Lower Extremity Measurements in the Prediction of Body Height of the Igbos

Jervas E; Anele TI; Uloneme GC; Okeke CU; Iwuoha G; Eke CC; Osuchukwu IW

doi:10.17140/antpoj-1-104

Abstract

Body height prediction is one of the biological profiles used in the identification of individuals in cases of natural and man made disasters, and in crime investigations. In the present study, an attempt was made to derive regression formula to determine Body Height (BH) from lower extremity dimensions. 123 females and 88 males belonging to the Igbo population of Nigeria who gave informed consent were studied. BH was measured using standard protocols and measurements of the lower extremity segments were taken using measuring tapecutaneously. The mean observed BH of the studied population was 167.55±9.10 cm. In sexing the population, all the variables obtained were statistically significant, p<0.0001 except Thigh Length (TL). All the lower limb segments measured, showed positive significant correlation with body BH except Leg Length (LL) in the females. The multiple linear regression models generated using Lower Leg Length (LLL), LL and Trochanteric height (TROCH) in the overall population, and using LL and TL in the males performed better than the simple linear regression models because of the values of ‘R’ used in testing model adequacy. In females, LLL performed better than the other lower limb segments.

Introduction

The human body may be subject to fragmentations by both natural and manmade disasters. Such conditions have lead researcher to continue to estimate Body Height (BH) using available human body parts. The documented results are used by Scientist either in the identification exercise or for body height reconstruction.

In humans, estimation of BH is also applied to check for normal or abnormal growth 1-4 especially in the young individuals.1-4

The long bones of the body have been used to estimate 6-14 stature6-14 and results obtained were reliable.

In this study, we attempted to investigate which of the sub-segments of the lower extremity will predict BH best in the overall population as well as in the males and females.

Materials And Methods

Subjects were randomly selected from a pull of 211 persons and attention was paid on the measurements of some of the sub-segments of the lower extremity after obtaining informed consents.

Demographics

Exclusion criteria

Individual with limb length discrepancies and other musculo-skeletal disorder affecting body height were excluded from this study. Pregnant women were also excluded.

Inclusion criteria

Subjects from the Igbo ethnic states who were apparently healthy were included in the study.

Anthropometrics

Body height (BH): The height was measured to the nearest 0.1 cm using an Anthropometer with subjects standing without shoes with the heels held together, toes apart, and the head held in the Frankfort plane.

The following lower extremity measurements were taken (see illustration of landmarks, figure 1 below)

Trochanteric height (TROCH) was measured from the lateral bulging of the greater trochanteric protrusion to the floor.

Thigh length (TL) was measured as the distance between greater trochanter and the knee joint.

Leg Length (LL): is the distance from the lateral supracondyl of the tibia to the distal part of the lateral malleolus. All the

Data Presentation And Analysis

In this study, analysis of data was carried out using statistical package for social sciences (SPSS 17.0 software). In summarizing the data, the Minimum, Maximum, Mean and Standard deviations were estimated and presented. A comparison of difference of variable in females and males was also carried out. To test the relationship between BH and lower extremity dimensions, Pearson correlation was performed. The prediction function was derived through linear regression for each of the measurement with BH for the overall population, males and females separately. Finally the predicted/estimated values of BH were compared with that of observed/actual values.

Overall population Variables N Min Max Mean SD N Min Max Mean SD N Min Max Mean SD

BH(CM) 211 149.00 190.00 167.55 9.10 123 149.00 190.00 163.17 7.64 88 156.00 190.00 173.66 7.30

TROCH 198 55.90 107.00 89.93 6.38 117 55.90 104.00 88.65 6.48 81 71.00 107.00 91.79 5.762

TL 198 33.00 59.00 43.11 4.36 117 35.00 55.00 42.88 4.40 81 33.00 59.00 43.44 4.32

LL 198 33.00 56.00 46.55 4.14 117 33.00 55.00 45.48 3.49 81 36.00 56.00 48.10 4.53

LLL 198 31.00 57.00 42.98 4.01 117 31.00 51.00 42.00 3.77 81 34.00 57.00 44.41 3.94

Table 1. Descriptive statistics of the overall population, females and males, for lower extremity anthropometry (cm).

