Clinical Biochemistry 48 (2015) 525–528
Contents lists available at ScienceDirect
Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem
Agreement between whole blood and plasma sodium measurements in profound hyponatremia☆,☆☆,★,★★ Pierce Geoghegan a,b,⁎, Christopher D. Koch c, Amy M. Wockenfus c, Andrew M. Harrison b,d, Yue Dong b,e, Kianoush B. Kashani b,f, Brad S. Karon c a
Department of Anaesthesia and Critical Care, Tallaght Hospital, Dublin, Ireland Multidisciplinary Epidemiology and Translational Research in Intensive Care (M.E.T.R.I.C.), Mayo Clinic, Rochester, MN, USA Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN, USA d Medical Scientist Training Program, Mayo Clinic, Rochester, MN, USA e Division of Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester, MN, USA f Division of Nephrology and Hypertension, Mayo Clinic, Rochester, MN, USA b c
a r t i c l e
i n f o
Article history: Received 10 January 2015 Received in revised form 2 March 2015 Accepted 3 March 2015 Available online 13 March 2015 Keywords: Hyponatremia Sodium measurement Agreement
a b s t r a c t Objectives: We compared two different methods of whole blood sodium measurement to plasma sodium measurement using samples in the profoundly hyponatremic range (Na b 120 mmol/L). Design and methods: Whole blood pools with a range of low sodium values were generated using combinations and dilutions of pooled electrolyte-balanced lithium heparin samples submitted for arterial blood gas analysis. Each pool was analyzed five times on a Radiometer 827 blood gas analyzer and iSTAT analyzer. Pools were centrifuged to produce plasma, which was analyzed five times on a Roche Cobas c501 chemistry analyzer. An additional 40 fresh (analyzed on day of collection) excess lithium heparin arterial blood gas samples from 36 patients were analyzed on the Radiometer 827, iSTAT, and Cobas c501 as described above. The setting was a tertiary referral center. Blood samples were collected from a combination of patients in the intensive care unit, operating theaters and emergency room. Results: All methods demonstrated excellent precision, even in the profoundly hyponatremic measurement range (Na b 120 mmol/L using a plasma reference method). However, agreement between the methods varied with the degree of hyponatremia. In the profoundly hyponatremic range, Radiometer whole blood sodium values were nearly identical to plasma reference sodium, while iSTAT whole blood sodium showed a consistent positive bias relative to plasma sodium in this range. Conclusion: If whole blood direct sodium measurements are compared with plasma sodium in profoundly hyponatremic patients consideration should be given to the use of Radiometer blood gas analyzers over iSTAT since the latter shows a positive bias relative to a plasma comparative method. © 2015 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Introduction Hyponatremia is a common electrolyte abnormality in hospitalized patients [1] and dysnatremia is an independent risk factor for hospital mortality [2]. Profound hyponatremia (sodium b 120 mmol/L) can lead to cerebral edema and correction of the hyponatremia is necessary to prevent or treat associated neurological complications or death [3]. However, excessively rapid correction of profound hyponatremia is well established to be associated with a risk of osmotic demyelination ☆ Institution: This work was performed at Mayo Clinic in Rochester, MN. ☆☆ Financial support: None. ★ Conflicts of interest: None. ★★ MeSH headings: Hyponatremia, blood gas analysis. ⁎ Corresponding author at: Department of Anaesthesia and Critical Care, Tallaght Hospital, Dublin, Ireland. E-mail address: [email protected] (P. Geoghegan).
syndrome (ODS), which may itself cause substantial morbidity or mortality [4]. Targeted rates of sodium correction are often difficult to achieve in profound hyponatremia and expert opinion has consistently emphasized the unpredictability of response to treatment [5,6]. Early identification or anticipation of non-optimal correction is important because there is convincing evidence from animal models that minimizing the duration of non-optimal correction reduces the risk of subsequent ODS and mortality [7]. Despite significant attention to the diagnostic workup in profound hyponatremia, there is little evidence-based guidance on how clinicians should monitor response to treatment. Many investigators have recommended very frequent monitoring of electrolytes during therapy [8,9], and others have even suggested the monitoring of sodium as frequently as every one to 2 h during acute therapy [5]. This raises the challenge of determining the optimal method and frequency of monitoring. Surprisingly, this issue has not
http://dx.doi.org/10.1016/j.clinbiochem.2015.03.001 0009-9120/© 2015 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
526
P. Geoghegan et al. / Clinical Biochemistry 48 (2015) 525–528
attracted significant attention. While the use of indirect ion-selective electrodes (indirect ISE) for testing sodium levels is the most common method for laboratory sodium measurement, direct (undiluted) ionselective electrode (direct ISE) whole blood methods have advantages in that they require smaller blood sample volumes and may be performed at or near the point of care, which may eliminate delays in specimen collection/processing, as well as communication of results to the treating physician [10]. Several small studies have investigated the agreement between direct and indirect ISE sodium measurements [11–15]. However, none have assessed the precision of the individual methods and the agreement between various methods in the profoundly hyponatremic range. Thus, our object was to assess the precision of indirect (plasma) and direct (whole blood) sodium measurements in the profoundly hyponatremic range and also assess the agreement between whole blood and plasma methods. Materials and methods