- Email: [email protected]
Estimation of stature and body weight in Slovak adults using static footprints: A preliminary study
Legal Medicine 34 (2018) 7–16
Contents lists available at ScienceDirect
Legal Medicine journal homepage: www.elsevier.com/locate/legalmed
Estimation of stature and body weight in Slovak adults using static footprints: A preliminary study
T
⁎
Zuzana Caplovaa, Petra Švábováa, , Mária Fuchsováb, Soňa Masnicovác, Eva Neščákováa, Silvia Bodorikováa, Michaela Dörnhöferováa, Radoslav Beňuša a
Department of Anthropology, Faculty of Natural Sciences, Comenius University, Mlynská dolina, 84215 Bratislava, Slovak Republic Department of Didactics of Natural Sciences in Primary Education, Faculty of Education, Comenius University, Račianska 59, 81334 Bratislava, Slovak Republic c Department of Criminalistics and Forensic Sciences, Academy of Police Forces, Sklabinská 1, 83517 Bratislava, Slovak Republic b
A R T I C LE I N FO
A B S T R A C T
Keywords: Forensic podiatry Diagonal axes Stature prediction Body weight prediction Regression equations
The stature and the body weight as part of the biological profile can aid the personal identification. The dimensions of the human foot, as well as the footprint, can be used for the prediction due to the existence of its positive correlation with the stature and body weight. Five diagonal axes and ball breadth of bilateral static footprints of 132 young Slovak adults were obtained. All diameters were larger in a male group than female group. No bilateral differences were found except the first diagonal axis and ball breadth. A positive correlations between the selected footprint diameters with the stature (r = 0.37–0.64) and the body weight (r = 0.29–0.71) were confirmed. The linear and multiple regression prediction equations were developed. A stature prediction equation using the most lateral diameters (the fourth and fifth diagonal axis) exhibited the highest accuracy ranging from 4 to 7.5 cm. Similar results were found for the body weight estimation of the male and mixed group. In the female group, the most medial axis (first and second) exhibited the highest accuracy. The body weight estimation accuracy ranges from 9.09 to 11.09 kg. The real and predicted stature and body weight were compared and found differences were lower than calculated SEEs. Thresholds and prediction trend of under- or overestimation was identified. The results of the present study show that selected measurements of static footprints could be used to predict stature and body weight but should be applied only for Slovaks due to population specificity.
1. Introduction The body height and body weight are the basic characteristics often used to describe another person and they are part of a biological profile. Together with the sex and age, if estimated correctly, may aid a personal identification. Multiple publications exploring mainly the stature estimation from a dimension of the various body parts can be found, among which the lower extremities show high correlations [1,2,3]. Similarly to the direct foot measurements, the footprints attracted attention and can be used for the predictions [4–5]. An extensive research and development of the stature prediction equations from the footprints were conducted in the countries where barefoot walking is common, for instance in India. On the other hand, only a few studies focus on the estimation of the body weight from the footprints [4–8]. Footprints are found in the majority of crime scenes but may be overlooked or underestimated [6]. Considering their individuality, the stature and body weight footprint estimations could be particularly
⁎
useful in the case of home violence. Alike the individual crease pattern of footprints [9], specific length and width measurements highlight their uniqueness [4,6–7]. Several quantitative footprint evaluation methods were developed utilizing the diagonal, parallel and horizontal measurements of the heel, ball, and toes or their combinations [10]. The positive correlations of the stature/body weight and various measurements of the footprint have been observed, but the population differences preclude the development of universal prediction equations [5–8,11]. Previously published literature confirmed the existence of the strong correlation between direct foot measurements and the stature of Slovak adults [12,13]. The presented research uses the diagonal footprint measurements based on Robbins [5] (five axes connecting the backmost point on the heel with the foremost point of each toe and breadth of the ball) to assess the footprint-stature and footprint-body weight correlations. The publication aims towards the development of the populationspecific stature and body weight prediction equations, as they have not
Corresponding author. E-mail address: [email protected] (P. Švábová).
https://doi.org/10.1016/j.legalmed.2018.07.002 Received 26 April 2018; Received in revised form 20 July 2018; Accepted 25 July 2018 Available online 25 July 2018 1344-6223/ © 2018 Published by Elsevier B.V.
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
been developed for Slovak adults yet.
2. Materials and methods The body weight and the stature of 132 young adults were measured, and a series of footprints were collected. The data collection was part of an anthropometric research project of young Slovak population conducted at the Faculty of Natural Sciences of Comenius University in Bratislava, Slovakia. All the participants received written instructions and details of the study explaining its aims, methods and expected involvement. Participation in the study was voluntarily and entirely based on a written informed consent, and each participant was ensured that he/she could withdraw from the study. All participants were measured approximately at the same time in the morning to avoid the diurnal variation in the stature [14,15]. Stature was measured using an anthropometer from the vertex to the floor in the anatomical position with the head oriented in the Frankfurt Plane by a single operator according to recommendations by [5]. The body weight was recorded using an analog body weight scale with a precision of 0.1 kg. The collection of the bilateral static footprint was part of the gait pattern and dynamic footprint collection. Each participant applied an even layer of non-coloring cream on both feet. Participants were asked to stand up on a prepared recycled paper (5 m in length and 1 m in width). The first set of the static bilateral footprint was obtained. Then participants walked naturally towards the end of the prepared paper and stopped with a left and right foot next to each other. The set of dynamic footprints reflecting the gait pattern and the second set of static footprints were obtained. Each footprint was immediately traced by a pencil for better preservation. Only bilateral static footprints were used for this study. Static (standing) footprints of 64 females (from 18 to 33 years old, with a mean age of 21.07 years old) and 68 males (from 18 to 25 years old, with a mean age of 20.59 years old) were collected. Six measurements comprising of five length diagonal dimensions and breadth of the ball were measured by a single operator. Firstly, the backmost point of the heel was identified (Pternion). Five diagonal axes connecting Pternion with the most anterior point of each toe was drawn according to [5]. The most lateral and medial points of the ball were used to obtain the ball breadth dimension (Fig. 1). The left footprint measurements were designed as DA_1L, DA_2L, DA_3L, DA_4L, DA_5L, and Br_L. The right footprint measurements were designed as DA_1R, DA_2R, DA_3R, DA_4R, DA_5R, and Br_R (Fig. 1).
Fig. 1. Five diagonal axes of the footprint. p – the backmost point of the heel; DA_1 - DA_5 – diagonal axes, Br – ball breadth.
3. Results Descriptive statistical analysis (mean, standard deviation, minimum, maximum) of stature, body weight, breadth of the ball and all diagonal axes of static footprints are shown in Table 1. Mean values of all measurements were found to be significantly larger in males than in females (p-value < 0.05). Statistically significant bilateral differences (p-value < 0.05) were only observed for the breadth of the ball and the diagonal axis of the first toe of females and males (Table 1). The results of the analysis of covariance confirmed the influence of sex on the relationship between stature and diagonal axes of static footprints (p-value < 0.05). Therefore, the results of the regression analysis are presented separately by sex as well as by side since it is possible to visually distinguish laterality of a footprint. The mixed group (males + females) is presented as well.
2.1. Statistic evaluation Data were statistically evaluated in SPSS Statistics 19.0. ShapiroWilk test confirmed a normal distribution of data (p-value > 0.05). Intersexual differences were tested by two-sample t-test and bilateral differences by paired t-test. Pearson correlation coefficient was employed to assess the stature-measurements and body weight-measurements correlations. An analysis of covariance (using GLM, general linear model) was done to test whether sex has an impact on the relationship between diagonal axes and body weight or stature. Prediction equations for stature and body weight were derived by linear regression analysis and multiple regression analyses [16]. Developed prediction equations were tested. Body weight and stature were calculated using every linear prediction equation in male, female and mixed group. Statistical differences between the measured and calculated data were verified by a two-sample t-test. Differences between the measured and calculated stature and body weight were calculated and the Pearson correlation coefficient was used to assess the relationship between the calculated difference and the stature or body weight and BMI.
3.1. Stature Correlation coefficients for the stature and static footprint measurements are shown in Table 2. All diagonal axes (p-value < 0.01) and ball breadth (p-value < 0.05) exhibited statistically significant correlation with the stature. In males, the highest correlation coefficient was between the stature and the fourth diagonal axis of the left footprint (r = 0.64). The lowest correlation coefficient was observed with the ball breadth of the right footprint (r = 0.37). In females, the fifth diagonal axis of the left footprint resulted in the highest correlation coefficient (r = 0.69), while the ball breadth of the left footprint showed the lowest correlation with the stature (r = 0.27). In the mixed group, the highest correlation coefficient was between the stature and the fourth diagonal axis of the left footprint (r = 0.84), and the lowest with the ball breadth of the left footprint (r = 0.60). Table 2 shows that 8
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Table 1 Descriptive statistics of stature (cm), body weight (kg) and selected measurements of static footprints (cm) and their bilateral differences (p-value). Females n = 64
Stature Weight DA_1 L DA_1 R DA_2 L DA_2 R DA_3 L DA_3 R DA_4 L DA_4 R DA_5 L DA_5 R Br_L Br_R
Bilateral differences
Males n = 68
Bilateral differences
Mean
SD
Min
Max
Mean
SD
Min
Max
173.2 58.16 23.54 23.24 22.97 22.75 21.96 21.91 20.71 20.72 19.27 19.19 8.79 8.58
9.55 8.41 1.26 1.23 1.34 1.25 1.28 1.20 1.12 1.17 1.10 1.06 0.65 0.66
152.1 45.00 21.00 20.80 19.80 19.70 19.20 19.20 18.30 18.00 16.80 17.10 7.50 6.60
190.7 80.00 26.80 26.80 26.60 25.80 25.50 25.00 23.70 23.80 21.70 22.00 10.90 10.30
173.7 78.27 26.15 25.82 25.64 25.47 24.49 24.48 23.23 23.15 21.51 21.58 9.85 9.66
9.55 12.89 1.43 1.60 1.43 1.64 1.36 1.66 1.25 1.66 1.23 1.46 0.77 0.80
161 57.00 23.50 22.90 22.40 22.10 21.30 21.40 20.80 20.40 19.30 19.00 8.30 8.20
198.4 115.0 29.60 30.70 29.30 31.20 28.00 29.90 26.70 28.20 24.50 25.50 11.50 11.30
0.004* 0.060 0.456 0.983 0.482 *
0.004
0.006* 0.113 0.615 0.174 0.926 0.020*
DA – diagonal axis, 1–5 – first to the fifth toe, R – right, L – left, Br – ball breadth, * – significance at < 0.05. Table 2 Correlation between stature and measurements of static footprints. Females n = 64
Males n = 68
Mixed group n = 132
Left DA_1 DA_2 DA_3 DA_4 DA_5 Br
R 0.65*** 0.65*** 0.63*** 0.65*** 0.69*** 0.19**
r 0.61*** 0.62*** 0.63*** 0.64*** 0.59*** 0.33***
r 0.82*** 0.82*** 0.82*** 0.84*** 0.82*** 0.58***
Right DA_1 DA_2 DA_3 DA_4 DA_5 Br
0.56*** 0.57*** 0.62*** 0.61*** 0.64*** 0.22***
0.49*** 0.54** 0.54*** 0.55*** 0.62*** 0.35***
0.76*** 0.78*** 0.78*** 0.77*** 0.81*** 0.60***
Table 3 Linear regression equations for stature estimation (cm) from measurements of the static footprint. Females n = 64 Left
R2
SEE
right
R2
SEE
96.58 + 2.95 DA_1 101.88 + 2.79 DA_2 103.27 + 2.85 DA_3 96.94 + 3.33 DA_4 96.32 + 3.60 DA_5 144.91 + 2.47 Br
0.42 0.42 0.40 0.42 0.48 0.07
4.44 4.46 4.50 4.43 4.14 5.85
104.48 + 2.64 DA_1 106.34 + 2.61 DA_2 100.57 + 2.98 DA_3 103.67 + 3.00 DA_4 98.89 + 3.49 DA_5 141.01 + 2.96 Br
0.32 0.32 0.39 0.37 0.41 0.11
4.80 4.76 4.53 4.58 4.48 5.70
Left
R2
SEE
right
R2
SEE
101.87 + 2.99 DA_1 101.57 + 3.06 DA_2 99.26 + 3.30 DA_3 95.15 + 3.66 DA_4 106.28 + 3.43 DA_5 144.84 + 3.06 Br
0.37 0.38 0.40 0.41 0.35 0.14
5.69 5.61 5.56 5.47 5.86 6.52
123.54 + 2.19 120.54 + 2.34 123.16 + 2.33 125.34 + 2.36 113.82 + 3.06 149.80 + 3.15
0.24 0.30 0.30 0.30 0.39 0.12
6.21 6.00 6.03 5.99 5.70 6.60
Left
R2
SEE
right
R2
SEE
69.34 + 4.18 DA_1 73.81 + 4.09 DA_2 73.17 + 4.30 DA_3 71.15 + 4.64 DA_4 72.64 + 4.90 DA_5 110.66 + 6.70 Br
0.67 0.67 0.67 0.70 0.67 0.35
5.59 5.55 5.53 5.29 5.60 7.54
80.35 + 3.78 DA_1 82.49 + 3.76 DA_2 82.84 + 3.89 DA_3 86.31 + 3.95 DA_4 81.57 + 4.48 DA_5 112.98 + 6.56 Br
0.57 0.60 0.61 0.60 0.66 0.36
6.31 6.10 6.09 6.16 5.60 7.51
Males n = 68
DA – diagonal axis, Br – ball breadth, 1–5 – first to fifth toe, r – correlation coefficient, ** – significant at < 0.01, *** – significant at < 0.001.
