### Introduction

### Materials and Methods

_{ICG}(t) = C

_{ICG}(0)·e

^{-kt}], was computed by linear regression from a semilogarithmic plot of ICG concentration versus time from approximately 2 min to 6 min after injection, where C

_{ICG}(t) = ICG concentration at t minutes after injection and e = natural logarithm [15]. Measurement of CO and blood volume (BV) was usually completed within 6 min. Propofol was then intravenously infused to attain a steady plasma concentration of 3 µg/ml using the TCI technique based on the pharmacokinetic parameters introduced by Schnider et al. [5] (RUGLOOP®, version 3.14, DeSmet T. and Struys M. Department of Anesthesia, University Hospital Ghent). The calculating parameter included patient age, sex, body weight and height. Changes in EEGs were assessed using the bispectral index (BIS: A-1050, version 3.4, Aspect Medical Inc., Natick, MA, USA. The averaging window was 15 seconds). The time required to achieve BIS values of 80 (BIS80) and 60 (BIS60) were determined. Patients who showed BIS values less than 96 at the beginning of induction were excluded from the analysis. The displayed BIS values in the window were decreased and fluctuated, and therefore, BIS80 and BIS60 were defined as the times when the displaying number of BIS value first reached each of the end point values, 80 and 60. We hypothesized BIS80 as the point of a very light anesthetic and hypnotic state, and BIS60 as the point of an adequate anesthetic state, with reference to a previous investigation [12].

^{2}), and the correlation between each of the variables was calculated using a linear regression model. Subsequent step-up selection determined candidate variables for the model that increased the correlation the most, and we stopped adding independent variables when the remaining variables were not significant (P > 0.05). A value of P < 0.05 was considered statistically significant. Data are presented as the mean ± SD. Statistical analysis was performed with the software package NCSS 2000 (Number Cruncher Statistical Systems, Kaysville, UT).

### Results

^{2}= 0.076, P < 0.020) (Table 3, Fig. 1). For BIS80, CO was determined as the only significant predictive variable. Increased age significantly prolonged the time taken to reach BIS60, as did increased LBM (R

^{2}= 0.220, P < 0.001). The plasma concentration of propofol was not a predictive variable for the time required to achieve either end point, and it was lowered by increased BV (R

^{2}= 0.073, P < 0.020) (Table 3, Fig. 1). Multicollinearity among independent variables was rejected in all models, because the variance inflation factor was < 10, and all the condition numbers for the eigenvalues of centered correlations were < 10. The power values of each accepted model were greater than 0.7.

### Discussion

^{2}value was rather small, except in the case of BIS60 in spite of a proper set of significant parameters and multivariable feedback control analysis used to calculate the biggest R

^{2}value. The usefulness of linear regression for the type of data that we explored would be limited [26,27]. It was only BV and not LBM that significantly affected the time taken to reach BIS60. However, it is possible that both variables may be correlated since our results leave a room for discussion concerning statistical completeness. The Cardiac Index (CI) rather than CO may be a preferable factor in the analysis; however, we had already explored factors including patient weight and height. CI is a product of weight, height and CO, and we selected CO as an appropriate and independent variable to elude the statistical problem of multicollinearity.