J Forensic Sci, 2018

doi: 10.1111/1556-4029.13972
PAPER Available online at: onlinelibrary.wiley.com

ANTHROPOLOGY

Allysha Powanda Winburn , Ph.D.

Validation of the Acetabulum As a Skeletal

Indicator of Age at Death in Modern
European-Americans†,‡

ABSTRACT: Progressive changes in the acetabulum have been used in modern skeletal age estimation, but they have not been completely
understood. If their age correlations are weakened by the influence of factors like physical activity and obesity, acetabular changes should not
be used for age estimation. To investigate their utility for aging, the acetabular variables of Rissech et al. (2006) were analyzed in 409 modern
European-Americans (Bass Collection, Tennessee). Correlation tests assessed potential associations between acetabular data, osteoarthritis scores
(collected per Jurmain, 1990), and documented demographic information (age, body mass index [BMI], metabolic intensity of physical activi-
ties). Acetabular changes had statistically significant, positive correlations with osteoarthritis (p < 0.001 in most joints/regions) and age
(p < 0.001), indicating their degenerative nature and relevance for age estimation. Acetabular changes showed no associations with BMI or
metabolic values, suggesting resistance to obesity and activity effects. These results suggest that acetabular degeneration is a valid skeletal age-
at-death indicator.

KEYWORDS: forensic science, forensic anthropology, age estimation, pelvis, skeletal degeneration, arthritis, obesity, physical activity

The acetabulum has been a focus of age-estimation research
for over a decade, and several acetabulum-based aging methods
have been proposed (1–3). However, the progressive changes
that occur in the acetabulum remain poorly understood. These
changes may constitute skeletal metamorphoses (akin to the for-
mation of the ventral rampart of the pubic symphysis), which
are generally believed to be linked with age. Alternately, acetab-
ular changes may represent osteoarthritis (OA)—generally
viewed as degenerative, less strongly correlated with age, and
more influenced by factors like physical activity and obesity.
This research investigated the nature of progressive changes in
the acetabulum. It aims to determine whether acetabular changes
are metamorphic or degenerative, ascertain whether they are
valid indicators of age at death, and scrutinize the overall utility
of a dichotomous, “metamorphic vs. degenerative” approach to
age estimation. By focusing on the skeletal biology underlying
acetabular changes, this research contributes to an understanding
of acetabular age progression that will enhance future method
testing and validation. Before acetabular aging methods continue
to be developed and refined, the fundamental question of
1
Department of Anthropology, University of West Florida, 11000 Univer-
sity Parkway, Bldg. 13, Pensacola, FL 32514.
Corresponding author: Allysha Powanda Winburn, Ph.D. E-mail: awinburn
@uwf.edu
†
Presented in part at the 70th Annual Scientific Meeting of the American
Academy of Forensic Sciences, February 19-24, in Seattle, WA.
‡
Financial support for this study was provided by the National Institute of
Justice Graduate Research Fellowship Program in Science, Technology,
Engineering, and Mathematics (Grant #: 2015-R2-CX-0009) and the Univer-
sity of Florida Graduate School Doctoral Research Travel Award.
Received 25 July 2018; and in revised form 8 Oct. 2018; accepted 13
Nov. 2018.
whether this joint surface is a valid age indicator must be
answered.
Background
Accurate and precise methods of skeletal age-at-death estima-
tion are vital to the generation of the biological profiles that
enable individual identifications. Broad age intervals contribute
little to the elimination of potential matches from the universe of
missing-persons, while narrow intervals may exclude the correct
missing-person from consideration. Still, adult age estimation
remains challenging.
Forensic anthropologists frequently examine joint surfaces to
inform their adult age estimates (1–42). All joints change with
age, but some exhibit metamorphic change, while others exhibit
degenerative change (39). Metamorphic skeletal changes occur
after the attainment of skeletal maturity but independent of
skeletal degeneration. They are often portrayed as more relevant
to age estimation than degenerative changes, since the latter pro-
cesses are complicated by intra-individual variation and inter-
individual differences—both hereditary and environmental (e.g.,
nutrition, bone density, mechanical loading, hormonal influences
[39,43,44]). The degenerative processes that complicate adult
age estimation can become more pronounced and variable over
an individual’s lifetime, and this trajectory effect (45) can trans-
late to further-decreased accuracy with advancing age (43,46; cf.
47). Thus, age estimates for elderly adults are often more diffi-
cult to achieve, less accurate, and less precise than age estimates
for younger adults, and anthropological aging methods tend to
underestimate age in the oldest individuals (46,48,49).
Yet, due to advances in health care and standards of living,
modern individuals are more likely than past peoples to reach

© 2018 American Academy of Forensic Sciences 1

2 JOURNAL OF FORENSIC SCIENCES
advanced ages at death (50). This presents a particular problem
for forensic anthropologists, who face the challenge of providing
the accurate age estimates that enable identifications for a world
population with a rising mean age at death. According to
missing-persons data curated by the FBI’s National Crime Infor-
mation Center—a dataset approximating the universe of uniden-
tified decedents analyzed by forensic scientists—of 43,646
active adult missing-persons cases, approximately 50% of miss-
ing adult females and 65% of missing adult males are 40 years
of age or older (51).
Forensic anthropologists must identify skeletal indicators of
adult aging processes and validate methods based on these indi-
cators, so that they can estimate age with accuracy and precision
even in older individuals. Much research on adult age estimation
has centered on the pelvis, due to the late-onset age-related
changes that occur in its amphiarthrodial joints (the relatively
immobile pubic symphysis and sacro-iliac joint). The age-related
changes of the pubic symphysis have been studied for over a
century (4,5,9,10,12–14,16,21,36,37,41). In particular, the meta-
morphic changes of the ventral rampart have been emphasized
for their role in enabling accurate and precise age estimates in
adults younger than 40 years (5,16). However, degenerative
changes of the pubic symphysis are also informative for estimat-
ing age in older individuals (4,14). Unfortunately, the often-poor
preservation of the pubic symphysis in adverse depositional
environments undermines its utility as an age indicator (7,18).
The auricular surface of the ilium has also been widely studied
(7,11,15,18,23–26,33,35,36). However, iliac auricular surface
methods either tend to downplay the variability of age-related
change in the joint (18) or classify it into broad phases resulting
in imprecise age estimates (26). The auricular surface of the
sacrum has also been considered for age estimation, though age-
related changes are even less pronounced in this region than in
the iliac auricular surface (6,27,28). The auricular surfaces lack
regions of metamorphic change analogous to the ventral rampart
of the pubic symphysis, but it is unclear whether the difficulties
of auricular surface aging stem from the degenerative nature of
their age changes.
Recently, biological anthropologists have also begun to
explore the age-estimation potential of the acetabulum (1–
3,8,19,20,22,29–34,40,42). The acetabulum is a robust joint with
a relatively protected anatomical location (8,31) and the potential
to survive the burial and fragmentation scenarios common in
forensic cases (42,52). Accurate acetabular age estimates have
been reported, even for older individuals (2,8,30). Evidence that
the changes observed in this joint are not degenerative but meta-
morphic—akin to the tightly age-correlated changes of the pubic
symphysis ventral rampart—might also constitute an argument
for the use of the acetabulum in age estimation. However, after
more than a decade of research, the nature of acetabular changes
(i.e., metamorphic vs. degenerative) remains obscure. Further,
the metamorphic vs. degenerative dichotomy itself is still rela-
tively unexplored. Thus, it also remains unclear to what degree,
if any, metamorphic changes truly outperform degenerative
changes as skeletal indicators of age.
Several factors have the potential to complicate the straightfor-
ward age correlation of the progressive changes observed in the
acetabulum. Like the pubic symphysis and auricular surface,
the acetabulum participates in the transfer of body mass from
the upper body to the lower limb. It is thus possible that obesity
—an excess of adipose tissue—may affect acetabular changes
(53). Unlike the amphiarthrodial joints of the pelvis, however,
the acetabulum is diarthrodial, mobile, and used in daily physical
activities. Thus, it is also possible that differing levels of physi-
cal activity can influence the way the acetabulum changes with
age. Further, diarthrodial joints commonly develop OA—the
breakdown of normal relationships among articular tissues that is
often visible in skeletal remains (via osteophytes, eburnation,
and ankylosis). Biological anthropologists have long noted that
OA increases with age (35,39); certainly, elderly individuals suf-
fer more frequently from OA than younger individuals (54).
However, the correlation between advancing age and increasing
OA is generally viewed as weak, due to the many other factors
influencing OA development (17,43). These factors include both
vigorous physical activity (55–57) and obesity (58–63). If the
changes observed in the acetabulum are akin to the processes of
OA observed in the other diarthrodial joints, then acetabular
changes could be strongly correlated with factors like activity
and obesity rather than with age.
The effects of activity and obesity on the age-related changes
of the acetabulum—and the relationship of these changes with
OA—have not been investigated. Before forensic anthropologists
can responsibly use the acetabulum to estimate age, research
must ascertain the following: Are acetabular changes affected by
patterns of physical activity and obesity? Are the changes truly
metamorphic, or are they degenerative? What does that mean in
terms of their correlation with age? If they merely constitute
degenerative change, are they still useful for adult age estima-
tion? Thus, this research aims to illuminate the potential utility
or unreliability of acetabular age changes and critically evaluate
the dichotomy between metamorphic and degenerative changes
for skeletal age estimation.
Materials and Methods
Skeletal Sample
In order to investigate these problems, acetabular changes and
OA were analyzed in a sample of European-American skeletal
individuals from the W.M. Bass Donated Skeletal Collection
(University of Tennessee [UT], Knoxville) with documented
antemortem data on age, ancestry, sex, height, body mass, and
habitual/occupational activities. A stratified random sample was
assembled from individuals previously determined to meet the
study’s documentary requirements. These 409 European-Ameri-
can females (n = 198) and males (n = 211) populated eight age
groups approximating 10-year intervals (Table 1, Fig. 1). A goal
of 30 individuals per age cohort was set for both females and
males. However, because of the demographic makeup of the
Bass Collection (fewer females than males; fewer extremely
young and extremely old individuals), some of these 10-year age
categories could not be fully populated (see Table 1; also see
below discussion of the study’s ancestral limitations).
Data Collection
For each individual in the sample, left and right acetabula
were observed and scored using the method developed by Ris-
sech et al. (2), in which seven acetabular variables are assigned
ordinal scores which can range from 0–3 to 0–6 (Table 2,
Fig. 2). In order to enable parametric, multivariate statistical test-
ing of normally distributed acetabular scores, composite acetabu-
lar scores (CAS) were also created (19). The ordinal acetabular
variable scores were transformed into a common scale of 0–10,
so that each variable would be weighted equally. For example, if
a variable received a score of 1 on a scale of 0–3 (2), this was
 WINBURN . VALIDATION OF THE ACETABULUM FOR AGING 3