correlation coefficients are higher in the mixed group (0.60–0.84) when compared to the male (0.37–0.64) or female group (0.27–0.69). Table 3 presents the linear regression equations for stature estimation together with the values of coefficients of determination (R2) and standard errors of estimates (SEE) representing the standard deviation of the stature estimation. The SEE ranges from ± 4.1 cm to ± 5.9 cm in the female group, from ± 5.4 cm to ± 6.6 cm in the male group, and from ± 5.3 cm to ± 7.5 cm in a mixed group. Regression parameters are statistically significant for all the measurements (p-value < 0.05) except for the ball breadth of the left and right footprint in the female group, and for the ball breadth of the left footprint in the male group. Table 4 illustrates multiple regression equations for the stature estimation based on various combinations of diagonal axes and ball breadth measurements for the left and right footprint. For the practical use, only multiple regression equations for the mixed group were calculated as it is not possible to visually estimate the sex from the footprint. The first and the fifth diagonal axis of the left footprint showed the lowest SEE ± 5.33 cm. For the right footprint, the combination of the fifth diagonal axis and the ball breadth showed the lowest SEE ± 5.45 cm.
DA_1 DA_2 DA_3 DA_4 DA_5 Br
Mixed Group n = 132
DA – diagonal axis, 1–5 – first to the fifth toe, Br – ball breadth, R2 – coefficient of determination, SEE – standard error of estimate.
show statistically significant correlation with the body weight (pvalue < 0.01). The highest correlation coefficient in the male group can be seen between the body weight and the fifth diagonal axis of the right footprint (r = 0.50), and the lowest between the body weight and the first diagonal axis of the right footprint (r = 0.40). In the female group, the second diagonal axis of the left footprint resulted in the highest correlation coefficient (r = 0.46), while the fifth diagonal axes of the right footprint showed the lowest correlation with the body weight (r = 0.29). In the mixed group, the highest correlation coefficient was between the body weight and the second, fourth and fifth diagonal axis of the left footprint (r = 0.71), and the lowest with the ball breadth of the right footprint (r = 0.65). The highest correlation coefficients for the sex-mixed group are shown in Table 5.
3.2. Body weight Correlation coefficients for the body weight and static footprint measurements are shown in Table 5. All diagonal axes and ball breadth 9
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Table 4 Multiple regression models for stature estimation (cm) from footprint measurements in a mixed group.
Table 6 Linear regression equations for body weight estimation (kg) from measurements of the static footprint.
Left
R2
SEE
Females n = 64
68.73 + 4.01 DA_1L + 0.51 Br 70.22 + 4.51 DA_5L + 1.15 Br 65.39 + 2.24 DA_1L + 2.54 DA_5 65.03 + 2.53 DA_5 + 2.15 DA_1 + 0.31 Br
0.67 0.68 0.70 0.70
5.60 5.56 5.33 5.35
Left
R2
SEE
Right
R2
SEE
Right 76.46 + 3.12 DA_1 + 2.20 Br 76.93 + 3.81 DA_1 + 2.01 Br 78.94 + 3.70 DA_5 + 0.75 DA_1 75.71 + 3.42 DA_5 + 1.92 Br + 0.41 DA_1
R2 0.60 0.68 0.67 0.68
SEE 6.14 5.45 5.60 5.46
−41.76 + 4.29 DA_1 −35.55 + 4.29 DA_2 −16.56 + 3.45 DA_3 −24.97 + 4.06 DS_4 −22.42 + 4.23 DA_5 −3.17 + 7.08 Br
0.24 0.25 0.16 0.17 0.18 0.17
9.65 9.68 10.15 10.09 10.15 10.05
−33.69 + 3.99 DA_1 −32.66 + 4.04 DA_2 −24.55 + 3.82 DA_3 −25.39 + 4.08 DA_4 −14.26 + 3.82 DA_5 −2.66 + 7.21 Br
0.19 0.21 0.17 0.19 0.13 0.18
9.92 9.94 10.16 10.07 10.46 9.98
Left
R2
SEE
Right
R2
SEE
−26.51 + 3.99 DA_1 −27.76 + 4.12 DA_2 −30.09 + 4.4 DA_3 −36.74 + 4.93 DA_4 −31.21 + 5.07 DA_5 4.90 + 7.41 Br
0.19 0.20 0.20 0.22 0.23 0.19
12.04 11.96 11.91 11.82 11.66 12.05
−11.57 + 3.46 −17.59 + 3.75 −12.49 + 3.69 −14.62 + 3.99 −24.03 + 4.72 7.57 + 7.27 Br
0.17 0.22 0.21 0.25 0.26 0.19
12.13 11.82 11.94 11.68 11.76 12.01
Left
R2
SEE
Right
R2
SEE
−71.02 + 5.62 DA_1 −65.71 + 5.52 DA_2 −61.65 + 5.6 DA_3 −64.81 + 6.06 DA_4 −63.86 + 6.49 DA_5 −34.11 + 11.02 Br
0.47 0.48 0.45 0.47 0.47 0.41
11.26 11.2 11.46 11.22 11.3 11.93
−61.36 + 5.30 DA_1 −58.35 + 5.26 DA_2 −54.51 + 5.31 DA_3 −54.45 + 5.56 DA_4 −55.00 + 6.06 DA_5 −30.18 + 10.83 Br
0.44 0.47 0.45 0.46 0.46 0.41
11.58 11.3 11.57 11.38 11.47 11.84
Males n = 68
DA – diagonal axis, Br – ball breadth, 1–5 – first to the fifth toe, R2 – coefficient of determination; SEE – standard error of estimate. Table 5 Correlation between body weight and measurements of static footprints. Females n = 64
Males n = 68
Mixed group n = 132
Left DA_1 DA_2 DA_3 DA_4 DA_5 Br
r 0.42** 0.46** 0.40** 0.38** 0.37** 0.44**
r 0.41** 0.44** 0.44** 0.45** 0.49** 0.42**
R 0.69** 0.71** 0.69** 0.71** 0.71** 0.66**
Right DA_1 DA_2 DA_3 DA_4 DA_5 Br
0.35** 0.41** 0.40** 0.41** 0.29** 0.42**
0.40** 0.45** 0.45** 0.48** 0.50** 0.41**
0.67** 0.70** 0.69** 0.70** 0.70** 0.65**
DA_1 DA_2 DA_3 DA_4 DA_5
Mixed group n = 132
DA – diagonal axis, Br– ball breadth, 1–5 – first to the fifth toe, R2 – coefficient of determination, SEE – standard error of estimate Table 7 Multiple regression models for body weight estimation (kg) from footprint measurements in a mixed group.
DA – diagonal axis, Br – ball breadth, 1–5 – first to the fifth toe, r – correlation coefficient, ** – significant at < 0.01.
The regression equations for the body weight estimation, standard errors of estimates (SEE) and values of coefficients of determination (R2) are listed in Table 6. The SEE ranges from ± 9.65 kg to ± 10.46 kg in the female group, from ± 11.66 kg to ± 12.13 kg in the male group, and from ± 11.20 kg to ± 11.90 kg in the mixed group. Regression parameters are statistically significant for all the measurements (p-value < 0.05). The results of the multiple regression analysis for the body weight estimation based on various combinations of diagonal axes and ball breadth measurements for the left and right footprint are presented in Table 7. As in the case of stature estimation, only multiple regression equations for the mixed group were calculated. The second, fifth diagonal axis and ball breadth of the left footprint showed the lowest SEE ± 10.64 kg. For the right footprint, three combinations showed the lowest SEE ± 10.57 kg, namely the first and second diagonal axis with the ball breadth; the second and third diagonal axis with the ball breadth; and the second, third and fourth diagonal axis with the ball breadth.
Left
R2
SEE
−77.57 + 5.3 Br + 3.85 DA_1 −78.62 + 4.96 Br + 2.10 DA_1 + 2.0 0 DA_2 −82.11 + 4.36 Br + 1.90 DA_1 + 1.74 DA_2 − 3.50 DA_3 + 3.70 DA_4 + 0.99 DA_5 −77.76 + 2.33 DA_2 + 2.23 DA_5 + 4.74 Br
0.52 0.53 0.54
10.75 10.74 10.65
0.53
10.64
Right −70.97 + 5.99 Br + 0.38 DA_1 + 3.14 DA_2 −71.28 + 3.00 DA_2 + 0.46 DA_3 + 6.25 Br −69.95 + 2.56 DA_2 − 2.65 DA_3 + 3.72 DA_4 + 6.09 Br −70.26 + 1.95 DA_4 + 1.92 DA_5 + 6.26 Br
R2 0.53 0.54 0.55 0.55
SEE 10.57 10.57 10.57 10.58
DA – diagonal axis, Br – ball breadth, 1–5 – first to the fifth toe, R2 – coefficient of determination, SEE – standard error of estimate.
lower in the mixed group, and are ranging from r = −0.55 to r = −0.79 (Table 8). Average differences between the calculated and real body weight for each group and measurement are listed in Table 9 together with its Pearson correlation coefficient (r) with the body weight. Additionally, the correlation with Body Mass Index (BMI) was calculated to better understand the relationship between the performance of the body weight prediction equations and the relative body size. The differences between the calculated and real body weight are lower than SEE values for each prediction equation listed in Table 6. Similarly, as in case of stature, all observed differences exhibited statistically significant negative correlation with the body weight (p-value < 0.01) and with BMI (p-value < 0.01). Correlation with the body weight ranges from r = −0.86 to r = −0.93 in the female group, r = −0.86 to r = −0.91 in the male group and r = −0.53 to r = −0.74 in the mixed group (Table 9). In the case of BMI, the correlation ranges from r = −0.84 to
3.3. Application Average differences between the calculated and real stature for each group and measurement are listed in Table 8 together with its Pearson correlation coefficient (r) with the stature. The differences between the calculated and real stature are lower than SEE values for each prediction equation listed in Table 3. All differences exhibited statistically significant negative correlation with the stature (p < 0.01). The correlation coefficient ranges from r = −0.72 to r = −0.96 in the female group and r = −0.77 to r = −0.94 in the male group. Correlations are 10
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
predicted body weight was calculated for the participants with low real body weight. The largest negative difference (underestimation) was calculated for the participants with the highest real body weight. The predicted body weight of females under 53 kg (or BMI 20.5) was always overestimated (positive difference between the estimated and the real body weight), and over 65 kg (or BMI 24) always underestimated (Fig. 5). In the male group, the predicted body weight of males weighing under 70 kg (or BMI 20.9) was always overestimated, and over 86 kg (or BMI 26.5) underestimated (Fig. 6). In the mixed group, the predicted weight of individual weighting under 52.5 kg (or BMI 18.3) was always overestimated and over 86 kg (or BMI 26.5) always underestimated (Fig. 7). More specifically, unisex equations overestimated the body weight of females under 52.5 kg (or BMI 18.3) and males under 60 kg (or BMI 20.8) and underestimated females over 65 kg (or BMI 24) and males over 86 kg (or BMI 26.5). Similarly, as in the stature prediction, both positive and negative errors were calculated for the participants with body weight between the thresholds.
