Optimal age selection for posterior hemivertebra resection and short fusion of a solitary simple lower thoracic or lumbar hemivertebra

Abstract

Hemivertebra (HV) is a leading cause of congenital scoliosis; however, the optimal timing for surgical intervention remains uncertain. This study aimed to compare surgical outcomes in children under 10 years old with scoliosis caused by a solitary simple lower thoracic or lumbar HV (T8–L5). From January 2015 to January 2024, we retrospectively analyzed 49 consecutive congenital scoliosis patients treated with posterior hemivertebra resection, fusion, and pedicle screw fixation. Of these, 35 patients met all inclusion criteria and were included in the final analysis. A minimum follow-up period of 12 months was maintained. We used ROC curve analysis to determine the age at which the rate of unplanned reoperation decreased, identifying a cutoff age of 5.07 years. Based on this criterion, we divided the pediatric population into two groups: the younger age group (≤ 5 years) and the older age group (> 5 years). No statistically significant differences were observed between the two groups in terms of correction rates in the coronal and sagittal planes. However, the unplanned reoperation rate was significantly higher in the younger age group (P = 0.016). Our findings suggest that delaying surgery until between five and ten years of age, combined with close follow-up, results in satisfactory outcomes with a lower rate of unplanned reoperation.

Introduction

Recent studies have estimated that congenital scoliosis (CS) has a prevalence of approximately 1 in 1,000 live births [1, 2]. Hemivertebra (HV) is one of the most common causes of this condition. CS primarily results from untreated fully segmented and semi-segmented non-incarcerated HV, which often leads to progressive spinal deformity as the spine grows [3, 4]. According to McMaster and David, the rate of progression and severity of spinal deformity are key factors influencing curve development in HV cases [5]. The severity of spinal deformities caused by HV depends on various factors, including the type, location, number, and interrelation of the affected vertebrae. Notably, a simple solitary HV during the early developmental stages is highly likely to result in a significant deformity [4]. The lower thoracic and thoracolumbar vertebrae are particularly prone to contributing to spinal curvature deterioration, highlighting the importance of timely surgical intervention [6].

Early surgical intervention in young children offers the advantage of preventing severe local deformities and secondary structural curves, thereby facilitating normal spinal growth in unaffected regions [7]. Timely execution of surgical procedures can effectively mitigate the progression of deformity while also alleviating psychological distress. Despite ongoing research, a consensus has yet to be reached regarding the optimal surgical timeframe for managing CS. Some clinicians advocate for early intervention as soon as the first year of life; however, in cases where patients with CS malformation reach an age beyond seven years, it remains plausible that they may still achieve favorable clinical outcomes following surgical intervention [8, 9].

Studies from various international healthcare settings including Europe, East Asia, and North America have shown differing preferences and outcomes, which highlights the influence of surgical culture, access to pediatric spine care, and long-term follow-up protocols [10].

Although the optimal timing for surgical intervention remains uncertain, literature reports indicate that favorable outcomes have been achieved in children under the age of 10 years with CS [11–13], as well as in those younger than 5 years [14, 15]. Recent systematic reviews and meta-analyses have further supported this notion, providing evidence-based insight into the benefits and limitations of early versus delayed surgical correction [16]. Although previous surgical practices have been widely endorsed by most surgeons, the ideal age for pediatric surgery remains a subject of debate. Although previous surgical practices have been widely endorsed by most surgeons, the ideal age for pediatric surgery remains a subject of debate. Through comparative analysis, it has been postulated that younger age and lower Risser’s grade, in combination with greater scoliotic curvature, may serve as risk factors for the loss of corrective treatment outcomes [17].

There is a paucity of literature on surgical outcomes based on the age at the time of surgery for children under 10 years old with CS caused by a specific type of HV. This study aimed to evaluate the correction effect of posterior-approach hemivertebrectomy and short-segment fusion in the management of solitary simple lower thoracic or lumbar HV (T8–L5) in pediatric patients across different age groups.

