The assessment of growth status is crucial for determining the optimal timing for orthodontic intervention, and it can be performed through the analysis of skeletal maturation. Fishman developed the Skeletal Maturation Index (SMI), which involved the utilization of hand and wrist radiographs. The SMI demonstrated its capacity for predicting skeletal maturation and growth status. Nevertheless, this method required additional radiographs, which in turn exposed the patient to a higher dose of radiation. ^{1, 2} In 1972, Lamparski introduced a method for analyzing skeletal maturation by examining the morphology of cervical bones, concluding that the morphological changes of C2 to C6 could be used to evaluate skeletal maturation and had the same diagnostic value as metacarpal radiographs. As the analysis is performed on cephalometric radiographs, which is a routine examination in orthodontic diagnosis, this method had the benefits of cost-saving and minimal X-ray exposure. ^{3– 6}

Baccetti et al. in 2002 determined skeletal maturation based on the morphologic characteristics of the second, third, and fourth cervical vertebrae (C2, C3, and C4). This Cervical Vertebrae Maturation (CVM) method comprises five maturational stages (CVMS I–CVMS V), with the mandibular growth peak occurring between CVMS II and CVMS III. ⁷ In 2005, this method was modified into six maturational stages (CS 1-CS 6), with the mandibular growth peak occurring between CS 3 and CS 4. Clinically, this method can be used to evaluate the growth status of subjects who require growth modification or orthognathic surgery and from there to determine the optimal time to intervene. ⁸ According to a systematic review and meta-analysis conducted by Cericato et al., the cervical vertebrae method by Baccetti et al. had been found to be a reliable alternative for hand-wrist x-ray in predicting the peak of a growth spurt. ⁹ When comparing the HW and CVM methods, Cunha et al. concluded that both methods presented good reproducibility for intra and inter-examiner correlation. The availability of lateral cephalograms as routinely diagnostic aid represent an advantage for CVM methods compare to the HW. ¹⁰

There are two widely used methods for determining cervical vertebrae maturation, CVMS and CS stage. Study by McNamara and Franchi (2018) demonstrated that not all maturation levels in CS stage are reproducible. The occurrence of spikes may appear to be extensions of the inferior border, leading to incorrect interpretation of an early concavity or notching when determining the stages. ¹¹ Given this data, this study aim to assess the reproducibility of two cervical vertebral maturation methods: the CVM method by Baccetti et al. (2002) and the CS method by Baccetti et al. (2005) with a particular focus on identifying which method exhibits superior reproducibility.

This research has been granted ethical permission by the Research Ethical Committee of the Faculty of Dentistry Universitas Indonesia Number 83/Ethical Approval/FKGUI/XII/2021. This is a retrospective observational study. Sample size calculation was performed according to Walter et al, where ρ0 = 0.8, ρ1 = 0.9, α = 0.05 and β = 0.2. The minimum sample size required was 45.8 or 46 cephalograms after rounding up. Ninety lateral cephalograms from Universitas Indonesia Dental Hospital database. ¹² The cephalograms were selected according to the criteria as follows: (1) Patient was between 8-18 years old; (2) Patient had class I malocclusion; (3) Patient had no history of facial trauma; (4) Patient did not present any craniofacial and cervical vertebrae malformations; (5) Patient did not present any syndrome or hormonal alteration affecting growth and development. Written Informed consent was obtained from all patients whose data included in our study.

Two orthodontists with five and ten years of experience evaluated the cephalometric radiographs. The skeletal maturation stage was determined by manually tracing the C2, C3, and C4 of cervical vertebrae with a pencil on acetate tracing paper. The patient information listed on each cephalogram was concealed. The maximum number of X-rays that each examiner could evaluate in a single day was ten. Examiners were provided explanatory text and illustrations of the CVMS and CS identification methods. Each examiner was instructed to analyze the cephalograms using CVMS method (5 stages) and the CS method (6 stages), both as described by Baccetti et al. ^{7, 8} For intra-examiner analysis, each examiner re-evaluated with a one-week interval between evaluations. The statistical analysis was performed using Datatab. Using the weighted kappa test, intra-examiner reliability between T1 and T2 and inter-examiner agreement were analyzed.

Two examiners analyzed ninety cephalograms, consisting of 49 females and 41 males. The age of females and males (as mean ± standard deviation) were 14.98 ± 1.59 and 15.37 ± 2.02 years. Both examiners performed staging in both methods at two different times (T1 and T2) to analyze the intra-examiner reliability. For intra-examiner reliability analysis, the weighted kappa statistic was applied. With a weighted kappa value of 0.86-0.95, both examiners had almost perfect agreement between T1 and T2 in both methods ( Table 1). We can conclude that each examiner’s assessment of skeletal maturation level was reliable.