Table 8 Average differences (cm) between the real and calculated stature and Pearson correlation coefficient (r) between the stature and average differences. Females n = 64
Males n = 68
Mixed group n = 132
Average difference
r
Average difference
r
Average difference
r
Br_L DA_1L DA_2L DA_3L DA_4L DA_5L
± 4.69 ± 3.38 ± 3.63 ± 3.57 ± 3.58 ± 3.16
−0.96** −0.76** −0.76** −0.77** −0.76** −0.72**
± 6.68 ± 4.48 ± 4.29 ± 4.32 ± 4.29 ± 4.46
−0.94** −0.80** −0.80** −0.78** −0.77** −0.81**
± 6.28 ± 4.49 ± 4.39 ± 4.31 ± 4.12 ± 4.20
−0.79** −0.58** −0.57** −0.57** −0.55** −0.58**
Br_R DA_1R DA_2R DA_3R DA_4R DA_5R
± 4.68 ± 3.50 ± 3.74 ± 3.68 ± 3.64 ± 3.53
−0.94** −0.83** −0.82** −0.78** −0.79** −0.77**
± 5.26 ± 4.88 ± 4.98 ± 4.57 ± 4.60 ± 4.27
−0.94** −0.87** −0.84** −0.84** −0.84** −0.79**
± 6.30 ± 4.94 ± 4.80 ± 4.79 ± 4.77 ± 4.18
−0.78** −0.65** −0.63** −0.63** −0.64** −0.58**
DA – diagonal axis, 1–5 – first to the fifth toe, R – right, L – left, Br – ball breadth, r – Person correlation coefficient, ** – significant at < 0.01.
4. Discussion
r = −0.94 in the female group, r = −0.83 to r = −0.91 in the male group, and from r = −0.54 to r = −0.89 in the mixed group. Moreover, the trend in a stature prediction can be identified for every footprint measurement in the female, male and mixed group. The largest positive error (overestimation) in the stature prediction can be seen in case of participants with the lowest real stature. When predicting the stature of the higher person, the negative error (underestimation) occurred. Similarly, the largest underestimation was observed in case of participants with the highest real stature. Figs. 2–4 demonstrate the trend of under/overestimation of the body stature. The average of all calculated differences (stature estimation from 10 diagonal axes and 2 ball breadths) shows that in the case of females under 164.4 cm in height the developed prediction equations only overestimated the stature, and over 171.2 cm only underestimated (Fig. 2). Likewise in the male group (Fig. 3), the only overestimation can be seen under the height of 176.1 cm and only underestimation over the height of 186.0 cm. The predicted stature between two thresholds resulted in both positive and negative errors. When unisex equations were applied to the mixed group, predicted equation only overestimated females under 164.7 cm and males under 169.9 cm, and only underestimated females over 171.2 cm and males over 186.0 cm (Fig. 4). Figs. 5–7 illustrate the similar trend in the body weight prediction. The largest positive difference (overestimation) between the real and
The results of the study showed the correlation of selected measurements of static footprints with the body weight and stature, and therefore the potential use for their estimation in the Slovak population. Similarly to the previous studies [5–7,17], developed prediction equations are population specific. All footprint dimensions of young Slovak adults are larger in comparison to the previous publications, which further highlights their population specificity. Differences were found between the obtained footprint measurements and measurements reported for Iban Ethnic of East Malaysia [18] and the Indian population [6–7]. In both cases, the measurements of diagonal axes of Slovak adults were larger. Differences in some measurements, namely ball breadth, and first diagonal axis, were found in comparison to the Australian population [17]. In this case, however, the ball breadth measurements in our sample were smaller. Differences were found in all measurements in comparison to Ghanaian [19] and Nigerian [20] population. All female dimensions reported by [19–20] were larger in comparison to our sample, and all male dimension were smaller. Likewise, the dimensions of ball breadth of females from Turkey were larger, and of males smaller than in our sample [21]. Significant intersexual differences were found and mean values of all measurements were significantly larger in the male group than in female (p-value < 0.05) which is in accordance with the previous studies [19–20,22]. All measurements were larger on the left footprints,
Table 9 Average differences (kg) between the real and calculated body weight and Pearson correlation coefficient (r) between the average differences and body weight and BMI. Females n = 64
Br_L DA_1L DA_2L DA_3L DA_4L DA_5L Br_R DA_1R DA_2R DA_3R DA_4R DA_5R
Males n = 68
Mixed group n = 132
Diff.
r BW
r BMI
Diff.
r BW
r BMI
Diff.
r BW
r BMI
± 6.66 ± 7.13 ± 8.23 ± 6.99 ± 7.10 ± 7.15 ± 6.79 ± 7.25 ± 7.06 ± 7.17 ± 7.13 ± 7.56
−0.91** −0.87** −0.86** −0.92** −0.91** −0.91** −0.90** −0.80** −0.89** −0.91** −0.90** −0.93**
−0.84** −0.91** −0.90** −0.93** −0.93** −0.94** −0.84** −0.91** −0.91** −0.93** −0.92** −0.94**
± 8.99 ± 9.51 ± 9.20 ± 8.98 ± 8.78 ± 8.67 ± 8.71 ± 9.01 ± 8.93 ± 8.89 ± 8.80 ± 8.50
−0.90** −0.90** −0.90** −0.89** −0.88** −0.88** −0.90** −0.91** −0.89** −0.89** −0.87** −0.86**
−0.83** −0.90** −0.90** −0.91** −0.90** −0.89** −0.83** −0.88** −0.88** −0.88** −0.87** −0.88**
± 10.86 ± 8.95 ± 8.91 ± 8.96 ± 8.35 ± 9.21 ± 11.86 ± 8.93 ± 8.51 ± 8.80 ± 9.32 ± 8.50
−0.56** −0.73** −0.63** −0.66** −0.73** −0.67** −0.53** −0.75** −0.73** −0.74** −0.65** −0.73**
−0.54** −0.85** −0.77** −0.87** −0.87** −0.74** −0.80** −0.84** −0.83** −0.84** −0.74** −0.89**
DA – diagonal axis, 1–5 – first to the fifth toe, R – right, L – left, Br – ball breadth, Diff. – average difference, r – Person correlation coefficient, ** – significant at < 0.01, BW – body weight, BMI – body mass index. 11
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Fig. 2. Correlation between stature (cm) and average calculated difference (cm) in the female group. a-axis – upper stature limit for overestimation, b-axis – lower stature limit for underestimation.
walking was previously reported [7,11,24] and causes greater muscle and bone development of the dominant leg to support the pressure, which results in the larger measurements. Static footprints often display missing print of the fifth toe, especially in non-athletes in comparison to athletes. Intrinsic muscles stabilizing the arch and helping to maintain toes on the ground are more
however, the differences were smaller than 3 mm and significant bilateral differences were found only for ball breadth and first diagonal axis. The intersexual and bilateral differences can be explained similarly as is in the case of direct foot measurements. Females are genetically shorter than males, and therefore their feet and footprints tend to be smaller [23]. The preference of the left leg for standing and
Fig. 3. Correlation between stature (cm) and average calculated difference (cm) in the male group. a-axis – upper stature limit for overestimation, b-axis – lower stature limit for underestimation. 12
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Fig. 4. Correlation between stature (cm) and average calculated difference (cm) in the mixed group. a-axis – upper stature limit for overestimation (females), c-axis – lower stature limit for underestimation (females), b-axis – upper stature limit for overestimation (males), d-axis – lower stature limit for underestimation (males), □ – males, ◆ – females.
and prediction equations were developed also for the mixed group (males and females) as it is not possible to visually estimate the sex from the footprint. Positive correlations of the stature with the direct foot measurements of young Slovak adults were previously described [12–13]. In our
active in the athlete group [25]. Fifth toe non-print of 14–27% [25] and 14–16% [26] of non-athletes were previously reported. In our study of non-athletic young Slovak adults, the fifth toe was missing in 7% of the right footprints and 3% of the left footprints. Despite the existence of the intersexual differences, the correlations
Fig. 5. Correlation between body weight (kg)/BMI and average calculated difference (kg) in the female group. a-axis – upper BMI limit for overestimation, b-axis – lower BMI limit for underestimation, c-axis – upper body weight limit for overestimation, d-axis – lower body weight limit for underestimation. 13
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Fig. 6. Correlation between body weight (kg) and BMI and average calculated difference (kg) in the male group. a-axis – upper BMI limit for overestimation, b-axis – lower BMI limit for underestimation, c-axis – upper body weight limit for overestimation, d-axis – lower body weight limit for underestimation.
this study are lower than reported by [27]. Correlation coefficients in the mixed group (regardless of the sex) were higher, similarly as in [5,28]. This finding further supports the fact that prediction equations without sex indicators are more reliable, especially when in the real life it is not possible to visually estimate the sex from the footprint. Stature prediction equations using the fourth and fifth diagonal axis also exhibited the lowest SEE in both sex groups. The same results were stated by [11,4]. Higher correlations with the statue and higher accuracy of prediction equations of lateral diagonal axes (fourth and fifth) occur as the longitudinal arch of the foot is more solid and supportive in its lateral part. Strong ligaments, tendons, and muscles preserve its integrity. The medial part of the longitudinal arch is more flexible and
study, the positive correlations with selected measurements of the static footprints were confirmed. The first, fourth and fifth diagonal axis exhibited the highest correlation with the stature, while the ball breadth the lowest. Findings are in accordance with the previous studies [4–5,11,17], which reported higher correlation coefficients with length footprint measurements than with breadth footprint measurements. The measurements of the left footprints exhibit higher correlation then the right footprints. In general, correlations reported in this study are higher than those reported by [11] for males, by [4,22] for a mixed group, and by [22] for females. On the other hand, correlations of stature and selected footprint dimension in all three groups in the Nigerian population [20] were higher. Similarly, the positive correlations with the ball breadth reported in
Fig. 7. Correlation between body weight (kg) and BMI and average calculated difference (kg) in the mixed group. ◆ – females, ○ – females, ◊ – males, □ – males. 14
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
and sex-mixed) and for their averages for each participant. High negative correlation between the stature and the calculated difference and between the body weight and the calculated difference resulted in a uniform trend: lower the stature or body weight (or BMI) – higher the possibility of overestimation (and higher the overestimation itself), and higher the stature or body weight (or BMI) – higher the possibility of underestimation (and higher the underestimation itself). Moreover, the upper and lower limits of over- and underestimation were identified for the female, male and sex-mixed group. Standard errors of estimates (SEE) of developed prediction equation for stature and body weight offer error margin in a positive and negative direction. Observed prediction trend illustrates the most likely scenario of over/underestimation for females, males and an unknown sex person of a certain body height or body weight which could potentially lead to more reliable prediction. Reported trends are truly the preliminary results as developed prediction equations were tested on the same sample. However, no previous studies reported such a trend to our knowledge. Therefore, more research is needed in this field. The results of the presented study show that selected measurements of footprints could be used as predictive values for stature and body weight estimation in the Slovak population. Standard errors of estimates of developed equations are relatively high, especially in the case of body weight estimation. On the other hand, when applied the differences between the real and calculated stature/body weight are lower than reported SEEs. Stature with the accuracy up to 3.16 cm and body weight with the accuracy up to 6.60 kg can be obtained. Described prediction equations should only be used only for Slovak population and only for the dimension of static footprints. Further study of a larger sample is needed, similarly, the study of dynamic (walking) footprints has to be undertaken to assess the differences between the static and dynamic footprints. Despite the fact, that the presented methodology using non-coloring cream is not validated, it was chosen over the traditional ink method. The cream method had been used in the past to study the gait pattern of rats [39]. The non-coloring cream method did not require any further feet cleaning and using recycled paper to obtain cream footprints prevented any slippery surface compared to non-recycled paper. However, in a real-life scenario, the footprints left on the slippery surface by the wet substance are found – for instance on tiles or wooden floor. The major limitation of the non-coloring cream method is the hand-drawn outline of the footprint as this could potentially add an unknown quantity of human error to the statistical analysis. Therefore, a further study of the presented footprint collection method needs to be undertaken. Less non-contact fifth toe (maximum 7%) was observed in our sample of non-athletic participants which may also be due to the used method. Further comparison in this matter is also needed.
