Comparison of serum sodium levels measured by blood gas analyzer and biochemistry autoanalyzer in patients with hyponatremia, eunatremia, and hypernatremia
a b s t r a c t
Background: blood gas analyzer (BGA) electrolyte measurements are frequently used in emergency departments (EDs) pending biochemistry laboratory autoanalyzer (BLA) results. There is lack of data in the literature in terms of agreement of these 2 measurement methods of sodium. We aimed to comprehensively evaluate the agree- ment in hyponatremia, eunatremia, and hypernatremia groups.
Methods: Retrospectively, adult subjects who presented to ED of a tertiary care teaching hospital and had simul- taneous BGA and BLA results were included in the study. Blood pairs were grouped into hyponatremia, eunatremia, and hypernatremia according to BLA results. Agreement of sodium measurements between the methods were evaluated by Bland-Altman plots and Passing and Bablok regression analysis.
Results: A total of 2557 blood pairs (1326 males [51.8%]) were included. Median age of the patients was 66 years (18-103). The numbers of patients with hyponatremia, eunatremia, and hypernatremia were 487 (19%), 1943 (76%), and 127 (5%), respectively. The minimum and maximum serum sodium levels measured by biochemistry analyzer were 106 and 171 mmol/L, respectively. The Pearson linear correlation coefficient between BGA and BLA for sodium measurements were 0.574, 0.358, and 0.562 in hyponatremia, eunatremia, and hypernatremia groups, respectively. The absolute mean difference for the 3 groups was greater than 4 mmol/L. Biochemistry lab- oratory autoanalyzer tended to measure serum sodium higher than BGA in all sodium groups. Passing and Bablok regression analysis showed significant differences between the 2 methods in all sodium groups.
Conclusion: This is the first comprehensive evaluation of agreement between BGA and BLA in distinct sodium
groups. Significant differences should be taken into account when these patients are managed in the ED.
(C) 2016
Introduction
Hyponatremia and hypernatremia are frequently encountered elec- trolyte disorders in emergency department (ED) [1]. Depending on the severity of the presenting illness, in many occasions, blood gas samples are studied in ED patients in addition to conventional central laboratory biochemistry analysis. The main impetus behind this practice is quicker attainment of the results of blood gas analyzer (BGA) compared with biochemistry study. Serum sodium values are among the routinely re- ported electrolytes included in a typical BGA result. Several studies in the past evaluated the agreement in serum electrolyte values between these 2 methods [2-7]. However, to the best of our knowledge, of stud- ies evaluated the difference in serum sodium levels between these 2 measurements, only 1 study assessed the agreement in patients with overt hyponatremia or hypernatremia [8]. However, this study did not
? Conflict of interest: No conflict of interest.
* Sakarya Egitim ve Arastirma Hastanesi, Nefroloji Klinigi, 54100, Sakarya, Turkey. Tel.: +90 5058899885.
E-mail address: yalcinsolakmd@gmail.com.
report the agreement in separate hyponatremia, eunatremia, and hypernatremia groups. To our opinion, we need this data of agreement in patients who are at the edges of serum sodium range most. Because these patients may need urgent therapeutic decision while pending bio- chemistry results, it is important to know the extent to which blood gas and biochemistry measurements are in line with each other from a clin- ical perspective.
Hence, we aimed to evaluate the agreement in serum Sodium levels between BGA and central biochemistry laboratory autoanalyzer (BLA) machines in a large sample of patients separately in hyponatremia, eunatremia, and hypernatremia in the ED.
Materials and methods
This was a retrospective study. Subjects older than 18 years who pre- sented to the ED of a tertiary care teaching hospital and had simulta- neous blood gas analysis and biochemistry analysis results were included in the study. Between November 2015 and January 2016 (3 months), we screened all ED presentations and included adult patients who had both blood gas analysis and biochemistry analysis results at
http://dx.doi.org/10.1016/j.ajem.2016.04.037
0735-6757/(C) 2016
that presentation. Patients who had only one of these measurements were excluded. We exerted great care to ensure that blood gas and bio- chemistry analyses were performed simultaneously or were close tem- porally to each other as much as possible. All adult patients presented during the aforementioned period were included irrespective of their presentation complaint or comorbid conditions.
