The preanalytic phase of laboratory medicine is the

Transcription

1 Correction Factors for Estimating Potassium Concentrations in Samples With In Vitro Hemolysis A Detriment to Patient Safety Mai M. H. Mansour, MS; Hassan M. E. Azzazy, PhD, DABCC; Steven C. Kazmierczak, PhD, DABCC Context. Correction factors have been proposed for estimating true potassium concentrations in blood samples with evidence of in vitro hemolysis. Objective. We used 2 different models of true (ie, nonsimulated) in vitro hemolysis to evaluate the clinical utility of correction factors for estimating potassium concentrations in samples with evidence of in vitro hemolysis. Design. Potassium correction factors were derived using 2 different models. In model 1, potassium and plasma hemoglobin were measured with the Hitachi 747 analyzer in 132 paired blood samples, with each pair consisting of 1 sample with evidence of hemolysis and 1 without, collected during the same phlebotomy procedure. The change in measured potassium concentration was plotted versus the change in plasma hemoglobin concentration for each pair of samples. In model 2, the potassium levels of blood samples and the corresponding hemolytic index values were measured with the Beckman LX20 analyzer. Potassium concentrations at the 10th, 25th, 50th, 75th, and 90th percentiles were calculated for each hemolysis index category. Results. From our 2 models, we derived correction factors expressing an increase in potassium concentration of 0.51 and 0.40 meq/l for every increase in plasma hemoglobin concentration of 0.1 g/dl. These correction factors are much higher than those reported in studies that simulated in vitro hemolysis by freeze-thaw lysis or osmotic disruption of whole blood. Conclusions. Use of correction factors for estimating the true potassium concentration in samples with evidence of in vitro hemolysis is not recommended. Derivation of correction factors by using samples with nonsimulated in vitro hemolysis suggests that the actual increase in potassium in hemolyzed samples is much higher than that reported in previous studies that produced hemolysis with artificial means. (Arch Pathol Lab Med. 2009;133: ) The preanalytic phase of laboratory medicine is the phase that can most adversely affect the accuracy of test results. 1 Poor specimen quality is recognized as the most frequent source of error in this phase of the testing process, with in vitro hemolysis accounting for the approximately 40% to 70% of samples received in the clinical chemistry laboratory that are unsuitable for analysis. 2 4 Although many analytes are subject to interference effects from in vitro hemolysis, the influence of hemolysis on measured potassium concentrations is probably most widely recognized. The frequency of potassium measurements, along with the serious consequences of misdiagnosis of hypokalemia or hyperkalemia, has led to the development of various options for handling hemolyzed samples. 4,5 These options include outright rejection of hemolyzed samples, analysis of hemolyzed samples and the reporting of potassium levels with a disclaimer stating that the result may be incorrect, or analysis of hemolyzed Accepted for publication September 15, From the Department of Chemistry, American University in Cairo, Cairo, Egypt (Drs Mansour and Azzazy); and the Department of Pathology, Oregon Health & Science University, Portland, Ore (Dr Kazmierczak). The authors have no relevant financial interest in the products or companies described in this article. Reprints: Steven C. Kazmierczak, PhD, DABCC, Department of Pathology, Oregon Health & Science University, 3181 SW Sam Jackson Park Rd, L471, Portland, OR ( kazmierc@ohsu.edu). samples and adjustment of the measured potassium result with a correction factor based on the magnitude of hemolysis. Several studies have advocated using correction factors to account for the effect of in vitro hemolysis on measured potassium concentrations. 6 9 While correction factors may help provide a better assessment of the true potassium concentration in patients, the variability in the correction factors that have been proposed is alarming. A variety of issues may account for this difference, including the mechanism used to simulate in vitro hemolysis, interindividual variability in erythrocyte hemoglobin concentrations, and the effect of erythrocyte age on intracellular potassium concentrations. 10,11 The variations in proposed correction factors suggest that adopting them to estimate potassium concentrations may be inappropriate and not in the best interest of patient safety. To better assess the clinical utility of correction factors, we investigated the effects of in vitro hemolysis by using 2 different models of true (ie, nonsimulated) in vitro hemolysis. In addition, we compared the results of our investigations with those of other studies that have derived correction factors by using a variety of different mechanisms to simulate in vitro hemolysis. MATERIALS AND METHODS We used 2 different models, 2 different instrument platforms, and 2 different sample types to evaluate the effects of in vitro hemolysis on measured potassium concentrations. Data were analyzed with Microsoft Excel 2007 (Microsoft, Redmond, Wash). 960 Arch Pathol Lab Med Vol 133, June 2009 Use of Correction Factors for Estimating Potassium Mansour et al