is influenced by genetic factors, body weight and age [29]. Contrarily, the previous studies of Nigerian population reported the first and third diagonal axes exhibited the strongest correlations with the stature [20,30]. Standard errors of estimates (SEE) are lower in the female group which indicates more reliable stature estimation in a comparison to the male or mixed group. The use of linear regression equations for stature estimation could be applied also in the case of incomplete footprints when the only limited number of measurements is obtainable. In a case of complete footprints, the multiple linear regression equations combining several footprint measurements resulted in lower SEEs. Multiple regression equations of the mixed group only were developed for the practical use. The combination of the first and fifth diagonal axis of the left footprint resulted in the lowest estimation error and therefore the most reliable stature estimation. In general, SEEs for the stature estimation calculated in this study were higher than reported by [7,18,20], and lower for ball breadth than in [27]. Only a few publications deal with the body weight estimation from the footprint measurements, despite the fact that their positive correlation was confirmed [5–6,11]. The correlation with the body weight is lower than with the stature. Body weight is easily altered and therefore the body weight estimation is less accurate in comparison to the stature. Enlarged length and width footprint dimensions and increased contact area were reported during loaded walking [31] but no longitudinal studies taking into consideration the weight gain and weight loss of the same individuals were conducted. Previous longitudinal and cross-sectional studies of pregnant women described characteristic changes of the plantar surface and the gait pattern as a compensatory mechanism to improve locomotor stability. Described gait and feet modifications in pregnant women could occur as a result of a mass gain, as similar gait characteristics were reported in overweight people [32–33]. Likewise, a number of publications report larger foot dimensions and increased arch index in obese children [34–36]. The structural foot changes, mainly the flat foot caused by the continuous bearing of extra weight can be seen from the early childhood through the adulthood [34,37,38]. Foot dimensions, and therefore the footprint measurements seem to increase to support the higher weight of the person similarly as is the case of higher stature [5,6]. In our study, the highest body weight correlations were found to be with the second, fourth and fifth diagonal axis of the static footprint. All reported correlation coefficients were lower than stature-measurements correlations. The measurements of the right footprints exhibit higher correlation with the body weight in oppose to the correlation with the stature. In comparison to the previously published studies including males only, the body weight-footprint measurements correlations of Slovak males are lower than for Indian [6] but similar to the correlations reported for males from Egypt [11]. The highest correlation coefficients were confirmed with the lowest SEE for linear regression equations using first and second diagonal axis in the female group, and fourth and fifth diagonal axis in the male and sex-mixed group. Differences between the female and male group may be caused by individuality in the enlargement of the feet with the higher body weight which is varying from anterio-posterior to mediolateral direction [6]. Similarly to the stature estimation, only multiple regression equations of the mixed group were developed for the practical use. The result of the multiple linear regression showed that a combination of the second, fifth diagonal axis and ball breadth of the left footprint in a mixed group exhibit the lowest SEE = ± 10.65 kg. Addition of the parameters decreases the SEE in the mixed group. The results of the application of the stature and body weight prediction equations allow us to draw interesting preliminary conclusions. Similar estimation trend can be seen for the linear regression of each parameter, each side (left and right) and each group (females, males,
5. Conclusion Selected measurements of static footprints were used to predict the stature and body weight of Slovak young adults. No previous studies in this field were conducted, and no forensic data for the stature and body weight estimation from footprint dimensions exist for Slovak population. To our knowledge, no similar studies were published in central Europe or in a similar ethno-language Slavic population. Significant intersexual differences were found in all parameters, and no bilateral differences were found between the measurements, except the ball breadth and the first diagonal axis. The high positive correlation was found between all the parameters and stature. As expected, the correlations with body weight were lower, but significant prediction equations were developed for the stature and body weight estimation. The body height prediction equations using the fourth and fifth diagonal axis exhibited the lowest SEE in both sex groups. The highest body weight correlation coefficients were confirmed with the lowest SEE for linear regression equations for the first and second diagonal axis in the 15
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
female group, and fourth and fifth diagonal axis in the male and sexmixed group. Stature and body weight prediction equations from static footprints have not been developed for Slovak population before. After their application, the trend in over/underestimation was identified.
Forensic Sci. Criminol. 2 (2) (2014) 201–208. [19] J.K. Abledu, G.K. Abledu, F.B. Offei, E.M. Antwi, Determination of sex from footprint dimensions in a ghanaian population, PLoS ONE 10 (10) (2015) e0139897. [20] E.A. Okubike, N.M. Ideabuchi, A.O. Olabiyi, M.E. Nandi, Stature estimation from footprint dimension in an adult Nigerian student population, J. Forensic Sci. Med. 4 (1) (2018) 7–17. [21] D. Atamturk, Estimation of sex from the dimensions of foot Footprints, and Shoe, Anthropol. Anz. 68 (1) (2010) 21–29. [22] T. Kanchan, K. Krishan, S. Shyamsundar, K.R. Aparna, S. Jaiswal, Analysis of footprint and its parts for stature estimation in Indian population, Foot 22 (3) (2012) 175–180. [23] B. Danborno, A. Elukpo, Sexual dimorphism in hand and foot length, indices, stature-ratio and relationship to height in nigerians, Internet J. Forensic Sci. 3 (1) (2008) 379–383. [24] N.G. Rao, M.S. Kotian, Footprint ratio (FPR) – a clue for establishing sex identity, J. Ind. Acad. Forensic Med. 12 (1990) 51–56. [25] T. Kulthanan, S. Techakampuch, N. Donphongam, A study of footprints in athletes and non-athletic people, J. Med. Assoc. Thai. 87 (7) (2004) 788–793. [26] T.N. Moorthy, W.N.Z.W.M. Samsudin, M.S. Ismail, A study on footprints of malaysian athletes and non-athletes for application during forensic comparison, Malaysian J. Forensic Sci. 2 (1) (2011) 29–35. [27] T. Kanchan, K. Krishan, D. Geriani, I.S. Khan, Estimation of stature from the width of static footprints – insight into an Indian model, Foot 23 (4) (2013) 136–139. [28] T.N. Moorthy, Y.L. Ang, A.S. Saufee, F.N.H. Nik, Estimation of stature from footprint and foot outline measurements in Malaysian Chinese, Aust. J. Forensic Sci. 46 (2) (2013) 136–159. [29] D.J. Cunninghan, G.J. Romanes, Cunninghaḿs Manual of Practical Anatomy, 14th ed., London, OUP, 1976. [30] U.U. Ukoha, O.A. Eqwu, M.G.C. Ezeani, A.E. Anyabolu, O.C. Ejimofor, H.C. Nzeako, K.E. Umeasalugo, Estimation of Stature using footprints in an adult student population in Nigeria, Int. J. Biomed. Adv. Res. 4 (11) (2013) 827–833. [31] C.M. Wall-Scheffler, J. Wagnild, E. Wagler, Human footprint variation while performing load bearing tasks, PLoS ONE 10 (3) (2015) e0118619. [32] J. Bertuit, C. Leyh, M. Rooze, V. Feipel, Plantar pressure during gait in pregnant women, J. Am. Podiatr. Med. Assoc. 106 (6) (2016) 398–405. [33] A.R. Bird, H.B. Menz, C.C. Hyde, The effect of pregnancy on footprint parameters. A prospective investigation, J. Am. Podiatr. Med. Assoc. 89 (8) (1999) 405–409. [34] A.M. Dowling, J.R. Steele, L.A. Baur, Does obesity influence foot structure and plantar pressure patterns in prepubescent children? Int. J. Obes. Relat. Metab. Disord. 25 (2010) 845–852. [35] D.L. Riddiford-Harland, J.R. Steele, L.H. Storlein, Does obesity influence foot structure in prepubescent children? Int. J. Obes. Relat. Metab. Disord. 24 (2000) 541–544. [36] J.C. Gilmour, Y. Burns, The measurement of the medial longitudinal arch in children, Foot Ankle Int. 22 (2001) 493–498. [37] T.R. Aurichio, J.R. Rebelatto, A.P. de Castro, The relationship between the body mass index (BMI) and foot posture in elderly people, Arch. Gerontol. Geriatr. 52 (2) (2011) e89–e92. [38] K.J. Mickle, J.R. Steele, B.J. Munro, The feet of overweight and obese young children: are they flat or fat? Obesity 14 (11) (2006) 1949–1953. [39] R. Rushton, H. Steinberg, C. Tinson, Effects of a single experience on subsequent reactions to drugs, Br. J. Pharmacol. 20 (1963) 99–105.
References [1] K. Krishan, A. Sharma, Estimation of stature from dimensions of hands and feet in a North Indian population, J. Forensic Leg. Med. 14 (6) (2007) 327–332. [2] S. Sanli, E. Kizlikanat, N. Boyan, E. Ozsahin, M. Bozkir, R. Soames, H. Erol, O. Oguz, Stature estimation based on hand length and foot length, Clin. Anat. 18 (8) (2005) 589–596. [3] G. Zaybek, I. Ergur, Z. Demiroglu, Stature and gender estimation using foot measurements, Forensic Sci. Int. 181 (1–3) (2008) 54.e1–54.e5. [4] S. Reel, S. Rouse, W. Vernon, P. Doherty, Estimation of stature from static and dynamic footprints, Forensic Sci. Int. 219:1-3 (2012) 283.e1-283e.5. [5] L. Robbins, Estimating height and weight from size of footprints, J. Forensic Sci. 31 (1) (1986) 143–152. [6] K. Krishan, Establishing correlation of footprints with body weight-forensic aspects, Forensic Sci. Int. 179 (1) (2008) 63–69. [7] K. Krishan, Estimation of stature from footprint and the foot outline dimensions in Gujjars of North India, Forensic Sci. Int. 175 (2–3) (2008) 93–101. [8] T.N. Moorthy, A.M.B. Mostapa, R. Boominaathan, R. Raman, Stature estimation from footprint measurements in Indian Tamils by regression analysis, Egypt J Forensic Sci 4 (2013) 7–16. [9] J. Kotzerke, A. Arakala, S. Davis, K. Horadam, J. McVernon, Ballprints as an infant biometric: A first approach, in: A. Jain, M. Tistarelli, P. Campisi, Vincenzo P. (Eds.), Proceedings of the 5th IEEE Workshop on Biometric Measurements and Systems for Security and Medical Applications (BIOMS) 2014, Rome, Italy, 17 October 2014, pp. 36–43. [10] W. Vernon, Identification from podiatry and walking, the foot, in: T. Thompson (Ed.), Forensic Human Identification: An Introduction, CRC Press, New York, 2007. [11] I.A. Fawzy, N.N. Kamal, Stature and body weight estimation from various footprints measurements among Egyptian population, J. Forensic Sci. 55 (4) (2010) 884–888. [12] P. Uhrová, R. Beňuš, S. Masnicová, Stature estimation from various foot dimensions among Slovak population, J. Forensic Sci. 58 (2) (2013) 448–451. [13] P. Uhrová, R. Beňuš, S. Masnicová, Z. Obertová, D. Kramárová, K. Kyselcová, M. Dornhoferová, S. Bodoríkova, E. Neščáková, Estimation of stature using hand and foot dimension in Slovak adults, Leg. Med. 17 (2) (2015) 92–97. [14] K. Krishan, K. Vij, Diurnal variation of stature in three adults and one child, Anthropologist 9 (2) (2007) 113–117. [15] A.R. Tyrrell, T. Reilly, J.D. Troup, Circadian variation in stature and the effects of spinal loading, Spine 10 (2) (1985) 161–164. [16] K. Krishan, T. Kanchan, A. Sharma, Multiplication factor versus regression analysis in stature estimation from hand and foot dimensions, J. Forensic Leg. Med. 19 (4) (2012) 211–214. [17] N. Hemy, A. Flavel, N. Ishak, D. Franklin, Estimation of stature using anthropometry of feet and footprints in Western Australian population, J. Forensic Leg. Med. 20 (5) (2013) 435–441. [18] M.A.K. Hairunnisa, T.N. Moorthy, Stature estimation from the anthropometric measurements of footprint in Iban ethnics of east malaysia by regression analysis, J.
