Long-term follow-up reveals non-utility of nonsurgical management in moderate adolescent idiopathic scoliosis: a comprehensive meta-analysis

Article information

Asian Spine J. 2025;19(4):638-651

Publication date (electronic) : 2025 June 24

doi : https://doi.org/10.31616/asj.2024.0358

Eduardo Hevia¹, Jesús Burgos², Vicente García³, María Teresa de Santos Moreno⁴, Ibrahim Khalil⁵, Gonzalo Mariscal^6,7, Carlos Barrios⁷

¹Spine Unit, University of Navarra Clinic, Madrid, Spain

²Vithas Internacional, Madrid, Spain

³Spine Surgery Section, Araba University Hospital, Vitoria, Spain

⁴Neuropediatrics Unit, San Carlos Clinic Hospital, Madrid, Spain

⁵Faculty of Medicine, Alexandria University, Alexandria, Egypt

⁶Mediterranean Observatory for Clinical and Health Research (OMEDICS), Valencia, Spain

⁷Institute for Research on Musculoskeletal Disorders, Valencia Catholic University, Valencia, Spain

Corresponding author: Ibrahim Khalil, Faculty of Medicine, Alexandria University, Alexandria 5372066, Egypt, Tel: +20-1141579270, E-mail: ibrahim.ibrahim1908@alexmed.edu.eg; ietfeiheh2016@gmail.com

Received 2024 September 2; Revised 2024 October 22; Accepted 2024 November 13.

Abstract

Graphical Abstract

Introduction

Adolescent idiopathic scoliosis (AIS), the most prevalent type of scoliosis, affects 2%–3% of adolescents worldwide, although its underlying causes remain unclear [1]. AIS is characterized by a lateral curvature of the spine exceeding 10°, typically emerging during puberty without an identifiable medical cause [1]. Clinical guidelines recommend bracing for moderate curves between 25°–30° and 40°–45° [2]. Bracing aims to prevent curve progression through the application of external force. However, its effects are complex, with studies showing post-treatment loss of correction and improvement. Its efficacy also depends on wear time, with greater adherence linked to superior radiographic outcomes [3].

The long-term efficacy of bracing and the impact of factors including patient age, curve flexibility, skeletal maturity, and adherence to wear protocols, require further investigation. Untreated AIS can have psychological consequences [4]. While bracing aims to avoid surgical intervention, its influence on curve behavior has not been fully elucidated. Studies have yielded mixed results, with some studies reporting decreased correction over time, while others have documented improvements [4]. Treatment strategies vary based on curve severity. Mild curves with minimal progression risk may require only observation after bracing, whereas severe curves often necessitate surgical intervention for spinal alignment. Although bracing seeks to correct spinal curvature, it is not without limitations and uncertainties. Bracing imposes restrictions on movement and positional freedom, and its long-term impact after removal remains unclear.

According to Gabos et al. [5], braces impose positional limitations, and the durability of correction after removal remains uncertain. Discomfort is a common complication, with incidence rates reaching up to 36% in various spinal regions [6,7]. Notably, the long-term impact of bracing on curve behavior is still poorly understood, as acknowledged by Aulisa et al. [8]. Furthermore, Danielsson et al. [9] in 2010 also highlighted the significant knowledge gaps in understanding the natural history of AIS after bracing and its quality-of-life implications. The current literature is characterized by predominantly short-term follow-up periods, whereas the long-term effects of bracing on curvature behavior and patient well-being remain understudied. Moreover, patient-reported outcome measures (PROMs) that capture satisfaction and perspective over extended periods are scarce. The marked heterogeneity in individual study results underscores the need for a well-powered meta-analysis to synthesize multicenter data, providing insight into critical issues such as long-term efficacy, pain impact, and natural history.

Previous reviews of bracing efficacy have significant limitations. A systematic review conducted in 2008 only provided descriptive comparisons of pre-post means across studies, omitting meta-analyses, long-term follow-up assessments, quality-of-life evaluations, or adherence to contemporary reporting standards such as Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) [10]. Another meta-analysis of 18 studies was restricted by its focus on skeletal maturity, resulting in follow-up periods as brief as 2–5 years, which is insufficient to address the durability of effects over decades [11].

Heterogeneous and predominantly short-term results from previous studies underscore the need for a rigorous meta-analysis synthesizing multicenter evidence. This review aims to systematically evaluate bracing effects, assessing curve progression using radiographic outcomes and patient-centered measures of functioning and well-being over an extended follow-up period of ≥10 years.

