Age estimation in a sub-adult Western Australian population based on the analysis of the pelvic girdle and proximal femur

Highlights

• Contemporary radiographic proximal femur and pelvic girdle age estimation study.

• Population standards for Western Australia based on MDCT imaging.

• Predictive models with associated accuracy of ±3.29–3.80 years.

Abstract

The accurate and precise estimation of skeletal age by a forensic anthropologist is both a professional and judicial requirement. When unknown skeletal remains are referred to the anthropologist, the estimation of the requisite biological attributes (e.g., age and sex) should accordingly be based on the application of population-specific standards (statistical data). Deviations from the latter practice may result in reduced accuracy and compromised identification. Towards informing appropriate forensic practice, the aim of the present study is to develop statistically quantified age estimation models for a contemporary sub-adult Western Australian population based on the timing of fusion in the os coxa and proximal femur.

The study sample comprises 562 known age and sex MDCT scans (292 male, 270 female) representing contemporary Western Australian individuals birth through 30 years of age. Scans are viewed in multi-planar reconstructed (MPR) and/or three-dimensionally reconstructed images using OsiriX^®. Fusion status is scored according to a three-stage system across a total of nine sites in the proximal femur and os coxae. Observer accordance, bilateral asymmetry and sex-specific variation in fusion timing are statistically quantified. Polynomial regression is used to formulate age prediction models; transition analysis is used to calculate age ranges and determine the mean age for transition between an unfused, fusing and fused status.

Observer accordance in stage assignation is acceptable (ϰ = 0.79) and there is no significant bilateral variation in fusion timing. It was found that the mean age of commencement of fusion is significantly earlier (∼2 years) in females. The accuracy (SEE) of the polynomial models ranges from ±3.29 to ±3.80 years and the transition analysis shows that fusion of the iliac crest is delayed in comparison to other attributes of os coxa and proximal femur. Results of the present study confirm that the pelvic girdle and proximal femur can be used to accurately estimate chronological age in the study population.

Introduction

Forensic anthropology is the application of physical anthropological theory and methods in medico-legal investigations [11], [18], [30]. The analysis of skeletal remains is a part of the traditional role of the forensic anthropologist, although in recent times these methodologies have been applied to the assessment of age in the living [11], [18], [60]. Physical and forensic anthropologists will assess skeletal remains and formulate biological profiles that include estimates of age, sex, stature and ancestry. Additionally, observation and interpretation of trauma or pathology are often provided [11], [15], [18], [25], [53]. The latter assessment can then contribute towards the process of identifying an unknown individual, or aid in the resolution of other issues, such as age estimation in undocumented individuals (e.g., illegal immigrants) [11], [60].

The methodologies available to the forensic anthropologist for estimating the age of juvenile (e.g., non-adult) individuals are numerous. The modern approach to age estimation typically evaluates multiple skeletal traits and elements, thus attempting to minimize potential variability in any one single age indicator [11], [60]. The initial choice of which method(s) to apply, however, is guided to some degree by the evidentiary age of the remains (e.g., whether juvenile or adult) and which element(s) are available for analysis. The latter is compromised when the cause of death involves fragmentation or thermal destruction (e.g., explosive force) or an extended post-mortem interval involving multiple damaging taphonomic processes [47], [18]. The above affords strong justification, both methodologically and practically, for the need to have statistically quantified data relative to age metamorphosis of multiple regions within the human skeleton.

The forensic value of statistically quantified data pertaining to the morphology of the human skeleton relative to age is, however, somewhat diminished if it is ‘misapplied’. For example, skeletal age assessments performed by a forensic anthropologist should be based on the application of standards (statistical data) that are both sex and population specific. In relation to human development, it is well established that male and female growth trajectories are relatively divergent, especially in relation to skeletal and dental (permanent teeth) maturity, with girls on average ahead of boys from birth to adulthood [29]. Thus skeletal maturity data from one sex, applied to the opposite sex, will inadvertently lead to either an over- or an under-estimation of actual age [31], [49]. This, however, is problematic in a forensic context because it is unlikely that the sex of an individual referred for assessment is known, and in the juvenile skeleton (especially pre-pubertal) it is highly unlikely that this particular biological factor could be reliably estimated—a consequence of the relatively androgynous morphology of the human skeleton prior to adolescence [14], [20], [49]. One approach towards minimising error due to sex-specific variation in skeletal maturity is to provide predictive models based on the analysis of individual and pooled-sex samples. Clearly age estimation error will be higher in the pooled, relative to individual, sex models, but it is reasonable to posit that it would be lower relative to the misapplication of male data to a female individual.