Variables (Cm)

Mean Std. Deviation Std. Error Mean Lower Upper

TROCH (F) - TRLOCH (M) -4.25641 6.55555 .74227 TL (F) - TL (M) -.83333 5.60593 .63475 LL (F) - LL (M) -3.14872 5.60980 .63518 LLL (F) - LLL(M) -2.92564 5.21438 .59041 BH: Body Height; TROCH: Trochanteric height; TL: Thigh Length; LL: Leg Length; LLL: Lower Leg Length. Variables are significant at p<0.05.

Table 2. Comparison of variables between females and males in lower extremity anthropometry.

Overall population Variables N Pearson Correlation N

TROCH 198 .454 .000 117 TL 198 .253 .000 117 LL 198 .365 .000 117 LLL 198 .500 .000 117 BH: Body Height, TROCH: Trochanteric height, TL: Thigh Length, LL: Leg Length, LLL: Lower Leg Length. *.Correlation is significant at the 0.05 level (2-tailed). **.Correlation is significant at the 0.01 level (2-tailed).

Table 3. Correlation coefficients between body height (Dependent variable) and lower extremities variables (Explanatory variables) in overall population, females and males.

Result

Table 1 shows the descriptive statistics of the overall population, females and males for the lower extremity anthropometry. The mean value for the dependent variable (BH) of the population under study was 167.55±9.10 while those of the explanatory variables are as in table 1.

Table 2 represents a comparison of difference of variables in females and males. It reveals that TROCH, TL, LL, LLL are highly significant in males than in females, P<0.0001, while TL is not significant.

The Pearson correlations (r) between the dependent variable and the explanatory variables for the overall population, females and males are given in table 3. In the overall population,

females males

Paired Differences Sig.

95% Confidence Interval of the Difference T Df (2-tailed)

-5.73446 -2.77836 -5.734 77 .000 -2.09727 .43061 -1.313 77 .193 -4.41353 -1.88390 -4.957 77 .000 -4.10130 -1.74998 -4.955 77 .000

Females Males Sig. Pearson Correla- Sig. Sig.

N Pearson Correlation (2-tailed) tion (2-tailed) (2-tailed)

** ** ** .328 .000 81 .515 .000 ** * ** .189 .041 81 .385 .000 ** ** .092 .322 81 .427 .000 ** ** ** .507 .000 81 .296 .007

all the four explanatory variables were significantly correlated with BH. The correlation between body height (YBH) and Lower Leg Length (XLLL), r=0.500 was better than that between YBH and trochanteric height (XLLLTROCH) r=0.454. The least significant correlation was observed in TL (r=0.253) in in the overall population.

The correlation between female’s body height (YFBH) and female’s lower leg length (XFLLL) i.e. r=0.507 was better than that between YFBH and females trochanteric height (XFTROCH), r=0.328; but the least significant correlation was observed in TL, r=0.189. LL in the females did not show any significant correlation with BH, as such LL could not be used to construct regression equations in the females. In the males, the correlation between males BH, (YMBH) and males trochanteric height (XMTROCH), r=0.515 was better than that betweenYMBH and males leg length (XMLL), r=0.481.

Females Variables Constant Regression 2 Variables Constant Regression 2

TROCH 128.336 .393 .107 .000 TROCH 114.567 .644 .265 .000 TL 148.858 .334 .036 .041 LLL 119.239 .166 .257 .000 Constant, Regression coefficient and Variation explained (R2) of lower extremity variables with BH (dependent) variables in females and

Table 5:

BH: Body Height; TROCH: Trochanteric height; TL: Thigh Length; LL: Leg Length; LLL: Lower Leg Length; R: Multiple correlation coefficient; R2: Coefficients of determinations; SEE: Standard error estimate.

Table 6:

Table 4 shows the constant, regression coefficient and the variation explained of lower extremity variables with the BH in the overall population. Twenty-Five percent of the variation in BH is inherent by lower leg length (LLL) while the combination of LLL, LL, and TROCH contributed 33.4% to the variation in the dependent variable.

