close

Вход

Забыли?

вход по аккаунту

?

Evaluation of juvenile stature and body mass prediction.

код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 136:387–393 (2008)
Evaluation of Juvenile Stature and
Body Mass Prediction
Paul W. Sciulli* and Samantha H. Blatt
Department of Anthropology, Ohio State University, Columbus, Ohio 43210
KEY WORDS
juvenile; stature; body mass; accuracy; bias
ABSTRACT
This investigation evaluates the performance of juvenile stature (from tibia and radius
lengths) and body mass (from breadth of the femoral
distal metaphysis) prediction equations based on the
Denver Growth Study sample (Ruff C. 2007. Am J
Phys Anthropol 133 698–716). The sample used here
for evaluation is an independent sample of juveniles
brought to the Franklin County (Ohio) Coroner in
1990–1991. The Ohio sample differs somewhat from
the Denver reference sample: it includes 25% AfricanAmericans (rather than all European-Americans), a
significant number of right limb bones were measured
(rather than all left side), it includes a wider range of
economic statuses and it includes individuals who died
from disease and trauma. As such the composition and
measures of the Ohio sample correspond more generally to that seen in skeletal samples so that the accuracy of the estimates from the present sample should
approach those found in practical applications of these
methods. Results indicate that both juvenile body mass
and stature are estimated relatively accurately. Accuracy of body mass estimates for 1-13-year-old juveniles
is similar for African-American and European-American males and females. The least accurate estimates
are for individuals in the 8–13 years age class
(excluding individuals with body mass indices greater
than the age specific 95th percentile): n 5 9, 6 2.9 kg,
95% confidence interval 1.4–4.4 kg. Accuracy of stature
estimates for 1-17-year-old juveniles is comparable for
the tibia and radius and, as with body mass estimates,
are similar for African-American and EuropeanAmerican males and females. For combined age,
sex, and ancestry groups average accuracies are in the
63.5 to 66.5 cm range. Some limitations of the methods are discussed. Am J Phys Anthropol 136:387–393,
2008. V 2008 Wiley-Liss, Inc.
Ruff (2007) presented juvenile stature and body mass
prediction equations based on skeletal measures for a
subset of the Denver Growth Study sample (McCammon,
1970). The prediction equations followed 20 individuals
longitudinally, 10 males and 10 females with the most
complete radiographic information, and covered yearly
intervals from 1 to 17 years of age. The results of the
analyses indicated that stature and body mass can be
estimated from juvenile skeletal measures with errors
equal to or less than errors associated with stature and
body mass estimates for adults.
As noted by Ruff (2007), variation in body size is commonly used to assess juvenile health status within and
among populations. Comparisons among living and
recent populations are facilitated by the data available
for a large number of living and recent populations (Eveleth and Tanner, 1990). The ability to estimate accurately juvenile stature and body mass in skeletal samples would allow more direct comparisons between these
populations and living populations. Furthermore, these
estimates would expand information on growth and development to include a greater range of genetic compositions, environmental situations, and genotype-environment interactions. Accurate estimates of juvenile stature
and body mass would also provide valuable forensic information for identification of recent skeletal remains.
However, even though the results of the analyses of the
Denver Growth Study sample indicated that juvenile
stature and body mass could be estimated relatively
accurately the investigation had a number of limitations
which in practice could affect the accuracy of the estimates when the methods are applied to other samples.
These limitations include the small sample sizes (n 5
18–20) on which the prediction equations are based, the
composition of the sample which consists of modern juveniles from the United States, mostly of northern European ancestry, and primarily middle to upper class, prediction equations derived from skeletal measures from
the left side only, and the necessity for an age determination to the nearest year in order to apply the prediction equations (Ruff, 2007). In addition, errors in estimating juvenile body size differ for the skeletal measures. For example, femoral head breadth has smaller
associated errors in estimating body mass than the distal femoral metaphyseal breadth in the 7–13 years age
range and lower limb bone lengths provide better stature
estimates than upper limb bone lengths (Ruff, 2007).
Thus, it would be useful to know how these prediction
equations perform for an independent sample of individuals that is known to differ in some respects from the
composition of the original study sample, whose skeletal
measures differ (right and left sides) from those in the
original sample, and does not have measures available
that have yielded body size estimates with the smallest
associated errors.
C 2008
V
WILEY-LISS, INC.
