Estimation of stature from facial measurements in northwest Indians

Legal Medicine 12 (2010) 23–27

Contents lists available at ScienceDirect

Legal Medicine journal homepage: www.elsevier.com/locate/legalmed

Estimation of stature from facial measurements in northwest Indians Daisy Sahni a,*, Sanjeev b, Parul Sharma a, Harjeet a, Gagandeep Kaur a, Anjali Aggarwal a a b

Department of Anatomy, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh 160012, India Central Forensic Science Laboratory, Sector-36, Chandigarh 160036, India

a r t i c l e

i n f o

Article history: Received 13 July 2009 Received in revised form 8 October 2009 Accepted 9 October 2009 Available online 11 December 2009 Keywords: Estimation Stature Facial measurements Regression equations Northwest India

a b s t r a c t Estimation of stature is one of the important component in identification of human remains in forensic anthropology. The present investigation attempts to estimate stature from seven facial measurements of 300 (173 males and 127 females) healthy subjects between the ages of 18–70 years from Northwest India. Height of all the subjects was measured and facial measurements were taken. Data was subjected to statistical analysis like mean, standard deviation, multiplication factors, Karl Pearson’s correlation coefficient (r), linear and multiple regression analyses using statistical package for social sciences (SPSS). The average height of the subjects was in the range of 154.3–178.3 cm in males and 155.1–168.4 cm in females. Estimated stature calculated by regression analysis of seven facial measurements was almost similar to mean actual stature in both males and females and the difference by using multiplication factors was found to be greater. Standard error of estimation (SEE) computed both by linear and multiple regression analyses was found to be low for the two sexes. Thus we can conclude that regression equations generated from facial measurements can be a supplementary approach for the estimation of stature when extremities are not available. Ó 2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Forensic anthropology deals with the identification of unrecognizable human remains usually in skeletal form by determination of age, sex, race and stature. Stature or body height is one of the primary and useful tool used in personal identification. Estimation of stature from various body parts like extremities is well documented in other countries [1–3] as well as in India [4–7]. However, it becomes difficult when only a bare skull is available for identification purposes and one has to estimate the stature of the deceased. Search of the available literature shows that some authors have given mathematical formulae to determine stature from cranial diameters [8] while others have formulated regression equations from somatometry of the skulls [9,10]. More recently, applicability of regression equations generated from the cephalofacial measurements for stature estimation has been greatly emphasized [11–15]. These equations are both population and sex specific, hence, the present study has been undertaken to investigate the usefulness of facial measurements in estimation of adult stature and to compare

* Corresponding author. Address: Department of Anatomy, Postgraduate Institute of Medical Education and Research (PGIMER), Research Block-B, Chandigarh 160012, India. Tel.: +91 172 2755201 (O); fax: +91 172 2744401/2745078. E-mail addresses: [email protected] (D. Sahni), sanjeevcfsl@hotmail. com ( Sanjeev), [email protected] (P. Sharma), [email protected] ( Harjeet), [email protected] (G. Kaur), [email protected] (A. Aggarwal). 1344-6223/$ - see front matter Ó 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.legalmed.2009.10.002

the reliability of stature estimation by multiplication factor, and regression analysis.

2. Materials and methods Data for the present study consisted of 300 adults (173 males, 127 females) belonging to Chandigarh zone of Northwest India (NWI) in the age group of 18–70 years (mean age 36.30 years). Seven facial measurements along with the stature of the subjects were taken according to standard anthropometric procedures [16]. The measurements taken are defined as follows: 1. Stature/height vertex: It measures the vertical distance from vertex (v) to floor. 2. Total face height (n–gn): It measures the straight distance between nasion and gnathion. 3. Upper face height (n–pr): It measures the straight distance between nasion and prosthion. 4. Height of lower face (sto–gn): It measures the straight projective distance between the chin and the opening of the mouth i.e. between stomion and gnathion. 5. Minimum frontal breadth (ft–ft): It measures the straight distance between the two frontotemporalia. 6. Bigonial breadth (go–go): It measures the straight distance between the two gonia. 7. Biocular breadth(ec–ec): It measures the straight distance between external canthi (ectocanthia) i.e. outer corners of the eyes.

