Accepted Manuscript

Title: Sex determination using the DSP (probabilistic sex

diagnosis) method on the coxal bone: efficiency of method
according to number of available variables

Author: Gérald Quatrehomme Irina Radoman Luı́sa Nogueira

Philippe du Jardin Véronique Alunni

PII: S0379-0738(16)30469-8
DOI: http://dx.doi.org/doi:10.1016/j.forsciint.2016.10.020
Reference: FSI 8630

To appear in: FSI

Received date: 15-6-2016

Revised date: 28-9-2016
Accepted date: 24-10-2016

Please cite this article as: G. Quatrehomme, I. Radoman, L. Nogueira, P. Jardin, V.

Alunni, Sex determination using the DSP (probabilistic sex diagnosis) method on the
coxal bone: efficiency of method according to number of available variables, Forensic
Science International (2016), http://dx.doi.org/10.1016/j.forsciint.2016.10.020

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
- The DSP method (probabilistic sex diagnosis) was applied to 100 coxal bones
- There was no mistake in sex determination of the 100 coxal bones
- The level of indetermination was higher in females than in males
- The level of indetermination varied a great deal according to the available combinations of variables

Sex determination using the DSP (probabilistic sex diagnosis) method on the coxal bone: efficiency of method

t
according to number of available variables

ip
Gérald Quatrehommea,b*gquatreh@unice.fr, Irina Radomana, Luísa Nogueiraa, Philippe du Jardinc, Véronique

cr
Alunnia,b

us
a
Laboratoire de Médecine Légale et d’Anthropologie médico-légale, Université de Nice Côte d’Azur, avenue de
Valombrose, 06107 Nice cedex 2, France
b
CEPAM, (UMR CNRS) n° 7264, Avenue de Valombrose, 06107 Nice cedex 2, France
c an
Edhec Business School, 393 Promenade des Anglais, BP 3116, 06202 Nice Cedex 3, France
M
Abstract
ed

The DSP method (probabilistic sex diagnosis) was applied to100 contemporary coxal bones from elderly
individuals of the South of France. Ten variables with a posterior probability greater or equal to a 0.95 threshold
were used. There was no statistical difference between right side and left side measurements. There was no
pt

mistake for sex assignment but the level of indetermination varied a great deal. It was higher in females than in
males. The best combinations were obtained when using all 10 variables, some combination of 9 variables
ce

(allexcept SS or SIS or VEAC) or the first 8 variables. We conclude that the DSP method is of great interest in
forensic anthropology, thanks to a very weak possibility of mistake when using the software for sex
determination of the coxal bone. The main drawback is the level of indetermination that can be high depending
Ac

on the available variables.

Keywords
Forensic anthropology population data; forensic sciences; DSP method; probabilistic sex diagnosis; Hip bone;
Coxal bone

1. Introduction
The coxal bone is considered to be the most dimorphic bone, a feature that is easily explained by the
requirements of parturition in females. Various methods of sex assessment have been described: (i) qualitative,
i.e. morphological methods dependent on the observation of various features (careful examination of features

Page 1 of 8
such as the subpubic angle, the great ischiatic notch, the subpubic concavity, the ventral arc, the double arc, the
pre-auricular and post-auricular sulcus, the pyramidal tubercle, the acetabulum, the iliac crest and the dorsal
pittings, among others allows for 90 to 95% of correct sex classification); (ii) semi-quantitative methods i.e.
scoring and (iii) quantitative methods. In 1969, Phenice [1] proposed a qualitative method based on the
observation of the ventral arch, the subpubic concavity and the medial area of the pubis ramus. Various methods
combining several morphological features have been described [2-11]. Ferembach et al.[2,3] created a score
based on 11 criteria enabling a classification into 5 groups (supra-males, males, indeterminate, female, supra-
female) and Bruzek [12] created a score to classify hip bones as "female", "male", or "intermediate" based on the

t
sacro-iliac complex (pre-auricular surface, great sciatic notch and the presence of a composite arch), and the