C
*Correspondence to: Paul W. Sciulli, Department of Anthropology,
The Ohio State University, 219 Lord Hall, 124 W. 17th Ave., Columbus, Ohio 43210. E-mail: sciulli.1@osu.edu
Received 9 November 2007; accepted 28 January 2008
DOI 10.1002/ajpa.20820
Published online 18 March 2008 in Wiley InterScience
(www.interscience.wiley.com).
388
P.W. SCIULLI AND S.H. BLATT
TABLE 1. Age and ancestry distribution of femur distal metaphysis breadth and the accuracy and bias of
estimates of body mass from this measure
Males
Females
Age
Euro-Am
N
FMETa
Afro-Am
N
FMET
Total
1
2
3
4
5
6
7
8
9
10
11
12
13
2
4d
4
2
1d
1
1d
–e
1
1
–
1
3
–
2
1
–
1
–
–
–
–
–e
1
–d
–
37.9
41.8
47.8
45.9
50.3
54.2
62.3
–
61.8
65.7
65.6
64.5
70.1
37.9
41.7
48.1
45.9
49.9
54.2
62.3
–
61.8
65.7
–
64.5
70.1
–
42.0
46.5
–
50.7
–
–
–
–
–
65.6
–
–
Euro-Am
N
FMET
Afro-Am
N
FMET
Total
Total
Mean
8
1
–
–
–
–d
–
–
–
–
2
2
–
–
–
–
–
–
–
–
–d
–
–
31.8
39.1
–
–
–
–
–
–
–
–
–
66.9
64.4
32.8
40.9
47.8
45.9
50.3
54.2
62.3
–
61.8
65.7
65.6
65.7
68.7
1
1
29.7
45.4
–
–
–
–
–
–
–
–
–
66.9
64.4
40.1
35.9
–
–
–
–
–
–
–
–
–
–
–
Combined sex and
ancestry
Accuracyb
Biasc
61.1
61.6
61.7
61.5
62.9
61.4
63.2
–
60.8
64.2
62.4
61.5
64.0
10.8
11.6
11.7
11.5
12.9
11.4
13.2
–
10.8
14.2
12.4
20.5
12.8
Body mass estimates are from ordinary least squares regressions of loge transformed data.
a
FMET is mean femur distal metaphysis breadth (mm); Total is combined ancestry mean of this measure; Total mean is combined
sex-ancestry mean for age class.
b
|Observed body mass – estimated body mass|; kilograms; average by age class, sexes, and ancestry combined.
c
(Observed body mass – estimated body mass); kilograms; average by age class, sexes, and ancestry combined.
d
One individual with BMI [ 95th percentile removed.
e
Two individuals with BMI [ 95th percentile removed.
TABLE 2. Accuracy and bias of body mass estimates from ordinary least squares regressions of loge data
by age groups, sex, and ancestry
Accuracya
Age
1–13
1–13
1–13
1–13
1–13
1–7
1–7
1–7
1–7
1–7
8–13
b
Sex
M, F
M, F
M
M
F
M, F
M, F
M
M
F
M,F
c
Ancestry
AA, EA
AA, EA
AA, EA
EA
AA, EA
AA, EA
AA,EA
AA, EA
EA
AA,EA
AA,EA
N
d
51
41
26
21
15
36d
32
19
15
13
9
Biasa
Mean
95% CI
Mean
95% CI
63.8
61.9
62.1
62.2
61.5
62.3
61.6
61.8
61.9
61.2
62.9
(2.0,5.6)
(1.5,2.3)
(1.6,2.6)
(1.6,2.8)
(0.6,2.4)
(1.4,3.2)
(1.2,2.0)
(1.3,2.3)
(1.3,2.5)
(0.7,1.7)
(1.4,4.4)
13.1
11.6
11.7
11.8
11.3
12.2
11.5
11.8
11.9
10.9
12.0
(1.2,5.0)
(1.0,2.2)
(1.0,2.4)
(1.0,2.6)
(0.3,2.3)
(1.3,3.3)
(1.1,1.9)
(1.3,2.3)
(1.3,2.5)
(0.2,1.6)
(20.3,4.3)
a
Kilograms; see Table 1 for definitions.
M is male, F is female.
c
AA is Afro-American, EA is Euro-American.
d
Includes individuals with BMI [ 95th percentile.
b
The purpose of the present investigation is to evaluate
the performance of a subset of Ruff ’s (2007) juvenile
stature and body mass prediction equations for an independent sample of recent juveniles of known age, sex,
ancestry, stature, and body mass and which does not
conform in all respects to the assumptions of the methods or the manner of measurement.