24

D. Sahni et al. / Legal Medicine 12 (2010) 23–27

8. Interocular breadth (en–en): It measures the straight distance between the internal canthus of the eye i.e. endocanthion to endocanthion, with the eyelids open.

Table 3 Correlation coefficient between stature and various facial measurements in males and females. Measurements (cm)

Collected data was subjected to statistical analysis like mean, standard deviation, multiplication factors, Karl Pearson’s correlation coefficient (r), linear and multiple regression analysis using statistical package for social sciences (SPSS).

Female (N = 127)

r-Value

r-Value

Total facial height (n–gn) Upper facial height (n–pr) Height of lower face (sto–gn) Minimum frontal breadth (ft–ft) Bigonial breadth (go–go) Biocular breadth (ec–ec) Interocular breadth (en–en)

3. Results The subjects were classified into six height categories (Table 1) according to Martin’s stature classification [16]. It is seen from this table that the maximum number of males (47.1%) falls in the category of medium (164.0–166.9 cm) and minimum (4.1%) in the category of short stature (150.0–159.9 cm). Similarly, in females, the maximum frequency (81.7%) was observed in the category of tall (159.0–167.9 cm) and minimum (0.8%) in the medium category (153.0–155.9 cm). No male was found in the category of very tall and no female was found in the category of short or lower medium stature. Means and standard deviations (SD) of seven facial measurements along with stature were found to be greater in males than the females (p < 0.001; p < 0.01) except for interocular breadth (Table 2). The correlation coefficients were found to be low and thus the correlations of facial measurements with stature were very poor (Table 3). However, significant positive correlation of stature with upper facial height (p < 0.001) and with total facial height (p < 0.01) was observed in both males and females and with height of lower face (p < 0.05) only in females. Linear regression equations were derived for each facial measurement in the two sexes (Table 4). The hypothetical regression equation is represented as: stature (S) = a + bX, where ‘a’ is the regression coefficient of the dependant variable i.e. stature, ‘b’ is the regression coefficient of the independent variable i.e. any facial measurement and ‘X’ is the mean of that particular measurement. Standard error of estimation (SEE) for all variables was low ranging 3.56–3.70 for males and 2.90–2.95 for females.

Male (N = 173)

* ** ***

0.219** 0.270*** 0.028 0.088 0.064 0.071 0.030

p-Value 0.002 0.000 0.358 0.124 0.201 0.178 0.347

0.181* 0.167* 0.195* 0.060 0.047 0.121 0.061

p-Value 0.021 0.030 0.014 0.253 0.299 0.088 0.246

p < 0.05. p < 0.01. p < 0.001.

Means and SD of the multiplication factors of facial measurements with regard to stature are given in Table 5. In both sexes, the multiplication factor was found to be higher for all the facial measurements. Table 6 gives a comparison of mean actual stature (MAS) and mean estimated stature (MES) calculated by using regression analysis and multiplication factors. Mean values of various facial measurements were substituted in their respective regression equations and multiplication factors to calculate MES. Using regression analysis MES was equal or slightly less (0.01 cm) than MAS in both males and females except in case of upper facial height in females, where MES was more than MAS with the difference of 4.93 cm. On the other hand using multiplication factors the MES was found to be greater than MAS in both the sexes and the difference varied from 0.57 to 8.93 cm in males and from 0.35 to 1.91 cm in females. Thus indicating that in the estimation of stature, linear regression equations are more reliable than multiplication factors. Tables 7 and 8 represent multiple regression coefficients, constant, SEE and correlation coefficients for all the seven combinations (i.e. when seven variables were considered and so on). SEE ranged from 3.569 to 3.610 in males and 2.880 to 2.914 in females. In the males, significant correlation coefficients were found for all

Table 1 Frequency of distribution of stature in males and females according to Martin’s classification [16]. Stature classification (cm)

Short Lower medium Medium Upper medium Tall Very tall

Male (N = 173)

Female (N = 127)