The regression coefficients (see Table 5) for the variables used for BH prediction in the females and males were highly significant, indicating that they contributed positively for the prediction of BH in the females as well as in the males. The variation explained (R2×100) showed that it ranges from 3.6% to 25.76% in the females while it ranges from 8.7% to 31.6% in the males. The best prediction power was observed in LLL in female while it was the combination of TL and LL in the males.

Constant Regression Coefficient R p value

TROCH 108.969 .651 .206 .000 TL 144.632 .530 .064 .000 LL 130.030 .805 .133 .000 LLL 118.569 1.138 .250 .000 LLL 88.969 .265 .334 .015

LL, .405 .007 TROCH .833 .000 BH: Body Height; TROCH: Trochanteric height; TL: Thigh Length; LL: Leg Length; LLL: Lower Leg Length.

Constant, Regression coefficient and Variation explained (R2) of lower extremity

Table 4:

Males

R p value R p value Coefficient Coefficient

TL 145.809 .642 .148 .000 LL 140.980 .680 .182 .000 LLL 149.694 .540 .087 .007 LL, 115.859 .609 .316 .000 TL .652 .000

2 Variables R R Adjusted R SEE

TROCH .454 .206 .202 8.165 TL .253 .064 .059 8.864 LL .365 .133 .129 8.532 LLL .500 .250 .246 7.937 LLL, .578 .334 .323 7.518 LL, TROCH

Variables R R Adjusted R SEE

TROCH .328 .107 .100 7.380 TL .189 .036 .027 7.671 LLL .507 .257 .251 6.732 R, R2, Adjusted R2, and SEE of lower extremities variables in females.

Table 7:

ficients R, of the coefficients of determinations R2 and R2adjusted, and the standard errors of the estimates (SEE),17 of all possible and multiple linear regression equations for each of the variables were tested for the best model (Table 6 and 7). The multiple linear regression model was found to be the best model with the highest values for multiple correlation coefficient ‘R’ as 0.578, 2 2 coefficient of determination R2 as 0.334 and R2adjusted as 0.323 and with lower SEE as 7.518 in the overall population.

The multiple linear regression model was also found to be the best model to predict BH in the males (R=0.562), while single linear regression model performed better in the females (R=0.507)

Considering all possible simple and multiple linear regression analysis, the best multiple linear equation to estimation BH from lower extremity dimensions in males is YMBH=115.859

Regression equation (overall population) ±SEE Regression equation (females) ±SEE Regression equation (males) ±SEE

BH=108.969+.651(TROCH) 8.165 BH=128.336+.393(TROHT) 7.380 BH=114.567+.644(TROCHT) 6.218

BH=118.569+1.138(LLL) 7.937 BH=127.775+1.401(FL) 7.443 BH=149.694 +.540(LLL) 6.929

BH=88.969+.265(LLL) +.405(LL) +.833(TROCH) 7.518 BH: Body Height: TROCH: Trochanteric height; TL: Thigh Length; LL: Leg Length; LLL: Lower;

Table 8. Regression Equations for BH in the overall population, females and males lower extremity measurements.

Observed Value For Bh 149.00 190.00 167.5460 9.1027 211

TROCH 145.3409 178.5899 167.4859 4.14785 198 E U L TL 162.1269 175.9106 167.4859 2.31359 198 AV D LL 156.5823 175.0888 167.4859 3.33295 198 E T LLL 153.8482 183.4376 167.4859 4.56587 198 CI D LLL, 146.7266 183.3746 167.4859 5.27892 198 E R LL, P TROCH BH: Body Height; TROCH: Trochanteric height; TL: Thigh Length; LL: Leg Length; LLL: Lower; Std: Standard deviation.

Table 9. Minimum, Maximum, Mean and standard deviations of the predicted Values of BH by regression functions using the lower extremities parameters in overall population.

