This review is aimed at clinicians appraising preclinical trauma studies and researchers investigating compromised bone healing or novel treatments for fractures. It categorises the clinical scenarios of poor healing of fractures and attempts to match them with the appropriate animal models in the literature.

We performed an extensive literature search of animal models of long bone fracture repair/nonunion and grouped the resulting studies according to the clinical scenario they were attempting to reflect; we then scrutinised them for their reliability and accuracy in reproducing that clinical scenario.

Models for normal fracture repair (primary and secondary), delayed union, nonunion (atrophic and hypertrophic), segmental defects and fractures at risk of impaired healing were identified. Their accuracy in reflecting the clinical scenario ranged greatly and the reliability of reproducing the scenario ranged from 100% to 40%.

It is vital to know the limitations and success of each model when considering its application.

Our aim was to develop a clinically relevant model of atrophic nonunion in the rat to test the hypothesis that the vessel density of atrophic nonunion reaches that of normal healing bone, but at a later time-point. Atrophic nonunion is usually attributed to impaired blood supply and is poorly understood. We determined the number of blood vessels at the site of an osteotomy using immunolocalisation techniques in both normally healing bones and in atrophic nonunion. At one week after operation there were significantly fewer blood vessels in the nonunion group than in the healing group. By eight weeks, the number in the atrophic nonunion group had reached the same level as that in the healing group. Our findings suggest that the number of blood vessels in atrophic nonunion reaches the same level as that in healing bone, but at a later time-point. Diminished vascularity within the first three weeks, but not at a later time-point, may prevent fractures from uniting