Range

N

%

Range

N

%

150.0–159.9 160.0–163.9 164.0–166.9 167.0–169.9 170.0–179.9 180.0–199.9

7 21 81 39 25 0

4.1 12.2 47.1 22.1 14.5 0

140.0–148.9 149.0–152.9 153.0–155.9 156.0–158.9 159.0–167.9 168.0–186.9

0 0 1 15 104 7

0 0 0.8 11.9 81.7 5.6

Table 2 Means and standard deviations of facial measurements along with stature in the two sexes. Measurements (cm)

Stature (height vertex) Total facial height (n–gn) Upper facial height (n–pr) Height of lower face (sto–gn) Minimum frontal breadth (ft–ft) Bigonial breadth (go–go) Biocular breadth (ec–ec) Interocular breadth (en–en) ** ***

p < 0.01. p < 0.001.

Male (N = 173)

Female (N = 127)

t-Test

Range

Mean ± SD

Range

Mean ± SD

154.3–178.3 9.70–13.0 5.5–7.8 2.5–4.5 8.8–12.2 9.0–12.4 5.0–10.9 1.7–4.7

165.90 ± 3.69 11.25 ± 0.67 6.85 ± 0.48 3.67 ± 0.37 10.59 ± 0.62 10.64 ± 0.63 9.86 ± 0.60 2.27 ± 0.33

155.1–168.4 9.5–12.4 5.5–7.4 2.9–4.3 8.0–11.5 8.2–11.3 8.5–10.9 1.7–2.7

163.24 ± 2.94 10.80 ± 0.54 6.53 ± 0.42 3.53 ± 0.30 9.90 ± 0.59 10.26 ± 0.68 9.68 ± 0.47 2.21 ± 0.23

6.704*** 6.233*** 5.799*** 3.506** 9.678*** 4.974*** 2.802** 1.845

25

D. Sahni et al. / Legal Medicine 12 (2010) 23–27 Table 4 Linear regression equation for estimation of stature (cm) from facial measurements in males and females. Stature

S= S= S= S= S= S=

Linear regression equation (S) = a + bX

SEE

Variables

Male

Female

Male

Female

152.281 + 1.210 X1 151.789 + 2.061 X2 164.879 + 0.277 X3 160.345 + 0.524 X4 161.883 + 0.377 X5 161.632 + 0.432 X6 166.655 + ( 0.335) X7

152.510 + 0.993 X1 155.657 + 1.160 X2 156.481 + 1.916 X3 160.310 + 0.296 X4 161.141 + 0.204 X5 155.854 + 0.763 X6 164.955 + ( 0.778) X7

3.61 3.56 3.70 3.69 3.69 3.69 3.70

2.90 2.91 2.90 2.95 2.95 2.93 2.95

X1 = Total facial height X2 = Upper facial height X3 = Height of lower face X4 = Minimum frontal breadth X5 = Bigonial breadth X6 = Biocular breadth X7 = Interocular breadth

SEE, standard error of estimation.

Table 5 Mean and SD of the multiplication factors (MF) for facial measurements with regard to the stature in both sexes. Measurements

Total facial height (n–gn) Upper facial height (n–pr) Height of lower face (sto–gn) Minimum frontal breadth (ft–ft) Bigonial breadth (go–go) Biocular breadth (ec–ec) Interocular breadth (en–en)

Mean ± SD of MF Male (N = 173)

Female (N = 127)

15.54 ± 9.79 24.35 ± 1.71 45.71 ± 4.77 15.72 ± 0.97 15.65 ± 0.98 16.91 ± 1.50 74.34 ± 9.86

15.15 ± 0.76 25.08 ± 1.64 46.61 ± 3.82 16.55 ± 1.06 15.99 ± 1.16 16.90 ± 0.85 74.73 ± 8.33

