Validity of the assessment method of skeletal maturation by cervical vertebrae: a systematic review and meta-analysis

Abstract

Objectives:

To perform a systematic review with meta-analysis to answer the question: is the cervical vertebrae maturation index (CVMI) effective to replace hand–wrist radiograph (gold standard) in determining the pubertal growth spurt in patients undergoing bone growth?

Methods:

A search in three databases was performed, in which studies were selected that compared one of the two main assessment methods for cervical vertebrae (Hassel B, Farman AG. Skeletal maturation evaluation using cervical vertebrae. Am J Orthod Dentofacial Orthop 1995; 107: 58–66, or Baccetti T, Franchi L, McNamara JA Jr. An improved version of the cervical vertebral maturation (CVM) method for the assessment of mandibular growth. Angle Orthod 2002; 72: 316–23) to a carpal assessment method. The main methodological data from each of the texts were collected and tabulated after. Later, the meta-analysis of the correlation coefficients obtained was performed.

Results:

19 articles were selected from an initial 206 articles collected. Regardless of the method used, the results of the meta-analysis showed that every article selected presented a positive correlation between skeletal maturation assessment performed by cervical vertebrae and carpal methods, with discrepancy of values between genders indicating higher correlation for the female gender (0.925; 0.878) than for the male (0.879; 0.842). When the assessment was performed without gender separation, correlation was significant (0.592; 0.688) but lower in the cases when genders were separated.

Conclusions:

With the results of this meta-analysis, it is safe to affirm that both CVMIs used in the present study are reliable to replace the hand–wrist radiograph in predicting the pubertal growth spurt, considering that the highest values were found in female samples, especially in the method by Hassel and Farman.

Introduction

Growth and development of the human being is a phenomenon that happens naturally and has acceleration periods that are called growth spurts.^1,2 From these, the most used and availed in orthodontic treatment is perhaps the one occurring during puberty, also known as pubertal growth spurt (PGS).^3,4 Chronological age has not proved to be the most reliable method to determine the biological maturity level of the individual, so other indicators are recommended to assess the development stage in which the individual lies.^3,5

The most popularly known assessment method of the PGS is the one that uses carpal radiograph and skeletal changes that occur in the hand and wrist region, including the gradual appearance of carpal bones and changes in fusions among epiphyses and diaphyses of the phalanges of the fingers.^4,6 However, although this is a reliable method that requires low doses of radiation, it is an additional examination to the orthodontic documentation, and thus an additional exposure to the patient. This goes against the principles of as low as reasonably achievable for ionizing radiation, which aims to minimize the doses that patients and workers are exposed to.^4,7

A more recent method to assess the PGS is the one that uses changes in shape and level of development of the cervical vertebrae C2, C3 and C4, the images of which show up on the profile teleradiographs—a routine test in clinical orthodontic practice.^8,9 Thus, the most used methods are the ones developed by Hassel and Farman,¹⁰ and Baccetti et al.¹¹ A distinct advantage of the assessment of cervical maturity is that it does not imply additional exposure to radiation.⁹ Therefore, if proved that carpal and cervical vertebrae methods present high correlation, an additional radiograph to assess skeletal maturation would not be needed, and such a request would not be justified.^2,7

Thus, this work aimed to perform a systematic review and meta-analysis to answer the question: is the cervical vertebrae maturation index effective to replace hand–wrist radiographs (gold standard) in determining the PGS in patients undergoing bone growth?

Methods and materials

Search strategy and eligibility criteria

To perform this work, a systematic search in literature was performed with no language restriction to identify relevant studies. Databases used were ScienceDirect, PubMed and SciELO. A search filter for the last 10 years was applied, with the aim to work with the most updated literature on the subject. Keywords used for the search were “cervical vertebrae”, “hand-wrist”, “skeletal maturation” and “skeletal age assessment”, as shown in Table 1.

Table 1.