16
Contents lists available at ScienceDirect
Legal Medicine journal homepage: www.elsevier.com/locate/legalmed
Estimation of stature and body weight in Slovak adults using static footprints: A preliminary study
T
⁎
Zuzana Caplovaa, Petra Švábováa, , Mária Fuchsováb, Soňa Masnicovác, Eva Neščákováa, Silvia Bodorikováa, Michaela Dörnhöferováa, Radoslav Beňuša a
Department of Anthropology, Faculty of Natural Sciences, Comenius University, Mlynská dolina, 84215 Bratislava, Slovak Republic Department of Didactics of Natural Sciences in Primary Education, Faculty of Education, Comenius University, Račianska 59, 81334 Bratislava, Slovak Republic c Department of Criminalistics and Forensic Sciences, Academy of Police Forces, Sklabinská 1, 83517 Bratislava, Slovak Republic b
A R T I C LE I N FO
A B S T R A C T
Keywords: Forensic podiatry Diagonal axes Stature prediction Body weight prediction Regression equations
The stature and the body weight as part of the biological profile can aid the personal identification. The dimensions of the human foot, as well as the footprint, can be used for the prediction due to the existence of its positive correlation with the stature and body weight. Five diagonal axes and ball breadth of bilateral static footprints of 132 young Slovak adults were obtained. All diameters were larger in a male group than female group. No bilateral differences were found except the first diagonal axis and ball breadth. A positive correlations between the selected footprint diameters with the stature (r = 0.37–0.64) and the body weight (r = 0.29–0.71) were confirmed. The linear and multiple regression prediction equations were developed. A stature prediction equation using the most lateral diameters (the fourth and fifth diagonal axis) exhibited the highest accuracy ranging from 4 to 7.5 cm. Similar results were found for the body weight estimation of the male and mixed group. In the female group, the most medial axis (first and second) exhibited the highest accuracy. The body weight estimation accuracy ranges from 9.09 to 11.09 kg. The real and predicted stature and body weight were compared and found differences were lower than calculated SEEs. Thresholds and prediction trend of under- or overestimation was identified. The results of the present study show that selected measurements of static footprints could be used to predict stature and body weight but should be applied only for Slovaks due to population specificity.
1. Introduction The body height and body weight are the basic characteristics often used to describe another person and they are part of a biological profile. Together with the sex and age, if estimated correctly, may aid a personal identification. Multiple publications exploring mainly the stature estimation from a dimension of the various body parts can be found, among which the lower extremities show high correlations [1,2,3]. Similarly to the direct foot measurements, the footprints attracted attention and can be used for the predictions [4–5]. An extensive research and development of the stature prediction equations from the footprints were conducted in the countries where barefoot walking is common, for instance in India. On the other hand, only a few studies focus on the estimation of the body weight from the footprints [4–8]. Footprints are found in the majority of crime scenes but may be overlooked or underestimated [6]. Considering their individuality, the stature and body weight footprint estimations could be particularly
⁎
useful in the case of home violence. Alike the individual crease pattern of footprints [9], specific length and width measurements highlight their uniqueness [4,6–7]. Several quantitative footprint evaluation methods were developed utilizing the diagonal, parallel and horizontal measurements of the heel, ball, and toes or their combinations [10]. The positive correlations of the stature/body weight and various measurements of the footprint have been observed, but the population differences preclude the development of universal prediction equations [5–8,11]. Previously published literature confirmed the existence of the strong correlation between direct foot measurements and the stature of Slovak adults [12,13]. The presented research uses the diagonal footprint measurements based on Robbins [5] (five axes connecting the backmost point on the heel with the foremost point of each toe and breadth of the ball) to assess the footprint-stature and footprint-body weight correlations. The publication aims towards the development of the populationspecific stature and body weight prediction equations, as they have not
Corresponding author. E-mail address: [email protected] (P. Švábová).
https://doi.org/10.1016/j.legalmed.2018.07.002 Received 26 April 2018; Received in revised form 20 July 2018; Accepted 25 July 2018 Available online 25 July 2018 1344-6223/ © 2018 Published by Elsevier B.V.
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
been developed for Slovak adults yet.
2. Materials and methods The body weight and the stature of 132 young adults were measured, and a series of footprints were collected. The data collection was part of an anthropometric research project of young Slovak population conducted at the Faculty of Natural Sciences of Comenius University in Bratislava, Slovakia. All the participants received written instructions and details of the study explaining its aims, methods and expected involvement. Participation in the study was voluntarily and entirely based on a written informed consent, and each participant was ensured that he/she could withdraw from the study. All participants were measured approximately at the same time in the morning to avoid the diurnal variation in the stature [14,15]. Stature was measured using an anthropometer from the vertex to the floor in the anatomical position with the head oriented in the Frankfurt Plane by a single operator according to recommendations by [5]. The body weight was recorded using an analog body weight scale with a precision of 0.1 kg. The collection of the bilateral static footprint was part of the gait pattern and dynamic footprint collection. Each participant applied an even layer of non-coloring cream on both feet. Participants were asked to stand up on a prepared recycled paper (5 m in length and 1 m in width). The first set of the static bilateral footprint was obtained. Then participants walked naturally towards the end of the prepared paper and stopped with a left and right foot next to each other. The set of dynamic footprints reflecting the gait pattern and the second set of static footprints were obtained. Each footprint was immediately traced by a pencil for better preservation. Only bilateral static footprints were used for this study. Static (standing) footprints of 64 females (from 18 to 33 years old, with a mean age of 21.07 years old) and 68 males (from 18 to 25 years old, with a mean age of 20.59 years old) were collected. Six measurements comprising of five length diagonal dimensions and breadth of the ball were measured by a single operator. Firstly, the backmost point of the heel was identified (Pternion). Five diagonal axes connecting Pternion with the most anterior point of each toe was drawn according to [5]. The most lateral and medial points of the ball were used to obtain the ball breadth dimension (Fig. 1). The left footprint measurements were designed as DA_1L, DA_2L, DA_3L, DA_4L, DA_5L, and Br_L. The right footprint measurements were designed as DA_1R, DA_2R, DA_3R, DA_4R, DA_5R, and Br_R (Fig. 1).
Fig. 1. Five diagonal axes of the footprint. p – the backmost point of the heel; DA_1 - DA_5 – diagonal axes, Br – ball breadth.
3. Results Descriptive statistical analysis (mean, standard deviation, minimum, maximum) of stature, body weight, breadth of the ball and all diagonal axes of static footprints are shown in Table 1. Mean values of all measurements were found to be significantly larger in males than in females (p-value < 0.05). Statistically significant bilateral differences (p-value < 0.05) were only observed for the breadth of the ball and the diagonal axis of the first toe of females and males (Table 1). The results of the analysis of covariance confirmed the influence of sex on the relationship between stature and diagonal axes of static footprints (p-value < 0.05). Therefore, the results of the regression analysis are presented separately by sex as well as by side since it is possible to visually distinguish laterality of a footprint. The mixed group (males + females) is presented as well.
2.1. Statistic evaluation Data were statistically evaluated in SPSS Statistics 19.0. ShapiroWilk test confirmed a normal distribution of data (p-value > 0.05). Intersexual differences were tested by two-sample t-test and bilateral differences by paired t-test. Pearson correlation coefficient was employed to assess the stature-measurements and body weight-measurements correlations. An analysis of covariance (using GLM, general linear model) was done to test whether sex has an impact on the relationship between diagonal axes and body weight or stature. Prediction equations for stature and body weight were derived by linear regression analysis and multiple regression analyses [16]. Developed prediction equations were tested. Body weight and stature were calculated using every linear prediction equation in male, female and mixed group. Statistical differences between the measured and calculated data were verified by a two-sample t-test. Differences between the measured and calculated stature and body weight were calculated and the Pearson correlation coefficient was used to assess the relationship between the calculated difference and the stature or body weight and BMI.
3.1. Stature Correlation coefficients for the stature and static footprint measurements are shown in Table 2. All diagonal axes (p-value < 0.01) and ball breadth (p-value < 0.05) exhibited statistically significant correlation with the stature. In males, the highest correlation coefficient was between the stature and the fourth diagonal axis of the left footprint (r = 0.64). The lowest correlation coefficient was observed with the ball breadth of the right footprint (r = 0.37). In females, the fifth diagonal axis of the left footprint resulted in the highest correlation coefficient (r = 0.69), while the ball breadth of the left footprint showed the lowest correlation with the stature (r = 0.27). In the mixed group, the highest correlation coefficient was between the stature and the fourth diagonal axis of the left footprint (r = 0.84), and the lowest with the ball breadth of the left footprint (r = 0.60). Table 2 shows that 8
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Table 1 Descriptive statistics of stature (cm), body weight (kg) and selected measurements of static footprints (cm) and their bilateral differences (p-value). Females n = 64
Stature Weight DA_1 L DA_1 R DA_2 L DA_2 R DA_3 L DA_3 R DA_4 L DA_4 R DA_5 L DA_5 R Br_L Br_R
Bilateral differences
Males n = 68
Bilateral differences
Mean
SD
Min
Max
Mean
SD
Min
Max
173.2 58.16 23.54 23.24 22.97 22.75 21.96 21.91 20.71 20.72 19.27 19.19 8.79 8.58
9.55 8.41 1.26 1.23 1.34 1.25 1.28 1.20 1.12 1.17 1.10 1.06 0.65 0.66
152.1 45.00 21.00 20.80 19.80 19.70 19.20 19.20 18.30 18.00 16.80 17.10 7.50 6.60
190.7 80.00 26.80 26.80 26.60 25.80 25.50 25.00 23.70 23.80 21.70 22.00 10.90 10.30
173.7 78.27 26.15 25.82 25.64 25.47 24.49 24.48 23.23 23.15 21.51 21.58 9.85 9.66
9.55 12.89 1.43 1.60 1.43 1.64 1.36 1.66 1.25 1.66 1.23 1.46 0.77 0.80
161 57.00 23.50 22.90 22.40 22.10 21.30 21.40 20.80 20.40 19.30 19.00 8.30 8.20
198.4 115.0 29.60 30.70 29.30 31.20 28.00 29.90 26.70 28.20 24.50 25.50 11.50 11.30
0.004* 0.060 0.456 0.983 0.482 *
0.004
0.006* 0.113 0.615 0.174 0.926 0.020*
DA – diagonal axis, 1–5 – first to the fifth toe, R – right, L – left, Br – ball breadth, * – significance at < 0.05. Table 2 Correlation between stature and measurements of static footprints. Females n = 64
Males n = 68
Mixed group n = 132
Left DA_1 DA_2 DA_3 DA_4 DA_5 Br
R 0.65*** 0.65*** 0.63*** 0.65*** 0.69*** 0.19**
r 0.61*** 0.62*** 0.63*** 0.64*** 0.59*** 0.33***
r 0.82*** 0.82*** 0.82*** 0.84*** 0.82*** 0.58***
Right DA_1 DA_2 DA_3 DA_4 DA_5 Br
0.56*** 0.57*** 0.62*** 0.61*** 0.64*** 0.22***
0.49*** 0.54** 0.54*** 0.55*** 0.62*** 0.35***
0.76*** 0.78*** 0.78*** 0.77*** 0.81*** 0.60***
Table 3 Linear regression equations for stature estimation (cm) from measurements of the static footprint. Females n = 64 Left
R2
SEE
right
R2
SEE
96.58 + 2.95 DA_1 101.88 + 2.79 DA_2 103.27 + 2.85 DA_3 96.94 + 3.33 DA_4 96.32 + 3.60 DA_5 144.91 + 2.47 Br
0.42 0.42 0.40 0.42 0.48 0.07
4.44 4.46 4.50 4.43 4.14 5.85
104.48 + 2.64 DA_1 106.34 + 2.61 DA_2 100.57 + 2.98 DA_3 103.67 + 3.00 DA_4 98.89 + 3.49 DA_5 141.01 + 2.96 Br
0.32 0.32 0.39 0.37 0.41 0.11
4.80 4.76 4.53 4.58 4.48 5.70
Left
R2
SEE
right
R2
SEE
101.87 + 2.99 DA_1 101.57 + 3.06 DA_2 99.26 + 3.30 DA_3 95.15 + 3.66 DA_4 106.28 + 3.43 DA_5 144.84 + 3.06 Br
0.37 0.38 0.40 0.41 0.35 0.14
5.69 5.61 5.56 5.47 5.86 6.52
123.54 + 2.19 120.54 + 2.34 123.16 + 2.33 125.34 + 2.36 113.82 + 3.06 149.80 + 3.15
0.24 0.30 0.30 0.30 0.39 0.12
6.21 6.00 6.03 5.99 5.70 6.60
Left
R2
SEE
right
R2
SEE
69.34 + 4.18 DA_1 73.81 + 4.09 DA_2 73.17 + 4.30 DA_3 71.15 + 4.64 DA_4 72.64 + 4.90 DA_5 110.66 + 6.70 Br
0.67 0.67 0.67 0.70 0.67 0.35
5.59 5.55 5.53 5.29 5.60 7.54
80.35 + 3.78 DA_1 82.49 + 3.76 DA_2 82.84 + 3.89 DA_3 86.31 + 3.95 DA_4 81.57 + 4.48 DA_5 112.98 + 6.56 Br
0.57 0.60 0.61 0.60 0.66 0.36
6.31 6.10 6.09 6.16 5.60 7.51
Males n = 68
DA – diagonal axis, Br – ball breadth, 1–5 – first to fifth toe, r – correlation coefficient, ** – significant at < 0.01, *** – significant at < 0.001.
