Ankle fractures

Last evidence check March 2009

A fracture classification should provide diagnostic information and guidance as to natural history, treatment and expected outcome. It should be simple enough to use in practice and be sufficiently reproducible for datasets created using it to be compared. Studies on fracture classifications at various body sites have shown that these criteria are not often met.

Pathomechanics and classification are brought together in this document because one of the main ankle classification systems (Lauge-Hansen) is based on pathomechanics and the other (Weber/AO) takes into account pathomechanical factors in describing fracture patterns.

Lauge-Hansen classification

Lauge-Hansen (1950) described cadaver experiments in which various forces were applied to specimens and fracture configurations produced. Each configuration was defined by two factors: the position of the foot (pronation or supination) and the force applied to the ankle (adduction, external rotation or abduction), and in addition each configuration had a number of stages describing sequential injuries as the force was applied:

Danis/Weber/AO classification

Danis (1949) and subsequently Weber (1972) classified fractures by the radiographic appearance, according to the relationship of the fibular fracture to the syndesmosis:

This was intended as a guide to the increasing likelihood of surgical treatment. Fracture patterns without a fibular fracture (10%) cannot be classified by this system (Kennedy et al 1998). The AO group used the Weber system as the basis of the malleolar section of the comprehensive fracture classification, providing groups and subgroups to include 27 fracture types, some with qualifiers.

Some practitioners simply describe ankle fractures on the basis of the number of malleoli fractured. While this is somewhat crude and excludes useful information, it is an important prognostic factor.

How good are the classifications?

Lauge-Hansen classification

Several biomechanical studies have attempted to reproduce the work and classification of Lauge-Hansen. Unfortunately, the fractures they produced did not correspond with Lauge-Hansen’s classification. If anything, the fracture patterns produced by a given force in modern experiments tend to be less severe than those described by Lauge-Hansen, and the relationship between forces exerted on the ankle and fracture patterns is not exact.

Gardner (2006) carried out MRI on a group of displaced ankle fractures. 10% could not be classified according to the Lauge-Hansen system. Over 50% did not have the patterns of ligament and bony injury predicted by their apparent Lauge-Hansen type.

AO/Weber classification

The Weber B group has been shown to include fractures of very different prognosis, grouping together fractures with both intact and disrupted medial malleolus and deep deltoid ligament; this probably also applies to the A and C groups to some degree. The Weber classification omits isolated fractures of the medial and posterior malleoli, although this is covered by the full AO classification.

Studies of the inter-observer and intra-observer reproducibility of both systems have found that reproducibility of both systems is moderate; if anything the Weber system is probably more reproducible, but this advantage may be lost at the group level and the subgroup level has not been tested.

Finally, studies relating the fracture configuration to prognosis show that fracture configuration by the Lauge-Hansen and Weber systems is not highly predictive of prognosis; the number of malleoli injured and initial talar displacement are probably more important.

The AO system would appear to have the following advantages:

A simpler approach to fracture mechanics

Although the Lauge-Hansen classification describes fracture mechanics and pathoanatomy in some detail, it has not stood up well to critical analysis over the last 15 years - the papers of Michelson and Gardner are particularly important. It seems that there are two main types of fracture mechanism:

Towards a stability-based fracture classification?

Neither system clearly presents the distinctions between injury configurations which are important for treatment; in particular the distinction between stable and unstable injury configurations are buried in the classification subgroups, especially in the AO classification. There is a need to describe fracture patterns, but key treatment decisions are based more on the stability of the fracture than any other factor, and prognosis (as in most fractures) is determined by the energy of the injury. Michelson (2007) reanalysed the literature on stability and treatment and found that unstable fractures did better with fixation and stable fractures with non-surgical treatment (this is not the easiest paper to read without prior knowledge of the articles analysed!)

Court-Brown et al (1998) reported the classification of 1500 ankle fractures according to the AO classification. Approximately half fell into groups which would be expected to be stable, although it is not clear to what extent they took into account clinical information in arriving at their conclusions. Fox et al (2005) studied 300 ankle fractures, grouping them by stability (on clinical and radiological grounds) and by displacement. They also found 50% to be stable and undisplaced; of the remainder, half were potentially unstable but undisplaced, while one quarter of all the fractures were displaced. Even AO type C fractures were more likely to be undisplaced.

McConnell et al (2004) carried out stress radiography on 97 undisplaced AO B fractures and found 61 to be stable and 37 unstable. Medial tenderness and bruising did not predict radiographic stability accurately. However, McConnell et al operated on all their stress-positive fractures; Fox et al’s finding imply that many would have healed anatomically in BKW casts. Further trials are required for clarification.

Classification should give an indication of prognosis. Stable fractures do not displace even with axial loading out of splintage (see below). Fractures which are undisplaced at presentation, but are considered potentially unstable (usually because of tenderness over the deltoid) had a risk of displacement of 2.3% in Fox et al’s study. Displacement which is not reduced has a significant risk of symptomatic degenerate change.