2 Table 1. Relationship Between Hemolysis Index Level and Hemoglobin Concentration Measured With the Beckman LX20 Analyzer Hemolysis Index Level Range of Approximate Hemoglobin Concentrations, g/dl 0 None detected to to to to to to to to to Model 1 The effects of in vitro hemolysis were investigated by using serum obtained from paired blood samples collected from patients during the same phlebotomy procedure. It is not uncommon to obtain 2 or more blood collection tubes from one patient during the same phlebotomy procedure that differ in the degree of hemolysis. If one of the samples so collected showed visual evidence of hemolysis, while another sample did not, the pair was analyzed for serum potassium concentrations. For statistical analysis, samples were considered hemolyzed if their hemoglobin concentration was greater than g/dl, while the control, or nonhemolyzed sample of each pair had a hemoglobin concentration of g/dl or less. We performed regression analysis by plotting the difference in potassium concentrations versus the difference in the hemolysis (H) index values measured in the hemolyzed and nonhemolyzed samples. All samples obtained for this phase of the investigation were collected by using serumseparator collection tubes (Becton, Dickinson and Company, Franklin Lakes, NJ). Potassium was measured with the Hitachi 747 analyzer (Roche Diagnostics, Indianapolis, Ind) by using an indirect ion selective electrode method. Plasma hemoglobin concentrations were estimated quantitatively with the H index measured with the Hitachi 747. The H index calculates plasma hemoglobin levels by using direct spectrophotometry and applying an algorithm to the resultant data. For each pair of samples, the H index value in the sample not showing visual evidence of hemolysis was subtracted from the H index value in the sample with evidence of hemolysis. Deming linear regression was used to analyze the difference between the potassium concentration in the hemolyzed and the nonhemolyzed samples versus the difference in the H index value measured in both. We also constructed a Bland-Altman difference plot to assess the ability of the regression equation to correct for the effects of hemolysis in samples. We plotted potassium concentrations in the sample with hemolysis against the difference between potassium concentrations in the sample without evidence of hemolysis and the potassium concentration estimated from the regression equation. Model 2 We evaluated patient results for potassium that were obtained from blood samples measured during a 1-year period with the Beckman LX20 analyzer (Beckman Instruments, Brea, Calif). Analyses were performed using the manufacturer s reagents. Hemolysis was assessed by using direct spectrophotometry to calculate a semiquantitative hemolysis index graded on a scale of 0 through 10. The hemolysis index levels and the hemoglobin range associated with each are shown in Table 1. Patient blood samples were assigned to 1 of 11 categories based on the hemolysis index value. We calculated the 10th, 25th, 50th, 75th, and 90th percentile values for potassium concentrations within each hemolysis index category. The potassium concentration for samples with no evidence of hemolysis (ie, hemolysis index 0) was considered to be the baseline or true potassium concentration. Potassium concentrations measured in samples with evidence of hemolysis (ie, hemolysis index of 1 10) were compared to potassium concentrations in samples with no evidence of hemolysis. Clinically significant differences between results of samples with and without evidence of hemolysis were determined by using the analytic precision of the assay and the normal intraindividual variation for potassium, according to the method of Fraser et al. 12 Differences in test results that were more than 95% likely to indicate a true change were calculated as follows: (Cv a) (Cv i) Significant Change (%) 100 where Cv a is the long-term analytic coefficient of variation for potassium measured with the Beckman LX20 analyzer and Cv i is the normal intraindividual variation of potassium in healthy individuals. In our laboratory, long-term analytic imprecision for potassium is 2.5% when measured with the Beckman LX20 at a potassium concentration of 4.1 meq/l. The within-subject biological coefficient of variation for potassium has been reported to be approximately 4.7%. 