Statistical analysis
. The sample size calculation, using the method described by Eberl et al. [17], was based on the retrospective sedation database of our hospital, in which the propofol requirement for gastrointestinal endoscopy was 300 (100) mg, presented as mean (standard deviation). Given power of 0.80 and type I error of 0.05, we will need to study 28 subjects per group to decrease propofol requirement by about 25%. And taking into consideration a potential dropout rate of 10%, 30 patients for each group will be required, thus a total of 90 adult patients should be randomly assigned.
SPSS 13.0 or the software R was performed for statistical analysis. We used the modified Dixon’s up-and-down methodology, described in 1965 [18], to determine the EC50 of TCI propofol with different doses of dexmedetomidine. We performed a chi-squared test for categorical variables, ANOVA test for continuous values and Kruskal-Wallis rank sum test for discrete values as appropriate. Data were presented as number (n), mean (standard deviation, SD), or median [Min, Max]. A 0.05 was set as the threshold of rejecting the null hypothesis.