Effectiveness of complete conservative treatment for adolescent idiopathic scoliosis (bracing and exercises) based on SOSORT management criteria: results according to the SRS criteria for bracing studies - SOSORT Award 2009 Winner

This is a retrospective study on a prospective database that started in March 2003, including all visits performed since September 2003 at our institute. When the study was carried out, 6,172 patients and 21,024 evaluations had been included in the database, 3,937 patients had idiopathic scoliosis, and 685 had reached the end of treatment.

According to the SRS paper on brace studies [8], inclusion criteria at the start of treatment were: AIS; age 10 years or older; Risser test 0-2; Cobb degrees 25-40°; no prior treatment; and less than one year post-menarchal. All patients who satisfied the inclusion criteria at the start and had finished their treatment were considered in the study.

We had 44 females and four males; at the start the age was 12.8 ± 1.6 years, while Cobb degrees were 30.4 ± 4.4°. According to SRS criteria, we had subgroups for curvature types, curve magnitude and skeletal maturity (Table 1). We performed subgroup analysis for gender.

In our everyday clinic, according to a complete evidence-based clinical practice, joining evidence to clinical expertise and patients' preferences, we do not set standard treatment methods [7, 13]. Consequently, the patients included in this study were treated on an individual basis according to their needs, and a therapeutic contract was established at each visit with the patient and his/her family. We apply a full set of conservative treatments, including exercises and braces: elastic (SpineCor since two years), rigid (Sibilla-Cheneau brace) and very rigid (Sforzesco brace, but also Risser cast until three years ago). We follow the "step-by-step" Sibilla theory of treatment of scoliosis [13‐15]. in which each step represents an increase not only in the strength of treatment but also in regard to the requirements placed on the patient (Figure 1).

Most of the patients respecting the SRS inclusion criteria (44) had been treated with braces full-time (from 18 to 23 hours per day) until they reached Risser 3 when, according to our protocol, brace weaning started and lasted an average 2.2 years until Risser 5 [16]. According to the individual needs, and respecting a discussion with the patient and family so as to maximally increase compliance, we defined the hours per day of bracing in each single child: 23 (14 patients), 21 (23 patients) or 18 (seven patients) hours per day. In this retrospective case series based on the SRS criteria, we also had two patients that had been treated with eight months of Risser cast, followed by Lyon brace, and two treated with exercises only: these have been excluded from further analysis and considered as a separated group. The characteristics of the patients according to the treatment proposed are listed in Table 1.

During treatment we follow the SOSORT criteria for several years. In the Additional File 1 the answer to the questionnaire for clinical studies proposed by SOSORT is reported [9]. According to the classification proposed, we had an excellent approach, with 43 of 44 questions receiving "yes" responses.

The treatments used in this study have been carefully described in an online open-access booklet http://​www.​isico.​it/​approach/​default.​htm[13]. A short description follows:

According to the SRS criteria [8], we verified the percentage of patients unchanged, worsened 6° or more, exceeding 45° at the end of treatment, and fused, with the required two years of follow-up in 40 patients (85%). For this study we also had radiographic (Cobb degrees) and clinical results (ATR, Aesthetic Index, plumbline distances), which were also considered in terms of the percentage of change among patients over the repeatability error.

Additionally, we propose the ISICO outcomes [7], which we use in our everyday clinic (Table 2); in fact, we usually define individual outcomes case by case, according to a general medical reference setting and a criterion for the acceptability of the patient and his/her family. The ISICO outcomes can be divided into absolute (avoiding surgery), minimal and optimal. The latter are based on data from the literature which indicates the need to be as far as we can from the two recognized thresholds of scoliosis (50 degrees, i.e. the near certainty of progression in adulthood; and 30 degrees, i.e. the possibility of progression) [23]. Given these goals, in everyday clinics we continuously adapt ourselves according to what we obtain, as well as to how the patient behaves and feels. We establish and constantly renew the contract with the patient and his/her parents, who in this way are fully integrated within the rehabilitation team. This allows one to obtain some minimal results even in the most difficult patients, particularly with those who do not comply with our prescriptions for best results.

Reported compliance has also been considered. Each patient and his or her parents was carefully queried at each visit about how many hours per day the patient had used the brace, as well as the average usage during the period reported. This was compared with the prescription, and a percentage of compliance was computed.

We used the paired ANOVA, the Tukey-Kramer test, the paired t-test and chi-square analysis according to what was appropriate. Evaluating the percentage of patients changed, we considered significant clinical changes if the repeatability error was exceeded, namely:

No patient progressed beyond 45°, nor was anyone fused, and this remained true in the 85% of patients who reached the two years' follow-up (Table 3). Due to our results, we decided to add the outcome "improved" for patients whose Cobb degrees were reduced 6° or more. Generally, the number of patients who had improved was higher than those who experienced no change, with the few exceptions of double major, 36-40° and Risser 1 subgroups. We had only two patients who worsened, and this should serve to interpret the results in the subgroups cautiously; in fact, both patients had thoracic curvatures with a magnitude between 25° and 30°, and were Risser 0, while they were treated with the brace for 23 or 21 hours per day.

We found highly statistically significant reductions of Cobb degrees (Figure 2), aesthetics and ATR (Table 4), but we also had a statistically significant negative impact on the sagittal profile for C7 (-7.7 mm) and SI (-13.0 mm). Clinically, no patient worsened for ATR; for all the other parameters, fewer than 10% of patients worsened; on the contrary, improvements were very common, at up to 100% (thoraco-lumbar curvatures' ATR). Sub-grouping only showed that double-major and thoracic curvatures had worse results; the Risser cast and exercise sub-groups were too small to allow statistical analysis. In Figure 3, 4 and 5 single clinical cases are reported.