ip
ischio-pubis complex (inferior margin of the coxal bone and ischium-pubis proportions). These visual methods
are time-efficient but are still potentially subjective. Quantitative methods, using single measurements, indices

cr
(ratio of two measurements) or combined measurements (discriminant analysis, logistic regression...) are
proposed (e.g., [13,14]). Each method has advantages and drawbacks. Morphological and scoring methods

us
require a certain amount of expertise on the part of the observer, are population dependent and may be
ambiguous if the features are similar between males and females. Quantitative methods are easy to learn and
implement, and more objective. Nevertheless they are not always superior to the morphological methods in terms
of correct sex determination. an
To estimate the sexual dimorphism of the skeleton and its magnitude we used a method called probabilistic sex
diagnosis method (named DSP method or "diagnostic sexuel probabiliste" in French) devised by Murail et al.
M
[15]. Murail et al.’s study was based on the measurements of 2040 coxal bones from Europe, Africa, North
America and Asia, dating from the 18th to the mid-20th centuries. This method is very original in that it is based
on the Bayes theorem with posterior probabilities greater or equal to a 0.95 threshold. Ten anthropological
ed

distances are measured on the coxal bone. Free software is available describing the measurements with
photographs and short explanations. The 10 variables are classified in decreasing order of interest. At least 4of
the first 8 variables are required. When it is impossible to obtain 4 of the first 8 variables, the last two variables
pt

may be added up. The software then indicates if the coxal bone belongs to the male or female category and
calculates the probability associated with the result. The software also indicates when the coxal bone cannot be
ce

classified using the DSP method. The software returns an answer of male, female, or indeterminate. The authors
state that all of the coxal bones are correctly classified using the DSP method, that there is a percentage of
indetermination, and that the method is not dependent of bio geographical origin. Steyn and Patriquin [16] also
Ac

claimed that “population-specific formulae may not be necessary for coxal bones data coming from widely
differing populations”.
Nevertheless there are few confirmatory studies of the results of this method on other bone collections. Chapman
et al. [17] focused on a comparison between actual bones and virtual (CT Scan) bones and stated that 100% of
the coxal bones of their collection were correctly classified, after ruling out the indeterminates, except for the
worst combination of 4 variables (SIS, SA, SS, and VEAC) where one observer failed to correctly classify 2 out
of 21 specimens, and another observer failed to correctly classify the same two specimens out of 11. The
percentage of indetermination ranged from 2.56% to 46.15%, according to the combination of variables used.
Mestekova et al. [18] reported 100% correct classification and an indetermination level of 2.8 to 73.1%,
according to the combination of variables applied to CT scans of virtual coxal bones.

Page 2 of 8
Therefore the aim of this study was to use the DSP method on a contemporary collection of coxal bones of the
south of France in order to observe the percentage of coxal bones that would be correctly classified using this
method, to state the percentage of indeterminate bones within this collection and to find the most important
variables to avoid sex indetermination.

1. Material and methods

The coxal bones were collected from people who “had donated their body to science”, through a specific French
legislation allowing for teaching and research. This collection (called the Nice Bone Collection) is made of

t
skulls, long bones, coxae and some whole skeletons. The sex of all of the bones is perfectly documented. The

ip
bones come from elderly individuals born after 1920 and having died between 1998 and 2016. 100 right and left
coxal bones from 51 individuals were studied. Two individuals provided just one coxal bone.

cr
The 10 variables described by Murail et al. [15] were measured as recommended by these authors by an observer
blind to the sex of the coxal bones. The measurements, in decreasing order of interest, are defined by Murail et

us
al. [15] as:
- PUM (acetabulo-symphyseal pubic length): Minimum distance from the superior and medial point of the pubic
symphysis to the nearest point on the acetabular rim at the level of the lunate surface.
an
- SPU (cotylo-pubic breadth): Pubic breadth between the most lateral acetabular point and the medial aspect of
the pubis.
- DCOX (maximum pelvis height): Maximum height of the coxal bone measured from the inferior border to the
M
most superior portion of the iliac crest.
- IIMT (depth of the great sciatic notch): Distance from the postero-inferior iliac spine to the anterior border of
the great sciatic notch.
ed