Methods

Study participants

This retrospective study was conducted from January 2015 to January 2024, A total of 75 patients were initially reviewed. After excluding 28 patients with follow-up periods less than 12 months and 12 with non-solitary or complex hemivertebrae, 35 patients were included in the final analysis. The surgical approach involved posterior HV resection, followed by fusion with pedicle screw fixation. In our center, posterior hemivertebra resection is the preferred technique for treating progressive congenital scoliosis caused by solitary HV, especially when segmental deformity or significant curve progression is present. Convex hemiepiphysiodesis was not employed in this cohort due to either limited correction potential, lack of adequate remaining growth, or the clinical judgment that definitive resection would provide more reliable long-term outcomes. The age at which the rate of unplanned reoperation decreased was determined using ROC curve analysis, identifying a cutoff age of 5.07 years (Youden’s index = 0.531, sensitivity: 100%, specificity: 53%) (Fig. 1). ROC curve analysis is a widely used method for establishing clinical cutoff points in surgical and diagnostic research, particularly when dichotomizing continuous predictors [18]. From an initial cohort of 75 consecutively treated children with congenital scoliosis, we applied strict inclusion and exclusion criteria to ensure homogeneity of the study population. Patients were excluded if they had follow-up less than 12 months (n = 28) or if they did not have a solitary simple hemivertebra (n = 12), resulting in a final analytic sample of 35 patients. The patient selection process is summarized in (Fig. 2). Based on this criterion, the pediatric population was divided into two groups: the younger age group (≤ 5 years) and the older age group (> 5 years). Each group underwent a minimum follow-up period of 12 months (Fig. 2). Unplanned reoperation was chosen as the dependent variable in the ROC curve analysis because it represents a clinically important and binary outcome. This includes events such as implant migration, infection, or progressive curve deformity requiring revision. The age threshold of 5.07 years was statistically derived using Youden’s index (sensitivity = 100%, specificity = 53%). While the statistical model identified an optimal cutoff, this threshold should be interpreted as exploratory and requires clinical validation before being used for decision-making. It reflects a failure of primary surgical treatment and encompasses complications such as implant migration, infection, or adding-on, all of which may necessitate revision surgery. ROC analysis was applied to determine the optimal age threshold at which the likelihood of these adverse events decreases.

Fig. 1.

ROC curve of age. ROC curve was used to identify the optimal age cutoff (5.07 years) for predicting unplanned reoperation

Fig. 2.

Flowchart depicting the process of participant inclusion and grouping

The study’s inclusion criteria comprised four main factors: (1) age under 10 years at the time of surgery, (2) congenital deformity caused by a solitary simple thoracolumbar HV, (3) posterior HV resection utilizing transpedicular instrumentation, and (4) a follow-up period of at least 12 months with comprehensive radiographic and clinical data. The exclusion criteria were as follows: (1) participants over 10 years old, (2) a history of prior spinal surgery, (3) presence of a complex congenital spinal deformity (including multiple HV), and (4) a diagnosis of syndromic scoliosis.

Radiographic data

Radiographic data were collected at three key time points: preoperatively, on postoperative day 7, and at the last follow-up. Data collection was conducted by a single trained professional, with random validation performed by a fellowship-trained pediatric spine surgeon. The measurements were based on the radiographic parameters established by Scheer et al. [19], which are widely validated in pediatric spinal deformity research. Standardized definitions and criteria for outcome measures such as curve correction (Cobb angle), ‘adding-on,’ and proximal junctional kyphosis (PJK) have been adopted from previous multicenter studies and guideline-based research [20]. These validated outcome measures are essential for ensuring reproducibility, comparability, and clinical relevance in pediatric spine surgery studies. A comprehensive preoperative assessment was performed, including standing long cassette anteroposterior (AP) and lateral (L) X-rays. Additionally, computed tomography (CT) scans with reconstruction were used to evaluate the shape and positioning of the HV, as well as the surrounding pedicle anatomy. Magnetic resonance imaging (MRI) was performed to identify any potential spinal cord abnormalities. All curve parameters were measured using the Cobb method, as described by Ruf and Harms [21].

