ORIGINAL ARTICLE Inter-population variation of histomorphometric variables used in the estimation of age-at-death D. Botha1 & N. Lynnerup2 & M. Steyn1 Received: 29 November 2018 /Accepted: 19 March 2019 # Springer-Verlag GmbH Germany, part of Springer Nature 2019 Abstract Population variation of several microscopic structures used in age-at-death estimation was assessed for three different population samples. The aim of the study was to determine if the need exists for population-specific standards when dealing with individuals of African and European origin. A total sample 223 bone sections from the anterior cortex of the femur (n = 99 black South Africans, n = 94 white South Africans and n = 30 Danish individuals) were analysed using a stereological protocol. Variables assessed included the average number of osteons per grid area (OPD), osteon size and Haversian canal size. ANCOVA was employed for assessment of statistically significant differences. The results indicated that OPD differed significantly between the three groups, but that osteon size was similar for all individuals. Haversian canal size showed unpredictable changes with age and high levels of variation, making it unsuitable to use for age estimation as a single factor. As there are conflicting opinions in the literature on whether to use population-specific equations for the estimation of age-at-death or not, this paper provided additional insight into the use of specific variables and its related variation between groups. Keywords Femur . Stereology . OPD . Osteon size . Haversian canal size Introduction Several microscopic features of bone change with age (e.g., an increase in osteon number and decrease in secondary osteon size) and have been used to develop age estimation techniques from bone histology [1–5]. These age estimation techniques have been developed for a large number of different popula- tions, but with most studies taking only one specific population into account. For example, population-specific studies have been published for the Dutch [6], American white and black groups [1, 3], Danish [7, 8] and black South Africans [9], to mention a few. Bone mineral density and strength are established during childhood and early adulthood (up to the point where skeletal maturity is reached). Various studies have found that differences in bone turnover rates and/or strength between different ancestral and geographical groups exist from childhood and continue into adulthood [10–12]. Although skel- etal growth and establishment of bone strength are a complicat- ed process affected by many factors, a general trend indicating bone mass acquisition and loss may be observed. Once peak bonemass (skeletal maturity) is reached around 30 years of age, the almost-linear relationship between age and bone mineral density begins to weaken as the effect of variation sets in. Variation seen within bone mineral density and variables used for age estimation may be linked to factors influencing skeletal growth, maturity, and maintenance. One should keep in mind that variation in bone microstructure is often multi-factorial and that no single cause should be accredited as the sole factor for its existence. Differences, specifically between black and white groups, have been reported in bone mass/density [13–16], mus- cle mass and obesity [13, 17, 18], remodelling rates [19, 20], bone histomorphometry [21, 22], bone geometry [13, 23], calci- ummetabolism [10, 24, 25] and othermetabolic occurrences [20, 26, 27]. Some of the observed differences are due to heredity [28, 29], but may also be the result of environmental factors. Bone mass/density has been reported to be higher in black than in white individuals [13, 16, 30]. In some studies, muscle mass and obesity, like bone density, were also found to be higher * D. Botha deona.botha@gmail.com 1 Human Variation and Identification Research Unit, School of Anatomical Sciences, University of the Witwatersrand, 2nd Floor, WITS Health Sciences Building, 7 York Road, Parktown,, Johannesburg 2193, South Africa 2 Department of Forensic Pathology, University of Copenhagen, Copenhagen, Denmark International Journal of Legal Medicine https://doi.org/10.1007/s00414-019-02048-7 /Published online: 9 2019April (2020) 134:709–719 http://crossmark.crossref.org/dialog/?doi=10.1007/s00414-019-02048-7&domain=pdf mailto:deona.botha@gmail.com in black than in white individuals and increased levels of body weight are known to stimulate bone formation, resulting in higher bone mass and strength [14, 27]. Bone remodelling rates are lower in black than in white groups [19]. A study by Qui et al. [31] stated that black individuals have longer bone forma- tion periods and diminished osteoblast apoptosis. When developing methods for age-at-death estimation, it is thus essential to consider population variation, as the factors causing variation may not only affect the time of development of specific skeletal features (e.g., osteon population density and osteon size) but may also alter its size and frequency. Socio- economic