- ISMM (post-acetabular ischium length): Distance from the most anterior and inferior point of the ischial
tuberosity to the furthest point of the acetabular border.
- SCOX (iliac breadth): Distance between the antero-superior iliac spine and the postero-superior iliac spine.
pt

- SS (spino-sciatic length): Minimum distance between the antero-inferior iliac spine and the deepest point in the
greater sciatic notch.
ce

- SA (spino-auricular length): Distance between the antero-inferior iliac spine and the auricular point
(intersection of the arcuate line with the auricular surface)
- SIS (cotylo-sciatic breadth): Distance between the lateral border of the acetabulum and the midpoint of the
Ac

anterior portion of the great sciatic notch.

- VEAC (vertical acetabular diameter): Maximum vertical diameter of the acetabulum, measured on the
acetabular rim, as the prolongation of the longitudinal axis of the ischium.

A comparison between the measurements of the right and left coxal bones was achieved (ANOVA). No intra or
inter-observer trial was achieved because the technical error of measurement is low [18] and excellent inter-
observer agreement of the measurements has already been demonstrated [17].

1. Results and discussion

Page 3 of 8
In this study we sought to confirm the results of the DSP method in terms of correct classification andnumber of
undetermined coxal bones, according to the number of variables. Our sample contained 100 coxal bones from 51
individuals (two individuals provided just one coxal bone).
Measurements of the right and left coxal bones were comparable and did not display any significant difference
(ANOVA with 1 factor, p=0.32 to 0.95, depending on the variable). Therefore some measurements may be
substituted by those obtained on the other side when all variables are not available.
The basic descriptive statistics are displayed in Table 1. As expected only the PUM and IIMT variables were
larger in females than in males, but there was no significant differences between males and females for

t
PUM,SCOX and SA.

ip
Various combinations of measures are easy to learn and to implement on bones and usually yield85-95% of
correct classification. In the current study, no matter what combination of variables was used, our main finding

cr
was that 100% of the 100 coxal bones were either correctly classified in terms of sex using the DSP method, or
classified as undetermined. There was no mistake in classification. Chapman et al. [17] had two specimens out of

us
21 (from observer 1) and two specimens out of 11 (from observer 2, working on a subsample of specimens also
evaluated by observer 1) that were incorrectly sexed; this occurred with the “worse combination” of four
variables (SS, SA, SIS, VEAC). The same two specimens were incorrectly sexed by both observers. This did not
an
occur in our series. But Chapman et al’s [17] study underlines that prudence is required when only the four
“worse” variables are available.
The main drawback of the DSP method is the number of indeterminate bones depending on the available
M
variables. This drawback is due to the choice of the authors [15] to use a threshold of 0.95 instead of the usual
0.50 threshold of discriminant analyses. This choice determines the high accuracy rate (usually 100% of correct
sex determination) but also a percentage of indeterminate bones that depends on the available variables and the
ed

degree of sexual dimorphism in a given population.

Table 2 displays the percentages of indeterminates in the current study and Table 3 the percentage of
indeterminates according to the data of the literature. In the current study the level of indetermination varied a
pt

great deal according to the measurements available. With all ten variables or with the first eight variables the
levels of indetermination were respectively 5.17%(N=58) and 5.08% (N=59). With 10 or 8 variables the
ce

percentage of indetermination was low in Chapman et al.’s [17]study (2.56%) but varied according to sex in
Mestekova et al.’s [18] study (7.7% for males and 2.8% for females) (Table 2).
Using only the “four best variables” (PUM, SPU, DCOX and IIMT), this percentage of indetermination
Ac

increases to 23.08% (N=65); and with the “worse combination” of 4 variables (SS, SA, SIS and VEAC), the
level of indetermination increases to 47.13% (N=87). The 4 variables deemed to provide the “worse
combination” (SS, SA, SIS and VEAC) is probably not the worse, as the variables concerning the central part of
the hip bone (IIMT, SS, SA, SIS) led to 67.41% of indetermination in the current study, and 73.1% of
indetermination in males and 63% in females in Mestekova et al’s [18] study.
It is also interesting to note the level of indetermination when one of the ten variables is not used: in the current
study the best results are found when DCOX, or SS, or SIS, or VEAC are discarded.
With a combination of the 3 first variables and one other, the level of indetermination is quite high in the current
study, between 17.19 and 23.08%. Conversely, with the combination of the first seven variables, the level of
indetermination is low (6.78%).