Creation of low sodium pools for bias and precision studies Electrolyte-balanced lithium heparin samples submitted for arterial blood gas (ABG) analysis from patients in the intensive care unit, operating theaters, or emergency department were pooled and allowed to sit at 4 °C for 1 to 3 weeks to create a pool of lithium heparin whole blood with a sodium concentration of approximately 90 mmol/L. Patient samples were allowed to sit at 4 °C to reduce sodium values due to cellular redistribution of water and sodium in the samples. Lithium heparin whole blood from this low sodium pool was mixed with various amounts of fresh lithium heparin whole blood from residual ABG samples (collected on the same day as analysis) to create pools 1 through 3. Pools 1–3 were visibly hemolyzed at the time of analysis. Pool 4 was created from excess fresh lithium heparin ABG samples and was diluted with either 10% (pool 5) or 20% (pool 6) half-normal saline to create fresh whole blood pools with low sodium values. Pools 4–6 were not visibly hemolyzed. The objective was to produce whole blood pools large enough for repeat testing on each platform to measure the precision of each method and bias between the methods across a wide range of sodium concentrations. Sodium analysis Each pool was analyzed five times on a Radiometer ABL-827© blood gas analyzer (Radiometer, Bronshoj, Denmark) and five times on Abbott iSTAT© analyzer (Abbott Diagnostics, Princeton, NJ) using an EC3 cartridge. Pools were centrifuged at 4000 ×g for 3 min to produce plasma, which was analyzed five times on a Roche Cobas c501© (Roche Diagnostics, Indianapolis IN) chemistry analyzer. Plasma sodium on the Roche Cobas© was considered the comparative method for sodium. Patient samples 40 fresh (analyzed on the same day as collection) excess lithium heparin ABG samples from 36 patients in either the intensive care unit, the operating theater or the emergency department were analyzed on the Radiometer 827© and iSTAT©, and following centrifugation as described under Materials and methods on the Roche Cobas c501©. Statistics For each whole blood pool, the mean sodium and coefficient of variation—defined as standard deviation divided by the mean value times 100—were calculated to allow measurement of assay bias (mean whole blood compared to plasma sodium) and precision. Statistical significance of differences in the mean sodium values were assessed by one-way ANOVA. Bland–Altman plots of Radiometer and iSTAT whole blood versus plasma sodium were constructed to
visually compare these differences. Statistical analyses were carried out using GraphPad Instat (GraphPad Software, San Diego) and MedCalc Statistical Software, version 14.8.1 (MedCalc, Software bvba, Ostend, Belgium). Where appropriate, results with p-values of less than 0.05 were regarded as statistically significant. Ethical review This study was deemed exempt by the Mayo Clinic Institutional Review Board. Results Sodium bias and precision from pooled and manufactured samples The sample pools at various sodium concentrations were prepared and analyzed as described under Materials and methods section. The precision of each method was very good, with an observed coefficient of variation (CV) ≤ 1.3% for all the sample pools (with the exception of one low sodium pool on the Radiometer), spanning sodium concentrations of 104 to 133 mmol/L (Table 1). Using ANOVA to test differences between the platforms for individual pools, for sodium concentrations of b130 mmol/L (pools 1, 2, 3, 5, and 6), the mean whole blood sodium as measured on the Radiometer 800 blood gas analyzer was not significantly different than the mean plasma reference value as measured on the Roche Cobas c501 (Table 1). In contrast, at these low sodium concentrations, the mean whole blood sodium on the iSTAT was significantly higher, by ~3 mmol/L for most pools, than reference plasma sodium (Table 1). Sodium accuracy from individual patient samples To determine if the agreement trends observed on whole blood pools could be reproduced with individual fresh lithium heparin blood gas patient samples, 40 residual blood gas samples were tested for sodium as described under the Materials and methods section. A Bland–Altman plot of whole blood versus plasma sodium demonstrates greater overall accuracy for Radiometer 800 as opposed to iSTAT whole blood sodium: 34/40 whole blood samples on the Radiometer 800 were within 2 mmol/L of the plasma reference value compared to 28/40 iSTAT samples meeting this criterion (Fig. 1). Discussion In this study we describe the bias (compared to a comparative plasma measurement) and the precision of two whole blood sodium methods across a wide range of sodium levels. Both the whole blood and plasma methods demonstrated very good precision (around 1% CV), even in the profoundly hyponatremic measurement range (Na b 120 mmol/L on plasma method). Thus method imprecision is not the cause of ≥3 mM discrepancies between whole blood and plasma sodium observed in this and previous (see below) studies at very low values. However, we did find that the agreement between the methods varied with the degree of hyponatremia. In the profoundly hyponatremic measurement range we found a good agreement between the Radiometer blood gas analyzer and the plasma comparative method, while iSTAT measurements exhibited a consistent positive bias relative to plasma sodium (Table 1). In the near normal sodium range, the limits of agreement were narrower for the Radiometer blood gas analyzer than for iSTAT (Fig. 1), though with a tendency for the Radiometer to show a slight negative bias compared to plasma sodium. The magnitude of the mean bias and limits of agreement when comparing iSTAT whole blood and plasma sodium in the hyponatremic range may be clinically significant for clinicians attempting to plot accurate sodium correction trajectories in this range (where differences of ≥3 mM matter).
