Rockwood, Green, and Wilkins' Fractures, 10e Package

117

CHAPTER 5 • Classification of Fracture

a dichotomous variable. 60 These authors focused on the chal lenge of interobserver variation and individual judgment in the use of fracture classification systems. 60 Categorizing injury severity presents interesting challenges because injury severity is a continuous variable, perhaps the key variable that surgeons face in treating fractures. Many fracture classifications are based on the severity of injury. The OTA/AO universal classification categorizes fractures into types A, B, and C based on the degree of comminution. Because injury severity is a continuous variable that occurs on a spectrum, breaking it into discrete categories will inevitably introduce error. These errors are increased by the fact that measuring injury severity is very imprecise (the amount of comminution, for example). Rank order techniques have been proposed as an alternate way to stratify continuous variables that are difficult to measure. Rank orders have been used in pathology and other medical fields where subjective opinion is used and is challenging to quantify. In one orthopaedic study, a rank order technique was used to stratify resident clinical performance, an important but difficult to mea sure variable. Rank order techniques have been recommended as methods to assess fractures and have been shown to have greater reliability than traditional classification. They have been used in the tibial plateau and tibial plafond to stratify injury severity. 11 Despite their advantages, there are significant disadvantages of rank order techniques that prevent wide applicability. First and foremost, they apply only to the series of cases under study. A rank order cannot be applied to an individual fracture out of that series, which limits the utility to research projects. CT imaging is a possible technique to measure injury severity through computational assessment in order to objectively and quantitatively measure fracture severity. This technique uses standard CT-based datasets of fractures taken at the time of injury and measures fracture surface area slice by slice. Liberated fracture surface area correlates with fracture energy. Although oversimplified, the technique essentially quantitatively mea sures the amount of comminution, which has been shown to correlate with clinical and radiographic outcomes of tibial pilon fractures. 3 This objective measurement has significant promise, but at its current stage, it is mostly a research tool. It does illus trate two important points. First, it shows the potential value of measuring something that previously could only be subjectively assessed. Secondly, it demonstrates that in the future computer ized analysis may have a big role in fracture classification.

Dislocation Compendium. 44 The updated version holds prom ise that it will be more widely accepted and utilized bringing the fracture community closer to a true international language of fractures. In many anatomic areas, time-honored fracture-­ specific classifications remain widely accepted and the most uti lized classification. These traditional classifications have stood the test of time by providing value for treatment and prognosis, and their terminology has become entrenched when fractures are discussed in these areas. The classification of proximal humerus fractures illustrates an area where a deeply entrenched classification has provided lan guage that is nearly universally utilized and accepted. However, scientific scrutiny shows that observers do not use the classifi cation and resulting language reliably, and this lack of a reliable classification may affect our ability to have clinical studies con vincingly demonstrate optimal treatments. The Neer classifica tion system (Fig. 5-2) for proximal humerus fractures involves a single anatomic area defined on radiographs. 41 This classifi cation identifies predictable fracture lines creating four poten tial fracture fragments: (1) the humeral head, (2) the greater tuberosity, (3) the lesser tuberosity, and (4) the humeral shaft. This classification is then based on how many “parts” are cre ated by the fracture lines. Neer defined a fracture as a part if it was displaced by more than 1 cm or angulated by more than 45 degrees. A one-part fracture could contain any number of fractures that are not displaced enough to meet either of these criteria. A proximal humerus fracture with enough displace ment to be called a part can be classified as either a two-part, three-part, or four-part fracture. When there is a dislocation, the fracture is first classified followed by the direction of the dis location, for example, three-part anterior fracture dislocation. While not part of Neer’s original classification, additional inju ries can also occur to the humeral head, such as head splitting and impaction fractures of the articular cartilage. 23 This is an example of how many classifications fail to classify and account for all potential fracture lines and patterns that can occur. Despite the fact that the proximal humerus fracture classifi cation according to Neer seems straightforward, poor interob server reliability has been shown. Issues with reliability of fracture classification were widely demonstrated for the Neer classification. 32,57,58 In one study by Kristiansen et al., 100 AP and lateral radiographs were reviewed and classified by four independent reviewers. A low degree of agreement was found between reviewers most notably between the least experienced reviewers. 32 In another study by Sidor et al., the radiographs of 50 fractures of the proximal humerus were reviewed by five observers at two time points 6 months apart. The reviewers included a shoulder specialist, trauma surgeon, skeletal radiolo gist, and two residents. All the five observers agreed on the final classification for only 32% and 30% of the fractures on the first and second viewings, respectively. Paired comparisons between the five observers showed a mean reliability coefficient (Kappa) of 0.48 for the first viewing and 0.52 for the second viewing. 57 FRACTURE-SPECIFIC CLASSIFICATIONS Proximal Humerus

CURRENT STATE OF CLASSIFICATION OF FRACTURES

While an exhaustive review of each anatomic-based classifi cation of fractures is beyond the scope of the present chapter, several classification systems will be described for illustrative purposes. Current classification of fractures has progressed to more uniformly accepted classifications that provide value in one or more of the ways described in the previous sections. The 2018 OTA/AO classification of fractures builds on its pre decessor compendiums 38 and especially the 1996 Fracture and