The causes of pseudohyperkalaemia can be classiiied into two main categories: disease-specific conditions and technical errors during blood collection, storage, separation or transport.1 Pseudohyperkalaemia arises from leakage of intracellular potassium, usually from red cells but occasionally from white cells, platelets or tissue cells. It has been reported that haemolysed samples account for 60% of rejected samples.2 For example, in our acute 750 tertiary bed hospital, based on 2003 data of the samples received for electrolytes. 6.1% had a haemolytic index (HI) high enough (>0.5 g/L) to cause significant artefactual increases in some analytes. The Emergency Unit (EU), whose specimens are collected in syringes, frequently by junior medical staff, and in relatively small numbers using small veins from the dorsum of the hand, represented 37% of these samples. However, the EU only provided 13% of the total number of specimens. The cell lysis may be due to in vitro damage but may also occur with spontaneous lysis of leucocytes or platelets, especially if these are present at high concentration or have increased fragility. These extra factors make it extremely difficult to implement an accurate mathematical correction mechanism for potassium or any other altered analytes using red cell analyte/haemoglobin ratios as they vary from sample to sample and between individuals.
Haemolysis affects many analytes but potassium (K+) is usually the most important, as the result obtained may falsely indicate a life-threatening abnormality and give rise to an immediate change in treatment. However, there is no standardized approach to the reporting of potassium results from haemolysed samples. The laboratory's options for handling haemolysed samples on which potassium is requested include:
1. rejection of all samples without analysis;
2. analysis of the samples and reporting of the potassium result with a qualitative comment;
3. analysis of the samples and rejection of the result when the haemolysis is above a defined limit with a qualitative comment:
4. analysis of the samples and correction for haemolysis by a correction formula and reporting of a quantitative result with or without qualitative comment.
Although a correction formula using a potassium/ haemoglobin ratio has been derived and published,3 this ratio has been demonstrated to be imprecise due to inter-individual red cell (haemoglobin) variability.4 Both of these studies used the Meites1 method or a modification of it to prepare the haemolysate. Here we describe a useful solution to the problem of reporting potassium on haemolysed samples, balancing clinical needs with analytical concerns. We use the degree of haemolysis and a potassium/haemoglobin ratio established by an improved method to provide a qualitative report as to the likely clinical significance of the measured potassium concentration.
We have developed a new method for preparing haemolysed samples, which is superior because it more faithfully reilects the way artefactual haemolysis usually occurs. Haemolysates were prepared from 7-10 mL of freshly collected lithium heparin blood (GreinerVacuette lithium heparin tubes with plasma separator, Product No. 455083, marketed by Interpath Services, Brisbane. Australia), which was obtained from 41 volunteers, and immediately five to eight equal aliquots (~1.5 mL) were prepared. The first aliquot was centrifuged, the plasma separated, and the potassium and HI measured. The HI on this first sample was required to be <0.5 g/L free Hb in order for further studies to be performed. The remaining aliquots were passed through a 21 G blood collection needle to mimic an actual clinical setting collection process. The number of times a sample was passed though a needle increased with each subsequent sample to produce an increasing range of haemolysis. This technique resulted in median plasma-free haemoglobin values of ~1 g/L in the second aliquot (range 1-2 g/L: one passage through a 21 G needle), to median haemoglobin values of ~4 g/L in the fifth aliquot (range 3-5 g/L; four passages through a 21G needle). The treated aliquots were then immediately centrifuged and the plasma separated from the cells. Analyses for HI and potassium were then performed on all samples on Hitachi Modular analysers (Roche Diagnostics, Sydney, Australia). The average change in potassium with increasing HI values was calculated and plotted against the red blood cell count, haemoglobin concentration, leucocyte and platelet counts.
Haemolysis was quantified spectrophotometrically on a Hitachi Modular analyser. The sample was diluted in saline, the absorbance measured at 570 nm, and the absorbance value converted to mg/dL by multiplication with a factor. The data are reported from this instrument as an HI. An HI of 100 represents approximately 1 g/L of free haemoglobin (100 mg/dL). Potassium concentrations were also measured using this analyser. At potassium concentrations of 3.0 mmol/L and 6.0mmol/L. the total coefficients of variation were 1.7% and 1.0%, respectively.
All collected samples also had haemoglobin, red blood cell, white cell and platelet counts measured on a Sysmex 9000 (Roche Diagnostics, Sydney, Australia). The ranges of concentrations tested were: haemoglobin 80-173 g/L; red blood cells 2.42-6.77 × 10^sup 12^/L; leucocytes 3.0-300 × 10^sup 9^/L and platelets 31-710 × 10^sup 9^/L.
In the three samples with varying leucocyte count, the individual sample haemolysis studies showed a linear relation between potassium and haemolysis increase, as indicated in Figure 1. The potassium increase ranged from 0.0029 - 0.0053 mmol/L per unit of HI. with a mean of 0.0036 mmol/L. Thus. 1 g/L of free haemoglobin or 100 units of HI will cause a 0.36 mmol/L (rounded off to 0.4 mmol/L) increase in potassium. The main contributor to this increase was leakage from red blood cells, as indicated by the linear relationship between plasma haemoglobin and potassium increment. Platelet count increase did not signilicantly contribute to potassium increase, but leucocyte count increase made a small increment. The increment in potassium concentration per HI unit was greater in anaemic subjects (see Figure 2).
|Figure 1. Example of the effects of increasing haemolysis on potassium concentration with varying leucocyte count ([black square] 3.0 × 10^sup 9^/L; * 10.8 × 10^sup 9^/L; [black triangle up] 85.6 × 10^sup 9^/L).|
|Figure 2. Effects of: (a) leucocyte count; (b) platelet count; (c) haemoglobin concentration and (d) red blood cell count on the delta K+ /HI. (a) Effects of leucocyte count on K+ increase with haemolysis, (b) Effects of platelet count on K+ increase with haemolysis, (c) Effects of haemoglobin concentration on K+ increase with haemolysis, (d) Effects of red blood cell count on K+ increase with haemolysis.|
The correction formula is applied via our Laboratory Information System (LIS) and the corrected potassium result is displayed only to laboratory staff.
|Table 1. Reporting procedure for corrected potassium with respect to HI and appropriate comment|
Depending upon the approximate potassium result after correction for haemolysis, one of a series of comments is released instead of a quantitative result, as shown in Table 1.