MATERIALS AND METHODS
The individuals employed in the present study are
taken from the sample of 186 juveniles from central
Ohio brought to the Franklin County (Ohio) Coroner and
who died between July 1, 1990 and June 30, 1991. (Pfau
and Sciulli, 1994; Sciulli and Pfau, 1994). Dates of birth
and death, sex, ancestry, weight, and stature were available for each individual. Weight and stature were
recorded from medical records (at most 3 days old) for
American Journal of Physical Anthropology
hospitalized individuals or individuals under a physician’s care. Six trauma cases were measured within 2 h
of death and results checked with medical records. European-American and African-American males and females
were used in the analyses. Tables 1–6 contain the samples sizes available for analysis by age, sex, and ancestry. Radiographs were obtained for the long bones of the
left limbs where possible using a General Electric Mobile
225 X-ray unit. Anterior–posterior radiographs were
taken at 40 in (101.6 cm) from the radiation source. A
radio-opaque meter stick was positioned parallel to and
at the mid-level of the limbs in each radiograph. All radiographs were made using 14 3 17 X-ray film placed in
a 14 3 17 cassette equipped with a detail screen to
ensure a high degree of resolution. Long bones exceeding
17 in (43.18 cm) were X-rayed with the film in a diagonal position. For the few cases where the long bone
exceeded 22 in (55.88 cm) two separate but overlapping
radiographs were obtained. The radio-opaque meter stick
389
JUVENILE BODY SIZE
TABLE 3. Age and ancestry distribution of tibia length and the accuracy and bias of estimates of statures from this measure
Males
Euro-Am
Age
c
1
2
3
4
5
6
7
8
9
10
11
12
13d
14
15
16
17
Females
Afro-Am
Euro-Am
Afro-Am
N
TIBIAa
N
TIBIA
Total
N
TIBIA
N
TIBIA
Total
Total
Mean
2
5
4
2
2
1
2
2
1
1
–
2
–
–
3
1
4
114.5
135.3
161.9
170.2
196.3
198.1
238.4
247.5
258.0
270.7
–
304.2
–
–
391.0
444.6
392.4
1
2
1
–
1
–
–
–
–
2
1
1
–
–
–
1
–
89.2
134.8
166.2
–
198.9
–
–
–
–
304.9
285.9
326.0
–
–
–
387.5
–
106.0
135.2
162.8
170.2
197.2
198.1
238.4
247.5
258.0
293.6
285.4
311.5
–
–
391.0
416.0
392.4
8
1
–
–
–
1
–
–
–
1
–
1
–
–
4
2
2
99.5
134.5
–
–
–
200.2
–
–
–
271.2
–
285.8
–
–
368.3
361.8
391.8
2
2
–
–
–
–
–
–
–
–
1
–
1
1
–
–
–
116.0
117.8
–
–
–
–
–
–
–
–
296.6
–
388.4
368.4
–
–
–
102.8
123.3
–
–
–
200.2
–
–
–
271.2
296.6
285.8
388.4
368.4
368.3
361.8
391.8
103.5
131.6
162.8
170.2
197.2
199.1
238.4
247.5
258.0
288.0
291.0
305.1
388.4
368.4
378.0
389.0
392.2
Combined sex and
ancestry
Accuracyb
Biasb
62.9
63.0
64.1
65.2
63.3
61.6
66.4
615.6
61.0
64.9
66.1
66.7
60.4
68.5
66.8
65.6
66.3
20.7
10.5
13.8
20.5
20.2
11.6
26.4
215.6
11.0
24.9
26.1
22.7
20.4
28.5
24.6
25.6
26.2
Stature estimates are from ordinary least squares regressions of untransformed data.
a
TIBIA is mean of tibia length (mm); TOTAL is combined ancestry mean of this measure; TOTAL MEAN is combined sex-ancestry
mean for age class.
b
Centimeters; see Table 1 for definitions.
c
For ages 1–12 diaphyseal length.
d
For ages 13–17 total length.
