Interchangeability of Sodium and Potassium Result Values of Arterial Blood Gas with Laboratory Analyzer: Narrative Review

ABSTRACT The major extracellular electrolytes, sodium, and potassium are often requested together and form a large percentage of the requested tests in routine clinical chemistry laboratories. Two types of devices that use direct and indirect ion-selective electrode (ISE) methods are used in hospitals for electrolyte measurements: blood gas analyzers (BGA), which use direct ISE technology, and the indirect ISE method, which is often used in a central-laboratory autoanalyzer (AA). We aimed to summarize the current scientific knowledge based on whether the electrolyte test results, using Na and K test results obtained with BGA and an AA, can be used interchangeably. We searched Medline (PubMed), Google Scholar, and Web of Science up to 31st March 2018. In addition, references of the included studies were also examined. Fourteen studies with a risk of bias were included in the analysis. Limits of agreement differences were variable among BGA and AA sodium and potassium test results in clinical practice. The results of both BGA and AA measures should not be used interchangeably under the assumption that they are equivalent to each other. How to cite this article Ustundağ Y, Huysal K, Ozgunay ŞE, Turkoğlu AR. Interchangeability of Sodium and Potassium Result Values of Arterial Blood Gas with Laboratory Analyzer: Narrative Review. Indian Journal of Critical Care Medicine, January 2019;23(1):35-42.


INTRODUCTION
The major extracellular electrolytes, sodium, and potassium are often requested together and form a large percentage of requested tests in routine clinical chemistry laboratories.
Sodium is responsible for the osmotic pressure of the extracellular fluid, and the physiological level in serum is 135 to 145 mmol/L. 1 Potassium is the major cation in intracellular fluid and, despite playing a physiological role in such processes as a contraction of skeletal muscles, it is necessary for normal cell function. 13][4][5][6] Hyponatremia, defined as a sodium concentration <135 mmol/L, is the most common electrolyte abnormality encountered in the emergency room and ICU and can lead to serious neurological complications. 3,4yperkalemia, which occurs with potassium higher than 5.0 mmol/L, is a common electrolyte disorder leading to morbidity and mortality due to cardiac dysrhythmia, especially in ED patients. 5ecause abnormal serum sodium and potassium levels are associated with mortality and morbidity, it is extremely important for patients to be diagnosed quickly and to start treatment early. 6

Laboratory Diagnosis of Electrolyte Disorders
Electrolytes are routinely measured by either direct or indirect ISE.The principle of the method is based on the determination of the electromotive power (potential) changes occurring between the measuring electrode and the reference electrode, whereas the ion to be measured interacts with the ISE membrane. 7wo different types of devices using direct and indirect ISE methods are used in hospitals for electrolyte measurements.BGA use direct ISE technology that measures electrolytes in undiluted sample types.Devices based on the indirect ISE method are often used in a high-efficiency central laboratory AA.Before measuring electrolyte concentrations with the indirect ISE method, the same diluent volume is used by estimating the amount of dilution by the expected solid fraction (7%).However, if the solid fraction is increased, as, during hyperproteinemia, the measured ion concentration is underestimated because of the higher dilution. 8lectrolyte values of the venous blood samples can be obtained after an average of 60 minutes in the AA in routine clinical chemistry laboratories.In the meantime, treatments depending on electrolyte values are required to be performed immediately, and are either done blindly or are delayed. 9In addition, the result times are even longer when the samples are hemolyzed, lipemic, inadequate, or lost, and while the devices are in the process of calibration.BGAs are especially advantageous in such places as emergency service units and ICUs because all measurement results are available in as little as 2 minutes.
1][12] With the observation of differences between the two results (BGAversus AA), even if samples are taken at the same time, physicians are often faced with the question of which test result to use in patient treatment, especially when therapy is to be initiated or frequent measurements are made to guide treatment.
We aimed to find whether the electrolyte test results using Na and K test results obtained with BGA and AA could be used interchangeably.