Materials and Methods

Assessment of risk of bias in the included studies

The Methodological Index for Non-Randomized Studies (MINORS) criteria were used to evaluate the quality and risk of bias in the included studies (Table 1) [4,6,7,9,19–28]. This 12-tem tool assesses methodological conduct and presentation, with scores ranging from 0–2, yielding a maximum score of 24 for comparative studies and 16 for single-group designs, indicating higher quality. The MINORS scores were categorized into four quality levels: very low, low, fair, and high. For non-comparative studies, scores of 0–4 corresponded to very low quality, 5–7 corresponded to low quality, 8–12 corresponded to fair quality, and ≥13 corresponded to high quality. For comparative studies, scores of 0–6 corresponded to very low quality, 7–10 corresponded to low quality, 11–15 corresponded to fair quality, and ≥16 correspond to high quality [19]. Two reviewers independently evaluated these scores. Any discrepancies in assessment between the two reviewers were resolved by consensus.

Table 1

Assessment of the quality of studies through MINORS

Results

Study selection

A total of 277 studies were retrieved through a search of electronic databases. After screening titles and abstracts, 251 studies were excluded due to duplication, non-relevance, or ineligibility. The remaining 26 studies underwent full-text review, resulting in the exclusion of 15 additional studies. These exclusions were made due to incorrect focus, incomplete data, duplicate studies with different follow-up periods (with the longest follow-up selected), or curve magnitudes outside the specified range of 30º–45º. This resulted in 11 studies. After screening the reference lists of the individual studies, two additional studies that met the inclusion criteria were also included. Finally, 13 studies were included in the meta-analysis (Fig. 1) [4,6,7,9,20,32–39]. To avoid duplication and ensure independent observations, studies reporting on the same patient sample as an already included publication were excluded if they offered shorter follow-ups or additional variables that did not provide significant new evidence. This approach prevented overlapping study cohorts from introducing unwarranted duplication into the analyses.

Fig. 1

Study selection flow diagram (Preferred Reporting Items for Systematic Reviews and Meta-analyses).

Study characteristics

The baseline characteristics of the included studies are shown in Table 2 [4,6,7,9,20–28]. All 13 studies included in this review had a retrospective design. These studies enrolled a total of 1,492 patients, with the follow-up periods ranging from 12 to 42 years. The average curves included at the initial time point of the studies ranged from 30° to 39°. Age and sex distributions are shown in Table 2.

Table 2

Baseline data of 13 included studies

The bracing protocols employed in studies that reported this information are detailed in Supplement 2 [4,6,7,9,21,23,24,26,28,40]. Study quality was rated as high in 9/13 studies and fair in 4/13 studies (Table 1).

Curve progression and surgery

The pooled mean progression across the included studies was −1.76 (95% CI, −4.97 to 1.46; studies=11; I²=92%) with no significant differences observed pre- and post-follow-up (Fig. 2) [4,6,7,9,20,21,23–26,28]. However, this estimate may be conservative, as patients with significant curve progression often undergo surgery, thereby preventing further deterioration. Consequently, the reported values likely underestimate the true progression, as surgical intervention effectively truncates the curve’s potential worsening. Ascani et al. [22] only reported the mean progression at the final follow-up, showing a progression of 11.6° at the final follow-up. Studies with follow-up duration of <20 years had less progression than those with >20 years of follow-up: (MD, 0.94; 95% CI, −3.58 to 5.47; studies=5; I²=95%) vs. (MD, −5.69; 95% CI, −11.66 to 0.29; studies=6; I²=87%). Analysis of studies with over 25 years of follow-up revealed significant curve progression at final follow-up (MD, −14.32; 95% CI, −20.14 to −8.50; studies=2; I²=0%).

Fig. 2

Forest plot showing mean curve progression before and after conservative treatment. SD, standard deviation; IV, inverse variance; CI, confidence interval; df, degree of freedom.

The percentage of patients considered as having significant progression (≥5°) was 35.68 (95% CI, 22.85 to 48.50; studies=6) (Fig. 3) [6,9,21,23,25,27]. Studies with <20 years of follow-up had a lower incidence of significant progression compared to those with over 20 years of follow-up: 32.89% (95% CI, 0.00 to 74.44) vs. 37.07% (95% CI, 15.25 to 58.90). On analysis of studies with >25 years of follow-up, the incidence of significant progression was 46.70% (95% CI, 46.52 to 46.88).

Fig. 3

Forest plot illustrating the percentage of patients with significant progression (≥5°). SE, standard error; IV, inverse variance; CI, confidence interval; df, degree of freedom.