Table 2 and Table 4 demonstrated the inter-examiner staging for the CVM method. Fifty-one of the ninety samples showed perfect agreement between the two examiners. Out of the total number of disagreements, 37 cases have single-stage disagreement. There were two stages of disagreement between CVMS II and CVMS III. According to weighted kappa analysis, the agreement fell into a substantial agreement with k = 0.66 ( Table 1).

Table 3 illustrated the interobserver staging in the CS method. The agreement between the two examiners resulted in almost perfect agreement ( Table 1). The percentage of agreement in this method was higher than in the CVM method (75,56%) as shown in Table 4. All of the disagreements between examiners had a difference of one stage. The weighted kappa value was also higher than the CVM method with k = 0.82. This study showed that the CS method provided better agreement, with 68 cases out of 90 having perfect agreement (65.6%).

A comprehensive understanding of orocraniofacial growth and development is crucial to facilitate accurate diagnosis, optimal treatment planning, and long-term stability of orthodontic treatments. Several methods have been investigated to identify the growth stage and determine the most appropriate timing for treatment. ¹³ These include chronological age, skeletal age, and skeletal maturation, the latter of which could be analyzed using a SMI derived from a hand-wrist radiograph or a cervical vertebral maturation index derived from a cephalometric radiograph. ^{6, 14}

Numerous cervical vertebral analysis techniques have existed. Jaqueira et al. assessed the reproducibility of three methods: Hassel and Farman, Baccetti et al., and Seedat and Forsberg. The results demonstrated that each method had clinical applicability. The method developed by Baccetti yielded the best results, followed by Hassel-Farman and Seedat-Forsberg. ^{4, 7, 15, 16} In contrast to this study, Cericato et al. revealed that Hassel-Farman’s methods are superior to Baccetti’s. ⁹

Baccetti et al. presented a skeletal maturation analysis method consisting of five maturational stages (CVMS I - CVMS V) based on the anatomical characteristics of C2, C3, and C4 cervical vertebrae. This method was then modified into six maturational stages identified as CS 1 through CS 6. According to previous research, when compared to Fishman’s hand-wrist method as the gold standard, both the Fishman and Baccetti CVM methods were valid for predicting growth status, with Fishman’s method being slightly more accurate. ^{17, 18} However, the CVM method has the advantage of less radiation exposure due to the utilization of lateral cephalometry in orthodontic diagnosis.

Despite Baccetti’s CVMS (five stages) and CS (six stages) methods having recently gained worldwide popularity among orthodontists, the reproducibility of these techniques remains debatable. Rainey et al. and Cunha et al found that the intra and inter-examiner agreement in cervical vertebrae method presented good reproducibility. ^{10, 19} In contrast to these results, Gabriel et al., and Sohrabi et al., showed that the reproducibility of this method is poor, with the level of agreement were below 50%. ^{20, 21}

In this study, both methods showed that the intra-observer reliability values for each examiner were “almost perfect agreement”. Flores-Mir et al. demonstrated a comparable result for the CVM method with an ICC coefficient of 0.889, which was categorized as “excellent agreement”. ⁵ In previous research by Gabriel et al. only moderate agreement was demonstrated in the CS method, while this present study showed better intra-observer reliability. ²⁰ This study involved two orthodontists as the examiners, each with varying levels of experience. The weighted kappa values indicated that all observers, regardless their level of experience, demonstrated good reliability in assessing cervical vertebrae staging of both methods. Notably, both observers exhibited higher kappa values for CS method in comparison to CVM method, reflecting a higher level of agreement in the former. Rongo et al. evaluated the reproducibility of the cervical vertebra maturation method among three observers with different levels of orthodontic experience, the junior group, postgraduate group, and specialist group. They concluded that the reproducibility of cervical vertebrae maturation method was not improved by the level or orthodontic experienced. The junior group, with the lowest level of orthodontic experience showed the best performance. ²²

The inter-observer agreement in this study, was greater in the CS method than in CVM method. The presence of alternative morphologies in several stages of the CVM method could potentially cause this issue. The operator was required to evaluate a greater range of morphologies and led to confusion. Contrary to that, the CS method provides a single morphological option at each stage, thereby simplifying the identification process. As opposed to the findings of this study, Duque et al. found that the CMV method had better inter-examiner agreement than the CS method. According to Duque et al., the appearance of the concavity in the inferior edge of C2, could have raised doubts during classification. ²³ However, we argued that although there are more stages in the CS method than in the CVM method, the number of morphological shapes that needed to be evaluated was greater in the CVM method.