13,14 Using these criteria for analytic imprecision and intraindividual biological variation, we defined a clinically significant change in potassium concentrations as a percent difference of greater than 15.0 between samples with hemolysis index values of 0 and those with index values of 1 through 10. If the difference between potassium concentrations in samples with a hemolysis index of 0 and those with a hemolysis index of 1 through 10 was not significant, as assessed by using the above formula, then hemolysis was considered not to be clinically significant. Literature Review We also reviewed the literature for articles in which investigators estimated the effect of hemolysis on measured potassium concentrations. We used Ovid MEDLINE to search the literature from 1960 through February We limited our search to those articles containing an abstract, written in the English language, and evaluating human subjects. We searched for peer-reviewed manuscripts by using the following search terms: potassium AND hemolysis, potassium AND haemolysis, hemolysis AND interference, and haemolysis AND interference. Information such as method used to simulate in vitro hemolysis, instrument platform used to measure potassium, and derived correction factor were recorded, where available. RESULTS Model 1 We collected a total of 132 paired blood samples from 129 hospitalized adult patients; 3 patients provided 2 pairs of samples each. Median age was 60 years (range, years), with 69 women and 60 men. Samples with evidence of in vitro hemolysis had a median hemoglobin concentration of g/dl (range, g/dl), while the median for the nonhemolyzed sample of each pair was g/dl (range, g/dl). Figure 1 shows the relationship between plasma hemoglobin concentration, as measured with the H index, and increment in potassium concentration. The regression equation obtained with all 132 samples (y 5.101x ; r ) corresponds to a correction factor of approximately 0.51; ie, potassium concentration increased by 0.51 meq/l for every increase in plasma hemoglobin concentration of g/dl. Since plasma hemoglobin concentrations greater than g/dl represent gross hemolysis, and samples with this degree of hemolysis are Arch Pathol Lab Med Vol 133, June 2009 Use of Correction Factors for Estimating Potassium Mansour et al 961

3 Figure 1. Relationship between plasma hemoglobin concentration and increment in measured potassium concentration as measured with the Hitachi 747 analyzer. often not analyzed in many laboratories, we also plotted data for samples containing plasma hemoglobin concentrations less than g/dl. The regression equation for this subset of 117 samples (y x ) corresponded to a correction factor of 0.50, not significantly different from the correction factor obtained with all 132 samples. We used Bland-Altman analysis to assess the ability of the regression equation to evaluate the agreement between potassium concentrations estimated with our derived correction factor and measured potassium concentrations in each of the 132 paired samples. The difference between measured potassium concentrations in the nonhemolyzed samples and estimated potassium concentrations in the hemolyzed samples is shown in Figure 2. Bland-Altman analysis showed that the applied correction factor performed similarly at all measured concentrations of potassium. The 95th percentile limits for estimated potassium concentrations ranged from 0.67 meq/l to 0.54 meq/l of the potassium concentration measured in the nonhemolyzed sample. Model 2 We collected data from patient blood samples and measured potassium levels and the hemolytic index value with the Beckman LX20 analyzer. Samples evaluated were from individuals 18 years and older. We used the long-term analytic precision of the Beckman LX20 method for measurement of potassium, and the normal intraindividual variation of potassium as the basis for determining clinically significant change in measured potassium levels due to in vitro hemolysis. At a potassium concentration of 4.0 meq/l, a percent difference of 15 corresponds to a change of 0.6 meq/l. We found clinically significant interference at a hemolysis index value of 3 or greater. This index value corre- Figure 2. Bland-Altman plot showing difference between the measured potassium concentration in nonhemolyzed sample and the estimated potassium concentration in the hemolyzed sample for each sample pair and plotted as a function of potassium concentration measured in the hemolyzed sample. Potassium concentration was estimated by using a correction factor of 0.51 meq of potassium per liter for each g of plasma hemoglobin per deciliter. 962 Arch Pathol Lab Med Vol 133, June 2009 Use of Correction Factors for Estimating Potassium Mansour et al

