Schroth physiotherapeutic scoliosis-specific exercises for adolescent idiopathic scoliosis: how many patients require treatment to prevent one deterioration? – results from a randomized controlled trial - “SOSORT 2017 Award Winner”

The evidence on physiotherapeutic scoliosis specific exercises (PSSE) is rapidly growing and getting stronger. Several randomized controlled trials (RCT) investigating the effect of PSSE in adolescents with idiopathic scoliosis (AIS) have been published. Monticone et al., in their long-term RCT found that PSSE consisting of active self-correction and task-oriented exercises, consistent with Scientific Exercise Approach to Scoliosis (SEAS) [1, 2] improved Cobb angles by 5.3° at skeletal maturity in patients with AIS, while traditional exercises were associated with stable curves. [3] Kuru et al., in their 6-month long RCT [4] in patients with AIS concluded that Schroth PSSE improved the Cobb angle by 2.5° in the supervised group, while the patients who were not supervised deteriorated by 3.3° which was similar to controls (deteriorate by 3.1°) receiving no treatment [4]. Our RCT found that patients with AIS receiving Schroth PSSE for six months in conjunction with standard of care consisting of observation or bracing improved curves by 1.2°, while the controls receiving only standard of care deteriorated by 2.3°. [5] Monticone et al. in an RCT conducted on adults with idiopathic scoliosis with childhood onset undergoing a 20-week long SEAS PSSE or usual physiotherapy, found that the Cobb angle in the experimental group improved by 2.9° and in the control group deteriorated by 2.1° [6]

All these RCTs reported statistically significant results favouring PSSE but none reported on clinical significance. Assessing the clinical significance is important to interpret the results in a way, which is understandable to the end users (clinicians and patients), as well as to facilitate appropriate knowledge transfer. [7–10]

Reporting statistical test results should be supplemented with methods for determining clinically significant change [11–14]. The concept of “clinical significance” basically consists of determining whether “this matters in the real world of clinical medicine.” [9] Several definitions have been proposed for “clinically meaningful effect”. One definition is “the extent to which therapy moves someone outside the range of scores typical of the dysfunctional population or within the range of the functional population” [12]. Another one states that “a clinically significant change is a difference score that is large enough to have an implication for the patient's treatment or care” [15]. These definitions are not associated with a definite statistic to determine the meaningfulness of an observed improvement or deterioration over time with clinical treatment. Therefore, in an effort to evaluate whether the treatment has had a clinically significant effect on a patient’s outcome variable, several approaches have been proposed.

Researchers most commonly assess the minimal clinically important difference (MCID), and use it as a threshold for determining if a patient experienced meaningful change. The MCID is “the smallest difference in score in the domain of interest that patients perceive as important, either beneficial or harmful, and which would lead the clinician to consider a change in the patient’s management” [7]. Two methods are used to establish the MCID: anchor- and distribution-based. Anchor-based methods use an external indicator (anchor) to assign participants into several groupings reflecting the importance/magnitude of their changes in outcome measurement (target). [16] The most commonly used anchor is the Global Rating of Change (GRC) [17]. Distribution-based methods rely on statistical properties of the outcome tool of interest to quantify how much change is deemed clinically important.

There is still no consensus as to which method is the most appropriate. [15, 18] Anchor-based methods because of their reliance on the GRC can be affected by recall bias. [18] Patients often do not remember accurately their health status before the treatment. Distribution-based methods depend only on the reliability of the measurement tool and generally ignore whether outcome changes are perceived as important. [18]

The number needed to treat (NNT) is the number of patients that need to be treated to prevent one bad outcome in a period. A low NNT indicates that a therapy has a positive outcome in most patients offered the therapy. The NNT is easier to interpret by the public than probabilities, and is therefore a useful measurement of the clinical effect.

Therefore, the objective of this secondary analysis of data from a RCT [5, 19, 20]study was to determine how many patients require Schroth PSSE added to standard care (observation or brace treatment) to prevent one progression of the Largest Curve (LC) by more than 5° or of the Sum of Curves (SOC) by more than 10° over a 6-month interval.

The ARR indicated that, with the Schroth intervention, there was a 28% and 32% absolute reduction in the risk of curve progression over the respective clinically important thresholds associated with the LC and SOC, respectively (Table 4). Further, the RRR indicated that the Schroth intervention reduced the risk of curve deterioration in excess of 5° for the LC by 70% and exceeding 10° for the SOC by 73% (Table 4).

This first study to assess the NNT in patients with AIS undergoing Schroth exercise intervention demonstrated the clinically important effect of the Schroth intervention. We calculated that over a 6-months period one additional person will avoid LC deterioration by >5° or SOC deterioration by >10°, respectively, for every four participants undergoing Schroth PSSE intervention in addition to standard of care compared to receiving standard of care only.

While the health consequences of progression of LC by >5° or of the SOC by >10° over six months are not fully documented, it is clear that the goal of every scoliosis therapy is to stop the curves from progression into a range where a change in care would be warranted. For a patient under observation, progression of the LC by >5^o might lead to a brace prescription, and for a participant in brace, progression might warrant a modification of the brace prescription or a recommendation for surgery. Therefore, avoiding a more aggressive treatment even if only for six months is clinically meaningful.