Page 4 of 8
Therefore the percentage of indetermination varies a great deal: between about 5% (when 10 variables or some
combination of 7 to 9variables are used), up to about 50% or even more, for some combinations (“four worse”
and “four central part”). The percentage of indetermination varies also according to the observer: Chapman et al.
[17] found (for observer 1 and observer 2, respectively) a level of indetermination of 2.56% and 0% (using all 10
variables or just the first 8 variables), 10.26% and 10% (4 best variables), 7.69% and 0% (5 first variables),
46.15% and 18.18% (4 worse variables).
Remarkably there was no indetermination with several combinations of variables in males in the current study:
10 variables, most of the combinations of 9 variables, 8 variables (without SIS and VEAC), 5 to 7 variables and

t
even some combinations of 4 variables (Table 2).If we analyze the results it is obvious that most of the

ip
indeterminate coxal bones come from females (Table 2). This result was also obtained by Mestekova et al.
[18]for one combination of variables, namely the “best four variables”. Conversely Chapman et al’s [17] found a

cr
higher level of indetermination in males than in females with a combination of 10 variables, 8 first variables, and
4 “central” variables. In the current study it would appear that male coxal bones are more easily classified and

us
that female coxal bones may be more ambiguous. This is an interesting fact but we have no clear explanation.
We have observed that the minimum measurements in the Nice Bone Collection for each of the ten variables was
always higher than that observed by Murail et al. [15], whereas the maximum observed in our collection was
an
nearly always less than that of the Murail’s sample (for all variables except SPU). In other words, we observed a
narrower range of measurements in our collection than in Murail et al’s sample. As such, sex determination is
probably more difficult in our collection. We make the hypothesis that the female coxal bones are closer to that
M
of the males in our collection, in the South of France area. Murail et al’s sample came from several geographical
regions so that no comparison of the basic descriptive statistics is possible.
The average age of the sample in the Nice Bone Collection is high, because the bones belong to individuals who
ed

“donated their body to science”. Old age may create difficulties and discrepancies when sexing bones (e.g.,
[19]). Nevertheless all of the coxal bones were correctly classified in this study.
pt

In conclusion, the DSP method is very efficient in forensic anthropology, because the risk of error is probably
extremely weak. If we accept a level of indetermination of 5% or less, we can use several combinations of
ce

variables: 10 variables, 9 variables (without SS or SIS or VEAC), or the first 8 variables. If we accept a level of
indetermination of less than 10%, combinations of the first 6 or 7 variables may also be acceptable. We must be
careful when only some combinations of four variables are available. The level of indetermination was higher in
Ac

females than in males in the current study. The best combinations of variables (translated by the lowest level of
indetermination) were the combination of 8 to 10 variables and the worst combinations were the four variables
representing the central part of the hip bone (IIMT, SS, SA, SIS).

1. Conflict of interest
There is no conflict of interest in this work.

References
[1] TW Phenice, A newly developed visual method of sexing the os pubis, Am J PhysAnthropol. 30 (1969) 297-301.
[2] D Ferembach, I Schwidetzky, M Stloukal, Recommandations pour déterminer l'âge et le sexe sur le squelette, Bull
Mém Soc Anthropol Paris. T6, série XIII (1979) 7-45.