The study protocol was approved by the university ethics board. Sodi- um levels were measured from venous whole blood samples by a BGA de- vice (ABL 700; Radiometer, Copenhagen, Denmark). Venous blood samples were centrifuged at the central biochemistry laboratory, and serum samples were then analyzed with a central laboratory biochemistry autoanalyzer (ICT [ISE] Module of ARCHITECT c16000; Abbott Laborato- ries, Abbott Park, IL). Blood samples were transported to the central labo- ratory in standard biochemistry tubes. Dedicated heparinized blood gas syringes were used to receive and transport venous blood gas samples.
Hyponatremia, eunatremia, and hypernatremia were defined as bio- chemistry serum sodium measurement results of less than 135, 135 to 145, and greater than 145 mmol/L, respectively.
Statistical analysis
Statistical analyses were performed via IBM SPSS Statistics version 20 (IBM SPSS Statistics for Windows; IBM Corp, Armonk, NY). We grouped all participants in terms of serum sodium values as hyponatremia, eunatremia, and hypernatremia. Serum sodium levels measured by bio- chemistry analyzer and BGA were not distributed normally; thus, we per- formed Wilcoxon signed test to compare the medians of differences between BGA and biochemistry analyzer in 3 sodium groups. We per- formed Bland-Altman plots to analyze the agreement between the 2 mea- surement methods in each sodium group. Because both sodium values measured by both devices and differences between the measurements in each sodium group were not normally distributed, we performed Pass- ing and Bablok regression model to understand the agreement. We also constructed a number of graphics to show correlation and level of agree- ment between the biochemistry and blood gas analysis sodium results. Pearson correlation was used to show correlation between the 2 mea- surement methods. Biochemistry autoanalyzer sodium measurement was accepted as criterion standard, and thus, measurement differences between the both devices were calculated as sodium measured by bio- chemistry autoanalyzer minus sodium measured by BGA. Because differ- ences between the 2 measurement methods had taken both positive and negative results depending on the device that measured sodium higher, we calculated absolute difference not taking into account the negativity and positivity, as well. P b .05 was accepted as significant.
Results
A total of 2557 blood pairs from 2557 patients (1326 males [51.8%]) were included. Median age of the patients was 66 years (18-103). The num- bers of patients with hyponatremia, eunatremia, and hypernatremia were 487 (19%), 1943 (76%), and 127 (5%), respectively. The minimum and max- imum serum sodium levels measured by biochemistry autoanalyzer were 106 and 171 mmol/L, respectively. The Pearson linear correlation coefficient between blood gas and biochemistry Laboratory analysis for sodium mea- surements were 0.574, 0.358, and 0.562 in hyponatremia, eunatremia, and hypernatremia groups, respectively (Fig. 1). The Table shows agreement be- tween the biochemistry laboratory autoanalyzer and BGA to classify blood samples as hyponatremia, eunatremia, autoanalyzer and hypernatremia. Frequency of sodium measurements that biochemistry result is less than, equal, or greater than the BGA in each sodium group is depicted in Fig. 2.
Hyponatremic patients
A total of 487 blood sample pairs were available in eunatremic group. Biochemistry autoanalyzer mean sodium value was 130.41 +-
4.40 mmol/L (range, 106-134), and median was 132 mmol/L. Blood
Fig. 1. A to C, Scatter and dot graphic of correlation between BGA and BLA for each sodium group. Blue and red lines denote regression line and line of equality (reference line), respectively. A, B, and C denote hyponatremia, eunatremia, and hypernatremia, respectively. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article).