the seven combinations while in the females, the correlation was found to be significant for the combination of total facial height, upper facial height and height of lower face and also when only total facial height was considered. 4. Discussion It becomes very difficult for a forensic scientist/anthropologist when isolated remains of head, face, or skull are brought for forensic examination, as the standards available in this direction are scanty. Therefore, facial measurements act as a useful in the absence of the other evidences for stature estimation. This study seems to be more useful in the case of fragmentary facial remains. With the help of the exact knowledge of how the soft tissue landmarks of face and their thicknesses relate to the craniometric landmarks of the skull [17], we can use this method when only skull is available for identification purposes. This requires a firm grasp of soft-hard tissue correlations and it is possible to accurately position the details of the eyes and the lips within and around their bony structures and such positioning alone may be sufficient to create the desired approximation for reconstruction purposes [18]. Regression analysis is considered to be the most easiest and reliable method for stature estimation [19]. Holland [20] derived

linear regression equations for estimation of stature from calcaneus and talus with relatively accurate results. Hauser et al. [21] estimated the stature from femur with considerable accuracy. Patil and Mody [22] found a high degree of reliability for estimating stature of Central Indian population from lateral cephalograms for the maximum length of the skull by regression analysis. Ryan and Bidmos [10] generated regression formulae for stature estimation for indigenous South Africans based on measurements of long bones of upper and lower extremities and the calcaneus. Using regression analysis Krishan [14,15] estimated stature from cephalo-facial measurements and concluded that the calculated regression formulae show good reliability and applicability not only for the population upon which these formulae are based but also for a sample of mixed population from North India. On comparing the results of estimation of stature using multiplication factors with the study on males of Manipur, India, Kumar and Chandra [13] found that the differences in the MES and MAS are higher in male Kabui Nagas (ranged from 2.637 to 14.70 cm) as compared to our study (ranged from 0.57 to 8.93 cm). Whereas, linear regression equations showed less variability i.e. 3.268– 5.23 cm as compared to multiplication factors. In the current study the difference between the MAS and MES was 0.01 cm, which is negligible indicating more reliability of regression equations than multiplication factors. In Table 9 the results of the males (N = 173) of the present study were compared with the similar studies available on different populations of North India. Krishan and Kumar [14] studied adolescent (12–18 year old) Koli males (N = 252) from District Sirmour of Himachal Pradesh, North India while Krishan [15] studied adult (18–30 year old) male Gujjars (N = 996) from 16 villages in Siwaliks and its adjoining plains near Chandigarh, North India. On comparing the mean stature of our population with these two populations, the mean stature of males was higher (165.90 cm) than the adolescent Koli males (152.647 cm) but lower than adult Gujjars (172.31 cm) while, the SD of the stature was found to be lower (3.69 cm) than both the populations (7.103 cm for Koli males and 6.83 cm for Gujjars). In case of the facial measurements

Table 6 Comparison of mean actual stature (MAS) with mean estimated stature (MES) calculated by using linear regression (LR) and multiplication factors (MF). MAS (cm)

Male 165.90

Female 163.24

MES calculated using LR Measurements Total facial height (n–gn) Upper facial height (n–pr) Height of lower face (sto–gn) Minimum frontal breadth (ft–ft) Bigonial breadth (go–go) Biocular breadth (ec–ec) Interocular breadth (en–en)

165.90 165.91 165.90 165.89 165.89 165.89 165.89

Male 165.90

Female 163.24

MES calculated using MF 163.23 168.17 163.24 163.24 163.23 163.24 163.24

174.83 166.80 167.76 166.73 168.75 166.47 166.52

163.62 163.77 164.53 163.59 165.15 163.85 164.06

26

D. Sahni et al. / Legal Medicine 12 (2010) 23–27

Table 7 Regression coefficients of combination of facial measurements using multiple regression analysis in males. Variables

Number of variables taken 7

1 2 3 4 5 6 7 Constant SEE r-Value * **

Total facial height (n–gn) Upper facial height (n–pr) Height of lower face (sto–gn) Minimum frontal breadth (ft–ft) Bigonial breadth (go–go) Biocular breadth (ec–ec) Interocular breadth (en–en)

6

0.931 1.391 1.489 0.173 0.580 0.0389 0.390 143.859 3.588 0.304*

5 0.926 1.386 1.499 0.144 0.579 0.105

142.756 3.579 0.302*

4

3

0.913 1.418 1.490 0.154 0.594

143.423 3.569 0.302*

2

1

0.844 1.357 1.213 0.227

0.788 1.476 1.203

0.072 1.982

1.210

149.154 3.577 0.286**

151.339 3.569 0.283**

151.522 3.572 0.270**

152.281 3.610 0.219*

p < 0.05. p < 0.01.