Females Males 2 2 2 2 Variables R R Adjusted R SEE

TROCH .515 .265 .256 6.218 TL .385 .148 .137 6.695 LL .427 .182 .172 6.559 LLL .296 .087 .076 6.929 LL, .562 .316 .298 6.040 TL

When the explanatory variables were considered one after the other, the best linear equation used to regress BH for the females is Y =119.239+.166(LLL), while that of the males FBH is Y =114.567+.644(TROCH) however stature could also be MBH estimated from any of the dismembered parts or sub-segments of the lower extremity using the formula constructed in the tables (Table 8).

The mean predicted value of BH through the regression function was similar to the mean observed value; however the minimum and maximum value indicated that there were differences in the predicted and observed value (Table 9).

The mean predicted value of BH through the regression function was similar to the mean observed value; however the

Minimum Maximum Mean Std. Deviation N

Observed Value For Bh

Table 10. Minimum, Maximum, Mean and standard deviations of extremity variables in females

Observed Value For

D TL 166.992 183.682 173.693 2.774 81 E T E CI U LL 165.466 179.069 173.693 3.079 81 L D E AV LLL 168.069 180.499 173.693 2.132 81 R P

Table 11. Minimum, Maximum, Mean and standard deviations of the predicted Values of BH by regression functions with lower extremity variables in males.

minimum and maximum value indicated that there were differences in the predicted and observed value; the minimum predicted value overestimates the minimum observed value and in both sexes while the maximum predicted value underestimates the maximum observed value in both sexes except the maximum predicted value from LL (Tables 10 and 11).

Discussion

Four lower extremity measurements including BH of the subjects were taken. The prediction function was derived through linear regression and multiple regressions for each of the measurement with BH, for the general population and for the males and females separately.

In sexing the lower extremity parameters, all the variables were highly significant (p<0.0001) except TL. These values were higher in the males than in the females. This agrees with the popular phenomena that most anthropometric variables are significantly higher in the males than in the females.

In this study, the value of coefficient of determination R2 for the multiple linear regression equations with BH as the dependent variable and LLL, LL and TROCH as explanatory variables in overall population was 0.334. The value of R2 with BH as the dependent variable and LL and TL as explanatory variables in males was 0.316. This means that 33.4% of the total variation in BH is explained by the explanatory variables LLL, LL and TROCH put together in the in overall population, while 32% of the total variation in BH is explained by the explanatory variables LL and TL in males. The values of multiple correlation coefficient R for the multiple linear regression equations for the overall population and males were 0.578 and 0.562 respectively,

Min Max Mean SD N

190 163.172 7.643 123

LLL 151.680 172.609 163.189 3.946 117 the predicted Values of BH by regression functions with lower

Min Max Mean SD N BH 156.00 190.00 173.660 7.296 88

while the SEE were 7.518 and 6.040 respectively. This means that the multiple linear regression models for the overall population as well as for the males fits very well to the observed value (compare tables 6, 7, 8 with 9, 10 and 11) unlike the linear regression models.

The findings above are not congruent with the report by Ozaslan A, et al13, who carried out similar work by measuring seven lower extremity variables on adult Turks (203 males and 108 females). Five (TROCH, LL, TL, Foot length, and Foot height) out of the seven variables were significantly correlated with stature. The highest value of coefficient of determination R2 was associated with TROCH (58%) and LL (56%) in the males while in females it was LLL (65%) and LL (63%).

2 2 The R, R2, Adjusted R2 and SEE obtained in this study 6,13 were lower than that obtained by.6,13 This means that the error arising from the predicted BH is minimal in the present study.

15,18-25 8 Using a Polish sample, Hauser R, et al were able to determine that the longest measurement of femur was the most useful for correlating with living height, atleast in males. Similarly26 reported that the femur was the most useful for estimating stature in a sample of adult Portuguese while indicated that the femur and tibia were the best predictors of stature in a Mesoamerican population. This is at variance with the result of this study because in the overall population the combination of TROCH, TL and LLL performed better than any of the independent lower extremity segments; this also applies to the males using LL and TL. However the independent/explanatory variables that ranks first in the prediction of BH in our study was LLL in the overall population as well as in the females while TROCH performed better than the rest of the other lower extremity seg-

Conclusion

The multiple regression model generated using LLL, LL and TROCH put together provided the best/highest accuracy of determination in the overall population while LL and TL gave that of the males.