TTSA was considered as endogenous variable and remaining anthropometrical Imatinib mw variables (i.e. body mass, body height, BCD, CSD and CP) as exogenous variables. The variables entered the equation if F�� 4.0 and removed if F�� 3.96 as suggested elsewhere (Barbosa et al., 2008). All assumptions to perform the selected multiple regression models were taken into account. For further analyses the equation computed, the coefficient of determination (R2), the adjusted coefficient of determination (Ra2), the error of estimation (s) and the probability of rejecting the null hypothesis (p �� 0.05). In each exogenous variables included in the final model, the t-value and the p-value were considered as well. Validation was made in the second sub-sample group (Baldari et al., 2009; Kristensen et al.

, 2009; Wolfram et al., 2010): (i) comparing mean data; (ii) computing simple linear regression models and; (iii) computing Bland Altman plots. Comparison between the mean TTSA assessed and the TTSA estimated, according to the equations previously developed, was made using paired Student��s t-test (p �� 0.05). Simple linear regression model between both assessed and estimated TTSA was computed. As a rule of thumb, for qualitative and effect size analysis, it was defined that the relationship was: (i) very weak if R2 < 0.04; weak if 0.04 �� R2 < 0.16; moderate if 0.16 �� R2 < 0.49; high if 0.49 �� R2 < 0.81 and; very high of 0.81 �� R2 < 1.0. In addition, the error of estimation (s) and the confidence interval for 95 % of the adjustment line in the scatter gram was computed.

The Bland Altman analysis (Bland and Altman, 1986) included the plot of the mean value of TTSA assessed and estimated versus the delta value (i.e. difference) between TTSA assessed and estimated. It was adopted as limits of agreement a bias of �� 1.96 standard deviation of the difference (average difference �� 1.96 standard deviation of the difference). For qualitative assessment, it was considered that TTSA estimated was valid and appropriate if at least 80% of the plots were within the �� 1.96 standard deviation of the difference. Results Morphometric characteristics Table 1 presents the descriptive statistics for all selected anthropometrical variables, according to gender groups. Overall, it can be verified that most mean values are higher in male than in female subjects.

Data dispersion can be considered as weak (i.e. CV �� 15%) or moderate Batimastat (i.e. 15% < CV �� 30%) within each gender group. Table 1 Anthropometrical characteristics of male (M) and female (F) subjects for body mass (BM), body height (H), biacromial diameter (BCD), chest sagital diameter (CSD), chest perimeter (CP) and measured trunk transverse surface area (TTSA) Computation of trunk transverse surface area prediction models For male gender, the final model (F2.75 = 17.143; p < 0.001) included the CP (t = 2.963; p < 0.001) and the CSD (t = 2.333; p = 0.