Table 8 Regression coefficients of combination of facial measurements using multiple regression analysis in females. Variables

Number of variables taken 7

1 2 3 4 5 6 7 Constant SEE r-Value *

Total facial height (n–gn) Upper facial height (n–pr) Height of lower face (sto–gn) Minimum frontal breadth (ft–ft) Bigonial breadth (go–go) Biocular breadth (ec–ec) Interocular breadth (en–en)

6

1.901 2.230 3.398 0.0261 0.247 0.724 0.699 148.965 2.905 0.280

5

4

3

1.907 2.257 3.482 0.031 0.250 0.639

1.971 2.571 3.375 0.053 0.232

1.955 2.536 3.330 0.168

147.749 2.897 0.275

152.932 2.899 0.258

154.381 2.890 0.254

2 1.975 2.600 3.332

155.834 2.880 0.251*

1 0.747 0.375

152.716 2.914 0.184

0.993

152.510 2.903 0.181*

p < 0.05.

Table 9 Comparisons of stature and facial measurements of males with other studies.

Mean ± SD of stature

Krishan and Kumar [14] Males (N = 252) 152.647 ± 7.103

Variables

Mean ± SD

Total facial height (n–gn) Upper facial height (n–pr) Height of lower face (sto–gn) Minimum frontal breadth (ft–ft) Bigonial breadth (go–go) Biocular breadth (ec–ec) Interocular breadth (en–en)

10.240 ± 0.823 6.305 ± 0.418 – 9.923 ± 0.790 8.347 ± 0.381 – –

r-Value of CC ***

0.345 0.364*** 0.515*** 0.449***

Krishan [15] Males (N = 996) 172.31 ± 6.83 SEE of LR 5.82 6.31 – 5.13 5.63 – –

Mean ± SD 10.81 ± 0.735 – – – 9.783 ± 0.377 – –

Present study Males (N = 173) 165.90 ± 3.69 r-Value of CC 0.455

***

0.462***

SEE of LR 5.82 – – – 5.13 – –

Mean ± SD 11.25 ± 0.67 6.85 ± 0.48 3.67 ± 0.37 10.59 ± 0.62 10.64 ± 0.63 9.86 ± 0.60 2.27 ± 0.33

r-Value of CC **

0.219 0.270*** 0.028 0.088 0.064 0.071 0.030

SEE of LR 3.61 3.56 3.70 3.69 3.69 3.69 3.70

CC, correlation coefficient; SEE, standard error of estimation; LR, linear regression. ** p < 0.01. *** p < 0.001.

although the means of all the measurements in the males of our population were found to be higher, their SD’s were found to be lower than the other two populations except for Upper facial height (0.48 cm) and Bigonial breadth (0.63 cm). On comparing the r-values of the correlation coefficients of our population with these two populations we found that the r-values of our sample was lower than the other two populations for all the facial measurements although significant for total facial height (p < 0.01) and for upper facial height (p < 0.001). Data on females cannot be compared as no such literature is available. In the present study, there was not much difference in the ranges of SEE obtained from multiple regression (3.569–3.610 in males and 2.880–2.914 in females) and linear regression equations

(3.56–3.70 in males and 2.90–2.95 in females). The results of multiple regression analysis by Ryan and Bidmos [10] having one similar parameter of upper facial height with our study showed higher values of SEE (4.37–4.70 in males and 6.09–6.24 in females) than our study. Krishan and Sharma [23] recorded SEE to be lower using multiple regression analysis as compared to linear regression analysis. They reconstructed stature from various dimensions of hands and feet using both linear and multiple regression analyses and found multiple regression equations to be better indicator of stature estimation. Whereas, in our study there were not much difference in the values of SEE computed using these regression analyses.