TABLE 1––Total sample of European-American adults (Bass Collection,

10 15 20 25 30 35
UT) analyzed in the current study, arranged by age group, with mean and
median ages.

Sex Age Group n Mean Age Median Age

Count
Female 20–29 3 27 27
30–39 14 35.7 36.5
40–49 33 45.3 46
50–59 33 54.7 55
60–69 33 64.8 66
70–79 33 74.7 75

5
80–89 33 83.9 84
90+ 16 92.4 92

0
Female total 198 64.3 65.5
Male 19–29 10 24.3 25.5 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90+
30–39 32 34.9 35
40–49 34 45.4 45.5 Female Age Group (Years)
50–59 31 54.5 55
60–69 32 64.2 65
70–79 33 73.9 74
80–89 33 83.5 83

35
90+ 6 94.7 95
Male total 211 58.8 59

30
Total 409 61.5 61

25
transformed to 3.33 (19). Transformed scores were summed and
20
Count
averaged to achieve a CAS for each individual.
Evidence of OA was also observed and scored in the major 15
appendicular joints (shoulder, elbow, wrist, hip, knee, ankle) and
10

the temporomandibular joint (TMJ). Using the Jurmain (64) ordi-

nal scoring system (Fig. 3), each element in a joint was assigned
5

an individual score (e.g., acetabulum, proximal femur), and then,

0

the individual scores were added and divided by the number of

available elements in order to achieve a combined joint score (e.g.,
“left hip” = [left femoral head score + left acetabulum score]/2).
Left and right combined joint scores were initially kept separate.
However, subsequent Wilcoxon rank-sum testing indicated no sta-
tistically significant differences between left and right mean joint
scores for females (largest difference 0.04) or males (largest differ-
ence 0.08). In 14 Wilcoxon rank-sum tests (significance level,
a = 0.004 with Bonferroni correction), female p-values ranged
from 0.42 to 0.74 (Spearman’s rho values ranged from 0.641 to
0.748), and male p-values ranged from 0.09 to 0.90 (rho values
ranged from 0.571 to 0.774). Thus, left and right joint scores were
combined and averaged following the protocol of Weiss (65). In
the case of missing data (i.e., damaged or missing articular sur-
faces), fewer surfaces were used in the calculation of the combined
joint variable. The OA scores were used in the analyses of individ-
ual joints (as ordinal data) and averaged to achieve overall, upper
limb, and lower limb OA scores (which approximated continuous
data and were normally distributed).
Coding Demographic Data
Subsequent to data collection, demographic data (e.g., age,
biological sex) were re-associated with each individual in the
sample. Documented height and body mass data were used to
calculate BMI using the following formula: BMI = mass [kg]/
height [m]2. Resulting BMI values were then categorized accord-
ing to National Institutes of Health standards current as of 2017:
obese (≥ 30); overweight (25–29.9); normal-weight (18.5–24.9);
and underweight (< 18.5).
All occupational and habitual physical activities documented
for the individuals in the sample were assigned a metabolic
equivalent (MET) intensity level using the activity codes and
MET values listed in the Compendium of Physical Activities
19-29 30-39 40-49 50-59 60-69 70-79 80-89 90+
Male Age Group (Years)
FIG. 1––Sample distribution by age group and sex (overall n = 409;
female n = 198; male n = 211).
(66). A MET value consists of the measured or estimated meta-
bolic rate of a given activity divided by the resting metabolic
rate for quiet sitting, classified as 1 MET (67,68). In this classifi-
cation system, a 3-MET activity involves three times the energy
expenditure of a 1-MET activity. For example, outdoor construc-
tion work is estimated as a 4-MET activity, and conducting
seated office work is estimated as a 1.5-MET activity (66). The
current version of the Compendium of Physical Activities con-
tains MET values for 821 activities, many of which were
obtained from published measurements (66).
The Compendium is typically used to assign standardized
codes to the physical activities self-reported by survey partici-
pants, in order to facilitate comparison among studies. When the
activity data recorded for a Bass Collection individual were
specific (e.g., describing a sport played during life), individual
activities and occupations were assigned to specific Com-
pendium categories. Even when reported activities were less
specific (e.g., reporting “exercise” rather than a specific sport),
they were sufficient to allow coding using a modified protocol
(68). Using this protocol, all associated example activities shar-
ing similar Compendium codes were averaged to create an esti-
mated MET value for the overall category (68). If an individual
listed more than one activity, the aggregate MET values for each
listed activity were averaged to create an overall estimated MET
level for that individual. This coding system was applied to both
reported occupations and habitual-activities. In order to enable
4 JOURNAL OF FORENSIC SCIENCES

TABLE 2––Descriptions of the seven acetabular variables and their states, summarized from Rissech et al. (2).

Variable Description of Variable Characteristics of the States

1. Acetabular groove Groove surrounding the internal acetabular rim Younger: no anatomical interruption
between lunate surface and rim.
Older: pronounced groove.
States 0–3
2. Acetabular rim shape Outer rim of the acetabulum Younger: dense, round and smooth.
Older: rough, with osteophytic growth.
States 0–6
3. Acetabular rim porosity Texture and activity of the outer rim Younger: smooth, little porosity.
Older: macroporosity and destruction.
States 0–5
4. Apex activity Osteophytic activity on the apex of the posterior cornu Younger: round, smooth apex.
Older: osteophytic growth of one (sometimes both)
horns of the lunate surface.
States 0–4
5. Fossa edge activity Crest formation between fossa and lunate surface Younger: smooth, no delineation between surfaces
Older: visible crest.
States 0–5
6. Fossa activity Level and activity of the acetabular fossa Younger: dense, level acetabular fossa.
Older: irregular with much activity.
States 0–5
7. Fossa porosity Texture and integrity of the acetabular fossa Younger: dense, smooth fossa.
Older: macroporosity and destruction.
States 0–6

interactions of age, activity, and obesity with acetabular changes

(Hypotheses 2, 3, and 4), and the relative contributions of these
factors to acetabular changes (Hypothesis 5). The following
hypotheses were designed to assess whether acetabular changes
are metamorphic or degenerative (Hypothesis 1) and whether
they are relevant to skeletal age estimation (Hypotheses 2
through 5).

Hypothesis 1 Acetabular Changes Correlate Positively with OA

Individuals with higher CAS (19) and higher individual
acetabulum aging scores for the seven variables (2) were
predicted to exhibit more OA in the other joints of the
body. Transformed CAS and scores for each of the seven
individual acetabular variables were tested for positive associa-
tions with OA scores (64) in individual joints and combined
joint regions (overall OA, upper limb OA, and lower limb
OA).