Strategies for database search

MeSH search terms	Database
	PubMed	ScienceDirect	SciELO	Total
“Cervical Vertebrae” [MeSH] AND “Hand-wrist” AND “Skeletal Maturation” OR “Cervical Vertebrae” AND “Hand-wrist” AND “Skeletal Age Assessment” [MeSH] OR “Cervical Vertebrae” AND “Hand-wrist” AND “Maturation”	36	153	X	189
“Cervical” AND “Vertebrae” AND “Maturation”	X	X	17	17
Total	36	153	17	206

The selection of studies included either cross-sectional or descriptive longitudinal studies, written in Portuguese, Spanish or English, that qualitatively or quantitatively compared the two main assessment methods of skeletal maturation of the cervical vertebrae (Hassel and Farman¹⁰ method and method of Baccetti et al¹¹) to the carpal assessment method, considered the gold standard.

Excluded from the selection were studies with samples made by unhealthy patients (patients with obesity, syndromes, cleft lip/palate, idiopathic scoliosis and other deformities), studies introducing a modified version of the cervical vertebrae method, abstracts, dissertations and/or theses, patents, convention abstracts, opinion articles, commentaries and editorials.

Initially, titles and abstracts of articles were identified, and when this information could not be collected solely based on the abstract, the full text of the scientific article was searched. Cases of publication duplicity in more than one of the databases consulted were only considered once. After the initial selection, the full texts were recovered for an assessment of the methodological strength by two reviewers for eligibility. The reviewers were not blinded to the authors or journals.

Data extraction

Using a standardized data extraction sheet, the following information was extracted and recorded from the studies: authors, year of publication, size and characteristics of the sample, age, assessment method of skeletal maturation by cervical vertebrae used as gold standard, inter- and intraexaminer correlation test used and correlation value between both methods assessed—these data were tabulated in a worksheet for later analysis. Moreover, each study received a score and was considered to be of low quality (0–5 points), medium quality (6–8 points) or high quality (9–11 points)—evaluation made from an adaptation by Santiago et al¹² (Table 2). The studies considered low quality were held as methodologically poor and were discarded. Disagreements between the reviewers were solved through a group reassessment of the article, until both reviewers were satisfied with the decision.

Table 2.

Quality scoring rubric for articles included in initial review, as suggested by Santiago et al¹²

Selection criteria	Score
Adequate study design	3 points
Adequate presentation of objectives	1 point
Adequate sample size	1 point
Adequate description of sample selection process and sample losses	1 point
Presentation of keywords inserted in MeSH	1 point
Method used for error analysis	1 point
Valid statistical methodology	1 point
Clear and objective presentation of results	1 point
Declared p-value	1 point

The study design was considered adequate when it was either cross-sectional or longitudinal, in which the conclusion clearly met the objectives. The sample was considered adequate when it mentioned the proper description of age and gender, ethnicity, systemic health problems, clearly excluding any growth or nutritional problem; any other condition, that is, drug use, which may change teeth and skeletal maturation; with no history of orthodontic treatment and size definition by sample calculation.

The use of method for error analysis was considered valid when mentioning the performance of intraexaminer or interexaminer correlation tests, regardless of the mention of the p-value, and the test used. The statistical methodology was considered valid when Pearson's or Spearman's correlation tests were presented between both assessment methods studied (vertebral and carpal).

Data analysis

Later, the meta-analysis of the correlation coefficients obtained was performed to analyse the results obtained in the studies that were part of the review. Calculations were performed in the program MedCalc v. 12.7.8 (MedCalc Software, Ostend, Belgium).

Results

Initial research resulted in a sample comprising 206 articles. After analysing them according to the inclusion and exclusion criteria, 16 articles were selected in the PubMed database, 9 articles in the ScienceDirect database and 3 articles in the SciELO database. The main reasons for exclusion are shown in Table 3 and Figure 1.

Table 3.

Criteria for exclusion after initial review

Reason for exclusion	PubMed (n = 36)	ScienceDirect (n = 153)	SciELO (n = 17)	Total
Exclusive use of VMI or CMI	6	12	2	20
Use of other VMI methods	5	5	3	13
Use of dental methods	1	9	1	11
Inadequate samples (syndromes, metabolic diseases)	1	2	0	3
Literature and/or systematic reviews	1	8	0	9
Articles in Chinese or French	1	2	0	3
Absence of correlation between VMI and CMI	4	3	2	9
Use of CT	1	3	0	4
Letters from the editor, opinion articles, abstracts for conferences	0	30	0	30
Other (glossary, indexes, appendices, bibliographies, responses to the authors, book chapters, handbooks)	0	70	6	76
Total exclusions	20	144	14	178
Articles included	16	9	3	28

CMI, hand–wrist maturation index; VMI, vertebral maturation index.

