Statistical Analysis
Data analysis was conducted with the Statistical Package for the Social Science (SPSS) V.21 (Armonk, NY, USA) for Mac OS. A Kolmogorov-Smirnoff test was done to analyze the normal distribution of the data (normal if p>0.05). After checking the sample homogeneity, student t-tests for independent samples were used to assess gender differences for age, height, weight, BMI, thorax circumference during maximum inspiration and expiration, or respiratory moment differences for ultrasonographic features.
Multiple linear regression analyses were conducted to determine variables significantly contributing to the variance within the pleura depth. First, a correlation analysis between pleura depth with anthropometric features and respiratory moment was performed using Pearson’s correlation coefficients (r) for normal distributed variables. Values ranging 0-0.3 were considered as poor correlation, 0.3-0.5 as fair, 0.6-0.8 as moderate and 0.8-1.0 as strong18. Those statistically significant variables (p<0.05) were included in a stepwise multiple linear regression model to estimate the proportion of variance explaining the pleura depth. To avoid risk of bias during the regression model, a multicollinearity and shared variance analyses between the variables (defined as r>0.80) were also calculated.
Hierarchical regression models were conducted to determine those variables that contributed significantly to pleura depth determination. The significance criterion of the critical F value for entry into the regression equation was set as p<0.05. The adjusted changes in R2 were reported step by step in the regression model to determine the association of each additional variable.