Hypothesis 2 Acetabular Changes Correlate Positively with Age

FIG. 2––Acetabular variables scored in the method of Rissech et al. (2):
the acetabular groove (Variable 1); acetabular rim shape and porosity
(Variables 2 and 3); activity on the apex of the posterior cornu (Variable 4);
crest formation on the edge of the acetabular fossa (Variable 5); and acetab-
ular fossa activity and porosity (Variables 6 and 7). Photograph by the
author.
Older individuals were expected to exhibit “older-looking”
acetabula. Transformed CAS and scores for each of the seven
individual acetabular variables were tested for positive associa-
tions with age.

an assessment of the impact of strenuous activities utilizing the

lower limb, habitual physical activities involving the lower limb
(e.g., running, wrestling) were noted during scoring so that they
could be tested separately.
Hypotheses
This research tested a series of hypotheses examining the rela-
tionship between acetabular changes and OA (Hypothesis 1), the
Hypothesis 3 Acetabular Changes Correlate Positively with
Activity
Individuals who had undertaken more vigorous occupational
and habitual physical activities were expected to exhibit “older-
looking” acetabula than age-matched individuals undertaking less
vigorous physical activities. Transformed CAS and scores for
each of the seven individual acetabular variables were tested for
positive associations with the MET level of documented physical
 WINBURN . VALIDATION OF THE ACETABULUM FOR AGING 5

FIG. 3––Images of the acetabulum (left) and distal femur (right), corresponding with OA scores in the Jurmain (64) system: 0 = no OA/slight OA; 1 = mod-
erate OA (small osteophyte and/or pitting over < 10% of articular surface); 2 = severe OA (very large osteophyte/remodeling and/or pitting over > 10% of
articular surface; or any evidence of eburnation); 3 = ankylosis (joint fusion; not pictured). Photographs by the author.

activities (66). In particular, tests assessed whether high

acetabular scores were associated with rigorous documented
habitual physical activities utilizing the lower limb (e.g.,
running, wrestling). To control for age, individuals were com-
pared within age-matched 10-year subsamples (e.g., 30–39 years,
40–49 years).
Hypothesis 4 Acetabular Changes Correlate Positively with
Obesity
Obese individuals were expected to exhibit “older-looking”
acetabula than age-matched normal-weight or underweight indi-
viduals. This hypothesis employed BMI as a proxy for obesity.
6 JOURNAL OF FORENSIC SCIENCES

Transformed CAS and scores for each of the seven individual preservation of the observed skeletal elements (Table 3). Prelimi-
acetabular variables were tested for positive associations with nary statistical analyses (Wilcoxon rank-sum tests) indicated no
BMI values. To control for age, individuals were compared statistically significant differences for left and right CAS
within age-matched 10-year subsamples. (p > 0.25) and acetabular variables (p > 0.02; a = 0.007; Bon-
ferroni correction: a = 0.05/7) for either sex. Spearman’s rank-
order correlation tests also indicated left and right scores to be
strongly correlated (CAS: female q = 0.849, male q = 0.864;
Hypothesis 5 Of the Above Factors, Age has the Most
acetabular variables: rho values ranged from 0.608 to 0.762 for
Influence on Acetabular Changes
females and from 0.592 to 0.793 for males). Thus, left and right
Once the above-proposed associations between acetabular CAS were averaged to create an overall CAS for each individ-
changes and age, activity, and obesity had been supported or ual. This averaging was deemed appropriate for a summary mea-
rejected, this research endeavored to determine the relative sure of acetabular changes, akin to the combined joint OA
importance of these contributing factors. Multivariate statistical scores averaged according to the protocol of Weiss (65). This
analyses assessed the influence of age, occupational activities, also created an approximation of continuous, normally
habitual-activities, and BMI on CAS. distributed data for these variables. Rather than averaging the
acetabular variables, only left-side variables were used in
the below analyses (with right-side variables substituted in cases
Statistical Testing
of damaged or pathological acetabula), following the protocol of
Statistical tests were conducted in R (69) to assess potential Rissech et al. (2). This preserved the original, ordinal nature of
associations between acetabular scores/CAS and OA scores, ages these individual acetabular variable data.
at death, MET values, and BMI. Wilcoxon rank-sum tests Wilcoxon rank-sum tests revealed statistically significant differ-
assessed sex and side differences, as well as differences between ences between the means for female and male CAS (p = 0.01) and
individuals engaging in strenuous physical activities utilizing the female and male MET values for occupations (p = 0.001). Female
lower limb and individuals who did not engage in these rigorous and male OA score means also differed significantly in several
activities. Associations between ordinal datasets (e.g., acetabular joints (seven Wilcoxon rank-sum tests; a = 0.007 with Bonferroni
and OA scores) were assessed via Spearman’s rank-order corre- correction). Because of these significant differences, tests of CAS,
lation tests (rho) and Spearman’s tests calculating p-values using occupational MET values, and TMJ, hip, knee, and ankle OA data
the conditional null distribution of the test statistic (asymptotic utilized separate female and male analyses. Sex-specific results are
approximation of the exact distribution), with ties in rank presented separately, below. Wilcoxon rank-sum tests revealed no
resolved using mid-ranks. Associations between continuous data- statistically significant differences in the female and male means
sets (e.g., CAS, ages at death, BMI, MET values) were visual- of: acetabular variables (a = 0.007; Bonferroni correction:
ized via scatterplots and assessed via Spearman’s rank-order a = 0.05/7); overall, upper limb, and lower limb OA; MET values
correlation tests (rho) and linear regression (p, adjusted r2). The for habitual-activities; or BMI values, so combined-sex analyses
alpha-level was set at 0.05, and the Bonferroni correction was were undertaken. For these data, combined-sex results are pre-
applied in cases of multiple test iterations. sented below, except where comparisons involved distributions
Multivariate statistical testing isolated the variables exerting that showed sex differences (e.g., OA scores).
the greatest influence on acetabular changes. Graphical examina-
tion of the variables suggested that some of the study data (i.e.,
Hypothesis 1
BMI and habitual and occupational activities) did not meet the
assumptions of normality within a variable or linearity when Hypothesis 1 was supported: Testing revealed statistically sig-
plotted against one another. Thus, these variables were each log- nificant positive correlations between CAS and OA in all
transformed prior to multiple regression (MR) analyses, using
the natural logarithm. Combined linear models were then con-
structed for females and males separately, including all potential TABLE 3––Sample sizes for each of the variables used in statistical analy-
ses.
contributing factors and their interaction effects. In some cases,
however, the transformed variables still appeared to violate the Variable n (female) n (male) n (combined-sex)
distributional assumptions for MR. Thus, the raw data were fit-
ted with generalized linear models (GLM), in hopes that the data CAS 198 210 408
Acetabular variables 195 210 405
might be better served by the relaxed distributional assumptions 1, 2, 3, 4, 6, and 7
of these models, and in light of the fact that GLM has previ- Acetabular variable 5 195 209 404
ously proved effective in analyzing the simultaneous impacts of TMJ OA 198 210 408
multiple contributing factors to processes of skeletal degenera- Shoulder OA 198 210 408
Elbow OA 198 210 408
tion (70). The R Gamma-family GLM function was utilized (69, Wrist OA 197 210 407
“stats” package), due to the right-skewness of the non-normally Hip OA 198 210 408
distributed data. In both cases (MR and GLM), the full models Knee OA 195 203 398
were then simplified via automated backward-stepwise selection Ankle OA 194 208 402
informed by Akaike’s Information Criterion (AIC) in order to Overall OA 198 210 408
Upper limb OA 198 210 408
achieve the minimal adequate models. Lower limb OA 198 210 408
BMI 187 188 375
MET level 178 192 370
Results (occupational activities)
MET level 71 69 140
Sample sizes for each of the variables tested in the below (habitual-activities)
analyses differed based on availability of demographic data and

