This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Story, D. A. Articles by Bellomo, R. Search for Related Content PubMed PubMed Citation Articles by Story, D. A. Articles by Bellomo, R. Related Collections Critical Care Physiology

---

CRITICAL CARE AND TRAUMA

The Effect of Albumin Concentration on Plasma Sodium and Chloride Measurements in Critically Ill Patients

David A. Story, MD, FANZCA^*, Hiroshi Morimatsu, MD^*, Moritoki Egi, MD^*, and Rinaldo Bellomo, MD, FJFICM^*

From the *Departments of Anaesthesia and Intensive Care, Austin Health, Heidelberg, Victoria, Australia; and Department of Surgery, University of Melbourne, Austin Health.

Address correspondence to David A. Story, Department of Anaesthesia, Austin Hospital, Studley Rd., Heidelberg, Vic., 3084, Australia. Address e-mail to david.story{at}austin.org.au.

	Abstract

Top Abstract Introduction METHODS RESULTS DISCUSSION APPENDIX A: CLINICAL PHYSICAL... APPENDIX B: ASSAY OUTLINES REFERENCES

 
BACKGROUND: We tested the hypothesis that the difference between indirect and direct sodium assays would be related to the plasma albumin concentration. Further, we proposed that differences between indirect and direct chloride assays might be explained by interference from other plasma constituents, particularly bicarbonate, and possibly albumin.

METHODS: We studied 300 critically ill patients at the time of admission to the intensive care unit (ICU) and compared each patient’s plasma sodium and chloride measurements from a central laboratory assay (indirect electrode) and an ICU blood gas machine assay (direct electrode).

RESULTS: The central laboratory sodium measurement was, on average, 2.1 mmol/L more than the ICU assay, limits of agreement 1.8–2.4 mmol/L greater, P < 0.001. The central laboratory chloride measurement was, on average, 1 mmol/L less than the ICU assay (limits of agreement 1.3–0.7 mmol/L less, P < 0.001). All correlations between the assay differences and plasma constituents were weak except for a moderately strong correlation between differences in sodium measurements and albumin. The difference in plasma sodium concentration between the assays (central laboratory – ICU) increased as the plasma concentration albumin decreased (difference = 6.2–0.16 albumin (g/L); P < 0.001, r = –0.46, r² = 0.22).

CONCLUSIONS: The central laboratory and ICUs assays are analytically, statistically, and clinically different for both sodium and chloride. Unless taken into account, the differences could be large enough in hypoalbuminemic populations (such as critically ill patients) to affect clinical diagnosis and decision making.

	Introduction

Top Abstract Introduction METHODS RESULTS DISCUSSION APPENDIX A: CLINICAL PHYSICAL... APPENDIX B: ASSAY OUTLINES REFERENCES

 
Plasma sodium and chloride measurements are used for a variety of clinical assessments. In the clinical setting, there are two methods for measuring the principal electrolytes in plasma (1). Both use ion-selective electrodes. One type incorporates pre-analytic dilution (indirect assay) and is often, but not always, used in high throughput central hospital laboratories (1). The other type (direct assay) does not have predilution and is used most commonly in blood gas machines. Both the direct and the indirect methods measure electrolyte activities (Appendices A and B) in plasma water, but report electrolyte concentrations in total plasma, assuming a normal solid phase (protein and lipids) of about 7%. The direct assay should report total plasma electrolyte concentrations that have a predictable and fixed relationship with the electrolyte concentration in plasma water, regardless of the solid phase component (1). The indirect methodology, however, has an apparent design flaw (1): With a normal electrolyte concentration in plasma water the indirect assay should report a reduced total plasma electrolyte concentration when the solid phase is increased. Further, the indirect assay may report increased plasma electrolyte concentration when the solid phase is decreased (2). These problems have been examined only with sodium assays, where, with hyperlipidemia or increased plasma proteins, the indirect method reports (from the plasma water perspective) pseudohypernatremia, and pseudonormonatremia with decreased plasma proteins (2). This has led clinical chemists to recommend the direct assay for sodium in increased solid phase conditions such as hyperlipidemia (1,3).