Figure 1.

Statement flowchart of article selection process.

From the 28 articles selected, 5 were excluded because of duplicity in the PubMed and ScienceDirect databases, 3 were excluded for receiving a methodological score between 0 and 5 points (low quality) and 1 article was excluded because it was simultaneously published in 2 journals. Thus, the total sample was 19 articles, as shown in Table 4.

Table 4.

Main characteristics of articles included in the study

Study	Article score	VMI	CMI	Sample (n)	Age (years)	Correlation intra	Correlation inter
Damian et al13	9	HF	Grave and Brown	105	7–18	0.939a; 0.804b	0.948a; 0.779b
Flores-Mir et al6	8	BFM	Fishman	79	ND	0.985a; 0.889b	NP
Gandini et al1	9	BFM	Bjork	30	7–18	VNP	NP
Grippaudo et al14	8	BFM	Grave and Brown	90	6–14	NP	NP
Kamal et al15	9	HF	Fishman	50	10–12	0.81	NP
Ozer et al2	9	HF	MP3	150	9–19	0.98b; 0.966a	NP
Uysal et al16	9	HF	Bjork	503	5–24	0.955b; 0.983a	0.987b; 1.00a
Alkhal et al3	9	BFM	Fishman	400	10–17	0.846	0.846
Lai et al4	9	BFM	NTUH	709	8–18	0.933	0.90
Moscatiello et al17	8	BFM	Greulich and Pyle	140	4–14	NP	NP
Kapoor et al5	9	HF	Fishman	90	9–12	VNP	NP
Wong et al18	9	BFM	MP3	400	10–17	0.574b; 0.083a	0.161
Litsas and Ari-Dermikaya8	9	BFM	Grave and Brown	393	8–18	0.160b; 1.00a	NP
Warmeling et al19	10	BFM	Martins	158	6–18	0.80	NP
Mahajan20	9	HF	Fishman	100	8–18	VNP	NP
Sachan et al21	9	HF	Fishman	90	9–13	VNP	NP
Shim et al7	10	HF	Fishman	62	11–17	0.997a; 0.844b	0.983a; 0.877b
Navlani and Makhija9	8	BFM	MP3	120	8–16	NP	NP
Prasad et al22	8	HF	MP3	200	10–19	NP	NP

BFM, Baccetti et al;¹¹ HF, Hassel and Farman;¹⁰ MP3, phalanx average third finger method; ND, not declared; NP, not performed. NTUH, National Taiwan University Hospital skeletal maturation index; VNP, test performed, but value not presented.

^aValues related to the carpal method used in the study.

^bValues related to the vertebral method used in the study.

Regarding the cervical assessment method selected, nine articles used Hassel and Farman,¹⁰ and ten articles used Baccetti et al.¹¹ Regardless of the method used, the results of the meta-analysis of this work (Tables 5–10, Figures 2 and 3) showed that every article selected presented a positive correlation between the maturation assessments performed by both methods and the assessment methods that used the hand–wrist radiographs (gold standard), with discrepancy of values between genders.

Table 5.

Values obtained in studies and meta-analysis results for the method by Baccetti et al¹¹ and male gender

Study	n	r	Confidence interval 95%
Grippaudo et al14	48	0.700	0.519–0.821
Alkhal et al3	200	0.920	0.896–0.939
Lai et al4	330	0.910	0.889–0.927
Moscatiello et al17	66	0.507	0.302–0.667
Wong et al18	200	0.952	0.937–0.963
Litsas and Ari-Dermikaya8	170	0.846	0.797–0.884
Navlani and Makhija9	60	0.693	0.533–0.805
Meta-analysis	1074	0.842	0.733–0.909

Table 10.

Values obtained in studies and meta-analysis results for the method by Hassel and Farman¹⁰ with no gender distinction

Study	n	r	Confidence interval 95%
Damian et al13	105	0.625	0.492–0.729
Shim et al7	62	0.530	0.323–0.689
Meta-analysis	167	0.592	0.483–0.683

Figure 2.