status (SES) is frequently linked to many of the fac- tors altering bone remodelling and maintenance. For example, people who are considered to be of low SES may suffer from malnutrition and poor health care [32]. In South Africa, people of low SES regularly travel away from home to find work. Job opportunities available in such cases often entail hard physical labour [33]. As these job opportunities provide a low income, these individuals are prone to less than optimal nutrition. Although increased physical activity for long time periods is considered healthy and is known to promote bone strength [34, 35], extreme physical labour can have the opposite effect, es- pecially when combined with poor nutrition [36–38]. In contrast to South Africa, people living in European coun- tries such as Denmark are mostly classified as being of moder- ate to high SES, with relatively few socioeconomic inequalities [39]. SES encompasses much more than only availability of food or adequate nutrition. A positive relationship between ed- ucation, SES, health and nutrition exists [40–42], with educa- tion labelled as the most important social variable influencing nutrition and dietary habits [43]. Individuals of vastly different socioeconomic backgrounds can thus be expected to have dif- ferent bone quality due to a variety of factors. There are thus a number of factors related to SES (such as education, nutrition and health care), physical activity, environ- ment (e.g. exposure to sunshine) and genetics that may account for differences seen in bone microstructure between population groups. The aim of this study was to assess the extent of inter- population variation in terms of ancestry and geographical loca- tion between three different population groups/samples with re- gard to histomorphometric variables of the anterior cortex of the femur used in age estimation. A better understanding of how histological variables vs. age differ between population groups may assist in determining if population-specific formulae are needed and which variables are reliable indicators of age. Materials and methods Sample size, demographics and slide preparation Three study samples were selected for analysis. South African black and white samples were used for which age estimation standards based on histomorphometric variables of the anteri- or cortex of the femur have been established [44, 45]. These two groups reside in broadly the same geographical area (South Africa), but differ in terms of ancestry and probably also socioeconomic status. Black South African individuals are of BAfrican origin,^ while white South Africans are pre- dominantly of BEuropean origin.^ A Danish sample was employed as comparative group that is geographically isolated from the South African samples, but are also obviously of European origin. The black South African sample originated from a previous study by Keough et al. [44] selected from the Pretoria Bone Collection [46] and was used in this study to prevent re- sectioning of bone specimens. The sample comprised 60 males and 39 females (total n = 99). Ages ranged from 21 to 82 years. The majority of black South African remains in the collection originated from unclaimed bodies. If an individual dies in a public hospital and the body is not claimed within a certain period of time, it is transported to the relevant tertiary institution for medical training and research. As many of these unclaimed bodies include migrant workers and homeless in- dividuals, these skeletons thus largely represent individuals of low socioeconomic status with no records regarding their gen- eral health and well-being (43). Details on bone slide prepa- ration of this sample are described by Keough [9] and Keough et al. [44] and involved manual grinding of bone sections according to the protocol by Maat et al. [47]. The white South African sample, also selected form the Pretoria Bone Collection, comprised 50 males and 44 females (total n = 94) with ages ranging from 21 to 94 years. Many of the white remains in the collection originated form donated bodies and mostly represent older individuals, with very few young adults represented in the lower age categories [46]. They were mostly of lower or middle SES. Bone slides were prepared by making cross-sections with a thickness of 2– 4 mm just below the midshaft of the anterior cortex of the femur (to not affect bone measurements in future) using an EXAKT 312 pathology saw fitted with a diamond blade. The sections were manually grinded by hand to a thickness of ± 1 mm according to the Manual for the Preparation of Ground Sections for theMicroscopy of Bone Tissue [47]. This method does not ensure a uniform thickness throughout the section. Therefore, machine grinding was employed from this point forward to ensure a smooth and uniform surface area. Sections were mounted onto perspex slides and ground to a minimum thickness of 100 μm (thicknesses ranged between 100 and 200 μm, depending on the brittleness of the bone) with an EXAKT 400 CS surface grinder. All equipment used