There is currently no panacea for the successful treatment of SNU, with failures occurring in up to 25% of cases [3, 10]. The main predictor for failure has been identified as the time elapsed between the initial injury and the treatment of the established SNU, with the success rates decreasing to 62% after delays of 5 years . To achieve clinical and radiological union the following principles have been previously proposed: (i) preservation of the blood supply; (ii) bone grafting to achieve the original bony alignment and correct any humpback deformity; (iii) stable internal fixation and correction of carpal instability; and (iv) the treatment of SNU before the development of degenerative change [6, 7, 9].
To this end, past SNU treatments have included bone grafting with or without internal fixation. Stable internal fixation with AO or Herbert screws has been shown to improve union rates when compared with K-wire fixation ; a quantitative meta-analysis has reported overall union rates of 94% following screw fixation with bone grafting, compared with 74% following K-wire fixation [9, 31]. The introduction of vascularized bone grafts has now also expanded the possibilities for SNU treatment to include proximal pole AVN and previous failed surgery [18–20], and has further improved union rates (to over 90%), though the harvesting and interposition of a viable vascularized bone graft requires great skill, and the placement of the fixation device is also technically demanding . Impressive results were also seen in a series of 15 SNU patients (7 fibrous unions and 8 nonunions) treated using an arthroscopically assisted percutaneous internal fixation without bone grafting at a mean of 8.5 months post-injury. 100% union rates and good clinical outcomes were seen at 14 weeks post procedure  though this technically challenging procedure, we feel, has the potential to cause further soft tissue damage and disruption to the local biology, in less experienced hands.
A recent systematic review reported union rates of 80% using bone graft without fixation, 85% using bone graft with fixation, and 91% using vascularized bone grafts .
In contrast, the Ilizarov technique performed in this series involved the application of a circular external fixator without the use of bone graft, and thus its main advantage was to eliminate the need to expose the nonunion site, avoid causing further soft-tissue damage, as well as avoiding the morbidity and technical difficulties of potential bone graft harvesting. We found that the use of this system was not particularly technically demanding, and would be fairly straight forward for surgeons trained in fine-wire fixator application.
The main disadvantages to this technique related to the size of the bulky apparatus and the prolonged immobilization of the wrist joint. Postoperative wrist immobilization, however, is advocated with most other fixation and treatment methods [3, 9], with periods of up to 80 weeks , and no patient in our series required the frame in situ for more than 9 weeks. Following intensive physiotherapy all patients achieved improved arcs of movement and no patient developed CRPS. Imprudent wire placement has the potential to cause a temporary tenodesis of the digital tendons during the distal-ring fixation, or damage to the ulnar nerve or radial artery when placing the proximal-ring K-wires, though no patient in our series had any problems with digital tightness, stiffness, tendon adherence or contractures in the MCP or IP joints.
Our initial results are encouraging, with bony union achieved in all fifteen patients after a mean of 89 days (70-130 days), comparing favourably to other standard techniques (42-112 days) [10, 15, 16, 19, 20]. Mean Mayo wrist scores (86 points) were also similar to those scores achieved in patients with vascularized bone grafts (82-92 points) . The patients tolerated the apparatus well, and though rather bulky found that they had good use of the operated hand with the frame in situ. The procedure had a low complication rate with 4 pin-tract infections in 3 patients which resolved with local saline washes, occlusive dressings and oral antibiotic therapy.
We noted that one patient in this series, with an SNU of 15 months duration, developed a humpback deformity of approximately 70 degrees during their Ilizarov treatment. The reasons for this remain unclear, though we postulate that it may relate to the compression having not been applied along the anatomical axis of the scaphoid, thus producing palmar angulation . This however was not seen in the other cases, and in fact the patient had a good clinical outcome with a Mayo score of 80, good grip strength and flexion-extension arc; and united their scaphoid nonunion in 95 days.
Our retrospective study has obvious limitations. We did not include SNU cases with humpback deformity, carpal instability, carpal collapse, AVN, or marked degenerative changes; these would have predisposed to an adverse outcome and therefore our results might not be directly comparable to those of other SNU series in the literature. In addition, we did not randomize the patients and compare the Ilizarov technique with other established methods for the treatment of SNU; thus it is difficult to draw any strong conclusions as to whether this technique is preferable.
However, the results of this study are promising and demonstrate that distraction-compression using the Ilizarov method without the use of bone graft is a safe technique, and that in selected cases may be an effective way of managing scaphoid nonunion. Further investigation should help to define a potential role for this technique in the management of scaphoid nonunion as well as to determine the mechanism by which distraction and compression applied through the Ilizarov fixator achieves successful bony union.