OUP user menu

Dental age assessment of adolescents and emerging adults in United Kingdom Caucasians using censored data for stage H of third molar roots

Teelana Boonpitaksathit , Nigel Hunt , Graham J. Roberts , Aviva Petrie , Victoria S. Lucas
DOI: http://dx.doi.org/10.1093/ejo/cjq101 503-508 First published online: 6 December 2010

Abstract

The root of the third permanent molar is the only dental structure that continues development after completion of growth of the second permanent molar. It is claimed that the lack of a clearly defined end point for completion of growth of the third permanent molar means that this tooth cannot be used for dental age assessment. The aim of this study was to estimate the mean age of attainment of the four stages (E, F, G, and H) of root development of the third molar. The way in which the end point of completion of stage H can be identified is described.

A total of 1223 dental panoramic tomographs (DPTs) available in the archives of the Eastman Dental Hospital, London, were used for this study. The ages of the subjects ranged from 12.6 to 24.9 years with 63 per cent of the sample being female. Demirjan’s tooth development stages (TDSs), for the first and second molars, were applied to the third molars by a single examiner. For each of stages E, F, and G and for stage H censored data, the mean ages of the males and females were compared, separately within each tooth morphology type using the two sample t-test (P < 0.01). The same test was used to compare the mean ages of the upper and lower third molars on each side, separately for each gender. The mean age of attainment and the 99 per cent confidence interval (CI) for each TDS were calculated for each third molar. The final stage H data were appropriately censored to exclude data above the age of completion of root growth.

The results showed that, for each gender, the age in years at which individuals attained each of the four TDSs was approximately normally distributed. The mean age for appropriately censored data was always lower than the corresponding mean age of the inappropriately censored data for stage H (male UR8 19.57, UL8 19.53, LL8 19.91, and LR8 20.02 and female UR8 20.08, UL8 20.13, LL8 20.78, and LR8 20.70). This inappropriately censored data overestimated the mean age for stage H. The appropriately censored data for the TDSs of the third molar may be used to estimate the age of adolescents and emerging adults assuming average growth and development and recent attainment of stage H.

Introduction

The assessment of an individual’s age has important roles in social, legal, and forensic settings. Interest in age determination of adolescents and emerging adults has risen in recent years due to an increasing number of mature-looking asylum seekers claiming to be less than 18 years of age. Various methods are available for age assessment (Lewis and Garn, 1960; Anderson et al., 1975; Gulati et al., 1990). One of the important indicators is tooth development (Garn et al., 1962), which is assessed using radiographs (Liversidge et al., 2003). Dental panoramic tomographs (DPTs) provide a useable image of all the tooth morphology types (TMTs) in the permanent dentition, including the third molars (Solari and Abramovitch, 2002).

It has been shown that tooth development is independent of hormonal and nutritional factors (Anderson et al., 1975) and comparative studies have also demonstrated that the age assessed by dental development correlates more closely with chronological age than other developmental measurements, such as skeletal development, height, and weight (Liliequist and Lundberg, 1971; Anderson et al., 1975; Demirjian et al., 1985). It has also been shown that age of attainment of maturity of third molars is closely linked to chronological age (Solari and Abramovitch, 2002). International comparisons indicate sufficient ethnic variation in the age of attainment of tooth development stages (TDSs) to justify the use of ethnicity as a factor in assessing third molar development as an indicator of age (Schmeling et al., 2000; Olze et al., 2004).

The most frequently used method of dental age assessment (DAA) is that of Demirjian et al. (1973) and Demirjian and Goldstein (1976). The disadvantage is that it is limited to the use of data from the development of the first seven permanent mandibular teeth (Jaffe et al., 1990; Davis and Hägg, 1994; Liversidge et al., 1999; Kostara et al., 2000). As this excludes the third molars, it is unsuited to assessing a subject’s dental age at the 18 year threshold. Thus, there are only a few publications on the third permanent molars that can be used in DAA from population samples in the USA (Mincer et al., 1993; Solari and Abramovitch, 2002), Belgium (Willems et al., 2002), Germany (Olze et al., 2004), and South Africa (Nortje, 1983). None of these deals appropriately with the final stage (H) of third molar root development and these studies do not relate to Caucasians. The third molars provide a measurement of development beyond the age of 15.75 years when the second permanent molar completes development (Demirjian et al., 1973).

