Association of proprioceptive deficits with curve progression and treatment outcomes in subjects with adolescent idiopathic scoliosis: A prospective cohort study with 1 year of follow-up

Abstract

Background Context

Proprioceptive deficits (PD) have been associated with adolescent idiopathic scoliosis (AIS). However, their long-term significance in curve progression and treatment outcomes remains uncertain. This study aimed to investigate whether spinal PD correlates with accelerated curve progression and to assess its influence on treatment outcomes.

Methods

Participants aged 10–25 years with AIS were recruited from a specialist outpatient clinic in Hong Kong and underwent baseline and 1-year follow-up assessments. Proprioception of the trunk, neck, elbow, and knee was measured using a 3-dimensional motion capture system, and spinal curvature was evaluated radiographically. Participants received observation, physiotherapy, bracing, or surgery according to clinical indications. Associations between PD and curve progression were examined using linear and logistic regression and chi-square tests, while changes in proprioception across treatment groups were assessed using 2-way analysis of variance.

Results

166 participants were evaluated at baseline, and 115 completed follow-up. Trunk PD was significantly associated with greater surgical correction. Although no significant interaction effects between PD and treatment modality were detected, significant main effects of PD were observed across all trunk tests, as well as in left neck rotation, right elbow flexion, and bilateral knee extension proprioception.

Conclusions

The current findings indicated that PD may identify a distinct subgroup with different responsiveness to surgical intervention. Further longitudinal research with extended follow-up periods and comprehensive clinical outcomes is necessary to elucidate the prognostic significance and underlying mechanisms of proprioception in AIS.

Introduction

Being the most prevalent form of spinal deformity, adolescent idiopathic scoliosis (AIS) occurs in children aged 10–18 for reasons that remain unknown [1]. In Hong Kong, approximately 3% of young individuals are diagnosed with AIS by age 19 [2]. Although the management of AIS involves a wide range of screening [3], monitoring [4], and treatment strategies [[5], [6], [7], [8]], clinical consultations are typically conducted only after the spinal curvature has progressed to a clinically significant degree [9,10]. A meta-analysis found that 42% of patients had a curve progression of more than 5 degrees [11]. Due to the unpredictable progression of the condition, both patients and their guardians often experience considerable anxiety [12]. To the best of our knowledge, AIS is considered a multifactorial disorder [13]. Among the various potential aetiologies [14]. proprioception has been identified as a significant factor [15,16]. However, the role of spinal proprioception has not been thoroughly investigated, particularly in longitudinal follow-up studies.

Impaired proprioception, which influences spinal posture and alignment, may contribute to the progression of spinal curvature in individuals with proprioceptive deficits (PD), but not in those with normal proprioception [17]. Proprioception refers to the body’s capacity to self-regulate body segments [18]. This process involves proprioceptors, which detect and transmit afferent signals to the central nervous system regarding the position, movement, and force [19]. Previous studies by our group have demonstrated an association between AIS and impaired spinal proprioception [20]. While genetic research has identified mutations in proprioception-related genes associated with idiopathic scoliosis aetiology [21], clinical investigations have also documented the presence of PD in scoliotic adolescents [22]. Our prior work has addressed the technical challenges of examining proprioception within scoliosis populations. We have developed a 3-dimensional motion analysis tool capable of assessing spinal proprioception [23], established normative cut-off values from healthy controls to identify abnormal proprioception [24], and found that 29% of subjects with AIS exhibited trunk PD [20]. Additionally, this subgroup showed a 2.2-fold increased likelihood of presenting a more severe curvature compared to those without PD [20].

To date, the relationship between curve progression and PD remains unclear. Consequently, a prospective study was designed to clarify this association. In this investigation, we planned to follow a cohort of scoliotic patients, both with and without PD, over 1 year to determine whether spinal PD contributes to accelerated curve progression. We also aimed to assess the impact of PD on treatment outcomes.

