The results of this study show the efficacy, in the correction phase, of the Risser cast and the Sforzesco brace. Only three patients (6%), out of a series of fifty with an average of 47° Cobb curves, worsened during a period of eighteen months of maximal effort of correction, while more than 50% reduced their maximal curve by at least 6° (and 14% by 15° or more). The Sforzesco brace did show results comparable to the Risser cast, having only minor differences in terms of scoliosis correction but major differences in terms of the reduction of specific spinal side effects. In fact, even if the sagittal parameters were reduced in both braces, the straightening of the spine was much higher (threefold for C7 and twofold for L3) in RCG, while it was not clinically significant in SBG.
Deepening the interpretation of our data, we must take note of the connection between three results appearing in almost all analysis:
The reduction of thoracic ATR and rib hump is higher in RCG than in SBG;
The flattening of the spine is higher in RCG than in SBG;
The thoracic curve reduction is higher in SBG than in RCG.
The above data clearly suggests that the correction action of the Risser cast is too posterior while the SPoRT concept [17, 18] of the Sforzesco brace is better balanced, since it serves to reduce the side effects while giving raise to better frontal-plane results with nevertheless good aesthetic ones (rib hump and Aesthetic Index).
The fact that thoraco-lumbar curves, as well as ATR and hump, have been better corrected in RCG than SBG should be carefully considered in the future, because this could suggest a possible subgroup of patients in which the mechanism of action of the Risser cast is more suitable than the SPoRT concept. Moreover, as a secondary result, in this study we confirmed what we had previously found regarding the higher efficacy of SEAS exercises versus the usual physiotherapy [26, 42].
Interestingly, splitting the results according to maturation parameters did not show any difference between the sub-groups (See additional file 4). A possible explanation of this unexpected result is that our data relate only to the correction phase, and that few patients are considered. Nevertheless, we do really lack more data on this respect, and future papers should carefully look at this point.
One question could be if the results of the "correction phase" remain with time and if these patients go to surgery or not. Provisionally, we can say today that 1 out of the 18 patients in the RCG went to surgery, while 1 dropped out. These results can be explained with a selection bias: in fact many of these patients already decided not to go to surgery before starting treatment. Another explanation could be that, until now, no patient in RCG worsened, while 67% improved of 5° or more: the average result at this point of treatment is a reduction of the worst curves of 7.8 ± 7.5° Cobb, with no difference according to bone age nor to treatment finished or not; 3 patients, still in treatment, have more than 50° Cobb curves (5 at start of treatment).
In the SBG, we have until now 2 drop-outs, that in a worst-case analysis should be considered as failures, while all the other patients are still in treatment and did not went to surgery (today 5 patients are over 50° Cobb versus 9 at start of treatment). The average result at this point of treatment is a reduction of the worst curves of 7.5 ± 7.5° Cobb, with a rate of improvement of 72% and of worsening of 9%: in SBG the rate of worsening is higher in R2 group (16%). These results must be considered totally provisional because patients are still in treatment, even if we will analyze them thoroughly in future studies.
Compliance to bracing is considered a key issue today [43, 44], even if measuring systems are still research tools [45–49] and not yet ready for everyday clinical usage, as should be needed in a study like this. Moreover, the RCG was a retrospective group and was therefore treated some years ago. Anyway, we must consider that in the RCG we inevitably had the highest compliance (99%) because for nearly twelve out of eighteen months of treatment the patients could not physically avoid wearing their casts, and after that the motivation is usually very high. The declared compliance obviously was not the same in the SBG (-10%), where we found all the 4 bad compliers: at any rate this was the new treatment, in which we also had fewer total hours of therapy due to the medical prescription and characteristics of a brace versus a cast. Because the compliance with the brace is really not known without some type of monitor, it is possible that the real compliance versus what was declared by the patients was much less, as has recently been shown , but this could have been an influence on the RCG for six months and on the SBG throughout the entire study. Nevertheless, the results for the two groups were totally comparable, and this gives even more strength to what was achieved in the SBG.
We are aware of very few studies in the indexed literature that compare the results of different concepts of bracing: These have shown the superiority of Boston over Charleston , TLSO over Milwaukee and Charleston , Chêneau over SpineCor , and the similarity of Milwaukee and Boston with a metal over-structure . All these studies related to patients with average Cobb angles of less than 35°. We recently proved the short-term superiority of the Sforzesco brace (SPoRT concept) over the Lyon (three-point concept) in a group of patients refusing fusion, with very-high-degree scoliosis (45 ± 7° for worst individual curves, 40 ± 10° for all curves) . To our knowledge, this is the first study to compare the efficacy of a brace versus a cast for AIS correction.
The main advantages of this study include the prospective data collection in the SBG, where all treated patients have been included. The fact that we have used a best clinical practice approach, because both treatments were as of that time considered the best possible ones by the treating team (an assessment that did not change during the study), along with the fact that we used a complete team approach, including treatment through [9, 54]. On the other hand, limitations include: the retrospective collection of data in RCG, but this is the only way to have a best clinical approach; not having included the dropouts, but this was not possible in the retrospective RCG, so that the study could not include an intent-to-treat analysis; the absence of data on reducibility of the curves through lateral bending radiographs, avoided in patients not surgically treated; and the fact of being focused only on the corrective phase (i.e., short-term results). In this regard we must consider that research in the field of bracing cannot be limited only to final results, otherwise we will have no possibility of understanding where we might possibly fail or where we have to focus in order to increase our knowledge and ability to treat patients. Obviously these results are not the final ones, but the corrective phase of AIS treatment is the starting one; and presumably the better the results of this phase, the better the final ones . The objective of this paper was not to prove the efficacy of bracing or casting but to verify whether the Sforzesco brace could be considered a valid option to substitute the Risser cast in this corrective phase of AIS treatment.