ceiling discrimination metrics may partially reflect the clustering of osteotomy requirements in advanced Etfal categories within this cohort. The diagnostic performance of the classification may vary in populations with different dysplasia severity distributions. Therefore, external validation across diverse cohorts is essential to confirm the generalizability of these findings.

Another implication of the Etfal classification is its utility in preoperative planning and its potential to reduce inter-surgeon variability in THA for DDH. Known challenges, such as differences in surgeon experience, varied tendencies toward extended soft tissue releases or controlled femoral cracking, and diverse interpretations of complex anatomy, are major contributors to inconsistent treatment approaches across centers [1,6]. Because the Etfal system is based on easily identifiable anatomical landmarks and integrates bone–soft tissue biomechanics, it may offer a more standardized and reproducible basis for surgical decision-making. Contemporary planning paradigms increasingly favor quantitative and reproducible metrics over subjective radiographic interpretations [13,19]. Regardless of whether an anatomic or high hip–center reconstruction is chosen, improved prediction of soft tissue tension may help reduce major complications, such as femoral fractures, dislocations, and neurological injuries, by allowing more precise and individualized planning [6,15,20].

The complications observed in this series, including sciatic nerve palsy, intraoperative fractures, and postoperative instability, have previously been associated with excessive soft tissue tension in dysplastic hips undergoing THA [2,10]. Greber et al and Atilla similarly highlighted abnormal proximal femoral morphology and substantial soft tissue contracture in high-riding dislocations as key contributors to neurological risk and reduction difficulty [1,3]. The clustering of complications within advanced Etfal categories in our study reinforces this biomechanical mechanism and demonstrates that proportional composite radiographic assessments predict surgical complexity and complication likelihood more effectively than displacement-based systems [6,17]. These findings support the concept that traditional classifications may inadequately distinguish clinically relevant subgroups, whereas tension-sensitive systems, such as Etfal, provide more meaningful stratification of THA in DDH.

Finally, the superior performance of the Etfal system suggests that it may serve not only as a classification tool but also as a conceptual model for future biomechanically driven prediction algorithms and computer-assisted planning systems. Modern arthroplasty emphasizes proportional relationships between soft tissue elasticity, limb length goals, and reduction tension over absolute displacement measurements [3,10,16]. The pragmatic findings of this study provide clear, reproducible evidence to guide surgical decision-making and align closely with the evolving understanding that classification systems must directly influence treatment strategies to be clinically meaningful [10,17]. Further refinement of ratio-based models and external validation across diverse populations and geographic regions are essential to establish the generalizability of the Etfal classification.

Conclusions

The Etfal classification demonstrated superior predictive capacity compared with the Crowe system for identifying the intraoperative requirements for subtrochanteric shortening osteotomy during THA for DDH. Nearly all Etfal type 3 and 4 hips required shortening, whereas Crowe type 3 hips rarely required it. ROC analysis further confirmed the clinical validity of the Etfal system, which showed a higher AUC and sharper discriminative ability.

However, the performance of the classification may depend on the severity distribution of the studied cohort, and further external validation studies are required to confirm its generalizability. Future studies incorporating broader populations and external validation frameworks are warranted to confirm the reliability of preoperative planning.