Purpose: Infection of a posterior fixation can lead to a therapeutic dilemma, particularly if the extensive fixation involves a demineralised spine.

Material and methods: From 1998 to 2001, seven patients aged 19 to 76 years (mean 58) were treated with an interbody cage and an autologous graft. Four patients with scoliosis had had prior posterior surgery (mean five operations, range 3 – 9 operations). All had exhibited non-union with repeated fistulisation at each prior anterior approach, in four cases with meti-R Staphylococcus associated once with a Streptococcus and twice with an Enterococcus. Three patients suffered severe radicular pain. Three of the post-trauma patients had undergone revision procedures to remove the posterior implants. All had developed nonunion with total loss of the initial correction in two cases, one with septic instability concerning the level above the fixation. Both infections were caused by meti-R Staphylococcus, associated with an Enterococcus in one case. Two persistent posterior fistulae had been reoperated earlier. The grafts involved one to four levels with no new posterior fixation except for one patient (two accesses to the thoracolumbar junction, five lumbosacral fusions, immobilisation for four to six months with a corset). Mean duration of postoperative antibiotics was four months (3–12 months). Fusion was confirmed by the radiological aspect of the grafts on the scan obtained at a mean 22 months follow-up (minimum follow-up 12 months).

Results: There were no cases of anterior infection except for one post-trauma patient where a posterior screw touched the disc (reactivation of infection without anterior abscess, posterior approach for revision and final fusion in kyphosis). There was no appreciable improvement in correction, but the six other patients fused with a clear clinical improvement (removal of rigid corset, reduction of antalgesics, mean time 6 months). Improvement was observed in the three patients with radicular pain. One patient who had undergone three prior anterior operations underwent the minimally invasive posterior revision with no particular intraoperative problem but later presented ureteral necrosis (secondary nephrectomy).

Discussion: This interbody grafting strategy is a possible solution to salvage mechanical failures subsequent to recurrent severe infection often due to multiple germs. The technique is more difficult in older patients with complex malformations. A positive disc sample is a factor of less satisfactory outcome. In the event of prior anterior revision, an ureteral catheter is advisable to limit the risk of necrosis. Use of intersomatic cages is not a problem and has allowed us to achieve primary and secondary stability in these patients with poor bone stock and this without supplementary infection problems.