- Email: [email protected]
Estimation of stature from body parts
Forensic Science International 3501 (2003) 1–6
Estimation of stature from body parts ¨ zaslana,*, M. Yas¸ar I˙s¸canb, ˙Inci O ¨ zaslanc, Harun Tug˘cud, Sermet Koc¸a Abdi O a˙ ¨
IU Cerrahpasa Tıp Faku¨ltesi, Adli Tıp Anabilim Dalı, I˙stanbul 34303, Turkey b˙ ¨ IU Adli Tıp Enstitu¨su¨, ˙Istanbul, Turkey c Adli Tıp Kurumu, I˙stanbul, Turkey d GATA, Adli Tıp Anabilim Dalı, Ankara, Turkey Received 11 January 2002; accepted 15 November 2002
Abstract Anthropometric technique commonly used by anthropologists and adopted by medical scientists has been employed to estimate body size for over a hundred years. With the increasing frequency of mass disasters, the identification of an isolated lower extremity and the stature of the person it belonged to has created problems for the investigation of the identity of some of the victims. In spite of a need for such a study, there is a lack of systematic studies to identify fragmented and dismembered human remains. The purpose of the paper is to analyze anthropometric relationships between dimensions of the lower extremity and body height. Analysis is based on a sample of middle class male (N ¼ 203) and female (N ¼ 108) adult Turks residing in Istanbul. The participants are mostly students and staff members of a medical school, and military personnel. Measurements taken are stature, trochanteric height, thigh length, lower leg length, leg length, and foot height, breadth, and length. Of the five variables entered into the regression analysis, all but foot breadth participate in the analysis with leg length as the first and followed by thigh and foot lengths, and finally foot height in males (R2). There were also individually calculated formulae for some of these measurements which provided smaller R2-values. Student’s t-test to assess if there was any intraobserver error in measurements take by individual anthropometrist did not show such any statistically significant difference. In conclusion, the study suggested that estimation of a living height can be made possible using various dimensions of the lower extremity. One must consider differences between populations in order to apply functions as such to others. # 2002 Published by Elsevier Science Ireland Ltd. Keywords: Stature estimation; Dismember lower extremity; Anthropometry; Regression analysis; Mass disaster; Turks
1. Introduction Anthropometric technique commonly used by anthropologists and adopted by medical scientists has been employed to estimate body size for over a hundred years [1–6]. In the last half of the 20th century, studies have been more vigorous when skeletal collections have been assembled in many parts of the world [7,8]. There has been an increase in the frequency of various mass disasters (air and train crash, bombing, mass suicide, flooding, powerful storm). To the list one should add the more recent tragedies such as the
* Corresponding author. Tel.: þ90-212-588-4800; fax: þ90-212-588-0011. ¨ zaslan). E-mail address: [email protected] (A. O
Marmara earthquake, Turkey and the attack on the World Trade Center and Pentagon in the US where thousands of people were killed. There have been already some attempts using hand [9–15], lower leg length and arm span [16] and foot dimensions [17–20] to regress on body height. The identification of an isolated lower extremity and the stature of the person it belonged to has created problems for the investigation of identity of some of the victims found in the rubbles of the bombed Oklahoma Federal building. In spite of a need as such, there is a lack of systematic studies to identify fragmented and dismembered human remains. The purpose of the paper is to analyze anthropometric relationships between various dimensions of the lower extremity and body height. Analysis is based on a sample of middle class male and female adult Turks using various statistical subroutines particularly that of regression analysis.
0379-0738/02/$ – see front matter # 2002 Published by Elsevier Science Ireland Ltd. PII: S 0 3 7 9 - 0 7 3 8 ( 0 2 ) 0 0 4 2 5 - 5
2
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
2. Materials and methods The study is composed of 311 (203 males and 108 females) adult Turks. The participants are mostly from the Cerrahpasa Medical Faculty staff, students, staff from the Council of Forensic Medicine and military personnel who lived mostly in Istanbul. These individuals are considered middle class with at least 5 years of elementary education. Measurements are all taken from the left side according to the procedure described by the International Biological Program [21–23]. The subjects are asked to remove their clothing except underwear and shoes. They are oriented in the standard anatomic position with the head on the Frankfort Horizontal Plane. Males were measured by ¨ and HT and females by I˙O ¨ . Anthropometrists are all AO initially trained in the technique by MYI. For the convenience of the readers the dimensions are taken with an anthropometer in millimeters and 100 g units using standard anthropometric instruments such as sliding and spreading calipers, anthropometer, steel tape, and scale. Extremity measurements are all taken from the left side (Fig. 1) [22].
