Significant changes in pre-dialytic partial pressure of CO2 (pCO2) during a week-long cycle of hemodialysis (HD) can be an effect of the intermittent supplementation of bicarbonate to correct chronic acidosis in patients. Mathematical modeling efforts carried out using the same parameters before each HD session might fail to produce accurate predictions of pCO2 and plasma bicarbonate concentration (CBic) because of this variability. A numerical model describing acid-base equilibrium changes during HD was applied to predict pCO2, pH, and CBic in 24 chronic HD patients, using both fixed parameters for the whole week and estimating a new value of minute ventilation (VE) and net acid generation rate (GH) for each interdialytic interval. Dialysances of bicarbonate and dissolved CO2 were also estimated independently for each HD session. The error of the model compared to the pre-dialytic data of CBic and pCO2 significantly decreased when VE and GH were estimated piecewise throughout the week. To fit the data, VE changed from 3.9 ± 1.0 mL/min before HD1, to 3.8e1 mL/min after HD1, 3.6 ± 1.0 mL/min after HD2, and 3.9 ± 1.1 mL/min after HD3 (p < 0.05). GH changes after each session were not statistically significant. VE values strongly correlated with pre-dialytic pCO2 (Spearman's ρ = -0.97), but GH only weakly correlated with pre-dialytic CBic (ρ = -0.30). Acid-base equilibrium is extremely sensitive to respiratory regulation. When attempting to predict the evolution of pCO2 a CBic during the HD cycle, changes in the respiration parameters must be accounted for by the model, at the risk of a significant loss of prediction accuracy.
Read full abstract