gas analyzer mean sodium value was 130.48 +- 7.92 (range, 103-179), and median was 130 mmol/L. The number of positive differences was
Number of blood samples categorized by BLA and BGA as hyponatremia, eunatremia, and hypernatremia
BGA |
BLA |
and median was 149 mmol/L. Blood gas analyzer sodium mean was |
|||
Hyponatremia |
Eunatremia |
Hypernatremia |
146.36 +- 7.52 mmol/L (range, 130-166), and median was 146 mmol/L. |
distributed normally so we used Wilcoxon Signed Rank Test to test the sta- tistical difference between sodium levels of the 2 methods. Biochemistry autoanalyzer sodium mean was 150.11 +- 4.93 mmol/L (range, 146-171),
Hyponatremia |
360 (73.9%) |
552 (28.4%) |
6 (4.7%) |
Eunatremia |
114 (23.4%) |
1207 (62.1%) |
55 (43.3%) |
Hypernatremia |
13 (2.7%) |
184 (9.5%) |
66 (52.0%) |
Total |
487 (100%) |
1943 (100%) |
127 (100%) |
265 (54.4%; mean, 4.01 +- 2.35 mmol/L), the number of negative differ- ences was 194 (39.8%; mean, 5.68 +- 6.65 mmol/L), and the number of no differences was 28 (5.8%) (Figs. 3 and 4). Mean difference was
-0.07 (range, -49 to 13; Wilcoxon test, P = .0123). Mean value of ab- solute difference between the measurement methods was 4.44 +- 4.73 mmol/L (median was 3 mmol/L). Bland-Altman plot showing agree- ment of sodium measurement by the 2 devices was shown in Fig. 5A. Passing and Bablok regression model showed that there was a signifi- cant deviation from linearity in the agreement of both measurement methods (P b .01), with a slope of 0.4 and intercept of 91.2 (Fig. 6A).
Eunatremic patients
A total of 1943 blood sample pairs were available in the eunatremic group. Biochemistry autoanalyzer mean sodium value was 139.40 +-
2.62 mmol/L (range, 135-145), and median was 139 mmol/L. Blood gas analyzer mean sodium was 137.82 +- 5.97 mmol/L (range, 96-170), and median was 137 mmol/L. The number of positive differences was 1291 (66.4%; mean difference, 4.58 +- 2.95 mmol/L), the number of negative differences was 542 (27.9%; mean difference, 5.25 +- 4.81 mmol/L), and the number of no differences was 110 (5.7%) (Figs. 3 and 4). Mean differ- ence was 1.58 (range, -32 to 39; Wilcoxon test, P b .001). Mean absolute difference between the 2 methods in the whole sample was 4.51 +- 3.68 (median, 4 mmol/L). Bland-Altman plot showing agreement of sodium measurement by the two devices was shown in Fig. 5B. Passing and Bablok regression model showed that there was a significant deviation from linearity in the agreement of both measurement methods (P b .01) with a slope of 0.4 and intercept of 91.2 (Fig. 6B).
Hypernatremic patients
A total of 127 blood sample pairs were available in hypernatremic group. Differences between the 2 measurement methods were not
Measurement differences between the 2 devices were as follows: number of positive differences, 95 (mean, 6.46 +- 4.00); number of negative differ- ences, 24 (mean, 5.75 +- 4.48); number of no differences, 8 (Figs. 2 and 4). Mean difference was 3.74 (range, -18 to 23) (P b .001). Mean of absolute differences in the whole group was 5.92 +- 4.23. Median value of differ- ence in the whole sample was 5 mmol/L. Bland-Altman plot showing agreement of sodium measurement by the 2 devices was shown in Fig. 5C. Passing and Bablok regression model showed that there was a sig- nificant deviation from linearity in the agreement of both measurement methods (P = .05) with a slope of 0.4 and intercept of 78.8 (Fig. 6C).