The present study utilized a formula to calculate the percentage change in spinal curvature from preoperative to postoperative evaluations. Specifically, the following formula is used: [(preoperative degree of Cobb angle - postoperative degree of Cobb angle)/preoperative degree of Cobb angle] *100%.

To assess shoulder balance, radiographic shoulder height (RSH) was measured. RSH is defined as a discrepancy in soft tissue shadows located directly above the bilateral acromioclavicular joints. This measurement was used as the evaluation parameter.

In the discipline of orthopedic pathology, the phrase “curve progression” refers to the development of a fresh curvature, demonstrating an elevation in the Cobb angle by a magnitude exceeding 20 degrees, and with the apex vertebra situated at least two spinal levels away from the lower instrumented vertebra. The present study utilized evaluation criteria that are in line with the definition provided. The most commonly employed diagnostic criteria for “adding-on” predominantly encompass either a gradual increase in the number of vertebrae included in the distal curve, or an elevation of the first vertebra deviation below the instrumentation from the center sacral vertical line (CSVL) by more than 5 mm, or an increase in the angulation of the first disc below the instrumentation by more than 5 degrees [22]. The condition known as proximal junctional kyphosis (PJK) is defined by the presence of a proximal junctional sagittal Cobb angle that exceeds either 10 degrees or the preoperative measurement by 10 degrees or more. This anatomical anomaly is a well-recognized complication of spinal surgery and is commonly utilized as an important clinical outcome measure [23].

The assessment of sagittal plane balance involves the utilization of segmental kyphosis measurement. In this context, thoracic kyphosis is denoted by a positive sign (+), whereas lumbar lordosis is represented by a negative sign (-).

Surgery technology

In this investigation, the surgical approach involved hemivertebra (HV) resection via a posterior single-stage procedure with fusion of adjacent vertebrae, utilizing transpedicular instrumentation [21]. The surgery was performed with the patient in the prone position, and a midline incision was made to gain access to the spine. Pedicle screws were inserted into the normal upper and lower vertebrae. For thoracic and lumbar lesions, following HV resection, short-segment fixation (within four segments) was performed. A pre-contoured rod was attached to the screws on the convex side, and the posterior HV elements along with adjacent levels were excised to expose the pedicle and nerve roots above and below. To protect the dural sac and spinal cord, resection of the posterior elements was performed in the final stage. The anterior HV was exposed using blunt dissection, and a wedge osteotomy was executed from the convex to the concave side. The anterior vertebral body was hollowed using a combination of rongeurs, osteotomes, curettes, and abrasive drilling. Additionally, the upper and lower discs, including the cartilage endplates, were removed from the bleeding bone. Intersomatic fusion was achieved following removal of the hemivertebra and the adjacent intervertebral discs. The cartilaginous endplates of the upper and lower vertebrae were carefully excised to expose bleeding subchondral bone. The anterior column defect created by the resection was packed with autologous bone graft harvested from the removed hemivertebra. Compression across the interbody space was applied via instrumentation to enhance fusion potential and mechanical stability. No interbody cages were used due to the young age of the patients and the size of the resected space. Gradual compression was applied while keeping the concave rod unlocked until the gap was completely closed. If a high compressive force was required to correct the deformity, particularly in cases of pronounced kyphosis, additional segments were included in the instrumentation to prevent excessive stress on the pedicles and reduce the risk of pedicle fractures. Special precautions were taken to ensure that the exiting nerve roots and dura were not impinged during the procedure. Finally, decortication of the posterior elements was performed before utilizing the autologous bone from the resected HV for posterolateral fusion. Upon achieving sufficient recovery to ambulate, the patient was provided with a custom-fabricated standard brace, which was worn for at least three months and discontinued only in the absence of radiographic evidence of pseudarthrosis. In addition, the specific timing of the spinal fusion was documented.

For patients experiencing postoperative complications, such as “adding-on” or implant migration, leading to loss of correction, the initial approach involved brace fixation and observation. If conservative management proved ineffective and significant curve progression was observed, surgical intervention was performed.

