International Journal of Legal Medicine (2025) 139:847–862

[Link]

ORIGINAL ARTICLE

Application of DSP2 for biological sex estimation in a Spanish sample:

analysis based on sex and side
Marta San-Millán1 · Varsha Warrier2 · Anna Carrera1 · Francisco Reina1

Received: 20 September 2024 / Accepted: 21 October 2024 / Published online: 15 November 2024
© The Author(s) 2024

Abstract
Applying existing sexing methodologies to different populations, and reporting these findings is important to enhance
their applicability and accuracy in real cases across the world. DSP was published in 2005 (Murail et al., 2005) and
updated as a DSP2 in 2017 (Brůžek et al., 2017) based on a database of 10 pelvic measurements from 2040 individuals
worldwide. These tools have been applied subsequently to various populations, however, its applicability to a dry Spanish
population is lacking. 303 hipbones belonging to 157 individuals from the School of Legal Medicine from the University
Complutense of Madrid (Spain), of which 140 individuals were documented, were analyzed to investigate the reliability,
applicability and accuracy of the DSP2 sex estimation methodology, examining side and sex-based potential differences
for the first time. In most of the DSP variables, intra-rater reliability showed excellent results and % applicability was
higher than 85.0%. Overall % accuracy was higher than 94.0% regardless of the number or discriminant power of the uti-
lized DSP variables. However, % sexing decreased when less variables or less discriminant ones were used for estimations,
reaching 45.51% (left) and 43.31% (right). Regarding sexual dimorphism, females’ results of % applicability, % sexing
and % accuracy were higher compared to males. In addition, left os coxae achieved better outcomes (aforementioned per-
centages) in most of the cases in the sex-pooled sample. Decreasing the mandatory posterior probability by 10% yielded
an increase in the % sexing but reduced % accuracy, and thus, does not seem to enhance the approach’s performance. The
present study validates the applicability and reliability of DSP for sexing a Spanish population. Future investigations will
attempt to assess its applicability within virtual anthropology.

Keywords Forensic anthropology · Sex estimation · Pelvis · Innominate · Os coxa · DSP · DSP2

Introduction sex estimation across literature [2–5]. Between these two

markers, the direct relationship of the pelvis with biologi-
Estimating the sex of adult skeletal remains is a critical cal sex, attributable to its active role in parturition, renders
step in the reconstruction of anthropological profiles, and pelvic characteristics more accurate for sexing, in com-
by extension, human identification, particularly in medico- parison to cranial parameters [6]. Although some enriching
legal or osteo-archaeological contexts. Despite numerous approaches have exhaustively described sexing variables
anatomical elements within the skeletal framework that using graphic explanations and introduced greater objec-
have been utilised for deriving sexing models [1], the skull tivity through score-based approaches [7], these visual or
and pelvis have demonstrated a greater predominance for morphognostic techniques, compared to metric ones, con-
tinue to remain problematic. Visual techniques, in addi-
tion to generally being highly subjective, which results in
higher intra-inter-observer errors, warrant prior experience
Marta San-Millán to be able to apply them accurately [7]. Morphometric tech-
[Link]@[Link]
niques, on the other hand, are more objective, repeatable,
1
Medical Sciences Department, Clinical Anatomy, and verifiable, which have led researchers to prefer them
Embryology and Neuroscience Research Group (NEOMA), over the former [8, 9].
Faculty of Medicine, University of Girona, Girona, Spain Murail and colleagues developed a new morphomet-
2
School of Sciences, College of Science and Engineering, ric technique (DSP: Diagnose Sexuelle Probabiliste = Proba-
University of Derby, Derby, UK

13
848 International Journal of Legal Medicine (2025) 139:847–862

bilistic Sex Diagnosis) in 2005 [10] to estimate the sex of sexing and % accuracy of DSP2 among sexes and sides; (3)
skeletal remains using ten pelvic measurements and obtained investigate how the combination of utilized DSP2 variables
an overall accuracy value of 99.63%. This methodology was and the posterior probability can change the accuracy out-
originally created based on a worldwide reference sample comes and (4) compare the sexing and accuracy percentages
of 2040 individuals from Europe (France, England, Portu- reported across different populations.
gal and Lithuania), Africa (South Africa), North America
(United States of America) and Asia (Thailand). In addition
to this geographical diversity, different ethnic groups (Zulu, Materials and methods
Soto, Afrikaner, African American, and European Ameri-
can) and temporal periods (from 18th to late 20th centu- Sample
ries) were also considered within the analyses to reinforce
its potential applicability in miscellaneous populations. In The sample used for this investigation is derived from
2017, a freely available updated software, the DSP2, was the modern documented skeletal collection housed in the
created and validated using two new samples from the Max- School of Legal Medicine at the Faculty of Medicine of the
well Museum Documented Collection (University of New Complutense University of Madrid (Madrid, Spain) [26]. As
Mexico, Albuquerque, United States) and the Simon Collec- of 2023 the collection includes 238 individuals and contin-
tion of identified skeletons (Department of Anthropology, ues to expand due to the ongoing agreement between the
University of Geneva, Switzerland) [11]. University and the funeral services of the Community of
Since its publication, this pelvic-based ten-variable met- Madrid. However, at the time of data acquisition (2010),
ric methodology has been tested in geographically diverse, this twentieth-century collection potentially included 195
documented collections with accuracy outcomes ranging accessible individuals (80 females and 115 males) ranging
from 88.34 to 100%: France [12], Greece [13], Mexico [14] from 3 to 97 years of age.
and Brazil [15–17]. This approach has also shown high reli- During the analysis, only mature individuals with the
ability and accurate results with virtual/ digital samples, i.e., three elements of the innominate fused were selected and
using CT images from Europe, specifically Belgium [18], remains displaying pelvic pathologies were excluded from
France [19], and Denmark [20]. Furthermore, these proven the study. This yielded a study sample comprising of 157
advantages led to its application to past population samples individuals. Within this sample, preservation permit-
from Neanderthals [21], Pre-Columbian mummies [22], ted applying the DSP2 methodology to at least one of the
Gravettian individuals [23], Joseon Dynasty Koreans [24] hipbones (left and/or right) of most individuals. However,
and medieval skeletal remains [25]. Although the original seven right hipbones and four left ones did not have coun-
studies advocated for the existence of a common sexual pat- terparts to undertake corresponding comparisons, attribut-
tern worldwide [10, 11], to the best of our knowledge, this is able to either poor preservation, or total absence. Thus, the
the first time the DSP approach is being tested in a Spanish final tally of analyzed sample included 303 coxal bones
documented collection. obtained from 157 individuals.
This study aimed to (1) analyze the intra-rater reliabil- Demographic information derived from obituary records
ity, applicability, and accuracy of the DSP2 method when (age-at-death and biological sex) was not available for
applied to a Spanish documented collection for sex esti- all individuals. Hence, accuracy investigation, for which
mation; (2) examine and compare the % applicability, % documented sex is critical, was undertaken with a reduced
dataset of 140 os coxae (74 males [54.4%] and 66 females
Table 1 Demographic information about the documented sample from [45.6%]; Table 1). Within this subset, 136 individuals were
the University Complutense of Madrid used for the investigation of
accuracy outcomes (N = 140) fully documented (age-at-death and biological sex), and
Age group Males Females Total 4 were partially identified (3 males and 1 female; whose
N % N % ages were unknown). Regarding sex-wise age distribu-
< 30 years old 04 5.4 01 1.5 06 tion of this subset, the female subsample was significantly
30–39 years old 12 16.2 05 7.6 19 older than males (♀ 68.82 ± 16.26 years vs. ♂ 57.52 ± 19.47
40–49 years old 11 14.9 04 6.1 16 years; Mann-Whitney U = 1566.5; p = 0.001), as is observed
50–59 years old 13 17.6 06 9.1 15 within osteological documented collections, attributable to
60–69 years old 05 6.8 10 15.2 19 differential life expectancy, amongst other factors.
70–79 years old 14 18.9 21 31.8 35
80–89 years old 10 13.5 16 24.2 25
> 90 years old 02 2.7 02 3.0 04
Unknown age 03 4.1 01 1.5 04
Total 74 100.0 66 100.0 140