Discussion
The main findings of this study were as follows: First, serum sodium measurements by BGA and biochemistry autoanalyzer differed signifi- cantly in all sodium groups. Second, biochemistry autoanalyzer tended to measure serum sodium higher compared with BGA in most patients in each sodium group. Third, mean absolute sodium difference between the measurement methods showed an upward trend from hyponatremia to hypernatremia. This meant that compared with hyponatremic patients, 2 measurement methods yielded more dispa- rate sodium results in hypernatremic patients. The median absolute dif- ference in the eunatremic group was 4 mmol/L. Moreover, range of difference was quite broad in hyponatremic and hypernatremic patients reaching up to 23 mmol/L. This difference may not lead to significant changes in the treatment plans; however, a difference at this magnitude may lead to different therapeutic actions in patients on the 2 edges of serum sodium spectrum. Lastly, to the best of our knowledge, this study is the first comprehensive analysis of sodium measurement pat- terns of BGA and biochemistry autoanalyzer in hyponatremic, eunatremic, and hypernatremic subgroups.
Several previous studies evaluated the agreement between BGA and central biochemistry laboratory autoanalyzer (BLA) measurements of serum sodium [2-7]. The great majority of these studies included pa- tients with normal or near-normal serum sodium levels. The results usually showed statistically significant differences between the 2 mea- surement methods [2-7].
Fig. 2. Bar graph showing frequency of Biochemistry laboratory analyzer and blood gas analyzer sodium measurements that equal to or greater than each other. BIO: Biochemistry labo- ratory autoanalyzer, BG: blood gas analyzer.
Fig. 3. A to C, Box and whisker graphics of sodium distribution measured by BLA and BGA in hyponatremia, eunatremia, and hypernatremia groups. A, B, and C denote hyponatremia, eunatremia, and hypernatremia, respectively.
The US Clinical Laboratory Improvement Amendment (USCLIA) finds
+-4 mmol/L sodium bias as acceptable [9]. The mean differences in all 3 sodium groups in our study exceeded the USCLIA recommendations.
However, most of the previous studies comparing two measurement methods in terms of sodium found a statistically significant difference. Mean difference (serum biochemistry sodium – BGA sodium) were found to be 4.07 [4], 4.9 [5], 2.1 [7], and 5.17 mmol/L [6] in several stud- ies. In all of these studies, BLA consistently measured serum sodium levels higher than those measured in whole blood by BGA [4-7]. Simi- larly, in all 3 groups in our study, there was a similar tendency as well.
Fig. 4. A to C, Histogram depicting frequency of differences between the 2 measurement methods. A, B, and C denote hyponatremia, eunatremia, and hypernatremia, respectively. Less than, equal, or greater than the BGA in each sodium group.
To the best of our knowledge, all studies included patients from critical care units. Only a few studies included patients from EDs. Bloom et al [10] evaluated the difference between sodium measured by a blood gas and a BLA. The authors found a mean difference of 3.36 mmol/L.
Some studies included patients with hyponatremia or hypernatremia by biochemistry autoanalyzer; however, sample sizes were relatively small, and the authors did not report biases in hyponatremia or hypernatremia groups separately [3,11].
One of the largest studies conducted in this field by Mirzazadeh et al
[8] evaluated records of approximately 11 000 paired samples. The range of the sodium of included patients was reported as 94 to 192
Fig. 5. A to C, Bland-Altman plots for each sodium group. Blue line shows mean of the observed differences with red dotted limits of agreement lines (+1.96 and -1.96 SD). Pink dotted line is regression line with accompanying confidence interval limits. A, B, and C denote hyponatremia, eunatremia, and hypernatremia, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
Fig. 6. A to C, Passing and Bablok regression lines for each sodium group. Blue line depicts regression line, whereas red dotted lines are confidence interval limits. A, B, and C denote hyponatremia, eunatremia, and hypernatremia, respectively.
with BLA (median, 139 mmol/L; mean, 140.05 mmol/L). The sodium differ- ences were distributed normally in +-5%. They also found that there was a small systematic bias between the 2 methods at different levels of sodium, namely, -0.57 mmol/L. Mean of the differences in whole study cohort in our study was 1.4 mmol/L. Systematic bias between the 2 methods in our study when we separately evaluated sodium groups were 3.7, 1.6, and
-0.1 mmol/L in hyponatremia, eunatremia, and hypernatremia groups, re- spectively. The numbers of sample pairs qualified as hyponatremia, eunatremia, and hypernatremia by both methods were 2353 (81.8%), 8060 (84.4%), and 2352 (81%), respectively, in the study by Mirzazadeh
et al [8]. These figures were 360 (73.9%), 1207 (62.1%), and 66 (52%) blood pairs for hyponatremia, eunatremia, and hypernatremia, respectively. However, the authors of this study did not report mean/median values and details of agreement separately in sodium groups.
