Fracture repair occurs by two broad mechanisms: direct healing, and indirect healing with callus formation. The effects of bisphosphonates on fracture repair have been assessed only in models of indirect fracture healing. A rodent model of rigid compression plate fixation of a standardised tibial osteotomy was used. Ten skeletally mature Sprague–Dawley rats received daily subcutaneous injections of 1 µg/kg ibandronate (IBAN) and ten control rats received saline (control). Three weeks later a tibial osteotomy was rigidly fixed with compression plating. Six weeks later the animals were killed. Fracture repair was assessed with mechanical testing, radiographs and histology. The mean stress at failure in a four-point bending test was significantly lower in the IBAN group compared with controls (8.69 Nmm. -2. (. sd. 7.63) vs 24.65 Nmm. -2. (. sd. 6.15); p = 0.017). On contact radiographs of the extricated tibiae the mean bone density assessment at the osteotomy site was lower in the IBAN group than in controls (3.7 mmAl (. sd. 0.75) vs 4.6 mmAl (. sd. 0.57); p = 0.01). In addition, histological analysis revealed progression to fracture union in the controls but impaired fracture healing in the IBAN group, with predominantly cartilage-like and undifferentiated mesenchymal tissue (p = 0.007). . Bisphosphonate treatment in a therapeutic dose, as used for risk reduction in fragility fractures, had an inhibitory effect on direct fracture healing. We propose that bisphosphonate therapy not be commenced until after the fracture has united if the fracture has been rigidly fixed and is undergoing direct osteonal healing. Cite this article: Bone Joint J 2013;95-B:1263–8

We treated 19 patients with established nonunion of the radius and/or ulna by the excision of avascular bone and the grafting of blocks of corticocancellous bone from the iliac crest, augmented by rigid plate fixation under compression. This allowed early mobilisation, and bony union was achieved between three and 24 months after operation in all but one of the patients. The single failure was attributed to the excessive length of the defect (100 mm) and inadequate fixation