Classifications in Brief: The Modified Neer Classification for Distal-third Clavicle Fractures

History

Clavicle fractures represent 2.6% to 4% of all fractures, 10% to 28% of which involve the distal clavicle [17, 20, 25]. Distal clavicle fractures occur after a direct blow or fall onto the acromion. The force is distributed medially to the clavicle, where the bone fails superiorly with tension and inferiorly with compression. Injuries are seen in a bimodal age distribution in men younger than 30 years and elderly persons older than 60 years [2, 23].

In 1847, Sir Robert William Smith [30] described the variable remodeling potential of the distal clavicle after posthumously examining eight clavicle fractures. Fractures of the distal clavicle were rarely considered for operative fixation until Charles Neer’s [16] 1960 study examining the incidence of clavicle nonunion as a function of disruption to the coracoclavicular ligaments. Neer originally described three fracture types in relation to the coracoclavicular ligaments and later revised his system to include avulsion fractures of the periosteal sleeve in pediatric patients and comminuted fractures with an inferior bony fragment attached to the coracoclavicular ligaments [17-19]. Craig [8] subdivided Type II fractures based on the individual components of the coracoclavicular ligaments. Together, these adjustments compose the modified Neer classification, which is commonly referenced today.

Other classification systems have been described. Allman [1] described fractures of the entire clavicle based on three anatomic locations: medial, middle, and lateral. However, his classification did not predict outcomes or guide treatment of distal clavicle fractures. Robinson [25] proposed a classification to address all clavicle fractures and specifically subclassified distal fractures, defined as the lateral fifth of the clavicle. Subgroups were broken down into A, nondisplaced, and B, displaced. These groups were further subdivided depending on acromioclavicular articular involvement. Despite high interobserver reliability, the Robinson classification is only occasionally cited [3], perhaps because the Robinson classification does not account for coracoclavicular ligament integrity in relation to fracture pattern. The Orthopaedic Trauma Association (OTA) revised their classification system to include the clavicle as bone “15” [14]. The OTA classification considered anatomic location (proximal, diaphyseal, and lateral to the medial edge of the base of the coracoid) and articular involvement (extraarticular, partial articular, and completely articular) [7]. Rauer et al. [24] found that among five orthopaedic trauma surgeons, a group who theoretically use the OTA classification system at a high rate, interobserver reliability was only fair (κ = 0.374), and so perhaps this classification’s complexity (which may be the cause of its limited interobserver reliability) accounts for what we perceive to be relatively limited uptake of the OTA system in practice. Despite the availability of alternatives, it has been our observation that the modified Neer classification remains the predominantly cited classification system.

Purpose

To our knowledge, Neer [16] was the first to discuss an increased proportion of nonunion with fractures of the distal clavicle. Others have reproduced similar findings in subsequent studies [21, 27]. The Neer classification aimed to describe the stability of a distal clavicle fracture as it relates to the anatomic insertion of the conoid and trapezoid ligaments. By predicting stability of the fracture fragments, Neer sought to guide surgeons in choosing which fractures can be managed nonoperatively and which might better be managed surgically.

The decision to operate on a patient cannot be taken lightly. Information gathered from radiographs in the decision-making process must be accurate to ensure surgical success and patient safety. The purpose of the Neer classification was well-intended. However, the reproducibly of the scheme has been called into question [3, 5, 24]. Furthermore, the clinical importance of a distal clavicle nonunion varies widely [10, 21, 26].

Description

An evaluation of distal clavicle fractures begins with an AP and axillary view of the shoulder. To classify fractures in his patients, Neer [18] used anterior and posterior 45° oblique views and a stress radiograph by having patients hold a 10-pound weight bilaterally. A 10° to 15° cephalic-tilt radiograph, also known as a Zanca view, can provide insight into potential intraarticular fracture patterns.

The modified Neer classification relies on determining the status of the fracture in relation to the coracoclavicular ligaments. The coracoclavicular ligaments are composed of two individual ligaments, the trapezoid ligament and conoid ligament. The trapezoid ligament is based laterally and attaches to the inferior clavicle on average 3 cm from the acromioclavicular joint, whereas the conoid ligament attaches 4.5 cm from the acromioclavicular joint. The conoid ligament sits posteromedial to the trapezoid ligament [32]. Together, these ligaments are responsible for vertical stability of the clavicle [9].

