Abstract
Aim
The purpose of this survey was to determine the cut-offs being used by Australian laboratories using their instrument’s Haemolysis Index (HI), whether these cut-offs vary, and at what level of haemolysis (or haemolysis index) did laboratories stop reporting one or more analytes. This was done in response to the large numbers of haemolysed samples reported in the RCPAQAP Key Incident Monitoring and Management System External Quality Assurance program (KIMMS EQA) and lack of information in the literature at the time regarding what to do once a haemolysed sample was identified. As it was known from discussions with laboratory personnel that different instruments reported their HI differently, we asked for the results to be provided in g/L free haemoglobin.
Method
An electronic survey was conducted with participants enrolled in the RCPA Quality Assurance Programs with a total of 68 laboratories responding to this survey. Some questions attracted a lower level of response.
Results
The responses showed a poor understanding of the relationship between HI units and haemoglobin concentration. There was wide variation in the way HI results were reported and thus comparing cut-off values for reporting specific analytes based on the HI was impossible to determine.
Conclusion
There is a need to harmonise the way laboratories report analytes in the presence of haemolysis. This would involve adopting a uniform definition of HI and a protocol for laboratories to confirm for themselves the level of HI at which each analyte is no longer reported, as this is method dependent and so will vary from laboratory to laboratory.
Introduction
The definition of haemolysis is “the breakdown of erythrocytes in blood, which frees the haemoglobin and intracellular contents from the cells.”1
Haemolysis is visible to the eye where the concentration of haemoglobin exceeds 0.3 g/L.2 Lippi et al found the upper limits for free haemoglobin to be 0.1 g/L for heparin plasma and 0.2 g/L for serum.3
Data being returned through the RCPAQAP KIMMS program (http://www.rcpaqap.com.au/kimms/) over the last six years has consistently highlighted haemolysis as a major source of test/specimen rejection. Other studies have also shown this with the frequency of haemolysis being as high as 3.3% of routine samples referred to a clinical chemistry laboratory and as high as 70% of all unsuitable specimens received.4–5 Published data indicates the proportion of haemolysed samples received by laboratories is variable but what is agreed is it is the commonest ongoing pre-analytical problem. Haemolysis can interfere with accurate measurement of some analytes in serum, plasma or other body fluids. There is more than one way for interference to occur,6 which means the degree of interference is particular to individual analysers, methods and causes of haemolysis, making harmonisation of the actual cut-off values for individual tests unrealistic.7
The haemolysis index (HI) is a (semi-)quantitative estimate of free haemoglobin and provides an alert to potential haemolysis interference. Studies have shown that, for the analysers studied, the HI closely approximates the range of haemoglobin concentrations.7–8 Automated HI can be used to provide a numerical value that can then be used to assess the integrity and quality of the sample. The numerical value can then be utilised to set ‘cut-off’ values for reporting certain analytes or rejecting a sample.1
The aim of the survey was to see if there was any consistency with regards to HI cut-off levels across instrument platforms, laboratories and analytes.
Method
A haemolysis survey was electronically distributed to all laboratories enrolled in the RCPA Quality Assurance Programs to determine the “cut-offs” being used, how these cut-offs vary and at what level of haemolysis laboratories stopped reporting one or more analytes. Sixty-eight laboratories responded to the survey, though some questions attracted a lower level of response.
Participants were asked to submit their cut-off value(s) for the levels of HI used in their laboratory in g/L. Unfortunately, this request was not followed making much of the data obtained difficult to interpret in any depth. Some conclusions, however, could be drawn.
Results
KIMMS data over a six year period (2009–2014) has consistently shown sample rejection rates of between 0.9 −1.2% of all accessions (Table 1). Of these, 24% are rejected due to haemolysis. In terms of the number of samples that are rejected due to pre-analytical errors, KIMMS data indicates 19% of all rejected samples are haemolysed (Table 1).
Table 1.
