Armstrong distinguished hip dislocations with an additional femoral head fracture or with an acetabular fracture. Acetabular fractures were further classified as rim fractures or acetabular floor fractures (quadrilateral surface involvement).6 Similar classifications were reported by Wiltberger et al in 194810 and Bonnin in 1958,11 but an exact description of the acetabular fracture morphology was not given.
Cauchoix and Truchet reported the first more detailed classification of acetabular fractures in 1951,12 distinguishing between posterior fracture dislocations, central fractures involving the acetabular floor, and the presence of a pelvic ring injury as a separate entity—later additionally proposed by Merle d’Aubigné.7 Central fractures were subgrouped into fractures with or without central hip dislocation and in posterior fracture dislocations. The size of the posterior rim or wall fragments as well as an additional involvement of the acetabular floor were further subgroup criteria.
Cagnoli divided acetabular fractures into fracture dislocations and pure acetabular fractures.13 In total, nine different fracture types were distinguished. For the first time, fractures of the superior and the anterior acetabulum were integrated in this classification.
The group of fracture dislocations consisted of pure acetabular fracture dislocations, fracture dislocations with an additional femoral head fracture, posterior fracture dislocations with a posterior rim/wall fragment, and involving the acetabular roof, the anterior acetabulum, or the acetabular fossa.
Pure acetabular fractures types were further subclassified as fractures involving the acetabular fossa, involvement of the ischiopubic rami, central fracture dislocations, and segmental acetabular fractures.
Creyssel et al classified fracture dislocations dependent on the remaining hip joint stability.14
A more precise classification of central fracture dislocations into four groups was proposed by Stewart and Milford.15 They distinguished between:
Linear or star-shaped fractures (multiple fracture lines) of the acetabular fossa without significant displacement
Acetabular comminution fractures with slight central dislocations
Central fracture dislocations with/without involvement of the superior acetabulum
Central fracture dislocations with additional femoral head or neck fracture
Fractures of the posterior wall were determined according to the size of the posterior rim fragments and any additional hip instability.
Böhler was the first to present clear radiological criteria for central acetabular fractures.16 Parameters were the involvement of the acetabular fossa and the severity of dislocation. Using a line connecting both acetabular roofs, the following fracture types were distinguished (▶ Fig. 4.1):
Posterior fractures with/without femoral head dislocations
Central fracture dislocations
Fractures involving the acetabular floor without displacement
Acetabular floor fractures starting anterior-cranial and extending posterior-caudal with medial subluxation without a height offset
Cranial (8–12 mm) and medial displacement of the acetabular roof together with the femoral head without subluxation
Cranial and medial displacement of the acetabular roof together with the femoral head with subluxation
Fig. 4.1 Classification according to Böhler. Pure central hip dislocation with simple acetabular floor fracture (left), central hip dislocation with acetabular comminution without height offset (middle), and central hip fracture dislocation with height offset (right).
Further suggestions for classification of acetabular fractures have shown no significant innovations and specification regarding fracture types were not made.
Early classifications distinguished between posterior and central fracture (dislocations), dependent on the available treatment possibilities, especially regarding treatment of the accompanying hip dislocation (closed reduction and traction treatment, supported by a lateral pull).
The first extended, anatomically and biomechanically based classification of acetabular fractures was proposed by Rowe and Lowell in 1961.17
The region of the acetabulum was anatomically divided into three areas, which corresponded to the primary ossification centers. Thus, an anterior pubic segment (inner wall), a posterior ischial segment (posterior acetabulum), and an iliac segment (superior dome) were distinguished (▶ Fig. 4.2).
Fig. 4.2 Classification of acetabular fractures according to Rowe and Lowell.
Based on the direction and magnitude of the injury forces acting on the acetabulum, this classification consisted on four groups with two to three subgroups.
As a main disadvantage, no clear criteria were reported to group and subgroup these injuries and the classification was only based on an anteroposterior (AP) X-ray of the pelvis. Thus, possible overlapping of bony structures could result in misinterpretations and the significance of displacement was not addressed.
