After exclusions, 265 patients with tuberculosis of the thoracic and/or lumbar spine were followed for three years from the start of treatment. They were randomly allocated to four daily regimens of chemotherapy: 1) isoniazid plus rifampicin for 6 months (6HR, 65); 2) the same drugs as in 1) but for 9 months (9HR, 71); 3) isoniazid plus paraaminosalicylic acid (PAS) or ethambutol for 9 months (9P/EH, 62); or 4) the same drugs as in 3) but for 18 months (18P/EH, 67). All patients were ambulatory from the start of chemotherapy and no form of splintage or support or operation was used in any case. Over half (55%) the patients were children and one-third had sinuses or clinically evident abscesses. At three years a favourable status, defined as no sinus nor clinically evident abscess, no myelopathy with functional impairment, no surgery nor additional chemotherapy, full physical activity with disease quiescent clinically and radiographically, was achieved in 203 patients (77%) and in another 41 (15%) in all respects except radiographically. Only 20 patients (8%) had an unfavourable status the proportion being highest (19%) in the 9P/EH series. Thirteen of these were classified as unfavourable solely because they had needed additional chemotherapy; only seven still had an unfavourable status at three years. The clinical results at three years were thus excellent in all series except the 9P/EH, in which more patients had required additional chemotherapy. In the 88 patients with sinuses or abscesses on admission, the rate of resolution was similar in all the series; most lesions (83%) had resolved by 12 months.(ABSTRACT TRUNCATED AT 250 WORDS)

One hundred and fifty patients in Hong Kong with a diagnosis of tuberculosis of the thoracic, thoracolumbar or lumbar spine were allocated a random to the "Hong Kong" radical resection of the lesion and the insertion of autologous bone grafts (Rad. series) or to debridement of the spinal focus without bone grafting (Deb. series). All patients received daily chemotherapy with para-aminosalicylic acid (PAS) plus isoniazid for 18 months, with streptomycin for the first three months. After exclusions, the main analyses of this report concern 119 patients (58 Rad., 61 Deb.) followed up for 10 years. During the first five years the allocated regimen was modified because of the spinal lesion in 14 patients, but there were no further modifications between five and 10 years. No patient developed a sinus or clinically evident abscess or a neurological abnormality between five and 10 years. Bony fusion occurred earlier and in a higher proportion of patients in the Rad, series but at five and 10 years there was vary little difference between the series. Over the period of 10 years there was a mean increase in vertebral loss of 0.05 of a vertebral body in the Rad. series and 0.23 in the Deb. series. In both series most of this loss occurred in the first 18 months, with very little subsequent change in the next eight and a half years. Over the 10 years there was a mean reduction in the angle of kyphosis in the Rad. series of 1.4 degrees for patients with thoracic and thoracolumbar lesions and 0.5 degrees for those with lumbar lesions. By contrast, in the Deb. series there were mean increases in the angle of 9.8 degrees and 7.6 degrees respectively. In both series most of the changes had occurred early, and persisted subsequently. At 10 years 57 of 58 Rad. and all 61 Deb. patients had a favourable status, 50 (86 per cent) and 54 (89 per cent) respectively on the allocated regimen without modification.