The influence of hyperchloraemia on acid base interpretation in diabetic ketoacidosis

Abstract

Objectives

During the acute treatment of diabetic ketoacidosis we (a) determined the temporal incidence of hyperchloraemia, and (b) quantified the influence of hyperchloraemia on interpretation of common blood gas derived acid base parameters, namely base deficit and bicarbonate.

Designand setting

Retrospective chart review in two regional paediatric intensive care units.

Measurements and Results

Stewart's physicochemical theory was used to develop regression equations quantifying the acidifying effect of hyperchloraemia on both base deficit and bicarbonate. These were then applied retrospectively to blood chemistry results from 18 children (median age 12.7 years, weight 43 kg) with diabetic ketoacidosis. Plasma ketonaemia was estimated using the albumin-corrected anion gap. The incidence of hyperchloraemia, as documented by a ratio of plasma chloride to sodium of greater than 0.79, increased from 6% at admission to 94% after 20 h of treatment. Correction for chloride produced a dramatic improvement in the relationship between changes in the anion gap vs. both base deficit (from R ² = 0.55 to R ² = 0.95) and bicarbonate (from R ² = 0.51 to R ² = 0.96) during treatment. After 20 h of treatment the mean base deficit had decreased from 24.7 mmol/l to 10.0 mmol/l however, the proportion that was due to hyperchloraemia increased from 2% to 98%.

Conclusions

It is now possible using a simple correction factor to quantify the confounding effect of hyperchloraemia on both base deficit and bicarbonate in diabetic ketoacidosis. This bedside tool may be a useful adjunct to guide therapeutic interventions.

Appendix

Law of mass action

When a weak acid dissociates incompletely in an aqueous solution, the relationship between the dissociated and undissociated components is constant. The Henderson equation is a well-known modification of this law, namely k = [H⁺] [HCO₃ ⁻]/[H₂CO₃], where k is defined as the hydrogen ion concentration at which half of the carbonic acid is dissociated. The Stewart approach also applies the law of mass action to carbonic acid but includes other major weak acids found in human plasma (albumin, phosphate), as well as incorporating the dissociation equilibrium for water: K _w = [H⁺] [OH⁻]/[H₂O].

Mass conservation

In an aqueous solution the amount of each component substance remains constant unless (a) a substance is added or removed or (b) the substance is generated or destroyed.

Preservation of electroneutrality

In an aqueous solution, the sum of all positively charged ion concentrations always equals the sum of all negatively charged ion concentrations.

AcidBasics II software uses the Stewart equations to calculate pH from its three independent variables, namely pCO₂, strong ion difference ([Na⁺] + [K⁺] + [Mg²⁺] + [Ca²⁺] − [Cl⁻] − [lactate⁻]) and total weak acid (phosphate and albumin charge) at a temperature of 37°C. If pH is known and all variables other than sodium and chloride are kept constant the magnitude of pH change as chloride is varied relative to sodium can be calculated. Since bicarbonate is dependent on pH, its theoretical charge (mEq/L) can be accurately calculated using across varying degrees of hypo/hyperchloraemia (chloride 70–130 mmol/L) by keeping all other variables in Stewart's equation constant. A linear regression equation can be plotted between (Na-Cl) and calculated bicarbonate to explain the degree to which bicarbonate is directly influenced by hyperchloraemia. For example sodium of 140 mEq/L and chloride 120 mEq/L (difference 20 mEq/L) will result in a bicarbonate concentration of 0.93 × 20 mEq/L − 5.9 = 12.7 mEq/L using this regression equation. A similar approach can be used to calculate the influence of hyperchloraemia on the base deficit. For example a difference of 20 mEq/L between Na and Cl results in a base deficit of −1.2 × 20 mEq/L + 42.5 = −18.5 mEq/L. Stewart's methodology is necessary since the traditional Henderson-Hasselbalch model assumes bicarbonate is an independent variable and is not influenced by chloride, hence it offers no method for the quantification of hyperchloraemic acidosis.