Stature is taken from the vertex to the floor obeying the anatomic position and Frankfort Plane. Trochanteric height is measured from the lateral bulging of the greater trochanteric protrusion to the floor. This measurement is preferred over the anterior iliac spine because of commonly S.E.E.n break and dismemberment in the hip joint. Thigh length is calculated from as distance between the trochanteric height and lower leg length. Lower leg length is the distance from the tibiale to the floor. Leg length is from the tibiale to the medial malleolus of the tibia. Foot height is calculated as the distance between lower leg length and leg length. Foot breadth is the distance between the lateral and medial sides at the metatarsal region using a sliding caliper. Foot length is the maximum distance between the most anterior and posterior projecting part of the foot with a sliding caliper. The data are analyzed using various subroutines of Statistical Package for the Social Sciences (SPSS) and
Fig. 1. Anthropometric dimensions used in the study: stature (a–e); trochanteric height (b–e); thigh length (b and c); lower leg length (c–e); leg length (c and d); foot height (d and e); foot breadth (f and g); foot length (h and i).
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
regression formulae are developed for various combinations to reach the best estimate possible. In case of stepwise regression analysis, entrance and removal of a variable are based on the statistical significance level of P < 0:05 and P < 0:10, respectively. Such a statistical approach has been commonly used in research of this nature [17,24–26]. Studies on bones have compensated for the loss of soft tissue when measured from cadavera. No adjustment as such has been needed for the present investigation. A test of intraobserver error was also made using randomly selected 10% of the samples for each sex. This test is useful because it assists if individual accuracy for a given dimension is consistent and is performed commonly among investigators [27].
3. Results Table 1 shows the descriptive statistics per sex. Females are about 5 years older than males and smaller in all dimensions. Overall males are more varied with higher standard deviations than females. Results of the primary stepwise regression analysis are given in Table 2. Measurements for the analysis consist of leg and foot dimensions and stature. Of the five variables entered into the analysis, all but foot breadth participate in the analysis with tibial length as the first and followed by thigh and foot lengths, and finally foot height in males. In females, the tibial length is selected first and followed by foot length and the rest was the same as those of males. R indicates the regression coefficient value. R2 explains the percentage that a dimension contributes to the variation in the dependent variable (that is, stature). In other words, nearly 55 and 63% of variation in stature are inherent by the tibial length for males and females, respectively. The addition of the others raise this figure to about 76% or so in both sexes. In order to consider dismembered body parts, Table 3 is developed to include those dimensions that correlate significantly with stature and can be used alone or in combination with others. Equations developed with the
3
Table 2 Stepwise regression showing the sequence of entrance of variables into the analysis, and standard error of estimate (S.E.E.), R, and R2 Variables
Males
Leg length Thigh length Foot length Foot height
Females
R
2
R
S.E.E.
R
R2
S.E.E.
0.74 0.82 0.86 0.86
0.55 0.68 0.74 0.75
44.62 37.79 34.26 33.80
0.79 0.86 0.83 0.87
0.63 0.73 0.69 0.76
39.34 33.75 36.33 32.42
direct method are trochanteric height, thigh length, lower leg length, tibial length and the combination of foot length and height. The highest R2 is associated with trochanteric height (58%) and lower leg length (56%) in males and lower leg length (65%) and leg length (63%) in females. Generated regression formulae from a stepwise and five direct analyses to estimate stature for both sexes from body parts are given in Table 4. Using the regression formulae of y (stature) ¼ a (constant) þ b1(regression coefficient for the first variable ) x1 (first variable) þ b2(regression coefficient for the second variable) x2 (second variable) þ bn (regression coefficient for the nth variable) xn (nth variable), one can calculate the height from the dismembered body part. The standard error of estimate refers the error that may arise from estimating stature. The use of an equation can be exemplified by considering a lower extremity of a male with a trochanteric height 989 mm. When this dimension is applied to the relevant formula as: stature ðmmÞ ¼ 989 ðtrochanteric heightÞ 1:10 þ 737:03 ðconstantÞ ¼ 1824:93 mm 58:88 ðS:E:E:Þ To test the possibility of intraobserver errors, a sample of 19 males and 12 females (10% of the total sample) using the SPSS random selection function is obtained. From the remaining cases, regression functions using trochanteric height and lower leg length are separately calculated. Stature
Table 1 Descriptive statistics of age (years) and various lower extremity dimensions (mm) in males (N ¼ 203) and females (N ¼ 108) Males Variables Age Stature Trochanteric height Thigh length Lower leg length Leg length Foot length Foot breadth Foot height
Females
Mean
S.D.
Minimum
Maximum
Mean
S.D.
Minimum
Maximum
30.7 1719.7 892.2 438.5 453.8 383.7 249.0 93.1 70.0
10.35 66.32 45.91 35.49 25.40 23.98 12.27 7.14 5.98
20 1546 774 332 384 323 218 63 58
59 1975 1004 518 560 492 288 111 92
35.3 1617.5 839.7 422.9 416.8 351.3 225.9 85.6 65.5
9.13 64.46 38.74 31.27 22.74 22.15 8.53 5.63 6.20
19 1462 738 320 344 267 206 70 57
71 1848 915 489 487 414 265 98 86
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
4
Table 3 Direct regression analysis including R, R2, and standard error of estimate Regression analysis
Males
Trochanteric height Thigh length Lower leg length Leg length Foot height and length
Females
R
2
R
S.E.E.
