Purpose: To examine the repeatability of Stewart's parameters and anion gap in a cohort of critically ill patients and to determine the smallest detectable changes in individual patients.
Methods: A total of 161 patients were included prospectively. They underwent two subsequent blood samplings within 10 min of each other and samples were analyzed using the same central laboratory analyzer. Measured and calculated parameters from the two samples were compared. The repeatability was expressed as the smallest detectable difference (SDD), coefficient of variation (CV) and intraclass correlation coefficient (ICC).
Results: The mean differences ± SD (mEq/L) for the repeated measurements were 0.1 ± 0.76, 0.12 ± 0.68, -0.02 ± 1.02, and -0.08 ± 1.05 for the apparent strong ion difference (SID(app)), effective strong ion difference (SID(eff)), strong ion gap (SIG), and albumin-corrected anion gap (AG(corr)), respectively. The SDDs (mEq/L) for SID(app), SID(eff), SIG, and AG(corr), were ±1.49, ±1.33, ±2, and ±2.06, respectively. The CVs (%) for these variables were 1.4, 1.45, 13.3, and 4.15, respectively. The ICCs for all these variables were high, largely above 0.75.
Conclusions: The repeatability of all these calculated variables was good. In repeated measurements, a change in value of these parameters exceeding 1.96√2 CV (%), the least significant change (LSC) or the SDD should be regarded as significant. Use of SDD is preferable to CV and LSC (%) because of its independence from the levels of variables and its expression in absolute units. Expressed as SDD, a SIG change value, e.g., of at least ±2 mEq/L should be significant.