METHOD
This article is neither a systematic review nor a metaanalysis.We searched Medline (Pubmed), Google Scholar, and Web of Science for English only in humans up to 31st March 2018; using the query 'blood gas analyzer or autoanalyzer in combination with sodium and/or potassium'.To eliminate as much unrelated research as possible, we determined that keywords must appear in the title or abstract.In addition, to include as many related studies as possible, references of the included studies were also examined.We then selected the publications with potential practical usefulness.We kept studies of adults but not children.
We excluded studies that use correlation and regression analysis, or the difference between the averages instead of Bland-Altman methods.The Bland-Altman plot quantifies the bias and a range of agreement within which 95% of the differences of the second method (as compared to the first one) fall.The Bland-Altman method objectively measures the differences between measurement techniques. 12,13e acceptability criteria of interchangeability of results were derived from The United States Clinical Laboratory Improvement Amendments (US CLIA) guidelines, which state that 95% of results should fall within 0.5 mmol/L for potassium levels and 4 mmol/L for measured sodium levels to assess the intralaboratory quality of clinical chemistry tests.13,14

RESULTS
There are some methodological problems when comparing BGA and AA electrolyte results.For example, some studies used the results of the heparinized arterial sample is plasma and the venous sample is serum the patient's database retrospectively, whereas others compared prospective sample results.As this may account for some of the differences.

Some Prospective Studies Investigate Whether the Electrolyte Test Results Using Arterial Whole Blood Versus Arterial Plasma or A Serum Specimen can be Used Interchangeably
6][17] However, 95% limits of agreement (LOA) were between -9.4 and12.6 mmol/L for Na and -0.58 and 1.24 mmol/L for K, and interchangeable use is unacceptable. 15,16n these studies, the devices of different manufacturers were compared (Table 1).Yilmaz et al. compared the results of the Abbott C 8000 Architect AA (Abbott Diagnostics, Abbott Laboratories, North Chicago, Illinois, USA) and Siemens Rapid Point 500BGA (Siemens Healthcare Diagnostics, Inc. Tarrytown, NY, USA). 15King et al. compared the results of the electrolytes with a Hitachi 717AA (Boehringer Mannheim, Lewes, West Sussex, UK) and a Radiometer ABL 505 BGA (Radiometer, Crawley, West Sussex, UK). 16llardet-Servent et al. demonstrated that 48% of the differences between the electrolyte test results of the two analyzers-an AU 580 AA (Beckman Coulter, Brea, CA, USA) and a RAPID Point 500 BGA(Siemens Healthcare Diagnostics Inc., Tarrytown, NY, USA)-were due to changes in the serum protein level. 17hacko et al. compared the sodium results of whole blood and serum samples; the mean bias was -4.07 mmol/L and 95% LOA -8.8 to 0.7. 18They used a GEM 3000 BGA (Instrumentation Laboratory, Werfen, Italy) and an Olympus AU2700 AA (Olympus Optical Company, Ltd., Japan).The mean difference in potassium values was -0.3 mmol/L and 95% LOA -0.72 to +0.13 mmol/L.However, individual differences were clinically signifi- cant, especially at low potassium levels (<3.0 mmol/L), and they suggested that delivery to the central laboratory by a pneumatic transport system may lead to hemolysis. 18uvet et al. compared 491 paired whole blood and plasma sample electrolyte levels with ABL 825 ® FLEX BGA (Radiometer, Copenhagen, Denmark) and AU2700 AA(Beckman Coulter Inc., Miami, FL, USA) devices.Because preanalytical conditions were near-perfect, the bias of the results obtained for potassium was 0.1 (95% LOA 0.1-0.5),and the bias for sodium was 1 (95% LOA minus 3 to 4); both results are interchangeable. 19n a recent study, Banerjee et al. compared ABL 800 (Radiometer, Copenhagen, Denmark) with an AU640 AA (Beckman Coulter Inc., Miami, FL, USA) analyzers to find a correction factor for the appropriation of the ABG value with the AA to minimize all errors.They concluded that a correction factor should be determined individually for each hospital.20

Jose et al. compared critical electrolytes run on a Bayer
Rapidlab 865 BGA (Siemens Healthcare Diagnostics Inc., Tarrytown, NY, USA) and an Olympus AU640 AAor an Olympus AU2700 AA (Beckman-Coulter, Inc., Fullerton, CA, USA).The difference between the potassium values of the two methods is 0.03 mmol/L and 95% LOA 0.011 to 0.056.The Bland-Altman statistical method shows that even in hypokalemia and hyperkalemia, 95% of the patient results bias is less than 0.5 mmol/L. 21tory et al. evaluated electrolytes with albumin levels and demonstrated that if the plasma albumin level was above 40 g/L, the bias was 0, and the indirect ISE Na value was found to be higher in hypoalbuminemia patients. 22

