A controlled trial of anterior spinal fusion and débridement in the surgical management of tuberculosis of the spine in patients on standard chemotherapy: a study in two centres in South Africa. Seventh Report of the Medical Research Council Working Party on tuberculosis of the spine.

In 2 centres in South Africa 159 patients with a diagnosis of tuberculosis of the thoracic and/or lumbar spine were allocated at random to operation by radical resection of the spinal lesion and insertion of autologous bone grafts (Rad. series) or by débridement of the spinal focus (Deb. series). All the patients were treated with isoniazid plus PAS daily for 18 months and daily streptomycin for the first 3 months. They have been followed up to 36 months. The main analysis of this report concerns 55 Rad. and 52 Deb. patients. The clinical and radiographic condition of the 2 series on admission was similar. Central nervous system involvement was common; 27% of the patients were paraplegic and 12% paraparetic. The mean period of initial stay in hospital was 179 days for the Rad. and 133 days for the Deb. patients and of strict recumbency 97 and 37 days, respectively. Seven per cent of the Rad. and 12% of the Deb. patients had a sinus and/or clinically evident abscess initially, all of which resolved by 18 months. Of 23 Rad. and 8 Deb. patients with a radiographyically evident mediastinal abscess initially, but never a sinus or clinically evident abscess, the shadow had disappeared by 12 months in 91% and 88%, respectively. It was still present in only 1 patient (Rad.) at 36 months. Radiographically complete bony fusion of the affected vertebral bodies had occurred in 11% of 46 Rad. and 13% of 45 Deb. patients by 6 months, in 39% and 22% by 12 months, in 59% and 33% by 18 months (P = 0.05) and 74% and 69% by 36 months. The mean vertebral loss on admission was 1.0 vertebra in the Rad. and 1.1 in the Deb. series. At 36 months there was a mean loss of a further 0.2 of a vertebra in each series. The mean angulation of the spine at the start of treatment was 33 degrees for the Rad. and 35 degrees for the Deb. series and the mean increase at 3 years was 15 degrees and 8 degrees (P = 0.06), respectively. At 18 months, 64% of the Rad. and 69% of the Deb. patients had a favourable response to treatment and at 3 years 81% and 88%, respectively, non-significant differences. Patients with central nervous system (CNS) involvement on admission stayed in hospital initially for an average of 204 days compared with 129 days for those with no CNS involvement, the period of strict recumbency being 84 days compared with 60 days. The prognosis was also less favourable, 14% compared with 2%, respectively, died (P = 0.03) and 32% compared with 3%, respectively, had an unfavourable response at 36 months (P = 0.0003). At 6 months the paralysis had resolved in 6 (21%) of the 29 paraplegic patients compared with 7 (54%) of the 13 paraparetic patients (P = 0.07). The corresponding proportions at 21 months were 72% and 69%. At 18 months, 28% of the paraplegics responded favourably compared with 62% of the paraparetics (P = 0.08) but at 36 months the proportions were similar, namely 65% and 67%. Specimens from spinal lesions were obtained at operation in 149 patients...