Values obtained in the studies and meta-analysis result for male gender, female gender and no gender distinction with Baccetti et al.¹¹

Figure 3.

Values obtained in the studies and meta-analysis result for male gender, female gender and no gender distinction with Hassel and Farman.¹⁰

Table 6.

Values obtained in studies and meta-analysis results for the method by Baccetti et al¹¹ and female gender

Study	n	r	Confidence interval 95%
Grippaudo et al14	42	0.840	0.720–0.911
Alkhal et al3	200	0.936	0.916–0.951
Lai et al4	379	0.937	0.923–0.948
Moscatiello et al17	74	0.731	0.603–0.822
Wong et al18	200	0.940	0.921–0.954
Litsas and Ari-Dermikaya8	223	0.716	0.645–0.775
Navlani and Makhija9	60	0.861	0.777–0.915
Meta-analysis	1178	0.878	0.789–0.930

Table 9.

Values obtained in studies and meta-analysis results for the method by Hassel and Farman¹⁰ and female gender

Study	n	r	Confidence interval
Kamal et al15	25	0.858	0.700–0.936
Uysal et al16	290	0.880	0.851–0.904
Kapoor et al5	45	0.932	0.879–0.962
Mahajan20	50	0.975	0.956–0.986
Sachan et al21	45	0.932	0.879–0.962
Prasad et al22	100	0.910	0.869–0.939
Meta-analysis	555	0.925	0.881–0.953

Because they were different methods, the statistical analyses of the meta-analysis were separated according to the vertebral method and gender used in the sample.

Regarding the sample, from ten articles that used the method by Baccetti et al,¹¹ seven used samples separated by gender so as to assess the correlation between carpal and vertebral indexes and three articles presented no gender distinction for analysis (Tables 5–7, Figure 2). The results indicate higher correlation for the female gender (0.878) than for males (0.842). When the assessment was performed without gender separation, correlation was significant (0.688) but lower than for the cases when genders were separated.

Table 7.

Values obtained in studies and meta-analysis results for the method by Baccetti et al¹¹ with no gender distinction

Study	n	r	Confidence interval 95%
Flores-Mir et al6	79	0.720	0.593–0.812
Gandini et al1	30	0.783	0.589–0.892
Warmeling et al19	158	0.616	0.509–0.704
Meta-analysis	267	0.688	0.579–0.772

Nine articles that used the method by Hassel and Farman¹⁰ were assessed in the same way. Thus, two articles presented no gender distinction for analysis, six articles used samples separated by gender and one article used an exclusively male sample (Tables 8–10, Figure 3). The results also indicated higher correlation for the female gender (0.925) than for males (0.879). When the assessment was performed without gender separation, correlation was significant (0.592) but lower than for the cases when genders were separated. Furthermore, it is possible to observe that in the analyses with samples separated by gender, the values found when using the method by Hassel and Farman¹⁰ are higher in relation to the method by Baccetti et al,¹¹ considering this relation is inverted when the sample had no gender distinction.

Table 8.

Values obtained in studies and meta-analysis results for the method by Hassel and Farman¹⁰ and male gender

Study	n	r	Confidence interval 95%
Kamal et al15	25	0.892	0.767–0.952
Ozer et al2	150	0.893	0.855–0.921
Uysal et al16	213	0.780	0.721–0.828
Kapoor et al5	45	0.849	0.740–0.915
Mahajan20	50	0.981	0.967–0.989
Sachan et al21	45	0.849	0.740–0.915
Prasad et al22	100	0.727	0.619–0.808
Meta-analysis	628	0.879	0.787–0.933

Discussion

Cervical vertebrae, which are many times ignored by orthodontists who are not familiar to its analysis, have been increasingly used to assess skeletal maturation of the individual because it comprises an easy assessment method and eliminates the need for an additional radiograph (hand and wrist) for such ends. This work aimed to perform a systematic review and meta-analysis of the literature to verify the correlation between the two main vertebral indexes mentioned in literature and carpal assessment, by answering a question developed with the patient, intervention, comparison, outcomes (PICO) strategy.