The agreement between the two methods relied on the examiner’s proficiency in distinguishing the curvature of the lower edge of cervical vertebrae C2, C3, and C4, as well as the shapes of C3 and C4. Notably, significant discrepancies in the CVM method were observed at stages CVMS III, IV, and V, where the differentiation between these stages primarily relied on the morphological characteristics of C3 and C4. Specifically, when both C3 and C4 exhibited a horizontally rectangular shape, they are classified as CVMS III. Misidentification might occur when the examiner found it challenging to distinguish between the horizontally rectangular shape of CVMS III and the trapezoidal shape of CVMS II, especially in cases where the anterior and posterior heights of cervical vertebrae are nearly the same. The identification of CVMS IV and V was subject to the same challenges. The shapes of C3 and C4 in CVMS V can be either square or vertically rectangular. It was difficult to differentiate the squared from the vertically rectangular, which could lead to the inaccurate identification of CVMS V as CVMS IV and vice versa. This issue was also evidenced in the CS method, where there were numerous disagreements between CS 3, CS 4, and CS 5. ^{9, 23, 24}

The findings of this study align with those reported by Sohrabi et al., who investigated the factors contributing to the poor reproducibility of CVM method. In their research, the observers were required to complete a questionnaire prior to evaluating the stages using the CVM method, which specifically addressed the concavity of the lower borders of cervical vertebrae CS, C3, and C4, as well as the morphological characteristics of C3 and C4. The study concluded that the challenges in achieving reproducibility with the CVM method stemmed primarily from the difficulties associated with accurately determining the morphology of C3 and C4. ²¹ Furthermore, Nestman et al. conducted a similar investigation focusing on the reproducibility of CS method. Their results indicated that the assessment of the curvature of the lower border C2, C3, and C4 exhibited a higher degree of agreement among observers, with moderate agreement for C2 and substantial agreement for C3 and C4. In contrast, the determination of the shapes of C3 and C4 yielded only fair agreement, highlighting the variability and potential inaccuracies in assessing these specific morphological features. ²⁵

According to Hassel and Farman, skeletal maturation is a continuous process characterized by a gradual transition between stages, which can complicate the determination of maturation stages in borderline cases. This complexity is particularly evident when assessing the anterior and posterior edges of the cervical vertebrae. Additionally, the presence of spikes along the inferior borders of the cervical vertebrae can further complicate this assessment. These spikes might be misinterpreted as extensions of the inferior border during stage determination, potentially leading to the incorrect identification of an early concavity. ^{4, 16, 26}

One of the limitations of this study was that the examiners did not receive formal training in the CVM and CS methods for analyzing skeletal maturation. Instead, the examiners were provided only with a directive and reference materials for both methods, which included definitions of each stage as their sole guidelines. The absence of thorough training might have negatively influenced the reproducibility of both methods. Rainey al. trained observers with no previous experience in the evaluation of cervical vertebrae, and the outcome was substantial agreement. ¹⁹ Perinetti et al. concluded that when specific training and clear guidelines were provided, the CS method could be more accurate and repeatable to an acceptable degree. ²⁴ Despite the benefits of training, Gabriel et al and Nestman et al. revealed only moderate agreement among observers, even when they had received training prior to staging. This suggested that the effectiveness of training might vary and that additional factors could influenced the reliability of the CVM and CS method. ^{20, 25}

According to a systematic-review and meta-analysis by Cericato et al., the CVM method has the potential to replace hand-wrist analysis for predicting the pubertal growth spurt due to its high reliability. ⁹ Ferrillo et al. and Lucchese et al. further supported this assertion, demonstrated that the CVM methods developed by Baccetti et al and Hassel-Farman are both reliable and effective for analyzing skeletal maturation during growth period. When using cervical vertebrae maturation as an aid to assess growth status, this study showed that CS method had better reproducibility. ^{27, 28} Another study stated that the CVM method could be used to determine the end of active growth and the optimal time for intervention in patients who require surgery for the late correction of facial disharmonies, which happened in CS 6, when growth completed. ⁸ However, a study by Pasciuti et al showed that 17% of females did not develop a CS 6 though active growth had ceased. ²⁹ Other study by Perinetti et al. also discovered that the sensitivity of the CS method in stages CS 4-CS 6 is less than in stages CS 1-CS 3. Among 450 participants between ages 20-45 years old in the study, 48 individuals were found to be at CS 4. Although the percentage is low, it appeared to be clinically significant as it may influence treatment decisions. ³⁰ These findings indicates that the cervical vertebral maturation may be more beneficial in growing patients and less reliable in predicting the post-puberty period.

The inter-observer reliability using the CVM method presented by Baccetti et al. in 2002 showed moderate agreement. While the CS method published by the same author in 2005 presented substantial agreement. Based on this research, CS method had better agreement than CVM method in assessing skeletal maturity.