Practical handling of hemolytic, icteric and lipemic samples for coagulation testing in European laboratories. A collaborative survey from the European Organisation for External Quality Assurance Providers in Laboratory Medicine (EQALM)

Statistics

Descriptive statistics (medians (10- and 90-percentiles) and percentages) were used. Percentages for each question were calculated based on number of laboratories responding to that particular question. Number of missing responses increased gradually throughout the questionnaire (319 (72.5 %) completed the entire survey). All laboratories were not expected to answer all questions, as some questions were shown or hidden, depending on answers to previous questions. SPSS version 22.0 (SPSS Inc.) and Excel (MicroSoft Office 365 v2302) were used for statistics and Figures.

Number and characteristics of the responding laboratories

A total of 440 laboratories responded to the questionnaire, of which 430 (98 %) came from 33 European countries (93 % from 16 of these countries) and 10 from non-European countries (Figure 1).

Figure 1:

Number of laboratories answering from each country.

The characteristics of the laboratories are shown in Table 1. The majority (95 %) used coagulation instruments from one of the following manufacturers: Stago, IL, Sysmex and Siemens (Figure 2 and Supplemental Table 1). Optical clot-detection for APTT, PT, INR and fibrinogen was most often used among participants (71 %), but only 11 % of these used mechanical clot-detection method or manual tilt tube method as an alternative or back-up method. More than half of the laboratories (54 %) stated using the same instrument both as main and alternative, while the rest stated different other instruments with optical clot-detection, clinical chemistry instruments (D-dimer) or point-of-care devices (INR, D-dimer) as their back-up method (Supplemental Table 1).

Figure 2:

The percentage of laboratories stated using a coagulation instrument with optical or mechanical clot-detection method for the clotting tests, APTT, PT, INR and fibrinogen as the main (middle upper circle) and alternative (lower middle circle) coagulation instrument. Circles to the left show the percentages using different manufacturers for instruments with optical clot-detection and to the right for instruments with mechanical clot-detection for the clotting tests. All instruments use optical methods for D-dimer and antithrombin. See Supplemental Table 1 for more details.

Automated HIL-measurement was available in approximately 2/3 of the main coagulation instruments and 1/4 of the alternative coagulation instruments, with the highest percentage among users of Sysmex and IL instruments (Table 2).

Frequency of HIL-detection

Participants stated they detect samples with hemolysis more frequently than icterus or lipemia. Detection of hemolysis was positively correlated with the number of processed coagulation samples in the laboratory (Supplemental Figure 1). Information regarding the frequency of detected HIL-samples was mainly based upon personal experience (64 %), with only 21 and 15 % obtaining the data from the laboratory information system (LIS) or from manual registration of HIL-samples, respectively.

Procedures for HIL-detection

A written procedure for detection of H, I and L was available in 67 , 55 and 61 % of the laboratories, respectively (Supplemental Figure 2A). All samples were checked by 83 , 73 and 80 %, for H, I and L, respectively, while 12 , 11 and 11 % checked some samples and 4 , 16 and 9 % none (Supplemental Figure 2B). The most frequent reasons given for not checking any samples or not checking all samples, were that the laboratory had “no written procedure for HIL” and/or “no automatic HIL-detection”. In addition, about 1/3 stated that “results are not affected by HIL” as reason for not checking any samples (Supplemental Tables 5 and 6). The percentage of laboratories not checking any samples, tended to be slightly higher for those using mechanical clot-detection (7 , 16 and 13 % for H, I and L, respectively) compared to optical clot-detection (3 , 12 and 5 %), and the same accounts for laboratories not checking all samples (20 , 8 and 13 % (mechanical) vs. 9 , 10 and 8 % (optical)). However, absolute numbers are small.

HIL was detected by visual inspection only, in 37–38 % of the laboratories, by automatic detection only, in 36–43 %, while the remaining used a combination (Figure 3A). Most laboratories using only visual inspection to detect HIL, used experience to determine HIL-levels or they did not determine HIL-levels (i.e. no levels=HIL positive or negative) (Figure 3B). Very few of these laboratories stated to use a visual chart/scale or to measure HIL-levels by coagulation or clinical chemistry instruments (Figure 3B).

Figure 3:

The percentage of laboratories stating to detect HIL by visual inspection only, by automatic (coagulation or chemistry instruments) analyses only or a combination (A), and the methods used to determine HIL-levels by those using only visual inspection to detect HIL (B). HIL-levels coag; measuring by coagulation instrument, HIL-levels chem; measuring by chemistry instrument, Hb/Bil/Tg conc; hemoglobin, bilirubin and/or triglyceride concentration by hematology and/or chemistry instrument, respectively, No levels (pos/neg); only stating HIL presence or not (no classification).

