Histological age-at-death estimation in white South Africans using stereology

Various methods are available for estimating age from skeletal remains, amongst them the use of histomorphometry. It is generally argued that age estimation standards are population specific, but this in itself creates problems as the reference samples used are often not large enough and/or lack substantial representation of all age cohorts. Traditional age methods have been shown to suffer from problems such as age mimicry. This paper aims at establishing histological age-at-death standards for the white South African population by supplementing the available sample (lacking an adequate number of young adults) with another sample of European descent to avoid over-estimation of age in younger individuals caused by age mimicry. Bone microstructures related to the number of osteons and fragments, osteon size and Haversian canal size that change with advancing age were used for the development of regression formulae. A histomorphometric assessment of the anterior cortex of the femur was done using stereology for the estimation of age at death. All sections were analysed using the optical fractionator and nucleator probes. A sample of 94 bone sections (n = 50 male, n = 44 females) of white South African individuals were used. A sample of Danish individuals (n = 14 males, n = 16 females) was combined with the South African sample to create a normal age distribution for the reference sample. Single and multiple regression equations were developed after randomly selecting a hold-out sample (n = 14) for validation. Osteon size (average length, surface area and volume) showed the highest correlation with age, followed by the number of osteons and fragments per grid area. Haversian canal size showed inconsistent changes with advancing age. Using the regression equations, predicted ages were obtained for the 14 individuals. RMSE values ranged between 14 and 17 years, which we deemed acceptable.