TABLE 4. Accuracy and bias of stature estimates from tibia based on ordinary least squares regressions of untransformed data
Accuracya
b
F
c
Biasa
Age
Sex
Ancestry
N
Mean
95% CI
Mean
95% CI
1–17
1–17
1–17
1–17
1–17
1–17
1–7
1–7
1–7
1–7
1–7
1–7
1–7
8–13
14–17
14–17
M,
M
F
M,
M,
M
M,
M
F
M,
M,
M
F
M,
M,
M,
AA, EA
AA, EA
AA, EA
EA
AA
EA
AA,EA
AA, EA
AA, EA
EA
AA
EA
EA
AA, EA
AA, EA
EA
69
42
27
52
17
32
37
23
14
28
9
18
10
14
18
16
64.8
64.5
65.3
64.8
65.0
64.6
63.4
63.2
63.6
63.3
63.6
63.6
62.7
66.5
66.5
66.2
(3.8,5.8)
(3.1,5.9)
(4.0,6.6)
(3.5,6.1)
(3.2,6.8)
(2.9,6.3)
(2.6,4.2)
(2.0,4.4)
(2.3,4.9)
(2.3,4.3)
(1.8,5.4)
(2.2,5.0)
(1.4,4.0)
(2.7,10.3)
(4.3,8.7)
(3.8,8.6)
22.5
22.2
22.9
22.3
23.1
21.7
10.1
10.1
10.1
10.2
20.1
10.2
10.1
25.2
25.6
25.2
(23.9,21.1)
(24.1,20.3)
(25.4,20.4)
(24.0,20.6)
(25.8,20.4)
(24.0,10.6)
(21.3,11.5)
(21.7,11.9)
(22.4,12.6)
(21.4,11.8)
(23.6,13.4)
(22.1,12.5)
(22.3,12.5)
(29.6,20.8)
(28.4,22.8)
(28.3,22.1)
F
F
F
F
F
F
F
F
By age groups, sex, and ancestry.
a
Centimeters; see Table 1 for definitions.
b
M is male, F is female.
c
AA is Afro-American, EA is Euro-American.
was included as a reference point in each case to ensure
accurate measures.
Juvenile body mass can be estimated from femoral
head breadth or femoral distal metaphyseal breadth
(Ruff, 2007). Since the proximal femur was not included
in the original radiographs, femoral distal metaphyseal
breadth was used to estimate body mass. Where possible
for each individual (n 5 51) in the age class range 1–13
years (0.5–13.5 years) the left distal metaphyseal
breadth was measured to the nearest 0.05 mm using a
Vernier caliper (capacity 60 cm). However, due to trauma
or unclear radiographs on the left side approximately
one-quarter of the sample consisted of right distal metaphyseal breadths. The age classes are 60.5 years from a
given integer year and correspond to Ruff ’s age classes.
For example, age class 1 includes individuals from 0.5 to
1.5 years of age. The age class range 1–13 was used
since body mass prediction equations for the femoral distal metaphyseal breadth are available for this range of
ages (Ruff, 2007).
Femur, tibia, radius, and humerus length can be used
to estimate juvenile stature (Ruff, 2007). Two measures
were available to estimate stature in this sample: tibia (n
5 69) and radius (n 5 77) length. Measurement of the tibia
and radius follow Ruff’s (2007) method. Diaphyseal lengths
of the left tibia and radius for individuals in the age range
1–12 years (0.5–12.5 years) were measured as maximum
lengths parallel to the long axes of the diaphyses between
American Journal of Physical Anthropology
390
P.W. SCIULLI AND S.H. BLATT
TABLE 5. Age and ancestry distribution of radius length and the accuracy and bias of estimates of statures from this measure
Males
Euro-Am
Age
c
1
2
3
4
5
6
7
8
9
10
11
12
13d
14
15
16
17
Females
Afro-Am
Euro-Am
Afro-Am
N
Radiusa
N
Radius
Total
N
Radius
N
Radius
Total
Total
Mean
2
5
4
2
2
1
2
2
1
1
1
2
–
–
2
2
8
82.4
95.5
109.7
118.8
129.8
126.0
156.5
159.7
168.1
180.0
175.4
197.0
–
–
247.3
256.1
248.4
1
2
1
–
1
–
–
–
–
1
1
1
–
–
2
2
2
66.1
93.6
114.9
–
126.8
–
–
–
–
170.4
176.4
210.2
–
–
228.3
257.4
279.6
77.0
95.0
110.7
118.8
128.8
126.0
156.5
159.7
168.1
175.2
175.9
201.4
–
–
237.8
256.7
254.7
8
1
–
–
–
1
–
–
–
1
–
1
–
–
4
2
2
72.9
97.4
–
–
–
134.7
–
–
–
161.2
–
189.8
–
–
232.7
222.4
237.2
2
2
–
–
–
–
–
–
–
–
1
–
1
–
–
–
–
82.3
82.4
–
–
–
–
–
–
–
–
193.4
–
234.9
–
–
–
–
74.8
87.4
–
–
–
134.7
–
–
–
161.2
193.4
189.8
234.9
–
232.7
222.4
237.2
75.3
92.7
110.7
118.8
128.8
130.3
156.5
159.7
168.1
170.5
181.8
198.5
234.9
–
235.3
245.3
251.7
Combined sex and
ancestry
Accuracyb
Biasb
63.5
63.0
64.7
66.4
63.6
62.1
67.0
615.2
60.6
63.0
67.7
65.5
66.7
–
64.6
63.8
66.0
20.2
10.3
14.2
23.3
10.7
12.1
27.0
215.2
20.6
20.3
25.7
23.5
16.7
–
23.6
22.8
21.9
Stature estimates are from ordinary least squares regressions of untransformed data.