4 Table 2. Hemolysis Index Value Potassium Concentrations at 5 Percentile Ranks for Samples in Each Hemolysis Index Category* Potassium Concentration at Different Percentile Ranks n 10th, meq/l (%) 25th, meq/l (%) 50th, meq/l (%) 75th, meq/l (%) 90th, meq/l (%) (0.0) 3.7 (0.0) 4.1 (2.5) 4.4 (2.3) 4.9 (4.3) (2.9) 3.9 (5.4) 4.2 (5.0) 4.7 (9.3) 5.3 (12.8) (8.8) 4.1 (10.8) 4.5 (12.5) 5.0 (16.3) 5.6 (19.2) (14.7) 4.3 (16.2) 4.7 (17.5) 5.2 (20.9) 5.8 (23.4) (14.7) 4.4 (18.9) 4.9 (22.5) 5.5 (27.9) 6.2 (31.9) (23.5) 4.7 (27.0) 5.2 (30.0) 5.7 (32.6) 6.2 (31.9) (32.4) 4.9 (32.4) 5.3 (32.5) 5.8 (34.9) 6.5 (39.2) (35.6) 5.1 (37.8) 5.5 (37.5) 6.1 (41.9) 6.4 (36.2) (44.1) 5.4 (46.0) 5.9 (47.8) 6.5 (51.2) 7.1 (51.1) (50.0) 5.4 (46.0) 5.9 (47.5) 6.6 (53.5) 7.3 (55.3) * Values in parentheses are percent differences in potassium concentrations for samples in each group relative to the potassium concentration of samples with a hemolysis index value of 0. Bolded values indicate clinically significant change (ie, 15.0%) in potassium concentrations due to in vitro hemolysis. sponds to a serum hemoglobin concentration of approximately g/dl to g/dl. Plotting the median potassium concentration for each of the 11 different hemolysis index groups showed that, on average, potassium concentrations increased by 0.20 meq/l with each increment in the hemolysis index value. Since an increment of 1 in the Beckman LX20 hemolysis index value corresponds to an increase in plasma hemoglobin concentration of approximately g/dl, potassium concentrations increased by approximately 0.40 meq/l (95% confidence interval, ) for every increase in plasma hemoglobin concentration of g/dl. Table 2 shows the number of patients with hemolysis index values of 0 thru 10, along with the 10th, 25th, 50th, 75th, and 90th percentile values for potassium concentrations in each of the 11 different hemolysis index groups. Figure 3 shows the same data displayed as a box-and-whisker plot. Clinically significant differences in measured potassium concentrations due to in vitro hemolysis did not occur consistently in all percentile ranks. Results for potassium levels in the 75th and 90th percentile ranks showed clinically significant interference at hemolysis index values of 3 and greater, while results in the 10th percentile did not show clinically significant interference until the hemolysis index was 6 or greater (see Table 2). Review of Studies Reporting Correction Factors for Potassium Associated With In Vitro Hemolysis We reviewed the literature and summarized the results of studies that evaluated the effect of in vitro hemolysis on potassium concentrations. Using the search terms potassium AND hemolysis, potassium AND haemolysis, hemolysis AND interference, and haemolysis AND interference, we identified 744 articles for consideration. Of these, 43 were identified for systematic review. Twelve of the 43 articles reported the increment in potassium as a function of plasma hemoglobin concentration (Table 3). The increase in potassium concentration varied as much as 2.5- fold, from an increase of 0.21 meq/l to an increase of 0.50 meq/l per plasma hemoglobin concentration of g/dl. The median value reported for potassium concentration was 0.31 meq/l per plasma hemoglobin concentration of g/dl. COMMENT The most common preanalytic factor affecting the clinical utility of specimens received in the clinical laboratory is in vitro hemolysis, which can occur by a variety of mechanisms including exposure of samples to extremes of heat and cold, collection of blood using small-bore nee- Figure 3. Box-and-whisker plot showing increase in potassium concentration relative to hemolysis index value as measured with the Beckman LX20 analyzer. Lower and upper ranges of box represent the 25th and 75th percentiles, respectively. Median is represented as horizontal line within box. Lower and upper limits of whiskers represent 10th and 90th percentiles, respectively. Figure based on data from Table 2. Arch Pathol Lab Med Vol 133, June 2009 Use of Correction Factors for Estimating Potassium Mansour et al 963