R
R2
S.E.E.
0.76 0.50 0.75 0.74 0.70
0.58 0.20 0.56 0.55 0.49
43.01 59.41 43.87 44.62 47.65
0.66 0.23 0.80 0.79 0.55
0.43 0.05 0.65 0.63 0.30
48.86 63.05 38.57 39.34 54.28
Table 4 Regression coefficients and standard error of estimate (mm) used to calculate stature (mm) Variables
Regression Equations Stepwise breadth
Males Lower leg length Trochanteric height Leg length Thigh length Foot length Foot height Constant S.E.E. Females Lower leg length Trochanteric height Leg length Thigh length Foot length Foot height Constant S.E.E.
Trochanter breadth
Thigh breadth
Lower leg breadth
Tibia breadth
Foot breadth
1.961 1.10 1.441 588 1.529 1.046 454.925
2.050
737.03
1352.124
829.750
933.030
3.700 3.700 757.567
56.09
58.88
51.81
55.22
50.34
72.03
0.84
2.277 1.092 1.969 492 2.035 1.605 152.789
700.293
103.08
102.49
1417.952 2.312 0.472
82.66
668.625
805.442
4.10 0.62 649.69
68.44
60.43
145.64
Table 5 Differences between the real and estimated stature (mm) using the regression formula based on a test and randomly selected samples and t-test of paired difference Males
Stature–stature from trochanteric height Stature–stature from lower leg length
Females
Differences between means
S.D.
t
Differences between means
S.D.
t
6.6 6.3
39.48 45.33
734 607
7.0 5.7
46.65 23.74
516 829
Degrees of freedom, 18 and 11 for males and females, respectively. Statistically not significant at P < 0:05 level.
of the individuals in the random sample for both sexes is calculated (Table 5). The t-test of difference between the paired main and random samples for each dimension did not show any statistically significant difference.
4. Discussion Estimation of stature is a major forensic anthropological concern used in the identification of unknown and com-
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
mingled human remains [6,28,29]. The procedure to estimate body height is to use its components. The accuracy is usually more reliable when the parts are situated along its length such as the lengths of the femur and tibia more so than the humerus and radius in the skeleton [8]. The present study takes this into the consideration and used similar measurements from the living people. There is an obvious need for a study as such. It could have been applied, for example, the identification of a victim who died in the Oklahoma City Federal Building bombing. As one of the authors (MYI) studied one of the legs of a victim to be identified, stature could have been estimated relatively simple instead of stripping the soft tissue to expose the bone. Anthropometry has enjoyed popularity over the years as a reliably technique to measure human body and its components [29]. Almost all human growth studies use it. The technique requires considerable experience so the measurements obtained are comparable with those taken by others. In osteological remains errors are relatively less especially in length dimensions because a long bone is relatively simpler to measure. Yet, there are numerous difficulties in the anthropometry of the living arising from the determining the correct landmark on the body. In the present study, male subjects were measured by male and female subjects by female investigators. In order to assess the impact of error arising from intraobserver variation over the same dimension (trochanteric height and lower leg length) a segment (10%) of the sample was randomly removed so that it can be used for intraobserver error. When two regression formulae were developed from these dimensions and test on the randomly selected sample the difference between the actual stature and the estimated stature were not found to be statistically significant. The population this study is carried differs from other European and North American populations upon which most of the studies are done. This difference may be particularly significant in the proportions of one body segment to another. Turks are shorter than Europeans and body proportions may also differ from others. One must consider differences between populations in order to apply functions as such to others. In conclusion, estimation of body height from its segments or dismember parts has important considerations for identifications of human remains recovered from mass disaster. The variables used in this analysis are from the lower extremity. It is shown that a single dimension can estimate the stature of an unknown person with a great accuracy and small standard error less than 7 cm. Dimensions from the lower extremity have greater association with the body height than those of the upper extremity.
Acknowledgements Authors are grateful to participants of this study. Without their cooperation this study could not have been carried out.