Complication Definitions

“Unplanned reoperation” was defined as any return to the operating room required due to a complication directly related to the index surgery. This included implant migration, pseudarthrosis, wound infection, or progressive deformity requiring revision.

“Implant migration” was defined as radiographic evidence of displacement or loosening of hardware, accompanied by clinical symptoms or progressive deformity.

“Adding-on” was defined as extension of the curve distally, evidenced by increased deviation of vertebrae beyond the lowest instrumented vertebra, or increased angulation below the construct.

“Pseudoarthrosis” was defined as radiographic absence of solid fusion at the operated segment, with or without clinical symptoms.

All complications were confirmed through a combination of radiographic review and clinical documentation in follow-up records.

Statistical analysis strategy

In this investigation, appropriate statistical methods were utilized to analyze both continuous and categorical variable distributions. Specifically, the independent sample t-test was applied for continuous variables, whereas the Pearson chi-squared test was used for categorical variables. All statistical analyses in this study were two-sided, with statistical significance defined as a P-value of 0.05 or less. In addition to p-values, effect sizes (Cohen’s d for continuous variables and absolute risk differences for categorical outcomes) and 95% confidence intervals were reported where relevant, to enhance interpretability of group differences. Data analysis was conducted using SPSS version 22.0 (Statistical Product and Service Solutions, NY, USA).

Radiographic measurements

Preoperative, one-week postoperative, and final follow-up whole-spine anteroposterior (AP) and lateral radiographs were analyzed to assess deformity correction and spinal balance. Additionally, computed tomography (CT) and magnetic resonance imaging (MRI) were utilized to evaluate the condition of the spine and spinal cord. All included patients had complete clinical and radiographic datasets, with no missing data. Therefore, no imputation or sensitivity analyses for missingness were required.

Results

The children’s ages ranged from 2.5 to 10 years, with an average age of 5.3 years. A comparison was conducted between early fusion (surgery at ≤ 5 years) and late fusion (surgery between 5 and 10 years). Patient characteristics are presented in Table 1.

Table 1.

Baseline characteristics of the patients

Items	≤ 5 years old (N = 18)	5–10 years old (N = 17)	P	95% CI
				Low	Up
Weight/kg	15.50 (2.78)	24.29 (6.81)	< 0.001	−12.49	−5.10
Surgical time/min	164.44 (61.66)	221.06 (68.62)	0.02	−101.43	−11.80
Blood loss/ml	187.72 (111.16)	308.24 (133.89)	0.01	−204.95	−36.08
Follow-up time/m	23.28 (16.19)	21.76 (9.69)	0.74	−7.74	10.76
Preoperative segmental scoliosis/°	27.52 (10.85)	27.54 (5.84)	1.00	−6.02	5.99
Preoperative main curve/°	32.42 (6.57)	28.97 (7.29)	0.15	−1.32	8.22
Preoperative compensatory cranial curve/°	10.84 (7.53)	12.81 (7.69)	0.45	−7.20	3.27
Preoperative compensatory caudal curve/°	11.33 (6.32)	10.89 (3.76)	0.84	−4.11	4.99
Preoperative segmental kyphosis/°	1.37 (14.03)	−3.50 (9.80)	0.27	−3.98	13.72
Preoperative RSH	0.82 (0.36)	1.11 (0.44)	0.04	−0.57	−0.01

RSH Relative spinal height

The study findings suggest that the late fusion group exhibited greater weight, higher RSH, longer surgical time, and increased intraoperative blood loss compared to the early fusion group (P < 0.05). However, no statistically significant differences were observed between the early and late fusion groups concerning preoperative radiographic data of scoliosis or segmental kyphosis (P > 0.05). Table 2 shows comparison of radiological data between the two groups of patients.

Table 2.