correlation coefficients are higher in the mixed group (0.60–0.84) when compared to the male (0.37–0.64) or female group (0.27–0.69). Table 3 presents the linear regression equations for stature estimation together with the values of coefficients of determination (R2) and standard errors of estimates (SEE) representing the standard deviation of the stature estimation. The SEE ranges from ± 4.1 cm to ± 5.9 cm in the female group, from ± 5.4 cm to ± 6.6 cm in the male group, and from ± 5.3 cm to ± 7.5 cm in a mixed group. Regression parameters are statistically significant for all the measurements (p-value < 0.05) except for the ball breadth of the left and right footprint in the female group, and for the ball breadth of the left footprint in the male group. Table 4 illustrates multiple regression equations for the stature estimation based on various combinations of diagonal axes and ball breadth measurements for the left and right footprint. For the practical use, only multiple regression equations for the mixed group were calculated as it is not possible to visually estimate the sex from the footprint. The first and the fifth diagonal axis of the left footprint showed the lowest SEE ± 5.33 cm. For the right footprint, the combination of the fifth diagonal axis and the ball breadth showed the lowest SEE ± 5.45 cm.
DA_1 DA_2 DA_3 DA_4 DA_5 Br
Mixed Group n = 132
DA – diagonal axis, 1–5 – first to the fifth toe, Br – ball breadth, R2 – coefficient of determination, SEE – standard error of estimate.
show statistically significant correlation with the body weight (pvalue < 0.01). The highest correlation coefficient in the male group can be seen between the body weight and the fifth diagonal axis of the right footprint (r = 0.50), and the lowest between the body weight and the first diagonal axis of the right footprint (r = 0.40). In the female group, the second diagonal axis of the left footprint resulted in the highest correlation coefficient (r = 0.46), while the fifth diagonal axes of the right footprint showed the lowest correlation with the body weight (r = 0.29). In the mixed group, the highest correlation coefficient was between the body weight and the second, fourth and fifth diagonal axis of the left footprint (r = 0.71), and the lowest with the ball breadth of the right footprint (r = 0.65). The highest correlation coefficients for the sex-mixed group are shown in Table 5.
3.2. Body weight Correlation coefficients for the body weight and static footprint measurements are shown in Table 5. All diagonal axes and ball breadth 9
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Table 4 Multiple regression models for stature estimation (cm) from footprint measurements in a mixed group.
Table 6 Linear regression equations for body weight estimation (kg) from measurements of the static footprint.
Left
R2
SEE
Females n = 64
68.73 + 4.01 DA_1L + 0.51 Br 70.22 + 4.51 DA_5L + 1.15 Br 65.39 + 2.24 DA_1L + 2.54 DA_5 65.03 + 2.53 DA_5 + 2.15 DA_1 + 0.31 Br
0.67 0.68 0.70 0.70
5.60 5.56 5.33 5.35
Left
R2
SEE
Right
R2
SEE
Right 76.46 + 3.12 DA_1 + 2.20 Br 76.93 + 3.81 DA_1 + 2.01 Br 78.94 + 3.70 DA_5 + 0.75 DA_1 75.71 + 3.42 DA_5 + 1.92 Br + 0.41 DA_1
R2 0.60 0.68 0.67 0.68
SEE 6.14 5.45 5.60 5.46
−41.76 + 4.29 DA_1 −35.55 + 4.29 DA_2 −16.56 + 3.45 DA_3 −24.97 + 4.06 DS_4 −22.42 + 4.23 DA_5 −3.17 + 7.08 Br
0.24 0.25 0.16 0.17 0.18 0.17
9.65 9.68 10.15 10.09 10.15 10.05
−33.69 + 3.99 DA_1 −32.66 + 4.04 DA_2 −24.55 + 3.82 DA_3 −25.39 + 4.08 DA_4 −14.26 + 3.82 DA_5 −2.66 + 7.21 Br
0.19 0.21 0.17 0.19 0.13 0.18
9.92 9.94 10.16 10.07 10.46 9.98
Left
R2
SEE
Right
R2
SEE
−26.51 + 3.99 DA_1 −27.76 + 4.12 DA_2 −30.09 + 4.4 DA_3 −36.74 + 4.93 DA_4 −31.21 + 5.07 DA_5 4.90 + 7.41 Br
0.19 0.20 0.20 0.22 0.23 0.19
12.04 11.96 11.91 11.82 11.66 12.05
−11.57 + 3.46 −17.59 + 3.75 −12.49 + 3.69 −14.62 + 3.99 −24.03 + 4.72 7.57 + 7.27 Br
0.17 0.22 0.21 0.25 0.26 0.19
12.13 11.82 11.94 11.68 11.76 12.01
Left
R2
SEE
Right
R2
SEE
−71.02 + 5.62 DA_1 −65.71 + 5.52 DA_2 −61.65 + 5.6 DA_3 −64.81 + 6.06 DA_4 −63.86 + 6.49 DA_5 −34.11 + 11.02 Br
0.47 0.48 0.45 0.47 0.47 0.41
11.26 11.2 11.46 11.22 11.3 11.93
−61.36 + 5.30 DA_1 −58.35 + 5.26 DA_2 −54.51 + 5.31 DA_3 −54.45 + 5.56 DA_4 −55.00 + 6.06 DA_5 −30.18 + 10.83 Br
0.44 0.47 0.45 0.46 0.46 0.41
11.58 11.3 11.57 11.38 11.47 11.84
Males n = 68
DA – diagonal axis, Br – ball breadth, 1–5 – first to the fifth toe, R2 – coefficient of determination; SEE – standard error of estimate. Table 5 Correlation between body weight and measurements of static footprints. Females n = 64
Males n = 68
Mixed group n = 132
Left DA_1 DA_2 DA_3 DA_4 DA_5 Br
r 0.42** 0.46** 0.40** 0.38** 0.37** 0.44**
r 0.41** 0.44** 0.44** 0.45** 0.49** 0.42**
R 0.69** 0.71** 0.69** 0.71** 0.71** 0.66**
Right DA_1 DA_2 DA_3 DA_4 DA_5 Br
0.35** 0.41** 0.40** 0.41** 0.29** 0.42**
0.40** 0.45** 0.45** 0.48** 0.50** 0.41**
0.67** 0.70** 0.69** 0.70** 0.70** 0.65**
DA_1 DA_2 DA_3 DA_4 DA_5
Mixed group n = 132
DA – diagonal axis, Br– ball breadth, 1–5 – first to the fifth toe, R2 – coefficient of determination, SEE – standard error of estimate Table 7 Multiple regression models for body weight estimation (kg) from footprint measurements in a mixed group.
DA – diagonal axis, Br – ball breadth, 1–5 – first to the fifth toe, r – correlation coefficient, ** – significant at < 0.01.
The regression equations for the body weight estimation, standard errors of estimates (SEE) and values of coefficients of determination (R2) are listed in Table 6. The SEE ranges from ± 9.65 kg to ± 10.46 kg in the female group, from ± 11.66 kg to ± 12.13 kg in the male group, and from ± 11.20 kg to ± 11.90 kg in the mixed group. Regression parameters are statistically significant for all the measurements (p-value < 0.05). The results of the multiple regression analysis for the body weight estimation based on various combinations of diagonal axes and ball breadth measurements for the left and right footprint are presented in Table 7. As in the case of stature estimation, only multiple regression equations for the mixed group were calculated. The second, fifth diagonal axis and ball breadth of the left footprint showed the lowest SEE ± 10.64 kg. For the right footprint, three combinations showed the lowest SEE ± 10.57 kg, namely the first and second diagonal axis with the ball breadth; the second and third diagonal axis with the ball breadth; and the second, third and fourth diagonal axis with the ball breadth.
Left
R2
SEE
−77.57 + 5.3 Br + 3.85 DA_1 −78.62 + 4.96 Br + 2.10 DA_1 + 2.0 0 DA_2 −82.11 + 4.36 Br + 1.90 DA_1 + 1.74 DA_2 − 3.50 DA_3 + 3.70 DA_4 + 0.99 DA_5 −77.76 + 2.33 DA_2 + 2.23 DA_5 + 4.74 Br
0.52 0.53 0.54
10.75 10.74 10.65
0.53
10.64
Right −70.97 + 5.99 Br + 0.38 DA_1 + 3.14 DA_2 −71.28 + 3.00 DA_2 + 0.46 DA_3 + 6.25 Br −69.95 + 2.56 DA_2 − 2.65 DA_3 + 3.72 DA_4 + 6.09 Br −70.26 + 1.95 DA_4 + 1.92 DA_5 + 6.26 Br
R2 0.53 0.54 0.55 0.55
SEE 10.57 10.57 10.57 10.58
DA – diagonal axis, Br – ball breadth, 1–5 – first to the fifth toe, R2 – coefficient of determination, SEE – standard error of estimate.
lower in the mixed group, and are ranging from r = −0.55 to r = −0.79 (Table 8). Average differences between the calculated and real body weight for each group and measurement are listed in Table 9 together with its Pearson correlation coefficient (r) with the body weight. Additionally, the correlation with Body Mass Index (BMI) was calculated to better understand the relationship between the performance of the body weight prediction equations and the relative body size. The differences between the calculated and real body weight are lower than SEE values for each prediction equation listed in Table 6. Similarly, as in case of stature, all observed differences exhibited statistically significant negative correlation with the body weight (p-value < 0.01) and with BMI (p-value < 0.01). Correlation with the body weight ranges from r = −0.86 to r = −0.93 in the female group, r = −0.86 to r = −0.91 in the male group and r = −0.53 to r = −0.74 in the mixed group (Table 9). In the case of BMI, the correlation ranges from r = −0.84 to
3.3. Application Average differences between the calculated and real stature for each group and measurement are listed in Table 8 together with its Pearson correlation coefficient (r) with the stature. The differences between the calculated and real stature are lower than SEE values for each prediction equation listed in Table 3. All differences exhibited statistically significant negative correlation with the stature (p < 0.01). The correlation coefficient ranges from r = −0.72 to r = −0.96 in the female group and r = −0.77 to r = −0.94 in the male group. Correlations are 10
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
predicted body weight was calculated for the participants with low real body weight. The largest negative difference (underestimation) was calculated for the participants with the highest real body weight. The predicted body weight of females under 53 kg (or BMI 20.5) was always overestimated (positive difference between the estimated and the real body weight), and over 65 kg (or BMI 24) always underestimated (Fig. 5). In the male group, the predicted body weight of males weighing under 70 kg (or BMI 20.9) was always overestimated, and over 86 kg (or BMI 26.5) underestimated (Fig. 6). In the mixed group, the predicted weight of individual weighting under 52.5 kg (or BMI 18.3) was always overestimated and over 86 kg (or BMI 26.5) always underestimated (Fig. 7). More specifically, unisex equations overestimated the body weight of females under 52.5 kg (or BMI 18.3) and males under 60 kg (or BMI 20.8) and underestimated females over 65 kg (or BMI 24) and males over 86 kg (or BMI 26.5). Similarly, as in the stature prediction, both positive and negative errors were calculated for the participants with body weight between the thresholds.