Some of the Prospective Studies Compare the Levels of Electrolytes Using Arterial Whole Blood Versus Venous Serum or Plasma
The study of Zhang et al. prospectively compared arterial and venous blood Na and K results obtained with an ABL 90 FLEX BGA (Radiometer Medical ApS, Copenhagen, Denmark) and a VT-5600AA (Johnson and Johnson Services, Inc., New Jersey, USA).The mean difference between the two devices for sodium was 3.04, and 95% LOA was minus 1.24 to 7.31 mmol/L.The paired test result biases of 32/50 were higher than the values accepted by the US CLIA.The average bias for potassium was 0.43 mmol/L (95% LOA minus 0.29-1.16mmol/L).The 44/50 pairs of values exceeded the acceptable range for US CLIA (0.5 mmol/L). 23ngyingsinn et al. studied whole blood electrolyte levels with a Bayer 348BGA (Bayer Diagnostics, Siemens Healthcare Diagnostics Inc., Tarrytown, NY, USA) and venous blood with a Roche Modular ISE 1800 AA (Roche Diagnostics, Indianapolis, IN, USA).The mean difference between the two methods was 0.49 mmol/L (95% CI of LOA 0.893-0.943).However, this difference is explained by the range of 1-54 minutes for taking blood from the patient. 24

Retrospective Studies Comparing the Levels of Electrolytes Using Arterial Whole Blood Versus Venous Serum or Plasma
In retrospective studies comparing the arterial whole blood and venous serum or plasma results, the mean bias for sodium was found to be 4.9 to 2.1 mmol/L, and the LOA was minus 0.97 to 10.05 mmol/L. 11,25][27] Johnston and colleagues analyzed K + in arterial and venous blood samples taken from 50 patients during cardiac arrest; and ran them on an IL1640 BGA(Instrumentation Laboratory System, Lexington, Mass., USA).The venous sample was run for analysis via a pneumatic tube delivery system through a central Olympus AA (Beckman Coulter Inc., Miami, FL, USA).Johnston explained that the differences between the results are due to unidentified hemolysis. 27n their retrospective analysis, Açıkgöz and colleagues compared 118 patients with acute potassium elevations analyzed with an ABL 700 radiometer (Radiometer, Copenhagen, Denmark) with the Architect's c16000 AA (Abbott Diagnostics, Abbott Laboratories, ILL, USA).The mean difference between the two methods was 0.62 ± 0.43 mmol/L (95% confidence intervals: 1.462 and -0.222). 28

DISCUSSION
There are several reasons for the differences observed in BGA and AA interchangeability studies for Na and K results.
The main reason for differences in the potassium value measured by the BGA device is that the hemolysis is not noticeable in the arterial specimen.Hawkins et al. reported that hemolysis is higher than predicted, and 33% of hypokalemic cases could not be detected with BGA. 29 Venous samples are always centrifuged and then visually examined for hemolysis in routine laboratories.When hemolysis is detected, the sample is discarded, or no measurement is made for potassium.Meanwhile, the release of K from the platelets during coagulation may cause an increase in serum. 30

IJCCM
The pneumatic tube system may lead to hemolysis, 31 and potassium levels may differ in the central laboratory. 18,27,325][36][37] In addition, heparin itself binds positively charged ions and lowers the value of the electrolytes measured. 14,38tudies have reported that when comparing BGA with an AA device, serum protein, and albumin levels are significantly important. 17The difference between the results was found to be correlated with serum albumin and total protein concentrations. 9,22t is important to detect small changes in the relative sodium concentrations in the follow-up of critical situations. 38Patients with symptomatic hyponatremia require slow correction, so patients' serum sodium levels should be monitored frequently. 39However, plasma protein levels may vary during treatment.In such cases, it is necessary to obtain the results from a single analyzer and minimize the analytical differences between the devices.
The accuracy and stability of the different calibrators used in each autoanalyzer are important for the reliability of the results, which can lead to differences in comparison studies. 40

Limitations
For the measured variables, it is not possible to determine which electrolyte values (BGA vs. AA) are closer to the true value.Although external or internal quality-control studies are performed in routine laboratories; it does not show the real value of the sample.

CONCLUSION
It seems inappropriate to draw a conclusion about the interchangeability of different device results.Electrolyte levels should be regularly monitored, and the results of both measures should not be used interchangeably under the assumption that they are equivalent to each other.

Table 1 :
Summary of the studies