We retrospectively reviewed 45 children treated between 1987 and 2002. Their mean age was 9 years (3 to 13). Fifteen patients had subacute osteitis. Only patients with Bledhill and Roberts type II, III and IV were included. Biopsy provided histological proof of subacute osteitis in nine patients, and six were successfully treated non-surgically with flucloxacillin. Six patients had Ewing’s sarcoma, 24 had osteosarcoma, 23 Enneking stage-IIB (extracompartmental) and one Enneking stage-IIA (intracompartmental).

The preoperative clinical signs, radiographs and MRI studies were reviewed. On plain radiographs, cortical destruction and periosteal reaction were assessed. On MRI the extent and nature of bone marrow involvement and the size of the soft tissue mass/oedema was analysed and correlated clinically. On plain radiographs, cortical destruction was present in all patients with Ewing’s sarcoma and stage-IIB osteosarcoma and in 50% of patients with subacute osteitis. An ill-defined zone of transition was found in all patients with Ewing’s sarcoma and osteosarcoma and in 50% of those with subacute osteitis. These findings therefore did not help to differentiate between the two groups.

The periosteal reaction was well-defined in subacute osteitis and lucencies between laminations were thin. In the malignant group the periosteal reaction was always ill-defined, with or without a Codman’s triangle, sunray spicules and hair-on-end. Lucencies between laminations were broad and broken. This was useful in differentiating between the two groups.

On MRI, patients with subacute osteitis had no soft tissue mass, with an infiltrative type of bone marrow involvement. In the malignant group, the soft tissue mass was large and the bone marrow involvement well demarcated.

We concluded that where there was a well-defined periosteal reaction on plain radiographs, and no soft tissue mass with infiltrative bone marrow involvement on MRI, patients could initially be treated as subacute osteitis without biopsy.

In an attempt to formulate a classification that might facilitate prognostication of outcome and possibly dictate early intervention, we conducted a retrospective review of fractures in which the extent of diaphyseal comminution was greater than 8 cm and resulted in nonunion. We looked at 150 femoral diaphyseal and 100 tibial diaphyseal fractures caused by gunshots and treated at our institution.