13
International Journal of Legal Medicine (2025) 139:847–862 849

Sex estimation ● ISMM - Ischium post-acetabular length: distance from

the most anterior and inferior point of the ischial tuber-
During the examination process, the innominate was sepa- osity to the furthest point on the acetabular border [29].
rated from the rest of the skeleton to prevent complemen- ● SCOX - Iliac or coxal breadth: distance between the
tary information from affecting the objectivity of the results. anterior-superior iliac spine and the postero-superior
Thus, the sex estimation technique was applied blindly by iliac spine [27].
the first author, without knowing any biological information ● SS - Spino-sciatic length: minimum distance between
for the individuals analyzed. Additionally, no other informa- the antero-inferior iliac spine and the deepest point in
tion that could affect the study (except for the individual’s the greater sciatic notch [28].
code) was available where the material is stored. ● SA - Spino-auricular length: distance between the an-
The DSP tool, originally created and published to enable tero-inferior iliac spine and the auricular point. Auricu-
non-population specific sex estimation in 2005, utilised a lar point is defined as the intersection of the arcuate line
simple spreadsheet program where the researcher could with the auricular surface [28].
type in the maximum possible number of variables (mea- ● SIS - Cotylo-sciatic breadth: distance between the lat-
surements in mm) out of the recommended ten, with a eral border of the acetabulum and the midpoint of the
required minimum of four variables. Based on the input anterior portion of the great sciatic notch. Fixed arm
data, the probability of each specimen being male or female of the sliding caliper is parallel to the acetabular plane
was automatically computed, with an equal prior probabil- [27].
ity for male and female groups (pMale = pFemale = 0.5). It ● VEAC - Vertical acetabular diameter: maximum vertical
is important to highlight that although the program worked diameter of the acetabulum, measured on the acetabu-
with a minimum of four variables, the more variables uti- lar rim, as a prolongation of the longitudinal axis of the
lised for analysis, the greater was the likelihood of obtain- ischium [27].
ing a significant probability. In addition, sex was determined
only if the posterior probability was ≥ 0.95 threshold, which The ten variables are displayed in decreasing order of dis-
equates to a risk of error of 0.05 i.e., the maximum required criminant interest. The authors of these pivotal studies [10,
for reliable palaeobiological studies [6]. If these probability 11] recommend the first eight variables for sex estimation.
and error conditions were not fulfilled, an individual was The last two dimensions (SIS and VEAC) should mainly
classified as undetermined. The ten measurements that the be reserved for incomplete bones. All the variables were
method employed were the following: measured with a standard non-digital sliding caliper and
a spreading caliper as proposed by Murail and colleagues
● PUM - Acetabulo-symphyseal pubic length: minimum [10].
distance from the superior and medial point of the pubic Following the creation of this spreadsheet in 2005, owing
symphysis to the nearest point on the acetabular rim at to its lack of feasibility, a new scoring software called DSP2
the level of the lunate surface [27]. was created in 2017 [11]. Consequently, the analysis within
● SPU - Cotylo- pubic width: pubic breadth between the the current study was performed using the DSP2 updated
most lateral acetabular point and the medial aspect of software ​(​​​h​t​​t​p​:​​/​/​p​r​​o​j​​e​t​s​.​p​a​c​e​a​.​u​-​b​o​r​d​e​a​u​x​.​f​r​/​l​o​g​i​c​i​e​l​/​D​S​P​
the pubis. Measurement is perpendicular to the major 2​/​d​s​p​2​.​h​t​m​l​​​​)​
​ , which utilizes Fisher’s linear discriminant
axis of the os pubis; arms of the sliding caliper are thus analysis (LDA). More detailed information about this math-
parallel to the plane of the obturator foramen [28]. ematical approach is described thoroughly in the original
● DCOX - Innominate or coxal length: maximum height studies [10, 11].
of os coxae measured from the inferior border of the os
coxae to the most superior portion of the iliac crest. Can Statistical analyses
be measured with sliding calipers or osteometric board
[27]. The sex estimation methodology was applied to 157 avail-
● IIMT - Greater sciatic notch height: distance from the able mature individuals at that time, using with whom intra-
postero-inferior iliac spine (defined as the point of in- rater reliability, asymmetry and applicability analyses were
tersection between the auricular surface and the pos- performed (refer to aims of the study).
terior portion of the sciatic notch) to anterior border of In order to address the objectives of this present research,
the great sciatic notch. Axis of the measurement must be three consecutive sets of measurements were taken in a row
perpendicular to the anterior border. Due to the configu- from the same individuals: the first set from the left innomi-
ration of hip bone, it is easier to use small arms of the nate (S1), the second set from the right innominate (S2), and
sliding caliper [27]. the third set from the left innominate again (S3).

13
850 International Journal of Legal Medicine (2025) 139:847–862

The first and the third sets of measurements (S1 vs. S3) Table 2 Intraclass correlation coefficient (ICC) for intra-observer error
analysis for the 10 variables of DSP methodology (N = 157) (results
were compared to analyze the intra-observer reliability obtained on comparing the sets 1 and 3 (S1 vs. S3))
(n = 157). To do this, a two-way random intraclass correla- Variable ICC 95% IC p - value
tion coefficient (ICC) was computed, due to the nature and PUM 0.885 0.806–0.928 < 0.001
characteristics of the data. Due to the data characteristics, SPU 0.949 0.928–0.964 < 0.001
absolute agreement ICC type was used and, as reported by DCOX 0.970 0.953–0.980 < 0.001
Daniel [30], single measures should be employed when IIMT 0.659 0.167–0.838 < 0.001
intra-observer performance was tested. Obtained ICC val- ISMM 0.969 0.956–0.978 < 0.001
ues were interpreted according to Koo and Li [31], wherein SCOX 0.942 0.969–0.984 < 0.001
ICC values lower than 0.5 are indicative of poor reliability, SS 0.975 0.962–0.983 < 0.001
values between 0.5 and 0.75 indicate moderate reliability, SA 0.929 0.902–0.949 < 0.001
SIS 0.972 0.960–0.980 < 0.001
values between 0.75 and 0.9 indicate good reliability and
VEAC 0.953 0.935–0.966 < 0.001
values higher than 0.90 indicate excellent reliability. The
PUM: Acetabulo-symphyseal pubic length; SPU: Cotylo- pubic
second and the third sets (S2 vs. S3) were compared to width; DCOX: Innominate or coxal length; IIMT: Greater sciatic
establish potential directional asymmetries (n = 157) within notch height; ISMM: Ischium post-acetabular length; SCOX: Iliac or
the os coxae. For this specific analysis, paired t-test and Wil- coxal breadth; SS: Spino-sciatic length¸SA: Spino-auricular length;
coxon test was used for normal and non-normal variables, SIS: Cotylo-sciatic breadth; VEAC: Vertical acetabular diameter.
Significant results are marked in bold
respectively, as and where applicable.
Regarding sexual differences within the DSP2 variables
(n = 140), student T-test or Mann-Whitney analysis was per- probability ≥ 0.95, leaving the undetermined ones out), and
formed depending on whether the variable follows a normal the latter takes into account the percentage of specimens
or a non-normal distribution, respectively. where specific variables could be measured based on the
To estimate accuracy for sexing, S2 (right side) and S3 individual preservation.
(left side) were analyzed. Two parameters were computed All tests were undertaken using SPSS 29.0. For all statis-
following the original publication [11]: tical assessments, a p-value lower than 0.05 was considered
statistically significant.
a) Percentage of sexing (% sexing), which constitutes the
percentage of specimens whose sex was estimated. To
establish the percentage of sexing, a posterior probabil- Results
ity equal or superior to 0.95 was considered to be the
sex classification threshold. Based on the values mentioned in the Methods Sect. [31],
b) Percentage of accuracy (% accuracy), which is the per- excellent reliability was achieved during intra-observer
centage of specimens whose sex has been correctly esti- analysis for 8 out of 10 DSP variables, with good reliability
mated among those calculated. for PUM and moderate reliability for IIMT (Table 2). Scat-
terplots of S1 vs. S3 for the variables PUM and IIMT have
The % sexing parameter is different to % applicabil- been displayed as Supporting Information (Supplementary
ity as the former considers the percentage of specimens Images 1 and 2) to provide a visual complement to the
that were classified as male or female (with a posterior obtained reliability results.

Table 3 Directional asymmetry analyses of the DSP2 variables (comparing sets 2 and 3 (S2 vs. S3)) (N = 157)
Variable S2 (right) S3 (left) T-test Fd Wilcoxon Z p - value
PUM 70.59 70.66 x x -0.423 0.672
SPU 26.29 26.31 x x -0.387 0.699
DCOX 202.28 202.89 x x -2.151 0.031
IIMT 43.06 43.83 -2.264 123 x 0.013
ISMM 106.39 105.42 x X -4.764 < 0.001
SCOX 154.32 154.49 -0.607 110 x 0.273
SS 69.82 70.00 -0.940 140 x 0.174
SA 74.77 75.80 x X -3.011 0.003
SIS 37.39 37.38 0.043 143 x 0.483
VEAC 54.57 54.72 x x -0.884 0.377
PUM: Acetabulo-symphyseal pubic length; SPU: Cotylo- pubic width; DCOX: Innominate or coxal length; IIMT: Greater sciatic notch height;
ISMM: Ischium post-acetabular length; SCOX: Iliac or coxal breadth; SS: Spino-sciatic length¸SA: Spino-auricular length; SIS: Cotylo-sciatic
breadth; VEAC: Vertical acetabular diameter. Significant results are marked in bold

13
International Journal of Legal Medicine (2025) 139:847–862 851

Regarding potential directional asymmetries, results are however, all female features achieved better results than
displayed in Table 3. Significant differences were observed male ones, with the exception of SCOX.
between the two sides in four variables (DCOX, IIMT, Concerning the accuracy of right (S2) and left (S3)
ISMM and SA), while no significant differences between datasets, general outcomes showed that accuracy percent-
left and right os coxae were achieved with the rest of the ages were high, being > 94% regardless of the number of
DSP variables. In the case of DCOX, IIMT and SA, mea- variables used (Table 6). Similar values (around 97%) were
surements for the left os coxa were significantly higher than reached with all variables, the first 8, or the most accurate
right ones, while the opposite was observed for ISMM. 4 variables. In fact, even with the worst 4 variables, the
A comparison of minimum and maximum values for accuracy % reached 94.74% and 94.55% for left and right
each of the ten DSP variables obtained in the present study, datasets, respectively. Thus, it appears that having the mea-
against those reported by Bruzek et al. [11], indicated surements of the first 4 variables is enough to obtain the
that all values are within the range reported in the origi- highest accuracy possible within this pooled sample. How-
nal study [11] (Supplementary Table 1). The only current ever, for the percentage of sexing, the results were different:
value that is not falling within the range variation shared while similar values were achieved with 10 or the first 8
by the original authors in the software platform was VEAC, variables, the values slightly decreased with the best 4 vari-
which exceeded the provided maximum score (69 mm vs. ables and reduced to half with the worst 4 variables. These
66.5 mm). Nevertheless, it is important to maintain cau- results were similar when sexes were analyzed separately
tion during such a comparison, as the original study did not (Table 6). Thus, accuracy values were always higher than
mention the side associated with these measurements, and 90% in males and higher than 95% in females, meanwhile
to maintain uniformity, comparison measurements within sexing percentages decreased when the number and qual-
the present study have been obtained by combining the right ity of variables decreased as well. However, if the sexes
(S2) and left (S3) values. are compared, the females’ results are always higher than
Regarding sexual dimorphism of the DSP variables for the males’ ones, both for sexing and accuracy percentages.
the right and left sides, descriptive statistics are displayed in The singular exception to this observation is the sexing per-
Table 4. Overall, higher mean values were found in males centage with the four worst variables on both sides, where
compared to females except for PUM and IIMT, where the the number of estimated males is superior to the estimated
opposite pattern was observed. On the right side (S2), all females. Lastly, considering sides in the pooled sample, the
the DSP variables were significantly sexually dimorphic left dataset achieved better accuracy results in all cases. The
barring PUM, while on the left side (S3), neither PUM nor same occurred in most of the cases regarding sexing percent-
SA mean values were significantly different between males ages. When sexes are considered in isolation, no clear side
and females. The rest of the variables exhibited significant pattern was identified, as better results were achieved for
sexual dimorphism. left and right datasets in different variable-based categories.
According to sample preservation, % applicability for With the aim of investigating the relevance of the poste-
every single DSP variable in a sex-pooled sample was rior probability threshold for accuracy estimations, previous
higher than 85% in all cases, except for the variable SCOX, outcomes were compared against sexing accuracy obtained
where the values were around 80% on both sides (Table 5). by decreasing the posterior probability from 95% (as origi-
Similar results were achieved in males and females sepa- nal authors advised) to 85% (Table 7). The percentage of
rately, with the exception of the right male subsample for sexing increased in all cases, with a significant increase
IIMT, where the percentage was 81.69%. In the combined observed in males. Nevertheless, decreasing the posterior
sample, larger applicability values were achieved on the probability by 10% also reduced the percentage of accuracy
left side in all cases, with the exceptions of PUM, SS and in most cases, with few right-side exceptions where similar
SIS. In the sex-specific samples, some variables achieved results were obtained.
right predominance and others demonstrated a left predomi-
nance, with no consistent pattern. Besides, some of these
side differences coincided in males and females separately Discussion
(PUM, SPU, ISMM, SCOX, SS, VEAC), whereas others
did not (DCOX, IIMT, SA, SIS). Regarding applicability Sex estimation is one of the first procedures towards human
for sexing between the two sexes, higher percentages were identification within medicolegal and forensic contexts.
found in females for most of the DSP variables in the right Accurate sex estimation can also contribute to providing
subsample, except for DCOX, SCOX and VEAC. In the important insights into population history and migration
case of SIS, both sexes achieved 100%. On the left sample, patterns. Therefore, this step is an essential tool for foren-
sic anthropologists and archaeologists, and it is crucial for