The use of the final stage H of third molar development is problematic as it is not possible to identify the attainment of the end point of root development from a single radiograph. A prevailing belief is that a tooth with completed growth (stage H) for any TMT should not be used for DAA. This is based on the premise that it is impossible, when viewing different TMTs, to identify precisely when all subjects in a reference sample have achieved stage H (Kullman et al., 1992). However, most investigators use stage H but in various ways. Usually, the age is capped at some defined value (i.e. any subject whose age is above this cap is excluded from the data set), although these values for capping vary from one study to another. For example, Nortje (1983) capped at 21 years, Kullman et al. (1992) at 25 years, Mincer et al. (1993) at 24.9 years, and Olze et al. (2004) at 26 years. Using this approach, the mean ages for attainment of stage H, for example, are 20.9 years (Mincer et al., 1993) and 22.5 years (Olze et al., 2004), a range of nearly 2 years. This is clearly unsuitable and is a consequence of those investigators failing to indicate how they identified the appropriate age at which to cap the data. For future reference, the use of a data set that has been capped in this way is referred to as ‘inappropriately censored’ data set.

The aim of this study was to explore the possibility of excluding data above the age at which apical closure is complete in all subjects in a population sample. It gives an indication of how reference data relating to third molar root development, including the final stage of apical closure, can be used to estimate the dental age of emerging adults.

Materials and methods

Ethical approval and radiographs

The protocol for the study received ethical approval from the University College Hospitals NHS Trust—reference 03/E023. The data for this study were obtained by reuse of DPTs provided for routine dental assessments. Patients with medical conditions that might have affected growth and development were excluded. The sample comprised female and male Caucasian adolescents and emerging adults who were resident in the London area of the UK. Radiographs obtained for clinical purposes from consecutive patients, with known dates of birth, attending the Eastman Dental Hospital between March 2005 and July 2006, were examined and, if of good diagnostic quality, the parents and/or patients were approached and asked if the radiograph could be included in the reference database for research purposes. In addition, radiographs from the Eastman Dental Hospital archives were included, giving a total of 1223 radiographs, which met the inclusion criteria. The information on third molars obtained from radiographs comprised the data used in this study. Ethnicity was obtained from the clinical records, demographic data sheets, or by direct examination and questioning of parents and/or patients.

Tooth development stages

The assessment of TDSs was carried out using the method described and defined by Demirjian and Goldstein (1976) (Figure 1). One investigator (TB) examined all the radiographs. The reference database comprised the ages of attainment of the TDSs for all 16 TMTs, by convention on the left side, present in the human dentition. Because of the relatively frequent developmental absence of third molars, data were also obtained for the lower and upper right third molars (LR8 and UR8). These reference data are designed to be used to estimate the age of emerging adults with no birth records (Olze et al., 2004).

Figure 1

The four tooth development stages of the molar redrawn according to the description of Demirjian et al. (1973).

Statistical analysis

A set of 30 randomly chosen DPTs from the database, each with four third molars, was used to assess intra- and inter-rater agreement of the TDSs present in the third molars. Two raters (TB and GJR) assessed all DPTs twice; the order of the TDSs was randomized and the time interval between the two ‘blind’ assessments was 1 week. In addition, the first assessment of all the DPTs by each of the two raters was used to determine inter-rater agreement. The degree of agreement in each instance was calculated using the Kappa statistic (Cohen, 1960) and assessed according to the categories suggested by Landis and Koch (1977).

For each of stages E, F, and G, for every third molar tooth, the patient’s chronological age together with other demographic information was entered onto a proforma, transferred into a Microsoft Access database, and exported to a Microsoft Excel spreadsheet. The frequency distributions of age were graphed using Stata (Stata Corporation, College Station, Texas, USA). As they each exhibited an approximately normal (Gaussian) distribution, the mean ages for these stages, with the data partitioned by gender and ethnicity, were calculated.