a
Radius is mean of radius length; Total is combined ancestry mean of this measure; Total mean is combined sex-ancestry mean.
b
Centimeters; see Table 1 for definitions.
c
For ages 1–12 diaphyseal length.
d
For ages 13–17 total length.
TABLE 6. Accuracy and bias of stature estimates from radius based on ordinary least squares regressions of untransformed data
Accuracya
Biasa
Age
Sexb
Ancestryc
N
Mean
95% CI
Mean
95% CI
1–17
1–17
1–17
1–17
1–17
1–17
1–7
1–7
1–7
1–7
1–7
1–7
1–7
8–13
14–17
14–17
14–17
M,
M
F
M,
M,
M
M,
M
F
M,
M,
M
F
M,
M,
M
M,
AA, EA
AA, EA
AA, EA
EA
AA
EA
AA,EA
AA, EA
AA, EA
EA
AA
EA
EA
AA, EA
AA, EA
AA, EA
EA
77
51
26
57
20
37
37
23
14
28
9
18
10
14
26
18
20
64.8
65.0
64.5
64.8
64.9
65.3
64.2
64.4
64.0
64.1
64.8
64.6
63.1
65.4
65.3
65.9
65.5
(3.9,5.7)
(3.8,6.2)
(3.2,5.8)
(3.7,5.9)
(3.6,6.2)
(3.7,6.9)
(3.2,5.2)
(3.1,5.7)
(2.3,5.7)
(3.0,5.2)
(2.4,7.2)
(3.1,6.1)
(1.2,5.0)
(2.5,8.3)
(3.6,7.0)
(3.6,8.2)
(3.3,7.7)
11.8
12.1
11.0
11.8
11.7
12.4
12.0
12.3
11.4
11.9
12.1
12.4
11.0
20.4
12.6
14.4
13.2
(0.4,3.2)
(0.4,3.8)
(21.2,13.2)
(0.2,3.4)
(20.2,15.0)
(0.2,4.6)
(0.4,3.6)
(0.2,4.4)
(21.4,14.2)
(0.1,3.7)
(22.1,16.3)
(20.1,14.9)
(21.9,13.9)
(24.8,14.0)
(0.0, 5.2)
(1.4,7.4)
(0.1,6.3)
F
F
F
F
F
F
F
F
F
By age group, sex, and ancestry.
a
Centimeters; see Table 1 for definitions.
b
M is male, F is female.
c
AA is Afro-American, EA is Euro-American.
the proximal and distal ends not including the epiphyses.
The total maximum lengths of the left tibia and radius,
including the epiphyses were measured from the most distal edge of the bone to the most proximal for individuals in
the age range 13–17 years (12.5–17.5 years). Again, due to
trauma or unclear radiographs on the left side the sample
consisted of about one-third right radii and tibiae. Measures for the specific age ranges reflect the availability of
prediction equations for the measures (Ruff, 2007).
For each of the three measures (femoral distal metaphyseal breadth, radius length, tibia length) a sub-sample of 20 individuals was measured a second time, following an interval of at least 2 weeks, to evaluate measurement error. A one-way analysis of variance was
American Journal of Physical Anthropology
performed for each measure. The error mean square provided the average variance between the two measures
for each individual and the square root of the error
mean square provided the average standard deviation in
the units of measurement (mm). The average difference
between the two measures was also noted. All measures
were taken by SHB.
Body mass was estimated for each individual using
the age-specific equations in Ruff ’s (2007) Table 2 for
loge (femoral distal metaphyseal breadth) and the correction for detransformation bias. Stature was estimated for
each individual using the age-specific equations in Ruff ’s
(2007) Tables 4 and 5 for the tibia and radius respectively.