Page 5 of 8
[3] D Ferembach, I Schwidetzky, M Stloukal, Recommandations for age and sex diagnoses of skeletons, J Hum
Evol. 9(1980) 517-549.
[4] MY Đşcan, K Derrick, Determination of sex from the sacro-iliac joint: a visual assessment technique, Florida Sci.
47 (1984) 94-98.
[5] MY Đşcan, SS Dunlap, Sexing the human sacro-iliac joint, Am J PhysAnthropol.60 (1983) 208-209.
[6] G Acsádi, J Nemeskéri, History of Human Lifespan and Mortality.AkadémiaiKiadó, Budapest,1970, 7, pp.
126-127.
[7] J Nemeskéri, L Harsanyi, G Acsádi, Merthodenzur Diagnose des Lebensalters von Skelettfunden,

t
AnthropologischerAnzeiger.24 (1960)70-95.

ip
[8] J Bruzek, D Ferembach, Fiabilité de la méthode visuelle de détermination du sexe à partir du bassin du
“Groupe de travail d’Anthropologues Européens”. Application surl’oscoxal, Arch AnthropolEthnol.72 (1992)

cr
145-161.
[9] V Novotný, Sex determination of the pelvis bone: a system approach, Anthropologie (Brno). 24 (1986) 197-206.

us
[10] T Rogers, S Saunders, Accuracy of sex determination using morphological traits of the human pelvis,J
Forensic Sci. 39 (1994) 1047-1056.
[11]K Krishan, PM Chatterjee, T Kanchan, S Kaur, N Baryah, RK Singh, A review of sex estimation techniques during
an
examination of skeletal remains in forensic anthropology casework, Forensic Sci Int. 261 (2016) 165.e1-165.e8.
[12] J Bruzek,A method for visual determination of sex, using the human hip bone, Am J PhysAnthropology.117
(2002) 157-168.
M
[13] WW Howells,Détermination du sexe du bassin par fonction discriminante: étude du matériel du Dr Gaillard, Bull
Mém Soc Anthropol Paris. XI série (7) (1965) 95-105.
[14] E Vacca, G Di Vella, Metric characterization of the human coxal bone on a recent Italian sample and multivariate
ed

discriminant analysis to determine sex, Forensic SciInt. 222 (2012) 401.e1-401.e9.

[15] P Murail, J Bruzek, F Houët, E Cunha, DSP: a tool for probabilistic sex diagnosis using worldwide variability in
hip bone measurements, Bull et MémSocd’Anthropol de Paris. t. 17 (3-4) (2005) 161-76.
pt

[16] M Steyn, ML Patriquin, Osteometric sex determination from the pelvis-Does population specific matter?, Forensic
Sci Int. 191(2009) 113.e1-113.e5.
ce

[17] T Chapman, P Lefèvre, P Semal, F Moiseev, V Sholukha, S Louryan, M Rooze, S Van Sint Jan, Sex
determination using the Probabilistic Sex Diagnosis (DSP: Diagnose SexuelleProbabiliste) tool in a virtual
environment, Forensic Sci Int. 234 (189) (2014) e1-8.
Ac

[18] S Mestekova, J Bruzek, J Veleminska, K Chaumoitre, A test of the DSP sexing method on CT images from
a modern French sample, J Forensic Sci. 60 (5) (2015) 1295-9.
[19] NC Lovell, Test of Phenice’s method for determining sex from the bone pubis, Am J PhysAnthropol. 79(1989)
117-120.

Page 6 of 8
Table1 Descriptive statistics and comparison of males vs females (average of right and left coxal bones)
(wilcoxon.test)

Variables Males Females p

mean (Sd, N) (mean, Sd, N)
PUM 72.15 (4.92,41) 72.92 (4.93,24) NS
SPU 33.00 (2.33,41) 27.61 (1.71,24) <0.0000
DCOX 225.25 (10.35,41) 205.47 (9.49,24) <0.0000
IIMT 41.76 (5.17,41) 46.11 (3.18,24) <0.001
ISMM 118.77 (5.24,23) 103.51 (4.27,22) <0.0000
SCOX 162.53 (9.98,21) 159.51 (8.54,22) NS

t
SS 79.35 (5.68,23) 71.82 (4.93,24) <0.0000

ip
SA 80.94 (5.64,23) 80.70 (5.87,24) NS
SIS 43.68 (3.32,23) 39.65 (3.73,22) <0.001
VEAC 57.93 (3.33,23) 50.48 (2.55,22) <0.0000

cr
us
Table 2 Correct sex assessment and non determination (ND) of coxal bones*

10
Combination of
variables
N
(M/F/Total)

37/21/58
Correct sex
assessment
(M/F/Total
31/18/55
an
ND (M/F)