The data for stage H were treated differently. Initially, the age data for each third molar tooth attaining stages G and H were arranged in ascending order with an indication of the associated TDS, either G or H. This enabled the identification of the youngest age beyond which there were only cases of stage H (i.e. no more cases of stage G) for the single third molar. All cases older than this age were excluded from the data set. The remaining data are referenced to the appropriately censored data as opposed to the inappropriately censored data which has an ‘arbitrary’ cut-off point. The appropriately censored data had a cut-off of 23.92 years.

To obtain inappropriately censored age, colleagues were informally asked the question ‘what is the age of the oldest patient who still has an immature root (stage G) of the third molar?’ The oldest age provided was 23.5 years. To ensure that all developing third molars were included in the sample, the age was arbitrarily capped at 24.99 years.

The appropriately censored age data were approximately normally distributed so the estimated mean age of stage H of each third molar in the appropriately censored sample was determined, together with the 99 per cent confidence interval (CI) for the mean (Altman, 1991). Previous investigators (Mincer et al., 1993; Willershausen et al., 2001; Solari and Abramovitch, 2002; Willems et al., 2002; Olze et al., 2004) have calculated the mean age of the inappropriately censored data, so, in order to assess the effect of censoring, the mean age (99 per cent CI) of the inappropriately censored data from the sample was also calculated and visually compared with the mean age of the appropriately censored data. It was not appropriate to perform a formal statistical hypothesis test to compare the mean values because the appropriately censored data were a subset of the inappropriately censored data.

For each of stages E, F, and G and for stage H censored data, the mean ages of males and females were compared, separately within each TMT using the two sample t-test. The same test was used to compare the mean ages of the upper and lower third molars on each side, separately for each gender. To avoid spurious results as a consequence of multiple testing, a significance level of 0.01 was used.

Results

All subjects were physically healthy, their chronological ages ranged from 12.6 to 24.9 years with females comprising 63 per cent of the sample. The intra-rater agreement using Cohen’s kappa was ‘almost perfect’: 0.94 for examiner 1 and 0.85 for examiner 2. The inter-rater agreement between the two investigators was 0.69, which is ‘substantial’.

The frequency distributions of age for stages E, F, and G of third molars for both females and males all approximated a normal distribution (Figure 2 shows one example; Figure 3 illustrates an example of a case with third molars present at stage F). The mean age (99 per cent CI) for the data for stages E, F, and G and for the appropriately censored data for stage H for both females and males are shown in Table 1.

View this table:
Table 1

Summary measurements for age of attainment in years of stages E, F, G, and H of third molar root formation in males and females.

ToothMean and 99% CIStage EStage FStage GStage H inappropriately censored data
MaleFemaleMaleFemaleMaleFemaleMaleFemale
UR8n123181811919212774101
Mean15.0914.9815.3316.2117.2417.5319.5720.08
99% CI14.84–15.3414.77–15.9115.15–15.5116.00–16.4216.99–17.5017.26–17.7919.07–20.0119.70–20.45
UL8n1242001311978712270101
Mean15.1414.9816.0116.1617.2017.6319.5320.13
99% CI14.88–15.3914.70–15.2015.78–16.2315.94–16.3816.94–17.4717.37–17.8818.67–19.6619.74–20.49
LL8n13718813116571995059
Mean15.3415.4516.5316.6817.5618.2519.9120.78
99% CI15.09–15.5815.21–15.6916.29–16.7816.41–16.9517.21–17.9117.85–18.6519.25–20.5720.32–21.24
LR8n149178129162691124960
Mean15.4015.5916.4216.4617.7118.1820.0220.70
99% CI15.13–15.6715.33–15.8516.18–16.6716.20–16.7217.34–18.0817.83–18.5119.35–20.6820.27–21.17
  • The age of attainment of tooth development stages (TDSs) was achieved earlier in males when compared with females by approximately 11 months. The order was reversed for UR8, UL8, and UL8. Of the 16 comparisons, only eight were statistically significant. The age of attainment of all TDSs of the upper teeth was always ahead of the lower teeth. This was statistically significant at P < 0.01 for 10 of the 16 comparisons, and between P < 0.05 and P < 0.01 for four comparisons. The remaining two differences were not statistically significant.