The percentage of patients requiring surgery at final follow-up was 14.20% (95% CI, 0.87 to 27.53; studies=2) (Fig. 4) [26,27]. The small number of studies prevented subgroup analysis. These studies did not provide the justification and indications for surgical intervention.

Fig. 4

Forest plot demonstrating the percentage of patients requiring surgery at the end of follow-up. SE, standard error; IV, inverse variance; CI, confidence interval; df, degree of freedom.

Sagittal profile

TK increased at the end of follow-up but not significantly (MD, −2.18; 95% CI, −5.26 to 0.91; participants=219; studies=3; I²=65%) (Fig. 5A) [20,23,24]. In all studies, the final TK remained within the normal range (20.9°–36.4°). LL decreased at the end of the follow-up but there were no significant differences (MD, 3.11; 95% CI, −3.18 to 9.39; participants=220; studies=3; I²=80%) (Fig. 5B) [20,23,24]. Similarly, all curves were within the normal range (44.0º–54.6º), except in the study by Simony et al. [20] (38.9º). The small number of studies providing this data prevented subgroup analysis.

Fig. 5

Forest plots showing thoracic kyphosis (A) and lumbar lordosis (B) at the end of the follow-up period. SD, standard deviation; IV, inverse variance; CI, confidence interval; df, degree of freedom.

Quality of life and functionality

The percentage of patients with global back pain at the end of follow-up was 63.35 (95% CI, 38.36 to 88.34; studies=2) (Fig. 6) [6,7]. Table 3 shows the final mean scores of different quality-of-life scales in studies reporting this data [7,20,24,25,27,28]. Analysis of the cutoff points revealed that GFS lacked defined thresholds to establish the degree of improvement. The ODI values ranged between 7.0 and 11.8 points, indicating minimal disability. The EQ-5D demonstrated high quality in two studies and acceptable quality in one, suggesting a favorable overall quality of life. Regarding the SRS-22 scores, no cutoff values were established. For the SF-36, studies reported these variables had a mean age range of 39–41 years [7,20]. According to the literature, patients in this age group presented a quality of life between the 10th and 20th percentile for the physical component summary, indicating significantly impaired physical health (scale: 0=worse quality of life; 100=better quality of life). The mental component summary values fell between the 30th and 40th percentile.

Fig. 6

Forest plot showing the percentage of patients with pain at the end of the follow-up period. SE, standard error; IV, inverse variance; CI, confidence interval; df, degree of freedom.

Table 3

Quality of life and functionality assessed by different scales^a)

Publication bias

Fig. 7 presents the funnel plots assessing publication bias. Evidence of publication bias was found for two primary variables: curve progression (Fig. 7A) and percentage of patients with significant progression (Fig. 7B). This was supported by the asymmetry and clustering of studies in the upper part of the funnel.

Fig. 7

Funnel plot assessing publication bias for curve progression (A) and percentage of patients with significant progression (B). SE, standard error; MD, mean difference.

Discussion

The curve progression and surgery rates in the studies included in this review varied depending on the follow-up duration. Longer follow-up periods were associated with an increased risk of progression over time. Studies with shorter follow-up durations (<20 years) reported less curve progression and lower surgical requirements than longer-term studies. However, studies with follow-up periods exceeding 25 years found a significant curve worsening in the final evaluation. The analysis revealed mixed results for quality of life and back pain outcomes. While some studies reported minimal disability and fair quality of life using the ODI, EQ-5D, and SF-36 scales, respectively, these patient-centered outcomes were assessed in only a limited number of studies. Bracing did not yield significant changes in sagittal profile, as measured by TK and LL. Both TK and LL remained within normal ranges at the end of treatment across the included studies.

Studies with follow-up periods of less than 20 years showed minimal curve progression, suggesting that the effects of conservative treatments might not be fully apparent in shorter study durations. Notably, when follow-up exceeded 25 years, significant progression was observed, with curves initially ranging from 30° to 45° progressing on average by 14°. By the end of 25 years, these curves reached between 44° and 59°, underscoring the potential need for reconsidering treatment strategies as patients age. This long-term progression underscores the need for extended monitoring beyond skeletal maturity to enable a comprehensive assessment of the efficacy of conservative therapies and better characterize their long-term impact on spinal health and quality of life. Given the increased life expectancy, it is imperative to continue ongoing vigilance and further long-term follow-up. These findings indicate that while initial treatments may delay progression, their effectiveness may wane over time, underscoring the need to evaluate long-term functional outcomes and patient satisfaction. Furthermore, similar brace protocol durations across studies, with limited variance, indicate a standardized approach to conservative treatments during the study period. However, the long-term impact of these treatments, particularly concerning patient functionality, quality of life, and satisfaction, requires further investigation and longer follow-up assessment.

