Context.—Elevated potassium concentrations due to in vitro hemolysis can lead to errors in diagnoses and treatment. Recently, we observed that potassium elevation in capillary samples appeared higher than expected, based on hemolytic index (H-index).

Objective.—To investigate the relation between potassium increase and H-index for capillary samples. As a control, the same analysis was performed for lactate dehydrogenase (LDH).

Design.—Potassium results of 332 760 venous and 2620 capillary samples were selected. For LDH, 135 974 venous and 999 capillary samples were included. Venous and capillary samples were differentiated by using patient age, as we perform mostly capillary blood sampling in children and venous sampling in adults. Results were obtained with Beckman-Coulter DxC800 analyzers.

Results.—The increase in potassium with increasing H-index was considerably higher for capillary samples than venous samples. Linear regression revealed a potassium increase of 0.38 mEq/L per increment in H-index for capillary samples, whereas a 0.17 mEq/L increase was found for venous samples. For LDH, no differences were found between venous and capillary samples.

Conclusions.—At identical H-index, capillary samples showed higher potassium elevations than venous samples. A possible explanation is that capillary sampling causes increased leakage of ions, such as potassium, from erythrocytes, compared with proteins such as hemoglobin and LDH. These results are especially important considering the increasing use of whole blood point-of-care analyzers, where the H-index is often not determined. Potassium results should therefore be interpreted with caution to avoid severe misdiagnosis of hypokalemia and hyperkalemia.

In vitro hemolysis is likely the most common preanalytic problem in laboratory medicine with a reported incidence of 3.3%.1 Plasma potassium concentrations are strongly affected by hemolysis owing to the more than 20-fold difference between intracellular and extracellular potassium levels.2 Consequently, spuriously elevated potassium concentrations caused by in vitro hemolysis can lead to misdiagnosis of hypokalemia and overdiagnosis of hyperkalemia, possibly resulting in serious errors in patient treatment.

Most modern-day clinical chemistry analyzers assess the extent of hemolysis automatically by determining the hemolytic index (H-index), a semiquantitative spectrophotometric measure that is directly related to the plasma free hemoglobin concentration.3,4 Based on the linear relation between potassium and H-index, correction factors might be derived to adjust the measured potassium level to the extent of hemolysis.315 Although this would produce representative potassium values in most cases, it fails in cases of intravascular hemolysis. Given that intravascular hemolysis is rare (less than 2% of the hemolytic samples, as shown for an Italian academic hospital6), potassium correction based on high H-indices is not justified in these cases. Alternatively, as we do in our laboratory, potassium results may be reported with a qualitative comment indicating the probable amount of potassium elevation due to hemolysis, thereby including the remark that this does not hold for in vivo hemolysis. However, comments may be overlooked by clinicians, for instance, in busy emergency departments.

During the daily clinical validation of laboratory results, we recently observed that potassium elevation in capillary samples appeared higher than expected based on H-index. Therefore, the relation between potassium concentration and H-index was further investigated for both capillary and venous samples. As a control, the elevation of L-lactate dehydrogenase (LDH, EC 1.1.1.27) activity as a function of H-index was also investigated for capillary and venous samples.

Venous and capillary samples were selected by using patient age, as we perform mostly capillary blood sampling in children and venous sampling in adults. Age-based selection criteria were 18 years or older and younger than 15 years for venous and capillary samples, respectively. Data were selected from the laboratory information system (GLIMS, MIPS Diagnostics Intelligence, Gent, Belgium) during a 4-year period. This resulted in 332 760 venous and 2620 capillary samples for the analysis of potassium versus H-index, and 135 974 venous and 999 capillary samples for LDH (Table). Capillary samples were obtained through heel or finger prick by using automated lancets and subsequently collected into pediatric tubes.16 Samples from the emergency department and intensive care units were excluded as these departments often draw blood directly from intravenous catheters. All samples were collected in lithium heparin gel tubes and were centrifuged for 4 minutes at 4190g within 1 hour after collection.

Summary of the Number of Included Venous and Capillary Samples Per Hemolytic Index as Applied for the Analysis of Potassium and Lactate Dehydrogenase (LDH)

Summary of the Number of Included Venous and Capillary Samples Per Hemolytic Index as Applied for the Analysis of Potassium and Lactate Dehydrogenase (LDH)
Summary of the Number of Included Venous and Capillary Samples Per Hemolytic Index as Applied for the Analysis of Potassium and Lactate Dehydrogenase (LDH)

All potassium, LDH, and H-index results were obtained on Beckman-Coulter DxC800 chemistry analyzers (Beckman-Coulter, Inc, Brea, California). Potassium concentrations were determined by using an indirect ion selective electrode, and LDH activity was measured according to the reference method of the International Federation of Clinical Chemistry (IFCC, lactate to pyruvate). H-indices ranged from 0 to 10, corresponding to plasma free hemoglobin concentrations of 0–500 mg/dL (0–0.31 mmol/L).3,15 Data were analyzed with Microsoft Access 2003 (Microsoft Corporation, Redmond, Washington) and SPSS 15.0 (SPSS, Chicago, Illinois).