In the modified Neer classification, distal clavicle fractures are divided into five types (Fig. 1). Type I fractures are lateral to the intact coracoclavicular ligaments and spare the acromioclavicular joint. Type III fractures are also lateral to the coracoclavicular ligaments but the fracture extends into the acromioclavicular joint. Types I and III fractures are inherently stable, imparted by the trapezoid and conoid ligaments medially and the deltotrapezial fascia laterally. These fractures are almost always treated nonoperatively [16]. Type II fractures involve the coracoclavicular ligaments and are subdivided into two types. Type IIA fractures occur medially to the trapezoid and conoid ligaments. Craig [8] modified Type II fractures to include the subtype IIB, in which the fracture exits between the conoid and trapezoid ligaments. This leaves an attached trapezoid ligament to the lateral fragment and a torn conoid ligament medially. Differentiating between Type IIA and IIB may be difficult to judge on routine images and is not clinically relevant. Both are inherently unstable because of the loss of ligamentous support of one or both coracoclavicular ligaments. Type IV fractures are only found in the pediatric population, with periosteal slippage of the proximal fragment at the epiphysis and intact coracoclavicular ligaments at the lateral fragment. Type V describes a comminuted fracture pattern with an inferior fracture fragment attached to the coracoclavicular ligaments.

Fig. 1.

This illustration shows the modified Neer classification [2]. Reprinted with permission from Banerjee R, Waterman B, Padalecki J, Robertson W. Management of distal clavicle fractures. J Am Acad Orthop Surg. 2011;19:392-401.

Validation

For any classification schema to justify widespread use, a high degree of interobserver and intraobserver reliability are critical; unfortunately, this has not been demonstrated for the modified Neer classification of distal clavicle fractures. Numerous studies have found its intra- and interobserver reliability to be inconsistent [3, 5, 24] which means that a large proportion of fractures would be classified differently if the same observer looked at the same radiographs on different days, or if two different observers graded the same radiographs. Consequently, when such a system is used in practice, it can result in incorrect treatment choices, incorrect prognosis estimates, and confusing communication between providers; when used in research, it can result in misclassifications that cause research results to mislead rather than enlighten.

Although it has been commonly used in clinical practice for some time, studies did not attempt to formally validate the modified Neer classification until 2015. Bishop et al. [3] aimed to establish the intraobserver and interobserver reliability of distal-third clavicle fractures on plain radiographs. Thirty radiographs were evaluated by 14 shoulder or sports medicine fellowship-trained orthopaedic surgeons at Timepoint 0 and again 3 months later. The authors evaluated the size of the distal fragment, stability of the fracture pattern, Neer classification, and the individual surgeon’s decision regarding surgical or nonsurgical management. Interrater reliability was fair for the Neer classification (κ = 0.11 to 0.35; p < 0.001), moderate for the decision to operate (κ = 0.55), and varied when judging the size of the remaining lateral fragment. Intrarater reliability was better for the size of the distal fragment (κ = 0.50) and Neer classification (κ = 0.42), and substantial for stability of the fracture pattern (κ = 0.65) and the subsequent decision to operate (κ = 0.65). The authors concluded that the most important clinical factor for the decision to operate was the stability of the fracture fragment rather than the Neer classification. Furthermore, they suggested that the size of the distal fragment can determine the fixation strategy [3].

Cho et al. [5] found similar results when measuring the modified Neer classification among fellowship-trained orthopaedic shoulder surgeons and fellows. Their results showed fair interrater reliability (κ = 0.344) and moderate intrarater reliability (κ = 0.496). The authors evaluated an additional CT image of the fracture patterns to assess whether 3-D CT would improve validity. No difference was found, suggesting three-dimensional (3-D) imaging does not improve the level of agreement [5]. Perhaps a reason for this finding is 3-D CT captures comminution missed on plain radiographs, leading to a more difficult interpretation. Nonetheless, the efficacy and clinical use of the Neer classification must be called into question if advanced imaging does not improve intra- and interrater reliability.

Rauer et al. [24] directly compared the reliability of the OTA, Neer, and Jager/Breitner classification systems among orthopaedic traumatologists and musculoskeletal radiologists. The overall interrater reliability was fair for both groups in all classification systems (κ = 0.238 to 0.374). The intrarater reliability was better for the OTA and Jager/Breitner classification systems than for the Neer classification (κ = 0.461, 0.426, and 0.298, respectively). Because of the overall low levels of agreement and reliability, the authors concluded that a great deal of subjectivity goes into classifying distal clavicle fractures with any of the classification systems [24].