P. Geoghegan et al. / Clinical Biochemistry 48 (2015) 525–528
527
Table 1 Pool # = whole blood pool number. These whole blood pools, which have a range of low sodium values, were generated using combinations and dilutions of pooled electrolyte-balanced lithium heparin samples submitted for arterial blood gas (ABG) analysis. CV = co-efficient of variation; * indicates statistically significant difference (p b 0.05) between mean whole blood and mean plasma (Roche Cobas c501) sodium using one-way ANOVA. Mean bias = mean difference between whole blood (iSTAT or Radiometer) and reference (Cobas c501) sodium (n = 5); 95% CI = 95% confidence interval around mean bias. Roche Cobas c501
iSTAT
Radiometer 800
Pool #
Mean sodium (mmol/L)
CV (%)
Mean sodium (mmol/L)
CV (%)
Mean bias (95% CI)
Mean sodium (mmol/L)
CV (%)
Mean bias (95% CI)
1
104
1.1%
107*
0.7%
103
1.3%
2
110
0.4%
113*
0.4%
111
2.1%
3
111
0.4%
114*
0.4%
111
0.5%
4
133
0.0%
134*
0.0%
131*
0.4%
5
126
0.0%
129*
0.4%
126
0.4%
6
120
0.0%
123*
0.4%
3.4 (1.8–5.0) 3.0 (0.6–5.4) 3.0 (2.2–3.8) 1.0 (0.5–1.5) 2.6 (1.8–3.4) 3.4 (2.6–4.2)
119
0.5%
−0.2 (−1.8–1.4) 1.2 (−1.2–3.6) −0.6 (−1.4–0.2) −1.6 (−2.1 to −1.1) −0.4 (−1.2–0.4) −0.6 (−1.4–0.2)
Comparisons of our data to the existing published data on agreement between indirect and direct ISE methods of sodium measurement are complicated by differences in the study design [12], differences in the devices used [14], and the wide variation in the statistical approach to method comparison [13,15]. A single study comparing both Radiometer and iSTAT whole blood sodium to plasma sodium measured on a Roche automated chemistry analyzer concluded that the iSTAT device produced whole blood sodium results in closer agreement to plasma sodium than the Radiometer [11]. However, this study did not challenge the methods with samples in the profoundly hyponatremic range; nor did the study address whether method imprecision may contribute to differences between whole blood and plasma sodium at low values. Similarly, a single observational study of patients in an intensive care unit (ICU) setting reported very poor agreement between Radiometer whole blood sodium and plasma sodium in the profoundly hyponatremic range, in contrast with our finding of a good agreement between these methods in this range [13]. However, this study had several limitations including an absence of data on method precision in the profoundly hyponatremic range, and failure to control for the presence of factors such as hyperproteinemia or hyperlipidemia, both common in the ICU cohort, which cause spurious differences between indirect and direct ISE sodium measurements [16]. Our study specifically challenged the methods with low sodium values, and used replicates with pools to eliminate random error effects. However, a limitation of our study is the lack of availability of fresh
samples in the profoundly hyponatremic range, and the possibility of a matrix effect from aging (with resulting cellular shifts in water and sodium) or diluting (with alteration of blood protein, lipid and glucose concentration) heparinized whole blood samples which could differentially affect sodium methods. We tried to mitigate this limitation by [1] using different techniques for pooling (aged blood, diluting with halfnormal saline) and [2] by confirming general findings with a set of fresh patient samples. Since many of these matrix effects would be expected to differentially impact indirect and direct ISE measurements, the very good agreement between Radiometer (direct) and plasma (indirect) sodium suggests that matrix effects were not significant. Additionally, issues beyond precision such as cost and turnaround time may influence choice of testing platform. The iSTAT device can cost 7000 to 10,000 United States Dollars (USD) with current per test consumable costs of 5 to 6 USD (depending on the analytes tested), while Radiometer blood gas analyzers can cost 20,000 to 40,000 USD with current per test consumable costs of 1 to 1.5 USD, depending on the analyzer model. In terms of turnaround times at our institution, iSTAT is a 90 second test performed at the bedside; blood gas analyzer turnaround time is 8 to 10 min, while a “stat” plasma sodium takes 25 to 35 min, with a “routine” plasma sodium taking 90–120 min. The results of our study are significant for clinicians who wish to monitor changes in sodium levels frequently to facilitate accurate mapping of sodium trajectories in patients with profound hyponatremia. In these cases, the turnaround time for plasma reference methods may limit the utility of these standard measurements and clinicians may seek alternatives such as blood gas analyzer or iSTAT. They should be aware that while the precision of testing platforms is good in the profoundly hyponatremic range, our data suggest that the use of a Radiometer blood gas analyzer may be superior to iSTAT, since iSTAT showed a systematic positive bias relative to plasma sodium in this range. This is particularly important when direct (whole blood) sodium measurements will be compared periodically to plasma sodium to determine the rate of sodium correction. Conclusion
Fig. 1. Bland Altman plot of iSTAT© (□) and Radiometer 800© (♦) whole blood sodium compared to plasma sodium (measured on Roche Cobas c501©) as a reference. In total 40 residual blood gas samples were tested for sodium on the Radiometer 800 and iSTAT, and following the centrifugation tested for plasma sodium on the reference method.