The Table summarizes the number of samples used for the analysis of potassium and LDH as a function of H-index. The percentage of samples with detectable hemolysis (ie, H-index > 0) was 19% and 63% for venous and capillary samples, respectively. This corresponds well with the general knowledge that hemolysis is a more common problem in capillary than in venous sampling.17 

The Figure (a) shows mean potassium concentrations in venous and capillary samples versus H-index. At the same H-index, capillary samples clearly show higher potassium increases than venous samples. This effect is seen over the whole H-index range. Linear regression analysis revealed a 0.38 mEq/L (95% confidence interval, 0.36–0.40; conversion to SI units: 1 mEq/L  =  1 mmol/L) potassium increase per increment in H-index for capillary samples, which is more than twice as high as the 0.17 mEq/L (95% confidence interval, 0.16–0.17) increase found for venous samples. The latter result corresponds well with the potassium increase of 0.14 mEq/L per H-index increment described by Vermeer et al,3 thereby supporting the validity of the current method. The identical offset of 4.1 mEq/L indicates that there are no intrinsic differences in potassium levels between venous and capillary samples, which is in correspondence with previously published results.17 For LDH, no significant differences between venous and capillary samples were found (Figure, b).

Spurious potassium elevations due to in vitro hemolysis are well described in literature. However, differences between venous and capillary blood samples have not been published before. The results presented here clearly demonstrate a higher potassium increase per increment in H-index for capillary samples than for venous samples, while no differences were found for LDH. A likely explanation is that capillary sampling results in increased erythrocyte membrane stress, thereby causing an increased leakage of cations such as potassium, whereas proteins such as LDH and hemoglobin remain intracellular.18 In addition, tissue cells surrounding the blood vessels may also experience this increased membrane stress during capillary sampling, thereby further increasing the potassium levels in capillary blood. In contrast, complete lysis of either blood or tissue cells cannot explain the observed differences between capillary and venous samples, as this would also lead to an increase in LDH.

The described database method has several advantages over experimental approaches where, for instance, volunteers are recruited for simultaneous capillary and venous blood sampling, or where hemolysis is induced artificially (eg, by freeze-thaw cycles or by repeated aspiration through fine needles). Database analysis is very robust, since extreme values for potassium and LDH level out owing to the large number of included samples (Table). In addition, it does not require additional blood sampling, and experiments using volunteers would never have resulted in these large numbers and would therefore be less accurate.

The fact that our hospital is a 1042-bed tertiary care academic medical center, where patients often suffer multiple illnesses, may have influenced the presented results as these patients may have erythrocytes that are intrinsically more permeable owing to the underlying disease. However, this effect is most likely similar for both venous and capillary groups and cannot explain the described differences. Intrinsic differences between erythrocytes obtained from children and adults, which might have biased the results, are considered unlikely because of the identical relationships between potassium concentrations and H-index described for neonates and adults.19 In addition, higher potassium elevations than expected, based on H-index, were also observed in capillary samples of adults (data not shown). This supports our finding that capillary sampling results in higher potassium elevations than for venous sampling, at the same detectable level of hemolysis, irrespective of patient age.

The importance of our results is underlined when considering the increasing use of capillary sampling due to the continuing trend toward smaller sample volumes and the more widespread use of point-of-care testing. Importantly, point-of-care analyzers, as well as dedicated blood gas analyzers, generally do not measure H-indices. Automatic corrections or comments indicating possible potassium elevations in hemolytic samples can therefore not be generated when bedside testing is applied. Additionally, these methods do not require centrifugation of the blood before analysis, thereby giving no visual indication of hemolysis. To overcome this problem, Lippi et al20 recently described centrifugation of arterial blood gas samples directly after analysis and subsequent assessment of the level of hemolysis in the plasma. However, this might be difficult to implement in routine laboratory practice as it requires additional sample handling and altered sample logistics. In addition, it is unsuitable for samples measured by point-of-care analyzers, where only minimal amounts of blood are drawn and analyzed immediately.