13
 852

13
Table 4 Descriptive statistics (by sex) for the os coxae variables in the documented sample (N = 140)
Variable N (F) Mean (F) SD Min (F) Max (F) N (M) Mean (M) SD Min (M) Max (M) p -value p – value
(F) (M) (t-test) (Mann-Whitney U)
PUM (R) 56 71.20 4.775 62 85 63 70.26 3.471 61 78 0.514
SPU (R) 56 23.03 2.339 19 31 63 28.89 2.813 21.4 35 < 0.001
DCOX (R) 55 190.49 8.117 172 210 66 212.03 11.093 175 236 < 0.001
IIMT (R) 57 44.71 4.810 33 57.6 58 41.52 4.416 32.6 51 < 0.001
ISMM (R) 57 99.12 4.722 90 113.6 65 112.52 5.730 96.6 125 < 0.001
SCOX (R) 49 149.98 6.969 133 168 57 157.79 8.546 136 174 < 0.001
SS (R) 62 65.45 3.754 58 76.4 70 73.45 4.482 62 83 < 0.001
SA (R) 62 73.26 5.750 57.6 89.7 68 75.43 5.992 61 90 0.019
SIS (R) 62 35.03 3.295 28 41 71 39.52 3.333 31 46 < 0.001
VEAC (R) 60 51.53 3.008 44 62 70 57.36 3.902 49 69 < 0.001
PUM (L) 57 71.07 4.651 61 84 63 70.34 4.151 59 81.6 0.182
SPU (L) 61 22.84 2.333 19 29.3 67 28.92 2.582 21 37 < 0.001
DCOX (L) 60 192.10 9.311 172 216 65 212.45 11.054 176 238 < 0.001
IIMT (L) 57 45.60 4.560 36 58.6 63 42.49 4.028 31 51 < 0.001
ISMM (L) 61 98.52 5.113 87.6 112 66 111.35 5.810 95 124 < 0.001
SCOX (L) 52 150.85 7.188 133 167 60 157.82 8.825 136 181 < 0.001
SS (L) 64 65.81 3.633 57.7 73.7 69 73.36 4.593 64 85 < 0.001
SA (L) 64 74.95 6.106 62 92 69 76.54 5.771 65.6 92 0.063
SIS (L) 65 35.11 3.289 28 42 71 39.32 3.287 31.6 46.4 < 0.001
VEAC (L) 65 51.45 2.657 45 58 71 57.79 3.944 48.8 67 < 0.001
PUM: Acetabulo-symphyseal pubic length; SPU: Cotylo- pubic width; DCOX: Innominate or coxal length; IIMT: Greater sciatic notch height; ISMM: Ischium post-acetabular length; SCOX:
Iliac or coxal breadth; SS: Spino-sciatic length¸SA: Spino-auricular length; SIS: Cotylo-sciatic breadth; VEAC: Vertical acetabular diameter. (F): female subsample; (M): male subsample;
(R): right innominate (set 2); (L): left innominate (set 3). Significant results are marked in bold
International Journal of Legal Medicine (2025) 139:847–862
International Journal of Legal Medicine (2025) 139:847–862 853

Table 5 Percentage of applicability of the individual variables of DSP2 on the studied sample
Males (%) Females (%) Total (%)
Right Left Right Left Right Left
PUM 88.73 87.50 90.32 87.69 89.33 88.24
SPU 88.73 93.06 90.32 93.85 88.67 93.46
DCOX 92.96 90.28 88.71 92.31 90.00 90.85
IIMT 81.69 87.50 91.94 87.69 86.67 88.24
ISMM 91.55 91.67 91.94 93.85 92.00 92.16
SCOX 80.28 83.33 79.03 80.00 79.33 81.70
SS 98.59 95.83 100.00 98.46 98.00 97.74
SA 95.77 95.83 100.00 98.46 96.67 97.39
SIS 100.00 98.61 100.00 100.00 100.00 97.39
VEAC 98.59 98.61 96.77 100.00 98.00 98.69
PUM: Acetabulo-symphyseal pubic length; SPU: Cotylo- pubic width; DCOX: Innominate or coxal length; IIMT: Greater sciatic notch height;
ISMM: Ischium post-acetabular length; SCOX: Iliac or coxal breadth; SS: Spino-sciatic length¸SA: Spino-auricular length; SIS: Cotylo-sciatic
breadth; VEAC: Vertical acetabular diameter. Global dataset (n = 157) was used for these calculations in the pooled sample while the docu-
mented sample (n = 140) was used to perform the sex-specific calculations. Applicability refers to the number of individuals out of the total
where the specific variable could be measured due to preservation

Table 6 Sexing accuracy results with various combinations of variables within the documented sample (N = 140)
Undetermined / N Number of errors / Undetermined / N Number of errors / Undetermined Number of errors
(% sexing) (F) determined N (% sexing) (M) determined N /N / determined N
(% accuracy) (F) (% accuracy) (M) (% sexing) (% accuracy)
(Total) (Total)
All available 1 / 65 0 / 64 11 / 72 2 / 61 12 / 137 2 / 125
variables (L) (98.46) (100) (84.72) (96.72) (91.24) (98.40)
All available 3 / 62 0 / 59 9 / 71 3 / 62 12 / 133 3 / 121
variables (R) (95.16) (100) (87.32) (95.16) (90.98) (97.52)
10 variables (L) 0 / 45 0 / 45 6 / 48 1 / 42 6 / 93 1 / 87
(100) (100) (87.50) (97.62) (93.55) (98.85)
10 variables (R) 0 / 44 0 / 44 5 / 46 2 / 41 5 / 90 2 / 85
(100) (100) (89.13) (95.12) (94.44) (97.65)
First 8 variables 0 / 45 0 / 45 6 / 49 1 / 43 6 / 94 1 / 88
(L) (100) (100) (87.76) (97.67) (93.62) (98.86)
First 8 variables 0 / 44 0 / 44 5 / 46 2 / 41 5 / 90 2 / 85
(R) (100) (100) (89.13) (95.12) (94.44) (97.65)
Best 4 variables 2 / 47 0 / 45 9 / 55 2 / 46 11 / 103 2 / 92
(L) (95.83) (100) (83.64) (95.65) (89.32) (97.83)
Best 4 variables 0 / 47 0 / 47 12 / 51 2 / 39 12 / 98 2 / 86
(R) (100) (100) (76.47) (94.87) (87.76) (97.67)
Worst 4 variables 40 / 64 0 / 24 34 / 67 3 / 33 74 / 131 3 / 57
(L) (37.50 (100) (49.25) (90.91) (45.51) (94.74)
Worst 4 variables 40 / 60 1 / 20 32 / 67 2 / 35 72 / 127 3 / 55
(R) (33.33) (95.00) (52.24) (94.29) (43.31) (94.55)
% sexing means the percentage of specimens whose sex has been determined (p ≥ 0.95) while % accuracy means the percentage of specimens
whose sex has been correctly determined among those determined. Results are given for both sides separately and, regarding biological sex,
for the pooled sample and each sex separately. First 8 variables: without SIS and VEAC. Best 4 variables: DCOX, PUM, SPU and IIMT. Worst
4 variables: SIS, VEAC, SA and SS. (L): left. (R): right. (F): female. (M): male

a comprehensive understanding of human evolution, biol- association with parturition and proven accuracy for being
ogy, and health. Numerous research has previously been the most dimorphic human bone, specific research on sex
published on metric sex estimation for diverse Spanish estimation using hipbone measurements is currently limited
samples, using different skeletal elements: the skull [32], for this biogeographical population [47, 54]. In this respect,
the dentition [33, 34], the clavicle [35–37], the sternum to the best of our knowledge, DSP2 has not been tested in
[38], the ribs [39, 40], the vertebrae [41–43], the sacrum any Spanish sample, so comparisons along this line are not
[44], the radius [45], the carpals [46], the femur [47, 48], the feasible.
patella [37, 49], the tibia [50, 51], the talus [45], the navicu-
lar [52] or the metacarpals [53]. However, despite its direct