Comparison of radiological data between the two groups of patients

Items		≤ 5 years old (N = 18)	5–10 years old (N = 17)	P	95% CI
					Low	Up
Segmental Scoliosis	Postoperative/°	7.67 (5.65)	6.41 (4.73)	0.48	−2.33	4.86
	Follow-up/°	9.56 (6.67)	9.24 (6.96)	0.89	−4.36	5.01
	Correction rate	0.65 (0.20)	0.70 (0.21)	0.44	−0.20	0.09
	Loss of correction rate	−2.30 (7.96)	−1.96 (3.38)	0.87	−4.59	3.91
Main Curve	Postoperative/°	7.86 (6.03)	8.09 (4.64)	0.90	−3.95	3.49
	Follow-up/°	10.36 (7.64)	8.39 (3.94)	0.34	−2.22	6.17
	Correction rate/%	0.69 (0.17)	0.70 (0.12)	0.91	−0.11	0.10
	Loss of correction rate/%	0.03 (0.66)	−0.08 (0.80)	0.64	−0.39	0.63
Compensatory Cranial Curve	Postoperative/°	5.38 (3.11)	6.45 (3.53)	0.35	−3.36	1.22
	Follow-up/°	6.23 (6.78)	7.59 (5.06)	0.51	−5.49	2.78
	Correction rate/%	−0.06 (2.10)	0.20 (0.71)	0.62	−1.37	0.83
	Loss of correction rate/%	−2.89 (11.30)	−0.15 (0.90)	0.33	−8.34	2.86
Compensatory Caudal Curve	Postoperative/°	4.32 (3.18)	4.55 (2.41)	0.84	−2.64	2.18
	Follow-up/°	7.09 (7.80)	4.68 (2.57)	0.36	−2.87	7.69
	Correction rate/%	0.22 (0.85)	0.44 (0.53)	0.47	−0.85	0.40
	Loss of correction rate/%	−1.07 (4.23)	−0.15 (0.75)	0.51	−3.72	1.88
Segmental Kyphosis	Postoperative/°	2.49 (11.78)	−3.76 (15.08)	0.19	−3.21	15.73
	Follow-up/°	1.89 (17.98)	−3.99 (14.75)	0.32	−5.87	17.62
	Correction rate/%	0.00 (1.89)	−1.20 (2.14)	0.11	−0.27	2.67
	Loss of correction rate/%	−0.43 (1.37)	−0.25 (1.07)	0.67	−1.02	0.67
Follow-up RSH/cm	Follow-up/°	0.83 (0.25)	0.98 (0.43)	0.20	−0.39	0.09
	Correction rate/%	−0.13 (0.56)	0.01 (0.60)	0.48	−0.54	0.26
Trunk Shift/cm		0.96 (0.72)	1.24 (0.71)	0.25	−0.78	0.21
Reoperation rate		0.22 (0.42)	0.00	0.02	0.01	0.43

RSH Relative spinal height

Among the four cases that required reoperation, one was due to postoperative infection, one resulted from the adding-on phenomenon, and the remaining two cases were attributed to implant migration (Table 3). These results should be interpreted with caution due to the small overall sample size and the reduced statistical power following subgroup stratification.

Table 3.

Postoperative complications

Postoperative complications	≤ 5 years old (%)	5–10 years old (%)
Implant migration	2 (11%)	0 (0%)
Wound infection	1 (5.5%)	0 (0%)
Major vascular injury	0 (0%)	0 (0%)
Neurologic deficit	0 (0%)	0 (0%)
Curve progression	0 (0%)	1 (0%)
Adding-on phenomenon	2 (5.5%)	1 (0%)
Crankshaft phenomenon	0 (0%)	0 (0%)
Unplanned reoperation	4 (22%)	0 (0%)

