Re-assessment of South African juvenile facial soft tissue thickness data for craniofacial approximation: A comparative analysis using central tendency statistics

Abstract

Facial soft tissue thickness (FSTT) data form the basis of craniofacial identification methods such as facial approximation in cases where unknown skeletal remains lack unique identifiers such as fingerprints, DNA and dental records. Appropriate FSTT data are said to be required to produce accurate facial approximations that may be recognised by relatives. This view led to a vast number of studies considering subdivisions of FSTT data according to ancestry, age and sex. The paucity of South African juvenile FSTT standards of particular age groups, sex and ancestry is therefore problematic as “accurate” facial approximations cannot be produced. However, the use of pooled datasets and central tendency statistics offers a unique opportunity to circumvent the problem of small or absent FSTT datasets. The aim of this study was to use central tendency statistics of previously published South African data in order to assess whether it is necessary to subdivide FSTT datasets into different subgroups. In addition, a meta-analysis using central tendency statistics of 11 datasets within the C-table repository using the free open source TDStats programme (available through CRANIOFACIALidentification.com) for midfacial landmarks was performed. These datasets comprised of raw juvenile and adult FSTT data gathered from 1895 to 2015 as measured by a variety of methods Scatter plots showed that FSTT correlation with age is rather weak, while Kernel density plots of FSTT by sex and landmark indicated no difference between South African juvenile males and females. In order to test the practical application of FSTT data, two facial approximations were constructed — one based on the shorth from South African data and C-tables and one based on an American dataset. When comparing the two facial approximations based on different datasets, geometric deviation indicated differences at midline and bilateral landmarks, but the visual presentation of the facial approximations was similar. Therefore it is suggested that differences of less than 3mm at any landmark do not result in profound practical differences in the juvenile face. Subcategorizing juvenile data is not necessary as the same result can be achieved by weighted means as presented in the sub-adult C-tables.