When measuring sodium trajectories in the profoundly hyponatremic range (b120 mmol/L), clinicians should be aware that there are limitations to using different measurement methods interchangeably, since agreement between the methods is not perfect. Ideally, a single method of measurement should be used and, in general, all these methods will be acceptably precise in this range. However, if the turnaround times or other concerns will result in both direct and indirect sodium measurements being compared to delineate sodium trajectories, consideration should be given to the use of the Radiometer blood gas analyzer over
528
P. Geoghegan et al. / Clinical Biochemistry 48 (2015) 525–528
iSTAT, as our data suggest that the results will agree more closely with the plasma sodium values. Acknowledgments The authors gratefully acknowledge the entire M.E.T.R.I.C. research group for their support, which made this project possible. We also thank Mayo Clinic's Center for Translational Science Activities, Grant Number UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS), for their support. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the NIH. References [1] DeVita MV, Gardenswartz MH, Konecky A, Zabetakis PM. Incidence and etiology of hyponatremia in an intensive care unit. Clin Nephrol 1990;34:163–6. [2] Darmon M, Diconne E, Souweine B, Ruckly S, Adrie C, Azoulay E, et al. Prognostic consequences of borderline dysnatremia: pay attention to minimal serum sodium change. Crit Care 2013;17:R12. [3] Schrier RW, Bansal S. Diagnosis and management of hyponatremia in acute illness. Curr Opin Crit Care 2008;14:627–34. [4] Karp BI, Laureno R. Pontine and extrapontine myelinolysis: a neurologic disorder following rapid correction of hyponatremia. Medicine (Baltimore) 1993;72:359–73. [5] Vaidya C, Ho W, Freda BJ. Management of hyponatremia: providing treatment and avoiding harm. Cleve Clin J Med 2010;77:715–26.
[6] Nguyen MK. Quantitative approaches to the analysis and treatment of the dysnatremias. Semin Nephrol 2009;29:216–26. [7] Gankam Kengne F, Soupart A, Pochet R, Brion J-P, Decaux G. Re-induction of hyponatremia after rapid overcorrection of hyponatremia reduces mortality in rats. Kidney Int 2009;76:614–21. [8] Adrogué HJ, Madias NE. The challenge of hyponatremia. J Am Soc Nephrol 2012;23: 1140–8. [9] Tzamaloukas AH, Malhotra D, Rosen BH, Raj DSC, Murata GH, Shapiro JI. Principles of management of severe hyponatremia. J Am Health Assoc 2013;2:e005199. [10] Kost GJ, Ehrmeyer SS, Chernow B, Winkelman JW, Zaloga GP, Dellinger RP, et al. The laboratory–clinical interface: point-of-care testing. Chest 1999;115:1140–54. [11] Leino A, Kurvinen K. Interchangeability of blood gas, electrolyte and metabolite results measured with point-of-care, blood gas and core laboratory analyzers. Clin Chem Lab Med 2011;49:1187–91. [12] Budak YU, Huysal K, Polat M. Use of a blood gas analyzer and a laboratory autoanalyzer in routine practice to measure electrolytes in intensive care unit patients. BMC Anesthesiol 2012;12:17. [13] Jain A, Subhan I, Joshi M. Comparison of the point-of-care blood gas analyzer versus the laboratory auto-analyzer for the measurement of electrolytes. Int J Emerg Med 2009;2:117–20. [14] Bingham D, Kendall J, Clancy M. The portable laboratory: an evaluation of the accuracy and reproducibility of i-STAT. Ann Clin Biochem 1999;36(Pt 1):66–71. [15] Papadea C, Foster J, Grant S, Ballard SA, Iv JCC, Michael W, Purohit DM, Carolina S. Evaluation of the i-STAT portable clinical analyzer for point-of-care blood testing in the intensive care units of a University Children's Hospital. 2002;32:231–43. [16] Liamis G, Liberopoulos E, Barkas F, Elisaf M. Spurious electrolyte disorders: a diagnostic challenge for clinicians. Am J Nephrol 2013;38:50–7.