It is important to stress the care that must be taken while applying the results presented in this systematic review, for although 14^{1–5,7,8,13,15,16,18–21} out of the 19 articles selected had received a score of either 9 or 10 (high quality), most of them work with convenience samples. Another limitation concerns the cutbacks of articles, considering that there are almost no studies with longitudinal assessment. Most articles use files from either courses or smaller samples owing to the radiographic exposure required to collect data. These cross-sectional studies present limitations regarding growth analysis, since it is relatively insensitive to individual variability. In this sense, longitudinal research would be the most indicated method for the detailed study of craniofacial growth, through which it would be possible to determine the type of development of the patient. Thus, it is suggested that longitudinal studies are performed in different populations, allowing a deeper understanding about the use of cervical vertebrae as an indicator of skeletal maturation.

A great inquiry regarding the validity of the cervical vertebrae method is about its reproducibility. From the 19 articles included in this study, most of them^{1,2,5,6,8,9,14,15,17,19–22} did not perform interexaminer statistical tests. However, in cases when this test was applied, the results showed significant correlations. Hence, further studies with the goal to assess the reproducibility of the method are required along with different dentistry expertise, aiming to obtain more knowledge about this subject that has not yet been clarified.

Another important aspect is the great variety of assessment methods of skeletal maturation by cervical vertebrae, including simple qualitative analysis of the vertebral shape, quantitative measurements of the vertebral shape (some limited to height and width) and depth of lower concavity, and others with more specific measurements performed through either morphometric analyses or linear regression. The results of this study refer to methods by Hassel and Farman¹⁰ and Baccetti et al,¹¹ both subjective methods of visual assessment of vertebrae C3 and C4 shapes, and the existence of lower concavity in C2, C3 and C4 vertebral bodies, since these two methods are the most used in literature, thus considerably increasing the possibilities of article search on the subject.

The authors investigated are unanimous in affirming that chronological age is not a good indicator to assess the rate of skeletal maturation of the individual and that vertebral methods may be used instead of carpal ones.^1–22 Thus, the results of the meta-analysis of this work showed that every article selected presented positive correlation between the two assessment methods of skeletal maturation, with discrepancy of values between genders. Regardless of the method used, the correlation values found by meta-analysis were higher for the female gender (0.878 for Baccetti et al¹¹ and 0.925 for Hassel and Farman¹⁰) than for males (0.842 for Baccetti et al¹¹ and 0.879 for Hassel and Farman¹⁰). In cases where genders were not separated, correlations found were lower (0.688 for Baccetti et al¹¹ and 0.592 for Hassel and Farman¹⁰).

Some studies^{1–3,8,14,20–22} highlight the need for special attention concerning gender distinction during the peak of the PGS, since the female gender always shows an advance regarding chronological age. Hence, in the present study care was taken to separate genders whenever possible, so as to perform the meta-analysis. The highest correlation found was for the female gender when using the method by Hassel and Farman.¹⁰ Perhaps, this might be explained by the fact that both vertebral methods used as inclusion criteria come from the Lamparski method (1972)—mentioned by Hassel and Farman,¹⁰ the first authors to describe the use of cervical vertebrae to assess skeletal maturation using an exclusively female sample to develop the method. The fact that correlation values were lower when samples were not separated (n = 5) could probably be explained by the small number of studies using a mixed sample.

The great advantage presented by the authors is the lack of need to require an additional radiograph,^1–22 since cervical vertebrae are visualized in profile teleradiographs, which is part of the basic orthodontic documentation.^3,8 Yet, several authors^{5–7,15,16,18–21} mention that the visualization of the vertebrae by orthodontists is more easily performed owing to the familiarity of these professionals with this type of radiograph.

Conclusion

Based on the results found, it is safe to affirm that both cervical vertebrae maturation indexes used in the present study are reliable to replace the hand–wrist radiographs in predicting the PGS, considering the highest values were found in female samples, especially in the method by Hassel and Farman.¹⁰ However, questions such as reproducibility of the methods assessed and the possibility of assessing growth and development of the individual has not yet been clarified, so the elaboration of more specific studies for elucidation are required.