Methods

The manuscript complies with the guidelines specified in the strengthening the reporting of observational studies in epidemiology (STROBE) statement. The detailed study protocol has been published elsewhere [20,23,24].

Design

This research was conducted as a prospective, observational cohort study. Ethical standards outlined in the Declaration of Helsinki were strictly adhered to. Approval was obtained from the institutional review board of the University of Hong Kong and Hospital Authority Hong Kong West Cluster (reference: UW 20-525), and the study protocol was registered on ClinicalTrials.gov (identifier: NCT04682379). Written informed consent and assent were obtained from all participants and their guardians prior to enrollment. Study subjects who were assessed at baseline underwent a follow-up assessment 1 year later. During this period, interventions such as physiotherapy, bracing, and surgery were administered based on the severity of the spinal curve. Follow-up data on treatment outcomes were systematically evaluated. Due to the variability in curve magnitudes, the sample sizes within treatment subgroups were expected to be limited, and compliance with treatment protocols was not monitored.

Setting

Before each testing session, participants were instructed to abstain from engaging in moderate or vigorous physical activity. They were advised to wear appropriate attire, such as a bra top for females or no top for males, in combination with shorts. The session consisted of 6 proprioception assessments, comprising 3 for the trunk and 1 each for the neck, elbow, and knee. All assessments were conducted in triplicate, with rest intervals provided both within and between tests. Outcome measurements were consistently recorded at both baseline and follow-up.

Experiment

Preparation

During the trunk proprioception assessment, participants were seated in a height-adjustable chair, with their feet shoulder-width apart and their knees flexed to 90 degrees. They crossed their arms over their chest and were instructed to maintain their gaze at a perceived horizontal level. For the neck proprioception assessment, participants rested their arms on their thighs in a similar seated posture. In the elbow proprioception assessment, participants sat with their arms beside their bodies and palms facing anteriorly. Regarding the knee proprioception assessment, participants were seated on a barstool chair with back support and their arms behind the chair.

Examination

All tests commenced with a 3-second period of stillness, during which participants memorized the specific testing region for each assessment with eyes open (ie, the upper body for trunk tests, the head for neck tests, the elbow for upper limb tests, and the knee for lower limb tests). Subsequently, they were instructed to perform designated motions with their eyes closed at their own pace. For the trunk flexion-extension repositioning test, participants were directed to bend forward to the maximum endpoint, extend back to an upright posture, and repeat this sequence. In the trunk lateral-flexion repositioning test, participants were asked to bend their trunks to the left and right, repeating the movements in the reverse directions. During the trunk axial-rotation repositioning test, participants were instructed to rotate their trunks to the left and right, then repeat the movements in reverse. Regarding the neck proprioception test, participants first memorized their neutral head position as a reference point. They then rotated their neck to the maximum extent on either the left or right side once before returning their head to the original reference position. The upper limb assessment included an elbow flexion repositioning test, in which participants slowly flexed their elbows until the examiner halted them, establishing this position as the reference. They then extended their elbows back to the natural position and repositioned to the reference with eyes closed. For the lower limb assessment, a knee extension repositioning test protocol similar to that used for the upper limb assessment was employed. After each specific movement, study subjects were required to revert to the reference position.

Participants

To identify potential participants for this study, consecutive screenings were conducted at the specialist outpatient clinics of the Duchess of Kent Children's Hospital in Hong Kong to include all scoliotic patients. Individuals aged 10–25 years diagnosed with AIS were invited to participate. Inclusion criteria were restricted to those who had not undergone any conservative treatments or had failed bracing interventions. These criteria were selected for several reasons. Since most participants were new patients to our clinic, it was essential to collect baseline data unaffected by previous scoliosis treatments. This included individuals unresponsive to bracing and awaiting surgical intervention, thereby enabling a comprehensive evaluation of treatment options. The focus on this age group was necessary because older adolescents and young adults tend to present with more severe curvatures, which typically require surgical intervention. This approach ensured inclusion of a wide range of curve severities, particularly those severe enough to require surgery. Participants were excluded if they had conditions such as spinal injury, fracture, tumor, neurological deficits, brain or spinal cord abnormalities, developmental delays, or psychological disorders. Suitability and demographic data were verified from the electronic medical records of the hospital.