Although our inclusion criteria specified curves measuring 30°–45º, based on the average curve measurements reported in the studies, individual patient data within these studies may have included curves slightly below 30º. Notably, there is no clear consensus regarding the impact of initial curve size on outcomes. For instance, Aulisa et al. [4] found no significant differences in outcomes based on the initial curve size. In contrast, Weinstein et al. [33] reported that curves exceeding 30° are more likely to progress. Meanwhile, Pellios et al. [41] observed that curves under 30° typically exhibit minimal progression. Our review revealed a critical knowledge gap in the literature, with remarkably few studies investigating the proportion of curves progressing beyond clinically meaningful thresholds, such as an increase of 5° or 6°. This metric is crucial to understanding the real-world impact of curve progression, as changes in average Cobb angle can be skewed by outliers. Unfortunately, only six of the 13 reviewed studies (less than 50%) provided this crucial data.

Our meta-analysis revealed that low skeletal maturity, assessed through various methods (Risser score, Sanders score, chronological age, menarche status, and triradiate cartilage status), consistently predicted brace treatment failure across six studies. This findings contrasts with non-meta-analytical literature reviews, such as that by Hawary et al. [13]. Additionally, several studies have identified an initial Cobb angle greater than 30°–40° as a risk factor for curve progression. However, the long-term implications of surgical interventions are crucial. For instance, procedures utilizing Cotrel-Dubousset instrumentation are associated with a significant long-term revision rate of 47.5% [36], highlighting some controversies surrounding this surgical method. These findings underscore the need for an in-depth evaluation of the impact of skeletal maturity on long-term outcomes. Our meta-analysis could not fully address this aspect due to insufficient data or lack of categorization of results based on this factor.

The existing literature lacks comprehensive information on complementary conservative strategies such as physiotherapy exercises. Although some studies have outlined brace protocols, none have thoroughly explicated complementary nonsurgical treatments that may impact outcomes. Furthermore, the included studies revealed that bracing does not significantly alter or correct sagittal alignment. This raises concerns about the potential risk of flatback syndrome without comprehensive multiplanar correction.

Studies have also not reported the specific causes necessitating surgery. It is reasonable to assume that progression drove many surgical indications, but quality-of-life impacts may also have factored during surgical decision-making. Capturing this context would enhance the comprehension of when and why treatment fails in different patients.

The correlation between brace adherence and stabilization of spinal conditions is well documented, with studies demonstrating less progression in patients wearing braces for over 20 hours daily [40,4,43]. However, the reported compliance rates, ranging from 54% to 70%, suggest that discomfort and psychological factors (such as body image concerns), which are often underreported, significantly limit adherence [44]. Suboptimal compliance presents a major challenge for achieving long-term stability after skeletal maturation. To address this issue, there is an ongoing debate and research into alternative therapies to enhance patient engagement and compliance. Innovative approaches, such as real-time feedback through sensor monitoring, more collaborative care models incorporating psychological support, and psychosocial interventions, are under exploration. Furthermore, researchers have suggested designing novel braces that enhance comfort and customization to improve adherence. While surgery is sometimes considered for non-compliant cases [37], the focus of our study and subsequent discussions should be on optimizing conservative treatment strategies to improve long-term outcomes for all patients, particularly by enhancing compliance through these innovative approaches.

Several studies have highlighted the importance of capturing PROMs to gain a comprehensive understanding of treatment impact. The study by Danielsson et al. [9], which included PROMs but only evaluated them at skeletal maturity, revealed that curves under 35° showed minimal differences from healthy controls. However, untreated or braced AIS patients reported significantly higher pain levels compared to controls. Danielsson et al. [9] also estimated that 70% of patients underwent unnecessary bracing, while surgery was avoided only in one in 10 cases (10%). Research on long-term PROMs for AIS patients during pregnancy is scarce, despite the majority being female. Most studies have focused on comparing conservative and surgical treatments, yielding no significant differences between the two approaches. Danielsson et al. [7] also reported greater functional limitations in braced AIS versus controls. SF-36 and ODI have emerged as reliable quality-of-life measures. Our findings indicate that patients with braced AIS experience impaired quality of life, as measured by SF-36, compared to the general population. Conversely, the ODI indicated minimal disability. Few studies have consistently evaluated the relationships between curve progression and patient-centered outcomes using PROMs. Although some studies have found no correlation at the individual level, capturing these metrics remains important for a comprehensive understanding of treatment impact.