Contents lists available at ScienceDirect
Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem
Agreement between whole blood and plasma sodium measurements in profound hyponatremia☆,☆☆,★,★★ Pierce Geoghegan a,b,⁎, Christopher D. Koch c, Amy M. Wockenfus c, Andrew M. Harrison b,d, Yue Dong b,e, Kianoush B. Kashani b,f, Brad S. Karon c a
Department of Anaesthesia and Critical Care, Tallaght Hospital, Dublin, Ireland Multidisciplinary Epidemiology and Translational Research in Intensive Care (M.E.T.R.I.C.), Mayo Clinic, Rochester, MN, USA Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN, USA d Medical Scientist Training Program, Mayo Clinic, Rochester, MN, USA e Division of Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester, MN, USA f Division of Nephrology and Hypertension, Mayo Clinic, Rochester, MN, USA b c
a r t i c l e
i n f o
Article history: Received 10 January 2015 Received in revised form 2 March 2015 Accepted 3 March 2015 Available online 13 March 2015 Keywords: Hyponatremia Sodium measurement Agreement
a b s t r a c t Objectives: We compared two different methods of whole blood sodium measurement to plasma sodium measurement using samples in the profoundly hyponatremic range (Na b 120 mmol/L). Design and methods: Whole blood pools with a range of low sodium values were generated using combinations and dilutions of pooled electrolyte-balanced lithium heparin samples submitted for arterial blood gas analysis. Each pool was analyzed five times on a Radiometer 827 blood gas analyzer and iSTAT analyzer. Pools were centrifuged to produce plasma, which was analyzed five times on a Roche Cobas c501 chemistry analyzer. An additional 40 fresh (analyzed on day of collection) excess lithium heparin arterial blood gas samples from 36 patients were analyzed on the Radiometer 827, iSTAT, and Cobas c501 as described above. The setting was a tertiary referral center. Blood samples were collected from a combination of patients in the intensive care unit, operating theaters and emergency room. Results: All methods demonstrated excellent precision, even in the profoundly hyponatremic measurement range (Na b 120 mmol/L using a plasma reference method). However, agreement between the methods varied with the degree of hyponatremia. In the profoundly hyponatremic range, Radiometer whole blood sodium values were nearly identical to plasma reference sodium, while iSTAT whole blood sodium showed a consistent positive bias relative to plasma sodium in this range. Conclusion: If whole blood direct sodium measurements are compared with plasma sodium in profoundly hyponatremic patients consideration should be given to the use of Radiometer blood gas analyzers over iSTAT since the latter shows a positive bias relative to a plasma comparative method. © 2015 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Introduction Hyponatremia is a common electrolyte abnormality in hospitalized patients [1] and dysnatremia is an independent risk factor for hospital mortality [2]. Profound hyponatremia (sodium b 120 mmol/L) can lead to cerebral edema and correction of the hyponatremia is necessary to prevent or treat associated neurological complications or death [3]. However, excessively rapid correction of profound hyponatremia is well established to be associated with a risk of osmotic demyelination ☆ Institution: This work was performed at Mayo Clinic in Rochester, MN. ☆☆ Financial support: None. ★ Conflicts of interest: None. ★★ MeSH headings: Hyponatremia, blood gas analysis. ⁎ Corresponding author at: Department of Anaesthesia and Critical Care, Tallaght Hospital, Dublin, Ireland. E-mail address: [email protected] (P. Geoghegan).
syndrome (ODS), which may itself cause substantial morbidity or mortality [4]. Targeted rates of sodium correction are often difficult to achieve in profound hyponatremia and expert opinion has consistently emphasized the unpredictability of response to treatment [5,6]. Early identification or anticipation of non-optimal correction is important because there is convincing evidence from animal models that minimizing the duration of non-optimal correction reduces the risk of subsequent ODS and mortality [7]. Despite significant attention to the diagnostic workup in profound hyponatremia, there is little evidence-based guidance on how clinicians should monitor response to treatment. Many investigators have recommended very frequent monitoring of electrolytes during therapy [8,9], and others have even suggested the monitoring of sodium as frequently as every one to 2 h during acute therapy [5]. This raises the challenge of determining the optimal method and frequency of monitoring. Surprisingly, this issue has not
http://dx.doi.org/10.1016/j.clinbiochem.2015.03.001 0009-9120/© 2015 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
526
P. Geoghegan et al. / Clinical Biochemistry 48 (2015) 525–528
attracted significant attention. While the use of indirect ion-selective electrodes (indirect ISE) for testing sodium levels is the most common method for laboratory sodium measurement, direct (undiluted) ionselective electrode (direct ISE) whole blood methods have advantages in that they require smaller blood sample volumes and may be performed at or near the point of care, which may eliminate delays in specimen collection/processing, as well as communication of results to the treating physician [10]. Several small studies have investigated the agreement between direct and indirect ISE sodium measurements [11–15]. However, none have assessed the precision of the individual methods and the agreement between various methods in the profoundly hyponatremic range. Thus, our object was to assess the precision of indirect (plasma) and direct (whole blood) sodium measurements in the profoundly hyponatremic range and also assess the agreement between whole blood and plasma methods. Materials and methods Creation of low sodium