These early classifications showed a lack of clear classification criteria.
4.3 Letournel Classification
In the early 60s, Judet and Letournel developed the still presently valid and accepted classification of acetabular fractures based on a fundamental anatomic and radiological analysis of the morphology of the acetabulum.18,19,20,21,22
These investigations led to a completely new understanding of acetabular fractures and is the basis of the most widely used classification systems. Anatomically, the understanding of the column principle forms this classification.22 The acetabulum is functionally composed of two columns, an anterior and a posterior column in shape of an inverted Y, with the hip socket integrated at its intersection.
The biomechanically more important posterior column (▶ Fig. 4.3) transfers the load from the trunk and the spine via the sacroiliac (SI) joint into the proximal thigh and therefore consists of a dense and very strong bone structure. This column consists of parts of the ilium and the ischium.
Fig. 4.3 Posterior and anterior column of the acetabulum.
The anterior column (▶ Fig. 4.3) consists of major parts of the anterior ilium and the pubic bones. It extends cranially from a variable point from the anterior superior iliac spine along the iliac crest over the apex of the crest and proceeds distally to the upper pubic rami and to the pubic symphysis.
The anatomic basis of these column structures was demonstrated on conventional radiographs. By analyzing special radiological lines, described in ▶ 3, five simple, elementary fracture types, with only a single main fracture line, and five associated fracture types were distinguished.
Elementary fracture types represent posterior wall fractures, posterior column fractures, anterior wall fractures, anterior column fractures, and pure transverse fractures (▶ Fig. 4.4).
Fig. 4.4 Elementary fracture types (top row) and associated fracture type (bottom row) according to Letournel.
Associated fracture types include associated posterior column and wall fractures, transverse and posterior wall fractures, T-type fractures, associated anterior column and posterior hemitransverse fractures, and both column fractures (▶ Fig. 4.4).
The detailed description of these types of fractures is given in the fracture type chapters, which focus on their peculiarities.
4.4 AO/OTA Classification
Müller et al integrated the Letournel classification into their standardized and complete AO-system.4,5,23 The resulting AO/OTA classification added additional relevant prognostic injuries of the hip joint, such as marginal impaction zones, femoral head injuries, and comminution zones into this “new” comprehensive classification system of acetabular fractures (Comprehensive Classification of Fractures [CCF] = AO/OTA classification).5 The main principle is a hierarchical classification of all fractures in triple groups. Three fracture types were defined:
Partial articular fractures: Type A fractures consist of posterior wall fractures, anterior wall fractures, posterior column fractures, anterior column fractures, and associated posterior column and wall fractures.
Factures with a transverse fracture component: Type B fractures consist of pure transverse fractures, associated transverse posterior wall fractures, T-type fractures, and associated anterior column plus posterior hemitransverse fractures.
Complete articular fractures (floating acetabulum): Type C fractures represent all types of both-column fractures.
Each fracture type is further divided into three fracture groups (A1, A2, A3, B1, B2, B3, C1, C2, C3), according to its proposed increasing fracture severity. If a group division is impossible, the fracture is classified as D.
Each fracture group is again subdivided into three subgroups (A1.1 to C3.3). If the fracture does not fit to one subgroup, it is marked as 4. Classification into subgroups (A.1.1 to C.3.3) is sometimes only possible intra- or postoperatively.
Additionally, for the exact description of the whole fracture pathology, modifiers were integrated into the classification system. These modifiers describe concomitant articular injuries, which potentially can be present in all fracture types and have potential prognostic value.24,25,26,27 Seven different modifiers were defined:
Defines the main fracture more precisely
Provides additional information to the main fracture
The other modifiers (joint injuries) often can only be analyzed intraoperatively:
Acetabular injury (superficial cartilage lesion, cartilage shear injury, marginal impaction)
Number of articulating fragments including the wall fragments (one fragment, two fragments, more than two fragments)
Extent of joint displacement (gaps/steps): undisplaced (0–1 mm), 1–5 mm, 6–10 mm, > 10 mm
Accompanying femoral head injury (superficial cartilage lesion, cartilage shear injury, marginal impaction)
Intraarticular fragments, requiring surgical removal
Overall, the AO/OTA classification allows approximately 20,000 different classification possibilities in acetabular fractures.
