Objectives In our study of the accuracy of blood gas/electrolyte analyzers, we have re-discovered that the the anion gap calculation (Na – Cl – HCO3) is an excellent indicator of electrolyte measurement inaccuracy. Calculations that incorporate component measurements will exhibit propagation of errors; inaccurate systems will demonstrate increased divergence of anion gap. While zero and negative anion gaps are usually attributed to laboratory error, such gaps may indicate significant pathophysiology. Astute clinicians will investigate these outlying gaps with tests for paraproteinemia and confirmation of calcium, albumin, potassium, magnesium and lithium levels. We compared the prevalence of zero/negative anion gaps of 3 different blood gas/electrolyte analyzers operated over 2 years in 3 different ICUs. Analyzers demonstrating high prevalence of zero and negative anion gaps predispose to overtesting and overdiagnosis.

Method Laboratory data repositories provided all the arterial blood gas, electrolyte and metabolite results generated by 1) two GEM 4000s on ICU patients in 2012 and 2013 at Calgary’s Foothills University Hospital, 2) two Radiometer ABL800 systems on ICU patients in 2012–2013 at Edmonton’s University of Alberta Hospital and 3) two Siemens Rapid 500 systems on ICU patients in October 2015–October 2017 at Dartmouth-Hitchcock Medical Center. At the two Alberta ICUs operating two different analyzers, we determined the cost of replicate central laboratory testing that was ordered within one hour of the ICU electrolyte measurements.

Results 935 (4.0%) of 23212 GEM gaps; 548 (1.3%) of the 42690 Radiometer gaps and 116 (0.9%) of 12951 Siemens gaps were zero/negative. The relative incidence of zero/negative gaps is higher for the GEM compared to the Radiometer (P

Conclusions Because of imprecise Na, Cl and HCO3, the GEM analyzer tends to produce larger and more artefactual zero and negative gaps than the Radiometer and Siemens. These gaps are associated with increased testing to confirm the electrolyte concentrations as well as other tests that are used to rule out serious pathophysiology. The costs of the increased testing and any overdiagnosis (including the discovery of incidental paraproteinemia) probably exceeds the savings achieved by acquiring the less accurate, lower cost system.