D. Sahni et al. / Legal Medicine 12 (2010) 23–27

5. Conclusions Estimation of stature from facial measurements is a supplementary approach when useful samples like extremities or other body parts are not available for examination. Regression equations are found to be more reliable than multiplication factors. Conflict of interest Authors have no financial or personal conflict of interest regarding this manuscript. Acknowledgements The authors are grateful to Director General, Indian Council of Medical Research, New Delhi and Directorate of Forensic Sciences, New Delhi for financial support. Corresponding author has full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. References [1] O’Connor WG. Briefly unidentified: a study of peculiar source of identification. J Forensic Sci 1999;44:713–5. [2] Tssokos M, Turk EE, Madea B, Koop E, Longauer F, Szabo M. Pathologic features of suicidal deaths caused by explosives. Am J Forensic Med Pathol 2003;24:55–63. [3] Ozaslan A, Koc S, Ozaslan I, Tugsu H. Estimation of stature from upper extremity. Mil Med 2006;171:288–91. [4] Duggal N, Nath S. Estimation of stature using percutaneous lengths of radius, ulna and tibia among Lodhas and Mundas of district Midnapore, West Bengal. Anthropologist 1986;24:23–7. [5] Nath S, Duggal N, Chandra NS. Reconstruction of stature on the basis of percutaneous lengths of forearm bones among Mundas of Midnapore district, W. Bengal. Human Sci 1988;37:170–5.

27

[6] Nath S, Rajni, Chibber S. Reconstruction of stature from percutaneous measurements of upper and lower extremity segments among Punjabi females of Delhi. Ind J Forensic Sci 1990;4:171–81. [7] Nath S. Estimation of stature through hand and foot lengths among Jats of Rajasthan. Police Res Dev J 1997;QE. IV:20–2. [8] Introna Jr F, Di Vella G, Petrachi S. Determination of height in life using multiple regression of skull parameters. Boll Soc Ital Biol Sper 1993;69:153–60. [9] Chiba M, Terazawa K. Estimation of stature from somatometry of skull. Forensic Sci Int 1998;97:87–92. [10] Ryan I, Bidmos MA. Skeletal height reconstruction from measurements of the skull in indigenous South Africans. Forensic Sci Int 2007;167:16–21. [11] Saxena SK, Jeyasingh P, Gupta AK, Gupta CD. The estimation of stature from head-length. J Anat Soc India 1981;30:78–9. [12] Jadav HR, Shah GV. Determination of personal height from the length of head in Gujarat region. J Anat Soc India 2004;53(1):20–1. [13] Kumar J, Chandra L. Estimation of stature using different facial measurements among the Kabui Naga of Imphal valley, Manipur. Anthropologist 2006;8(1):1–3. [14] Krishan K, Kumar R. Determination of stature from cephalo-facial dimensions in a North Indian population. Leg Med 2007;9:128–33. [15] Krishan K. Estimation of stature from cephalo-facial anthropometry in north Indian population. Forensic Sci Int 2008;181:52e1–e6. [16] Singh IP, Bhasin MK. A manual of biological anthropology. India: Kamla-Raj Enterprises; 2004. p. 161, 179–83. [17] Singh D, Sanjeev, Singh G, Jit I, Singh P. Facial soft tissue thickness in northwest Indian adults. Forensic Sci Int 2008;176:137–46. [18] George RM. Anatomical and artistic guidelines for forensic facial reconstruction. In: Iscan MY, Helmer RP, editors. Forensic analysis of the skull craniofacial analysis, reconstruction, and identification. New York: Wiley-Liss, Inc.; 1993. p. 215–27. [19] Krogman WM, Iscan MY. The human skeletal in forensic medicine. Springfield, IL: Charles C. Thomas; 1986. [20] Holland TD. Brief communication: estimation of adult stature from the calcaneus and talus. Am J Phys Anthropol 1995;96:315–20. [21] Hauser R, Smolinski J, Gos T. The estimation of stature on the basis of measurements of the femur. Forensic Sci Int 2005;147(1–2): 185–90. [22] Patil KR, Mody RN. Determination of sex by discriminant function analysis and stature by regression analysis: a lateral cephalometric study. Forensic Sci Int 2005;147:175–80. [23] Krishan K, Sharma A. Estimation of stature from dimensions of hands and feet in a north Indian population. J Forensic Leg Med 2007;14:327–32.