This systematic review summarizes the evidence on the distribution of chronological age from pre-defined stages of clavicular ossification studied by MRI or CT in living individuals.
We included ten observational studies, all published within the last five years. The studies were conducted in five different countries, involving 4190 participants. Nine studies used CT and one used MRI. The QUADAS-2 checklist was used to assess risk of bias. The mean chronological age and its 95 % confidence interval (95% CI) were presented for males and females separately in each stage and substage of medial clavicle epiphysis.
Most of the studies showed a high risk of selection bias due to age mimicry, which could influence the distribution of age from each clavicular stage. Due to the limitations, we are unable to provide a reliable distribution of chronological age from each stage of medial clavicular development.
Every year, young asylum seekers come to Norway without legal documentation of their chronological age. To ensure that children receive their entitled rights and that adults are not treated as children, it is necessary to estimate their chronological age. Assessments of hand skeleton and third molar teeth using X-ray have been used for age estimation in Norway for years. We have previously published systematic reviews assessing age distribution using radiographs based on the Greulich & Pyle atlas for hand-wrist skeleton and the Demirjian’s stages for the third molar teeth. Here we present a systematic review to evaluate the use of computed tomography (CT) and magnetic resonance imaging (MRI) on the medial clavicle epiphysis for age estimation in the living.
We searched for studies in the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase and Google Scholar. This was a joint search conducted for studies using X-ray of teeth and hand, CT and MRI of the clavicle, knee and ankle in both males and females. An update search was conducted in April 2017 for clavicle, knee and ankle only. Two people screened the literature independently and assessed the risk of the included studies based on the QUADAS-2 checklist. The mean chronological age and the standard deviation in the included studies were extracted for each stage and substage for each sex separately.
We found 10059 abstracts in the first search and 663 in the second search. In total 63 potentially relevant publications were forwarded for full-text screening. Eventually ten studies were included, all published within the last five years. Nine studies used CT and one study used MRI. The sample sizes in the included studies ranged from 152 to 752 participants. Six studies were from Turkey, and one from Australia, China, Germany and Thailand, respectively.
The age distribution for each clavicular stage varied largely in and between the included studies. According to the QUADAS-2 checklist, all the included studies were associated with high risk of age mimicry bias. This particular bias lead to unreliable estimates of chronological age within each stage of medial clavicular ossification. In addition, we summarized information on the developmental asymmetry of right and left clavicles. Five relevant studies were found and 11% of included cases showed different epiphysis stages of clavicular ossification.
These findings are consistent with our previous systematic reviews regarding estimation of chronological age by development of the hand-wrist skeleton and third molar teeth, which also revealed high risks of age mimicry bias. Our findings raise concern about the study design and sample selection process used in age estimation research. When studying the direct probability of chronological age from stage we highlight the need to include populations with uniform distribution and sufficiently wide spectrum of chronological age, or using statistical techniques like transition analysis to overcome the age mimicry bias. Another suggested possibility is to observe the minimum age for each stage (minimum age concept), which is data that is not affected by the age mimicry bias. However, this cannot give an estimate of the individual’s probable age (which is centred on the mean value).
We are unable to assess the validity of using CT or MRI as a method to show the distribution of chronological age from each clavicular stage due to the high risk of age mimicry bias in the included studies and the observed high heterogeneity among included studies. None of the studies included a study population with a uniform distribution of chronological age, which leads to potential age mimicry bias of the results (the distribution of chronological age from each clavicular stage). Studies avoiding the age mimicry selection bias are warranted for a better description of the age distribution of the medial clavicular epiphysis development stages.