are situated in the Bone Research Laboratory at the University of the Witwatersrand. The Danish sample originated from a number of autopsies performed between 1988 and 1990 at the Department of Forensic Pathology, University of Copenhagen. These Int J Legal Med (2020) 134:709–719710 individuals represent a group of people of middle to higher SES, living in a region with relatively low UV light levels. Complete cross-sections from the femoral mid-shaft were made. Specimens were then boiled in water and detergent for a few hours, after which sections (± 100 μm)were cut with a Buehler Isomet water saw and mounted on glass slides with- out staining [7]. The total sample is shown in Table 1 and comprised 223 individuals. Their ages ranged from 21 to 94 years. Stereological analysis A MicroBrightField (MBF, Colchester, Vermont, USA) sys- tem with a three-plane motorised stage, Zeiss.Z2 Vario Axioimager was used for the scanning of sections. For quan- tification of secondary osteons and its related Haversian ca- nals, StereoInvestigator software (MBF, version 11.08.1; 64- bit) was used. The optical fractionator and nucleator probes were employed. Methodology and stereological criteria as described by Botha et al. [45] were used to analyse all sections. Stereology is preferred above traditional two-dimensional his- tological techniques because it allows for systematic random sampling, avoiding bias [45, 48]. Grid size was set at 1250 × 1250 μm2 (1562 500 μm2 ≈ 1.6 mm2) with an average of 55 counting frames covering the entire section. Variables assessed included average number of osteons per grid area (Avg_OPD), average osteon length (Avg_Ost_L), surface area (Avg_Ost_Ar) and volume (Avg_Ost_Vol), as well as average Haversian canal length (Avg_Hav_L), surface area (Avg_Hav_Ar) and volume (Avg_Hav_Vol). Statistical analysis ANCOVA (analysis of covariance), classified as a univariate general linear model, is similar to ANOVA (analysis of vari- ance) but is used to detect a difference in the means of three or more independent groups for datasets where it is necessary to control for a scale covariate. If a covariate is present (i.e. age) that could influence the dependent variable, it should be con- trolled for. ANCOVA is a method that combines ANOVAwith linear regression to point out group differences after account- ing for the effects of the covariate [49]. ANCOVA was employed to detect statistically significant differences in variable number and size between the samples, whilst controlling for age. To better understand where the differences occurred between the groups, pairwise compari- sons (post hoc tests) were used. Probabilities reported formed part of the ANCOVA analysis. The Bonferroni post hoc test was done to account for inflated error/results due to numerous tests being conducted. ANCOVAwas performed for age sub- groups 20–39 (young adults), 40–59 (middle aged adults) and 60+ (older individuals) years to assess the histomorphometric variation between groups across all life phases. To illustrate the relationship between the independent var- iables and age, least squares regression was performed. Figures depicting the sample regression lines for each variable are given. All analyses were performed using SPSS (v. 24). Results The descriptive statistics for all variables are summarised in Table 2. Regression slopes are illustrated by Figs. 1, 2, 3, 4, 5, 6 and 7. Regression analysis indicated that a significant cor- relation exists between osteon number and size and age for all groups, but not between age and Haversian canal size. In order to perform ANCOVA, preliminary checks (as- sumption testing) should be carried out to ensure that none of the covariates are highly correlated with one another, resid- uals are normally distributed, no outliers are present, homo- geneity of regression slopes and equality of variance exist. Covariates were not highly correlated (all correlation values were smaller than 0.7). Kolmogorov-Smirnov and Shapiro- Wilk tests were non-significant, indicating normality. No out- liers were detected (tested by means of Cook’s distance). Levene’s test was non-significant (p > 0.05) for all analyses, representing homogeneity of variance. Amodel for ANCOVA was performed to test the interaction between the independent variables and age. Interactions were found to be non- significant for the 20–39, 40–59 and 60+ age categories. There was thus no violation of the assumptions for ANCOVA the data were found to be suitable for its employment. To establish if statistically significant differences exist be- tween the three samples, ANCOVA (Table 3) and pairwise Table 1 Sample size and age distribution Age group Age cohort SA black SAwhite Danish white n Mean SD n Mean SD n Mean SD Young adults 20–39 25 30.4 6.7 4 29.5 9.3 16 27.8 5.3 Middle aged adults 40–59 38 48.9 5.4 21 51.5 5.2 7 46.6 5.8 Older individuals 60+ 36 69.1 7.0 69 74.3 9.0 7 78.6 8.0 Total 99 51.6 16.4 94 67.3 14.9 30 44.1 21.8 Int J Legal Med (2020) 134:709–719 711 comparisons (Table 4) were examined. A significant