TABLE 4––Spearman’s rank-order correlations for CAS and OA in the vari-
ous joints and body regions in females and males.
Joint/Region rho (Female) rho (Male)
TMJ 0.331*
Shoulder 0.456*
Elbow 0.428*
Wrist 0.525*
Hip 0.745*
Knee 0.435*
Ankle 0.282*
Overall 0.664*
Upper limb 0.572*
Lower limb 0.606*
*Statistically significant at a = 0.007 for joints (Bonferroni correction:
a = 0.05/7) and a = 0.02 for body regions (Bonferroni correction: a = 0.05/
3).
individual joints and body regions, in both sexes (Table 4). Rho
values were particularly high in the hip: The 95% confidence
intervals for female and male hip OA did not contain the rho
values of any other joints (see Table 4). Regression of CAS on
overall OA, upper limb OA, and lower limb OA revealed statis-
tically significant positive associations between these composite
acetabular and OA scores in both females and males (Fig. 4).
Thus, OA seems to be linked with the acetabular aging processes
captured by CAS.
Many statistically significant positive correlations were also
present between the individual acetabular variables and OA in
the individual joints (Tables 5 and 6); and in the three body
regions, all positive correlations with the variables were statisti-
cally significant (Table 7). This underscores the link between
acetabular aging processes and OA. Further, the fact that both
female and male hip OA exhibited statistically significant posi-
tive correlations with all acetabular variables (see Table 5)
argues for similarities between the acetabular variables and the
processes of hip OA. If the variables being scored in the acetab-
ulum per Rissech et al. (2) were metamorphic, this correlation
with OA would not be expected.
Hypothesis 2
Hypothesis 2 was supported: Testing revealed statistically sig-
nificant positive correlations between CAS and age in both
females (q = 0.565; p < 0.001) and males (q = 0.713;
p < 0.001), indicating the applicability of acetabular age estima-
tion to both sexes (Fig. 5). Statistically significant positive corre-
lations were also noted between age and each of the seven
individual acetabular variables in the combined-sex sample, with
rho values of 0.405, 0.414, 0.507, 0.484, 0.434, 0.343, and
0.435, respectively (a = 0.007; Bonferroni correction: a = 0.05/
7).
Hypothesis 3
Hypothesis 3 was rejected: No statistically significant correla-
tions were detected between acetabular changes (CAS, acetabular
variables) and MET values for occupational or habitual physical
activities in females or males of any age group. For CAS and
occupational activities, the highest r2 value was 0.070, and
Spearman’s rho values ranged from 0.179 to 0.170, none of
which were significant at a = 0.01 for females and a = 0.008
for males (Bonferroni correction for females: a = 0.05/5; for
WINBURN . VALIDATION OF THE ACETABULUM FOR AGING 7
males: a = 0.05/6). For the individual acetabular variables and
occupational activities, Spearman’s rho values ranged from
0.408 to 0.438, none of which were statistically significant at
a = 0.001 (Bonferroni correction for females: a = 0.05/35; for
0.345* males: a = 0.05/42). For CAS and habitual-activities, the highest
0.654* r2 value was 0.144, and Spearman’s rho values ranged from
0.520*
0.603*
0.395 to 0.409, none of which were statistically significant at
0.772* a = 0.01 (Bonferroni correction for combined-sex sample:
0.564* a = 0.05/5). For the individual acetabular variables and habitual-
0.498* activities, Spearman’s rho values ranged from 0.565 to 0.443,
0.779* none of which were statistically significant at a = 0.001 (Bonfer-
0.703*
0.765* roni correction for combined-sex sample: a = 0.05/35).
Further, Wilcoxon rank-sum tests detected no statistically sig-
nificant differences between CAS and acetabular variable means
from individuals engaging in strenuous physical activities utiliz-
ing the lower limb compared with scores from individuals who
did not engage in these rigorous activities (p > 0.188 for all
groups; a = 0.01; Bonferroni correction: a = 0.05/5). These
findings indicate that neither differences in MET values nor vari-
ation in mechanical loading of the lower limb contribute signifi-
cantly to acetabular degeneration.
Hypothesis 4
Hypothesis 4 was rejected: No statistically significant correla-
tions were detected between acetabular changes (CAS, acetabular
variables) and BMI for females or males of any age group. For
CAS and BMI, the highest r2 value was 0.159, and Spearman’s
rho values ranged from 0.384 to 0.300, none of which were
statistically significant at a = 0.01 for females and a = 0.008 for
males (Bonferroni correction for females: a = 0.05/5; for males:
a = 0.05/6). For the individual acetabular variables and BMI,
Spearman’s rho values ranged from 0.172 to 0.410, none of
which were statistically significant at a = 0.001 (Bonferroni cor-
rection: a = 0.05/42). These findings indicate that obesity does
not contribute significantly to acetabular degeneration.
Hypothesis 5
Hypothesis 5 was supported: After the process of model sim-
plification based on AIC values, age remained in every minimal
adequate model in both of these multivariate approaches (MR,
GLM). When the data were examined graphically via pairwise
comparisons of variables, scatterplots showed strong linear rela-
tionships between CAS and age. Visualizing the MR models as
trees also indicated that age was the most important explanatory
variable in every model. In contrast, the other variables appeared
to be contributing little more than noise. These results lend sup-
port for the use of the acetabulum in age estimation.
In the female sample, a linear model was created with age,
log-transformed habitual-activities, and log-transformed BMI as
the explanatory variables and CAS as the response variable
(p = 0.008; r2 = 0.18). In the simplified model, age was the
only remaining term, and it was statistically significant
(p < 0.001; r2 = 0.23). A similar model was created with the
larger sample size available for testing the effects of age, log-
transformed occupational activities, and log-transformed BMI on
female CAS (p < 0.001; r2=0.32). Again, in the simplified
model, age was the only remaining term, and it was statistically
significant (p < 0.001; r2 = 0.33).
In the male sample, a linear model was created with age, log-
transformed habitual-activities, and log-transformed BMI as the
explanatory variables and CAS as the response variable
8 JOURNAL OF FORENSIC SCIENCES

Females Males
10

10
r 2 =0.329* r 2 =0.478*

9
9

8
8

7
CAS

CAS
7

6
6

5
5

4
4

3
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5
Upper Limb OA

10
10

r 2 =0.39* r 2 =0.595*

9
9

8
8

7
CAS
CAS
7

6
6

5
5

4
4

0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5

Lower Limb OA
10

10

r 2 =0.449* r 2 =0.619*
9
9

8
8

7
CAS

CAS
7

6
6

5
5

4
4

0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5

Overall OA
FIG. 4––Linear regression revealed statistically significant correlations between acetabular changes (CAS) and OA in females and males, suggesting similari-
ties between these degenerative processes (*p < 0.001).

(p < 0.001; r2 = 0.57). In the simplified model, age and BMI
remained in the model, and both were statistically significant
(age p < 0.001; BMI p = 0.004; model p < 0.001; r2 = 0.59). A
similar model was created with the larger sample size available
for testing the effects of age, log-transformed occupational activ-
ities, and log-transformed BMI on male CAS (p < 0.001;
r2 = 0.41). In the simplified model, age and BMI again
remained in the model, and both were statistically significant
(age p < 0.001; BMI p = 0.03; model p < 0.001; r2 = 0.42).
The nontransformed data were then fitted to Gamma-family
GLMs, with age, activities, and BMI as the explanatory variables
and CAS as the response variable. In the simplified female
habitual-activity GLM, age, activity, and the age-activity interac-
tion remained, but only age was statistically significant
(p < 0.001; null deviance: 2.23 on 68 degrees of freedom [df];
residual deviance: 1.67 on 65 df). In the simplified female occu-
pational-activity GLM, age was the only remaining term, and it
was statistically significant (p < 0.001; null deviance: 6.56 on
170 df; residual deviance: 4.53 on 169 df). In the simplified
male habitual-activity model, age and BMI remained in the
model, and both were statistically significant (p < 0.001; null
deviance: 3.44 on 65 df; residual deviance: 1.44 on 63 df). In
 WINBURN . VALIDATION OF THE ACETABULUM FOR AGING 9

TABLE 5––Spearman’s rank-order correlations for acetabular variables 1 through 7 and OA in the various joints in females and males.

Summary OA rho (1) rho (2) rho (3) rho (4) rho (5) rho (6) rho (7)
Female TMJ 0.309* 0.089 0.331* 0.206 0.188 0.169 0.229
Female hip 0.515* 0.553* 0.510* 0.472* 0.499* 0.417* 0.371*
Female knee 0.265* 0.302* 0.227 0.326* 0.300* 0.310* 0.241
Female ankle 0.254* 0.245 0.170 0.192 0.209 0.173 0.112
Male TMJ 0.281* 0.215 0.252* 0.234 0.198 0.194 0.158
Male hip 0.537* 0.629* 0.576* 0.645* 0.673* 0.330* 0.363*
Male knee 0.380* 0.445* 0.432* 0.485* 0.414* 0.249* 0.300*
Male ankle 0.357* 0.319* 0.221 0.394* 0.478* 0.328* 0.176
*Statistically significant at a = 0.0007 (Bonferroni correction: a = 0.05/70). Note that separate female and male samples were only used to test joints with
sex differences in OA distributions.

TABLE 6––Spearman’s rank-order correlations for acetabular variables 1 through 7 and OA in the various joints in the combined-sex sample.

Summary OA rho (1) rho (2) rho (3) rho (4) rho (5) rho (6) rho (7)
Shoulder 0.426* 0.372* 0.422* 0.424* 0.368* 0.323* 0.329*
Elbow 0.371* 0.329* 0.346* 0.286* 0.306* 0.300* 0.268*
Wrist 0.424* 0.353* 0.371* 0.381* 0.397* 0.318* 0.265*
*Statistically significant at a = 0.0007 (Bonferroni correction: a = 0.05/70). Note that the combined-sex sample was only used to test joints with no
detected sex differences in OA distributions.

TABLE 7––Spearman’s rank-order correlations for acetabular variables 1 through 7 and summary OA in the combined-sex sample.

Summary OA rho (1) rho (2) rho (3) rho (4) rho (5) rho (6) rho (7)
Overall 0.532* 0.483* 0.495* 0.533* 0.521* 0.401* 0.367*
Upper limb 0.481* 0.417* 0.445* 0.442* 0.430* 0.377* 0.344*
Lower limb 0.469* 0.520* 0.438* 0.531* 0.538* 0.384* 0.331*
*Statistically significant at a = 0.002 (Bonferroni correction: a = 0.05/21).

the simplified male occupational-activity model, age and BMI
again remained in the model, and both were statistically signifi-
cant (age p < 0.001; BMI p = 0.01; null deviance: 6.69 on 178
df; residual deviance: 3.98 on 176 df).
Discussion
Correlation of Acetabular Changes with OA
The statistically significant positive correlations observed
between acetabular changes and OA indicate that the changes
occurring in the acetabulum are degenerative rather than meta-
morphic. There may be a functional component to this correla-
tion with degenerative joint disease. Higher rho values were
always seen between the individual acetabular variables and
overall and lower limb OA (compared with upper limb OA); in
many cases, differences were marked enough that the 95% confi-
dence intervals for the overall and lower limb rho values did not
contain the upper limb rho value. In linear regression analyses
of CAS, r2 values were also higher for lower limb OA than
upper limb OA in both females and males, as would be expected
if the regular stresses of bipedal locomotion were causing both
acetabular age change and lower limb OA.
However, there may also be a systemic component to the posi-
tive relationship between the acetabular variables and OA. While
female and male hip OA correlated significantly with all acetabu-
lar variables, female and male knee and ankle OA did not. In the
ankle, and especially in females, associations with acetabular vari-
ables were particularly weak. In contrast, combined-sex shoulder,
elbow, and wrist OA had statistically significant positive correla-
tions with all acetabular variables. This indicates a systemic com-
ponent to both processes—OA and acetabular changes—rather
than a purely biomechanical etiology. Preliminary research on the
same European-American subset of the UT Bass Collection has
revealed statistically significant OA-age correlations in nearly
every joint analyzed, with the exception of the ankle (42). Degen-
erative processes of the shoulder and hip emerge from this and
other research as particularly strongly correlated with age (42,71),
possibly due to their proximal and relatively anatomically
protected positions within the upper and lower limb (72). That
acetabular changes did not correlate with physical activity or obe-
sity in the current study (see below) lends further support to the
possibility that hip degeneration may be relatively resistant to
biomechanical effects and influenced largely by systemic effects
(i.e., age).
Correlation of Acetabular Changes with Age
The statistically significant positive correlations observed
between acetabular changes and age indicate that, in spite of
their degenerative nature, acetabular changes are useful for age
estimation. The positive age correlations observed in both
females and males are consistent with recent research demon-
strating the pattern of acetabular aging to be similar in both
sexes (3,34), despite the fact that the original acetabular aging
method was developed and tested on all-male samples (2,30).
The current results are consistent with previous research by
San-Mill
an et al. (3) indicating the strong age correlation of the
10 JOURNAL OF FORENSIC SCIENCES

(19) if the variable descriptions had been more clearly elucidated

r 2 = 0.338*
in the original method (2). To this end, recent research has
90

emended the existing descriptions of the acetabular variables (3),

80

modifying Variables 5 through 7 for clarity and repeatability.