In a previous study of critically ill patients (4), we found that plasma sodium concentrations from a direct assay were significantly lower than those from an indirect assay. Because critically ill patients are often hypoalbuminemic, this could have been due to the different handling of solid phase corrections by the two methodologies. However, we also found that direct chloride concentrations were slightly, but significantly, larger than indirect values, the opposite of the sodium discrepancy and contrary to the expected change for a solid phase decrease. After further consideration of findings from that study (4), we tested the hypothesis that the difference between the indirect (central laboratory) and direct intensive care unit (ICU) blood gas machine sodium assays would be related to the plasma albumin concentration. Further, we proposed that differences between the central laboratory and ICU chloride assays may be explained by interference from other plasma constituents, particularly bicarbonate (5,6), and possibly albumin (1). We used the same clinical chemistry database as in our previous study (4).

	METHODS

Top Abstract Introduction METHODS RESULTS DISCUSSION APPENDIX A: CLINICAL PHYSICAL... APPENDIX B: ASSAY OUTLINES REFERENCES

 
Data were retrospectively collected from ICU records at the Austin Hospital, Melbourne, Australia. All samples were taken from arterial lines. The Austin Health Human Research Ethics Committee waived the need for informed consent. Data were collected on samples taken on admission to the ICU where the records indicated that simultaneous arterial samples were sent for blood gas and general chemistry analysis. Arterial blood samples were collected in heparinized blood–gas syringes (Rapidlyte, Chiron Diagnostics, East Walpole, MA) and analyzed in a bench-top blood–gas analyzer (Ciba Corning 865, Ciba Corning Diagnostics, Medfield, MA) in the ICU laboratory (ICU assay). The sodium and chloride assays were direct ion-specific electrodes (Appendices A and B) (3,6–8).

A further sample was drawn at the same time from the same arterial sampling point using a vacuum technique with lithium heparin tubes or clot-activating tubes (Vacuette, Greiner labortechnik, Kremsmunster, Austria). These samples were sent to the Division of Laboratory Medicine (central laboratory assay). Plasma and serum underwent a multicomponent analysis (Hitachi 747, Roche Diagnostics, Sydney, Australia). The sodium and chloride assays used indirect ion-specific electrodes (Appendices A and B) (1,7). We also recorded the plasma albumin concentration, which is reported without the total protein concentration.

A paired t-test was performed comparing the central laboratory and ICU measurements. A Bland–Altman analysis (9) was performed for the central laboratory minus ICU difference; we defined 4 mmol/L as an acceptable upper limit for limits of agreement for sodium and 4.5 mmol/L for chloride (10). Correlation analysis was performed between plasma pH, bicarbonate, lactate, and albumin and the differences between the central laboratory (indirect) and ICU (direct) assays for sodium and chloride. Regression analysis was performed between the albumin concentration and the difference between the central laboratory (indirect) and ICU (direct) sodium measurements. We used GraphPad Prism Version 4 software (GraphPad Software, San Diego, CA). A P value of <0.05 was considered statistically significant.

	RESULTS

Top Abstract Introduction METHODS RESULTS DISCUSSION APPENDIX A: CLINICAL PHYSICAL... APPENDIX B: ASSAY OUTLINES REFERENCES

 
We studied clinical chemistry from 300 critically ill patients (Table 1). The results for the sodium and chloride assays were quite different. Bland–Altman analysis of the central laboratory result minus the ICU sodium measurement had limits of agreement of –1.6 to 6.2 mmol/L. This upper limit exceeded an acceptable difference of 4 mmol/L (10). The mean difference was statistically significant: 2.1 mmol/L (95% CI: 1.8–2.4 mmol/L, P < 0.001). Bland–Altman analysis of the central laboratory result minus the ICU chloride measurement had limits of agreement of –6.5 to 4.5 mmol/L. The lower limit exceeded an acceptable difference of 4.5 mmol/L (10). The mean difference was statistically significant: –1.0 mmol/L (95% CI: –1.3 to –0.7 mmol/L, P < 0.001).