Measurements

Proprioception

This study employed a 3-dimensional motion capture system consisting of 8 optical cameras (Nexus 2.16 and MX-T40, VICON, United Kingdom). The system demonstrated an error of less than 2 mm following calibration and operated at a sampling frequency of 100 Hz. A total of 34 retro-reflective markers, each measuring 9.5 mm in diameter, were strategically placed at specific anatomical landmarks. These included the seventh cervical vertebra, third, fifth, seventh, ninth, and 11th thoracic vertebrae, the first, second, third, fourth, and fifth lumbar vertebrae, and the sacrum. Additional markers were affixed bilaterally to the front head, anterior superior iliac spine, posterior superior iliac spine, clavicle, scapula, rib cage, upper limbs (wrist, elbow, and shoulder), and lower limbs (anterior superior iliac spine, knee, and ankle).

Radiology

The severity of spinal curvature and the pattern of the curve were assessed through radiographic evaluation (EOSedge, EOS imaging, France). Standing posteroanterior radiographs of the entire spine were captured. The Cobb method was utilized to quantify the magnitude of both structural and nonstructural curves.

Variables

The data-capturing software was the same as the data-processing system. Raw data were processed using a fourth-order low-pass Butterworth filter with a cut-off frequency of 6 Hz. Subsequently, the data were analyzed with MATLAB (R2021a, MathWorks, United Kingdom). The angles of the spine in the coronal and sagittal planes were measured at each thoracic and lumbar level, determined by the intersection of specific markers with their adjacent proximal and distal markers. Rotational elements of the trunk were measured relative to the lower level compared to the upper level. The neck angle was defined by the line connecting the midpoint of the forehead and the seventh cervical vertebra. Additionally, the angles of elbows and knees were derived from specific marker points. The proprioceptive outcome measure was obtained by calculating the absolute repositioning errors between the initial (reference) and final (reproduced) postures for each test. Demographic and radiographic data, including Cobb angle measurements of the major curve, proximal thoracic curve, main thoracic curve, and thoracolumbar or lumbar curve, as well as the convexity of the major curve, were extracted from electronic medical records. The severity of the curves was categorized as mild (10–19 degrees), moderate (20–39 degrees), or severe (over 40 degrees). The operational definition of a PD was established as a test result exceeding the normative value, defined as the mean plus 1 standard deviation of results from a healthy population. Curve progression was evaluated by comparing the baseline and follow-up Cobb angles. Surgical correction was quantified by subtracting the postoperative Cobb angle of the fusion segment from the preoperative major Cobb angle, then dividing the result by the preoperative major Cobb angle.

Analysis

Statistical analyses were performed using the statistical product and service solutions (SPSS) software (version 29.0, IBM), with a significance level set at 0.05. Descriptive statistics summarized the distribution of PD at baseline and follow-up. One-way analysis of variance (ANOVA) was conducted to compare differences in demographic and radiological parameters among treatment groups. The relationships between curve progression and PD were evaluated using linear and logistic regression models and chi-square tests. Adjustments for confounders identified in previous research [20], specifically age, height, and gender, were incorporated into the regression models. Two-way ANOVA was used to assess proprioceptive changes across treatment groups and PD status.

Results

A total of 166 participants were recruited for the baseline assessment between September 2020 and August 2023. Of these, 115 individuals (a dropout rate of 30.7%) participated in the follow-up data collection conducted from June 2022 to September 2024. The study framework is illustrated in Fig. 1. Additionally, a sensitivity analysis was conducted to compare baseline characteristics between participants who completed the study and those lost to follow-up. No statistically significant differences were observed in any baseline variables, including age, gender, height, weight, body mass index (BMI), and major Cobb angle. Following initial data collection, 56.5%, 14.8%, 16.5%, and 12.2% of study subjects received management for AIS through observation, physiotherapy, bracing, and surgery, respectively. Demographic information at the final follow-up included mean changes of 1.9±3.2 cm in height, 1.8±4.9 kg in weight, and 0.3±1.4 kg/m² in BMI. Participant characteristics are detailed in Table 1.