Future method tests should determine whether these changes
70

have improved the method sufficiently to reduce issues of intra-

Age

and inter-observer error.

60

The lack of age correlation reported by previous authors for

50

acetabular fossa variables could also reflect biological differ-

ences in the aging processes taking place in the region of the
40

acetabular fossa, compared with the more strongly age-correlated

30

acetabular rim. Previous research has indicated that marginal

joint changes (e.g., osteophyte formation) are more strongly cor-
4 5 6 7 8 9 10 related with age than changes to the joint’s articular surfaces
(73). One recent study found that osteophytes of the acetabulum
CAS (Females)
and scapular glenoid exhibit stronger age correlations than osteo-
phytes in other locations (71). Reevaluations of acetabular age-
100

r 2 = 0.52* estimation methods consistently conclude that acetabular rim and

apex variables correlate with age (1,8,19). A recent geometric
morphometric shape analysis of age-related changes to the
80

acetabulum revealed that bone production along the borders of

the lunate surface (i.e., the acetabular rim, apex, and outer edge
Age

of the acetabular fossa) is associated with age in both females

60

and males (34). In the current study, the two acetabular variables
with the highest rho-value correlations with age were Variables
40

3 and 4, and sometimes these differences in age correlation were

marked (e.g., the 95% confidence interval for Variable 3 did not
include the rho values for Variables 1, 2, and 6; the 95% confi-
20

3 4 5 6 7 8 9 10
CAS (Males)
FIG. 5––Linear regression revealed statistically significant correlations
between acetabular changes (CAS) and age in females and males
(*p < 0.001).
seven acetabular variables (2). Other research, however, has indi-
cated that variables of the acetabular fossa are less strongly cor-
related with age than variables of the acetabular rim (1,19). This
discrepancy could reflect population differences among the sam-
ples analyzed by Calce (1; males from the early 20th-century
Grant Collection, Toronto, Canada), Mays (19; females and
males from the 18th- to 19th-century Spitalfields Collection,
London, U.K.), San-Mill
an et al. (3; females and males from the
19th- and 20th-century Lisbon Collection, Portugal), and in the
current research (females and males from the modern Bass Col-
lection, UT, U.S.A.).
However, discrepancies in the pattern of age correlations of
the acetabular variables could also reflect difficulties of applying
the original acetabular aging method (2), possibly representing
an artifact of inter-observer error. For example, Mays (19)
reported modifying the method protocol when scoring Variables
6 and 7; it is thus unsurprising that the modified scores do not
correlate as strongly with age as the original authors reported
(2). Other authors have critiqued previously published descrip-
tions of acetabular variables, stating that difficulties in interpret-
ing the descriptions lead to observer subjectivity (8,40). During
the current study, scoring difficulties were noted for the variables
of the acetabular fossa (Variables 5–7), which assess subtle and
often-ambiguous skeletal traits like bone texture and porosity (2;
Table 2). It is possible that fossa variables would have correlated
more strongly with age in the research of Calce (1) and Mays
dence interval for Variable 4 did not include the rho values for
Variables 1 and 6). Variables 3 and 4 describe porosity of the
acetabular rim and osteophyte formation of the apex, respec-
tively. Both of these strongly age-correlated variables describe
marginal joint changes.
Interestingly, some of the Rissech (2) acetabular rim and
apex variables (Variables 1, 2, and 4) have also proved resis-
tant to the effects of previous traumatic hip injuries and surgi-
cal interventions, while the variables of the acetabular fossa
(Variables 5 through 7) have showed statistically significant
trauma effects (42). Since acetabular aging methods should not
be applied to hip joints showing evidence of skeletal trauma,
surgery, or pathological conditions (2), these limited trauma
effects can easily be avoided during normal forensic anthropo-
logical practice (i.e., by substituting right for left acetabula or
simply utilizing different age indicators). Still, the resistance of
Variables 1, 2, and 4 to traumatic injuries and/or surgical inter-
ventions argues in favor of these OA-related rim and apex vari-
ables, and the trauma effects exhibited by Variables 5 through
7 argue against the use of these problematic fossa variables
(42). Future acetabular method revisions could focus only on
the variables of the acetabular rim and apex—or better eluci-
date the processes occurring in the poorly understood acetabu-
lar fossa.
Lack of Correlation Between Acetabular Changes and Habitual/
Occupational Activities
This study’s results suggest that acetabular changes are resis-
tant to physical activity effects, lending support to the use of
acetabular changes for age estimation. The lack of correlation
between activity and acetabular degeneration accords with cur-
rent medical and anthropological research, which contradicts the
“intuitive” idea that more frequently used joints will age prema-
turely and indicates instead the benefit of moderate physical
 WINBURN . VALIDATION OF THE ACETABULUM FOR AGING 11