391
JUVENILE BODY SIZE
Two measures were used to evaluate the performance
of the prediction equations: accuracy and bias. Accuracy
is the absolute value of the difference between observed
and predicted body mass or stature and bias is signed
difference between observed and predicted. Average accuracy and bias were calculated for each age class and
average accuracy, bias and their 95% confidence intervals (ci) were calculated for combined age classes by sex
and ancestry.
RESULTS
The analysis of measurement error for femoral distal
metaphyseal breadth, radius length, and tibia length
indicates that measurement error is not a major source
of variation. The average difference between the first
and second measures and the average standard deviation of the two measures (n 5 20) for the three features
is 0.3 mm and 0.3 mm, 0.6 mm and 0.7 mm, and 1.1 mm
and 1.3 mm, respectively.
Table 1 contains the average femoral distal metaphyseal breadth by age class, sex, and ancestry. The means
in all cases except age class 7 of European-American
males fall within the ranges of the Denver reference
sample (Ruff, 2007; Table 1). However European-American male age class 7 is represented by a single individual and his measure is only 2.5 mm above the range of
the Denver reference sample. The accuracy of body mass
estimates by age class for combined sex and ancestry
ranges from 60.8 kg in age class 9 to 64.0 kg in age
class 13. Bias by age class for pooled sex and ancestry is
positive for all but age class 12 (20.5 kg) and the positive values range from 0.8 kg for age classes 1 and 9 to
4.2 for age class 10.
Table 2 contains the mean accuracy and bias of body
mass estimates by sex and ancestry for combined age
classes. Age classes were combined into 1–7 and 8–13
groups as Ruff (2007) noted body mass estimation errors
are smallest in years 2–7 (slightly higher in year 1) and
greatly increases from year 8 onwards. Small sample
sizes precluded combining age classes into smaller
groups. In the combined age class 1–13 accuracy (61.5
to 2.1 kg) and bias (11.3 to 11.6 kg) are similar for
males and females and for African-Americans and European-Americans when individuals (n 5 10) with a body
mass index greater than the 95th percentile for an age
class are excluded (Must et al., 1991a,b). The combined
age class1–7 show somewhat greater accuracies (61.2 to
1.9 kg) than the total combined sample (1–13; 61.5–2.2
kg) but in general the results for both combined age
classes are comparable for accuracy and bias. As
expected (Ruff, 2007), accuracy of body mass prediction
is somewhat lower in the older juveniles (combined age
classes 8–13; 62.9 kg).
Table 3 contains the average tibia length by age
class, sex, and ancestry. The total mean for each age
class falls within the range of the Denver reference
sample (Ruff, 2007; Table 1). The only values that fall
outside of the Denver reference sample ranges are for
single males in age class 16 (European-American, 57.3
mm above range) and age class 1 (African-American,
8.8 mm below range) and the two African-American
females in age class 2 (7.2 mm below range).The accuracy of stature estimates by age class for combined sex
and ancestry ranges from 60.4 cm in age class 13 to
615.6 cm in age class 8. Bias by age class for pooled
sex and ancestry is negative for 13 of the 17 age classes
and the negative values range from 0.7 cm for age class
1 to 15.6 cm for age class 8.
Table 4 contains the mean accuracy and bias of stature
estimates from tibia length by sex and ancestry for combined age classes. In the combined age class 1–17 accuracy (64.5 to 5.3 cm) and bias (21.7 to 23.1 cm) are
similar for males and females and for African-Americans
and European-Americans. The combined age class 1–7
shows a somewhat greater accuracy (62.7 to 3.6 cm)
than age classes 14–17 and 8–13 (66.2 to 6.5 cm) for sex
and ancestry groupings. Bias is low in the 1–7 combined
age class (20.1 to 10.2 cm) and larger but similar in the
8–13 and 14–17 combined age classes (25.2 to 25.6 cm).
Bias in the total combined age class (1–17) is negative
for all sex and ancestry groupings (21.7 to 23.1).
Table 5 contains the average radius length by age
class, sex, and ancestry. The total mean for each age
class falls within the range of the Denver reference sample (Ruff, 2007; Table 1). The only values that fall outside of the Denver reference sample ranges are AfricanAmericans males in age class 1 (n 5 1, 5.9 mm below
range) and age class 17 (n 5 2, 10.6 mm above range)
and African-American females in age class 2 (n 5 2, 4.6
mm below range). The accuracy of stature estimates by
age class for combined sex and ancestry ranges from
60.6 cm in age classes 9 to 615.2 cm in age class 8. Bias
by age class for pooled sex and ancestry is negative for
11 of the 16 age classes for which data are available and
the negative values range from 0.2 cm for age class 1 to
15.2 cm for age class 8.