All correlations (Table 2) between the assay differences and plasma constituents were weak except for a moderately strong correlation (11) between differences in sodium measurements and albumin. Linear regression analysis (Fig. 1) showed that the difference in plasma sodium concentration between the assays increased as the plasma concentration albumin decreased (Y = 6.2–0.16X; Y intercept 95% CI: 5.3–7.1; slope 95% CI: –0.12 to –0.19; P < 0.001, r² = 0.22).

View larger version (15K): [in this window] [in a new window]   Figure 1. Regression analysis, with 95% confidence intervals, of the difference between the central laboratory (indirect) and intensive care unit (direct) sodium assays (difference) and the plasma albumin concentration (albumin). The difference is 6.2–0.16 (albumin). The difference was zero when the plasma albumin was 40 g/L.

For 39 samples (13%), the central laboratory estimate for sodium was equal to or more than 135 mmol/L and the ICU estimate was <135 mmol/L, possible pseudnormonatremia. For 20 samples (7%) the central laboratory estimate was more than 145 mmol/L and the ICU estimate was less than or equal to 145 mmol/L, possible pseudohypernatremia.

	DISCUSSION

Top Abstract Introduction METHODS RESULTS DISCUSSION APPENDIX A: CLINICAL PHYSICAL... APPENDIX B: ASSAY OUTLINES REFERENCES

 
In an extension of a previous study (4), we found that the difference between plasma sodium concentration estimates from the central laboratory and ICU blood gas machines increased as the plasma albumin decreased in critically ill patients. Although we predicted that the same should happen with the corresponding chloride assays (1), this was not the case. Further, the differences between direct and indirect assays for both sodium and chloride were not related to other elements in plasma chemistry, particularly bicarbonate (5,6). The central laboratory and ICU assays, for both sodium and chloride, were statistically different, and had limits of agreement exceeding an acceptable difference of 4 and 4.5 mmol/L, respectively (10).

Many clinical chemists (1,3,6,12,13) conclude that the concentrations of electrolytes in plasma water (3,6) are more physiologically important than the total concentration in plasma (Appendix A). For sodium, increased lipids and paraproteinemias may produce "pseudohyponatremia" that is the low total plasma sodium leads to the false impression that the concentration of sodium in plasma water is also decreased (13) (Appendices A and B). This may lead clinicians to under-estimate the sodium concentration in plasma water, with clinical consequences (13).

When the plasma solids are increased and the water phase is decreased, clinical chemists have proposed that the direct (ICU) sodium and chloride assays should be used (1), because the plasma solid content is assumed constant (Appendix B) in the direct assay calculations. Therefore, the total plasma sodium and chloride concentrations (Appendix A) reported by direct assays should maintain a fixed relationship (93%) with the sodium and chloride concentrations in plasma water when plasma solids are increased. There are only a few references in the clinical chemistry literature (2,14,15), and none in the critical care or anesthesia literature, to the point that direct assays of total plasma sodium are likely to maintain a fixed relationship with the plasma concentration when plasma solids, including albumin, are decreased. Many critically ill patients have decreased albumin.

On the basis of our data and the indirect assay methodology (1), as plasma albumin decreases, the central laboratory assay will more closely approximate the plasma water sodium (3) concentration, because water will constitute a larger proportion of the total plasma volume. The estimate from the ICU assay will continue to be 93% of the ionized concentration and will deviate more from the plasma water value than the corresponding central laboratory value. This may seem to argue for using the central laboratory assay in patients with decreased plasma albumin, because the central laboratory measurement is closer than the ICU assay to what we really want to know: the plasma water concentration of sodium. However, using the central assay, changes in plasma solids can lead to the false conclusion that plasma water sodium has changed, whereas the ICU assay would not. Further, a value for the plasma water sodium can be calculated from the direct assay result by using the fixed value (7%) for plasma solids (Appendices A and B).