5
Authors thanks Dr. Sadi C¸ag˘ dir for drawing Fig. 1 and Can Medical Inc. for their effort to supply the anthropometric equipment used in the study. This investigation is supported by the Istanbul University Research Fund (Project No. 1135/05052000). References [1] J. Beddoe, On the stature of the older races of England, as estimated from the long bones, J. R. Anthropol. Inst. 17 (1887–1888) 202–207. [2] F. Rollet, De la mensuration de os longs du members. These pour le doctorat en medicine, Ph.D. Dissertation, vol. 43, Paris, 1888, pp. 1–128. [3] L. Manouvrier, De´ termination de la taille d’apre`s les grands os des members, Rev. l’Ecole d’Anthropol. 2 (1892) 227–233. [4] K. Pearson, Mathematical contributions to the theory of evolution. V. On the reconstruction of the stature of prehistoric races, Philos. Trans. R. Soc. Lond. 192 (1899) 169–244. [5] E. Breitinger, Zur Berechnung der Korperho¨ he aus den langen Gliedmassenknochen, Anthropol. Anz. 14 (1937) 249–274. [6] W.M. Krogman, M.Y. I˙s¸can, The Human Skeleton in Forensic Medicine, Charles C. Thomas, Springfield, IL, 1986. [7] C.W. Dupertuis, J. Hadden, On the reconstruction of stature from long bones, Am. J. Phys. Anthropol. 9 (1951) 15–54. [8] G. Fully, Une nouvelle methode de determination de la taille, Ann. Me´ d Le´ gale 36 (1956) 266–273. [9] D. Bhatnagar, S. Thapar, M. Batish, Identification of personal height from the somatometry of the hand in Punjabi males, Forensic Sci. Int. 24 (1984) 137–141. [10] S. Saxena, A study of correlations and estimation of stature from hand length, Anthropol. Anz. 42 (1984) 271–276. [11] N. Attallah, W. Marshall, The estimation of stature from anthropometric and photogrammetric measurements of the limbs, Med. Sci. Law 26 (1986) 53–59. [12] A.K. Abdel-Malek, A.M. Ahmed, S.A. el-Sharkawi, N.A. elHamid, Prediction of stature from hand measurements, Forensic Sci. Int. 46 (1990) 181–187. [13] K. Shintaku, Y. Furuya, Estimation of stature based on the proximal phalangeal length of Japanese women’s hands, Sangyo Ika Daigaku Zasshi 12 (1990) 215–219. [14] K. Kimura, Estimation of age at death from second metacarpals, Z. Morphol. Anthropol. 79 (1992) 169–181. ¨ zaslan, El ve 2. metakarp o¨ lc¸u¨ mlerinden vu¨ cut boy [15] A. O uzunlug˘ unun hesaplanması. Uzmanlık Tezi (specialty thesis), ¨ niversitesi Cerrahpas¸a Tıp Faku¨ ltesi, Adli Tıp I˙stanbul U Bilim Dalı, Istanbul, 1996. [16] T. Han, M. Lean, Lower leg length as an index of stature in adults, Int. J. Obes. Relat. Metab. Disord. 20 (1996) 21–27. [17] E. Giles, P.H. Vallandigham, Height estimation from foot and shoeprint length, J. Forensic Sci. 36 (1991) 1134–1151. [18] C.C. Gordon, J.E. Buikstra, Linear models for the prediction of stature from foot and boot dimensions, J. Forensic Sci. 37 (1992) 771–782. [19] T. Singh, M. Phookan, Stature and footsize in four Thai communities of Assam, India, Anthropol. Anz. 51 (1993) 349–355.
6
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
[20] K. Ashizawa, C. Kumakura, A. Kusumoto, S. Narasaki, Relative foot size and shape to general body size in Javanese, Ann. Hum. Biol. 24 (1997) 117–129. [21] G. Olivier, Practical Anthropology, C.C. Thomas, Springfield, IL, 1969. [22] J.S. Weiner, J.A. Lourie, Human Biology: A Guide to Field Methods, Blackwell Scientific Publications, Oxford, 1969. [23] S.J. Ulijaszek, C.G.N. Mascie-Taylor (Eds.), Anthropometry: The Individual and the Population, Cambridge University Press, Cambridge, 1994. [24] M. Trotter, Estimation of stature from intact limb bones, in: T.D. Stewart (Ed.), Personal Identification in Mass Disasters, National Museum of Natural History, Washington, DC, 1970, pp. 71–84.
[25] T.D. Holland, Estimation of adult stature from fragmentary tibias, J. Forensic Sci. 37 (1992) 1223–1229. [26] T. Sjøvold, Stature estimation from the skeleton, in: J. Siegel, P. Saukko, G. Knupfer (Eds.), Encyclopedia of Forensic Sciences, Academic Press, London, 2000, pp. 276–284. [27] C.C. Gordon, B. Bradtmiller, Interobserver error in a large scale anthropometric survey, Am. J. Hum. Biol. 4 (1992) 253–263. [28] M.Y. I˚s¸can, Rise of forensic anthropology, Yrbk. Phys. Anthropol. 31 (1988) 203–230. [29] M.Y. I˚s¸can, Anthropometry, in: J. Siegel, P. Saukko, G. Knupfer (Eds.), Encyclopedia of Forensic Sciences, Academic Press, London, 2000, pp. 284–290.