Limitations

The most serious limitation of this classification is its inconsistent (and often only fair to moderate) intra- and interobserver reliability [3, 5, 24], which undermines its utility both to clinicians and researchers. Radiographs are commonly used to classify distal clavicle fractures. Cho et al. [5] attempted to assess whether evaluating 3-D CT images would improve the agreement between orthopaedic shoulder surgeons regarding the classification of distal clavicle fractures. The addition of a CT image did not improve intrarater or interrater reliability or change surgical decision-making [5]. The degree of ambiguity in classifying distal clavicle fractures after radiography and CT leaves much of the decision-making to the experience of the surgeon rather than defaulting to a classification-based algorithm.

Other researchers have developed classification systems with improved intrarater and interrater reliability. Cho et al. [4] devised a novel classification based on fracture stability and bony displacement. The authors found a moderate (κ = 0.434) and substantial (κ = 0.644) intrarater and interrater reliability among experienced shoulder surgeons and fellows. Robinson [25] proposed a classification to address all clavicle fractures and specifically subclassified distal fractures. His classification considered location, displacement, and intraarticular involvement. The classification demonstrated substantial interobserver (κ= 0.77) and excellent intraobserver (κ= 0.84) reliability. It must be noted that both the Cho and Robinson classification systems were reported by their respective designers. Before wide acceptance, the reliability of their classifications must be reproduced by other investigators.

The modified Neer classification attempts to guide treatment by identifying the location of the fracture in relation to the coracoclavicular ligaments. However, the decision to operate on a distal clavicle fracture depends mostly on the stability of the fracture pattern rather than its Neer classification [3, 4]. Fixation strategies for the distal clavicle include locking plate technology [29], tension band wiring [31], hook plate [13, 15], open coracoclavicular ligament reconstruction [2] and arthroscopically assisted coracoclavicular ligament reconstruction [6]. The decision of which modality to use—or indeed whether to operate at all—also depends on the size of the lateral fracture fragment. Fragment size of greater than 1 cm is reported to be necessary for bony union using locking plate technology [3, 29]. Open coracoclavicular ligament stabilization with nonabsorbable sutures passed around the coracoid process, secured to the proximal fragment through drill holes, and supplemented with a dorsal sided suture tension band can be implemented when the distal fragment cannot accommodate a locking plate [2]. The Neer classification omits the size of the lateral clavicle fracture fragment.

The nonunion risk for Neer type II fractures ranges from 21% to 33% [17, 21, 26] and is frequently cited as a reason to operate [11, 12, 15, 17]. However, distal clavicle nonunion may not be clinically relevant because radiographic nonunion does not always correlate with symptomatic nonunion. The risk of symptomatic nonunion ranges from 0% to 34% [10, 21, 26]. Nonoperative management of displaced Neer type II fractures allows the patient to avoid the risks and complications associated with surgery and provides adequate functional outcomes. In a small series, Rokito et al. [28] found no difference in the mean American Shoulder and Elbow Surgeons and Constant scores for patients with Neer Type II distal clavicle fractures between those treated operatively and those treated nonoperatively. Similarly, a systematic review of 435 distal clavicle fractures found overall adequate functional outcomes and a low complication proportion with nonoperative management of Type II distal clavicle fractures [22].

Conclusion

The modified Neer classification has limited utility in guiding the treatment of distal clavicle fractures. Despite its fair-to-moderate reliability, it is still widely accepted and used by orthopaedic surgeons. However, we believe it should not be. Kappa values less than 0.4 represent an unacceptable rate of disagreement between observers. Using the modified Neer classification system could lead to errors in communication, insufficiently justified or unjustified surgical recommendations, and misleading research. Distal clavicle nonunion is commonly used to support the decision to treat Type II fractures operatively. However, distal clavicle nonunion is not always symptomatic and surgery may result in postoperative pain, infection, symptomatic hardware, and risk of revision surgery [13, 22]. Future work is required to appropriately predict which fractures will progress to symptomatic nonunion and could be treated with surgery and those that are best treated nonoperatively. We recommend surgeons evaluate individual patient age, functional requirements, specific fracture pattern in reference to the coracoclavicular ligaments, and the amount of distal bone remaining when treating distal clavicle fractures, rather than using the modified Neer classification, which has demonstrated itself to be insufficiently reliable for clinical or research use.