Our results demonstrate that the principle cause of potassium increase is red cell haemolysis but that large increases in leucocyte count may also contribute. In the samples studied, platelet count had little effect upon potassium concentration. The Meite's method for the preparation of haemolysate removes most of the platelets and white cells during the washing process and does not reflect the in vitro process of haemolysing samples in the clinical setting. Hence, this method can significantly underestimate the true extent of the potassium increase in vitro.
Of the inter-individual variability factors, haemolysis or contribution from red cells is the only measurable cause of arlefactual increase in plasma potassium. The degree of potassium contribution from red blood cells is dependent on the red blood cell count and inter-sample variability factors. The data show that decrease in red blood cell counts leads to increased potassium leakage. It is possible that reduced numbers of red blood cells lead to faster flow through the needle, hence leading to increased shear and cell membrane rupture. When contributions from other non-measurable inter-individual factors (leucocyte and platelet counts, etc.) and inter-sample variability factors (collection process, storage, transport, delay, centrifugation, etc.) are considered, the calculation of a quantitative corrected potassium result may prove unreliable. As indicated by our results, increase in leucocytes count results in increase in potassium leakage. In particular, in conditions where leucocytes are more mechanically fragile, such as in chronic lymphocytic leukaemia (CLL), the degree of potassium increase is even more significant.6 In thrombocytosis, where platelet counts exceed 750 × 10^sup 9^/L. particularly with serum collections, the degree of potassium leakage increases.7 It is estimated that with serum collections potassium increases by 0.07-0.15 mmol/L per 100 × 10^sup 9^/L platelets due to leakage.8 The degree of contribution from the leucocytes or platelets is unmeasurable and therefore cannot be corrected.
In considering the options for handling potassium requests on haemolysed samples, there are reasons why the solution presented here offers many advantages for both the laboratory and the patient. Repeat sampling is not always possible for many reasons, for example critical status of the patient, patient in transit to an operating theatre or other ward, discharged, collection from an outpatient clinical or general practice medical centre. Reporting a quantitative corrected potassium result with a qualitative comment warning the clinicians that the result may not be reliable is in our opinion unsafe, as clinical staff tend to act on a reported number regardless of the means by which it is derived, rather than the reported number and a qualitative comment in unison. For this reason, we release a qualitative statement only, as shown in Table 1, together with a recommendation for re-collection. If a critical value is suggested, the assessment is communicated urgently by telephone to the clinical unit. This availability of a qualitative potassium result therefore allows the patient to receive more timely and appropriate intervention when needed.
We advocate that samples with an HI > 6 g/L should have no results released. We acknowledge that the potassium increment is a continuous variable and the choice for the cut-off point is arbitrary. The value of the upper limit of acceptance (if any) is one for each laboratory to determine. We consider an HI of 6 g/L as grossly haemolysed and unsuitable for a comment on the likely corrected potassium. In our hospital, of the total samples received for electrolytes, only 0.14% had an HI > 6 g/L. of which 54% were from the EU. At this level of haemolysis, there is a clinically significant dilutional effect on all analytes from the volume of actual cellular content release. It is therefore best not to release any results and to seek a further sample. All results not reported because of haemolysis, including HI results, are saved on our LIS and remain viewable only by laboratory staff as a reference. However, in extremely rare circumstances with an HI > 6 g/L where potassium concentration is required by clinical staff and corrected potassium is critically low or high, it is communicated verbally to clinical staff (with a warning that it may be unreliable) in order to aid patient management.
Rarely, haemolysed samples are due to in vivo haemolysis. If the patient is known to have intravascular haemolysis (i.e. indicated by clinical notes or information provided by haematology). all results are reported regardless of the HI, with the following comment: 'Patient has intravascular haemolysis. This may cause a true elevation of potassium.'
This procedure has been implemented successfully across our network of 33 laboratories where both multi-skilled and specialized biochemistry laboratory staff operate. It has received very favourable feedback from laboratory and medical staff, especially from EUs, who are the largest source of haemolysed samples.
All new analysers now on the market are capable of estimating haemolysis in some form. The approach used here can be easily implemented on any analyser or LIS. For example, the equation can be implemented with the Hitachi series of analysers in the Calculated Parameters field if it is impossible to set it up via the LIS.
In conclusion, we have assessed the problem of pseudohyperkalaemia due to leakage from cells. We believe our method to be superior to that in the current literature as it reflects the reality of how haemolysis occurs in clinical practice. We found larger increments of potassium than previously reported. We present a robust qualitative mechanism for dealing with haemolysed samples, which still allows useful clinical information to be released, without making unreasonable extrapolations from in vitro data as occurs with quantitative corrections. It is a powerful tool in ensuring that clinically significant and particularly, critical results in haemolysed samples are reported and acted on promptly in the interest of patient care.