Fig. 1.

Flow diagram of subject recruitment.

Table 1.

Characteristics of participants.

Variable	Observation	Physiotherapy	Bracing	Surgery	Significance
Sample size (n)	65	17	19	14	/
Age at baseline (y)	15.6±2.6	15.0±3.6	12.4±0.8	16.3±5.2	<0.001*
Gender (proportion)	81.8% females	76.5% females	84.2% females	85.7% females	0.909
Height 1 (cm)	160.8±7.3	161.1±7.9	156.7±10.0	157.4±9.1	0.146
Height 2 (cm)	161.8±7.0	163.3±7.9	159.8±8.2	161.6±8.0	0.584
Weight 1 (kg)	50.1±10.7	45.9±5.2	44.7±11.9	49.5±11.4	0.158
Weight 2 (kg)	51.6±10.4	48.5±5.7	46.9±7.6	50.9±10.2	0.245
Body mass index 1 (kg/m2)	19.3±3.3	17.7±1.8	18.0±3.1	19.8±3.6	0.111
Body mass index 2 (kg/m2)	19.6±3.3	18.2±1.9	18.3±2.3	19.4±3.2	0.179
Cobb angle 1 (°)	28.4±11.5	30.0±12.3	33.5±8.2	59.6±12.8	<0.001*
Cobb angle 2 (°)	28.8±11.7	31.3±12.4	31.6±12.8	13.4±8.7	<0.001*
No. of attendance (session)	NA	2.9±1.7	NA	NA	/
Length of wear (month)	NA	NA	13.7±2.4	NA	/
Postoperative duration (month)	NA	NA	NA	14.4±2.7	/
Segment fused (level)	NA	NA	NA	9.6±2.4	/
Type of surgery (proportion)	NA	NA	NA	71.4% fusion 28.6% tethering	/

Note. 1 = baseline; 2 = follow-up.

NA, not applicable.

^⁎Significant.

The relationship between curve progression and PD was examined using linear regression models. No statistically significant effect of trunk PD on the progression of the major curve was identified across conservative treatment groups. Notably, the presence of trunk PD was significantly correlated with improved surgical correction of the curve (coefficient=0.200, p=.004; odds ratio=1.221, 95% CI=1.064–1.402). Details of the regression analyses are provided in Table 2. We also examined whether baseline PD was associated with the development of severe curves during follow-up using chi-square tests. In both the observation and bracing groups, no significant differences were observed between the subjects with and without trunk PD (p=.640; p=.510). Similarly, in the physiotherapy group, no significant association was found (p=.052). The distribution of severe curves across treatment groups is illustrated in Fig. 2.

Table 2.

Regression models on the progression or correction of the major curve.

Parameter	Model without adjustments			Model with adjustments
	Coefficient	Confidence intervals	Significance	Coefficient	Confidence intervals	Significance
Observation
Trunk PD	−0.870	−2.875 to 1.135	0.395	−1.008	−3.036 to 1.020	0.365
Age	/	/	/	−0.001	−0.372 to 0.370	0.999
Height	/	/	/	−3.874	−18.062 to 10.313	0.593
Gender	/	/	/	−0.675	−3.186 to 1.841	0.600
Physiotherapy
Trunk PD	−1.564	−6.230 to 3.101	0.511	−1.764	−6.531 to 3.004	0.468
Age	/	/	/	−0.047	−0.610 to 0.516	0.870
Height	/	/	/	13.045	−14.346 to 40.436	0.351
Gender	/	/	/	−1.251	−6.492 to 3.989	0.640
Bracing
Trunk PD	−0.007	−5.804 to 5.817	0.998	0.820	−3.666 to 5.306	0.720
Age	/	/	/	−1.692	−4.484 to 1.101	0.235
Height	/	/	/	22.885	−4.442 to 50.213	0.101
Gender	/	/	/	7.289	0.071 to 14.507	0.048
Surgery
Trunk PD	0.165	0.023 to 0.307	0.023*	0.200	0.062 to 0.338	0.004*
Age	/	/	/	−0.005	−0.023 to 0.012	0.548
Height	/	/	/	1.091	0.262 to 1.921	0.010
Gender	/	/	/	0.030	−0.151 to 0.211	0.747

Note. PD, proprioceptive deficits.

