La valeur prédictive du cliché en traction, par rapport à la réduction postopératoire, a été évaluée sur 40 courbures idiopathiques de l’adulte (37 thoraciques et 3 thoraco-lombaires) raides (angle de Cobb préopératoire supérieur à 60° et réduction sous traction de moins de 35 %) traitées par libération/réduction postérieure et instrumentation segmentaire. De plus, la valeur prédictive obtenue pour le seul groupe des 37 thoraciques raides a été comparée à celle obtenue avec 56 courbures thoraciques souples.

L’angle de Cobb postopératoire des courbures raides était fortement corrélé avec l’angle préopératoire en traction (r = 0,91 ; P < 0,0001). Les clichés en traction ont prédit moins de correction de l’angle de Cobb que celle obtenue en post-opératoire. Cette différence était de 17,5° (± 7°) ce qui correspond au gain supplémentaire en réduction obtenu après chirurgie. Pour les seules courbures thoraciques, les différences entre les angles de Cobb sous traction et les angles de Cobb postopératoires étaient de 14,5° pour les courbures raides et de 6,5° pour les courbures souples.

L’un des intérêts du cliché en traction, du fait que l’angle postopératoire peut être prédit avec une faible marge d’erreur, serait de pouvoir juger un résultat postopératoire en fonction, par exemple, de l’association ou non d’une libération antérieure.

The aim of this study was to determine the predictive value of the traction radiograph in adults with stiff curve (preoperative Cobb angle >60° and reduction of less than 35% with traction) thoracic scoliosis. We wanted to compare this predictive value with that observed in reducible scoliosis.

A traction radiograph was obtained using a standard protocol with dynamometric control of the force applied. Patients with stiff scoliosis had 47 thoracic curves and 11 thoracolumbar curves (with primary anterior release for ten thoracic curves and eight thoracolumbar curves) and patients with reducible scoliosis had 56 thoracic curves. Cortre-Dubousset instrumentation was used for treatment in all patients.

The postoperative Cobb angle for the stiff curves (without anterior release) was strongly correlated with the preoperative angle with traction (r=0.91; p<0.0001). The correlation between the postoperative Cobb angle and the preoperative Cobb angle without traction was less pronounced (r=0.86; p<0.0001). The traction radiographs were less predictive of the Cobb angle correction than the postoperative Cobb angle. The difference was 17.5±7°, which corresponds to a supplementary gain in reduction after surgery. For the thoracic curves alone, the differences between the traction Cobb angle and the postoperative Cobb angle was 14.5° for stiff curves and 6.5° for reducible curves. Furthermore, the correlation between the Cobb angle with traction and the postoperative Cobb angle was stronger for stiff curve thoracic scoliosis (r=0.90) than reducible thoracic scoliosis (r=0.78).

The standard traction radiographs were highly predictive of postoperative reduction of stiff thoracic and thoracolumbar curves treated by segmental instrumentation. The postoperative result can thus be estimated with a margin of error of ±7°. For the stiff curves, the postoperative Cobb angle was 17.5° on average less than predicted on the traction radiograph (on average 20% supplementary gain in reduction). This angle gain, which was greater for stiff than reducible curves, corresponds to the determining effect of release occurring with stiff curves. Finally, because of the stronger correlation between the traction Cobb angle and the postoperative Cobb angle for stiff curves, the predictive value of the traction radiograph is greater for stiff curves than for reducible curves. In conclusion, one of the contributions of the traction radiograph, which results from the predictability of the postoperative angle with a small margin of error, is to enable adequate prediction of the postoperative outcome for a given patient or a specific sub-group of patients, e.g. with or without anterior release.