Table 6 contains the mean accuracy and bias of stature
estimates from radius length by sex and ancestry for
combined age classes. In the combined age classes 1–17
accuracy (64.5 to 5.3 cm) and bias (11.0 to 12.4 cm) are
similar for males and females and for African-Americans
and European-Americans. The combined age class 1–7
shows a somewhat greater accuracy (63.1 to 4.8 cm)
than age classes 14–17 and 8–13 (65.3 to 5.9 cm). Bias
for all combined age classes except combined age class
8–13 is positive and comparable for the sex and ancestry
groups.
DISCUSSION AND CONCLUSIONS
The present sample differs from the specific models
on which the Denver reference sample age-specific prediction equations for juvenile body mass and stature
are based. In the present sample a significant number
of bones from right limbs were measured (rather than
all left). The present sample consists of 25% AfricanAmericans (rather than all European-Americans) and
includes a wider cross section of economic statuses
than the Denver reference sample. Finally, the present
sample consists of individuals who died of acute and
chronic illnesses or trauma. However, evaluation of the
methods of estimation of juvenile body mass and stature by the present sample using ordinary least squares
regression from the Denver reference sample is appropriate since the present sample shows very little difference from the Denver sample in age-specific size of all
three measures used for estimation (femoral distal
metaphyseal breadth and radius and tibia lengths). In
addition, the wider applicability of these estimation
American Journal of Physical Anthropology
392
P.W. SCIULLI AND S.H. BLATT
methods is evaluated here since this is an independent
sample and measurement error is relatively small.
The inclusion of individuals who died from illness or
trauma and the wider cross section of economic statuses
in the present sample make the present sample somewhat more similar in composition than the Denver reference sample to what might be expected in skeletal samples. The accuracy of body size estimates for the present
sample should approach those in the practical applications of these methods in skeletal samples and in many
forensic cases.
Body mass estimates in the present sample, considering the inherent variability of body mass (Eveleth and
Tanner, 1990), are reasonably accurate across age
classes, sexes, and ancestral groups when individuals
with body mass indices greater than the 95th percentile
(obese) are excluded. Across age classes for combined
sexes and ancestral groups the least accurate estimate of
average body mass is 64.0 kg in age class 13. For combined age classes the least accurate estimate is 62.9 kg
for the combined male-female, African-American and European-American age class 8–13. Thus body mass estimates from the Denver reference sample equations yield
reasonably accurate estimates for nonobese males and
females and for African-Americans and European-Americans in the present sample.
The major limitation in the application of Denver reference sample body mass estimation equations to the
present sample concerns obese juveniles. Although
these individuals have femoral distal metaphyseal
breadths within the range of the Denver reference sample their body mass estimates are much lower than
their measured body mass. This limitation will probably not result in systematically under-estimating body
mass in most previous populations in which nutritional
stress was prevalent (Steckel and Rose, 2002) but it
may systematically affect body mass estimates for
recent juveniles due to the increased prevalence of obesity (Popkin et al., 2006). This latter affect is reflected
in the generally small but positive biases of the body
mass estimates for virtually all ages, sexes, and ancestral groups in the present sample. The application of
these methods of body mass estimation will have to
assume the individuals were not obese.
The accuracy of juvenile stature estimates in the present sample is comparable for tibia and radius length.
Aside from two short (for age) 8-year old EuropeanAmerican males (116 and 119 cm) and a tall thirteen
year old African-American female (175 cm) the mean accuracy of stature estimates by age class for combined
sexes and ancestry ranged from 61.0 to 8.5 cm (median
5.0 cm) for the tibia and 60.6 to 7.7 cm (median 4.6 cm)
for the radius. Mean accuracy of stature estimates for
the tibia and radius by age classes (1–13 and 15–17) for
combined sexes and ancestry (Table 3 and 5) are
strongly correlated (r 5 0.83). Mean accuracy by sex and
ancestry for combined age classes (Table 4 and 6) are
similar for the radius and tibia and show overlapping
95% confidence intervals for all mean accuracies. For the
combined age classes the average accuracy is greatest
for sex and ancestral groupings in age class 1–7 (less
than 64.8 cm) followed by age class 14–17 and 8–13
(both less than 66.5 cm). These accuracy estimates are
similar to the confidence intervals that can be calculated
from the Denver reference sample (Ruff, 2007). For
example, an individual in the 13 year age class whose
tibia length is the same as the mean length for the 13
American Journal of Physical Anthropology
year age class has a confidence interval of 66.9 cm; if
the tibia length is two standard deviations above the
mean of the reference sample the confidence interval is
67.5 cm.
