WHY ORDER TWO WHEN YOU CAN GET ONE FOR HALF THE PRICE?: EVALUATING THE RELIABILITY OF POINT OF CARE ARTERIAL BLOOD GAS ELECTROLYTE MEASUREMENTS FOR USE IN STANDARD SUPPLEMENTATION PROTOCOLS

Background: Frequent blood draws are implicated in hospital-acquired anemia as well as rising costs. Critically ill patients undergo frequent venous and arterial blood sampling, both providing electrolyte concentrations. In this study, we sought to examine electrolyte values obtained by arterial blood gas testing (ABG) and central laboratory testing (VCL) and determine the extent of essentially duplicative testing. We hypothesized minimal ABG-VCL differences and that ABG values could safely be used to guide electrolyte supplementation in our intensive care unit.

Methods: Using the electronic health record, we obtained potassium and ionized calcium values from patients admitted between June 2013 and July 2016 who had paired ABG and nonhemolyzed VCL samples drawn within five minutes of one another. All ABG results were obtained using Radiometer ABL800 FLEX instruments and were compared to potassium VCL results from Siemens Advia 1800 analyzers or ionized calcium VCL results from Instrumentation Laboratory GEM 4000s. We examined how often management by ABG would result in different supplementation compared to VCL based on our standard electrolyte management protocol.

Results:  Of the 63,646 paired potassium samples the mean value of ABG potassium was 4.0g/dL and VCL potassium was 4.1g/dL. Linear regression considering VCL as the independent variable yielded a slope of 0.95 (p<0.001), intercept of 0.05, and adjusted R squared of 0.90. There was no difference in potassium supplementation in 63% of events (n=40,376) and 20 mEq or less difference in 93% (n=59,355) of events. (Figure 1).

Of the 19,892 paired ionized calcium samples the mean value of ABG ionized calcium was 1.20g/dL and VCL ionized calcium was 1.21 g/dL. Linear regression considering VCL as the independent variable yielded a slope of 0.82 (p<0.001), intercept of 0.2, and adjusted R squared of 0.84. There was no difference in calcium gluconate supplementation in 89% of events (n=17,654), and 1g or less difference in 99% (n=19,789) of events (Figure 2).

Conclusions: Our findings suggest minimal differences between calcium and potassium measurement using point-of-care blood gas compared to central laboratory. This supports the use of blood gas electrolytes alone to guide potassium and calcium management using a standard protocol, improving time to treatment and potentially reducing iatrogenic anemia and laboratory testing costs.