Almost half of the laboratories (40–45 %) did not classify into HIL-levels, but only stated HIL-presence or not (positive or negative for HIL). When determination of HIL-level was performed, most (30 %) classified into qualitative HIL-levels (e.g. 1–5) (Figure 4A). Among laboratories detecting HIL only automatically, the percentage classifying into qualitative HIL-levels (44–54 %) was more frequent than not classifying into levels (29–36 %) (Figure 4C). Among laboratories detecting HIL only visual or by visual and automatic, the percentage not classifying into HIL-levels was higher (Figure 4B) or rather similar (Figure 4D), respectively.

Figure 4:

The percentage of laboratories stating to classify into different categories for hemolysis, icterus and lipemia for all laboratories (A), for laboratories using only visual HIL-check (B), for laboratories using only automatic HIL-check (C), and for laboratories using both visual and automatic HIL-check (D). No levels (pos/neg); only stating HIL presence or not (no classification). HIL-levels; semi-quantitative expression of HIL into levels (e.g., 1–5). Hb/Bil/Tg conc; hemoglobin, bilirubin and/or triglyceride concentration by hematology and/or chemistry instrument, respectively. +/++/+++; semi-quantitative expression of HIL into three levels (merged laboratories answering +/++/+++ and slight/moderate/gross).

When looking more closely into which manufacturers were used, Stago and Sysmex users tended to classify into HIL-levels more often than the others, whilst Siemens users tended to more often classify into +/++/+++ (slight/moderate/gross), in particular for hemolysis (Figure 5).

Figure 5:

The percentage of laboratories stating they classify into different HIL-categories for hemolysis (A), icterus (B) and lipemia (C) according to instrument manufacturer in use. No levels (pos/neg); only stating HIL-presence or not (no classification). HIL-levels; semi-quantitative levels (e.g., 1–5). Hb/Bil/Tg conc; measurement of hemoglobin, bilirubin and triglycerides concentration by hematology or chemistry instrument, respectively. +/++/+++; semi-quantitative expression of HIL into three levels (merged laboratories answering +/++/+++ and slight/moderate/gross).

Comment- and reject-levels

The HIL-levels at which laboratories stated to comment (report test-result with a comment) or reject (report comment without test-result) were very heterogeneous for the coagulation tests evaluated, even within each manufacturer group (Supplemental Tables 2A–4A). Several laboratories answered this question with text, rather than the required quantitative or semiquantitative HIL-levels at which they comment and/or reject results (Supplemental Tables 2B–4B).

Among laboratories not reporting comment- and/or reject-levels, several stated using “Individual assessment” (45–55 %) or that they evaluated coagulation curves (35–45 %), while some stated switching to other wavelengths or using an alternative instrument (Supplemental Figure 3A–C). “Individual assessment” was stated more often for samples with hemolysis, while “Individual assessment” and “Evaluation of coagulation curves” were stated equally for icterus and lipemia. Minor differences were seen between handling of the different coagulation tests. Stago users tended to state “Individual assessment” more often, while Sysmex and IL users stated “Individual assessment” and/or “Evaluation of coagulation curve”, and the majority of Siemens users stated “Alternative wavelength” (80 % for APTT, PT, INR and fibrinogen and 60 % for D-dimer and antithrombin) (Supplemental Figure 4–6). Some Stago (mechanical clot-detection) users stated that their coagulation tests were not affected by HIL, while only one optical clot-detection user stated this.

When HIL-levels are higher than their own reject-level, 45–60 % of the responding laboratories answered that they would never release the results, with a tendency that results more often were released in cases of icterus and less often for D-dimer and antithrombin (Figure 6). Some laboratories would release results in case of in vivo hemolysis (≈30 %), liver failure (≈30 %) or chronic lipemia (≈20 %), for H, I or L, respectively. When difficult to draw a new blood sample, 25–30 % of the laboratories would release the potentially biased results, approximately 10 % if it was urgent (STAT) and 10 % for results within the reference interval (Figure 6). Multiple answers could be given for this question.

Figure 6:

The percentage of laboratories stating never to report the results if the HIL-level is above their own rejection-level (never report) and the percentages reporting in different specific situations for hemolysis (A), icterus (B) and lipemia (C), for each of the different coagulation tests. STAT; urgent sample.