Discussion

There were statistically significant differences in preoperative weight among the pediatric patients (P < 0.001), which was expected since younger children generally weigh less. The surgical duration for the younger age group was significantly shorter compared to the older age group (P = 0.015). Despite their smaller anatomical structures, young children did not pose significant challenges for internal fixation placement in clinical practice. The younger age group had fewer screws inserted on average (5 vs. 5.8), and HV removal was easier because of minimal progression. Additionally, scoliosis correction was facilitated by the greater flexibility of the spine in younger children. The younger age group also experienced less intraoperative blood loss (P = 0.007), which may be attributed to their lower total blood volume due to their smaller body size. However, there was no statistically significant difference in the ratio of surgical blood loss to total blood volume between the two groups (8.58% vs. 8.99%). In accordance with previous studies, the blood loss rate was calculated by dividing blood loss (in mL) by total blood volume (in mL), with an average blood loss rate of 14.1 mL/kg body weight [24]. There was also a statistically significant difference in the preoperative RSH between the younger and older age groups (P = 0.042), likely due to the height disparity between the two groups.

Our results indicate that postoperative outcomes did not differ significantly between the younger (≤ 5 years old) and older (5–10 years old) age groups. The correction rate of the main curve was consistent with the findings of Ruf and Harms [24], reaching 75.8% in the younger age group and 72.1% in the older age group. The correction rate for segmental scoliosis was 72.1% in the younger age group and 76.7% in the older age group, aligning with the findings of Zhang et al. [25].

Despite previous research demonstrating favorable correction rates and minimal impact on spinal and spinal cord growth in young children undergoing HV resection surgery [26], our study revealed a notable increase in unplanned reoperation rates among children in the younger age group. It is worth noting that the average curve magnitude in our cohort was relatively modest. In select cases—particularly those with milder curves and younger ageconvex hemiepiphysiodesis may have served as a suitable alternative to resection. However, in our institution, posterior hemivertebra resection has been the preferred approach when early curve progression and segmental anomalies are evident, as it allows for immediate deformity correction and definitive treatment. We acknowledge that future comparative studies incorporating both resection and growth-guiding procedures would help clarify the optimal approach based on patient age, curve magnitude, and skeletal maturity. Additionally, patients in the younger age group exhibited a significant increase in postoperative complications and unplanned reoperation rates (P = 0.016). For patients requiring unplanned reoperation, we used a surgical approach involving vertebral extension, either upward or downward, to achieve subsequent fixation (Fig. 3).

Fig. 3.

Example of a patient requiring unplanned reoperation due to postoperative “adding-on” phenomenon. a–b Preoperative radiographs showing main curvature and location of hemivertebra; (c–d) Immediate postoperative alignment after HV resection and short fusion; (e–f) Radiographs at 16 months of operation showing curve progression and adding-on below instrumentation; (g–h) Post-revision radiographs after extended fusion; (i–j) Follow-up at 2 years after reoperation showing stable alignment. Surgical hardware (pedicle screws and rods) and hemivertebra have been labeled. All images are shown in standard AP and lateral views

Previous studies have reported similar findings, indicating a significant increase in the reoperation rate among younger children following surgery [27]. Specifically, the complication risks associated with vertebrectomy in younger children have been documented to reach up to 44% [28]. Although a post hoc power analysis revealed limited statistical power to detect significant differences possibly due to the infrequent occurrence of these outcomes patients undergoing early fusion exhibited a higher risk of intraoperative and long-term complications compared to those undergoing late fusion. The ROC-derived cutoff age of 5.07 years demonstrated high sensitivity but only moderate specificity, limiting its predictive precision. However, this threshold aligns with findings from recent systematic reviews and meta-analyses, which similarly caution against early surgical intervention (before age 5) due to increased risks of complications and hardware failure in younger children [29]. These studies support the rationale that delaying surgery until children are biomechanically more stable improves outcomes. Thus, our ROC finding not only reflects existing clinical intuition but is also reinforced by current literature and should be further validated through large-scale, multicenter prospective research.

In the younger age group (n = 18), implant migration was observed in two cases, while one case presented with incision infection. Additionally, two cases demonstrated the adding-on phenomenon, with one requiring reoperation. The reoperation rate in the younger age group was 22%. In the older age group (n = 17), one case exhibited curve progression and one case displayed the adding-on phenomenon; however, reoperation was not required in either case. Subsequent assessments did not reveal any instances of neurological or vascular impairment resulting from the surgical procedure or the use of implants. Ruf and Harms previously reported a group of patients with an average age of 3 years and 4 months, in which the incidence of complications following HV resection was 28.5%. In subsequent studies, a cohort of 41 cases under the age of 6 years, with an average follow-up period of 6.2 years, exhibited a complication rate of 29.3% [7].