Category 0: Isolated wall fractures
Category 1: Isolated column fractures
Category 2: Involvement of both columns
2A: Fracture only at the joint level
2B: Fracture extension into the ilium
2C: Fractural extension into the obturator foramen
Category 3: Floating acetabulum (both-column fracture)
Ultimately, this classification is also based on the original Letournel classification and therefore thinking in Letournel’s classic 10 fracture types remains the present basis of fracture understanding.
4.5 Reliability of Acetabulum Fracture Classification
Visutipol et al compared the relevance of three-dimensional (3D) illustration of acetabular fractures with conventional radiographs.36 Five orthopaedic surgeons classified conventional X-rays and computed tomography (CT) scans of two groups of 20 patients at two times (2-month interval). Kappa-statistics37 showed no difference in intra- and interobserver reliability. Kappa values for conventional X-rays and 3D images were 0.4 and 0.24, respectively, and thus a very low to moderate result was observed. 3D images were found to have no advantage in classifying acetabular fractures.
Beaulé et al analyzed 30 X-ray sets at two times (2-month interval) by nine investigators.30 The examiners were divided into three groups: experts, educated by Letournel, with 15–18 years of experience in surgical treatment of acetabular fractures; experienced acetabular surgeons with 5–10 years of experience; and less experienced surgeons with less than 50 acetabular stabilizations performed within the last 5–10 years. A significant increase of consistency was observed with increasing experience, but no significant difference was found between experts and very experienced surgeons.
Average Kappa values of 0.65 and 0.7 were found at the two examination times, corresponding to a good agreement. Analysis of the rate of percentage agreement for particular fracture types showed difficulties in the assessment of T-type fractures, both-column fractures, anterior wall fractures, and associated anterior column plus posterior hemitransverse fractures.
However, this group of examiners represents an expert group as even the most inexperienced investigators operated on up to 50 acetabular fractures within 5–10 years.
In another report from 2003, 30 radiographic X-ray sets (pelvic AP view, Judet views) were analyzed by six investigators at two intervals.34 Although surgeons in training showed results close to probability by chance, orthopaedic surgeons showed moderate agreement (intraclass correlation of 0.56). The best agreement was observed when analyzing the iliopectineal line, the ilioischial line, and the presence of a posterior wall fragment, whereas the Judet views had no influence on improvement of agreement. Overall, an experience-dependent agreement was observed.
Similar results were reported by two musculoskeletal radiologists who analyzed conventional radiographs and CT scans of 101 acetabular fractures at different times.33 The consistency analyzing conventional X-rays was only 0.42 (moderate consistency), whereas the CT diagnostics showed good agreement (k = 0.70). The addition of Judet views did not result in an optimization of agreement.
In contrast, nonspecialist/nonexperienced surgeons had worse results. Ten orthopaedic residents evaluated 50 acetabular fractures and intra- and interobserver agreement was analyzed. Very low mean kappa coefficients were found with an overall poor agreement. Methodical training in the interpretation of radiographs was not effective.38
The iliopectineal line, the ilioischial line, and the presence of a posterior wall fragment seem to be the most relevant line for analysis on the pelvic AP X-ray, whereas the value of the Judet views is questionable for classification.
Durkee et al proposed using an algorithm that focuses more on these relevant lines for acetabular fracture classification.39 The presented algorithm using these lines clearly separated 5 of the 10 Letournel fracture types. However, the other fracture types remained unconsidered.