13
854 International Journal of Legal Medicine (2025) 139:847–862

Table 7 Results with various combinations of variables based on the posterior probability (95% vs. 85%) within the documented sample (N = 140)
Posterior Posterior Results N
probability ≥ 0.95 probability ≥ 0.85 (total)
% sexing % accuracy % sexing % accuracy Sexing Sexing Number of Number of
(M) (F) new errors new errors
(M) (F)
All available variables (L) 91.24 98.40 94.90 97.69 + 5 (11) + 0 (1) + 1 (5) + 0 (0) 137
All available variables (R) 90.98 97.52 96.24 97.66 + 7 (9) + 0 (3) + 0 (7) + 0 (0) 133
10 variables (L) 93.55 98.85 97.85 97.80 + 4 (6) + 0 (0) + 1 (4) + 0 (0) 93
10 variables (R) 94.44 97.65 97.78 97.73 + 3 (5) + 0 (0) + 0 (3) + 0 (0) 90
First 8 variables (L) 93.62 98.86 97.87 97.83 + 4 (6) + 0 (0) + 1 (4) + 0 (0) 94
First 8 variables (R) 94.44 97.65 97.78 97.73 + 3 (5) + 0 (0) + 0 (3) + 0 (0) 90
Best 4 variables (L) 89.32 97.83 95.15 96.94 + 4 (9) + 2 (2) + 1 (4) + 0 (2) 103
Best 4 variables (R) 87.76 97.67 96.94 96.84 + 9 (12) + 0 (0) + 1 (9) + 0 (0) 98
Worst 4 variables (L) 45.51 94.74 76.34 93.00 + 17 (34) + 26 (40) + 3 (17) + 1 (26) 131
Worst 4 variables (R) 43.31 94.55 66.14 92.86 + 19 (40) + 10 (32) + 3 (19) + 0 (10) 127
% sexing means the percentage of specimens whose sex has been determined (p ≥ 0.95) while % accuracy means the percentage of specimens
whose sex has been correctly determined among those determined. Results are given for both sides separately and, regarding biological sex,
for the pooled sample and each sex separately. First 8 variables: without SIS and VEAC. Best 4 variables: DCOX, PUM, SPU and IIMT. Worst
4 variables: SIS, VEAC, SA and SS. L: left. R: right. F: female. M: male
In the “Sexing” columns belonging to “Results”: 1) Plain numbers means the number of individuals who were determined with the 0.85 thresh-
old but not with the 0.95 one. The number between parentheses shows the total number of undetermined individuals with the 0.95 threshold. In
the “Number of new errors” columns belonging to “Results”: 1) Plain numbers means the number of new committed errors in sex estimation
when using the 0.85 threshold instead of the 0.95 one. The number between parentheses shows the number of individuals who were determined
with the 0.85 threshold but not with the 0.95 one

previous literature results [11, 13, 14]. However, those

Reliability authors reported similar lower values for PUM, contrasting
to our findings. Furthermore, overall better female preser-
Overall, the current study shows excellent results for intra- vation (% applicability: possibility of taking the measure-
rater reliability analysis, coinciding with previous litera- ments) compared to males was found (except SCOX),
ture for pelvic measurements [15, 55, 56]. However, good contrasting to the higher gracility and associated fragility of
and moderate results were achieved with PUM and IIMT, skeleton of women. This could be linked to their belonging
respectively, yielding lower ICC values compared to the to a documented collection from modern cemeteries, with
ones reported by de Almeida in a Brazilian sample [15]. different taphonomic processes or artificial barriers such as
The comparatively lower values for sciatic greater notch coffin protection, slowing down the natural human decom-
(IIMT) have already been reported during previous reliabil- position. In addition, the female mean age was significantly
ity analyses [15, 55], and with CT-based studies [19, 20]. higher than the male’s in this study sample, rendering their
These results are likely due to different factors such as the skeletons more prone to be vulnerable and fragmentary,
greater difficulty associated with identifying the anatomical so the current results regarding applicability should be
landmarks and placing the sliding caliper in the correct posi- explained by other different factors. The SCOX exception,
tion, which requires the user to visually check a right angle on the other hand, may be due to both postero-superior and
between the postero-inferior iliac spine and the anterior bor- antero-superior spines involved in the measurement. These
der of the greater sciatic notch [27]. In addition, it is worth landmarks are likely to be more robust, and so preservable
highlighting that these outcomes contrast with the fact that in males due to anatomical muscular attachments.
both features, PUM and IIMT, belong to the first four DSP
variables, considered to have the highest discrimination Sexing and accuracy percentages
power [10]. Consequently, special care should be taken
when using these specific variables, and better descriptions Overall results
and/or images could be provided in future DSP updates.
Current results show that, whereas the sexing % decreased
Applicability progressively when the number of variables or their dis-
crimination power was reduced, the accuracy % remained
The current percentage applicability for SCOX was lower in high irrespective of the combination of employed variables.
comparison to other variables. This outcome corroborated A comprehensive comparison to previously reported similar

13
International Journal of Legal Medicine (2025) 139:847–862 855

literature is shown in Table 8. Overall, results from all ten, or due to its smaller sample size, the results ranged between
the first eight variables are almost identical across the table, 90.48% [18] and 100% [10–12]. In the case of the % sex-
so it appears that when the first eight variables are avail- ing, the scenario is different. The minimum outcome was
able for analysis, incorporating the remaining two (SIS and reported in a British subsample published by the original
VEAC) does not render better outcomes. As advocated by authors [11] as 32.79%, while the maximum was also found
Murail et al. and Bruzek et al. [10, 11], the last two features within the original publication [10] as 76%. Murail and col-
are useful within degraded and/or fragmentary contexts, leagues’ original publication reported a sexing % of 40.23%
where other more accurate variables are non-viable. In such in the worldwide sample (n = 2040) but, surprisingly, higher
scenarios, these variables with relatively lower accuracy can outcomes were reported in its different subsamples (even
aid in identification by helping achieve the minimum four- when 7 out of 9 of them were testing reference models on
variable requirement mandated by the DSP software. geographically different target samples), ranging from 63
Sexing % values obtained with all ten variables in the to 76%. The rest of the revised literature, including the
present study were 94.44% and 93.55% for the right and left updating of DSP as DSP2 [11], reported % sexing between
halves, respectively, which is slightly lower than the 97.40% 32.79% [11] and 61.0% [15]. A scientific explanation for
reported by Chapman et al. [18], the maximum value found Murail’s differential findings is currently lacking.
in the literature, and higher than the 85.43% reported by One of the drawbacks of DSP compared with other dis-
Salles Machado et al. [17], the minimum value achieved criminant analysis is the high number of undetermined cases
across referenced studies (Table 8). Regarding % accuracy, that the software generates, especially when a reduced num-
the present study achieved 97.65 and 98.85 (right and left, ber of variables are considered. These results arise of the
respectively), within the range from the minimum 86.20% original authors decision about employing a 0.95 threshold
[24] to the maximum, 100.0% [11, 12, 14, 18–20]. A com- instead of the 0.50 value. This settlement, however, guaran-
parative analysis (Table 8) indicates that general % accu- tees very high levels of accuracy, with 100% or close val-
racy results are consistently high in literature, with most of ues with a combination of the most discriminant variables.
the authors reporting values higher than 95% with only two Interestingly, even taking into account just the four variables
exceptions: a Brazilian miscegenated population [17] and a with relatively lower discrimination power (SS, SA, SIS
very small sample from South Korea [24]. In fact, several and VEAC), accuracy percentages were lower than 94.0%
authors found no cases of misclassification during their sex in just 3 out of 32 revised literature outcomes, including the
estimation by DSP [11, 12, 14, 18–20]. In the current study, Korean one with a sample size of 29 individuals. Neverthe-
this sort of success rate was achieved only in females. less, as other authors have suggested [20], the poorer results
The present results of % accuracy with the best 4 vari- with the worst 4 combination of variables (or with a reduced
ables were 97.67% and 97.83% for right and left, respec- number of variables) need to be interpreted with caution and
tively. According to the previous studies (Table 8), the complemented with non-metric methodologies.
reported results ranged from 80% [24] to 100% [10–12, In the present study, the posterior probability threshold
15, 18, 20]. However, the lowest value (80%) came from was decreased from 95 to 85% to assess its potential impact
an already mentioned small sample from Korea compris- on % sexing and % accuracy. This resulted in an increase in
ing of only 29 individuals. The second lowest result gen- the % sexing in all cases in exchange of reduced % accuracy
erated with a larger sample size (n = 103) was reported as in most of the cases. Since the current female outcomes were
90.29% [17]. In this specific case [17], the results for % comparably better than males, most of the changes were
accuracy should be taken with caution since some reported associated with sexing in males: improvement by increas-
values are unexpected and difficult to justify; i.e. they ren- ing the sexing rate but gaining imprecision by incorporat-
dered 88.34%, 90.29% and 93.20% using 10 variables, best ing errors. While this reduced accuracy still has acceptable
4 variables and worst 4 variables, respectively. Continuing values, the reduction in posterior probability is not recom-
with best 4 variables results, all other studies yielded accu- mended within future forensic practice from a reliability
racy percentages higher than 90%. However, outcomes of point of view, as a decrease in the number of indeterminate
% sexing are quite different: while the present study’s ones bones does not necessarily make up for decreased reliability
were 87.76% and 89.32% for right and left sides, respec- and accuracy. Given the medico-legal and forensic contexts
tively, previously literature results ranged from 71.09% [16] of interest here, an unreliable sex estimation approach can
to 92.8% for a specific European subsample [10]. be more problematic than an unapplicable one [57, 58]. Fur-
Results of the worst 4 variables achieved good accuracy thermore, the reduction in posterior probability here, by and
% outcomes in the present study: 94.55% and 94.74% for large, also resulted in a decrease in % accuracy, negating the
the right and left sides, respectively. Regarding previous need for this reduction. Moreover, even with a 95% poste-
studies, if the Korean small sample study is not considered rior probability, % accuracy obtained in the present study