^⁎Significant.

Fig. 2.

Proportion of severe curves.

Further, we sought to determine whether the treatments improved proprioception (Table 3). The interaction effect between treatment and PD was not statistically significant for trunk proprioception. However, significant main effects of PD were observed (trunk flexion-extension test: p<.001; trunk lateral-flexion test: p<.001; trunk axial-rotation test: p=.030). The group comparisons for trunk proprioception are shown in Fig. 3. Additionally, no interaction effects between scoliosis treatment and PD were detected for proprioception of the neck, elbow, and knee. A significant main effect of PD was noted in the neck rotation proprioception test for the left turn (p<.001), with group comparisons presented in Fig. 4. Likewise, the right elbow flexion test demonstrated a significant main effect of PD (p=.001), as shown in Fig. 5. Similar main effects were observed in the knee extension tests, in which PD had a significant effect on both the left (p<.001) and right (p<.001) knees. Comparisons of knee proprioception across groups are depicted in Fig. 6.

Table 3.

Differences in proprioception test results from baseline to follow-up assessments.

Proprioception tests	Interaction effect		Main effect
	Significance	Partial eta-squared	Significance	Partial eta-squared
Trunk flexion-extension repositioning test	0.522	0.021	<0.001*	0.100
Trunk lateral-flexion repositioning test	0.272	0.035	<0.001*	0.105
Trunk axial-rotation repositioning test	0.385	0.028	0.030*	0.043
Neck rotation repositioning test in left turn	0.912	0.005	<0.001*	0.232
Neck rotation repositioning test in right turn	0.203	0.042	0.711	0.001
Elbow flexion repositioning test on left side	0.130	0.051	0.255	0.012
Elbow flexion repositioning test on right side	0.416	0.026	0.001*	0.090
Knee extension repositioning test on left side	0.567	0.018	<0.001*	0.147
Knee extension repositioning test on right side	0.081	0.060	<0.001*	0.135

Note.

^⁎Significant.

Fig. 3.

Treatment effects on trunk proprioception.

Fig. 4.

Treatment effects on neck proprioception.

Fig. 5.

Treatment effects on elbow proprioception.

Fig. 6.

Treatment effects on knee proprioception.

Discussion

The present study elucidated a sophisticated association between AIS and PD. Although our findings did not reveal a statistically significant correlation between trunk PD and the degree of curve progression under conservative treatment, an association between trunk PD and curve correction was identified in subjects who underwent surgery. Additionally, we observed that subjects with PD across all body regions showed substantial improvements in proprioception over time, irrespective of treatment modality, compared to those without PD. While numerous questions remain regarding the relationship between scoliosis and proprioception, this study was a pioneering effort to advance research in this area.

Unfortunately, conservative treatments for scoliosis have not been shown to affect proprioception. Since these interventions aim to correct spinal curvature and prevent further progression of deformity, they primarily do not address the underlying PD. For instance, bracing can restrict spinal movement and provide external support, but it does not necessarily improve proprioception [25]. Evidence suggests that patients using a brace may experience no effect or increased displacements of their center of pressure [26,27], indicating a possible adverse effect on proprioception [28]. This may also be attributable to significant mutations in proprioceptive genes [21], which could render them unresponsive to external stimuli that affect proprioception.