The comment- and reject-levels for HIL for each coagulation test were stated mainly to be derived from the manufacturers (70–80 %), but approximately 35 % reported the laboratories’ own experience as one of the sources. Another 10–20 % claimed to use experts, guidelines, or published studies, but very few relied on studies carried out in their own laboratory (1–6%) (Supplemental Figure 7). Multiple answers could be given for this question.

Type of comments given

Standardized comments for H, I and L were used by 72 , 59 and 64 % laboratories, respectively, while fewer used individualized comments (17 , 19 and 21 %) or did not provide comments (11 , 22 and 15 %). The direction of the potential bias caused by HIL (e.g. increased/decreased) in the sample was included in the comments by less than 10 %, with no agreement on increase/decrease for any coagulation test or interfering substance (Supplemental Figure 8).

Reporting of HIL-levels and internal/external QC

Some laboratories (31–33 %) stated to include results of HIL-checks in the laboratory system available for the laboratory staff, but few (7–8%) included this in reports to the clinicians. Only 1.2–2% measure in-house IQC material and about 5.5 % use commercially available IQC. EQA programs for HIL in coagulation testing were joined by 22 % of the laboratories measuring HIL, while 67 % stated to be interested in participating in EQA for HIL in coagulation testing.

Frequency of HIL-samples

The laboratories observed hemolysis more frequent than samples with icterus and lipemia, similar to other studies [1], 7], 26]. Hemolysis may originate both in vivo – hemolytic disease or mechanical (e.g. intravascular devices or artificial heart valve) or in vitro – preanalytical issues like e.g. difficult venipuncture, thin needles and vigorous mixing [1], 26], while icterus (e.g. caused by cholestasis or inherited diseases) and lipemia (caused by high fat meals, metabolic diseases, parenteral infusion or propofol anesthesia [7], 27]) – only occur in vivo. Continuous improvement in the quality of sample handling would eventually reduce the number of in vitro hemolytic samples. Some in vivo lipemic samples may be avoided with fasting samples and stopping parenteral infusion, but ultracentrifugation may be an option to clear lipemic samples [17], 28], 29].

Procedures for HIL-detection

The percentage of laboratories checking for HIL-samples in coagulation testing seem to have increased over the last few years [30], probably because of increasing availability of automatic HIL-check on the newer coagulation instruments [18], 19], 31], 32] and increasing use of transport bands [33]. However, it is of great concern that visual checks alone, still are used to a large extent (almost 40 %), as also shown by another survey [34]. Automatic HIL-detection is recommended as visual checks have been shown to be unreliable [9], 15], [35], [36], [37].

The reason for the slightly lower percentage of Stago users performing HIL-checks may be that mechanical clot-detection is not affected by the analytical interference (optical/spectrophotometric). However, biological interference (mainly hemolysis, but also lipemia) may affect results from coagulation tests both for instruments with mechanical (Stago) and optical (IL, Sysmex and Siemens) clot-detection methods [2], 4]. In addition, immunological and chromogenic methods (e.g. D-dimer and antithrombin) use optical detection, independently upon the instrument’s clot-detection method, and lipemia might also affect antibody binding [7]. Consequently, HIL-checks are important regardless of the instrument used.

Comment- and reject-levels

The large heterogeneity in the comment- and reject-levels for HIL and how these are reported, cannot only be explained by using different methods, as a large heterogeneity was also found between laboratories using the same manufacturer. Since most laboratories state to use manufacturers information, part of the heterogeneity could be caused by different interpretation of the information given in the package inserts (e.g. some laboratories stated to report the result with a comment at the same HIL-level/concentration as others would reject). How manufacturers performed their HIL-studies and decided upon the acceptability criteria (allowable bias) as basis for stating not to be affected by HIL up to certain cut-offs, are usually not available [7], 15], 16]. For some methods (mechanical clot-detection) HIL-interference is not always mentioned in the package inserts for the clot-methods. This also makes it more difficult to understand and adhere to manufacturers’ information, which is actually legally binding.

Automatic HIL-detection by coagulation instruments is performed by optical absorbance measurements at different wavelengths. Stago and Sysmex instruments reports this in HIL-levels (e.g. 0–5), each corresponding to an approximate concentration-interval, while IL-instruments transforms optical absorbance measurement into concentration intervals (clouds) for hemolysis and bilirubinemia and milliabsorbances (mAbs) for lipemia/turbidity. Since HIL cut-offs are given in mg/dL or mAbs, respectively, more IL users give comment- and reject-levels in these units, instead of HIL-levels as for other instruments. However, qualitative, quantitative and semi-quantitative cut-offs were seen among all laboratories, regardless of manufacturer.