Adding the disruption or integrity of the obturator ring as a fracture criteria to such an algorithm led to an improvement of consistency rates from 59.9% to 71.2%.35
The importance of CT examination with or without 3D images, in combination with conventional recordings, was elaborated recently. Interobserver reliability was rated poor when analyzing only conventional X-rays compared to examinations with additional axial CT or 3D-CT. Interestingly, analysis of conventional views generated from CT data showed identical results compared to pure CT analyzes.32
Hurson et al performed an inter- and intraobserver analysis based on conventional X-rays and CT images as well as of bone models generated from the CT data set.31 Consultants and residents showed better results using these models compared to conventional radiographs (k = 0.76 vs. 0.51, and k = 0.71 vs. 0.42, respectively).
Using the clinical image library,40 40 sets of acetabular fracture X-rays were chosen, with all cases having conventional X-ray diagnostics (pelvic AP view, iliac and obturator oblique views, iliac oblique view [IOV], obturator oblique view [OOV]). In 27 sets, axial two-dimensional CT images were present and 19 of these sets had additional 3D images.
These 40 cases were independently analyzed and classified by eight investigators (interobserver reliability) and six examiners performed a reevaluation after at least 6 weeks (intraobserver reliability), with the X-ray sets in a different order. In the meantime, the cases could not be seen by any of the investigators. Statistical assessment was performed using kappa statistics.
The average percentage agreement in different fracture types was 81% (48–93%). Thus, an average of 19% of the cases was assessed differently (7–52%).
The average kappa value for intraobserver reliability for all six investigators was 0.77 (0.56–0.91), corresponding to a good agreement.
The average kappa value of intraobserver reliability for cases with only conventional X-ray diagnostics (n = 13) was 0.78 (0.63–1.00), corresponding to a good agreement.
For eight patients with additional two-dimensional CT, the average kappa value was 0.61 (0.42–0.70), corresponding to a good agreement.
In 19 patients with additional 3D-CT, the average kappa value was 0.81 (0.56–0.95), corresponding to an almost perfect agreement.
The type of radiological diagnostics had no significant effect on agreement in the intraobserver analysis.
Although the agreement rates for conventional radiological diagnostics and the presence of a 3D-CT were nearly identical, using additional two-dimensional CT led to a certain deterioration of the results.
The average kappa value for interobserver reliability in the assessment of all 40 cases was 0.59, corresponding to a moderate agreement. All examiners classified the cases identically in only 12 cases. These fracture types were six posterior wall fractures, three both-column fractures, one posterior column fracture, one pure transverse fracture, and one associated posterior column posterior wall fracture.
Further analysis was performed to identify the transition cases.
In 11 fractures two different fracture types were rated; in three cases no agreement was found between pure transverse fractures and an associated transverse posterior wall fracture nor in three other cases between an anterior column fracture and an associated anterior column plus posterior hemitransverse fracture/both-column fracture.
In 14 fractures, three different fracture types, and in three fractures, four different fracture types were rated.
The average kappa value of interobserver reliability for 13 cases with only conventional X-ray diagnostics was 0.62, corresponding to a good agreement. In four of these cases, all examiners completely agreed. There were two posterior wall fractures, one posterior column fracture, and one pure transverse fracture. In five cases, two fracture types were chosen, in two cases three or four fracture types. Using conventional X-rays, only elementary simple fracture types showed uniform agreement.
For the eight patients with additional two-dimensional CT, the average kappa value was 0.63, corresponding to a good agreement. Two cases (one posterior wall fracture, one both-column fracture) had consistent agreement (25%).
In 19 patients with additional 3D-CT, the average kappa value was 0.59, corresponding to moderate agreement. Six cases (32%) were assessed identically by all investigators (three posterior wall fractures, one posterior column and posterior wall fracture, two both-column fractures).
In summary, the type of radiological diagnostics does not affect the rate of consistency by different investigators.
Analysis of each category (fracture type) resulted in good agreement in isolated posterior wall fractures and both-column fractures, whereas poor agreement was observed in associated transverse posterior wall fractures, T-type fractures, and associated anterior column plus posterior hemitransverse fractures.