differ- ence (p < 0.001) for Avg_OPD was observed between the three groups for all age classifications (subgroups 20–39, 40–59 and 60+ years). Mean values (Table 2) and Fig. 1 indi- cate that the black South African population presented with the least number of secondary osteons vs. age (± 17), followed by the white South African group (± 26). The Danish sample had the largest number of osteons vs. age (± 28). Pairwise comparisons suggested that significant differences exist be- tween all age groups for this variable, except for SAB/SAW comparison in the young adult age group. The small sample size of the SAW group in the 20–39 years age cohort may account for this outcome, as there may be too few individuals to reach statistical significance. No significant differences were observed in osteon size (Avg_Ost_L, Avg_Ost_Ar and Avg_Ost_Vol; Figs. 2, 3 and 4) between the samples for all age groups, except in average osteon volume in the young adult age group (p < 0.01). Pairwise comparisons suggested that a significant difference in Avg_Ost_Vol is found between SAWandDAN individuals, with white South Africans having larger average volumes than Danish individuals. Quantification of the Haversian canal (Avg_Hav_L, Avg_Hav_Ar and Avg_Hav_Vol; Figs. 5, 6 and 7) showed significant differences (p < 0.001) between the groups for the young adult age cohort. Pairwise comparisons showed that these differences were present between black South Africans and Danish individuals, with black South African individual Haversian canal sizes being larger on average than that of the Danish. However, no significant difference was present for Haversian canal size in the middle and older aged groups, with the exception of Avg_Hav_L in middle aged individuals (p < 0.01). This difference was observed between SAW and DAN samples, with white South Africans showing larger Haversian canal lengths on average. SAB South African black, SAW South African white, DAN Danish. Discussion Previous studies have demonstrated a relationship between histological variables and age [2, 3, 6, 8, 44, 50] and reported on inter- and intra-observer error [6, 51], as well as cross- sectional variability [7] related to these methods. However, inter-population/group variation has not received much atten- tion. Some studies stated that differences between black and white individuals are present on a microstructural level [12, 19, 21], but these have not been explored in detail. Although these studies deal with bone microstructure differences be- tween black and white groups, literature on differences be- tween groups of the same ancestry, but geographically isolated Table 2 Descriptive statistics of single variables for black South African, white South African and Danish samples Variable Group Mean SD SE 95% CI for mean Lower Upper OPD SAB 16.88 4.49 0.45 15.98 17.78 SAW 25.78 6.27 0.65 24.50 27.07 DAN 28.32 7.78 1.42 25.42 31.23 Avg_Ost_L SAB 108.33 11.19 1.12 106.09 110.56 SAW 106.16 14.22 1.47 103.70 109.52 DAN 108.16 12.01 2.19 103.67 112.65 Avg_Hav_L SAB 31.85 4.30 0.43 30.99 32.71 SAW 30.32 6.17 0.64 28.96 31.49 DAN 27.50 3.97 0.72 26.02 28.98 Avg_Ost_Ar SAB 40,293.74 8295.19 833.69 38,639.29 41,948.18 SAW 38,743.62 10,648.98 1098.36 36,562.49 40,924.74 DAN 40,001.23 8609.64 1571.89 36,786.34 43,216.12 Avg_Hav_Ar SAB 3832.51 1111.21 111.98 3610.88 4054.14 SAW 3488.05 1638.89 169.04 3152.37 3823.73 DAN 2800.45 820.09 149.73 2494.22 3106.68 Avg_Ost_Vol SAB 6,958,100.87 2,188,730.01 219,975.64 6,521,566.37 7,394,635.38 SAW 6,494,934.47 2,799,871.62 288,784.71 5,921,465.27 7,068,402.66 DAN 6,828,080.53 2,130,795.58 389,028.27 6,032,428.39 7,623,732.68 Avg_Hav_Vol SAB 235,831.84 101,061.07 10,157.02 215,675.56 255,988.12 SAW 210,775.74 189,757.06 19,571.95 171,909.73 249,641.74 DAN 143,285.63 64,189.14 11,719.28 119,317.01 167,254.24 Int J Legal Med (2020) 134:709–719712 is almost non-existent. The intention of this study was to ex- plore inter-population variation of histomorphometric variables of cortical bone used in age-at-death estimation be- tween groups in terms of ancestry and geographical location. Fig. 1 Regression between average number of osteons per grid area and age Fig. 2 Regression between average osteon length and age Int J Legal Med (2020) 134:709–719 713 When considering the average number of osteons (Avg_OPD), significant differences were observed in all of the different age cohorts. Black South Africans presented with the least number of osteons per grid area, followed white Fig. 4 Regression between average osteon volume and age Fig. 3 Regression between average osteon surface area and age Int J Legal Med (2020) 134:709–719714 South Africans that had significantly higher OPD. Danish individuals showed the highest number of osteons per grid area. This outcome is supported by the results of other studies [19, 20] stating that bone remodelling rates are higher in white Fig. 5 Regression between average Haversian canal length and age Fig. 6 Regression between average Haversian canal surface area and age Int J Legal Med (2020) 