Estimation of stature from body parts ¨ zaslana,*, M. Yas¸ar I˙s¸canb, ˙Inci O ¨ zaslanc, Harun Tug˘cud, Sermet Koc¸a Abdi O a˙ ¨
IU Cerrahpasa Tıp Faku¨ltesi, Adli Tıp Anabilim Dalı, I˙stanbul 34303, Turkey b˙ ¨ IU Adli Tıp Enstitu¨su¨, ˙Istanbul, Turkey c Adli Tıp Kurumu, I˙stanbul, Turkey d GATA, Adli Tıp Anabilim Dalı, Ankara, Turkey Received 11 January 2002; accepted 15 November 2002
Abstract Anthropometric technique commonly used by anthropologists and adopted by medical scientists has been employed to estimate body size for over a hundred years. With the increasing frequency of mass disasters, the identification of an isolated lower extremity and the stature of the person it belonged to has created problems for the investigation of the identity of some of the victims. In spite of a need for such a study, there is a lack of systematic studies to identify fragmented and dismembered human remains. The purpose of the paper is to analyze anthropometric relationships between dimensions of the lower extremity and body height. Analysis is based on a sample of middle class male (N ¼ 203) and female (N ¼ 108) adult Turks residing in Istanbul. The participants are mostly students and staff members of a medical school, and military personnel. Measurements taken are stature, trochanteric height, thigh length, lower leg length, leg length, and foot height, breadth, and length. Of the five variables entered into the regression analysis, all but foot breadth participate in the analysis with leg length as the first and followed by thigh and foot lengths, and finally foot height in males (R2). There were also individually calculated formulae for some of these measurements which provided smaller R2-values. Student’s t-test to assess if there was any intraobserver error in measurements take by individual anthropometrist did not show such any statistically significant difference. In conclusion, the study suggested that estimation of a living height can be made possible using various dimensions of the lower extremity. One must consider differences between populations in order to apply functions as such to others. # 2002 Published by Elsevier Science Ireland Ltd. Keywords: Stature estimation; Dismember lower extremity; Anthropometry; Regression analysis; Mass disaster; Turks
1. Introduction Anthropometric technique commonly used by anthropologists and adopted by medical scientists has been employed to estimate body size for over a hundred years [1–6]. In the last half of the 20th century, studies have been more vigorous when skeletal collections have been assembled in many parts of the world [7,8]. There has been an increase in the frequency of various mass disasters (air and train crash, bombing, mass suicide, flooding, powerful storm). To the list one should add the more recent tragedies such as the
* Corresponding author. Tel.: þ90-212-588-4800; fax: þ90-212-588-0011. ¨ zaslan). E-mail address: [email protected] (A. O
Marmara earthquake, Turkey and the attack on the World Trade Center and Pentagon in the US where thousands of people were killed. There have been already some attempts using hand [9–15], lower leg length and arm span [16] and foot dimensions [17–20] to regress on body height. The identification of an isolated lower extremity and the stature of the person it belonged to has created problems for the investigation of identity of some of the victims found in the rubbles of the bombed Oklahoma Federal building. In spite of a need as such, there is a lack of systematic studies to identify fragmented and dismembered human remains. The purpose of the paper is to analyze anthropometric relationships between various dimensions of the lower extremity and body height. Analysis is based on a sample of middle class male and female adult Turks using various statistical subroutines particularly that of regression analysis.
0379-0738/02/$ – see front matter # 2002 Published by Elsevier Science Ireland Ltd. PII: S 0 3 7 9 - 0 7 3 8 ( 0 2 ) 0 0 4 2 5 - 5
2
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
2. Materials and methods The study is composed of 311 (203 males and 108 females) adult Turks. The participants are mostly from the Cerrahpasa Medical Faculty staff, students, staff from the Council of Forensic Medicine and military personnel who lived mostly in Istanbul. These individuals are considered middle class with at least 5 years of elementary education. Measurements are all taken from the left side according to the procedure described by the International Biological Program [21–23]. The subjects are asked to remove their clothing except underwear and shoes. They are oriented in the standard anatomic position with the head on the Frankfort Horizontal Plane. Males were measured by ¨ and HT and females by I˙O ¨ . Anthropometrists are all AO initially trained in the technique by MYI. For the convenience of the readers the dimensions are taken with an anthropometer in millimeters and 100 g units using standard anthropometric instruments such as sliding and spreading calipers, anthropometer, steel tape, and scale. Extremity measurements are all taken from the left side (Fig. 1) [22].