5 Table 3. Summary of Studies Evaluating Interference Effects of In Vitro Hemolysis on Measured Potassium Concentrations* Correction Factor Method Used to Simulate In Vitro Hemolysis Source, y 0.51 Model 1: Analysis of paired samples by indirect potentiometry with the Hitachi 747 analyzer (Roche Diagnostics, Indianapolis, Ind) Model 2: Increase in potassium concentration of 0.20 meq/l with each increment in the hemolysis index value as measured by indirect potentiometry with the Beckman LX analyzer (Beckman Instruments, Brea, Calif) Osmotic lysis followed by addition of saponin. Indirect potentiometer with Technician Stat Ion analyzer (Technicon Instruments, Tarrytown, NY) and atomic emission spectroscopy Lysis by freezing and thawing of whole blood from 12 healthy volunteers. Indirect potentiometry with Roche Modular Analyzer (Roche Diagnostics) Osmotic lysis followed by freezing and thawing of whole blood. Indirect potentiometry with the Hitachi 717 analyzer (Roche Diagnostics) Osmotic lysis followed by freezing and thawing of whole blood. Atomic emission spectroscopy Osmotic lysis followed by freezing and thawing of whole blood. Blood from 100 (range, ) hospitalized patients studied. Indirect potentiometry with the Hitachi 917 analyzer (Roche Diagnostics) Blood collected from 3 healthy volunteers and homogenized in a blender. WBCs removed before homogenization. Atomic emission spectroscopy Physical disruption by forcing whole blood from 41 volunteers through a 21-gauge needle. Indirect potentiometry with Roche Modular analyzer (95% CI, ) Physical disruption by repeatedly forcing blood from 20 healthy volunteers through a 27-gauge needle. Indirect potentiometry with Dade Behring Dimension RXL (Siemens Healthcare Diagnostics, Deerfield, Ill). Current study Current study Frank et al, Lippi et al, Jay & Provasek, Mather & Mackie, Hawkins, Brydon & Roberts, Dimeski et al, Owens et al, Physical disruption (clot lysis with wooden applicator stick). Pai & Cyr-Manthey, Lysis by freezing and thawing of whole blood from 1 healthy volunteer. Potassium Sonntag, measured by atomic emission spectroscopy (0.25), indirect potentiometry (0.26), and direct potentiometry (0.24) Potassium measurement repeated within 24 h after receipt of hemolyzed specimen. Shepard et al, Thirty-five patients studied. Difference in potassium between 2 samples correlated to serum index. Potassium concentration of 0.16 meq/l for each increment in hemolysis index value as measured by indirect potentiometry with Beckman LX (Beckman Instruments) Potassium concentration of 0.14 meq/l for each increment in the hemolysis index value as measured by indirect potentiometry with Beckman LX (Beckman Instruments). Evaluation of 307 patient samples with in vitro hemolysis. Vermeer et al, * CI indicates confidence interval; WBC, white blood cells. Increase in potassium concentration in meq/l per plasma hemoglobin concentration of g/dl. dles, centrifugation of blood before clotting is completed, and agitation of collection tubes. 4,5 Intracellular potassium concentrations are reportedly 22-fold greater than those found in the extracellular fluid. 16 Thus, in vitro hemolysis can release large quantities of potassium. Guidelines have been established for the assessment of interference effects due to in vitro hemolysis, 17 and various experimental designs have been advocated. 16,18,19 Unfortunately, these methods often ignore the appropriateness of artificially prepared samples as representative of true in vitro hemolysis. Interference from in vitro hemolysis is most often evaluated by the addition of whole blood lysate prepared by freezing and thawing of whole blood, or osmotic lysis of cells after the addition of distilled water to packed cells. These techniques result in lysis of all cells, including reticulocytes, mature erythrocytes, platelets, and leukocytes present in the sample. In vitro hemolysis due to mechanical disruption of cells during blood collection typically does not result in lysis of all cells. In addition, aged erythrocytes are more prone to shear stresses that occur during sample collection than are younger cells. 20 Shear-induced deformation of erythrocytes results in the loss of cellular potassium in proportion to shear forces, and loss of erythrocyte potassium occurs at shear stresses far below the threshold needed to cause cell lysis. 21 Erythrocyte shear stress that occurs during blood collection can result in the formation of erythrocyte membrane pores that allow the egress of small ions such as potassium, but block passage of large molecules such as hemoglobin. 22 Thus, the use of plasma hemoglobin as a marker for in vitro hemolysis will not account for erythrocyte potassium loss due to shear stress during blood collection that does not result in complete cell lysis. Variability in membrane permeability as a result of disease may also cause significant differences in potassium loss from erythrocytes. 