The inclusion of African-Americans in the present
sample raises the possibility that errors in estimation
may have been increased due to differences from the
Denver reference sample in limb length to stature proportions. The approach suggested to assess the similarity
in proportions between the Denver reference sample and
cases for estimation is to calculate the crural index (tibia
length/femur length) in the cases for estimation and
compare them to the Denver reference sample (Ruff,
2007). This is not possible for the present study as femur
length was not systematically collected. However, the
results (Tables 4 and 6) show that both accuracy and
bias of stature estimates are similar for African-Americans and European-Americans indicating that differences in body proportions had little or no effect on stature estimates in the present sample or that the AfricanAmericans in this sample have body proportions similar
to European-Americans.
Bias by age class for combined sex and ancestry
groups for stature estimated from tibia and radius
length in the present sample is negative for virtually all
age classes but similar in magnitude (Table 3 and 5).
Bias of stature estimates from radius and tibia length in
age classes (1–13 and 15–17) for combined sexes and
ancestry are, like the corresponding accuracy estimates,
strongly correlated (r 5 0.84). Bias of stature estimates
by sex and ancestry for combined age classes however
differs for tibia and radius lengths (Table 4 and 6). In
the total combined age class (1–17) bias is negative for
all sex and ancestry groupings for stature estimated
from the tibia but positive for all groupings for stature
estimated from the radius. Much of this difference
appears to stem from differences in the (14–17) year age
class. The (1–7) year age class shows virtually all positive biases for stature estimates from the tibia and radius and the 95% confidence intervals of the mean biases
all overlap. The (8–13) year age class shows negative
biases for stature estimates from both tibia and radius
length and, again, the 95% confidence intervals of the
mean biases overlap. It is only in the (14–17) year age
class where comparable sex and ancestry groupings
show bias of stature estimates from the tibia (negative)
and radius (positive) with opposite signs and nonoverlapping 95% confidence intervals of mean bias estimates.
The differences between stature estimate biases in the
(14–17) year class may be due to the small samples in
this age class.
The present sample, while appropriate, is not optimal
for evaluating all components of Ruff ’s (2007) system for
estimating juvenile body mass and stature. However, the
present sample does evaluate the wider applicability of
some of the prediction equations, at least for modern
samples from the United States. Also because the composition of the present sample is closer than the Denver
reference sample to the composition expected in most
skeletal samples, errors in body size estimation for the
present sample should approach the errors of estimation
found in practical applications of the methods evaluated
here. The predicted measures of body size from the Denver reference equations are reasonably accurate for the
present sample. Errors associated with predictions
should be contained in the 95% confidence intervals that
can be calculated for individual or sample estimates of
body size (Ruff, 2007).
JUVENILE BODY SIZE
LITERATURE CITED
Eveleth PB, Tanner JM. 1990. Worldwide variation in human
growth, 2nd ed. Cambridge: Cambridge University Press.
McCammon RW. 1970. Human growth and development.
Springfield, IL: Charles C. Thomas.
Must A, Dallal GE, Dietz WH. 1991a. Reference data for obesity: 85th and 95th percentiles of body mass index (wt/ht2)—a
correction. Am J Clin Nutr 54:773.
Must A, Dallal GE, Dietz WH. 1991b. Reference data
for obesity: 85th and 95th percentiles of body mass index (wt/
ht2) and triceps skinfold thickness. Am J Clin Nutr 53:839–
846.
393
Pfau RO, Sciulli PW. 1994. A method for establishing the age of
subadults. J Forensic Sci 39:165–176.
Popkin BM, Conde W, Hou N, Monteiro C. 2006. Is there a lag
globally in overweight trends in children compared with
adults? Obesity 14:1846–1853.
Ruff C. 2007. Body size prediction from juvenile skeletal remains. Am J Phys Anthropol 133:698–716.
Sciulli PW, Pfau RO. 1994. A method for estimating weight in
children from femoral midshaft diameter and age. J Forensic
Sci 39:1280–1286.
Steckel RH, Rose JC. 2002. The backbone of history: health and
nutrition in the western hemisphere. Cambridge: Cambridge
University Press.
American Journal of Physical Anthropology
Документ
Категория
Без категории
Просмотров
1
Размер файла
91 Кб
Теги
mass, stature, evaluation, body, juvenile, prediction
1/--страниц
Пожаловаться на содержимое документа