KIMMS data covering 6 years showing the % of haemolysed samples per accessions and per Total samples rejected.
| Year | Accessions | Total samples rejected | %Samples rejected | Haemolysed samples rejected | Haemolysis as % of Total samples rejected | Haemolysis as % of total accessions | Total pre-analytical errors | Haemolysis as % of Total pre-analytical errors |
|---|---|---|---|---|---|---|---|---|
| 2009 | 13049463 | 117492 | 0.90% | 28161 | 24% | 0.22% | 136351 | 21% |
| 2010 | 20736679 | 202745 | 0.98% | 29875 | 15% | 0.14% | 234135 | 13% |
| 2011 | 28211117 | 316471 | 1.12% | 79172 | 25% | 0.28% | 398167 | 20% |
| 2012 | 28798023 | 344335 | 1.20% | 90570 | 26% | 0.31% | 427789 | 21% |
| 2013 | 33945026 | 348657 | 1.03% | 70947 | 20% | 0.21% | 435856 | 16% |
| 2014 | 39196320 | 352337 | 0.90% | 105746 | 30% | 0.27% | 519197 | 20% |
| Total 6 years | 163936628 | 1682037 | 1.03% | 404471 | 24% | 0.25% | 2151495 | 19% |
A literature search conducted in 2010 demonstrated little information regarding appropriate actions for laboratories to follow with a haemolysed sample. This led KIMMS to investigate contemporary Australian practices.
In response the question “What is the approximate haemoglobin concentration cut-off at each of the following haemolysis levels in g/L? (Please answer in whole numbers (no decimals))”, where we asked for the cut-off levels for H1 through to H6, we received answers ranging from 1–1500 for H1 and 3–1000 for H6 (see Table 2 for all results received).
Table 2.
The range of cut-off values reported for HI levels and the number of laboratories that reported answers for each level. Participants were asked to report these in g/L. 44 laboratories entered replies to this question.
| HI level | n | Range of cut-off values reported |
|---|---|---|
| 1 | 12 | 1–1500 |
| 2 | 15 | 1–3000 |
| 3 | 10 | 2–850 |
| 4 | 12 | 2–1000 |
| 5 | 6 | 8–500 |
| 6 | 6 | 3–1000 |
Although the question could have been worded better, as not all laboratories use H1–H6, it is apparent that not all results were in g/L. Asking for the results to be in whole numbers was also problematic, as cut-offs generally start at well below 1 g/L of free haemoglobin.2 Work done by others show that there is no consensus in the way instruments express haemolytic indexes, making this question even more difficult to answer.6,8
As noted, there is a lack of standardisation as to how many levels of HI are reported. Whilst 14 laboratories reported that they used 6 levels and 15 reported using 4 levels, there were many variations of this (see Table 3).
Table 3.
Laboratories were asked to report how many HI levels they measured. (n=44)
| Number of HI levels | Number of Laboratories |
|---|---|
| 6 | 14 |
| 5 | 4 |
| 4 | 15 |
| 3 | 10 |
| 2 | 1 |
We also sought to determine at which HI level laboratories stopped reporting at least one analyte (Table 4), and at which level they stopped reporting all analytes (Table 5). Whilst the level of haemoglobin is not reflected by the HI level across laboratories, and the cut-off is method dependent, the variation appears to be quite high.
Table 4.
This shows the range of responses to the question “at what haemolysis level do you cease to report at least one analyte”. (n=68)
| Response | Response Percent Level at which 1 analyte not reported |
|---|---|
| H1 | 32% |
| H2 | 77% |
| H3 | 98% |
| H4 | 100% |
| H5 | 0% |
| H6 | 0% |
Table 5.
This shows the range of responses to the question “at what haemolysis level do you cease to report all analytes”. (n=68)
| Response | Response Percent Level at which no results reported |
|---|---|
| H1 | 0% |
| H2 | 2% |
| H3 | 26% |
| H4 | 58% |
| H5 | 75% |
| H6 | 100% |
Table 6 shows the number of participants and the range of HIs used as a cut-off for certain analytes. The data shows wide variation in the level of HI being utilised by laboratories and was impossible to interpret with any accuracy.
Table 6.
Results from the question “at what haemolysis level do you cease reporting the following analytes”.
| Analyte | Number of labs reporting they do have a haemolysis cut off | Range of reported haemolysis indexes |
|---|---|---|
| Albumin | 34 | H4–H5 |
| ALT | 36 | H1–H6 |
| Amylase | 30 | H1–H5 |
| AST | 37 | H1–H6 |
| Bicarb | 32 | H2–H6 |
| Bilirubin (conj) | 32 | H1–H6 |
| Bilirubin (total) | 38 | H1–H6 |
| Calcium | 33 | H2–H6 |
| Cholesterol | 32 | H2–H6 |
| CK | 36 | H1–H6 |
| Creatinine | 33 | H2–H6 |
| GGT | 32 | H2–H6 |
| Potassium | 44 | H1–H6 |
| LDH | 38 | H1–H6 |
| Magnesium | 35 | H1–H6 |
| Sodium | 32 | H2–H6 |
| Phosphate | 34 | H1–H6 |
| Total Protein | 34 | H1–H6 |
| Triglyceride | 30 | H3–H6 |
| Urate | 32 | H1–H6 |
Discussion
Automated HI can be used to provide a numerical value that can then be used to assess the integrity and quality of the sample. The numerical value can then be utilised to set ‘cutoff’ values for rejecting a sample,1 however, the user must have information regarding the amount of free haemoglobin each numerical value is related to (unless already given in g/L).8 Currently, these numerical values are not subjected to any calibration or QC check requirements.