Few laboratories have the resources to perform in-house HIL-studies, potentially leading to sample-rejection although HIL-levels may be too low to induce clinically significant bias [20] or no sample-rejection even though HIL-levels may cause clinically significant bias [38]. Both the complexity of performing in-house HIL-studies and lack of manufacturers’ transparency may contribute to the large heterogeneity seen in comment- and reject-levels in this survey. Published studies comparing instruments and reagents are needed to make more detailed guidelines, as published studies, with a few exceptions [18], 21], represent one instrument and a few reagents [4], 9], 10], 19], 20], [22], [23], [24]. However, a far easier approach would be if instrument and assay manufacturers would share the raw data on which their recommendations in their package inserts/instrument applications are based upon, a call made by the working group “Preanalytics” (WG-PRE) of the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) already in 2018 [16]. Harmonizing HIL-levels and reporting in the same unit (preferentially as quantitative concentrations) would also make studies comparing different methods feasible and possibly encourage manufacturers to develop tools to decrease susceptibility to HIL when possible.

In special cases, some laboratories release results even if HIL-levels are higher than the reject-levels. Such decisions may be taken after risk assessment, as laboratories should balance the risk of reporting (sometimes slightly) erroneous results against the risk of withholding clinically important results (causing delayed results, diagnosis and treatment), both potentially affecting patient safety [39]. In situations, in which a re-draw of samples is difficult or in very urgent situations, a biased result may in some instances still be clinically useful, if information on the direction and magnitude of the measurement error are available [39]. Unfortunately, for coagulation tests, the magnitude and direction of changes are not always easy to assess (see below Type of comments given).

Several of the laboratories without comment- and reject-criteria reported to assess HIL-samples individually (“Individual assessments”), which may lead to heterogeneity of handling HIL-samples within the laboratory. Written procedures should be present to harmonize handling of HIL-samples within each laboratory. Flagging of erroneous coagulation curves and/or switch of wavelength may be performed by manufacturers’ instructions for some of the optical instruments, but few published studies are available to evaluate such procedures [40].

Type of comments given

It is understandable that most laboratories do not add the direction of the bias in their comments for HIL-samples. Stating the direction of the biased results for a particular coagulation test may be possible if the interference is only optical (analytical), based on the method of that test, but may be impossible for clot-based assays when both analytical and biological interferences may be present. Manufacturers do not provide information on the direction and performing in-house studies are challenging.

Results from published studies may be difficult to implement in own laboratory as the extent of interference found will depend upon several factors; among others the HIL-level in the sample, the analytical method (i.e. clot-, chromogenic- or immunological method), the coagulation test, reagent and instrument used, the type of clot-detection method (e.g. optical or mechanical) for clot-based assays and the concentration/level of the coagulation parameter measured (normal levels in healthy and/or pathological in disease or anticoagulant treatment) [4], 9], 10], 18], 19], [21], [22], [23], [24]. In addition, different methods for producing HIL-samples [41] (e.g. artificially hemolysis by freeze-thaw cycles or aspiration of blood through a thin needle, or using real-patient paired samples with and without hemolysis [4], 9]), may affect the measurement differently depending on the method used in the preparation of samples. Furthermore, conflicting conclusions may be caused by different methods used to define comment- and reject-levels, with some using an arbitrary ±10 % change in concentration from baseline, and others using analytical change limit and reference change value [7], 15] and sometimes the reflection of the needed setting is not available.

Reporting of HIL-levels and internal/external QC

As recommended for chemistry tests, HIL-levels in samples for coagulation testing should be reported in the laboratory report-system and IQC and EQA should be implemented [7], 15], 42], 43]. Increasing availability of automated HIL-detection in coagulation instruments makes reporting of HIL-levels feasible [33]. However, solutions for automatic transfer of HIL-results and material for IQC and EQA may not yet be widely available. Manufacturers should be encouraged to develop systems for this [16] and EQA organisations should be encouraged to develop EQA-programs for HIL-detection in samples for coagulation testing as very few currently exist [34]. As an added benefit, results from EQA-programs may be used to benchmark between manufacturers, showing potential biases among them, which hopefully will encourage harmonization [16], 44].

Strengths and limitations

One strength of this survey is that multiple European countries were included with as many as 440 responding laboratories in total, and more than 70 % of the participants finishing the survey despite its length (58 questions in total, although some questions were skipped depending upon former answers). It is assumed that laboratories with most interest in HIL and preanalytical issues participated, and the large heterogeneity may therefore also be generalizable.

The survey is limited in that the response rate to the survey could not be calculated because the number of laboratories which received the questionnaire is not fully known. Country specific results could also not be assessed since the percentage of laboratories per country is unknown.