Table 8 Average differences (cm) between the real and calculated stature and Pearson correlation coefficient (r) between the stature and average differences. Females n = 64
Males n = 68
Mixed group n = 132
Average difference
r
Average difference
r
Average difference
r
Br_L DA_1L DA_2L DA_3L DA_4L DA_5L
± 4.69 ± 3.38 ± 3.63 ± 3.57 ± 3.58 ± 3.16
−0.96** −0.76** −0.76** −0.77** −0.76** −0.72**
± 6.68 ± 4.48 ± 4.29 ± 4.32 ± 4.29 ± 4.46
−0.94** −0.80** −0.80** −0.78** −0.77** −0.81**
± 6.28 ± 4.49 ± 4.39 ± 4.31 ± 4.12 ± 4.20
−0.79** −0.58** −0.57** −0.57** −0.55** −0.58**
Br_R DA_1R DA_2R DA_3R DA_4R DA_5R
± 4.68 ± 3.50 ± 3.74 ± 3.68 ± 3.64 ± 3.53
−0.94** −0.83** −0.82** −0.78** −0.79** −0.77**
± 5.26 ± 4.88 ± 4.98 ± 4.57 ± 4.60 ± 4.27
−0.94** −0.87** −0.84** −0.84** −0.84** −0.79**
± 6.30 ± 4.94 ± 4.80 ± 4.79 ± 4.77 ± 4.18
−0.78** −0.65** −0.63** −0.63** −0.64** −0.58**
DA – diagonal axis, 1–5 – first to the fifth toe, R – right, L – left, Br – ball breadth, r – Person correlation coefficient, ** – significant at < 0.01.
4. Discussion
r = −0.94 in the female group, r = −0.83 to r = −0.91 in the male group, and from r = −0.54 to r = −0.89 in the mixed group. Moreover, the trend in a stature prediction can be identified for every footprint measurement in the female, male and mixed group. The largest positive error (overestimation) in the stature prediction can be seen in case of participants with the lowest real stature. When predicting the stature of the higher person, the negative error (underestimation) occurred. Similarly, the largest underestimation was observed in case of participants with the highest real stature. Figs. 2–4 demonstrate the trend of under/overestimation of the body stature. The average of all calculated differences (stature estimation from 10 diagonal axes and 2 ball breadths) shows that in the case of females under 164.4 cm in height the developed prediction equations only overestimated the stature, and over 171.2 cm only underestimated (Fig. 2). Likewise in the male group (Fig. 3), the only overestimation can be seen under the height of 176.1 cm and only underestimation over the height of 186.0 cm. The predicted stature between two thresholds resulted in both positive and negative errors. When unisex equations were applied to the mixed group, predicted equation only overestimated females under 164.7 cm and males under 169.9 cm, and only underestimated females over 171.2 cm and males over 186.0 cm (Fig. 4). Figs. 5–7 illustrate the similar trend in the body weight prediction. The largest positive difference (overestimation) between the real and
The results of the study showed the correlation of selected measurements of static footprints with the body weight and stature, and therefore the potential use for their estimation in the Slovak population. Similarly to the previous studies [5–7,17], developed prediction equations are population specific. All footprint dimensions of young Slovak adults are larger in comparison to the previous publications, which further highlights their population specificity. Differences were found between the obtained footprint measurements and measurements reported for Iban Ethnic of East Malaysia [18] and the Indian population [6–7]. In both cases, the measurements of diagonal axes of Slovak adults were larger. Differences in some measurements, namely ball breadth, and first diagonal axis, were found in comparison to the Australian population [17]. In this case, however, the ball breadth measurements in our sample were smaller. Differences were found in all measurements in comparison to Ghanaian [19] and Nigerian [20] population. All female dimensions reported by [19–20] were larger in comparison to our sample, and all male dimension were smaller. Likewise, the dimensions of ball breadth of females from Turkey were larger, and of males smaller than in our sample [21]. Significant intersexual differences were found and mean values of all measurements were significantly larger in the male group than in female (p-value < 0.05) which is in accordance with the previous studies [19–20,22]. All measurements were larger on the left footprints,
Table 9 Average differences (kg) between the real and calculated body weight and Pearson correlation coefficient (r) between the average differences and body weight and BMI. Females n = 64
Br_L DA_1L DA_2L DA_3L DA_4L DA_5L Br_R DA_1R DA_2R DA_3R DA_4R DA_5R
Males n = 68
Mixed group n = 132
Diff.
r BW
r BMI
Diff.
r BW
r BMI
Diff.
r BW
r BMI
± 6.66 ± 7.13 ± 8.23 ± 6.99 ± 7.10 ± 7.15 ± 6.79 ± 7.25 ± 7.06 ± 7.17 ± 7.13 ± 7.56
−0.91** −0.87** −0.86** −0.92** −0.91** −0.91** −0.90** −0.80** −0.89** −0.91** −0.90** −0.93**
−0.84** −0.91** −0.90** −0.93** −0.93** −0.94** −0.84** −0.91** −0.91** −0.93** −0.92** −0.94**
± 8.99 ± 9.51 ± 9.20 ± 8.98 ± 8.78 ± 8.67 ± 8.71 ± 9.01 ± 8.93 ± 8.89 ± 8.80 ± 8.50
−0.90** −0.90** −0.90** −0.89** −0.88** −0.88** −0.90** −0.91** −0.89** −0.89** −0.87** −0.86**
−0.83** −0.90** −0.90** −0.91** −0.90** −0.89** −0.83** −0.88** −0.88** −0.88** −0.87** −0.88**
± 10.86 ± 8.95 ± 8.91 ± 8.96 ± 8.35 ± 9.21 ± 11.86 ± 8.93 ± 8.51 ± 8.80 ± 9.32 ± 8.50
−0.56** −0.73** −0.63** −0.66** −0.73** −0.67** −0.53** −0.75** −0.73** −0.74** −0.65** −0.73**
−0.54** −0.85** −0.77** −0.87** −0.87** −0.74** −0.80** −0.84** −0.83** −0.84** −0.74** −0.89**
DA – diagonal axis, 1–5 – first to the fifth toe, R – right, L – left, Br – ball breadth, Diff. – average difference, r – Person correlation coefficient, ** – significant at < 0.01, BW – body weight, BMI – body mass index. 11
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Fig. 2. Correlation between stature (cm) and average calculated difference (cm) in the female group. a-axis – upper stature limit for overestimation, b-axis – lower stature limit for underestimation.
walking was previously reported [7,11,24] and causes greater muscle and bone development of the dominant leg to support the pressure, which results in the larger measurements. Static footprints often display missing print of the fifth toe, especially in non-athletes in comparison to athletes. Intrinsic muscles stabilizing the arch and helping to maintain toes on the ground are more
however, the differences were smaller than 3 mm and significant bilateral differences were found only for ball breadth and first diagonal axis. The intersexual and bilateral differences can be explained similarly as is in the case of direct foot measurements. Females are genetically shorter than males, and therefore their feet and footprints tend to be smaller [23]. The preference of the left leg for standing and
Fig. 3. Correlation between stature (cm) and average calculated difference (cm) in the male group. a-axis – upper stature limit for overestimation, b-axis – lower stature limit for underestimation. 12
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Fig. 4. Correlation between stature (cm) and average calculated difference (cm) in the mixed group. a-axis – upper stature limit for overestimation (females), c-axis – lower stature limit for underestimation (females), b-axis – upper stature limit for overestimation (males), d-axis – lower stature limit for underestimation (males), □ – males, ◆ – females.
and prediction equations were developed also for the mixed group (males and females) as it is not possible to visually estimate the sex from the footprint. Positive correlations of the stature with the direct foot measurements of young Slovak adults were previously described [12–13]. In our
active in the athlete group [25]. Fifth toe non-print of 14–27% [25] and 14–16% [26] of non-athletes were previously reported. In our study of non-athletic young Slovak adults, the fifth toe was missing in 7% of the right footprints and 3% of the left footprints. Despite the existence of the intersexual differences, the correlations
Fig. 5. Correlation between body weight (kg)/BMI and average calculated difference (kg) in the female group. a-axis – upper BMI limit for overestimation, b-axis – lower BMI limit for underestimation, c-axis – upper body weight limit for overestimation, d-axis – lower body weight limit for underestimation. 13
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
Fig. 6. Correlation between body weight (kg) and BMI and average calculated difference (kg) in the male group. a-axis – upper BMI limit for overestimation, b-axis – lower BMI limit for underestimation, c-axis – upper body weight limit for overestimation, d-axis – lower body weight limit for underestimation.
this study are lower than reported by [27]. Correlation coefficients in the mixed group (regardless of the sex) were higher, similarly as in [5,28]. This finding further supports the fact that prediction equations without sex indicators are more reliable, especially when in the real life it is not possible to visually estimate the sex from the footprint. Stature prediction equations using the fourth and fifth diagonal axis also exhibited the lowest SEE in both sex groups. The same results were stated by [11,4]. Higher correlations with the statue and higher accuracy of prediction equations of lateral diagonal axes (fourth and fifth) occur as the longitudinal arch of the foot is more solid and supportive in its lateral part. Strong ligaments, tendons, and muscles preserve its integrity. The medial part of the longitudinal arch is more flexible and
study, the positive correlations with selected measurements of the static footprints were confirmed. The first, fourth and fifth diagonal axis exhibited the highest correlation with the stature, while the ball breadth the lowest. Findings are in accordance with the previous studies [4–5,11,17], which reported higher correlation coefficients with length footprint measurements than with breadth footprint measurements. The measurements of the left footprints exhibit higher correlation then the right footprints. In general, correlations reported in this study are higher than those reported by [11] for males, by [4,22] for a mixed group, and by [22] for females. On the other hand, correlations of stature and selected footprint dimension in all three groups in the Nigerian population [20] were higher. Similarly, the positive correlations with the ball breadth reported in
Fig. 7. Correlation between body weight (kg) and BMI and average calculated difference (kg) in the mixed group. ◆ – females, ○ – females, ◊ – males, □ – males. 14
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
and sex-mixed) and for their averages for each participant. High negative correlation between the stature and the calculated difference and between the body weight and the calculated difference resulted in a uniform trend: lower the stature or body weight (or BMI) – higher the possibility of overestimation (and higher the overestimation itself), and higher the stature or body weight (or BMI) – higher the possibility of underestimation (and higher the underestimation itself). Moreover, the upper and lower limits of over- and underestimation were identified for the female, male and sex-mixed group. Standard errors of estimates (SEE) of developed prediction equation for stature and body weight offer error margin in a positive and negative direction. Observed prediction trend illustrates the most likely scenario of over/underestimation for females, males and an unknown sex person of a certain body height or body weight which could potentially lead to more reliable prediction. Reported trends are truly the preliminary results as developed prediction equations were tested on the same sample. However, no previous studies reported such a trend to our knowledge. Therefore, more research is needed in this field. The results of the presented study show that selected measurements of footprints could be used as predictive values for stature and body weight estimation in the Slovak population. Standard errors of estimates of developed equations are relatively high, especially in the case of body weight estimation. On the other hand, when applied the differences between the real and calculated stature/body weight are lower than reported SEEs. Stature with the accuracy up to 3.16 cm and body weight with the accuracy up to 6.60 kg can be obtained. Described prediction equations should only be used only for Slovak population and only for the dimension of static footprints. Further study of a larger sample is needed, similarly, the study of dynamic (walking) footprints has to be undertaken to assess the differences between the static and dynamic footprints. Despite the fact, that the presented methodology using non-coloring cream is not validated, it was chosen over the traditional ink method. The cream method had been used in the past to study the gait pattern of rats [39]. The non-coloring cream method did not require any further feet cleaning and using recycled paper to obtain cream footprints prevented any slippery surface compared to non-recycled paper. However, in a real-life scenario, the footprints left on the slippery surface by the wet substance are found – for instance on tiles or wooden floor. The major limitation of the non-coloring cream method is the hand-drawn outline of the footprint as this could potentially add an unknown quantity of human error to the statistical analysis. Therefore, a further study of the presented footprint collection method needs to be undertaken. Less non-contact fifth toe (maximum 7%) was observed in our sample of non-athletic participants which may also be due to the used method. Further comparison in this matter is also needed.