These features of the assays should apply to chloride (1); however, this has not been investigated. We found that there is no clear relationship between the difference between the chloride assays and the plasma albumin concentration. It is unclear why this is so, but the relationship between the plasma electrolyte activity (Appendices A and B) and the total plasma concentration of an electrolyte is complex, as is the electrode chemistry (6). Nor could we find a relationship between the chloride difference and other plasma constituents, including bicarbonate, which has been previously implicated in discrepancies between assays (5,6). Other ions that may affect chloride assays (due to hydration energy) include heparin, salicylate, thiocyanate, bromide, and iodide (6). We did not, however, have data on these ions. By another mechanism (Donnan exclusion failure), cations, including drugs with quaternary ammonium, can also affect chloride assays (6). Again, we have no data on this.

Our results are clinically important because, worldwide, thousands of critically ill patients have plasma sodium and chloride measurements taken each day. Reviewers in internal medicine journals (16,17) have incorrectly stated that the problem of psuedohyponatremia had been largely eliminated due to (presumed) almost universal use of direct assays, and that pseudohypernatremia does not exist. In fact, indirect assays are still widely used around the world (1), and we found pseudohypernatremia in 7% of our patients. Further 13% had pseudonormonatremia. Based on the values of our 300 patients, if the central laboratory sodium estimates were used and sodium concentrations <130 mmol/L are triggers for physician intervention (10), then in seven patients hyponatremia would have been missed. Further, in a patient with an albumin of 22 g/L, the ICU sodium estimate would be expected to be 3 mmol/L less than the central laboratory estimate. Using the central laboratory sodium estimate, the osmolality would be over-estimated by 6 mmol/L (18), and water balance would be under-estimated by 1 L (19). Combining the central laboratory sodium and chloride estimates, the anion gap would be overestimated by about 4 mmol/L.

We conclude that the differences in sodium measurements made with indirect and direct methods were correlated with plasma albumin, and increased with the severity of hypoalbuminemia. This is consistent with the theoretical effects of altered solid phase constituents on the two methodologies (1,3). The differences between the two methods were sufficient to affect clinical tonicity decisions. Because the findings are consistent with the theory behind solid phase effects, they support a preference for the direct method in hypoalbuminemia, such as that found in critical illness. Further, caution is required when interpreting indirect (central laboratory) sodium values when albumin concentrations are reduced.

Our findings concerning chloride concentrations, however, were inconsistent with the theoretical effect of a reduced plasma solid phase on the indirect chloride assay (1). The values obtained by subtracting the direct from the indirect chloride measurements were unrelated to albumin concentrations (or to pH, bicarbonate and lactate), and were highly variable. The fact that the chloride difference is the opposite of the sodium difference, and that only the sodium difference increases with worsening hypoalbuminemia, has major implications for acid–base interpretation, especially for the scanning tools used to detect unmeasured ions. Anion gap and strong ion gap may differ slightly between the two methodologies, even in populations with normal solid phase concentrations, due to a chloride discrepancy. More importantly, there will be increasing differences as albumin decreases (about 1.5 mmol/L per 10 g/L albumin decrease), due to increasing direct versus indirect sodium discrepancy. Unless taken into account, the differences could be large enough in hypoalbuminemic populations (such as critically ill patients) to affect clinical diagnosis and decision making.

	APPENDIX A: CLINICAL PHYSICAL CHEMISTRY TERMS

Top Abstract Introduction METHODS RESULTS DISCUSSION APPENDIX A: CLINICAL PHYSICAL... APPENDIX B: ASSAY OUTLINES REFERENCES

 
Concentration in Plasma

Also called the total electrolyte concentration (3,6), this is the reported variable of mmol of sodium or chloride per liter of total plasma that is used clinically. Total plasma includes both water (93%) and solids (7%) consisting of protein and lipids (1,3).