Our findings indicated that trunk PD is associated with enhanced surgical correction of the spinal curve. Given that spinal alignment depends on precise afferent input from paraspinal muscle spindles and associated mechano-transduction pathways [29], diminished proprioceptive accuracy may lead to persistent asymmetrical muscle activation, impaired postural correction, and progressive coronal and rotational malalignment [30]. From this perspective, surgical intervention may improve outcomes for affected patients, as corrective instrumentation and spinal fusion realign the vertebral column and mechanically stabilize segments that previously relied on a defective feedback-control system. This stabilization likely reduces the influence of inaccurate spinal position sense on deforming forces and postural instability. Although this interpretation remains hypothesis-generating rather than definitive, it offers a coherent rationale for why patients with a more pronounced proprioceptive component of their deformity might derive greater benefit from surgical correction compared to those whose deformity is primarily due to fixed structural stiffness.

Longitudinal results showed alterations in proprioceptive abilities, particularly improvements in participants with PD. Initially, the potential influence of age was considered, and ad hoc analyses were conducted to examine it. Two-way ANOVA analyses assessing the interaction effects of treatment and PD status on age differences revealed that younger individuals were exclusively present among surgically treated candidates with trunk PD (p<.001), neck PD (p<.001), elbow PD (p=.004), and knee PD (p=.001). Consequently, it is proposed that these improvements reflect an age- or growth-related effect. We hypothesized that this phenomenon could be attributable to disruptions within the proprioceptive system. Blecher et al. [17] reported that both Golgi tendon organs and muscle spindles are involved in maintaining spinal alignment, with muscle spindles generally associated with a less severe phenotype.[] It is plausible that some study subjects developed PD due to impaired muscle spindles, resulting in a milder form of scoliosis that is susceptible to environmental influences.

Furthermore, we observed that changes in peripheral proprioception affected only 1 side. This may be related to the convexity of the major curve. Therefore, additional ad-hoc analyses were performed. The chi-square test results indicated that only individuals with elbow PD exhibited a higher proportion of right-sided major curves (With PD=68.1% vs. Without PD=49.5%, p=.033). Given the convexity of the major curve, it may cause an imbalance in muscle activation on 1 side of the body [31], which could contribute to proprioceptive adaptation in the limbs. Additionally, factors contributing to curvature may alter how the nervous system processes sensory information, for example, through abnormal brain oscillatory activity and cerebrospinal fluid flow abnormalities [32,33]. These alterations could differently affect proprioceptive abilities. Nonetheless, further research is necessary to elucidate the underlying mechanisms.

Several limitations were identified in this study. The loss of participants to follow-up may have introduced bias and potentially limited the generalizability of the findings. Although proprioception was the primary outcome measure, we did not evaluate other critical clinical outcomes, such as spinal pain and functional limitations, thereby limiting the comprehensiveness of the results. Additionally, potential confounding variables, such as coexisting health conditions, were not considered, which may limit the ability to attribute observed effects solely to PD or the specific treatments administered. Finally, the study design involved only a 1-year follow-up, which may not adequately reflect the long-term progression of spinal curvature and treatment effects.

Conclusions

In this prospective cohort study of subjects with AIS, trunk PD was not associated with increased curve progression or a greater likelihood of developing severe curves among those receiving conservative management. However, it was significantly correlated with improved radiographic correction in patients who underwent surgical intervention. Participants with baseline proprioceptive deficits also exhibited greater improvements in proprioceptive performance over time across various body regions, regardless of treatment modality, suggesting that these changes may reflect maturation or adaptive processes. Overall, these findings indicated that impaired spinal proprioception may serve as a marker for a clinically distinct subgroup with differential responsiveness to surgical correction.

Funding

KC received the National Clinical Key Specialty Construction Program of China (SZXK2020084) and the Sanming Project of Medicine in Shenzhen (SZSM202211004).

Ethics approval

Approval for the study was granted by the institutional review board of the University of Hong Kong and Hospital Authority Hong Kong West Cluster (reference: UW 20-525), and the study protocol was registered on ClinicalTrials.gov (identifier: NCT04682379).

Declaration of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.