23 Older erythrocytes have greater permeability to cations than do younger cells 24 and contain at most one half the potassium content of younger cells. 25 Because of the relationship between erythrocyte potassium concentration and cell age and between permeability to cations and cell age, hemolysis due to mechanical trauma may cause much different changes in serum potassium concentration when compared with hemolysis induced by freezing and thawing of whole blood or osmotic disruption of cells. The former process is more selective for the efflux of potassium from predominantly older cells; the latter process results in nonselective disruption of all blood cells. Because of the greater tendency for older cells 964 Arch Pathol Lab Med Vol 133, June 2009 Use of Correction Factors for Estimating Potassium Mansour et al

6 to hemolyze or lose potassium without undergoing complete cellular lysis, correction factors derived from interference studies that prepare hemolysate with techniques that lyse all cells will not account for this tendency. Another factor that may affect increases in potassium levels after in vitro hemolysis is erythrocyte concentration. Dimenski et al 8 found that decreased erythrocyte concentrations were associated with an increase in potassium release following in vitro hemolysis. They hypothesized that decreased numbers of erythrocytes may lead to faster flow through the needle during phlebotomy, thus resulting in increased shear forces and cell membrane rupture. In addition to the effects of erythrocyte concentration on potassium, increases in leukocyte concentration cause increases in potassium levels. In certain conditions where leukocytes are more mechanically fragile, such as in individuals with chronic lymphocytic leukemia, the degree of potassium increase is even more significant. 26 The ability of automated analyzers to measure plasma hemoglobin in patient blood samples has proved to be an efficient and reliable means for identifying in vitro hemolysis. We evaluated the effects of in vitro hemolysis by using 2 models that represent a more realistic approach for investigating these effects. Hemolyzed samples obtained from actual patients are representative of the population for which in vitro hemolysis of blood samples is seen in a hospital-based clinical laboratory setting. In addition, we obtained samples from many patients with a variety of disease states. Studies that evaluate the effects of in vitro hemolysis with limited number of samples collected from healthy volunteers, or pooled samples, may not give results that are representative of the interference effects seen in hospitalized patients. We defined clinically significant interference by using the analytic precision of the assay and the normal intraindividual biological variation of potassium. As seen in Table 2, our results suggest that in vitro hemolysis causes much greater increases in measured potassium concentrations than have been reported in other studies that have attempted to address this issue. It should be noted that the correction factors obtained with our 2 models, 0.51 and 0.40 meq of potassium per liter for every g of plasma hemoglobin per deciliter, are considerably different. The reason for this difference is unclear, although samples used in model 1 were sera collected using gel-separator tubes, while samples used in model 2 were from heparinized plasma, collected without separator gel. We are not aware of any studies that have attempted to evaluate the effect of sample collection tubes on in vitro hemolysis and induced increases in potassium levels. A wide variety of mechanisms have been used to simulate in vitro hemolysis. As shown in Table 2, these include osmotic lysis, 6,10,27,28 freeze-thaw cycles, 29,30 physical disruption by forcing blood through a small-bore needle, 7,8 physical disruption of clotted blood with wooden applicator sticks, 31 or homogenization of whole blood in a blender. 23 In addition, some studies removed platelets and leukocytes before inducing hemolysis, 23 whereas others did not. 6 8,10,27 30 The use of these different methods to hemolyze erythrocytes may explain the wide variability in correction factors reported by studies that have attempted to quantify the effects of in vitro hemolysis on potassium concentrations. Interestingly, one study that used physical disruption of cells to produce hemolysis derived a correction factor of 0.50, similar to that derived by our first model. 