Most chemistry analysers now provide an HI to assess levels of haemolysis. These are initially set by the manufacturer. From a brief follow-up phone survey conducted across some Australian Laboratories in 2014, it is apparent that there is no consensus regarding verification of these levels. As the HI is utilised more by laboratories, it might be expected that, as with other ‘assays’, the HI becomes subject to calibration and QC.
More importantly, our data shows that there is a need to standardise the way in which HI is reported i.e. units used. This can vary from actual levels of free haemoglobin, to arbitrary numerical values or a system of ordinal numbers.6,8
Currently, laboratories set their own rules to determine which tests they will report or not based on the level of haemolysis and, most often, the manufacturers’ guidelines. There is a need to harmonise the approach to this, as others have shown that the information supplied by manufacturers regarding the effects of haemolysis does not always match reality.7
This survey and subsequent discussions has led to the AACB forming a working party to develop guidelines on how haemolysis should best be treated by laboratories.
Conclusion
Results show that laboratories do not have a good understanding of what level of haemolysis in g/L free haemoglobin relates to their HI. This would then make deciding cut-off levels for analytes difficult when haemoglobin, as an interference factor, is quoted in g/L. If there is no harmonisation of the HI unit, then a clinician may misinterpret results from one laboratory if they assume a given HI indicates the same level of interference as that reported by another laboratory.
Acknowledgments
The authors would like to acknowledge the Australian Government, Department of Health and Ageing for their support via a grant through the Quality Use of Pathology Program.
Footnotes
Competing Interests: None declared.
References
- 1.Clinical and Laboratory Standards Institute . Approved Guideline. CLSI C56-A. Wayne, PA, USA: CLSI; 2012. Haemolysis, icterus, and lipemia/turbidity indices as indicators of interference in clinical laboratory analysis. [Google Scholar]
- 2.Thomas L. Haemolysis as influence and interference factor. eJFCC. 13(4) http://www.ifcc.org/ejifcc/vol13no4/130401002.htm. [PMC free article] [PubMed] [Google Scholar]
- 3.Lippi G, Giavarina D, Gelati M, Salvagno GL. Reference range of hemolysis index in serum and lithium-heparin plasma measured with two analytical platforms in a population of unselected outpatients. Clin Chim Acta. 2014;429:143–6. doi: 10.1016/j.cca.2013.12.010. [DOI] [PubMed] [Google Scholar]
- 4.Carraro P, Plebani M. Errors in a stat laboratory: types and frequencies 10 years later. Clin Chem. 2007;53:1338–42. doi: 10.1373/clinchem.2007.088344. [DOI] [PubMed] [Google Scholar]
- 5.Vecellio E, Li L, Mackay M, Gay S, Lake R, Georgiou A, et al. Report to Royal College of Pathologists Australasia. Sydney: Australian Institute of Health Innovation, Macquarie University; 2015. A benchmark study of the frequency and variability of haemolysis reporting across pathology laboratories – the implications for quality use of pathology and safe and effective patient care. [Google Scholar]
- 6.Dolci A, Panteghini M. Harmonization of automated hemolysis index assessment and use: Is it possible? Clin Chim Acta. 2014;432:38–43. doi: 10.1016/j.cca.2013.10.012. [DOI] [PubMed] [Google Scholar]
- 7.Fernandez P, Llopis MA, Perich C, Alsina MJ, Alvarez V, Biosca C, et al. Harmonization in hemolysis detection and prevention. A working group of the Catalonian Health Institute (ICS) experience. Clin Chem Lab Med. 2014;52:1557–68. doi: 10.1515/cclm-2013-0935. [DOI] [PubMed] [Google Scholar]
- 8.Lippi G, Luca Salvagno G, Blanckaert N, Giavarina D, Green S, Kitchen S, et al. Multicenter evaluation of the hemolysis index in automated clinical chemistry systems. Clin Chem Lab Med. 2009;47:934–9. doi: 10.1515/CCLM.2009.218. [DOI] [PubMed] [Google Scholar]