activity (and the detriment of physical inactivity) for joint health Revising the Metamorphic-Degenerative Dichotomy
(54,74–77). Rather than supporting the idea that the daily wear
The current research project has indicated that acetabular
and tear of bipedal locomotion causes premature acetabular
changes constitute skeletal degeneration. However, it also indi-
aging, acetabular variables appeared resistant to mechanical load-
cates that acetabular degeneration has a significant, positive cor-
ing. However, these activity findings must be considered tenta-
relation with age. One of the study’s assumptions was that the
tive, due to the difficulties of assessing the metabolic and
utility of degenerative changes for age estimation would be pre-
biomechanical impacts of occupational and habitual physical
cluded by the multiple other influences on the progression of
activities in a skeletal sample of donated remains (see Discussion
skeletal degeneration. Yet, acetabular degeneration also proved
of study limitations, below).
relatively resistant to the effects of physical activity and obesity
—and, in the same skeletal sample, even overall measures of
Lack of Correlation Between Acetabular Changes and Obesity OA correlated strongly with age (42). This suggests that the
dichotomy between metamorphic and degenerative change (and
This study’s results suggest that acetabular changes are resis-
their probative vs. nonprobative value for age estimation) is a
tant to obesity effects, supporting their use in age estimation.
false one. If both metamorphic and degenerative changes are
These findings are particularly relevant in a forensic setting,
strongly positively correlated with age, the former should not be
considering the prevalence of obesity in modern populations
given preference over the latter in age-estimation studies. After
(78). The findings are also interesting in light of the fact that
all, while some regions of metamorphic change enable reliable
previous research has determined the iliac auricular surface to
age predictions (at least in younger individuals—e.g., the ventral
be affected by obesity—yielding less accurate and more biased
rampart of the pubic symphysis), others have been discounted as
age estimates for obese adults compared with normal-weight
age-relevant for decades (e.g., the cranial sutures [39]). One
adults (53). While the pubic symphysis has been found to be
might speculate that it is the anabolic nature of the formation of
more resistant to obesity effects than the auricular surface (53),
the pubic symphysis ventral rampart that makes this feature more
the acetabulum has been determined to be one of the least
relevant to aging. Yet, there is an anabolic component to “de-
biased and most reliable skeletal age indicators for obese indi-
generative” OA as well: abnormal subchondral bone production,
viduals (79). The results of the present study support this con-
sclerosis, and the formation of marginal osteophytes are integral
clusion, indicating that the acetabulum is resistant to the
facets of the disease in addition to the destructive processes of
increasingly prevalent condition of obesity. However, the cur-
cartilage degradation and bone erosion.
rent results must be compared with these previous findings
Further, the ventral rampart is not the only component of the
(53,79) with caution, as the effects of obesity on the acetabular
pubic symphysis that enables reliable age estimation (4,14).
age estimates themselves were not tested in the current study.
Most pubic symphyseal aging methods also assess degenerative
These results also merit a caveat in light of the fact that multi-
changes (e.g., 5,14)—including both anabolic traits (i.e., osteo-
variate modeling reveals that BMI contributes to variance in
phytes) and catabolic traits (i.e., porosity [80]). Degenerative and
CAS—albeit in males only (see below).
metamorphic processes may also feedback into one another. As
demonstrated in a recent study by Calce et al. (80), OA severity
Age as the Major Contributor to Acetabular Changes has an impact on the age estimates generated by the various pel-
vic joints. Biomechanical differences in the various joints must
Age remained in every minimal adequate MR model and
also be considered in assessing differing patterns of age-related
GLM investigating the contribution of the various factors to
change (42,81). Dichotomous thinking simply may not be an
CAS. Age was a statistically significant term in every minimal
appropriate approach to these related, overlapping, and difficult-
adequate model for both sexes, using both approaches. For
to-distinguish processes of skeletal change. Certainly, there
males, BMI also remained as a statistically significant term in
seems no legitimate basis on which to tout “metamorphic”
all simplified models. The consistency of the results from MR
changes as “good” and dismiss “degenerative” changes as “bad.”
and GLM indicates that the potential violation of distributional
The very assumption that bone production represents a benefi-
assumptions is not impacting the performance of tests—per-
cial process and bone destruction a maladaptive process must be
haps due to the significance of the impact of age on CAS. In
questioned. If bone is building a structure (e.g., a ventral rampart
MR models, age alone explained approximately one-quarter-to-
or osteophyte) where one is not functionally necessary, it could
one-third of the variance in CAS observed in females, and,
be argued that this process is maladaptive regardless of whether
along with BMI, age explained approximately half of the vari-
the formation is classified as metamorphic or degenerative. The
ance in CAS observed in males. It must be noted, however,
fact remains that metamorphic change is a normal part of human
that over two-thirds of the variance in female CAS was not
skeletal aging, while OA is a pathological condition for which
explained by the female age-only MR models, and approxi-
age is a major predisposing factor (82,83). However, if this
mately half of male CAS variance was not explained by the
pathological process is as strongly age correlated as this study
male combined (age-BMI) models. Additional contributing fac-
suggests, then it too should be considered in skeletal estimates
tors undoubtedly influence acetabular changes—they simply
of age.
were not measured in the current study. Future research could
Indeed, it seems imprudent to dismiss degenerative processes
examine the influences of such factors as heredity and (partic-
like OA as “ancillary (if not a last resort)” to other skeletal
ularly in the case of females) hormones on variance in CAS.
indicators of age (43, p. 59). For the purposes of age estima-
Still, these results support the above conclusions that, of the
tion, this dichotomous distinction may be meaningless. Both
three factors investigated herein, age is always a consistent
metamorphic and degenerative skeletal changes are age corre-
and significant contributor to acetabular changes. This is in
lated, and both should be incorporated into skeletal age-estima-
accordance with other recent research on pelvic degeneration
tion methods (14,71,80,84,85). A potential venue for this
(74) and lends support for the use of the acetabulum in age
incorporation might be the multifactorial Bayesian age-
estimation.
12 JOURNAL OF FORENSIC SCIENCES
estimation methods that are currently gaining support among
forensic anthropologists (33,86–89).
Study Limitations and Future Avenues for Research
This research employed an ordinal scoring system to analyze
skeletal evidence of OA. While such systems are widely used in
biological anthropology, they oversimplify the complexity of
anabolic and catabolic processes that comprise the OA disease
process—which is nonlinear in its progression, differing from
joint to joint and involving both degenerative and reparative pro-
cesses (80,90,91). Further, the sensitivity of ordinal scores to
identifying clinical markers of OA remains unknown, as does
the correspondence of those scores with actual OA symptoms in
living patients. In absence of the articulated-joint radiographs
and patient-reported pain experiences used in clinical diagnoses
of OA (63,92), anthropologists rely on the presence of macro-
scopically observable osteophytes, eburnation, and porosity to
diagnose OA indirectly in a deceased individual (90). Yet, the
correlation of these indicators with OA symptoms is not clear
even in the clinical setting, where radiographic evidence of the
disease does not always correlate with actual symptoms of dis-
ability (82,93). Future analyses should explore the correlation of
skeletal indicators of OA with actual experiences of pain and
impairment.
This study’s activity coding system is subject to similar limita-
tions. Simply, it is unclear how well (or poorly) averaged MET
values from the Compendium of Physical Activities reflect the
actual energetic exertions of the deceased individuals in the Bass
Collection. The Compendium was designed for use in self-
reported activity surveys in which participants quantify their
daily activities in a standardized fashion (66,67). In contrast, the
occupational and habitual-activity data tested in the current study
were nonstandardized—captured by open-ended prompts on
skeletal donor forms. It is also difficult to assess whether aver-
aged MET values can differentiate among real, in vivo activity
differences to a degree that would be meaningful biomechani-
cally and in terms of skeletal response. Metabolic equivalency
values reflect energy exertion, not skeletal biomechanics, so the
use of MET data as a proxy for the actual joint forces to which
an individual’s joints were regularly subjected may be ques-
tioned. For these reasons, this study’s lack of positive correlation
between physical activity and acetabular degeneration should be
considered tentative. Still, this finding is consistent with recent
medical research on the benefits of physical activity for aging
individuals (54,75,76) and anthropological research indicating a
negative correlation between activity and OA (74).
This study approximated obesity using BMI—a description of
the relationship of an individual’s height to body mass. How-
ever, BMI does not measure an individual’s percent body fat,
differentiate adipose tissue from muscle, bone, or organs, or
describe an individual’s abdominal girth (94,95). A muscular
athlete may have a high BMI but a low percent body fat, for
example; conversely, an older individual may have a higher per-
cent body fat than a younger individual with the same BMI.
Body fat proportions may also differ among individuals of dif-
ferent ancestries (96,97) and biological sexes (94). Thus, critics
assert that BMI is not sensitive enough to identify the dangerous
levels of adiposity that have been correlated with morbidity and
mortality in obese individuals (95)—at least in non-European
populations, for whom relationships between BMI and adiposity
have not been sufficiently established (97). However, methods
that do directly measure or approximate body fat composition
(e.g., underwater body density studies) are often time-consuming
and expensive (95). Calculating BMI remains a conveniently
simple approximation for adiposity (98), and while imperfect, its
lower and upper limits likely do capture extremes of adiposity
(i.e., underweight, BMI <18.5; obese, BMI ≥30). More accurate
methods of measuring adiposity are impractical when dealing
with deceased individuals and impossible when analyzing skele-
tal remains. Further, the lack of ancestral diversity in the devel-
opment and testing of BMI equations is not a factor in the
current study. Percent body fat is strongly correlated with BMI
in European-American populations (99), and this sample is com-
prised exclusively of European-American individuals. Thus,
BMI is an indirect indicator of adiposity valid for use this study.
It should also be stated that the steps taken in the current
study to control for the effects of age in analyses of activity and
obesity (restricting testing to within age-matched subsamples)
reduce the statistical power of those analyses. This increases the
chances of achieving results that support the null hypotheses
of no activity and obesity effects. In turn, this makes the results
of these analyses less directly comparable with the analyses of
Hypotheses 1 and 2 (investigating correlations between acetabu-
lar changes and OA and age), which include larger sample sizes.
Future studies could attempt to replicate these results across lar-
ger, age-matched samples that are more comparable with this
study’s overall sample size.
Finally, there are ancestral limitations to the current study.
Within the overall Bass Collection (which, at the time of data
collection, included over 1500 individuals), approximately 700
individuals included information on height, body mass, and
occupational/habitual-activities. Most of these completely docu-
mented individuals were of European-American ancestry.
Because this documentation disparity prevented the assemblage
of sufficient sample sizes for non-European individuals, all indi-
viduals in the current sample were of European-American des-
cent. This precluded an analysis of the effects of ancestry—a
factor linked with heredity that almost undoubtedly contributes
both to OA and to acetabular changes (30). As previous acetabu-
lar age-estimation research has also focused almost entirely on
individuals of European ancestry (2,3,8,19,30), the ancestral
makeup of this skeletal sample is appropriate to the current capa-
bilities of the age-estimation methods. Still, the absence of
ancestry data is a limitation of the current study—one that
should be remedied in future studies of acetabular aging.
Conclusions
In summary, the degenerative changes of the acetabulum are
valid skeletal age indicators that are relatively resistant to the
effects of physical activity and obesity in European-Americans.
This study’s results demonstrate the utility of the acetabulum for
age estimation and indicate that the metamorphic-degenerative
dichotomy of skeletal age change is in need of revision. Further,
the tentative finding that physical activity has no effect on
acetabular degeneration suggests that an activity-led etiology for
joint degeneration is overly simplistic.
The current research is particularly relevant to forensic anthro-
pological age estimation in an increasingly elderly U.S. popula-
tion, since acetabular degeneration continues with advancing
age. By demonstrating the validity of this skeletal age indicator,
this research paves the way for future method revisions and pop-
ulation validations. Acetabular age changes could prove particu-
larly relevant to the development of multifactorial aging
methods. This study’s results will enable more effective method