Stature is taken from the vertex to the floor obeying the anatomic position and Frankfort Plane. Trochanteric height is measured from the lateral bulging of the greater trochanteric protrusion to the floor. This measurement is preferred over the anterior iliac spine because of commonly S.E.E.n break and dismemberment in the hip joint. Thigh length is calculated from as distance between the trochanteric height and lower leg length. Lower leg length is the distance from the tibiale to the floor. Leg length is from the tibiale to the medial malleolus of the tibia. Foot height is calculated as the distance between lower leg length and leg length. Foot breadth is the distance between the lateral and medial sides at the metatarsal region using a sliding caliper. Foot length is the maximum distance between the most anterior and posterior projecting part of the foot with a sliding caliper. The data are analyzed using various subroutines of Statistical Package for the Social Sciences (SPSS) and
Fig. 1. Anthropometric dimensions used in the study: stature (a–e); trochanteric height (b–e); thigh length (b and c); lower leg length (c–e); leg length (c and d); foot height (d and e); foot breadth (f and g); foot length (h and i).
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
regression formulae are developed for various combinations to reach the best estimate possible. In case of stepwise regression analysis, entrance and removal of a variable are based on the statistical significance level of P < 0:05 and P < 0:10, respectively. Such a statistical approach has been commonly used in research of this nature [17,24–26]. Studies on bones have compensated for the loss of soft tissue when measured from cadavera. No adjustment as such has been needed for the present investigation. A test of intraobserver error was also made using randomly selected 10% of the samples for each sex. This test is useful because it assists if individual accuracy for a given dimension is consistent and is performed commonly among investigators [27].
3. Results Table 1 shows the descriptive statistics per sex. Females are about 5 years older than males and smaller in all dimensions. Overall males are more varied with higher standard deviations than females. Results of the primary stepwise regression analysis are given in Table 2. Measurements for the analysis consist of leg and foot dimensions and stature. Of the five variables entered into the analysis, all but foot breadth participate in the analysis with tibial length as the first and followed by thigh and foot lengths, and finally foot height in males. In females, the tibial length is selected first and followed by foot length and the rest was the same as those of males. R indicates the regression coefficient value. R2 explains the percentage that a dimension contributes to the variation in the dependent variable (that is, stature). In other words, nearly 55 and 63% of variation in stature are inherent by the tibial length for males and females, respectively. The addition of the others raise this figure to about 76% or so in both sexes. In order to consider dismembered body parts, Table 3 is developed to include those dimensions that correlate significantly with stature and can be used alone or in combination with others. Equations developed with the
3
Table 2 Stepwise regression showing the sequence of entrance of variables into the analysis, and standard error of estimate (S.E.E.), R, and R2 Variables
Males
Leg length Thigh length Foot length Foot height
Females
R
2
R
S.E.E.
R
R2
S.E.E.
0.74 0.82 0.86 0.86
0.55 0.68 0.74 0.75
44.62 37.79 34.26 33.80
0.79 0.86 0.83 0.87
0.63 0.73 0.69 0.76
39.34 33.75 36.33 32.42
direct method are trochanteric height, thigh length, lower leg length, tibial length and the combination of foot length and height. The highest R2 is associated with trochanteric height (58%) and lower leg length (56%) in males and lower leg length (65%) and leg length (63%) in females. Generated regression formulae from a stepwise and five direct analyses to estimate stature for both sexes from body parts are given in Table 4. Using the regression formulae of y (stature) ¼ a (constant) þ b1(regression coefficient for the first variable ) x1 (first variable) þ b2(regression coefficient for the second variable) x2 (second variable) þ bn (regression coefficient for the nth variable) xn (nth variable), one can calculate the height from the dismembered body part. The standard error of estimate refers the error that may arise from estimating stature. The use of an equation can be exemplified by considering a lower extremity of a male with a trochanteric height 989 mm. When this dimension is applied to the relevant formula as: stature ðmmÞ ¼ 989 ðtrochanteric heightÞ 1:10 þ 737:03 ðconstantÞ ¼ 1824:93 mm 58:88 ðS:E:E:Þ To test the possibility of intraobserver errors, a sample of 19 males and 12 females (10% of the total sample) using the SPSS random selection function is obtained. From the remaining cases, regression functions using trochanteric height and lower leg length are separately calculated. Stature
Table 1 Descriptive statistics of age (years) and various lower extremity dimensions (mm) in males (N ¼ 203) and females (N ¼ 108) Males Variables Age Stature Trochanteric height Thigh length Lower leg length Leg length Foot length Foot breadth Foot height
Females
Mean
S.D.
Minimum
Maximum
Mean
S.D.
Minimum
Maximum
30.7 1719.7 892.2 438.5 453.8 383.7 249.0 93.1 70.0
10.35 66.32 45.91 35.49 25.40 23.98 12.27 7.14 5.98
20 1546 774 332 384 323 218 63 58
59 1975 1004 518 560 492 288 111 92
35.3 1617.5 839.7 422.9 416.8 351.3 225.9 85.6 65.5
9.13 64.46 38.74 31.27 22.74 22.15 8.53 5.63 6.20
19 1462 738 320 344 267 206 70 57
71 1848 915 489 487 414 265 98 86
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
4
Table 3 Direct regression analysis including R, R2, and standard error of estimate Regression analysis
Males
Trochanteric height Thigh length Lower leg length Leg length Foot height and length
Females
R
2
R
S.E.E.