In conclusion, the results of this study showed that BGA systematically and consistently measures sodium lower compared with BLA. We, for the first time in the literature, evaluated agreement of 2 measurement methods comprehensively in a large sample of blood sample pairs. Mean absolute sodium difference between the measurement methods showed an upward trend from hyponatremia to hypernatremia. In all 3 sodium groups, observed differences between the methods were higher than USCLIA recommendations. Clinicians dealing with patients whose serum sodium levels are incorporated in therapeutic decisions should take into account these likely differences between the measurement methods.
References
- Brouns SH, Dortmans MK, Jonkers FS, Lambooij SL, Kuijper A, Haak HR. Hyponatraemia in elderly emergency department patients: a marker of frailty. Neth J Med 2014;72(6):311-7.
- Zhang JB, Lin J, Zhao XD. Analysis of bias in measurements of potassium, sodium and hemoglobin by an emergency department-based blood gas analyzer relative to hos- pital laboratory autoanalyzer results. PLoS One 2015;10(4):e0122383. http://dx.doi. org/10.1371/journal.pone.0122383 [PONE-D-14-39797 [pii].
- Uyanik M, Sertoglu E, Kayadibi H, Tapan S, Serdar MA, Bilgi C, et al. Comparison of blood gas, electrolyte and metabolite results measured with two different Blood gas analyzers and a core laboratory analyzer. Scand J Clin Lab Invest 2015;75(2): 97-105. http://dx.doi.org/10.3109/00365513.2014.981854.
- Chacko B, Peter JV, Patole S, Fleming JJ, Selvakumar R. Electrolytes assessed by point- of-care testing–are the values comparable with results obtained from the central laboratory? Indian J Crit Care Med 2011;15(1):24-9. http://dx.doi.org/10.4103/ 0972-5229.78219 [IJCCM-15-24 [pii].
- 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 pa- tients. BMC Anesthesiol 2012;12:17. http://dx.doi.org/10.1186/1471-2253-12-17 [1471-2253-12-17 [pii].
- 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(2):117-20. http://dx.doi.org/10.1007/s12245-009-0091-1.
- Morimatsu H, Rocktaschel J, Bellomo R, Uchino S, Goldsmith D, Gutteridge G. Com- parison of point-of-care versus central laboratory measurement of electrolyte con- centrations on calculations of the anion gap and the strong ion difference. Anesthesiology 2003;98(5):1077-84 [00000542-200305000-00009 [pii].
- Mirzazadeh M, Morovat A, James T, Smith I, Kirby J, Shine B. Point-of-care testing of electrolytes and calcium using blood gas analysers: it is time we trusted the results. Emerg Med J 2016;33(3):181-6. http://dx.doi.org/10.1136/emermed-2015- 204669emermed-2015-204669 [pii].
- USA Clinical laboratory Improvement Amendment. Regulations part 493 laboratory requirements, subpart I proficiency testing programs by specialty and subspecialty, section 493.931 routine chemistry; 2004.
- Bloom BM, Connor H, Benton S, Harris T. A comparison of measurements of sodium, potassium, haemoglobin and creatinine between an emergency department-based point-of-care machine and the hospital laboratory. Eur J Emerg Med 2014;21(4): 310-3. http://dx.doi.org/10.1097/MEJ.0000000000000076.
- Geoghegan P, Koch CD, Wockenfus AM, Harrison AM, Dong Y, Kashani KB, et al. Agreement between whole blood and plasma sodium measurements in profound hyponatremia. Clin Biochem 2015;48(7-8):525-8. http://dx.doi.org/10.1016/j. clinbiochem.2015.03.001 [S0009-9120(15)00073-9 [pii].