Pulmonary function is a vital, yet frequently neglected aspect of evaluating AIS treatment outcomes. Despite this oversight, studies addressing this dimension have yielded crucial insights into the broader implications of both conservative and surgical interventions. Misterka et al. [21] found that vital capacity remained stable at the final follow-up in their study cohort; however, these findings provide only a limited window into true long-term respiratory outcomes [45].

A knowledge gap persists regarding the long-term effects of bracing on disc health. Given that AIS primarily affects adolescents during active growth, with a lengthy life expectancy ahead, understanding spinal degeneration risks is crucial. While bracing aims to mitigate progression by modulating vertebral biomechanics, the potential effects on intervertebral discs warrant investigation. Whether conservative options influence this outcome is unknown [46].

The studies reviewed in our analysis exhibited a strong element of selection bias. Frequently, patients exhibiting significant worsening of their scoliotic curves while undergoing bracing quickly transition to surgical interventions, which excludes them from the cohort analyzed for long-term curve progression under bracing alone. This practice likely skews the data, leading to an under-representation of the true extent of curve progression associated with bracing alone. Owing to this selection bias, the outcomes reported in these studies might reflect a best-case scenario rather than a comprehensive depiction of natural disease progression under nonsurgical management. It is likely that including more severe cases that progress despite bracing would significantly increase the reported average curve progression. However, withholding surgical intervention from deteriorating patients solely to observe natural disease progression raises profound ethical concerns. This selection bias undermines the validity of study outcomes, making it challenging to accurately evaluate the long-term efficacy of bracing as a standalone treatment.

Some limitations of this study should be acknowledged. The small sample size hindered robust subgroup analyses and adjusted modeling approaches. Furthermore, the retrospective nature of the data introduces potential bias compared to prospective designs. Furthermore, there was high heterogeneity (I²=92%) among the included studies regarding curve progression outcomes. Many studies focused on mean Cobb angle changes rather than reporting the clinically relevant proportion of patients progressing over specific thresholds, such as 5° or 6°. Future studies should prioritize this outcome to enable a direct comparison. Additionally, important factors such as skeletal maturity, curve location, and side were not adequately assessed. Furthermore, the exact timing of patient follow-ups was not consistently reported. Moreover, approximately half of the studies did not report data on curve progression, limiting the power of the meta-analysis. Furthermore, critical patient-reported factors were generally lacking and the determinants of low compliance were under-researched. The psychosocial impact of bracing and quality of life metrics were infrequently reported despite their essential role in enabling in-depth characterization of the treatment burden. Another major limitation of this study was the lack of baseline and very long-term PROMs, which preempts a comprehensive evaluation of the impact and effectiveness of conservative treatments on quality of life over time. Cost analyses that consider health spending and absenteeism related to nonsurgical care would provide valuable insights for decision-makers. Pulmonary function and the risk of spinal degeneration have rarely been addressed, preventing the evaluation of their systemic implications. Moreover, the rationale behind surgical decisions is often unclear, highlighting the need for distinguishing between surgeries driven by curve progression and those aimed at improving quality of life. Clarifying these distinctions would enhance clinical understanding, allowing for more informed treatment choices. Overall, follow-up tended to end at skeletal maturity, without a long-term assessment of adult outcomes. While individual studies offer preliminary insights, the presence of high unexplained heterogeneity and inadvertent omissions hinders conclusive comparisons of optimal management strategies across diverse presentations and time frames.

Conclusions

This review indicates that a significant proportion of patients with moderate curves (30°–45º) treated conservatively experienced a progression of their curvatures that surpassed established thresholds, raising concerns that the current treatment algorithms may not adequately address such variations. Long-term monitoring over 25 years revealed that conservative treatment often failed to halt progression, with a significant proportion of the patients ultimately requiring fusion surgery. These findings challenge the efficacy of existing treatment approaches and underscore the need for re-evaluation. Standardized criteria for transitioning from nonsurgical to definitive correction must be reassessed, with a focus on tailoring interventions to individual risk profiles. Surgical interventions with fusion require careful consideration. Additionally, there is suboptimal adherence to conservative options, highlighting the need for optimized approaches to minimize the need for bracing and surgical procedures with definitive fusion. Finally, the exclusion of patients fast-tracked for surgery due to significant curve worsening under bracing alone from original studies may have underestimated true curve progression. A more comprehensive study incorporating these cases could provide valuable insights. Considering factors such as patient selection, quality of life, cost-effectiveness, and technological advancements in fusionless techniques can provide deeper insights informing a personalized and balanced treatment approach.

Supplementary Materials

Article information Continued