R
R2
S.E.E.
0.76 0.50 0.75 0.74 0.70
0.58 0.20 0.56 0.55 0.49
43.01 59.41 43.87 44.62 47.65
0.66 0.23 0.80 0.79 0.55
0.43 0.05 0.65 0.63 0.30
48.86 63.05 38.57 39.34 54.28
Table 4 Regression coefficients and standard error of estimate (mm) used to calculate stature (mm) Variables
Regression Equations Stepwise breadth
Males Lower leg length Trochanteric height Leg length Thigh length Foot length Foot height Constant S.E.E. Females Lower leg length Trochanteric height Leg length Thigh length Foot length Foot height Constant S.E.E.
Trochanter breadth
Thigh breadth
Lower leg breadth
Tibia breadth
Foot breadth
1.961 1.10 1.441 588 1.529 1.046 454.925
2.050
737.03
1352.124
829.750
933.030
3.700 3.700 757.567
56.09
58.88
51.81
55.22
50.34
72.03
0.84
2.277 1.092 1.969 492 2.035 1.605 152.789
700.293
103.08
102.49
1417.952 2.312 0.472
82.66
668.625
805.442
4.10 0.62 649.69
68.44
60.43
145.64
Table 5 Differences between the real and estimated stature (mm) using the regression formula based on a test and randomly selected samples and t-test of paired difference Males
Stature–stature from trochanteric height Stature–stature from lower leg length
Females
Differences between means
S.D.
t
Differences between means
S.D.
t
6.6 6.3
39.48 45.33
734 607
7.0 5.7
46.65 23.74
516 829
Degrees of freedom, 18 and 11 for males and females, respectively. Statistically not significant at P < 0:05 level.
of the individuals in the random sample for both sexes is calculated (Table 5). The t-test of difference between the paired main and random samples for each dimension did not show any statistically significant difference.
4. Discussion Estimation of stature is a major forensic anthropological concern used in the identification of unknown and com-
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
mingled human remains [6,28,29]. The procedure to estimate body height is to use its components. The accuracy is usually more reliable when the parts are situated along its length such as the lengths of the femur and tibia more so than the humerus and radius in the skeleton [8]. The present study takes this into the consideration and used similar measurements from the living people. There is an obvious need for a study as such. It could have been applied, for example, the identification of a victim who died in the Oklahoma City Federal Building bombing. As one of the authors (MYI) studied one of the legs of a victim to be identified, stature could have been estimated relatively simple instead of stripping the soft tissue to expose the bone. Anthropometry has enjoyed popularity over the years as a reliably technique to measure human body and its components [29]. Almost all human growth studies use it. The technique requires considerable experience so the measurements obtained are comparable with those taken by others. In osteological remains errors are relatively less especially in length dimensions because a long bone is relatively simpler to measure. Yet, there are numerous difficulties in the anthropometry of the living arising from the determining the correct landmark on the body. In the present study, male subjects were measured by male and female subjects by female investigators. In order to assess the impact of error arising from intraobserver variation over the same dimension (trochanteric height and lower leg length) a segment (10%) of the sample was randomly removed so that it can be used for intraobserver error. When two regression formulae were developed from these dimensions and test on the randomly selected sample the difference between the actual stature and the estimated stature were not found to be statistically significant. The population this study is carried differs from other European and North American populations upon which most of the studies are done. This difference may be particularly significant in the proportions of one body segment to another. Turks are shorter than Europeans and body proportions may also differ from others. One must consider differences between populations in order to apply functions as such to others. In conclusion, estimation of body height from its segments or dismember parts has important considerations for identifications of human remains recovered from mass disaster. The variables used in this analysis are from the lower extremity. It is shown that a single dimension can estimate the stature of an unknown person with a great accuracy and small standard error less than 7 cm. Dimensions from the lower extremity have greater association with the body height than those of the upper extremity.
Acknowledgements Authors are grateful to participants of this study. Without their cooperation this study could not have been carried out.