is influenced by genetic factors, body weight and age [29]. Contrarily, the previous studies of Nigerian population reported the first and third diagonal axes exhibited the strongest correlations with the stature [20,30]. Standard errors of estimates (SEE) are lower in the female group which indicates more reliable stature estimation in a comparison to the male or mixed group. The use of linear regression equations for stature estimation could be applied also in the case of incomplete footprints when the only limited number of measurements is obtainable. In a case of complete footprints, the multiple linear regression equations combining several footprint measurements resulted in lower SEEs. Multiple regression equations of the mixed group only were developed for the practical use. The combination of the first and fifth diagonal axis of the left footprint resulted in the lowest estimation error and therefore the most reliable stature estimation. In general, SEEs for the stature estimation calculated in this study were higher than reported by [7,18,20], and lower for ball breadth than in [27]. Only a few publications deal with the body weight estimation from the footprint measurements, despite the fact that their positive correlation was confirmed [5–6,11]. The correlation with the body weight is lower than with the stature. Body weight is easily altered and therefore the body weight estimation is less accurate in comparison to the stature. Enlarged length and width footprint dimensions and increased contact area were reported during loaded walking [31] but no longitudinal studies taking into consideration the weight gain and weight loss of the same individuals were conducted. Previous longitudinal and cross-sectional studies of pregnant women described characteristic changes of the plantar surface and the gait pattern as a compensatory mechanism to improve locomotor stability. Described gait and feet modifications in pregnant women could occur as a result of a mass gain, as similar gait characteristics were reported in overweight people [32–33]. Likewise, a number of publications report larger foot dimensions and increased arch index in obese children [34–36]. The structural foot changes, mainly the flat foot caused by the continuous bearing of extra weight can be seen from the early childhood through the adulthood [34,37,38]. Foot dimensions, and therefore the footprint measurements seem to increase to support the higher weight of the person similarly as is the case of higher stature [5,6]. In our study, the highest body weight correlations were found to be with the second, fourth and fifth diagonal axis of the static footprint. All reported correlation coefficients were lower than stature-measurements correlations. The measurements of the right footprints exhibit higher correlation with the body weight in oppose to the correlation with the stature. In comparison to the previously published studies including males only, the body weight-footprint measurements correlations of Slovak males are lower than for Indian [6] but similar to the correlations reported for males from Egypt [11]. The highest correlation coefficients were confirmed with the lowest SEE for linear regression equations using first and second diagonal axis in the female group, and fourth and fifth diagonal axis in the male and sex-mixed group. Differences between the female and male group may be caused by individuality in the enlargement of the feet with the higher body weight which is varying from anterio-posterior to mediolateral direction [6]. Similarly to the stature estimation, only multiple regression equations of the mixed group were developed for the practical use. The result of the multiple linear regression showed that a combination of the second, fifth diagonal axis and ball breadth of the left footprint in a mixed group exhibit the lowest SEE = ± 10.65 kg. Addition of the parameters decreases the SEE in the mixed group. The results of the application of the stature and body weight prediction equations allow us to draw interesting preliminary conclusions. Similar estimation trend can be seen for the linear regression of each parameter, each side (left and right) and each group (females, males,
5. Conclusion Selected measurements of static footprints were used to predict the stature and body weight of Slovak young adults. No previous studies in this field were conducted, and no forensic data for the stature and body weight estimation from footprint dimensions exist for Slovak population. To our knowledge, no similar studies were published in central Europe or in a similar ethno-language Slavic population. Significant intersexual differences were found in all parameters, and no bilateral differences were found between the measurements, except the ball breadth and the first diagonal axis. The high positive correlation was found between all the parameters and stature. As expected, the correlations with body weight were lower, but significant prediction equations were developed for the stature and body weight estimation. The body height prediction equations using the fourth and fifth diagonal axis exhibited the lowest SEE in both sex groups. The highest body weight correlation coefficients were confirmed with the lowest SEE for linear regression equations for the first and second diagonal axis in the 15
Legal Medicine 34 (2018) 7–16
Z. Caplova et al.
female group, and fourth and fifth diagonal axis in the male and sexmixed group. Stature and body weight prediction equations from static footprints have not been developed for Slovak population before. After their application, the trend in over/underestimation was identified.
Forensic Sci. Criminol. 2 (2) (2014) 201–208. [19] J.K. Abledu, G.K. Abledu, F.B. Offei, E.M. Antwi, Determination of sex from footprint dimensions in a ghanaian population, PLoS ONE 10 (10) (2015) e0139897. [20] E.A. Okubike, N.M. Ideabuchi, A.O. Olabiyi, M.E. Nandi, Stature estimation from footprint dimension in an adult Nigerian student population, J. Forensic Sci. Med. 4 (1) (2018) 7–17. [21] D. Atamturk, Estimation of sex from the dimensions of foot Footprints, and Shoe, Anthropol. Anz. 68 (1) (2010) 21–29. [22] T. Kanchan, K. Krishan, S. Shyamsundar, K.R. Aparna, S. Jaiswal, Analysis of footprint and its parts for stature estimation in Indian population, Foot 22 (3) (2012) 175–180. [23] B. Danborno, A. Elukpo, Sexual dimorphism in hand and foot length, indices, stature-ratio and relationship to height in nigerians, Internet J. Forensic Sci. 3 (1) (2008) 379–383. [24] N.G. Rao, M.S. Kotian, Footprint ratio (FPR) – a clue for establishing sex identity, J. Ind. Acad. Forensic Med. 12 (1990) 51–56. [25] T. Kulthanan, S. Techakampuch, N. Donphongam, A study of footprints in athletes and non-athletic people, J. Med. Assoc. Thai. 87 (7) (2004) 788–793. [26] T.N. Moorthy, W.N.Z.W.M. Samsudin, M.S. Ismail, A study on footprints of malaysian athletes and non-athletes for application during forensic comparison, Malaysian J. Forensic Sci. 2 (1) (2011) 29–35. [27] T. Kanchan, K. Krishan, D. Geriani, I.S. Khan, Estimation of stature from the width of static footprints – insight into an Indian model, Foot 23 (4) (2013) 136–139. [28] T.N. Moorthy, Y.L. Ang, A.S. Saufee, F.N.H. Nik, Estimation of stature from footprint and foot outline measurements in Malaysian Chinese, Aust. J. Forensic Sci. 46 (2) (2013) 136–159. [29] D.J. Cunninghan, G.J. Romanes, Cunninghaḿs Manual of Practical Anatomy, 14th ed., London, OUP, 1976. [30] U.U. Ukoha, O.A. Eqwu, M.G.C. Ezeani, A.E. Anyabolu, O.C. Ejimofor, H.C. Nzeako, K.E. Umeasalugo, Estimation of Stature using footprints in an adult student population in Nigeria, Int. J. Biomed. Adv. Res. 4 (11) (2013) 827–833. [31] C.M. Wall-Scheffler, J. Wagnild, E. Wagler, Human footprint variation while performing load bearing tasks, PLoS ONE 10 (3) (2015) e0118619. [32] J. Bertuit, C. Leyh, M. Rooze, V. Feipel, Plantar pressure during gait in pregnant women, J. Am. Podiatr. Med. Assoc. 106 (6) (2016) 398–405. [33] A.R. Bird, H.B. Menz, C.C. Hyde, The effect of pregnancy on footprint parameters. A prospective investigation, J. Am. Podiatr. Med. Assoc. 89 (8) (1999) 405–409. [34] A.M. Dowling, J.R. Steele, L.A. Baur, Does obesity influence foot structure and plantar pressure patterns in prepubescent children? Int. J. Obes. Relat. Metab. Disord. 25 (2010) 845–852. [35] D.L. Riddiford-Harland, J.R. Steele, L.H. Storlein, Does obesity influence foot structure in prepubescent children? Int. J. Obes. Relat. Metab. Disord. 24 (2000) 541–544. [36] J.C. Gilmour, Y. Burns, The measurement of the medial longitudinal arch in children, Foot Ankle Int. 22 (2001) 493–498. [37] T.R. Aurichio, J.R. Rebelatto, A.P. de Castro, The relationship between the body mass index (BMI) and foot posture in elderly people, Arch. Gerontol. Geriatr. 52 (2) (2011) e89–e92. [38] K.J. Mickle, J.R. Steele, B.J. Munro, The feet of overweight and obese young children: are they flat or fat? Obesity 14 (11) (2006) 1949–1953. [39] R. Rushton, H. Steinberg, C. Tinson, Effects of a single experience on subsequent reactions to drugs, Br. J. Pharmacol. 20 (1963) 99–105.
References [1] K. Krishan, A. Sharma, Estimation of stature from dimensions of hands and feet in a North Indian population, J. Forensic Leg. Med. 14 (6) (2007) 327–332. [2] S. Sanli, E. Kizlikanat, N. Boyan, E. Ozsahin, M. Bozkir, R. Soames, H. Erol, O. Oguz, Stature estimation based on hand length and foot length, Clin. Anat. 18 (8) (2005) 589–596. [3] G. Zaybek, I. Ergur, Z. Demiroglu, Stature and gender estimation using foot measurements, Forensic Sci. Int. 181 (1–3) (2008) 54.e1–54.e5. [4] S. Reel, S. Rouse, W. Vernon, P. Doherty, Estimation of stature from static and dynamic footprints, Forensic Sci. Int. 219:1-3 (2012) 283.e1-283e.5. [5] L. Robbins, Estimating height and weight from size of footprints, J. Forensic Sci. 31 (1) (1986) 143–152. [6] K. Krishan, Establishing correlation of footprints with body weight-forensic aspects, Forensic Sci. Int. 179 (1) (2008) 63–69. [7] K. Krishan, Estimation of stature from footprint and the foot outline dimensions in Gujjars of North India, Forensic Sci. Int. 175 (2–3) (2008) 93–101. [8] T.N. Moorthy, A.M.B. Mostapa, R. Boominaathan, R. Raman, Stature estimation from footprint measurements in Indian Tamils by regression analysis, Egypt J Forensic Sci 4 (2013) 7–16. [9] J. Kotzerke, A. Arakala, S. Davis, K. Horadam, J. McVernon, Ballprints as an infant biometric: A first approach, in: A. Jain, M. Tistarelli, P. Campisi, Vincenzo P. (Eds.), Proceedings of the 5th IEEE Workshop on Biometric Measurements and Systems for Security and Medical Applications (BIOMS) 2014, Rome, Italy, 17 October 2014, pp. 36–43. [10] W. Vernon, Identification from podiatry and walking, the foot, in: T. Thompson (Ed.), Forensic Human Identification: An Introduction, CRC Press, New York, 2007. [11] I.A. Fawzy, N.N. Kamal, Stature and body weight estimation from various footprints measurements among Egyptian population, J. Forensic Sci. 55 (4) (2010) 884–888. [12] P. Uhrová, R. Beňuš, S. Masnicová, Stature estimation from various foot dimensions among Slovak population, J. Forensic Sci. 58 (2) (2013) 448–451. [13] P. Uhrová, R. Beňuš, S. Masnicová, Z. Obertová, D. Kramárová, K. Kyselcová, M. Dornhoferová, S. Bodoríkova, E. Neščáková, Estimation of stature using hand and foot dimension in Slovak adults, Leg. Med. 17 (2) (2015) 92–97. [14] K. Krishan, K. Vij, Diurnal variation of stature in three adults and one child, Anthropologist 9 (2) (2007) 113–117. [15] A.R. Tyrrell, T. Reilly, J.D. Troup, Circadian variation in stature and the effects of spinal loading, Spine 10 (2) (1985) 161–164. [16] K. Krishan, T. Kanchan, A. Sharma, Multiplication factor versus regression analysis in stature estimation from hand and foot dimensions, J. Forensic Leg. Med. 19 (4) (2012) 211–214. [17] N. Hemy, A. Flavel, N. Ishak, D. Franklin, Estimation of stature using anthropometry of feet and footprints in Western Australian population, J. Forensic Leg. Med. 20 (5) (2013) 435–441. [18] M.A.K. Hairunnisa, T.N. Moorthy, Stature estimation from the anthropometric measurements of footprint in Iban ethnics of east malaysia by regression analysis, J.
16