Electrolyte Concentration in Plasma Water

This variable is more physiologically important than the total electrolyte concentration (3,6). Because virtually all the plasma sodium and chloride is in this water phase, which is about 93% of plasma volume, the plasma water concentrations will be about 108% of the total plasma concentrations. Clinically, the total sodium and chloride concentration is usually assumed to reflect the plasma water sodium concentration (1).

Plasma Water Electrolyte Activity

Activity is an effective concentration (20); it is a thermodynamic variable and is important for understanding chemical reactions. Activity is related to the concentration by the activity coefficient: Activity = concentration x activity coefficient. The value of the activity coefficient depends on several factors, including the ion in question (sodium) and the chemistry of the surrounding solution. Plasma has particularly complex chemistry (6). Ion-selective electrodes, including the hydrogen ion (pH) electrode, measure the activity of solutions. The direct and indirect sodium assays use different approaches to convert the sodium and chloride activities in plasma water to total sodium concentration. Many clinical chemists would prefer that clinicians use activities, as we effectively do with pH, but recognize that such expectations are unrealistic (3,6,20).

	APPENDIX B: ASSAY OUTLINES

Top Abstract Introduction METHODS RESULTS DISCUSSION APPENDIX A: CLINICAL PHYSICAL... APPENDIX B: ASSAY OUTLINES REFERENCES

 
Indirect Sodium and Chloride Ion-Specific Electrodes (Central Laboratory)

A known volume of serum or plasma is diluted with a known volume of potassium solution before analysis (1). The diluted plasma has an activity coefficient of one. This means the measured sodium activity equals the concentration of sodium in total plasma. The concentration of sodium in total plasma is reported.

Direct Sodium and Chloride Ion-Specific Electrodes (ICU Blood Gas Machine)

The electrodes (1,8) come into contact with heparinized blood. The electrode measures the activity of sodium and chloride in plasma water. The machine software converts the activity to the concentration of sodium in plasma water by dividing the activity by the activity coefficient. The concentration of sodium in plasma water is converted to the concentration of sodium in total plasma by multiplying the concentration of sodium in plasma water by the proportion of total plasma that is water, assumed to be a fixed value of about 0.933. The concentration of sodium and chloride in total plasma is reported.

	Footnotes

 
Accepted for publication December 28, 2006.

Supported by the Research Funds of the Departments of Anaesthesia and Intensive Care, Austin Health.

	REFERENCES

Top Abstract Introduction METHODS RESULTS DISCUSSION APPENDIX A: CLINICAL PHYSICAL... APPENDIX B: ASSAY OUTLINES REFERENCES

 