134:709–719 715 than in black individuals. The results of this study showed that both white/European samples had a higher number of osteons than the black sample, but also that there is a difference be- tween European samples that are geographically unrelated. Whether this is related to climatic conditions and in particular the exposure to ultraviolet light remains to be explored. Osteon population density has been widely used in age estimation studies as it correlates well with age. It is also the most Buser-friendly^ variable, as counting may be performed fairly easily. Apart from the advantages of using OPD for age estimation, the results of this study showed that significant differences are present between the study samples, which have implications when it comes to the use of general age estima- tion regression equations. This suggests that population stan- dards may be necessary for the employment of OPD as a single variable. Alternatively, the reference population (if be- ing applied to more than one ancestral group) for the construc- tion of a regression equation should include a sub-sample of the population group of the individual in question. Osteon size showed almost no significant differences be- tween the sample for all age categories. Some variation (al- though not statistically significant) is present in the young and middle-aged adults, but once older age is reached, osteon size is very similar for all samples. Secondary osteon size, which also correlates relatively well with age [45, 52], appears to be a more dependable variable for age estimation, specifically where ancestry is unknown. Significant differences were seen between the samples for Haversian canal size in younger adults, but as age advances the size of the canals became more uniform between the dif- ferent groups (refer to Figs. 5, 6 and 7 and Table 4). Haversian canal size has been reported to show a low correlation with age and due to its considerable variation, specifically in the younger ages, should not be used for age estimation as a single factor. As Haversian canals are often irregular and highly inter-connected [53], its shape and size vary much more than that of the external surface of the Haversian system (osteon). The outcome of this study is similar to other studies on age estimation [9, 44, 45] that also found the Haversian canal to show unpredictable changes with age. Pratte and Pfeiffer [54] tested age prediction equations de- veloped on a sample of European descent and found that it performed poorly when used to predict age-at-death in a South African sample comprising of white, black and coloured indi- viduals. Given these results, as well as evidence for population variation related to bone turn-over rates, density and strength [13, 21, 27] between different population groups, it is expect- ed that population-specific regression formulae for age-at- death estimation provide higher accuracy and precision. However, Pfeiffer et al. [52] reported that equations developed based on an American sample of white, black and ethnicity unknown individuals [50] may be applied to a South African sample of white, black and coloured individuals. It was also found that population-specific standards did not perform Fig. 7 Correlation between average Haversian canal volume and age Int J Legal Med (2020) 134:709–719716 better than the Bethnicity unknown^ equation. The outcome of this study agrees with Pratte and Pfeiffer [54] that European populations are dissimilar from white South Africans. The white South African population, although having European ancestry, possibly has genetic admixture with other popula- tions. Also, adaptation to the environment may have contrib- uted in variation seen between them and the Danish. It is possible that South African groups and American groups (such as the reference sample by Cho et al. [50] and applied by Pfeiffer et al. [52]) may be more similar to one another, but further investigation is needed to ascertain this possibility. Although bone loss over time is affected by environmental factors, genetic make-up largely contributes to skeletal varia- tion related to bone metabolism (bone formation and loss). Genetic studies have shown that skeletal growth and density acquired during childhood has a strong genetic foundation, and that the subsequent development and timing of peak bone mass is a result of specific genetic factors [27]. Black individ- uals tend to reach a higher peak bone mass than white indi- viduals [55, 56], and often present with higher bone mineral density as adults [57, 58], which assist in maintaining higher bone mass throughout life. On a molecular level, differences between populations have been reported that relate to calcium metabolism [10, 24], sex hormone levels [59], vitamin D levels [60], sun ex- posure and vitamin D [61] and vitamin D receptor genes [62] and body composition [14, 63, 64]. These studies found a higher calcium retention rate, bone formation rate and oestrogen levels in black individuals. All these factors were linked to a greater