5
Authors thanks Dr. Sadi C¸ag˘ dir for drawing Fig. 1 and Can Medical Inc. for their effort to supply the anthropometric equipment used in the study. This investigation is supported by the Istanbul University Research Fund (Project No. 1135/05052000). References [1] J. Beddoe, On the stature of the older races of England, as estimated from the long bones, J. R. Anthropol. Inst. 17 (1887–1888) 202–207. [2] F. Rollet, De la mensuration de os longs du members. These pour le doctorat en medicine, Ph.D. Dissertation, vol. 43, Paris, 1888, pp. 1–128. [3] L. Manouvrier, De´ termination de la taille d’apre`s les grands os des members, Rev. l’Ecole d’Anthropol. 2 (1892) 227–233. [4] K. Pearson, Mathematical contributions to the theory of evolution. V. On the reconstruction of the stature of prehistoric races, Philos. Trans. R. Soc. Lond. 192 (1899) 169–244. [5] E. Breitinger, Zur Berechnung der Korperho¨ he aus den langen Gliedmassenknochen, Anthropol. Anz. 14 (1937) 249–274. [6] W.M. Krogman, M.Y. I˙s¸can, The Human Skeleton in Forensic Medicine, Charles C. Thomas, Springfield, IL, 1986. [7] C.W. Dupertuis, J. Hadden, On the reconstruction of stature from long bones, Am. J. Phys. Anthropol. 9 (1951) 15–54. [8] G. Fully, Une nouvelle methode de determination de la taille, Ann. Me´ d Le´ gale 36 (1956) 266–273. [9] D. Bhatnagar, S. Thapar, M. Batish, Identification of personal height from the somatometry of the hand in Punjabi males, Forensic Sci. Int. 24 (1984) 137–141. [10] S. Saxena, A study of correlations and estimation of stature from hand length, Anthropol. Anz. 42 (1984) 271–276. [11] N. Attallah, W. Marshall, The estimation of stature from anthropometric and photogrammetric measurements of the limbs, Med. Sci. Law 26 (1986) 53–59. [12] A.K. Abdel-Malek, A.M. Ahmed, S.A. el-Sharkawi, N.A. elHamid, Prediction of stature from hand measurements, Forensic Sci. Int. 46 (1990) 181–187. [13] K. Shintaku, Y. Furuya, Estimation of stature based on the proximal phalangeal length of Japanese women’s hands, Sangyo Ika Daigaku Zasshi 12 (1990) 215–219. [14] K. Kimura, Estimation of age at death from second metacarpals, Z. Morphol. Anthropol. 79 (1992) 169–181. ¨ zaslan, El ve 2. metakarp o¨ lc¸u¨ mlerinden vu¨ cut boy [15] A. O uzunlug˘ unun hesaplanması. Uzmanlık Tezi (specialty thesis), ¨ niversitesi Cerrahpas¸a Tıp Faku¨ ltesi, Adli Tıp I˙stanbul U Bilim Dalı, Istanbul, 1996. [16] T. Han, M. Lean, Lower leg length as an index of stature in adults, Int. J. Obes. Relat. Metab. Disord. 20 (1996) 21–27. [17] E. Giles, P.H. Vallandigham, Height estimation from foot and shoeprint length, J. Forensic Sci. 36 (1991) 1134–1151. [18] C.C. Gordon, J.E. Buikstra, Linear models for the prediction of stature from foot and boot dimensions, J. Forensic Sci. 37 (1992) 771–782. [19] T. Singh, M. Phookan, Stature and footsize in four Thai communities of Assam, India, Anthropol. Anz. 51 (1993) 349–355.
6
¨ zaslan et al. / Forensic Science International 3501 (2003) 1–6 A. O
[20] K. Ashizawa, C. Kumakura, A. Kusumoto, S. Narasaki, Relative foot size and shape to general body size in Javanese, Ann. Hum. Biol. 24 (1997) 117–129. [21] G. Olivier, Practical Anthropology, C.C. Thomas, Springfield, IL, 1969. [22] J.S. Weiner, J.A. Lourie, Human Biology: A Guide to Field Methods, Blackwell Scientific Publications, Oxford, 1969. [23] S.J. Ulijaszek, C.G.N. Mascie-Taylor (Eds.), Anthropometry: The Individual and the Population, Cambridge University Press, Cambridge, 1994. [24] M. Trotter, Estimation of stature from intact limb bones, in: T.D. Stewart (Ed.), Personal Identification in Mass Disasters, National Museum of Natural History, Washington, DC, 1970, pp. 71–84.
[25] T.D. Holland, Estimation of adult stature from fragmentary tibias, J. Forensic Sci. 37 (1992) 1223–1229. [26] T. Sjøvold, Stature estimation from the skeleton, in: J. Siegel, P. Saukko, G. Knupfer (Eds.), Encyclopedia of Forensic Sciences, Academic Press, London, 2000, pp. 276–284. [27] C.C. Gordon, B. Bradtmiller, Interobserver error in a large scale anthropometric survey, Am. J. Hum. Biol. 4 (1992) 253–263. [28] M.Y. I˚s¸can, Rise of forensic anthropology, Yrbk. Phys. Anthropol. 31 (1988) 203–230. [29] M.Y. I˚s¸can, Anthropometry, in: J. Siegel, P. Saukko, G. Knupfer (Eds.), Encyclopedia of Forensic Sciences, Academic Press, London, 2000, pp. 284–290.