pools for bias and precision studies Electrolyte-balanced lithium heparin samples submitted for arterial blood gas (ABG) analysis from patients in the intensive care unit, operating theaters, or emergency department were pooled and allowed to sit at 4 °C for 1 to 3 weeks to create a pool of lithium heparin whole blood with a sodium concentration of approximately 90 mmol/L. Patient samples were allowed to sit at 4 °C to reduce sodium values due to cellular redistribution of water and sodium in the samples. Lithium heparin whole blood from this low sodium pool was mixed with various amounts of fresh lithium heparin whole blood from residual ABG samples (collected on the same day as analysis) to create pools 1 through 3. Pools 1–3 were visibly hemolyzed at the time of analysis. Pool 4 was created from excess fresh lithium heparin ABG samples and was diluted with either 10% (pool 5) or 20% (pool 6) half-normal saline to create fresh whole blood pools with low sodium values. Pools 4–6 were not visibly hemolyzed. The objective was to produce whole blood pools large enough for repeat testing on each platform to measure the precision of each method and bias between the methods across a wide range of sodium concentrations. Sodium analysis Each pool was analyzed five times on a Radiometer ABL-827© blood gas analyzer (Radiometer, Bronshoj, Denmark) and five times on Abbott iSTAT© analyzer (Abbott Diagnostics, Princeton, NJ) using an EC3 cartridge. Pools were centrifuged at 4000 ×g for 3 min to produce plasma, which was analyzed five times on a Roche Cobas c501© (Roche Diagnostics, Indianapolis IN) chemistry analyzer. Plasma sodium on the Roche Cobas© was considered the comparative method for sodium. Patient samples 40 fresh (analyzed on the same day as collection) excess lithium heparin ABG samples from 36 patients in either the intensive care unit, the operating theater or the emergency department were analyzed on the Radiometer 827© and iSTAT©, and following centrifugation as described under Materials and methods on the Roche Cobas c501©. Statistics For each whole blood pool, the mean sodium and coefficient of variation—defined as standard deviation divided by the mean value times 100—were calculated to allow measurement of assay bias (mean whole blood compared to plasma sodium) and precision. Statistical significance of differences in the mean sodium values were assessed by one-way ANOVA. Bland–Altman plots of Radiometer and iSTAT whole blood versus plasma sodium were constructed to
visually compare these differences. Statistical analyses were carried out using GraphPad Instat (GraphPad Software, San Diego) and MedCalc Statistical Software, version 14.8.1 (MedCalc, Software bvba, Ostend, Belgium). Where appropriate, results with p-values of less than 0.05 were regarded as statistically significant. Ethical review This study was deemed exempt by the Mayo Clinic Institutional Review Board. Results Sodium bias and precision from pooled and manufactured samples The sample pools at various sodium concentrations were prepared and analyzed as described under Materials and methods section. The precision of each method was very good, with an observed coefficient of variation (CV) ≤ 1.3% for all the sample pools (with the exception of one low sodium pool on the Radiometer), spanning sodium concentrations of 104 to 133 mmol/L (Table 1). Using ANOVA to test differences between the platforms for individual pools, for sodium concentrations of b130 mmol/L (pools 1, 2, 3, 5, and 6), the mean whole blood sodium as measured on the Radiometer 800 blood gas analyzer was not significantly different than the mean plasma reference value as measured on the Roche Cobas c501 (Table 1). In contrast, at these low sodium concentrations, the mean whole blood sodium on the iSTAT was significantly higher, by ~3 mmol/L for most pools, than reference plasma sodium (Table 1). Sodium accuracy from individual patient samples To determine if the agreement trends observed on whole blood pools could be reproduced with individual fresh lithium heparin blood gas patient samples, 40 residual blood gas samples were tested for sodium as described under the Materials and methods section. A Bland–Altman plot of whole blood versus plasma sodium demonstrates greater overall accuracy for Radiometer 800 as opposed to iSTAT whole blood sodium: 34/40 whole blood samples on the Radiometer 800 were within 2 mmol/L of the plasma reference value compared to 28/40 iSTAT samples meeting this criterion (Fig. 1). Discussion In this study we describe the bias (compared to a comparative plasma measurement) and the precision of two whole blood sodium methods across a wide range of sodium levels. Both the whole blood and plasma methods demonstrated very good precision (around 1% CV), even in the profoundly hyponatremic measurement range (Na b 120 mmol/L on plasma method). Thus method imprecision is not the cause of ≥3 mM discrepancies between whole blood and plasma sodium observed in this and previous (see below) studies at very low values. However, we did find that the agreement between the methods varied with the degree of hyponatremia. In the profoundly hyponatremic measurement range we found a good agreement between the Radiometer blood gas analyzer and the plasma comparative method, while iSTAT measurements exhibited a consistent positive bias relative to plasma sodium (Table 1). In the near normal sodium range, the limits of agreement were narrower for the Radiometer blood gas analyzer than for iSTAT (Fig. 1), though with a tendency for the Radiometer to show a slight negative bias compared to plasma sodium. The magnitude of the mean bias and limits of agreement when comparing iSTAT whole blood and plasma sodium in the hyponatremic range may be clinically significant for clinicians attempting to plot accurate sodium correction trajectories in this range (where differences of ≥3 mM matter).