31 Results obtained with model 2 showed that clinically significant hemolysis is present when the hemolysis index, measured with the Beckman LX20, is 3 or greater. This index value corresponds to a plasma hemoglobin concentration of approximately to g/dl. The same observation has also been made independently by 3 other groups of investigators, who found that samples with hemolysis index values of 3, measured with the Beckman LX20, have an increase in potassium concentration of approximately 0.4 meq/l or more, an increase that was considered to be clinically significant. Several studies have been published that advocate the use of correction factors to estimate potassium concentrations in samples with evidence of in vitro hemolysis. 6 9 Dimenski et al 8 recommended reporting a qualifying comment along with a corrected potassium concentration. They considered samples with plasma hemoglobin concentrations greater than g/dl to be grossly hemolyzed and not amenable to correction. Jay and Provasek 6 developed a computer algorithm to estimate the magnitude of increase in potassium on the basis of plasma hemoglobin concentration. A comment alerting clinicians to the magnitude of the false increase in potassium due to in vitro hemolysis is appended to potassium results. Owens et al 7 derived correction factors after physically disrupting whole blood by repeatedly forcing samples, collected from 20 healthy adults, through a 27-gauge needle. The derived correction factor was used to estimate the magnitude of false increase in hemolyzed pediatric blood samples. They suggest that corrected potassium results falling into the normal reference interval obviate the need for repeated venipuncture. Hawkins et al 10 derived correction factors after evaluating in vitro hemolysis on 100 hospitalized patients. They simulated in vitro hemolysis by osmotic disruption, followed by freezing of whole blood. They reported a wide range of correction factors expressing an increase in potassium concentration of 0.20 to 0.35 meq/l per plasma hemoglobin concentration of g/dl. These authors suggest a more practical approach in that they advocate reporting a range of corrected potassium results based on the magnitude of in vitro hemolysis. For example, samples with plasma hemoglobin concentrations between and g/dl would have corrected potassium concentrations in the range of 0.2 meq/l to 0.7 meq/l lower than those measured in the hemolyzed sample. This approach has been suggested as being more appropriate since it takes into account the uncertainty of the derived correction factor. 9 Our results suggest that it is inappropriate to apply correction factors to estimate potassium concentrations in samples with evidence of in vitro hemolysis. The multitude of factors that can affect erythrocyte potassium loss suggests that the use of correction factors based on a semiquantitative measurement of plasma hemoglobin is an overly simplistic solution to a complex process. In addition, legal ramifications concerning the release of test results on samples are known to be inaccurate. Appending a disclaimer to a laboratory test result known to be adversely affected by in vitro hemolysis does not absolve the laboratory of any adverse medical event that may occur as a result of this test result. 35 Laboratories should adopt practices that are legally defensible for reporting of hemolyzed samples. These approaches include reassessment of what constitutes an unacceptable level of hemolysis and integrating this definition into laboratories policies and Arch Pathol Lab Med Vol 133, June 2009 Use of Correction Factors for Estimating Potassium Mansour et al 965

7 procedures. If however, laboratories do use correction factors for reporting potassium concentrations of samples with evidence of in vitro hemolysis, results should probably be reported as a range, thereby indicating the degree of uncertainty. One of the shortcomings of our study was that none of the samples used were obtained from neonates or from specimens collected via heel stick or finger stick. Hemolysis in blood samples collected via heel stick or finger stick is often attributed to excessive squeezing of the puncture site to obtain a sufficient volume of blood. 