In all reoperations, except for those involving infection, pseudoarthrosis formation was observed at the fused segments, suggesting that young children may require prolonged brace protection postoperatively. The loosening or breakage of internal fixation may be related to the mismatch between bone quality and screw stiffness in younger children. Therefore, meticulous surgical planning is essential to minimize reoperations and complications during early fusion procedures. These findings reflect longstanding biomechanical concerns associated with performing hemivertebra resection in very young children. Previous surgical experience has shown that immature pedicles may lack the structural strength to reliably anchor pedicle screws, increasing the risk of hardware failure and loss of correction. As a result, several adaptive strategies have historically been used, including delaying surgery until after age 4–5, increasing the number of fused segments (although this affects spinal length), applying postoperative bracing, or using advanced constructs such as the 3-rod technique to distribute corrective forces through the laminae. The elevated reoperation rate observed in our younger cohort aligns with this rationale, reinforcing the importance of careful patient selection and construct planning in early surgical interventions. Several factors can contribute to complications in congenital scoliosis (CS), including limitations in surgical techniques, insufficient precision, inadequate posterior fusion, incomplete HV removal, and inappropriate fusion level selection. Therefore, it is crucial to implement appropriate strategies to address these challenges. Accurate identification of the HV location and careful preparation of the pedicle tunnel are paramount for successful screw insertion. Preoperative spinal CT scans aligned with the pedicle axis supplemented by 3D reconstruction are recommended to assess pedicle morphology and facilitate optimal instrumentation selection.

This study has several limitations inherent to its retrospective design. The absence of randomization and reliance on historical data introduce potential selection and information biases, despite our use of standardized surgical techniques and clearly defined inclusion criteria. While all surgeries were performed by a consistent pediatric spine team, we could not account for subtle changes in surgical experience or perioperative management that may have evolved over the 9-year study period. Additionally, the strict inclusion and exclusion criteria particularly the focus on solitary, non-syndromic hemivertebra may introduce selection bias and limit the generalizability of our findings to the broader congenital scoliosis population.

The sample size was modest, with only 35 patients meeting the inclusion criteria after strict screening. Further subdivision into age groups limited the statistical power to detect small but potentially meaningful differences in surgical outcomes. This limitation is particularly important when interpreting subgroup comparisons between the ≤ 5 and > 5 year-old groups, where subtle differences may not have reached statistical significance due to insufficient power. Additionally, although reoperation rates differed significantly between groups, the lack of a priori power calculation means these findings should be interpreted cautiously. The use of ROC curve analysis to determine an age cutoff for risk stratification provided a statistically derived threshold (5.07 years), but the moderate specificity of this value (53%) limits its clinical applicability. This threshold should be considered exploratory and requires external validation.

The follow-up period, while sufficient to capture early complications and unplanned reoperations, was relatively short for a pediatric population. A minimum of 12 months was required, but many patients did not reach skeletal maturity during the observation window. Consequently, the study could not assess long-term outcomes such as spinal growth, hardware longevity, or curve progression through adolescence. Although some patients had follow-up exceeding two years, the variability in duration precluded meaningful long-term subgroup analysis. This limitation is particularly relevant in the context of congenital scoliosis, as many complications such as the adding-on phenomenon, junctional deformities, or secondary curve progression tend to emerge during the adolescent growth spurt. Since our study cohort included mostly pre-adolescent patients and follow-up was limited in duration, these later-occurring events may have been underrepresented. Future longitudinal studies following patients into adolescence and skeletal maturity are essential to fully assess the durability and evolution of surgical outcomes.