P. Geoghegan et al. / Clinical Biochemistry 48 (2015) 525–528
527
Table 1 Pool # = whole blood pool number. These whole blood pools, which have a range of low sodium values, were generated using combinations and dilutions of pooled electrolyte-balanced lithium heparin samples submitted for arterial blood gas (ABG) analysis. CV = co-efficient of variation; * indicates statistically significant difference (p b 0.05) between mean whole blood and mean plasma (Roche Cobas c501) sodium using one-way ANOVA. Mean bias = mean difference between whole blood (iSTAT or Radiometer) and reference (Cobas c501) sodium (n = 5); 95% CI = 95% confidence interval around mean bias. Roche Cobas c501
iSTAT
Radiometer 800
Pool #
Mean sodium (mmol/L)
CV (%)
Mean sodium (mmol/L)
CV (%)
Mean bias (95% CI)
Mean sodium (mmol/L)
CV (%)
Mean bias (95% CI)
1
104
1.1%
107*
0.7%
103
1.3%
2
110
0.4%
113*
0.4%
111
2.1%
3
111
0.4%
114*
0.4%
111
0.5%
4
133
0.0%
134*
0.0%
131*
0.4%
5
126
0.0%
129*
0.4%
126
0.4%
6
120
0.0%
123*
0.4%
3.4 (1.8–5.0) 3.0 (0.6–5.4) 3.0 (2.2–3.8) 1.0 (0.5–1.5) 2.6 (1.8–3.4) 3.4 (2.6–4.2)
119
0.5%
−0.2 (−1.8–1.4) 1.2 (−1.2–3.6) −0.6 (−1.4–0.2) −1.6 (−2.1 to −1.1) −0.4 (−1.2–0.4) −0.6 (−1.4–0.2)
Comparisons of our data to the existing published data on agreement between indirect and direct ISE methods of sodium measurement are complicated by differences in the study design [12], differences in the devices used [14], and the wide variation in the statistical approach to method comparison [13,15]. A single study comparing both Radiometer and iSTAT whole blood sodium to plasma sodium measured on a Roche automated chemistry analyzer concluded that the iSTAT device produced whole blood sodium results in closer agreement to plasma sodium than the Radiometer [11]. However, this study did not challenge the methods with samples in the profoundly hyponatremic range; nor did the study address whether method imprecision may contribute to differences between whole blood and plasma sodium at low values. Similarly, a single observational study of patients in an intensive care unit (ICU) setting reported very poor agreement between Radiometer whole blood sodium and plasma sodium in the profoundly hyponatremic range, in contrast with our finding of a good agreement between these methods in this range [13]. However, this study had several limitations including an absence of data on method precision in the profoundly hyponatremic range, and failure to control for the presence of factors such as hyperproteinemia or hyperlipidemia, both common in the ICU cohort, which cause spurious differences between indirect and direct ISE sodium measurements [16]. Our study specifically challenged the methods with low sodium values, and used replicates with pools to eliminate random error effects. However, a limitation of our study is the lack of availability of fresh
samples in the profoundly hyponatremic range, and the possibility of a matrix effect from aging (with resulting cellular shifts in water and sodium) or diluting (with alteration of blood protein, lipid and glucose concentration) heparinized whole blood samples which could differentially affect sodium methods. We tried to mitigate this limitation by [1] using different techniques for pooling (aged blood, diluting with halfnormal saline) and [2] by confirming general findings with a set of fresh patient samples. Since many of these matrix effects would be expected to differentially impact indirect and direct ISE measurements, the very good agreement between Radiometer (direct) and plasma (indirect) sodium suggests that matrix effects were not significant. Additionally, issues beyond precision such as cost and turnaround time may influence choice of testing platform. The iSTAT device can cost 7000 to 10,000 United States Dollars (USD) with current per test consumable costs of 5 to 6 USD (depending on the analytes tested), while Radiometer blood gas analyzers can cost 20,000 to 40,000 USD with current per test consumable costs of 1 to 1.5 USD, depending on the analyzer model. In terms of turnaround times at our institution, iSTAT is a 90 second test performed at the bedside; blood gas analyzer turnaround time is 8 to 10 min, while a “stat” plasma sodium takes 25 to 35 min, with a “routine” plasma sodium taking 90–120 min. The results of our study are significant for clinicians who wish to monitor changes in sodium levels frequently to facilitate accurate mapping of sodium trajectories in patients with profound hyponatremia. In these cases, the turnaround time for plasma reference methods may limit the utility of these standard measurements and clinicians may seek alternatives such as blood gas analyzer or iSTAT. They should be aware that while the precision of testing platforms is good in the profoundly hyponatremic range, our data suggest that the use of a Radiometer blood gas analyzer may be superior to iSTAT, since iSTAT showed a systematic positive bias relative to plasma sodium in this range. This is particularly important when direct (whole blood) sodium measurements will be compared periodically to plasma sodium to determine the rate of sodium correction. Conclusion
Fig. 1. Bland Altman plot of iSTAT© (□) and Radiometer 800© (♦) whole blood sodium compared to plasma sodium (measured on Roche Cobas c501©) as a reference. In total 40 residual blood gas samples were tested for sodium on the Radiometer 800 and iSTAT, and following the centrifugation tested for plasma sodium on the reference method.