Monticone et al. [2] also reported results favouring SEAS PSSE at maturity. None of the patients from the SEAS PSSE group deteriorated by >5^o, 62% improved, and 38% remained stable suggesting that 100% were treated successfully [2]. In the control group receiving general physiotherapy, none improved, 4 (8%) deteriorated, and 47 (92%) remained stable. We calculated NNT based on those results as 12.8 with CI containing “0”, suggesting that the treatment effect was not significant. However, there were 61.5% more patients who improved beyond 5° in the PSSE group, clearly suggesting the benefit of PSSE over the long term. This lack of clinical importance in the treatment effect in this study is not clinically significant in terms of prevention of deterioration by >5°, could be contributed to the fact that the girls under investigation had low risk of progression (35% risk of progression according to Lonstein and Carlson’s formula) [30], and the comparator was an active therapy.

While ours was the first study reporting NNT for exercise treatment, NNTs have been reported for brace management. The Bracing in Adolescent Idiopathic Scoliosis Trial (BrAIST) study, which was partly a randomized and partly a patient-preference trial, reported that 3.0 (95% CI, 2.0 to 6.2) patients needed to be treated in order to prevent one case of curve progression warranting surgery based on the results from the randomized cohort and applying intention-to-treat methodology. [31] This is not in line with Sanders et al. [32] results in a prospective follow-up of a cohort of 126 immature patients (Risser 0–2) with AIS and curves between 25° and 45° treated with a Boston brace and brace wear to the end point of progression to surgery. The noncompliant patients were compared both with highly compliant patients and with the entire cohort. The authors reported a NNT of 7 to avoid surgery. However, the 95% CI included “0”, indicating that the treatment effect was not significant, so the NNT cannot be directly interpreted. For the highly compliant patients (bracewear >14 h) compared with non-compliant patients, the NNT (3.0; 95% CI, 2.0 to 7.0) was similar to that reported in the BrAISt and the present study.

Nachemson and Peterson multicenter study compared observation and bracing for girls with thoracic major curves and Cobb angles of 25° to 35°. [33] In their study they found that 17 of 88 braced patients with full follow-up deteriorated by >6°, and 58 of 120 observed patients, which gives NNT of 3.4 (95% CI 2.5 to 6.2). However, in the worst-case analysis when the dropouts are considered failures, there were 36% failures among braced patients and 56% among observed cohort, producing the NNT of 6.3 (3.6 to 30.3).

Danielsson et al. used some of the subjects from Nachemson and Peterson study, and determined the treatment failure at a mean of 16-year follow-up, found that none of the 41 initially braced patients progressed by >6°, whereas 40% of those initially under observation progressed. However, there was no difference in progression between the two groups after maturity (5.7° in the braced group and 7.0° in the observed patients). The authors concluded that since “70% of the observed patients during the original study period did not require any other treatment, 70% of the initially braced patients can therefore be regarded as having been treated unnecessarily.” [34] Of the remaining observed patients, who needed a treatment, only 10% required surgery, meaning that 10 patients need to be brace-treated to avoid 1 surgery (NNT = 10; 95% CI 5.7 to 23.9).

In our sample, 17 (68%) patients wore a brace in each group. When a good compliance was assumed as brace wear of >16 h, there were 8 (47%) compliant patients in the Schroth PSSE group and 7 (41%) in the control group. There were 20 (80%) patients in the experimental group with >75% compliance with daily home exercise program. Of those who wore a brace in the Schroth PSSE group, 8 (47%) non-compliant and 7 (41%) compliant patients with brace had improved or stable curves. There was 1 (6%) compliant and 1 (6%) non-compliant patient with a brace who deteriorated. Of the 5 (20%) patients with <75% compliance with the home exercise program, 3 (60%) did not have a brace; of 2 that did have a brace, 1 (50%) was compliant with it and 1 (50%) was not. Of the eight patients in the Schroth PSSE group who did not wear a brace 7 (88%) improved, and 1 (12%) deteriorated.

While our sample size was relatively small, the distribution of compliant and non-compliant patients with a brace between the groups was similar. In addition, there was a similar number of compliant and non-compliant patients with brace who improved or deteriorated in the experimental group. Therefore, the compliance with a brace did not have a direct influence on the effect of the intervention, as long as the patients were compliant with the exercises. However, this study did not compare Schroth PSSE with bracing alone, so this should be interpreted with caution.

There is a general consensus that current brace indications lead to overtreatment. [32, 34] Many patients wearing a brace do not need to wear one because their curves would naturally not increase to the surgical range, and only estimated 10% of braced patients would avoid surgery. [34] We showed that four participants undergoing Schroth PSSE intervention in addition to standard of care and not standard of care only would need to be treated in order to see one additional curve improvement over a 6-months period.

We observed that adding Schroth PSSE might address a need and offer a treatment complement in patients who are not fully compliant with brace treatment. In our sample, despite promoting both exercise and brace compliance, of nine patients reporting wearing their a brace less than 16 h a day in the exercise group, 8 (89%) were highly compliant with exercises.

Our study was designed to determine whether adding Schroth PSSE to standard of care (observation and bracing) would lead to better outcomes, as compared to standard of care alone. To assess the differences between brace vs. exercises alone, we would need to deny the brace treatment to the patients who meet the SRS bracing criteria, which would raise ethical concerns. Despite the clinically significant results in our study, the overarching question of who will benefit from wearing a brace, who from doing the Schroth PSSE, and who from a combined treatment still remains.

The short-term Schroth PSSE intervention added to standard care provided a clinically important benefit illustrated by low NNT when compared to standard care alone. Results suggest that four (LC, SOC) patients require treatment for the additional benefit of a 6-month long Schroth PSSE intervention added to standard of care to be observed beyond the standard of care in at least one patient.