13
856 International Journal of Legal Medicine (2025) 139:847–862

Table 8 Comparison of results across literature and the present study when applying DSP methodology in pooled samples belonging to diverse
populations based on the number of variables used
TOTAL 10 variables 8 variables “Best” 4 variables “Worst” 4 variables
Reference Geographical N % sexing % % sexing % % sexing % % sexing % accu-
origin accuracy accuracy accuracy racy
Murail et al., 2005 [10] European 454 - - 95.9 100 92.8 100 69.7 98.3
(UK, France,
Portugal)
Murail et al., 2005 [10] African Ameri- 329 - - 92 98.6 86.9 98.6 66.1 98.6
can 1
Murail et al., 2005 [10] European 311 - - 93 100 89.5 99.6 65 97.4
American 1
Murail et al., 2005 [10] Euro- 1094 - - 99.7 99.3 86.9 99.7 76 99.6
pean + North
American
Murail et al., 2005 [10] Thailand 2 198 - - 94.1 100 90.5 100 75.5 100
Murail et al., 2005 [10] Lithuania 2 220 - - 94.4 100 91.7 100 71.6 98.7
Murail et al., 2005 [10] South Africa – 306 - - 88.7 98.8 84.6 98.8 66.2 99
Zulu 2
Murail et al., 2005 [10] South Africa – 110 - - 86 100 84.4 100 63 100
Soto 2
Murail et al., 2005 [10] South Africa – 112 - - 95.1 100 88.8 100 70.8 100
Afrikaner 2
Murail et al., 2005 [10] Worldwide 2040 90.71 99.63 90.76 99.63 86.69 99.61 40.23 98.75
Sánchez-Mejorada et al., Mexico 250 89.2 100 - - - - - -
2011 [14]
Chapman et al., 2014 Belgium 39 97.4 100 97.44 100 89.74 100 53.85 90.48
[18]
Bruzek et al., 2017 [11] France 160 89.93 100 90.07 100 83.22 99.19 45.57 100
Bruzek et al., 2017 [11] Portugal 232 89.64 100 89.69 100 86.54 98.33 44.78 99.03
Bruzek et al., 2017 [11] United 62 86.54 100 86.54 100 80.7 100 32.79 100
Kingdom
Bruzek et al., 2017 [11] Lithuania 220 95.39 100 95.41 100 92.73 100 39.91 100
Bruzek et al., 2017 [11] South Africa 306 88.44 99.23 88.78 99.23 84.85 100 33.44 98.04
- Zulu
Bruzek et al., 2017 [11] South Africa 110 85.44 100 85.44 100 80.73 100 43.4 95.65
- Soto
Bruzek et al., 2017 [11] South Africa 112 93.62 100 94 100 89.72 100 43.27 100
- Afrikaner
Bruzek et al., 2017 [11] European 112 93.62 100 93.68 100 90.48 100 42.45 100
American
Bruzek et al., 2017 [11] African 113 90.2 98.91 90.2 98.91 87.04 100 40.18 97.78
American
Bruzek et al., 2017 [11] European 199 88.24 100 89.01 100 84.02 99.39 49.22 97.89
American
Bruzek et al., 2017 [11] African 216 91.71 98.4 91.39 98.43 89.57 98.41 45.02 96.84
American
Bruzek et al., 2017 [11] Thailand 198 94.62 100 94.62 100 91.1 100 38.14 100
Bruzek et al., 2017 [11] North America 120 93.46 99 93.58 99.02 87.27 98.96 50.86 94.92
Bruzek et al., 2017 [11] Switzerland 503 94.74 96.03 94.78 96.06 90.91 96.88 55.25 98.06
Bruzek et al., 2017 [11] Worldwide 2040 90.84 99.65 90.98 99.65 87.17 99.53 41.49 98.67
Quatrehomme et al., France 100 94.83 100 94.92 100 76.92 100 52.87 100
2017 [12]
Salles Machado et al., Brazil 103 85.43 88.34 - - 82.52 90.29 60.19 93.20
2018 [17]
Rodriguez Paz et al., Denmark 116 93.9 100 93.1 100 81.9 100 49.7 98.2
2019 [20]
Kranioti et al., 2019 [13] Greece 133 88.00 97.43 - - - - - -
De Almeida et al., 2020 Brazil 301 94.00 99.3 94.7 99.3 90.7 100 61.0 98.9
[15]

13
International Journal of Legal Medicine (2025) 139:847–862 857

Table 8 (continued)
TOTAL 10 variables 8 variables “Best” 4 variables “Worst” 4 variables
Reference Geographical N % sexing % % sexing % % sexing % % sexing % accu-
origin accuracy accuracy accuracy racy
Rodrigo de Oliveira Brazil 128 - - - - 71.09 92.97 - -
Lopes et al., 2023 [16]
Oh et al., 2023 [24] South Korea 29 86.2 86.2 89.66 86.2 85.00 80 31.3 31.03
Current study (right) Spain 140 94.44 97.65 94.44 97.65 87.76 97.67 43.31 94.55
Current study (left) Spain 140 93.55 98.85 93.62 98.86 89.32 97.83 45.51 94.74
1
Results reported from Murail et al. testing the European model (reference) on North American target samples
2
Results reported from Murail et al. testing the European and North American model (reference) on different targets: Asian (Thailand), African
(South Africa) and European (Lithuania) samples
% sexing means the percentage of specimens whose sex has been determined (p ≥ 0.95) while % accuracy means the percentage of specimens
whose sex has been correctly determined among those determined

is high (even with the worst four variables), demonstrating right sides both achieved 100%. In males, on the other hand,
an excellent applicability for the DSP2 approach when it right side displayed higher % sexing in all cases except
comes to sex estimation. for the best 4 variables, while left side achieved better %
In relation to misclassification cases, some documentary accuracy than right side with one exception: the worst 4
or human error should not be completely ruled out in the variables case. The only previous investigation to compare
osteological collection itself, since worldwide researchers those results was from Rodrigo de Oliveira Lopes et al. [16],
and curators work with them all the time and misplacing who reported accuracy comparison for the best 4 variables
skeletal elements or typographical errors could be contribut- (PUM, SPU, DCOX, IIMT). These authors found higher
ing factors [18, 59]. % sexing and % accuracy in the left side in females and
in the right side in males, i.e., contrasting findings in com-
Side differences parison to the present study (Table 6). Accuracy % results
in females were not comparable because both left and right
Most of the authors who have analyzed DSP performance in sides achieved 100% in the present study.
diverse populations have used the left side throughout to be
consistent [13, 15, 17, 18, 20]. However, some researchers Sexual differences
used left and right hip bones in their studies, combining both
sides without analyzing for potential differences [14, 19]. Neither PUM nor SA mean metric values were found to
Whereas other studies only partially examined them [12, have significant differences between sexes in the left side
16]. The present study has found higher values in all left coinciding with Salles Machado et al. [17]. In the right
cases compared to the right-side results, apart from ISMM, side, PUM alone did not exhibit significant sex differences,
with opposite results, and a quite identical mean value in a result unreported in the literature. Other authors describe
SIS between sides. However, these differences were only non-significant differences between males and females in
statistically significant in DCOX, IIMT, ISMM and SA. SA [11, 13–15, 19] and in the combination of PUM, SCOX
This finding could not be corroborated as no similar results and SA [12]. These PUM results highly contrast with the
have been reported previously, warranting further investiga- fact that this variable has the highest discrimination power
tion into the causality of this anomaly. In fact, other authors across the analyzed ten variables according to the DSP cre-
concluded that the DSP values from the right and left coxal ators [10], data which contrasts with Kranioti et al., who
bones were comparable [12], even suggesting that single described the four most discriminant variables as ISMM,
measurements may be substituted in cases of non availabil- SPU, DCOX and VEAC in a Greek sample [13].
ity of both halves of the pelvis. This absence of asymmetry Tables 9 and 10 display the comparison of sex-specific
agrees with previous results from a CT-based investigation percentages of sexing and accuracy depending on the
which utilised different pelvic measurements [60]. number of DSP variables used across scientific literature.
Overall, current results did not indicate any clear side- In the female comparative analysis (Table 9), the present
based results regarding the % sexing in different number of study achieved 100% in most of the cases of % sexing in
variables sets. However, left side achieved better % accu- the first three sets (10, 8 and best 4 variables), exceeding
racy compared to the right in all cases (Table 6). In females, the reported outcomes in other samples. Regarding % accu-
similar results were found. No differences were displayed racy, no misclassification cases were reported here, agreeing
in most of the cases of % accuracy in females as left and with the ones rendered in Mexican [14], French [12, 19] and

13
858 International Journal of Legal Medicine (2025) 139:847–862

Table 9 Comparison of literature results with current ones when applying DSP methodology in female samples belonging to diverse populations
based on the number of variables used
FEMALES 10 variables 8 variables “Best” 4 variables “Worst” 4 variables
Reference Geographi- N % sexing % % sexing % % sexing % % sexing % accu-
cal origin accuracy accuracy accuracy racy
Sánchez-Mejorada et al., Mexico 118 98.31 100 - - - - - -
2011 [14]
Mestekova et al., 2015 [19] France 54 97.2 100 - - 85.2 100 - -
Bruzek et al., 2017 [11] Worldwide 1023 90.96 99.29 91.21 99.30 87.55 99.53 42.54 98.82
Quatrehomme et al., 2017 France - 85.71 100 94.91 100 41.66 100 40.48 100
[12]
Salles Machado et al., 2018 Brazil 50 98.05 86.00 - - 92.00 88.00 38.00 82.00
[17]
Rodriguez Paz et al., 2019 Denmark 67 97.0 - - - 93.3 - - -
[20]
De Almeida et al., 2020 [15] Brazil 136 94.4 100 95.8 100 88.7 100 81.2 99.0
Rodrigo de Oliveira Lopes et Brazil 50 - - - - 82.0 96.0 - -
al., 2023 [16]
Current study (right) Spain 66 100 100 100 100 100 100 33.33 95.0
Current study (left) Spain 66 100 100 100 100 95.83 100 37.50 100
% sexing means the percentage of specimens whose sex has been determined (p ≥ 0.95) while % accuracy means the percentage of specimens
whose sex has been correctly determined among those determined

