The predictive performance of infusion strategy nomogram based on a fluid kinetic model

BACKGROUND In a previous study, fluid kinetic models were applied to describe the volume expansion of the fluid space by administration of crystalloid and colloid solutions. However, validation of the models were not performed, it is necessary to evaluate the predictive performance of these models in another population. METHODS Ninety five consenting patients undergoing elective spinal surgery under general anesthesia were enrolled in this study. These patients were randomly assigned to three fluid groups i.e. Hartmann's solution (H group, n = 28), Voluven® (V group, n = 34), and Hextend® (X group, n = 33). After completion of their preparation for surgery, the patients received a loading and maintenance volume of each fluid predetermined by nomograms based on fluid pharmacokinetic models during the 60-minute use of an infusion pump. Arterial samples were obtained at preset intervals of 0, 10, 20, and 30 min after fluid administration. The predictive performances of the fluid kinetic modes were evaluated using the fractional change of arterial hemoglobin. The relationship between blood-volume dilution and target dilution of body fluid space was also evaluated using regression analysis. RESULTS A total of 194 hemoglobin measurements were used. The bias and inaccuracy of these models were -2.69 and 35.62 for the H group, -1.53 and 43.21 for the V group, and 9.05 and 41.82 for the X group, respectively. The blood-volume dilution and target dilution of body-fluid space showed a significant linear relationship in each group (P < 0.05). CONCLUSIONS Based on the inaccuracy of predictive performance, the fluid-kinetic model for Hartmann's solution showed better performance than the other models.