Another limitation is the study’s focus on radiographic correction and unplanned reoperations, without incorporating validated patient-reported outcome measures (PROMs) such as the EOSQ-24 or PedsQL. This lack of functional and quality-of-life assessment limits the ability to evaluate the broader clinical impact of surgery, such as pain relief, activity level, or psychosocial outcomes, which are especially important in pediatric patients. Future studies should incorporate standardized PROMs to ensure a more comprehensive evaluation of patient well-being. These metrics are essential to understanding the functional and quality-of-life impact of surgical intervention, especially in children. Unfortunately, such data were not consistently available due to the retrospective design and the age of the patients involved.

Generalizability is also limited. The study was conducted at a single institution with a homogenous patient population, all from one country and cultural context. This enhances procedural consistency but restricts the applicability of our findings to more diverse populations, healthcare settings, and ethnic groups. The lack of a non-surgical control group further limits the ability to compare surgical outcomes with the natural progression of untreated solitary hemivertebra or with alternative interventions such as growth modulation or convex hemiepiphysiodesis.

Another limitation is the absence of formal interobserver reliability analysis for radiographic measurements. Although data collection was performed by a single trained evaluator and randomly validated by a pediatric spine surgeon, we did not calculate interobserver agreement metrics such as the intraclass correlation coefficient (ICC). This may affect the reproducibility of radiographic outcomes, especially in a small sample with modest measurement differences. Future studies should include formal reliability testing to improve consistency and transparency.

Finally, we acknowledge that unmeasured anatomical or physiological confounders such as bone mineral density, pedicle morphology, or spinal flexibility could have influenced complication rates and reoperation risk, particularly across different age groups. While preoperative curve magnitude and kyphosis were similar between groups, these other variables were not quantifiable in our dataset and should be addressed in future prospective, multivariable studies. Additionally, due to the modest sample size, we did not perform multivariable statistical adjustments to control for confounders such as curve severity, comorbidities, or fusion level. Future studies with larger cohorts should include such adjustments to strengthen causal inference. Furthermore, we acknowledge that the strict selection criteria used to ensure a homogeneous study population (e.g., solitary HV, exclusion of syndromic cases) may reduce the generalizability of our findings to the broader congenital scoliosis population. Similarly, the minimum 12-month follow-up duration, while adequate for capturing early complications and radiographic changes, is not sufficient to evaluate long-term spinal growth, fusion durability, or implant longevity particularly through adolescence. These limitations inherently restrict the strength of causal conclusions and underscore the need for prospective, long-term, multicenter studies with broader inclusion criteria to validate and expand upon these findings.

Conclusion

There was no statistically significant difference in correction outcomes between children aged ≤ 5 years and those aged 5–10 years who underwent posterior HV resection and short-segment fusion for solitary simple lower thoracic or lumbar HV (T8–L5). However, the postoperative complication rate and unplanned reoperation rate were significantly higher in the younger age group compared to the older age group.

Abbreviations

HV

Hemivertebra

CS

Congenital scoliosis

CSVL

Sacral vertical line

PJK

Proximal junctional kyphosis

Authors’ contributions

Conceptualization, Shangyu Guo and Haodong Li; methodology, Zhiqiang Zhang; formal analysis, Yiming Zheng; writing original draft preparation, Shangyu Guo; writing review and editing, Junfeng Wang and Maoxiang Qian; visualization, Dong Fu; project administration, Dahui Wang; funding acquisition, Chuang Qian. All authors have read and agreed to the published version of the manuscript.

Funding

This study was supported by The National Key Research and Development Program of China: Construction and Demonstrative Application of a Prevention and Intervention System for Structural Birth Defects in Children (No. 2021YFC2701003), and the 2022 Annual Open Fund Project of the Shanghai Key Laboratory of Birth Defects: General Project - Clinical and Fundamental Exploration of Optimal Surgical Age for Congenital Spinal Deformities (No. 2022CSQX1008).

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author (Yiming Zheng or Dahui Wang) upon reasonable request.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Children’s Hospital of Fudan University (Approval No. 2019 − 184). Informed consent was obtained from all the participants involved in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.