Even if our programmes met the criteria published in 2008 by the SOSORT , we were not able to follow the guidelines proposed by the SOSORT in 2006 , and especially the prognostic risk  as this study began in 1999. We use the CMCR brace to treat combined (thoracic or thoraco-lumbar) scolioses, without prior casting in 95.6% of cases (if scoliosis angulation is moderate: mean Cobb between 20° and 24,1°). Treatment is initiated as soon as the clinical and radiological examinations show a significant increase of the hump and the Cobb angle (> 5°).
Our population was similar to the one Wong studied in 2008 (randomized controlled trial)  to compare the effectiveness of a rigid and an elastic brace. His patients had a 20 to 30 ° Cobb angle, but lung capacity was not included in the study. The other studies we found in the literature [2, 26–29] focused on smaller groups of patients, and with a greater Cobb angulation. As lumbar scoliosis patients may be successfully treated with other types of rigid or elastic braces (3-points brace, St Etienne brace), we decided to exclude them from our study . In our daily practice, the choice of a brace is determined by the type of curvature, its severity, and age at the beginning of treatment . In scoliosis treated with a CMCR brace, the mean reducibility for all curvature types is 42% (thoracic: 51%, thoraco-lumbar: 53%, combined: 37%).
We noticed that our patients did not have a normal FVC prior to the CMCR setting up, as it was only 75% of the theoretical value (-25% change of predicted value)[26, 30].
Korovessis observed that before the introduction of the brace, this change was -11% in a comparable population treated by Boston brace and -16% in a population treated by Kennedy brace. However these two populations were different, which make their study and comparison difficult [31, 32]. It would have been interesting to compare the absolute FVC values of our patients with those of a control group, rather than with the theoretical value given by the spirometer. We found a loss of 26% in girls, whereas boys were less concerned, with a 16% FVC loss, compared to the theoretical value. Barrios and al. studied a group of scoliotic teenagers (mean age 13, mean angulation 33°) , in which the FVC was 3.04 litres with a standard deviation of ± 0.5 litre. Barrios concluded that there was no significant difference in the basic ventilatory parameters measured in static conditions, compared to his control group of 10 girls. On the other hand, he underlined a loss in the respiratory maximal exercise tolerance test response in his scoliotic group, which could not be related to the brace. In our patients, all curvature types (thoracic, thoraco-lumbar and combined) were concerned by this FVC loss, whereas in literature only thoracic curvatures were reportedly concerned . We observed that moderate-angulation scoliosis (mean Cobb between 20° and 24,1°) seems to have a negative impact on pulmonary capacity, before any conservative treatment. This might be put down to a disturbance of pulmonary physiology during the growth period (FVC loss should be considered as a scoliosis disease symptom), or to a disturbance in the biomechanics of the thoracic cage , originating from bone, muscle, or from an underdevelopment of the alveoli and pulmonary vasculature, encountered in patients with scoliosis .
Several hypotheses have been put forward to explain this disturbance: for Chu and al.  thorax deformity is responsible for the pulmonary impact of scoliosis, whereas disturbances of diaphragm or ribcage mobility are ruled out. For Jones and al.  this is due to a dysfunction between inspiratory muscles and ribcage deformity. Kotani et al.  found that respiratory movements are decreased in scoliotic patients, whereas Caro and Dubois  think that pulmonary compliance is reduced even if the ribcage is not particularly stiffened. For Giordano and al.  the movements of the hemi-diaphragm are reduced on the concave side. For Chu and al. , only patients with a severe thoracic scoliosis (from 40 to 98°) have a modified respiratory function (height of diaphragmatic domes and lung volumes while breathing in and out, explored by MRI). They found no difference between the three groups of patients (severe thoracic scoliosis, moderate thoracic scoliosis from 10 to 30°, and patients without scoliosis), whether regarding the antero-posterior and transverse thoracic diameters, or for diaphragmatic domes mobility. Chu et al. concluded that the pulmonary function is disturbed in severe scoliotic patients because lung volume is reduced on both the concave and the convex sides. Adam et al.  used a three-dimensional scanner to study the right - left ratios of pulmonary volumes in scoliotic patients and found a positive correlation between the increase of this ratio and the size of the hump. Pulmonary volumes are found to be correlated with the size of the hump and with the number of vertebrae included in the major curve. The shorter, the higher and the higher-rotated the major curve is, the more this pulmonary volume is decreased.
At the CMCR brace setting up, we found that the loss of real vital capacity (0.3 litre) compared to the value without a brace was 10% of the theoretical value. The higher the correction rate of the brace is (e.g. thoracic scoliosis), the greater the FVC is and the lower the correction rate is (e.g. in combined scoliosis), the lower the FVC reduction is. Refsum has shown that the vital capacity with Boston brace as well as Kennedy brace was significantly reduced to about 15% to 20% of the prebracing level .
In his study about the impact of bracing on the ventilatory function, Lacheretz [16, 17] compared the Lyon brace to the Milwaukee brace and found only a 5% decrease of vital capacity in a group of 33 children wearing a Lyon brace. We find this result surprising in view of our experience (unpublished data, study in progress), according to which wearing a Lyon brace has much more impact on breathing than wearing a CMCR brace.
Apart from the restriction of thoracic expansion, other factors are likely to be involved in the vital capacity decrease at the brace setting up, such as abdominal compression and its consequences on diaphragmatic function. At the CMCR brace removal, the real FVC had increased of 0.5 litre, i.e. an increase of 21 ± 4.2% compared to the initial value (related to the theoretical value, this represents a 4% increase). This positive evolution is most important in girls at a low Risser stage when the treatment was started before the age of 11 (in compliance with the SRS recommendations) .
According to the SOSORT guidelines, the aim of conservative treatment is to stop curvature progression and improve pulmonary function . We can draw the conclusion that the CMCR brace treatment does not restrain chest and lung development, even in young patients (under 11), since constraints exerted on the growing thorax preserve pulmonary and thoracic capacity.