Abstract
Background
Haemolysis is defined as the release of cellular components of erythrocytes and other blood cells into the extracellular space of blood. These cellular components can cause interference in laboratory measurements, potassium being a commonly measured analyte to be affected. A number of factors have been implicated in the aetiology of haemolysis. We undertook this study to enable us to identify and hence rectify causes of haemolysis in samples from patients on acute medical and surgical wards.
Methods
We performed a prospective study of 353 blood sampling events during February and March 2007. A proforma was used to obtain detailed information of each blood-taking episode. Information from the proforma was linked to the incidence of haemolysis obtained from the hospital computer system.
Results
The incidence of haemolysis among the samples studied was 6.5%. While staff group, method of sampling, tourniquet time and number of attempts at venepuncture were each univariately associated with haemolysis, stepwise logistic regression resulted in a final model which only included tourniquet time (odds ratio for haemolysis if tourniquet time >1 min was 19.5 [95% confidence interval [CI] 5.6–67.4%]).
Conclusion
Tourniquet time of more than a minute is associated with a significant increase in risk of haemolysis. Advice on tourniquet time is included in phlebotomy training within the hospital; hence a campaign of appropriately channelled continuing education on this issue may be successful in reducing the haemolysis rate.
Introduction
Haemolysis is defined as the release of intracellular components of erythrocytes and other blood cells into the extracellular space of blood. 1 It usually occurs in vitro, e.g. during blood sampling, transport or storage, but can also occur in vivo, e.g. as a result of a transfusion reaction.
Haemolysis can be detected by visual inspection for reddish discolouration of serum or plasma. However, many modern laboratory analysers are set to detect haemolysis automatically, lessening the likelihood of any resulting interference going undetected. Haemolysis can cause interference due to leakage of components normally found in high concentrations within cells, such as potassium, aspartate transaminase and lactate dehydrogenase, or due to interference in the analytical process itself e.g. released adenylate kinase affecting enzymatic reactions.
The effect of haemolysis is most widely appreciated with respect to potassium. Some laboratories assist clinicians in the interpretation of potassium results from haemolysed samples by issuing a qualitative report. 2 However, where knowledge of a parameter, known to be altered by haemolysis, is essential for clinical management repeat venepuncture is required. This leads to increased workload for clinical, nursing and phlebotomy staff, increased costs to laboratories and increased discomfort for patients.
We performed this study to assess the incidence of haemolysis in samples received in the laboratory and to identify any associated factors that might be amenable to correction.
Methods
We performed a prospective study of blood sampling during February and March 2007 at 26 locations within Derriford Hospital. All personnel taking blood at these locations during this time were asked to complete a proforma at the time of sampling. This detailed patient demographics, number of attempts at venepuncture, tourniquet time, mode of sample transport to the laboratory and staff group of the person taking blood.
Samples were assayed on a Roche Modular system and haemolysis was detected by using the automated haemolytic index (HI) function, which uses spectrophotometry to detect differences in absorbance at 570–600 and 600–700 nm. An HI corresponding to 30 μmol/L of haemoglobin was used to define samples as haemolysed.
The potential univariate relationships between presence/absence of haemolysis and each of the factors of interest (staff group, mode of sampling, tourniquet time, number of attempts at venepuncture and method of transport to laboratory) were assessed using chi-squared tests, with exact P values presented (StatXact 8, Cytel Inc, Cambridge, MA, USA, 2007). Stepwise logistic regression (SPSS v15, SPSS Inc, Chicago, IL, USA, 2006) was used to identify which combination of these factors significantly influenced the presence of haemolysis. Missing data (i.e. ‘unknown’) were excluded from the chi-square testing and logistic regression analysis and resulting odds ratio (OR) calculations.
Results
During the study period 353 samples were received accompanied by a completed proforma, of which 23 (6.5%, 95% confidence interval [CI] 4.2% to 9.6%) were haemolysed. Data from the proformas are summarized in Table 1.
Summary of data obtained from the proforma and results of statistical analysis showing exact P values for parameters studied
There was evidence of statistically significant univariate associations between haemolysis and the staff group, tourniquet time and number of attempts at venepuncture (Table 1). The final logistic regression model only included tourniquet time (P < 0.001), with an increased risk of haemolysis with increased tourniquet time (OR for haemolysis if tourniquet time >1 min of 19.5 (95% CI 5.6–67.4%)).
Discussion
Understanding the factors associated with haemolysis is important in order for them to be appropriately addressed. The staff group of the person taking blood was associated with haemolysis in our study. This result may have come about due to confounding as there was a significant difference in the tourniquet times between each staff member group. While this might relate to differences in the patient population allotted to each group, it may relate to differences in training and/or experience of the groups concerned. A previous study found a significantly higher rate of haemolysis on wards where phlebotomists had not undergone formal training and certification. 3 We found a tourniquet time of more than 1 min to be significantly associated with haemolysis. This is in keeping with some studies, 3,4 although not with others. 5,6 Stasis at venepuncture from prolonged tourniquet application increases the potassium concentration in serum because of release of potassium-rich fluid from red blood cells and muscle cells with haemoconcentration. 7
Guidelines suggest that the maximum time a tourniquet should remain in place is 1 min; this requires the tourniquet to be applied twice during the venepuncture procedure – once during selection of an appropriate vein and then again just prior to puncture. 8 The tourniquet should be released when blood flows into the blood collection tube.
Our hospital has a written policy for phlebotomy, which includes appropriate advice regarding tourniquet time, and this is used when providing instruction in the technique. There is an important role for continuing education in order to reduce preanalytical errors and this may be particularly relevant with respect to use of the tourniquet.
Conclusion
A tourniquet time of more than 1 min was significantly associated with an increase in the risk of haemolysis in blood samples from hospital patients. Advice on tourniquet use is included in all phlebotomy training, but we believe that the simple measure of providing continuing education on this issue should help to reduce the haemolysis rate in our patient population.
Footnotes
ACKNOWLEDGEMENTS
The authors acknowledge the help of staff within Derriford Combined Laboratory and all those who took part in data collection.