revision in the future, informed by a more complete understand-
ing of the complexity of acetabular changes.
In a STEM-based field where standards matter, it is important
not only to test and refine age-estimation methods, but also to
understand the biology underlying method success or failure.
This research constitutes a response to the call for studies estab-
lishing the scientific bases on which forensic methods are built
(100,101). The results indicate that the use of the acetabulum for
age estimation indeed has scientific merit.
Acknowledgments
To the faculty and students of the University of Florida
Department of Anthropology and Institute on Aging—especially
Dr. David Daegling and Dr. Michala Stock—thank you for your
kind help and support. Many thanks go to the terrific team at the
University of Tennessee Forensic Anthropology Center who
facilitated my access to the Bass Collection. Special thanks are
due to the skeletal donors and their families, without whose gift
this type of research would not be possible.
References
1. Calce SE. A new method to estimate adult age-at-death using the
acetabulum. Am J Phys Anthropol 2012;148:11–23.
2. Rissech C, Estabrook GF, Cunha E, Malgosa A. Using the acetabulum
to estimate age at death of adult males. J Forensic Sci 2006;51:213–29.
3. San-Mill
an M, Rissech C, Turb
on D. New approach to age estimation
of male and female adult skeletons based on the morphological charac-
teristics of the acetabulum. Int J Legal Med 2016;131:501–25.
4. Berg GE. Pubic bone age estimation in adult women. J Forensic Sci
2008;53:569–77.
5. Brooks S, Suchey JM. Skeletal age determination based on the os
pubis: a comparison of the Asc
adi-Nemesk
eri and Suchey-Brooks meth-
ods. J Hum Evol 1990;5:227–38.
6. Brown CA. A test of the Passalacqua sacral age estimation method in a
Japanese sample. Proceedings of the 67th Annual Scientific Meeting of
the American Academy of Forensic Sciences; 2015 Feb 16-21; Orlando,
FL. Colorado Springs, CO: American Academy of Forensic Sciences,
2015.
7. Buckberry JL, Chamberlain AT. Age estimation from the auricular sur-
face of the ilium: a revised method. Am J Phys Anthropol
2002;119:231–9.
8. Calce SE, Rogers TL. Evaluation of age estimation technique: testing
traits of the acetabulum. J Forensic Sci 2011;56:302–11.
9. Cleland J. On certain distinctions of form hitherto unnoticed in the
human pelvis, characteristic of sex, age, and race. Mem Memor Anat
1889;1:95–103.
10. Djuric M, Djonic D, Nikolic S, Popovic D, Marinkovic J. Evaluation of
the Suchey-Brooks method for aging skeletons in the Balkans. J Foren-
sic Sci 2007;52:21–3.
11. Falys CG, Schutkowski H, Weston DA. Auricular surface aging: worse
than expected? A test of the revised method on a documented historic
skeletal assemblage. Am J Phys Anthropol 2006;130:508–13.
12. Gilbert BM, McKern TW. A method for aging the female os pubis. Am
J Phys Anthropol 1973;38:31–8.
13. Hanihara K, Suzuki T. Estimation of age from the pubic symphysis by
means of multiple regression analysis. Am J Phys Anthropol
1978;48:233–40.
14. Hartnett KM. Analysis of age-at-death estimation using data from a
new, modern autopsy sample—Part I: pubic bone. J Forensic Sci
2010;55:1145–51.
15. Igarashi Y, Uesu K, Wakebe T, Kanazawa E. New method for estima-
tion of adult skeletal age at death from the morphology of the auricular
surface of the ilium. Am J Phys Anthropol 2005;128:324–39.
16. Katz D, Suchey JM. Race differences in pubic symphyseal aging pat-
terns in the male. Am J Phys Anthropol 1989;80:167–72.
17. Listi GA, Manhein MH. The use of vertebral osteoarthritis and osteo-
phytosis in age estimation. J Forensic Sci 2012;57:1537–40.
18. Lovejoy CO, Meindl RS, Pryzbeck TF, Mensforth RP. Chronological
metamorphosis of the auricular surface of the ilium: a new method for
WINBURN . VALIDATION OF THE ACETABULUM FOR AGING 13
the determination of adult skeletal age at death. Am J Phys Anthropol
1985;68:15–28.
19. Mays S. An investigation of age-related changes at the acetabulum in
18th-19th century AD adult skeletons from Christ Church Spitalfields.
London. Am J Phys Anthropol 2012;149:485–92.
20. Mays S. A test of a recently devised method of estimating skeletal age
at death using features of the adult acetabulum. J Forensic Sci
2014;59:184–7.
21. McKern TW, Stewart TD. Skeletal age changes in young American
males (Technical Report EP-45). Natick, MA: Quartermaster Research
and Development Command, 1957.
22. Miranker M. A comparison of different age estimation methods of the
adult pelvis. J Forensic Sci 2016;61:1173–9.
23. Moraitis K, Zorba E, Eliopoulos C, Fox SC. A test of the revised auric-
ular surface aging method on a modern European population. J Forensic
Sci 2014;59:188–94.
24. Mulhern DM, Jones EB. Test of revised method of age estimation
from the auricular surface of the ilium. Am J Phys Anthropol
2005;126:61–5.
25. Murray KA, Murray T. A test of the auricular surface aging technique.
J Forensic Sci 1991;36:1162–9.
26. Osborne DL, Simmons TL, Nawrocki SP. Reconsidering the auricular
surface as an indicator of age at death. J Forensic Sci 2004;49:905–11.
27. Passalacqua NV. Forensic age-at-death estimation from the human
sacrum. J Forensic Sci 2009;54:255–62.
28. Passalacqua NV. Applications of the human sacrum in age at death esti-
mation. In: Latham KE, Finnegan M, editors. Age estimation of the
human skeleton. Springfield, IL: Charles C Thomas, 2010;102–17.
29. Powanda AI. A comparison of pelvic age-estimation methods on two
modern Iberian populations: bioarchaeological and forensic applications
[thesis]. New York, NY: New York University, 2008.
30. Rissech C, Estabrook GF, Cunha E, Malgosa A. Estimation of age-at-
death for adult males using the acetabulum, applied to four Western
European populations. J Forensic Sci 2007;52:774–8.
31. Roug
e-Maillart C, Jousset N, Viella B, Gaudin A, Telmon N. Contribu-
tion of the study of the acetabulum for the estimation of adult subjects.
Forensic Sci Int 2007;171:103–10.
32. Roug
e-Maillart C, Telmon N, Rissech C, Malgosa A, Roug
e D. The
determination of male adult age at death by central and posterior coxal
analysis—a preliminary study. J Forensic Sci 2004;49:208–14.
33. Roug
e-Maillart C, Vielle B, Jousset N, Chappard D, Telmon N, Cunha
E. Development of a method to estimate skeletal age at death in adults
using the acetabulum and the auricular surface on a Portuguese popula-
tion. Forensic Sci Int 2009;188:91–5.
34. San-Mill
an M, Rissech C, Turb
on D. Shape variability of the adult
human acetabulum and acetabular fossa related to sex and age by geo-
metric morphometrics. Implications for adult age estimation. Forensic
Sci Int 2017;272:50–63.
35. Sashin D. A critical analysis of the anatomy and pathological changes
of the sacro-iliac joints. J Bone Joint Surg 1930;12:891–910.
36. Schmitt A. Age-at-death assessment using the os pubis and the auricular
surface of the ilium: a test on an identified Asian sample. Int J Osteoar-
chaeol 2004;14:1–6.
37. Sinha A, Gupta V. A study on estimation of age from pubic symphysis.
Forensic Sci Int 1995;75:73–8.
38. Snodgrass JJ. Sex differences and aging of the vertebral column. J
Forensic Sci 2004;49:458–63.
39. Stewart TD.Essentials of forensic anthropology. Springfield, IL: Charles
C Thomas, 1979.
40. Stull KA, James DM. Determination of age-at-death using the acetabu-
lum. In: Latham KE, Finnegan M, editors. Age estimation of the human
skeleton. Springfield, IL: Charles C Thomas, 2010;134–46.
41. Todd TW. Age changes in the pubic bone. Am J Phys Anthropol
1920;3:285–337.
42. Winburn AP. Skeletal age estimation in modern European-American
adults: the effects of activity, obesity, and osteoarthritis on age-related
changes in the acetabulum [dissertation]. Gainesville, FL: University of
Florida, 2017.
43. Aykroyd RG, Lucy D, Pollard AM, Roberts CA. Nasty, brutish, but not
necessarily short: a reconsideration of the statistical methods used to
calculate age at death from adult human skeletal and dental age indica-
tors. Am Antiq 1999;64:55–70.
44. Meindl RS, Russell KF. Recent advances in method and theory in pale-
odemography. Annu Rev Anthropol 1998;27:375–99.
45. Nawrocki SP. The nature and sources of error in the estimation of age
at death from the skeleton. In: Latham KE, Finnegan M, editors. Age
14 JOURNAL OF FORENSIC SCIENCES
estimation of the human skeleton. Springfield, IL: Charles C Thomas,
2010;79–101.
46. Latham KE, Finnegan M, editors. Age estimation of the human skele-
ton. Springfield, IL: Charles C Thomas, 2010.
47. Milner GR, Boldsen JL. Transition analysis: a validation study with
known-age modern American skeletons. Am J Phys Anthropol
2012;148:98–110.
48. Buikstra JE, Ubelaker DH, editors. Standards for data collection from
