Comparison of a new simplified acid–base tool to the original Stewart–Figge approach: a study on cardiac surgical patients

Abstract

Purpose

To suggest a simplified method for strong ion gap ([SIG]) calculation.

Patients and methods

To simplify [SIG] calculation, we used the following assumptions: (1) the major determinants of apparent strong ion difference ([SID_a]) are [Na⁺], [K⁺] and [Cl⁻] (2) [Ca²⁺] and [Mg²⁺] do not contribute significantly to [SID_a] variation and can be replaced by their reference concentrations (3) physiologically relevant pH variation is at the order of 10⁻² and therefore we can assume a standard value of 7.4. In the new model, [SID_a] is replaced by its adjusted form, i.e. [SID_a,adj] = [Na⁺] + [K⁺] − [Cl⁻] + 6.5 and [SIG] is replaced by “bicarbonate gap”, i.e. [BIC_gap] = [SID_a,adj] − (0.25·[Albumin]) − (2·[Phosphate]) − [HCO₃⁻]. The model was tested in 224 postoperative cardiac surgical patients.

Results

Strong correlations were observed between [SID_a,adj] and [SID_a] (r = 0.93, p < 0.0001) and between [BIC_gap] and [SIG] (r = 0.95, p < 0.0001). The mean bias (limits of agreement) of [SID_a,adj] − [SID_a] and of [BIC_gap]–[SIG] was − 0.6 meq/l (− 2.7 to 1.5) and 0.2 meq/l (− 2 to 2.4), respectively. The intraclass correlation coefficients between [SID_a,adj] and [SID_a] and between [BIC_gap] and [SIG] were 0.90 and 0.95, respectively. The sensitivities and specificities for the prediction of a [lactate⁻] > 4 meq/l were 73.4 and 82.3% for a [BIC_gap] > 12.2 meq/l and 74.5 and 83.1% for a [SIG] > 12 meq/l, respectively.

Conclusions

The [BIC_gap] model bears a very good agreement with the [SIG] model while being simpler and easier to apply at the bedside. [BIC_gap] could be used as an alternative tool for the diagnosis of unmeasured ion acidosis.