Figure 2

Lower right third permanent molar stage F, female (LR8Ef).

Figure 3

Example of a subject with an unknown date of birth with the lower third molars at stage F.

Upper third molar stages E, F, and G for both females and males showed a tendency for completion of growth earlier than the comparable lower third molar (Table 1). The differences in mean age between the upper and lower third molars varied from approximately 3 to 9 months, with almost all differences statistically significantly different (P < 0.01). There was no consistent pattern in the gender comparison and few of the differences in mean age between males and females were statistically significantly (Table 1).

The mean age for attainment of stage H for each third molar in the appropriately and inappropriately censored data sets were compared and the estimated mean ages for theses data sets are presented in Table 2. As expected, comparison showed that, for each gender, the mean age of the appropriately censored data was consistently lower than the corresponding mean age from the inappropriately censored data. The differences in the mean ages ranged from 0.20 to 0.67 years.

View this table:
Table 2

Comparison between summary data of stage H using the inappropriately censored data set and the appropriately censored data set for males and females.

ToothStatisticStage H—malesStage H—females
Inappropriately censoredAppropriately censoredInappropriately censoredAppropriately censored
UR8n747210193
Mean19.5719.4220.0819.81
99% CI19.07–20.118.96–19.8819.70–20.4519.47–20.15
UL8n706810191
Mean19.5319.3820.1319.77
99% CI18.67–19.6618.91–19.4519.74–20.4919.43–20.09
LL8n50455951
Mean19.9119.4620.7820.34
99% CI19.25–20.5718.87–20.0520.32–21.2419.94–20.74
LR8n49476058
Mean20.0219.8120.7019.71
99% CI19.35–20.6819.18–20.4520.27–21.1720.24–21.16
  • Statistical analysis was not carried out as the appropriately censored data set was a subset of the inappropriately censored data.

Discussion

The sample data in this study indicate that, on average, males achieved most of the TDSs for third molar teeth significantly later than females (Tables 1 and 2). This is consistent with other studies of third molar maturation (Haavikko, 1974; Levesque et al., 1981; Mincer et al., 1993).

This investigation, which is the first to apply appropriate censoring to the stage H data set, appears to raise the issue of the impact of inappropriately censored data. It is clear that for both males and females in stage H, each mean age derived from the appropriately censored data set is lower by approximately 2 to 4 months than the corresponding ‘pseudo’ mean age derived from the inappropriately censored data set (Table 2). This is not surprising as the act of appropriately censoring the data excludes the highest ages and the values previously included which are beyond the true end point of stage H.

Age assessment for subjects up to the age of 15.75 years with multiple developing teeth, including third molars, using TDSs (Demirjian and Goldstein, 1976) and the mathematical techniques of meta-analysis, has recently been published (Roberts et al., 2008). In older subjects who have only stage H of their third molars, the appropriately censored data can be used to estimate the dental age of the emerging adult. It is important to recognize that when estimating age, an overall assessment of the subject is required. This should take account physical appearance, demeanour, and further radiographic information such as a postero-anterior chest view, which visualizes the stage of development of the sterno-clavicular joint (Kreitner et al., 1998). Further research is required to extend these procedures for stage H to examine the relationship between third molar development and ethnicity.

Conclusions

The results showed that, for each gender, the age in years at which individuals attained each of the four TDSs was approximately normally distributed. The mean age for appropriately censored data was always lower than the corresponding mean age of the inappropriately censored data for stage H. This inappropriately censored data overestimates the mean age for stage H. The appropriately censored data for the TDSs of the third molar may be used to estimate the age of adolescents and emerging adults: this assumes average growth and development and recent attainment of stage H. In the absence of any developing TDSs, a subject’s age may be estimated using the mean age of attainment of stage H of the third molars from an appropriately censored data set.

Acknowledgments

We are grateful to the dental records staff of the Eastman Dental Hospital for providing the radiographs used in this study.

References

View Abstract