Table 3. Multiple linear regression analysis to explain the relationship between gas exchange values and variables considered.
Variables | Regression model | Fitted regression model | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Dependent | Independent | VIF | EC | SE | t- value | p-value | AR2 | p-value | EC | SE | t- value | p-value | AR2 | p-value |
PaO2PRE | Sex (M) | 1.257 | -4.239 | 1.84 | -2.291 | 0.024 | 0.138 | <0.001 | -4.287 | 1.83 | -2.339 | 0.021 | 0.145 | <0.001 |
Age | 1.044 | 0.025 | 0.09 | -0.260 | 0.794 | - | - | - | - | |||||
BMI | 1.117 | -0.426 | 0.14 | -2.899 | 0.004 | -0.420 | 0.14 | -2.906 | 0.004 | |||||
Leptin | 1.318 | -0.148 | 0.06 | -2.292 | 0.024 | -0.147 | 0.06 | -2.295 | 0.023 | |||||
PaO2POST | Sex (M) | 1.257 | -5.115 | 2.379 | -2.149 | 0.0338 | 0.037 | 0.089 | -5.885 | 2.098 | -2.804 | 0.0059 | 0.058 | 0.006 |
Age | 1.044 | 0.004 | 0.128 | -0.036 | 0.971 | - | - | - | - | |||||
BMI | 1.117 | -0.103 | 0.189 | -0.547 | 0.585 | - | - | - | - | |||||
Leptin | 1.318 | 0.057 | 0.083 | 0.687 | 0.493 | - | - | - | - | |||||
PaCO2PRE | Sex (M) | 1.257 | 1.539 | 0.856 | 1.798 | 0.074 | 0.128 | <0.001 | 1.423 | 0.767 | 1.856 | 0.066 | 0.136 | <0.001 |
Age | 1.044 | 0.131 | 0.046 | 2.845 | 0.006 | 0.130 | 0.045 | 2.848 | 0.005 | |||||
BMI | 1.117 | 0.191 | 0.068 | 2.816 | 0.005 | 0.197 | 0.065 | 3.033 | 0.003 | |||||
Leptin | 1.318 | 0.009 | 0.029 | 0.310 | 0.757 | - | - | - | - | |||||
PaCO2POST | Sex (M) | 1.257 | 0.778 | 1.138 | 0.683 | 0.496 | -0.017 | 0.712 | - | - | - | - | - | - |
Age | 1.044 | -0.006 | 0.061 | -0.105 | 0.916 | - | - | - | - | |||||
BMI | 1.117 | 0.045 | 0.090 | 0.498 | 0.619 | - | - | - | - | |||||
Leptin | 1.318 | -0.032 | 0.040 | -0.812 | 0.418 | - | - | - | - |
Multiple linear regression analysis was performed to explain the relationship between one dependent normally distributed variable (gas exchange parameters) and independent normally distributed variables (sex, age, BMI, and serum leptin). Multicollinearity was not detected using variance inflation factors. Using the Akaike information criterion, backward/forward stepwise regression analysis was then performed to choose the best model. BMI, body mass index; EC, estimate coefficient; PaCO2POST, postoperative arterial partial pressure of carbon dioxide (PaCO2); PaCO2PRE, preoperative PaCO2; ΔPaCO2, PaCO2POST minus PaCO2PRE; PaO2POST, postoperative arterial partial pressure of oxygen (PaO2); PaO2PRE, preoperative PaO2; ΔPaO2, PaO2POST minus PaO2Pre; SE, standard error; VIF, variance inflation factor.