Estimating the net effect of unmeasured ions in human extracellular fluid using a new mathematical model. Part I: Theoretical considerations

A head to head evaluation of 8 biochemical scanning tools for unmeasured ions

We aimed to evaluate the sensitivity and specificity of 8 biochemical scanning tools in signalling the presence of unmeasured anions. We used blood gas and biochemical data from 15 patients during and after cardio-pulmonary bypass. Sampling time-points were pre-bypass (T1), 2 min post equilibration with priming fluid containing acetate and gluconate anions (T2), late bypass (T3) and 4 h after surgery (T4). We calculated the anion gap (AG), albumin-corrected anion gap (AGc), whole blood base excess (BE) gap, plasma BE gap, standard BE gap and the strong ion gap (SIG), plus 2 new indices-the unmeasured ion index (UIX) and unmeasured plasma anions according to the interstitial, plasma and erythrocyte acid-base model (IPEua). Total measured plasma concentrations of acetate and gluconate [XA] were proxies for unmeasured plasma anions. [XA] values (mmol/L) were 1.41 (0.87) at T1, 11.73 (3.28) at T2, 4.80 (1.49) at T3 and 1.36 (0.73) at T4. Corresponding [albumin] values (g/L) were 32.3 (2.0), 19.8 (2.6), 21.3 (2.5) and 29.1 (2.3) respectively. Only the AG failed to increase significantly at T2 in response to a mean [XA] surge of >10 mEq/L. At an [XA] threshold of 6 mEq/L, areas under receiver -operator characteristic curves in rank order were IPEua and UIX (0.88 and 0.87 respectively), SIG (0.81), AGc (0.79), standard BE gap (0.77), plasma BE gap (0.71), BE gap (0.70) and AG (0.59). Similar ranking hierarchies applied to positive and negative predictive values. We conclude that during acute hemodilution UIX and IPEua are superior to the anion gap (with and without albumin correction) and 4 other indices as scanning tools for unmeasured anions.

A comprehensive, computer-model-based approach for diagnosis and treatment of complex acid-base disorders in critically-ill patients

We have developed a computer-model-based approach to quantitatively diagnose the causes of metabolic acid-base disorders in critically-ill patients. We use an interstitial-plasma-erythrocyte (IPE) model that is sufficiently detailed to accurately calculate steady-state changes from normal in fluid volumes and electrolyte concentrations in a given patient due to a number of causes of acid-base disorders. Normal fluid volumes for each patient are determined from their sex, height and weight using regression equations derived from measured data in humans. The model inputs (electrolyte masses and volumes) are altered to simulate the laboratory chemistry of each critically-ill patient. In this process, the model calculates changes in body-fluid volumes, osmolality and yields the individual values of IPE base excess (BE(IPE)) attributed to changes due to: (1) fluid dilution/contraction, (2) gain or loss of Cl(-), (3) hyper- or hypoalbuminemia, (4) presence of unmeasured ions, (5) gain of lactate, (6) gain or loss of phosphate, (7) gain or loss of calcium and magnesium, (8) gain or loss of potassium and (9) gain or loss of sodium. We use critically-ill patient data to show how our new approach is more informative and much simpler to interpret as compared to the approaches of Siggaard-Andersen or Stewart. We demonstrate how the model can be used at the bedside to diagnose acid-base disorders and suggest appropriate treatment. Hence, this new approach gives clinicians a new tool for diagnosing disorders and specifying fluid-therapy options for critically-ill patients.

Estimating the net effect of unmeasured ions in human extracellular fluid using a new mathematical model. Part II: Practical issues

Building on the theoretical considerations developed in Part I of this two-part series, a population range was established for a new extracellular fluid acid-base parameter using Monte Carlo simulation. This new parameter the unmeasured ion excess, had a normal range of +/- 2.50 mEq/l which was slightly narrower than the normal range for the standard base excess at +/- 3.20 mmol/l. In both cases, most of this variation occurred as a result of the estimation of the bicarbonate concentration. Finally, several short clinical vignettes were explored to highlight the differences between the unmeasured ion excess and the standard base excess.