bone mass and maintenance in black adult individuals compared to their white counterparts. Many of the factors, documented to play a role in bone microstructure and variation thereof, are linked to environ- mental circumstances (e.g. climate), physical activity and Table 3 Summary of ANCOVA results Variable F-ratio Significance (p value) Ages 20–39 F(2,41) Avg_OPD 34.807 0.000** Avg_Ost_L 4.572 0.016 Avg_Ost_Ar 5.104 0.010 Avg_Ost_Vol 5.785 0.006* Avg_Hav_L 9.320 0.000** Avg_Hav_Ar 6.142 0.005* Avg_Hav_Vol 9.469 0.000** Ages 40–59 F(2,62) Avg_OPD 49.656 0.000** Avg_Ost_L 2.940 0.060 Avg_Ost_Ar 2.814 0.068 Avg_Ost_Vol 2.701 0.075 Avg_Hav_L 5.583 0.006* Avg_Hav_Ar 3.888 0.026 Avg_Hav_Vol 3.063 0.054 Ages 60+ F(2,108) Avg_OPD 27.993 0.000** Avg_Ost_L 0.151 0.860 Avg_Ost_Ar 0.074 0.929 Avg_Ost_Vol 0.018 0.982 Avg_Hav_L 0.823 0.442 Avg_Hav_Ar 1.608 0.205 Avg_Hav_Vol 1.534 0.220 **p < 0.001 *p < 0.01 Table 4 Pairwise comparisons (significance; p values) between groups for all variables assessed Avg_OPD Avg_Ost_L Avg_Ost_Ar Avg_Ost_Vol Avg_Hav_L Avg_Hav_Ar Avg_Hav_Vol Young adults (20–39 years) SAB/SAW 1.000 0.015 0.010 0.010 1.000 1.000 0.914 SAB/DAN 0.000** 1.000 1.000 1.000 0.000** 0.004* 0.000** SAW/DAN 0.001** 0.024 0.015 0.005* 0.166 0.337 0.398 Middle aged adults (40–59 years) SAB/SAW 0.000** 0.252 0.317 0.397 1.000 1.000 1.000 SAB/DAN 0.000** 0.606 0.544 0.487 0.011 0.038 0.105 SAW/DAN 0.000** 0.080 0.083 0.086 0.005* 0.026 0.051 Older individuals (60+ years) SAB/SAW 0.000** 1.000 1.000 1.000 1.000 1.000 1.000 SAB/DAN 0.000** 1.000 1.000 1.000 1.000 1.000 1.000 SAW/DAN 0.001** 1.000 1.000 1.000 1.000 1.000 1.000 **p < 0.001 *p < 0.01 Int J Legal Med (2020) 134:709–719 717 socio-economic status (e.g. nutrition and health care). The exact influences of everyday life on the health of bone are complex. As bone loss throughout life is multifactorial, all these aspects most likely play a part in the variation we ob- serve in the studied groups. It is clear that variation between black and white, and individuals of shared ancestry but geo- graphically isolated, is associated with various factors. Most of these factors are strongly related to a genetic basis that has been shaped by long-term environmental influences. However, due to a lack of information on general well-being and health on many of the individuals in the study samples, the exact nature of these factors and how it contributed to the variation seen is unclear. Conclusion Overall, the highest variation in the studied variables was ob- served in the younger age groups. The average number of osteons showed significant differences between the three sam- ples across all age categories. The opposite was true for osteon size, with population samples showing similar osteon dimen- sions over age. Haversian canal size showed variation in the younger age groups, but not in older individuals. As OPD and osteon size correlates well with age, it is recommended that these variables (rather than Haversian canal size) be employed for age estimation using regression analysis. It is important to consider population differences in terms of OPD before using this variable in age estimation. Osteon size appears to be a more generally applicable variable to employ as it showed the least amount of inter-population var- iation. In cases where the ancestry is unknown, it is suggested that osteon size, rather than OPD, be used for estimating age. As there are conflicting views on whether age estimation equations should be generally applied, further studies need to be done on inter-population variation and the use of specific histological age estimation standards. Acknowledgments The authors would like to thank the curators of the collections for granting permission to access the bone samples, Prof. Ripamonti for the use of the Bone Research Laboratory (WITS), Prof. Manger for the use of the stereology system and Dr. Bhagwandin for technical assistance in producing the images for analysis. Funding information Funding for this project was provided by the NRF. References 1. Thompson DD (1979) The core technique in the determination of age at death of skeletons. J Forensic Sci 24:902–915 2. Ericksen MF, Stix AI (1991) Histologic examination of age of the first African Baptist church adults. Am J Phys Anthropol 85:247– 252 3. Stout SD, Paine RR (1992) Histological age estimation using rib and clavicle. Am J Phys Anthropol 87:111–115 4 . E r ik sen EF, Axe l rod DW, Mel sen F (1994 ) Bone Histomorphometry. Raven Press, New York 5. Crowder C, Heinrich J, Dominquez V (2009) Histological age es- timation. In: Blau S, Ubelaker D (eds) Handbook of forensic an- thropology and archaeology. Left Coast Press, Walnut Creek, CA, pp 222–235 6. Maat GJR,Maes A, Aarents J, Nagelkerke NJD (2006) Histological age prediction from the femur in a contemporary Dutch sample: a decrease of nonremodeled bone in the anterior cortex. J Forensic Sci 51:230–237 7. Villa C, Lynnerup N (2009) A stereological analysis of the cross- sectional variability of the femoral osteon population. Am