Table 10 Comparison of literature results with current ones when applying DSP methodology in male samples belonging to diverse populations
based on the number of variables used
MALES 10 variables 8 variables “Best” 4 variables “Worst” 4 variables
Reference Geographi- N % sexing % % sexing % % sexing % % sexing % accu-
cal origin accuracy accuracy accuracy racy
Sánchez-Mejorada et al., Mexico 132 81.06 100 - - - - - -
2011 [14]
Mestekova et al., 2015 [19] France 52 92.3 100 - - 96.2 100 - -
Bruzek et al., 2017 [11] Worldwide 1017 90.72 100 90.75 100 86.79 99.53 40.44 98.51
Quatrehomme et al., 2017 France - 100 100 100 100 97.56 100 64.44 100
[12]
Salles Machado et al., 2018 Brazil 53 89.32 90.57 - - 73.58 73.58 79.24 100
[17]
Rodriguez Paz et al., 2019 Denmark 49 89.7 - - - 86.3 - - -
[20]
De Almeida et al., 2020 [15] Brazil 165 93.7 98.6 93.7 98.6 92.4 100 43.5 98.5
Rodrigo de Oliveira Lopes et Brazil 78 - - - - 64.10 91.03 - -
al., 2023 [16]
Current study (right) Spain 74 89.13 95.12 89.13 95.12 76.47 94.87 52.24 94.29
Current study (left) Spain 74 87.5 97.62 87.76 97.67 83.64 95.65 49.25 90.91
% sexing means the percentage of specimens whose sex has been determined (p ≥ 0.95) while % accuracy means the percentage of specimens
whose sex has been correctly determined among those determined

Brazilian [15] samples. However, the low % sexing exhib- by de Almeida and colleagues [15] in a Brazilian sample
ited in French (41.66%) compared to the rest of analyzed (81.2%), is more similar to the results of 10, 8 or best 4
samples using the best four variables, ranging from 85.2 to variables in other geographical regions.
100%, is hard to explain, and is more similar to results from On the other hand, male-specific analysis is showed
the set of the worst 4 variables. Since some of the values in Table 10. Regarding the 10-variable and 4-best cases,
reported by Quatrehomme et al. [12] are internally incon- where more references were reported, current study %
sistent within their own tables, these results could have sexing are lower than the previous publications, except in
been caused by formatting typos and should be compared Mexican [14] and Brazilian populations [16, 17]. For %
with caution. In the case of the worst 4 variables, present accuracy, similar overall lower accuracy was achieved in
results are slightly lower than other studies [11, 12, 17]. It the current study when compared with other publications
is remarkable that the high percentage achieved in this case with the exception of the Brazilian population [16, 17]. As

13
International Journal of Legal Medicine (2025) 139:847–862 859

mentioned previously, Salles Machado et al. [17] results authors [10, 11] with the exception of VEAC, where the cur-
are anomalous and should be considered cautiously; e.g. % rent data was higher than the worldwide-based maximum
accuracy using the worst 4 variables (100%) is higher than threshold reported by Bruzek et al. [11]. Some research [62]
the value resulted from using 10 variables (90.57%). With stated that skeletal sexually dimorphic characters show inter
regards to the worst 4 variables results, present % sexing and intra-population variability and other authors stated that
results are higher than Northeastern Brazilian [15] and the pelvic-based sex determination may be impacted by popula-
worldwide sample utilised by Bruzek and colleagues [11] tion differences and sample variability [56, 63]. However,
but lower than the outcomes achieved for France [12] and other researchers advocate for the opposite, specifically
Southeast Brazil populations [17]. It is worthy to highlight regarding the pelvis [64]. In keeping with this, Bruzek and
the high percentage achieved by Salles Machado et al. [17] colleagues [11] stated that the pelvis shows a similar pat-
in a Brazilian sample (79.24%), comparable to the results tern of sexual dimorphism across diverse geographical areas
of 10, 8 or best 4 variables in other geographical regions. which appeared approximately 100–150 ky ago in early
The existence of an outlier outcome had been found and modern humans, defended by previous literature [65–69].
previously commented in Brazilian females, but in a differ- Thus, they offered a worldwide database (software DSP2)
ent sample [15]. These differences could be explained by of pelvic measurements to any anthropologist who needs
the great geographical extent and human diversity of this to sex skeletal remains, regardless of their geographical
country, likely existing intrapopulation differences between origin. This global reference not only comes from samples
Northeastern [15] and Southeast [17] Brazilian samples. To around the world but also from different ethnical groups
conclude the comparison, the current % accuracy results (Zulu, Soto, Afrikaner, African American, and European
for the worst 4 variables in males were lower than all the American) and temporal periods (from 18th to late 20th
revised literature ones [11, 12, 15, 17]. centuries), enhancing their potential applicability in miscel-
Comparing sexes across populations, in the case of the laneous populations. At this respect, it is worthy to stress
performance of the DSP methodology, both the sexing and that when reference models were used to determine the sex
accuracy results were better for females than males in most of geographically different target populations [10], accuracy
of the cases in the current sample. Higher % sexing, with results from 97.4 to 100% were displayed (Table 8). How-
less undetermined individuals in females, has already been ever, some large geographical areas were not considered in
reported in Mexicans [14], Danish [20], Brazilians [16] the reference sample used to calculate the posterior proba-
and the worldwide sample offered by Bruzek et al. [11]. bilities on DSP2, such as Central or South America. In addi-
Contrasting results were found for a French population tion, Africa and Asia were scarcely represented, with just
[12]. Furthermore, outcomes in favor of females or males Thai and South African samples analyzed. Although pos-
depending on the number of used variables, with no clear terior tests on Mexican [14] and Brazilian [15, 16] samples
sex-specific trend, were found in French [19], Brazilian reported comparable percentages of sexing and accuracy,
[15, 17], and Belgians [22]. According to % accuracy, on application to miscegenated Brazilian-identified sample
the other hand, the female predominance compared to males [17] yielded lower values compared to published literature
found here agreed with some Brazilian outcomes [15, 16] (Table 8). Furthermore, similar results arose when DSP2
and were also reported in pre-Columbian mummies [22]. and population-specific formulae were applied to a Greece
Besides, Bruzek et al. (worldwide sample) found more sample [13]. However, Kranioti et al. recommended popu-
accurate results in males in the first two subsamples (10 lation-specific formulations whenever possible to maximize
and 8 variables) and coinciding or similar outcomes with the outcomes. This existing inter-population variation was
the best and worst 4 variables subsamples, respectively [11]. also suggested by the application of DSP on Pre-Columbian
No differences were found in % accuracy between sexes in mummies [22], which garnered lower measurements for
Mexicans [14], and French [12, 19], where the systematic the DSP variables when compared to the original reference
shared results between sexes were 100%. Contrasting with ranges [11].
current findings, Gonzalez et al. examined sexual dimor-
phism in the great sciatic notch and ischiopubic complex in
a Portuguese sample [61], concluding females are misclas- Conclusions and future directions
sified more often than males.
This study investigated thoroughly the application of DSP2
Inter-population differences to a Spanish dry skeletal sample. Percentage of sexing was
high when 10, first 8 or 4 best variables were considered
Current DSP variables values were within the ranges and reduced to half when the 4 variables with least dis-
described by the worldwide sample used by the original crimination power, called worst 4 variables, were utilized.

13
860 International Journal of Legal Medicine (2025) 139:847–862

adaptation, distribution and reproduction in any medium or format,

Regarding % accuracy, however, they were high in all cases, as long as you give appropriate credit to the original author(s) and the
especially in females, regardless of the number of variables source, provide a link to the Creative Commons licence, and indicate
used. As the original authors described [10], the “worst” if changes were made. The images or other third party material in this
combination of four variables only refers to the decreasing article are included in the article’s Creative Commons licence, unless
indicated otherwise in a credit line to the material. If material is not
percentage of sexed specimens, while the accuracy remains included in the article’s Creative Commons licence and your intended
extremely high. The benefits and utility of the DSP method use is not permitted by statutory regulation or exceeds the permitted
seem to have driven to consolidate also in bioarchaeological use, you will need to obtain permission directly from the copyright
contexts [21–23, 25, 70–72]. However, their actual accu- holder. To view a copy of this licence, visit ​h​t​t​​p​:​/​/​​c​r​e​​a​t​i​​v​e​c​o​m​m​o​n​s​.​o​
r​g​/​l​i​c​e​n​s​e​s​/​b​y​/​4​.​0​/​​​​.​​​
racy on past populations is poorly understood, so further
comparison between DSP outcomes on genetic-based docu-
mented samples should be implemented in paleontological
References
and archaeological remains.
Despite being restrictive, the suggested posterior prob- 1. Krishan K, Chatterjee PM, Kanchan T et al (2016) A review of
ability of 0.95 seems to make the DSP reliable and accurate, sex estimation techniques during examination of skeletal remains
rendering it a meaningful tool in forensic practice. Besides, in forensic anthropology casework. Forensic Sci Int 261. ​h​t​t​​p​s​:​/​​/​d​
its high flexibility regarding the variables used gives it a o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​n​t​.​2​0​1​6​.​0​2​.​0​0​7​​​​.​ :165.e1-165.e8
2. Buikstra J, Ubelaker D (1994) Standards for data collection from
special place in forensic anthropology lab routinely working human skeletal remains. Ark Archaeol Surv Res Ser 44:44
on fragmentary or degraded samples. DSP has also yielded 3. Ferembach D, Schwidetzky I, Stoukal M (1980) Recomendations
good accuracy results when applied to CT scan images for age and sex diagnosis of skeletons. J Hum Evol 9:517–549
[18–20]. These virtual-based results open the possibility to 4. Ubelaker DH (1999) Human Skeletal Remains: Excavation,
Analysis, Interpretation. Washington
new population-specific tests, which may be performed in 5. White TD, Folkens PA (2005) The Human Bone Manual. Amster-
geographical areas hardly examined so far, such as Asia or dam; Boston
Africa, where documented collections are scarce or work- 6. Scheuer L (2002) Application of osteology to forensic medicine.
ing with skeletal remains is more difficult for international Clin Anat 15:297–312. https:/​/[Link]​g/10.10​02/ca​.10028
7. Bruzek J (2002) A method for visual determination of sex, using
researchers. the human hip bone. Am J Phys Anthropol 117:157–168. ​h​t​t​​p​s​:​/​​/​
d​o​​i​.​o​​r​g​/​1​0​.​1​0​0​2​/​a​j​p​a​.​1​0​0​1​2​​​​​​​
Supplementary Information The online version contains 8. Christensen AM, Passalacqua NV, Bartelink EJ (2019) Forensic
supplementary material available at ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​0​7​/​s​0​0​4​1​4​-​0​ Anthropology: current methods and practice. Academic
2​4​-​0​3​3​5​8​-​1​​​​​.​​ 9. Klales AR (2020) Chap. 2 - Practitioner preferences for sex esti-
mation from human skeletal remains. In: Klales AR (ed) Sex Esti-
Acknowledgements We would like to thank Dr. José Antonio Sán- mation of the Human Skeleton. Academic Press, pp 11–23
chez and his team for allowing us access to the collection housed by 10. Murail P, Bruzek J, Houët F, Cunha E (2005) DSP: a tool for
the School of Legal Medicine at the Faculty of Medicine of the Com- probabilistic sex diagnosis using worldwide variability in hip-
plutense University of Madrid (Madrid, Spain). bone measurements. Bull Mém Société Anthropol Paris 17:167–
176. https:/​/[Link]​g/10.40​00/bm​sap.1157
Funding Open Access funding provided thanks to the CRUE-CSIC 11. Brůžek J, Santos F, Dutailly B et al (2017) Validation and reli-
agreement with Springer Nature. ability of the sex estimation of the human os coxae using freely
available DSP2 software for bioarchaeology and forensic anthro-
pology. Am J Phys Anthropol 164:440–449. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​
Data availability N/A.
0​2​/​a​j​p​a​.​2​3​2​8​2​​​​​​​
12. Quatrehomme G, Radoman I, Nogueira L et al (2017) Sex deter-
Declarations mination using the DSP (probabilistic sex diagnosis) method on
the coxal bone: efficiency of method according to number of
Ethical approval N/A. available variables. Forensic Sci Int 272:190–193. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​
g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​n​t​.​2​0​1​6​.​1​0​.​0​2​0​​​​​​​
Informed consent N/A. 13. Kranioti EF, Šťovíčková L, Karell MA, Brůžek J (2019) Sex esti-
mation of os coxae using DSP2 software: a validation study of a
Greek sample. Forensic Sci Int 297. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​
Competing interests The authors have no competing interests to de-
c​i​i​n​t​.​2​0​1​9​.​0​2​.​0​1​1​​​​​. :371.e1-371.e6
clare that are relevant to the content of this article.
14. Sánchez-Mejorada G, Gómez-Valdés J, Herrera P et al (2011) Val-
oración Del método de Diagnóstico sexual Probabilístico (dsp)
Conflict of interest The authors declare that they have no known com- en una colección osteológica mexicana. Estud Antropol Biológica
peting financial interests or personal relationships that could have ap- 15. https:/​/[Link]​g/10.22​201/i​ia.14055066p.2011.42780
peared to influence the work reported in this paper. 15. De Almeida SM, De Carvalho MVD, De Lyra Menezes MCT et
al (2020) Validation of the DSP2 Tool in a contemporary identi-
Research involving human participant and/or animals N/A. fied skeletal Collection from Northeastern Brazil. Adv Anthropol
10:169–180. https:/​/[Link]​g/10.42​36/aa​.2020.102010
Open Access This article is licensed under a Creative Commons 16. Rodrigo de Oliveira Lopes A, Silva EML, da Nascimento MM
Attribution 4.0 International License, which permits use, sharing, S, et al (2024) DSP2 for sex determination of miscegenated