There are, however, further confounding issues in relation to the estimation of skeletal age, from which actual (chronological) age, is inferred. The use of non-population specific data, whether geographically and/or temporally removed, introduces a further source of error of an unknown magnitude into age estimation [18], [35]. In some jurisdictions the latter cannot be avoided due to the lack of contemporary data required to model skeletal growth and development, in that instance the closest available standards are the most appropriate to apply. In recognition of such perceived limitations in forensic practice, there has been an increasing global trend towards devising anthropological standards for specific populations. Formulation of such standards, especially for contemporary adolescent individuals, is readily achievable based on the analysis of radiographic images drawn from medical databases. The quantification of the morphological appearance (size and shape) of ossification centres and their progressive fusion is one such example of what can be a relatively accurate and straightforward method of estimating skeletal age [65], albeit not without its own limitations (see Ref. [18]; and also Ref. [14]:11–13).

In recognition of the growing awareness of the need for population specific standards, the Australian forensic community has been proactive in exploring alternate approaches for acquiring contemporary skeletal data, which in the general absence of documented collections of skeletons, has involved the analysis of various medical imaging modalities that facilitate the virtual analysis of human anatomy (e.g., Refs. [21], [22], [23], [24], [40], [41], [44]). Volumetric scans (e.g., MDCT) afford higher resolution and thus more accurate visualisation of growth markers in the human skeleton relative to earlier studies limited to plain film radiography. The ability to view an ossification centre(s) of interest in cross-section and any orientation through multiple planes in axial and multiplanar reconstructed (MPR) images and/or 3D volumetric reconstructions, has been demonstrated to facilitate more accurate interpretations of fusion relative to the physical inspection of dry-bone specimens [56], [23], [24]; these modalities are also not affected by projection artifacts or superimposition of foreign material [16].

The fusion of ossification centres, which depending on the specific centre(s) involved, occurs anywhere from prenatal life (e.g., most centres of the sphenoid fuse prenatally) through adulthood (e.g., medial clavicle; basi-occipital synchondrosis) [55], [56], [23], [24], [14]. With specific reference to the pelvic girdle and the proximal femur, each os coxa comprises three primary (pubis, ilium and ischium) and several secondary ossification centres (e.g., acetabular, iliac crest and the ischial epiphyses – amongst others). The proximal femur comprises the single primary centre (diaphysis) and three (sometimes four) secondary ossification centres (head, greater and lesser trochanter) [14], [5]. The timing and sequence of osteogenesis, and fusion of the respective primary and secondary centres of ossification, can be used to estimate skeletal age (e.g., Refs. [10], [12], [32], [52], [58], [62]).

Based on the analysis of physical remains from the Lisbon documented skeletal collection, Cardoso [10] quantified the timing of epiphyseal union at the os coxa and lower limb (femur, tibia, fibula) in a total of 18 anatomical locations. It was demonstrated that age in adolescent and young adult individuals could be estimated to within 5–6 years. The data presented are sex-specific time-points (in years) for fusion according to the three-stage system (no union, partial union and completed union) of Johnston [33]. The study concludes that it is fundamental to estimate sex prior to age because it was demonstrated that epiphyseal union in males is delayed relative to females by 1–2 years, albeit in adolescent individuals this may not be practically feasible (see above). A similar observation regarding precocious female development was demonstrated in another Portuguese population based on the analysis of 64 ossification loci of individuals drawn from the Coimbra documented skeletal collection [12]. That study further demonstrated little bilateral asymmetry in epiphyseal fusion overall.

In the exploration of novel possible criterion for forensic age estimation in the living, Wittschieber et al. [63] demonstrated that based on large sample of conventional radiographic images (566 cases) and assessment according to a 10 stage (4 stage and 6 sub-stage) scoring system, complete union of the anterior iliac crest apophysis occurred at 16.42 and 17.90 years of age in boys and girls respectively. The authors highlight some issues relating to projection artefacts and the scoring system (feasibility of sub-stages), but establish the iliac crest apophysis as a reliable criterion for forensic age estimation. Wittschieber et al. [64] continued their investigation of the iliac crest by evaluating age assessment using magnetic resonance imaging (MRI). In their preliminary male sample (152 individuals) they conclude that MRI limits further classification into sub-stages, but overall that specific modality appears to be generally suitable for age diagnostics of living individuals. In their analysis of the iliac crest apophysis based on MDCT evaluation of living individuals, Ekizoglu et al. [16] made some methodological conclusions, recommending that where there is the opportunity of case specific retrospective pelvic CT assessment, it is a ‘useful method for researchers’ and that it provides “…the advantages of high resolutions and minimization of various sources of artifacts that could be seen with X-ray…” (p.1106).

In the present study the timing of ossification and fusion in the os coxae and proximal femur are statistically quantified relative to a contemporary Western Australian population based on the retrospective analysis of high resolution pelvic MDCT scans. Such analysis facilitates the forensic estimation of age based on the assessment of multiple skeletal elements (multifactorial approach), which in turn assists in formulating a more accurate biological profile overall [13], [36]. The primary objective of the present study is, therefore, to formulate a set of Western Australian forensic age estimation standards based on the statistical analysis of nine fusion sites in the os coxa and proximal femur.