0/3
ND (%)
(M+F)

5.17
9 without PUM 42/32/74 41/24/65 1/8 12.16
M
9 without SPU 38/21/59 36/18/54 2/3 8.47
9 without DCOX 38/21/59 38/17/55 0/5 8.47
9 without IIMT 38/21/59 38/16/54 0/5 8.47
9 without ISMM 38/22/60 38/14/52 0/8 13.33
ed

9 without SCOX 40/24/64 40/12/58 0/6 9.37

9 without SS 38/21/59 38/18/56 0/3 5.08
9 without SA 38/21/59 38/16/54 0/5 8.47
9 without SIS 38/21/59 38/18/56 0/3 5.08
pt

9 withoutVEAC 38/21/59 38/18/56 0/3 5.08

8 without SIS et VEAC 38/21/59 38/18/56 0/3 5.08
8 without DCOX et 40/24/64 39/19/58 1/5 9.38
SCOX
ce

7 without SA, SIS et 38/21/59 38/17/55 0/4 6.78

VEAC
6first 37/21/58 40/16/56 0/8 12.50
5 first 40/24/64 40/16/56 0/8 12.50
Ac

4 best 41/24/65 40/10/50 1/14 23.08

4 (central part) 47/42/89 24/4/29 22/38 67.41
4 worse 45/42/87 29/17/46 16/25 47.13
3+IIMT 40/25/65 10/10/50 0/15 23.08
3+ISMM 40/24/64 40/13/53 0/11 17.19
3+SCOX 39/21/60 39/9/48 0/12 20.00
3+SS 42/23/65 41/10/51 1/13 21.54
3+SA 43/21/64 42/11/53 1/10 17.19
3+SIS 42/23/65 41/10/51 1/13 20.00
3+VEAC 41/24/65 40/10/50 1/14 36.59
* M:males ; F: females ; ND: non determination of sex

Page 7 of 8
Table 3 Percentages of indeterminate coxal bones according to the data of the literature in comparison with the
current study (M: males; F: females)
Combinations of Chapman et al. [7] Mestekova et al. [8] Current study (N=100)
variables (N=39) * (N=106 ) (M; F; Total)
Researcher 1 (N=39)/
Researcher 2 (N=20)
10 variables 2.56/5.0 7.7 (M); 2.8 (F) 0.0 (M); 5.17 (F)

t
9 variables - - Between 5.08 and 13.33
(see Table 2)

ip
8 first variables (without 2.56/5.0 - 0.0 (M); 5.08 (F); 5.08
SIS and VEAC)
8 first variables (without - 11.5 (M); 3.7 (F) 1.56 (M); 7.81 (F); 9.37

cr
DCOX and SCOX)
7 variables (all except - - 0.0 (M); 6.78 (F); 6.78
SA, SIS and VEAC)

us
6 variables (all except - - 0.0 (M); 6.90 (F); 6.90
SS, SA, SIS and VEAC)
5 variables (Dcox, PUM, 7.69/0.0 - 0.0 (M); 12.5 (F); 12.50
SPU, IIMT, ISMM)
Best 4 variables (PUM,
SPU, DCOX, IIMT)
4 variables (central part
10.26/10.0

-
an
3.8 (M); 14.8 (F)

73.1 (M); 63.0 (F)

1.54 (M); 21.88 (F);
23.08
24.72 (M); 42.70 (F);
of hip bone) (IIMT, SS, 67.41
M
SA, SIS)
Worst 4 variables (SS, 46.15/18.18 - 18.39 (M); 28.73 (F);
SA, SIS, VEAC) 47.13
3 first variables and - - Between 17.19 and 23.08
another one (see Table 2)
ed

* Researcher 2 studied a subsample of 20 randomly chosen coxal bones from the first researcher’s sample of 39
bones.
pt
ce
Ac

Page 8 of 8