13
International Journal of Legal Medicine (2025) 139:847–862 861

contemporary hip bones. Anat Histol Embryol 53:e12979. ​h​t​t​​p​s​ Sci Int 317:110549. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​n​t​.​2​0​2​0​.​1​1​0​5​
:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​1​1​1​/​a​h​e​.​1​2​9​7​9​​​​​​​ 4​9​​​​​​​
17. Salles Machado MP, Costa ST, Freire AR et al (2018) Applica- 34. Viciano J, López-Lázaro S, Alemán I (2013) Sex estimation
tion and validation of Diagnose Sexuelle Probabiliste V2 tool in a based on deciduous and permanent dentition in a contemporary
miscegenated population. Forensic Sci Int 290. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​ Spanish population. Am J Phys Anthropol 152:31–43. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​
.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​n​t​.​2​0​1​8​.​0​6​.​0​4​3​​​​.​ :351.e1-351.e5 o​​r​g​/​1​0​.​1​0​0​2​/​a​j​p​a​.​2​2​3​2​4​​​​​​​
18. Chapman T, Lefevre P, Semal P et al (2014) Sex determination 35. Alcina M, Rissech C, Clavero A, Turbón D (2015) Sexual dimor-
using the probabilistic sex diagnosis (DSP: diagnose Sexuelle phism of the clavicle in a modern Spanish sample. Eur J Anat
Probabiliste) tool in a virtual environment. Forensic Sci Int 19:73–83
234:189e1–189e8. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​n​t​.​2​0​1​3​.​1​0​.​0​3​ 36. Ruiz Mediavilla E, Pérez B, Labajo González E et al (2016)
7​​​​​​​ Determining sex with the clavicle in a contemporary Spanish ref-
19. Mestekova S, Bruzek J, Veleminska J, Chaumoitre K (2015) A erence collection: a study on 3D images. Forensic Sci Int 261.
test of the DSP sexing method on CT images from a modern https:/​/[Link]​g/10.10​16/j.​forsciint.2016.01.029. :163.e1-163.e10
French sample. J Forensic Sci 60:1295–1299. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​ 37. Ruiz Mediavilla E, Labajo González E, Perea Pérez B et al (2017)
.​1​1​1​1​/​1​5​5​6​-​4​0​2​9​.​1​2​8​1​7​​​​​​​ Sex determination by bone volume in Spanish population: a study
20. Rodriguez Paz A, Banner J, Villa C (2019) Validity of the proba- on 3D images of talus, radius, clavicle and patella. Rev Médecine
bilistic sex diagnosis method (DSP) on 3D CT-scans from mod- Légale 8:188. https:/​/[Link]​g/10.10​16/j.​medleg.2017.10.027
ern Danish population. Rev Médecine Légale 10:43–49. ​h​t​t​​p​s​:​/​​/​d​ 38. García-Parra P, Pérez Fernández Á, Djorojevic M et al (2014)
o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​m​e​d​l​e​g​.​2​0​1​8​.​0​8​.​0​0​2​​​​​​​ Sexual dimorphism of human sternum in a contemporary Spanish
21. Rmoutilová R, Brůžek J, Gómez-Olivencia A et al (2024) Sex population. Forensic Sci Int 244. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​
estimation of the adult Neandertal Regourdou 1 (Montignac, n​t​.​2​0​1​4​.​0​6​.​0​1​9​​​​.​ :313.e1-313.e9
France): implications for sexing human fossil remains. J Hum 39. Macaluso PJ, Rico A, Santos M, Lucena J (2012) Osteometric
Evol 189:103470. https:/​/[Link]​g/10.10​16/j.​jhevol.2023.103470 sex discrimination from the sternal extremity of the fourth rib in a
22. Chapman T, Tilleux C, Polet C et al (2020) Validating the proba- recent forensic sample from Southwestern Spain. Forensic Sci Int
bilistic sex diagnosis (DSP) method with a special test case on 223. https:/​/[Link]​g/10.10​16/j.​forsciint.2012.09.007. ​:​3​7​5​.​e​1​-​3​7​5​.​
pre-columbian mummies (including the famous Rascar Capac). J e​5​​​
Archaeol Sci Rep 30:102250. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​j​a​s​r​e​p​.​2​0​2​ 40. Partido Navadijo M, Fombuena Zapata I, Borja Miranda EA,
0​.​1​0​2​2​5​0​​​​​​​ Alemán Aguilera I (2021) Discriminant functions for sex estima-
23. Villotte S, Santos F, Courtaud P (2015) In situ study of the Gra- tion using the rib necks in a Spanish population. Int J Legal Med
vettian individual from Cussac cave, locus 2 (Dordogne, France). 135:1055–1065. https:/​/[Link]​g/10.10​07/s0​0414-021-02537-8
Am J Phys Anthropol 158:759–768. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​0​2​/​a​j​p​a​.​ 41. Amores A, Botella MC, Alemán I (2014) Sexual dimorphism in
2​2​8​3​1​​​​​​​ the 7th cervical and 12th thoracic vertebrae from a Mediterranean
24. Oh KC, Hwang T, Choi GO et al (2022) Evaluation through the Population. J Forensic Sci 59:301–305. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​1​1​1​/​1​
Use of DSP2 Program for Sex Estimation by Measuring Human 5​5​6​-​4​0​2​9​.​1​2​3​2​0​​​​​​​
Hip bones in a Joseon Dynasty Bone Collection. Anat Biol 42. Azofra-Monge A, Alemán Aguilera I (2020) Morphometric
Anthropol 35:9–19. https:/​/[Link]​g/10.11​637/a​ba.2022.35.1.9 research and sex estimation of lumbar vertebrae in a contempo-
25. Jerković I, Bašić Ž, Kružić I, Anđelinović Š (2018) Creating ref- rary Spanish population. Forensic Sci Med Pathol 16:216–225.
erence data on sex for ancient populations using the probabilistic https:/​/[Link]​g/10.10​07/s1​2024-020-00231-6
sex diagnosis method: a validation test using the results of aDNA 43. Gama I, Navega D, Cunha E (2015) Sex estimation using the sec-
analysis. J Archaeol Sci 94:44–50. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​j​a​s​.​2​ ond cervical vertebra: a morphometric analysis in a documented
0​1​8​.​0​4​.​0​0​7​​​​​​​ Portuguese skeletal sample. Int J Legal Med 129:365–372. ​h​t​t​​p​s​:​
26. Villoria Rojas C, Mata Tutor P, Labajo González E et al (2023) /​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​0​7​/​s​0​0​4​1​4​-​0​1​4​-​1​0​8​3​-​0​​​​​​​
The identified skeletal Collection of the School of Legal Medi- 44. Gaya-Sancho B, Alemán Aguilera I, Navarro-Muñoz JJ, Botella
cine: a contemporary osteological collection housed in Universi- López M (2018) Sex determination in a Spanish population based
dad Complutense De Madrid, Spain. Int J Legal Med. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​ on sacrum. J Forensic Leg Med 60:45–49. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​1​
o​​r​g​/​1​0​.​1​0​0​7​/​s​0​0​4​1​4​-​0​2​3​-​0​3​0​4​7​-​5​​​​​​​ 6​/​j​.​j​f​l​m​.​2​0​1​8​.​1​0​.​0​0​1​​​​​​​
27. Bräuer G (1988) Osteometrie. Anthropologie, handbuch des ver- 45. Ruiz Mediavilla E, Perea Pérez B, Labajo González E et al (2012)
gleichenden biologie des menschen, Knussmann. Gustav Fischer, Determining sex by bone volume from 3D images: discriminating
Stuttgart, pp 160–232 analysis of the tali and radii in a contemporary Spanish reference
28. Gaillard J (1960) Détermination sexuelle d’un os coxal fragmen- collection. Int J Legal Med 126:623–631. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​0​7​
taire. Bull Mém Société Anthropol Paris 1:255–267. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​ /​s​0​0​41​ ​4​-​0​1​2​-​0​7​1​5​-​5​​​​​​​
r​g​/​1​0​.​3​4​0​6​/​b​m​s​a​p​.​1​9​6​0​.​1​1​4​5​​​​​​​ 46. Mastrangelo P, De Luca S, Alemán I, Botella MC (2011) Sex
29. Schulter-Ellis FP, Schmidt DJ, Hayek L-A, Craig J (1983) Deter- assessment from the carpals bones: discriminant function analysis
mination of sex with a Discriminant Analysis of New Pelvic Bone in a 20th century Spanish sample. Forensic Sci Int 206. ​h​t​t​​p​s​:​/​​/​d​o​​
measurements: part I. J Forensic Sci 28:169–180. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​ i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​n​t​.​2​0​1​1​.​0​1​.​0​0​7​​​​.​ :216.e1-216.e10
1​0​.​1​5​2​0​/​J​F​S​1​2​2​4​9​J​​​​​​​ 47. Clavero A, Salicrú M, Turbón D (2015) Sex prediction from the
30. Daniel WW (1998) Biostatistics: A Foundation for Analysis in the femur and hip bone using a sample of CT images from a Spanish
Health Sciences, 7th edn. Wiley population. Int J Legal Med 129:373–383. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​0​
31. Koo TK, Li MY (2016) A Guideline of selecting and reporting 7​/​s​0​04​ ​1​4​-​0​1​4​-​1​0​6​9​-​y​​​​​​​
Intraclass correlation coefficients for Reliability Research. J Chi- 48. Sobreira LB, Cunha E, Curate F (2023) Biological sex estimation
ropr Med 15:155–163. https:/​/[Link]​g/10.10​16/j.​jcm.2016.02.012 with femoral dimensions: a study of an adult sample from the
32. Amores-Ampuero A (2017) Sexual dimorphism in base of skull. osteological collection of Granada (Spain). Aust J Forensic Sci
Anthropol Anz 74:9–14. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​1​2​7​/​a​n​t​h​r​a​n​z​/​2​0​1​7​/​0​ 0:1–12. https:/​/[Link]​g/10.10​80/00​450618.2023.2270648
6​0​3​​​​​​​ 49. Peckmann TR, Meek S, Dilkie N, Rozendaal A (2016) Deter-
33. Daniele G, Matilde S-SA, María M et al (2020) Sex estimation by mination of sex from the patella in a contemporary Spanish
tooth dimension in a contemporary Spanish population. Forensic