According to the ISICO outcomes, 96% of patients had minimal and 65% optimal results (Figure 6); optimal results were obtained mainly in thoraco-lumbar and lumbar curvatures, in scoliosis up to 35 degrees, and the youngest patients.

In all analysis performed, gender or patients braced only did not show any difference with the total population. The results at 2 years follow-up were not different from those at the end of treatment.

The other two published papers that followed the SRS criteria reported results which were completely different. According to Janicki [10], using a TLSO in 48 AIS, 85% of patients worsened 6° or more, 62.5% were in excess of 45° at the end of treatment, and 79% were fused; using a Providence brace in 35 patients, the outcomes were 69%, 42.8% and 60%, respectively. This was a retrospective paper, and suffered the same possible bias of the study presented here, albeit with a significant difference: Janicki and colleagues are surgeons, while we are conservative specialists. It could be hypothesized that the drop-outs in our series were mainly patient candidates for surgery who did not achieve good results (and this could explain to some extent why we did not have any patient treated surgically in our series) [7], while the opposite could be true for a surgical group (even if we don't believe in this assumption). Moreover, we had a very high compliance rate using braces full-time, while Janicki et al. had the best adherence with nightly bracing. Consequently, they concluded by stating their preference for the latter solution to increase compliance. Nevertheless, we must consider that adherence to treatment is not only a matter of the braces used but also of the total management of patients [9]. Obviously, raising the point of surgeons versus conservative specialists, we are not stating anything about competence, but only about general settings and attitudes toward treatment and interpretation of obtained results, that are possibly understood by patients and can drive their behaviours.

Another significant reason for the differences found with Janicki et al. [10] could be the type of braces used (i.e. mechanisms of action) and/or their quality (i.e. single-brace efficacy in the context of single patients). Unfortunately, we do not have any measurement system for this, apart from in-brace correction [27], which is not reported in these papers.

The other study following SRS criteria, by Coillard et al. [11], was performed on 170 patients wearing the SpineCor brace and reported at the end of treatment (efficacy analysis) that 22.9% of patients were fused, 33.5% had progression of at least 6° and 24.1% finished treatment in excess of 45°. There were 12 patients (7.1%) who withdrew from treatment and were not included in the efficacy analysis. These results were much better than those reported by Janicki [10] but worse than those reported in this study. Nevertheless, the comparison of Coillard's results with ours is coherent with what has been reported elsewhere in the literature. In fact, the effectiveness of the SpineCor brace has been reported to be inferior to rigid TLSO braces in a randomised controlled trial [2], as well as in a study with an historical control group [28].

The compliance rate can serve as a general justification for these results. SpineCor brace results have been reported by the developers of the brace; moreover, the SpineCor approach requires systematic, frequent contact with the patient by a well-trained team [29]. All these points are part of the SOSORT criteria of brace treatment management [9], which we followed in this study. Consequently, the study by Coillard et al. is more comparable to our results than to those of Janicki, given the management applied during treatment.

Another characteristic of our study should be pointed out: Apart from the application of the SOSORT criteria, each treatment has been tailored on each single patient so as to maximise compliance, as well as to allow the best possible inclusion of patients and their parents in the treating team. Not only the starting point of treatment (23 versus 21, or 18 hours per day of bracing, or even Risser cast or exercises), but also the final possible results were tailored during treatment (i.e. we decreased brace wearing individually, according to the need) in terms of the optimal or minimal results following the ISICO outcomes [7]. Finally, exercises were used as a means to increase compliance, not simply as a way to increase bracing results, as has been proved in some studies [16, 21, 22]. In practice, all treatment management was focused on the patient not only in terms of SOSORT criteria but also in terms of treatment planning.

Another possible explanation for the high compliance rate observed could be the private setting of our Institute, versus the usual Health National Service one used in the remaining of our country (Italy): nevertheless, in our view the SOSORT criteria, the psychological approach we used, the presence of a complete and well trained team play the most important role in increasing compliance.

This is the third study published with respect to SRS criteria, and it is the first one that has also fulfilled the SOSORT criteria for bracing studies. The former criteria provide the methodological framework while the latter give the clinical framework so as to gather the best possible data on this kind of treatment. The number of patients is low, but the population is comprehensively selected and cohesive.

This is a retrospective study. Ideally, we should have performed a prospective study, but our institute was established in 2003 and we have been collecting a prospective database since that time. We verified, in a preliminary analysis, that the population with respect to the SRS criteria was too low at this stage to perform an adequate prospective study. Consequently, we decided upon a retrospective analysis of all patients who completed their treatment which, at the start, respected the SRS criteria.

Another characteristic of the paper is that not simply has one standardized treatment been proposed but patients have been treated with different braces and some have been treated with exercises exclusively (even if these have been excluded from the whole group and considered as a group per se). Nevertheless, this research concentrates on a complete, conservative approach, focused on the increase of compliance through management and a clinical everyday approach. It represents the everyday clinical reality instead of a "laboratory" setting as it could be in other studies. This could be considered a weakness of the paper as well as a strength.

Other possible limits to be considered include: the high prevalence of females in this population, but this is typical of the everyday clinical reality in scoliosis treatment (moreover we did not find differences according to genders); the inclusion of two patients treated with exercises only and two with casting before bracing, and not only of braced patients, but we did not find any difference in the subgroup analyses, and the retrospective design required in our view to include all patients respecting the SRS criteria without introducing any other possibly confounding inclusion criterion; the fact that only 85% of patients reached the 2 years follow-up, but this subgroup was not different from the entire population for any basal characteristic nor any final result.

It must also be stated that for many years we have conducted our work with consideration for the SOSORT criteria [9], in fact well before they had been established, because they are part of our everyday clinical approach. In this respect, they are not something new to our work, which is totally focused on compliance that exceeds the SOSORT criteria in various respects.