When measuring sodium trajectories in the profoundly hyponatremic range (b120 mmol/L), clinicians should be aware that there are limitations to using different measurement methods interchangeably, since agreement between the methods is not perfect. Ideally, a single method of measurement should be used and, in general, all these methods will be acceptably precise in this range. However, if the turnaround times or other concerns will result in both direct and indirect sodium measurements being compared to delineate sodium trajectories, consideration should be given to the use of the Radiometer blood gas analyzer over
528
P. Geoghegan et al. / Clinical Biochemistry 48 (2015) 525–528
iSTAT, as our data suggest that the results will agree more closely with the plasma sodium values. Acknowledgments The authors gratefully acknowledge the entire M.E.T.R.I.C. research group for their support, which made this project possible. We also thank Mayo Clinic's Center for Translational Science Activities, Grant Number UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS), for their support. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the NIH. References [1] DeVita MV, Gardenswartz MH, Konecky A, Zabetakis PM. Incidence and etiology of hyponatremia in an intensive care unit. Clin Nephrol 1990;34:163–6. [2] Darmon M, Diconne E, Souweine B, Ruckly S, Adrie C, Azoulay E, et al. Prognostic consequences of borderline dysnatremia: pay attention to minimal serum sodium change. Crit Care 2013;17:R12. [3] Schrier RW, Bansal S. Diagnosis and management of hyponatremia in acute illness. Curr Opin Crit Care 2008;14:627–34. [4] Karp BI, Laureno R. Pontine and extrapontine myelinolysis: a neurologic disorder following rapid correction of hyponatremia. Medicine (Baltimore) 1993;72:359–73. [5] Vaidya C, Ho W, Freda BJ. Management of hyponatremia: providing treatment and avoiding harm. Cleve Clin J Med 2010;77:715–26.
[6] Nguyen MK. Quantitative approaches to the analysis and treatment of the dysnatremias. Semin Nephrol 2009;29:216–26. [7] Gankam Kengne F, Soupart A, Pochet R, Brion J-P, Decaux G. Re-induction of hyponatremia after rapid overcorrection of hyponatremia reduces mortality in rats. Kidney Int 2009;76:614–21. [8] Adrogué HJ, Madias NE. The challenge of hyponatremia. J Am Soc Nephrol 2012;23: 1140–8. [9] Tzamaloukas AH, Malhotra D, Rosen BH, Raj DSC, Murata GH, Shapiro JI. Principles of management of severe hyponatremia. J Am Health Assoc 2013;2:e005199. [10] Kost GJ, Ehrmeyer SS, Chernow B, Winkelman JW, Zaloga GP, Dellinger RP, et al. The laboratory–clinical interface: point-of-care testing. Chest 1999;115:1140–54. [11] Leino A, Kurvinen K. Interchangeability of blood gas, electrolyte and metabolite results measured with point-of-care, blood gas and core laboratory analyzers. Clin Chem Lab Med 2011;49:1187–91. [12] Budak YU, Huysal K, Polat M. Use of a blood gas analyzer and a laboratory autoanalyzer in routine practice to measure electrolytes in intensive care unit patients. BMC Anesthesiol 2012;12:17. [13] Jain A, Subhan I, Joshi M. Comparison of the point-of-care blood gas analyzer versus the laboratory auto-analyzer for the measurement of electrolytes. Int J Emerg Med 2009;2:117–20. [14] Bingham D, Kendall J, Clancy M. The portable laboratory: an evaluation of the accuracy and reproducibility of i-STAT. Ann Clin Biochem 1999;36(Pt 1):66–71. [15] Papadea C, Foster J, Grant S, Ballard SA, Iv JCC, Michael W, Purohit DM, Carolina S. Evaluation of the i-STAT portable clinical analyzer for point-of-care blood testing in the intensive care units of a University Children's Hospital. 2002;32:231–43. [16] Liamis G, Liberopoulos E, Barkas F, Elisaf M. Spurious electrolyte disorders: a diagnostic challenge for clinicians. Am J Nephrol 2013;38:50–7.