13
862 International Journal of Legal Medicine (2025) 139:847–862

population. J Forensic Leg Med 44:84–91. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​ 62. Benazzi S, Maestri C, Parisini S et al (2008) Sex assessment from
1​6​/​j​.​j​f​l​m​.​2​0​1​6​.​0​9​.​0​0​7​​​​​​​ the acetabular rim by means of image analysis. Forensic Sci Int
50. Garcia S (2012) Is the circumference at the nutrient foramen of 180. https:/​/[Link]​g/10.10​16/j.​forsciint.2008.06.007. :58.e1-58.e3
the tibia of value to sex determination on human osteological col- 63. Karakas HM, Harma A, Alicioglu B (2013) The subpubic angle
lections? Testing a new method. Int J Osteoarchaeol 22:361–365. in sex determination: anthropometric measurements and analyses
https:/​/[Link]​g/10.10​02/oa​.1202 on Anatolian caucasians using multidetector computed tomogra-
51. Kranioti EF, Apostol MA (2015) Sexual dimorphism of the tibia phy datasets. J Forensic Leg Med 20:1004–1009. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​
in contemporary greeks, italians, and Spanish: forensic implica- 1​0​.​1​0​1​6​/​j​.​j​f​l​m​.​2​0​1​3​.​0​8​.​0​1​3​​​​​​​
tions. Int J Legal Med 129:357–363. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​0​7​/​s​0​0​4​ 64. Steyn M, Patriquin ML (2009) Osteometric sex determination
1​4​-​0​1​4​-​1​0​4​5​-​6​​​​​​​ from the pelvis—does population specificity matter? Forensic Sci
52. Saldías E, Malgosa A, Jordana X, Isidro A (2016) Sex estimation Int 191. https:/​/[Link]​g/10.10​16/j.​forsciint.2009.07.009. ​:​1​1​3​.​e​1​-​1​
from the navicular bone in Spanish contemporary skeletal collec- 1​3​.​e​5​​​
tions. Forensic Sci Int 267. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​n​t​.​2​0​1​ 65. Betti L (2014) Sexual dimorphism in the size and shape of the os
6​.​0​8​.​0​0​2​​​​​. :229.e1-229.e6 coxae and the effects of microevolutionary processes. Am J Phys
53. Barrio PA, Trancho GJ, Sánchez JA (2006) Metacarpal sexual Anthropol 153:167–177. https:/​/[Link]​g/10.10​02/aj​pa.22410
determination in a Spanish Population. J Forensic Sci 51:990– 66. Bruzek J, Murail P (2006) Methodology and reliability of sex
995. https:/​/[Link]​g/10.11​11/j.​1556-4029.2006.00237.x determination from the Skeleton. In: Schmitt A, Cunha E, Pin-
54. Djorojevic M, Roldán C, García-Parra P et al (2014) Morpho- heiro J (eds) Forensic Anthropology and Medicine: Complemen-
metric sex estimation from 3D computed tomography os coxae tary sciences from Recovery to cause of death. Humana, Totowa,
model and its validation in skeletal remains. Int J Legal Med NJ, pp 225–242
128:879–888. https:/​/[Link]​g/10.10​07/s0​0414-014-1033-x 67. Hager LD (1989) The evolution of sex differences in the Hominid
55. Brůžek J, Murail P, Houët F, Cleuvenot E (1994) Inter- and Intra- Bony Pelvis. University of California, Berkeley
observer Error in pelvic measurements and its implication for the 68. Hager LD (1996) Sex differences in the sciatic notch of
methods of sex determination. Anthropol 1962- 32:215–223 great apes and modern humans. Am J Phys Anthropol
56. Vacca E, Di Vella G (2012) Metric characterization of the human 99(199602):287–300. [Link]
coxal bone on a recent Italian sample and multivariate discrimi- ::AID-AJPA6>[Link];2-W
nant analysis to determine sex. Forensic Sci Int 222. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​ 69. Rosenberg K, Trevathan W (2002) Birth, obstetrics and human
r​g​/​1​0​.​1​0​1​6​/​j​.​f​o​r​s​c​i​i​n​t​.​2​0​1​2​.​0​6​.​0​1​4​​​​.​ :401.e1-401.e9 evolution. BJOG Int J Obstet Gynaecol 109:1199–1206. ​h​t​t​​p​s​:​/​​/​d​
57. Avent PR, Hughes CE, Garvin HM (2022) Applying posterior o​​i​.​o​​r​g​/​1​0​.​1​0​4​6​/​j​.​1​4​7​1​-​0​5​2​8​.​2​0​0​2​.​0​0​0​1​0​.​x​​​​​​​
probability informed thresholds to traditional cranial trait sex 70. Oelze VM, Koch JK, Kupke K et al (2012) Multi-isotopic anal-
estimation methods. J Forensic Sci 67(2):440–449. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​ ysis reveals individual mobility and diet at the early iron age
r​g​/​1​0​.​1​1​1​1​/​1​5​5​6​-​4​0​2​9​.​1​4​9​4​7​​​​​​​ monumental tumulus of magdalenenberg, Germany. Am J Phys
58. Pilmann Kotěrová A, Santos F, Bejdová Š et al (2024) Prioritiz- Anthropol 148:406–421. https:/​/[Link]​g/10.10​02/aj​pa.22063
ing a high posterior probability threshold leading to low error rate 71. Quintelier K (2009) Calcified uterine leiomyomata from a post-
over high classification accuracy: the validity of MorphoPASSE medieval nunnery in Brussels, Belgium. Int J Osteoarchaeol
software for cranial morphological sex estimation in a contempo- 19:436–442. https:/​/[Link]​g/10.10​02/oa​.971
rary population. Int J Legal Med 138:1759–1768. ​h​t​t​​p​s​:​/​​/​d​o​​i​.​o​​r​g​/​ 72. Scorrano G, Viva S, Pinotti T et al (2022) Bioarchaeological and
1​0​.​1​0​0​7​/​s​0​0​4​1​4​-​0​2​4​-​0​3​2​1​5​-​1​​​​​​​ palaeogenomic portrait of two pompeians that died during the
59. Fily M, Jaroslav B, Cunha E, Ludes B (2000) Researching ambig- eruption of Vesuvius in 79 AD. Sci Rep 12:6468. ​h​t​t​​​​p​​s​:​​/​​/​d​​o​i​​.​​o​​r​g​/​​1​
uous sex cases in ancient skeletons of the series of Coimbra (Por- 0​.​1​0​3​8​/​s​41​ ​5​9​8​-​0​2​2​-​1​0​8​9​9​-​1​​​​​​​
tugal). Prog Forensic Genet 8:558–560
60. Decker SJ, Davy-Jow SL, Ford JM, Hilbelink DR (2011) Virtual Publisher’s note Springer Nature remains neutral with regard to juris-
determination of sex: Metric and nonmetric traits of the adult dictional claims in published maps and institutional affiliations.
pelvis from 3D computed tomography Models*†. J Forensic Sci
56:1107–1114. https:/​/[Link]​g/10.11​11/j.​1556-4029.2011.01803.x
61. Gonzalez PN, Bernal V, Perez SI (2009) Geometric morphomet-
ric approach to sex estimation of human pelvis. Forensic Sci Int
189:68–74. https:/​/[Link]​g/10.10​16/j.​forsciint.2009.04.012

13