1. Scott MG, Klutts JS. Electrolytes and blood gases. In: Burtis CA, Ashwood ER, Bruns DE, eds. Tietz textbook of clinical chemistry and molecular diagnostics, 4th ed. Philadelphia: Elsevier, 2005:983–1018.
2. Lang T, Prinsloo P, Broughton AF, et al. Effect of low protein concentration on serum sodium measurement: pseudohypernatraemia and pseudonormonatraemia! Ann Clin Biochem 2002;39:66–7.[Web of Science][Medline]
3. Burnett RW, Covington AK, Fogh-Andersen N, et al. International Federation of Clinical Chemistry and Laboratory Med (IFCC). IFCC Scientific Division Working Group on Selective Electrodes. Recommendations for measurement of and conventions for reporting sodium and potassium by ion-selective electrodes in undiluted serum, plasma or whole blood. Clin Chem Lab Med 2000;38:1065–71.[Web of Science][Medline]
4. Morimatsu H, Rocktaschel J, Bellomo R, et al. Comparison of point-of-care versus central laboratory measurement of electrolyte concentrations on calculations of the anion gap and the strong ion difference. Anesthesiology 2003;98:1077–84.[Web of Science][Medline]
5. Dimeski G, Clague AE. Bicarbonate interference with chloride-ion-selective electrodes. Clin Chem 2004;50:1106–7.[Free Full Text]
6. Rayana MC, Burnett RW, Covington AK, et al. Recommendation for measuring and reporting chloride by ISEs in undiluted serum, plasma or blood. Clin Chem Lab Med 2006;44:346–52.[Web of Science][Medline]
7. Levy GB. Determination of sodium with ion-selective electrodes. Clin Chem 1981;27:1435–8.[Abstract/Free Full Text]
8. Burnett RW, Covington AK, Fogh-Andersen N, et al. International Federation of Clinical Chemistry and Laboratory Med (IFCC). Scientific Division Working Group on Selective Electrodes. Recommendations for nomenclature, definitions and conventions. Clin Chem Lab Med 2000;38:363–70.[Web of Science][Medline]
9. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10.[Web of Science][Medline]
10. Linnet K, Boyd JC. Selection and analytical evaluation of methods – with statistical techniques. In: Burtis CA, Ashwood ER, Bruns DE, eds. Tietz textbook of clinical chemistry and molecular diagnostics, 4th ed. Philadelphia: Elsevier, 2005:353–407.
11. Myles PS, Gin T. Statistical methods for anaesthesia and intensive care. Oxford: Butterwoth-Heinemann, 2000.
12. Kulpmann WR. Determination of sodium with ion-selective electrodes: a new method or a new quantity? J Clin Chem Clin Biochem 1990;28:813–5.[Web of Science][Medline]
13. Weisberg LS. Pseudohyponatremia: a reappraisal. Am J Med 1989;86:315–8.[Web of Science][Medline]
14. Broughton PM, Smith SC, Buckley BM. Calibration of direct ion-selective electrodes for plasma Na⁺ to allow for the influence of protein concentration. Clin Chem 1985;31:1765–6.[Web of Science][Medline]
15. Kau N, Gunther M, Fahnenstich H, et al. Measurements of sodium and potassium in newborns and prematures by ion-selective electrodes and flame photometry: influence of lipid and protein content of blood. Acta Anaesthesiol Scand Suppl 1995;107:107–11.[Medline]
16. Adrogue HJ, Madias NE. Hyponatremia. N Engl J Med 2000;342:1581–9.[Free Full Text]
17. Oster JR, Singer I. Hyponatremia, hyposmolality, and hypotonicity: tables and fables. Arch Intern Med 1999;159:333–6.[Abstract/Free Full Text]
18. Rose BD. New approach to disturbances in the plasma sodium concentration. Am J Med 1986;81:1033–40.[Web of Science][Medline]
19. Scheinkestel CD, Tuxen DV, Cade JF, Shann FA. Fluid management of shock in critically-ill patients. Med J Aust 1989;150:508–10, 513–7.
20. Lewenstam A. Electric potential measured, concentration reported: how to get mmols from mV. Scand J Clin Lab Invest Suppl 1996;224:135–9.[Medline]

This article has been cited by other articles:

				M Sherlock, E O'Sullivan, A Agha, L A Behan, D Owens, F Finucane, D Rawluk, W Tormey, and C J Thompson Incidence and pathophysiology of severe hyponatraemia in neurosurgical patients Postgrad. Med. J., April 1, 2009; 85(1002): 171 - 175. [Abstract] [Full Text] [PDF]

				B J Jones and P J Twomey Relationship of the absolute difference between direct and indirect ion selective electrode measurement of serum sodium and the total protein concentration J. Clin. Pathol., May 1, 2008; 61(5): 645 - 647. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Story, D. A. Articles by Bellomo, R. Search for Related Content PubMed PubMed Citation Articles by Story, D. A. Articles by Bellomo, R. Related Collections Critical Care Physiology