human skeletal remains. Fayetteville, AK: Arkansas Archeological Sur-
vey Research Series (No. 44), 1994.
49. Komar DA, Buikstra JE. Forensic anthropology: contemporary theory
and practice. New York, NY: Oxford University Press, 2008.
50. Vaupel JW. The biodemography of aging. Pop Dev Rev 2004;30:48–
62.
51. Center NationalCrimeInformation. Missing persons dataset. Washington,
DC: Federal Bureau of Investigation, 2014.
52. Haglund WD, Reay DT, Swindler DR. Canid scavenging/disarticulation
sequence of human remains in the Pacific Northwest. J Forensic Sci
1989;34:587–606.
53. Wescott DJ, Drew JL. Effect of obesity on the reliability of age-at-death
indicators of the pelvis. Am J Phys Anthropol 2015;156:595–605.
54. Hunter DJ, Eckstein F. Exercise and osteoarthritis. J Anat
2009;214:197–207.
55. Felson DT, Zhang Y. An update on the epidemiology of knee and hip
osteoarthritis with a view to prevention. Arthritis Rheum 1998;41:1343–
55.
56. Kennedy KAR. Skeletal markers of occupational stress. In: Is _ßcan MY,
Kennedy KAR, editors. Reconstruction of life from the skeleton. New
York, NY: Alan R. Liss Inc., 1989;129–60.
57. Larsen CS. Bioarchaeology: interpreting behavior from the human
skeleton. Cambridge, U.K.: Cambridge University Press, 1997.
58. Coggon D, Reading I, Croft P, McLaren M, Barrett D, Cooper C. Knee
osteoarthritis and obesity. Int J Obes 2001;25:622–7.
59. Couchman J. Past, present and future insights into the understanding
and treatment of osteoarthritis: molecular and mechanical approaches.
Int J Exp Pathol 2009;90:A30–84.
60. Felson DT, Anderson JJ, Naimark A, Walker AM, Swift M, Meenan
RF. Obesity and knee osteoarthritis: the Framingham Study. Ann Intern
Med 1988;109:18–24.
61. Felson DT, Lawrence RC, Dieppe PA, Hirsch R, Helmick CG, Jordan
JM, et al. Osteoarthritis: new insights. Part 1: the disease and its risk
factors. Ann Intern Med 2000;133:635–46.
62. Fransen M, Bridgett L, March L, Hoy D, Penserga E, Brooks P. The
epidemiology of osteoarthritis in Asia. Int J Rheum Dis 2011;14:113–
21.
63. Mandl LA. Epidemiology of osteoarthritis. In: Sharma L, Berenbaum F,
editors. Osteoarthritis: a companion to rheumatology. Philadelphia, PA:
Mosby, Inc. (Elsevier), 2007;1–14.
64. Jurmain RD. Paleoepidemiology of a Central California prehistoric pop-
ulation from CA-ALA-329: II. Degenerative disease. Am J Phys
Anthropol 1990;83:83–94.
65. Weiss E. Osteoarthritis and body mass. J Archaeol Sci 2006;33:690–5.
66. Ainsworth BE, Haskell WL, Herrmann SD, Meckes N, Bassett DR Jr,
Tudor-Locke C, et al. 2011 Compendium of Physical Activities: a sec-
ond update of codes and MET values. Med Sci Sports Exerc
2011;43:1575–81.
67. Ainsworth BE, Haskell WL, Whitt MC, Irwin ML, Swartz AM, Strath
SJ, et al. Compendium of Physical Activities: an update of activity
codes and MET intensities. Med Sci Sports Exerc 2000;32(S1):S498–
516.
68. Tudor-Locke C, Washington TL, Ainsworth BE, Troiano RP. Linking
the American Time Use Survey (ATUS) and the Compendium of Physi-
cal Activities: methods and rationale. J Phys Act Health 2009;6:347–53.
69. R Core Team. R: A language and environment for statistical computing
(Version 3.2.2, “Fire Safety”), 2015; http://www.R-project.org/ (ac-
cessed November 13, 2018).
70. Nikita E, Mattingly D, Lahr MM. Methodological considerations in the
statistical analysis of degenerative joint and disc disease. Int J Paleopath
2013;3:105–12.
71. Alves-Cardoso F, Assis S. Can osteophytes be used as age at death esti-
mators? Testing correlations in skeletonized human remains with known
age-at-death. Forensic Sci Int 2018;288:59–66.
72. Jurmain RD. Stress and the etiology of osteoarthritis. Am J Phys
Anthropol 1977;46:353–66.
73. Weiss E, Jurmain R. Osteoarthritis revisited: a contemporary review of
aetiology. Int J Osteoarchaeol 2007;17:437–50.
74. Calce SE, Kurki HK, Weston DA, Gould L. The relationship of age,
activity, and body size on osteoarthritis in weight-bearing skeletal
regions. Int J Paleopathol 2018;22:45–53.
75. Tak E, Staats P, Van Hespern A, Hopman-Rock M. The effects of an
exercise program for older adults with osteoarthritis of the hip. J
Rheumatol 2005;32:1106–13.
76. Urquhart DM, Tobing JFL, Hanna FS, Berry P, Wluka AE, Ding C,
et al. What is the effect of physical activity on the knee joint? A sys-
tematic review. Med Sci Sports Exerc 2011;43:432–42.
77. Wallace IJ, Worthington S, Felson DT, Jurmain RD, Wren KT, Maija-
nen H, et al. Knee osteoarthritis has doubled in prevalence since the
mid-20th century. Proc Natl Acad Sci USA 2017;114:9332–6.
78. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of childhood
and adult obesity in the United States, 2011-2012. JAMA
2014;311:806–14.
79. Merritt CE. Inaccuracy and bias in adult skeletal age estimation: assess-
ing the reliability of eight methods on individuals of varying body sizes.
Forensic Sci Int 2017;275(315):e1–11.
80. Calce SE, Kurki HK, Weston DA, Gould L. Effects of osteoarthritis on
age-at-death estimates from the human pelvis. Am J Phys Anthropol
2018;167:3–19.
81. Brennaman AL. Examination of osteoarthritis for age-at-death estima-
tion in a modern population [thesis]. Boston, MA: Boston University,
2014.
82. Loeser RF. Aging cartilage and osteoarthritis cartilage: differences and
shared mechanisms. In: Sharma L, Berenbaum F, editors. Osteoarthritis:
a companion to rheumatology. Philadelphia, PA: Mosby, Inc. (Elsevier),
2007;77–84.
83. Wollheim FA, Lohmander LS. Pathology and animal models of
osteoarthritis. In: Sharma L, Berenbaum F, editors. Osteoarthritis: a
companion to rheumatology. Philadelphia, PA: Mosby, Inc. (Elsevier),
2007;104–112.
84. Brennaman AL, Love KR, Bethard JD, Pokines JT. A Bayesian
approach to age-at-death estimation from osteoarthritis of the shoulder
in modern North Americans. J Forensic Sci 2017;62:573–84.
85. Falys CG, Prangle D. Estimating age of mature adults from the degen-
eration of the sternal end of the clavicle. Am J Phys Anthropol
2015;156:203–14.
86. Boldsen JL, Milner GR, Konigsberg LW, Wood JW. Transition anal-
ysis: a new method for estimating age from skeletons. In: Hoppa
RD, Vaupel JW, editors. Paleodemography: age distributions from
skeletal samples. Cambridge, U.K.: Cambridge University Press,
2002;73–106.
87. Bullock M, M
arquez L, Hern
andez P, Ru
ız F. Paleodemographic age-
at-death distributions of two Mexican skeletal collections: a comparison
of transition analysis and traditional aging methods. Am J Phys Anthro-
pol 2013;152:67–78.
88. Cappella A, Cummaudo M, Arrigoni E, Gibelli DM, Porta D, Cattaneo
C. The issue of age estimation in a modern skeletal population: are cur-
rent aging methods satisfactory for the elderly? Proceedings of the 67th
Annual Scientific Meeting of the American Academy of Forensic
Sciences; 2015 Feb 16-21; Orlando, FL. Colorado Springs, CO: Ameri-
can Academy of Forensic Sciences, 2015.
89. Langley-Shirley N, Jantz RL. A Bayesian approach to age estimation in
modern Americans from the clavicle. J Forensic Sci 2010;55:571–83.
90. Calce SE, Kurki HK, Weston DA, Gould L. Principal component analy-
sis in the evaluation of osteoarthritis. Am J Phys Anthropol
2017;162:476–90.
91. Dieppe P. Developments in osteoarthritis. Rheumatology 2011;50:245–
7.
92. Kellgren JH, Lawrence JS. Radiological assessment of osteoarthritis.
Ann Rheum Dis 1957;16:494–502.
93. Rothschild BM. Porosity: a curiosity without diagnostic significance.
Am J Phys Anthropol 1997;104:529–33.
94. Smalley KJ, Knerr AN, Kendrick ZV, Colliver JA, Owen OE.
Reassessment of body mass indices. Am J Clin Nutr 1990;52:405–8.
95. Wellens RI, Roche AF, Khamis HJ, Jackson AS, Pollock ML, Siervogel
RM. Relationships between the body mass index and body composition.
Obes Res 1996;4:35–44.
96. Rush EC, Freitas I, Plank LD. New standards of weight and body com-
position by frame size and height for assessment of nutritional status of
adults and the elderly. Br J Nutr 2009;102:632–41.
97. Wang J, Thornton JC, Russell M, Burastero S, Heymsfield S, Pierson
RN. Asians have lower body mass index (BMI) but higher percent body
fat than do whites: comparisons of anthropometric measurements. Am J
Clin Nutr 1994;60:23–8.

98. Keys A, Fidanza F, Karvonen MJ, Kimura N, Taylor HL. Indices of
relative weight and obesity. J Chronic Dis 1972;25:329–43.
99. Flegal KM, Carroll MD, Ogden CL, Curtin LR. Prevalence and trends
in obesity among US adults, 1999-2008. JAMA 2010;303:235–41.
100. Daubert V. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).
WINBURN . VALIDATION OF THE ACETABULUM FOR AGING 15
101. National Research Council Committee on Identifying the Needs of the
Forensic Sciences Community. Strengthening forensic science in the
United States: a path forward (Document No. 228091). Washington,
DC: The National Academies Press, 2009.