As indicated above, at our hospital we include a comment with potassium results in hemolytic samples to indicate the level of potassium elevation due to in vitro hemolysis. Since the sampling method is not registered in our laboratory information system, these comments now include the statement that potassium elevation is approximately twice as high when capillary blood sampling is performed. Ideally, specific comments for capillary and venous samples should be generated automatically when the sampling method is registered by the phlebotomist. However, this might be difficult to implement in daily practice. In addition, it should be stressed that more factors can influence the linear relationship between potassium and the hemolytic index, including analyzer type, reagents, patient population, and sample logistics.1 Therefore, laboratories should determine the relationship between potassium and hemolysis for their own local setting and not blindly implement formulas from the literature, including the ones presented here.

The consequences of the presented results were broadly communicated within our hospital to increase the physicians' awareness about falsely elevated potassium levels in capillary sampling, especially in the context of point-of-care testing, where the hemolytic index cannot be determined, and therefore no automatic comments can be generated.

In summary, capillary samples show a higher potassium increase than venous samples at the same hemolytic index. Even minimal in vitro hemolysis results in a substantially higher potassium elevation in capillary samples. This is especially important when using whole blood point-of-care analyzers, for instance in emergency departments, where the level of hemolysis remains undetected. If unaware of this, serious errors in diagnoses and treatment may occur. Since more than half of all capillary samples are hemolytic (Table), each potassium result obtained using point-of-care analyzers should be interpreted with caution, even when results are within the reference range. In addition, standard correction factors for potassium, based on the hemolytic index, are not valid for capillary blood samples and may lead to severe misinterpretation of potassium status. The presented results are therefore relevant not only for laboratory professionals; they should be broadly communicated to all physicians in order to correctly interpret potassium levels in capillary samples.

Average potassium concentrations (a) and lactate dehydrogenase (LDH) activities (b) as a function of the hemolytic index (H-index) for venous (closed diamonds, straight line) and capillary (open circles, dotted line) samples. Linear regression analysis revealed the following equations, with the 95% confidence intervals for slope and intercept indicated in parentheses: Kvenous (mEq/L)  =  0.17 (0.16–0.17) × H-index + 4.1 (4.1–4.1); Kcapillary (mEq/L)  =  0.38 (0.36–0.40) × H-index + 4.1 (4.0–4.1); LDHvenous (U/L)  =  69.9 (68.2–74.5) × H-index + 245.8 (243.8–245.9); LDHcapillary (U/L)  =  65.2 (59.2–71.2) × H-index + 252.3 (243.9–260.7). Note that linear regression was performed by using all data points, whereas only mean values are shown in the figure for clarity. Therefore, regression lines seem to deviate at higher H-index owing to the lower number of data points at these indices.

Average potassium concentrations (a) and lactate dehydrogenase (LDH) activities (b) as a function of the hemolytic index (H-index) for venous (closed diamonds, straight line) and capillary (open circles, dotted line) samples. Linear regression analysis revealed the following equations, with the 95% confidence intervals for slope and intercept indicated in parentheses: Kvenous (mEq/L)  =  0.17 (0.16–0.17) × H-index + 4.1 (4.1–4.1); Kcapillary (mEq/L)  =  0.38 (0.36–0.40) × H-index + 4.1 (4.0–4.1); LDHvenous (U/L)  =  69.9 (68.2–74.5) × H-index + 245.8 (243.8–245.9); LDHcapillary (U/L)  =  65.2 (59.2–71.2) × H-index + 252.3 (243.9–260.7). Note that linear regression was performed by using all data points, whereas only mean values are shown in the figure for clarity. Therefore, regression lines seem to deviate at higher H-index owing to the lower number of data points at these indices.

Close modal

The authors thank Dirk Koppenaal, PhD, for his valuable contribution in retrieving the required data from the laboratory information system and for construction of the database.