J Phys Anthropol 142:491–496 8. Lynneryp N, Frohlich B, Thomsen JL (2006) Assessment of age at death by microscopy: unbiased quantification of secondary osteons in femoral cross sections. Forensic Sci Int 159S:S100–S103 9. Keough N (2007) Estimation of age at death from the microscopic structure of the femur. MSc Dissertation, University of Pretoria 10. Bryant RJ, Wastney ME, Martin RR et al (2003) Racial differences in bone turnover and calcium metabolism in adolescent females. J Clin Endocrinol Metab 88:1043–1047 11. Wetzsteon RJ, Hughes JM, Kaufman BC, Vazquez G, Stoffregen TA, Stovitz SD, Petit MA (2009) Ethnic differences in bone geom- etry and strength are apparent in childhood. Bone 44:970–975 12. Pollock NK, Laing EM, Taylor RG, Baile CA, Hamrick MW, Hall DB, Lewis RD (2011) Comparisons of trabecular and cortical bone in late adolescent black and white females. J BoneMiner Metab 29: 44–53 13. Nelson DA, Jacobsen G, Barondess DA, Parfitt AM (1995) Ethnic differences in regional bone density, hip axis length, and lifestyle variables among health black and white men. J BoneMiner Res 10: 782–787 14. Aloia JF, Vaswani A, Ma R, Flaster E (1997) Comparison of body composition in black and white premenopausal women. J Lab Clin Med 129:294–299 15. Gilsanz V, Roe TF, Mora S, Costin G, Goodman WG (1991) Changes in vertebral bone density in black and white girls during childhood and puberty. N Engl J Med 325:1597–1600 16. Hochberg MC (2007) Racial differences in bone strength. Trans Am Clin Climatol Assoc 118:305–315 17. Cohn SH, Abesamis C, Zanzi I et al (1977) Body elemental com- position: comparison between black and white adults. Am J Phys 232:E419–E422 18. Wang M, Bachrach L, Loan V et al (2005) The relative contribu- tions of lean tissue mass and fat mass to bone density in young women. Bone 37:474–481 19. Weinstein R, Bell N (1988) Diminished rates of bone formation in normal black adults. N Engl J Med 319:1698–1701 20. Kleerekoper M, Nelson DA, Peterons EL et al (1994) Reference data for bone mass, calciotropic hormones and biochemical markers of bone remodelling in older (55-75) postmenopausal white and black women. J Bone Miner Res 9:1267–1276 21. Schnitzler CM, Pettifor JM, Mesquita JM et al (1990) Histomorphometry of the iliac crest bone in 346 normal black and white south African adults. Bone Miner 10:18–199 22. Schni tz ler CM, Mesquita JM (2006) Cort ica l bone histomorphometry of the iliac crest in normal black and white south African adults. Calcif Tissue Int 79:373–382 23. Mikhail MB, Vaswani AN, Aloia JF (1996) Racial differences in femoral dimensions and their relationship to hip fracture. Osteoporos Int 6:22–24 24. BraunM, Palacios C, Wigertz K, Jackman LA, Bryant RJ, McCabe LD, Martin BR, McCabe GP, Peacock M, Weaver CM (2007) Racial differences in skeletal calcium retention in adolescent girls Int J Legal Med (2020) 134:709–719718 with varied controlled calcium intakes. Am J Clin Nutr 85:1657– 1663 25. Heaney RP (2002) Ethnicity, bone status and the calcium require- ment. Nutr Res 22:153–178 26. Aloia JF, Talwar SA, Pollack S, Yeh J (2005) A randomized con- trolled trial of vitamin D3 supplementation in African American women. Arch Intern Med 165:1618–1623 27. Aloia JF (2008) African Americans, 25-hydroxyvitamin D and os- teoporosis: a paradox. Am J Clin Nutr 88(suppl):545–550 28. Fleet JC, Harris SS, Wood RJ, Dawson-Hughes B (1995) The BsmI vitamin D receptor restriction fragment length polymorphism (BB) predicts low bone density in premenopausal black and white wom- en. J Bone Miner Res 10:985–990 29. Peacock M, Turner CH, Econs MJ, Foroud T (2002) Genetics of osteoporosis. Endocr Rev 23:303–326 30. Daniels ED, Pettifor JM, Schnitzler CM, Russell SW, Patel DN (1995) Ethnic differences in bone density in female south African nurses. J Bone Miner Res 10:359–367 31. Qiu S, Rao DS, Palnitkar S, Parfitt AM (2006) Differences in oste- ocyte and lacunar density between black and white American wom- en. Bone 38:130–135 32. Williams DR (1999) Race, socioeconomic status and health: the added effects of racism and discrimination. Ann N Y Acad Sci 896:173–188 33. Steyn M, Meiring JJ, Nienaber WC (1997) Forensic anthropology in South Africa: a profile of cases from 1993 to 1995 at the Department of Anatomy, University of Pretoria. S Afr J Ethnol 20:23–26 34. Mundy GR (1999) Bone remodelling. In: Favus MJ (ed) Primer on the metabolic bone diseases and disorders of mineral metabolism. Lippencott. Williams and Wilkins, Philadelphia, pp 30–38 35. Kohrt WM, Bloomfield SA, Little KD et al (2004) Physical activity and bone health. Am Coll Sports Med:1985–1996 36. Lanyon LE, Rubin CT, Baust G (1986) Modulation of bone loss during calcium insufficiency by controlled dynamic loading. Calcif Tissue Int 38:209–216 37. Palacios C (2006) The role of nutrients in bone health, from a to Z. Crit Rev Food Sci Nutr 46:621–628 38. Cashman KD (2007) Diet, nutrition and bone health. J Nutr 137: 2507S–2512S 39. Ullits LR, Ejlskov L, Mortensen RN, Hansen SM, Kræmer SRJ, Vardinghus-Nielsen H, Fonager