36 Thus, the correction factors obtained in our studies may not be appropriate for use with hemolyzed specimens obtained via heel stick or finger stick. In addition, all samples were obtained from individuals 18 years or older. Further studies need to be performed on younger patient populations. References 1. Plebani M, Carraro P. Mistakes in a stat laboratory: types and frequency. Clin Chem. 1997;43: Carraro P, Servidio G, Plebani M. Hemolyzed specimens: a reason for rejection or a clinical challenge? Clin Chem. 2000;46: Jones BA, Calam RR, Howanitz PJ. Chemistry specimen acceptability: a College of American Pathologists Q-Probes study of 453 laboratories. Arch Pathol Lab Med. 1997;121: Lippi G, Blanckaert N, Bonini P, et al. Haemolysis: an overview of the leading cause of unsuitable specimens in clinical laboratories [review]. Clin Chem Lab Med. 2008;46: Hawkins RC. Poor knowledge and faulty thinking regarding hemolysis and potassium elevation. Clin Chem Lab Med. 2005;43: Jay DW, Provasek D. Characterization and mathematical correction of hemolysis interference in selected Hitachi 717 assays. Clin Chem. 1993;39: Owens H, Siparsky G, Bajaj L, Hampers LC. Correction of factitious hyperkalemia in hemolyzed specimens. Am J Emerg Med. 2005;23: Dimeski G, Clague AE, Hickman PE. Correction and reporting of potassium results in haemolysed samples. Ann Clin Biochem. 2005;42: Hawkins RC. Correction and reporting of potassium results in haemolysed samples. Ann Clin Biochem. 2006;43: Hawkins R. Variability in potassium/hemoglobin ratios for hemolysis correction [letter]. Clin Chem. 2002;48: Romero PJ, Romero EA, Winkler MD. Ionic calcium content of light dense human red cells separated by percoll density gradients. Biochim Biophys Acta. 1997;1323: Fraser CG, Hyltoft Peterson P, Larsen ML. Setting analytical goals for randon analytical error in specific clinical monitoring situations. Clin Chem. 1990; 36: Biological Variability Data Bank. Available at: biobank1.htm. Accessed May, Ricos C, Alvarez V, Cava F, et al. Current database on biological variation: pros, cons and progress. Scand J Clin Lab Invest. 1999;59: Guder WG. Haemolysis as an influence and interference factor in clinical chemistry. J Clin Chem Clin Biochem. 1986;24: Kroll MH, Elin RJ. Interference with clinical laboratory analyses. Clin Chem. 1994;40: Clinical and Laboratory Standards Institute. Interference Testing in Clinical Chemistry: Approved Guideline Second Edition. Wayne, Pa: Clinical and Laboratory Standards Institute; CLSI document EP7-A Galteau MM, Siest G. Drug effects in clinical chemistry. Part 2. Guidelines for evaluation of analytical interferences. J Clin Chem Clin Biochem. 1984;22: Letellier G, Desjarlais F. Analytical interference of drugs in clinical chemistry. 1. Study of twenty drugs on seven different instruments. Clin Biochem. 1985; 18: Rifkind JM, Araki K, Hadley EC. The relationship between the osmotic fragility of human erythrocytes and cell age. Arch Biochem Biophys. 1983;222: Ney PA, Christopher MM, Habel RP. Synergistic effects of oxidation and deformation on erythrocyte monovalent cation leak. Blood. 1990;75: Kinosita K Jr, Tsong TY. Hemolysis of human erythrocytes by a transient electric field. Proc Nat Acad Sci. 1977;74: Brydon WG, Roberts LB. The effect of hemolysis on the determination of plasma constituents. Clin Chim Acta. 1972;41: Hentschel WM, Wu LL, Tobin GO, et al. Erythrocyte cation transport activities as a function of cell age. Clin Chim Acta. 1986;157: Joiner CH, Lauf PK. Ouabain binding and potassium transport in young and old populations of human red cells. Membr Biochem. 1978;1: Colussi G, Ciprini D. Pseudohyperkalemia in extreme leukocytosis. Am J Nephrol. 1995;15: Frank JJ, Bermes EW, Bickel MJ, Watkins BF. Effect of in vitro hemolysis on chemical values for serum. Clin Chem. 1978;24: Mather A, Mackie NR. Effects of hemolysis on serum electrolyte values. Clin Chem. 1960;6: Sonntag O. Haemolysis as an interference factor in clinical chemistry. J Clin Chem Clin Biochem. 1986;24: Lippi G, Salvagno GL, Montagnana M, Brocco G, Guidi GC. Influence of hemolysis on routine clinical chemistry testing. Clin Chem Lab Med. 2006;44: Pai SH, Cyr-Manthey M. Effects of hemolysis on chemistry tests. Lab Med. 1991;22: Vermeer HJ, Steen G, Naus AJM, Goevaerts B, Agricola PT, Schoenmakers CHH. Correction of patient results for Beckman Coulter LX20 assays affected by interference due to hemoglobin, bilirubin or lipids: a practical approach. Clin Chem Lab Med. 2007;45: Shepard J, Warner MH, Poon P, Kilpatrick ES. Use of haemolysis index to estimate potassium concentration in in-vitro haemolysed serum samples. Clin Chem Lab Med. 2006;44: Steen G, Vermeer HJ, Naus AJM, Goevaerts B, Agricola PT, Schoenmakers CHH. Multicenter evaluation of the interference of hemoglobin, bilibrubin and lipids on Synchron LX-20 assays. Clin Chem Lab Med. 2006;44: Harty-Golder B. Legal dangers of testing unacceptable specimens. Med Lab Observer. 2004;36: Kazmierczak SC, Robertson AF, Briley KP. Comparison of hemolysis in blood samples collected using an automated incision device and a manual lance. Arch Pediatr Lab Med. 2002;156: Arch Pathol Lab Med Vol 133, June 2009 Use of Correction Factors for Estimating Potassium Mansour et al