1.
Lippi
G
,
Blanckaert
N
,
Bonini
P
, et al.
Haemolysis: an overview of the leading cause of unsuitable specimens in clinical laboratories
.
Clin Chem Lab Med
.
2008
;
46
(
6
):
764
772
.
2.
Scott
M
,
LeGrys
V
,
Klutts
J
.
Electrolytes and blood gases
.
In
:
Burtis
C
,
Ashwood
E
,
Bruns
D
,
eds
.
Tietz Textbook of Clinical Chemistry and Molecular Diagnostics. 4th ed
.
St Louis, MO
:
Elsevier-Saunders
;
2006
:
984
.
3.
Vermeer
HJ
,
Steen
G
,
Naus
AJ
,
Goevaerts
B
,
Agricola
PT
,
Schoenmakers
CH
.
Correction of patient results for Beckman Coulter LX-20 assays affected by interference due to hemoglobin, bilirubin or lipids: a practical approach
.
Clin Chem Lab Med
.
2007
;
45
(
1
):
114
119
.
4.
Plebani
M
,
Lippi
G
.
Hemolysis index: quality indicator or criterion for sample rejection
?
Clin Chem Lab Med
.
2009
;
47
(
8
):
899
902
.
5.
Brescia
V
,
Tampoia
M
,
Mileti
A
.
Evaluation of factitious hyperkalemia in hemolytic samples: impact of the mean corpuscular hemoglobin concentration
.
LabMed
.
2009
;
40
:
224
226
.
6.
Carraro
P
,
Servidio
G
,
Plebani
M
.
Hemolyzed specimens: a reason for rejection or a clinical challenge
?
Clin Chem
.
2000
;
46
(
2
):
306
307
.
7.
Dimeski
G
,
Clague
AE
,
Hickman
PE
.
Correction and reporting of potassium results in haemolysed samples
.
Ann Clin Biochem
.
2005
;
42
(
pt 2
):
119
123
.
8.
Hawkins
R
.
Variability in potassium/hemoglobin ratios for hemolysis correction
.
Clin Chem
.
2002
;
48
(
5
):
796
.
9.
Jay
DW
,
Provasek
D
.
Characterization and mathematical correction of hemolysis interference in selected Hitachi 717 assays
.
Clin Chem
.
1993
;
39
(
9
):
1804
1810
.
10.
Lippi
G
,
Luca Salvagno
G
,
Blanckaert
N
, et al.
Multicenter evaluation of the hemolysis index in automated clinical chemistry systems
.
Clin Chem Lab Med
.
2009
;
47
(
8
):
934
939
.
11.
Lippi
G
,
Salvagno
GL
,
Montagnana
M
,
Brocco
G
,
Guidi
GC
.
Influence of hemolysis on routine clinical chemistry testing
.
Clin Chem Lab Med
.
2006
;
44
(
3
):
311
316
.
12.
Mansour
MM
,
Azzazy
HM
,
Kazmierczak
SC
.
Correction factors for estimating potassium concentrations in samples with in vitro hemolysis: a detriment to patient safety
.
Arch Pathol Lab Med
.
2009
;
133
(
6
):
960
966
.
13.
Meites
S
.
Reproducibly simulating hemolysis, for evaluating its interference with chemical methods
.
Clin Chem
.
1973
;
19
(
11
):
1319
.
14.
Owens
H
,
Siparsky
G
,
Bajaj
L
,
Hampers
LC
.
Correction of factitious hyperkalemia in hemolyzed specimens
.
Am J Emerg Med
.
2005
;
23
(
7
):
872
875
.
15.
Steen
G
,
Vermeer
HJ
,
Naus
AJ
,
Goevaerts
B
,
Agricola
PT
,
Schoenmakers
CH
.
Multicenter evaluation of the interference of hemoglobin, bilirubin and lipids on Synchron LX-20 assays
.
Clin Chem Lab Med
.
2006
;
44
(
4
):
413
419
.
16.
Demir
AY
,
van Solinge
WW
,
Kemperman
H
.
Handling of and direct sampling from primary barcode-labeled pediatric tubes on Vitros clinical chemistry analyzers integrated into an enGen work cell
.
Clin Chem
.
2005
;
51
(
5
):
920
921
.
17.
Kupke
IR
,
Kather
B
,
Zeugner
S
.
On the composition of capillary and venous blood serum
.
Clin Chim Acta
.
1981
;
112
(
2
):
177
185
.
18.
Johnson
RM
.
Membrane stress increases cation permeability in red cells
.
Biophys J
.
1994
;
67
(
5
):
1876
1881
.
19.
Jeffery
J
,
Sharma
A
,
Ayling
RM
.
Detection of haemolysis and reporting of potassium results in samples from neonates
.
Ann Clin Biochem
.
2009
;
46
(
pt 3
):
222
225
.
20.
Lippi
G
,
Ippolito
L
,
Fontana
R
.
Prevalence of hemolytic specimens referred for arterial blood gas analysis
.
Clin Chem Lab Med
.
2011
;
49
(
5
):
931
932
.

Author notes

The authors have no relevant financial interest in the products or companies described in this article.

Portions of this manuscript were presented as a poster at the conference of the Dutch Society of Clinical Chemistry (NVKC), Veldhoven, The Netherlands, April 14–15, 2011, and as a poster at the conference of the International Federation of Clinical Chemistry (IFCC), Berlin, Germany, May 16–19, 2011.