K, Bøggild H, Torp-Pedersen C, Overgaard C (2015) Socioeconomic inequality and mortality – a regional Danish cohort study. BMC Public Health 15:490 40. Smith AM, Baghurst KI (1992) Public health implications of die- tary differences between social status and occupational category groups. J Epidemiol Community Health 46:409–416 41. Smith GD, Brunner E (1997) Socio-economic differentials in health: the role of nutrition. Proc Nutr Soc 56:75–90 42. Mackenbach JP, Kunst AE, Cavelaar AEJM et al (1997) Socioeconomic inequalities in morbidity and mortality in western Europe. Lancet 349:1655–1659 43. Groth MV, Fagt S, Brøndsted L (2001) Original communication: social determinants of dietary habits in Denmark. Eur J Clin Nutr 55:959–966 44. Keough N, L’Abbé EN, Steyn M (2009) The evaluation of age- related histomorphometric variables in a cadaver sample of lower socioeconomic status: implications for estimation of age at death. Forensic Sci Int 191:114.e1–114.e6 45. Botha D, Bhagwandin A, Lynnerup N, Steyn M (2018) The use of stereological methods in the histomorphometric assessment of bone for age-at-death estimation. Forensic Sci Int 290:353.e1–353.e7 46. L’Abbé EN, Loots M, Meiring JH (2005) The Pretoria bone collec- tion: a modern south African skeletal sample. HOMO – J Comp Hum Biol 56:197–205 47. Maat GJR, Van den Bos RPM, Aarents MJ (2002) Manual for the Preparation of Ground Sections for theMicroscopy of Bone Tissue. Published by Barge’s Anthropologica Nr. 7. Leiden University Medical Centre, Leiden 48. Howard CV, Reed MG (2010) Unbiased stereology, 2nd edn. QTP Publications, Liverpool 49. Huitema BE (2011) The analysis of covariance and alternatives. Wiley, Hobokon 50. Cho H, Stout SD, Madsen RW, Streeter MA (2002) Population- specific histological age-estimating method: a model for known African-American and European-American skeletal remains. J Forensic Sci 47:12–18 51. Lynnerup N, Thomsen JL, Frohlich B (1998) Intra- and inter- observer variation in histological criteria use in age at death determi- nation based on femoral cortical bone. Forensic Sci Int 91:219–230 52. Pfeiffer S, Heinrich J, Beresheim A, Alblas M (2016) Cortical bone histomorphometry of known-age skeletons from the Kirsten collec- tion, Stellenbosch University, South Africa. Am J Phys Anthropol 160:137–147 53. Kim J-N, Lee J-Y, Shin K-J, Gil YC, Koh KS, Song WC (2015) Haversian system of compact bone and comparison between end- osteal and periosteal sides using three-dimensional reconstruction in rat. Anat Cell Biol 48:258–261 54. Pratte DG, Pfeiffer S (1999) Histological age estimation of a cadaveral sample of diverse origins. Can Soc Forensic Sci J 32:155–167 55. Berenson AB, Rahman M, Wilkinson G (2009) Racial differences in the correlates of bone mineral content/density and age at peak among reproductive-aged women. Osteoporos Int 20:1439–1449 56. Hui SL, Perkins AJ, Harezlak J, Peacock M, McClintock CL, Johnston CC Jr (2010) Velocities of bone mineral accrual in black and white American children. J Bone Miner Res 25:1527–1535 57. Ettinger B, Sidney S, Cummings SR et al (1997) Racial differences in bone density between young adult black and white subjects per- sist after adjustment for anthropometric, lifestyle and biochemical differences. J Clin Endocrinol Metab 82:429–434 58. Wilkin LD, Jackson MC, Sims TD, Haddock BL (2010) Racial/ ethnic differences in bone mineral density of young adults. Int J Exerc Sci 3:197–205 59. Perry HM III, HorowitzM,Morley JE et al (1996) Ageing and bone metabolism in African American and Caucasian women. J Clin Endocrinol Metab 81:1108–1117 60. Armas LA, Dowell S, Akhter M et al (2007) Ultraviolet-B radiation increases serum 25-hydroxyvitamin D levels: the effect of UVB dose and skin color. J Am Acad Dermatol 57:588–593 61. Gloth FM 3rd, Alam W, Hollis B (1999) Vitamin D vs broad spec- trum phototherapy in the treatment of seasonal affective disorder. J Nutr Health Aging 3:5–7 62. Harris SS, Eccleshall TR, Gross C et al (1997) The vitamin D receptor start codon polymorphism (FokI) and bonemineral density in premenopausal American black and white women. J BoneMiner Res 12:1043–1048 63. Aloia JF, Vaswani A, Ma R, Flaster E (1996) Body composition in normal black women: the four-compartment model. J Clin Endocrinol Metab 81:2363–2369 64. Shaffer JR, Kammerer CM, Reich D, McDonald G, Patterson N, Goodpaster B, Bauer DC, Li J, Newman AB, Cauley JA, Harris TB, Tylavsky F, Ferrell RE, Zmuda JM, for the Health ABC study (2007) Genetic markers for ancestry are correlated with body composition traits in older African Americans. Osteoporos Int 18:733–741 Publisher’s note Springer Nature remains neutral with regard to jurisdiction- al claims in published maps and institutional affiliations. Int J Legal Med (2020) 134:709–719 719 Inter-population variation of histomorphometric variables used in the estimation of age-at-death Abstract Introduction Materials and methods Sample size, demographics and slide preparation Stereological analysis Statistical analysis Results Discussion Conclusion References