Effect of automated feedback laboratory comments promoting good phlebotomy practice on the frequency of ethylenediaminetetraacetic acid (EDTA) contamination of serum samples

Abstract

Background

Ethylenediaminetetraacetic acid (EDTA) contamination of serum samples is common but under-recognized unless EDTA is measured. Incorrect order of draw with closed phlebotomy (vacutainer) does not cause EDTA contamination. EDTA contamination occurs largely or solely during open phlebotomy due to syringe tip or needle-tip contamination when delivering blood into EDTA sample tubes before other sample tubes or direct transfer of blood from EDTA containing tubes to other tubes. Therefore, preference for closed phlebotomy or following the order of tube fill when open phlebotomy is used may reduce EDTA contamination.

Methods

The laboratory’s comments for EDTA-contaminated serum samples were amended to encourage closed phlebotomy and with open phlebotomy filling of serum tubes before EDTA and fluoride-EDTA tubes. The weekly frequency of EDTA sample contamination, normalized for weekly urea and electrolyte (U&E) requests, was studied 52 weeks before and 43 weeks after amending the comments.

Results

Median (IQR) frequency of EDTA-contaminated samples per week per 10,000 U&Es decreased by 58% [5.6 (3.1–9.2) versus 2.3 (1.1–4.4); P < 0.001] after the introduction of the new comment.

Conclusion

Explicit automated laboratory feedback comments promoting closed phlebotomy and order of tube fill with open phlebotomy were associated with a 58% reduction in EDTA-contaminated samples and thus may play a role in improving phlebotomy practise.

Background

Ethylenediaminetetraacetic acid (EDTA) contamination of serum samples is under-recognized unless EDTA is measured.^1–3 In a recent national survey, gross EDTA contamination, largely identified using surrogate markers, has been estimated to affect 50–100 samples per annum per laboratory. ⁴ Studies have reported that rates of EDTA-contaminated samples range from non-significant to 0.46% and may depend on phlebotomy devices, sample tube type, training and practice of phlebotomists, and contamination detection methods ^1–3,5. In the authors’ laboratory, EDTA is measured and at least 250 EDTA-contaminated samples per annum were detected from 2015 to 2019.

Three mechanisms leading to in vitro EDTA contamination have been proposed: ⁶ (a) incorrect order of draw while using closed phlebotomy systems (vacutainers), (b) transfer of blood collected in EDTA tube to other tubes, and (c) syringe or needle-tip contamination resulting from incorrect order of sample tube fill.

By measuring EDTA, studies have shown that incorrect order of draw under ideal phlebotomy conditions using various closed phlebotomy systems does not cause EDTA contamination.^7–9 More recently, it has been reported that EDTA sample contamination almost always, if not exclusively, occurs following open phlebotomy methods, most likely due to syringe tip or needle contamination with EDTA when delivering blood into EDTA sample tubes before other tubes. ¹⁰ A tiny amount of potassium-EDTA (K-EDTA), transferred via a syringe or needle tip, can cause spurious hyperkalaemia.^6,9,11,12 Easier access to open phlebotomy equipment compared to closed phlebotomy equipment, non-familiarity with the closed phlebotomy system and preference for open phlebotomy were identified as leading reasons for EDTA contamination at our centre. ¹⁰

Although the identification of K-EDTA contamination mitigates clinical management errors resulting from erroneous results, careful phlebotomy preventing K-EDTA sample contamination is the best strategy to minimize the adverse effect on patient care. We, therefore, evaluated the effect on the frequency of EDTA contamination of automated feedback comments promoting closed phlebotomy or correct order of tube fill with open phlebotomy when reporting results of EDTA contamination.

Methods

In our laboratory, EDTA measurement is reflexed when serum potassium is ≥6.0 mmol/L and the haemolysis index (H-index) is <1.25 (equivalent to 1.25 g/L haemoglobin). Haemoglobin at a concentration above 4.0 g/L has significant positive interference in the EDTA assay. ⁶ EDTA, however, is not auto-reflexed when H-index is >1.25 as this is the laboratory’s cut-off for not reporting potassium from haemolysed samples. If EDTA is <0.15 mmol/L, results are validated according to laboratory policy.

Prior to the intervention, if EDTA was ≥0.15 mmol/L and <0.20 mmol/L, the results of the requested tests on the serum sample were released with an automated comment stating ‘Possibility that sample is contaminated with potassium-EDTA, suggest urgent repeat’. Test results were not reported when EDTA was ≥0.20 mmol/L and an automated comment reported stating ‘Sample contaminated with potassium-EDTA, suggest urgent repeat’. These EDTA cut-offs were selected based on the change in serum potassium with K-EDTA contamination. ⁶

The comments were amended to provide explicit feedback and their impact on the frequency of EDTA contamination was assessed. If EDTA was ≥0.15 and <0.20 mmol/L test results were released with the comment ‘Possible potassium EDTA contamination, please send repeat specimen. EDTA contamination occurs with open phlebotomy. Recommend closed phlebotomy (vacutainer) or with open phlebotomy fill gold top tube before purple and grey top tubes’. EDTA results ≥0.20 nmol/L received the comment ‘Potassium EDTA contamination, suggest urgent repeat. EDTA contamination occurs with open phlebotomy. Recommend closed phlebotomy (vacutainer) or with open phlebotomy fill gold top tube before purple and grey top tubes’.

The new automated comments were implemented in the laboratory information system (LIMS) in the first week of December 2020. A 4-week run-in period was excluded from analysis to verify that any effects were attributable to the intervention. Data on EDTA sample contamination were collected for 52 weeks (01/12/2019–28/11/2020) before and 43 weeks (03/01/2021–30/10/2021) after the 4-week run-in period following the intervention.

The COVID-19 pandemic and national blood tube shortage ¹³ may have impacted the laboratory’s workload during the period studied. Therefore, the frequency of EDTA-contamination was expressed as the number of samples with EDTA ≥0.15 mmol/L per week per 10,000 renal profiles (U&Es).

Greiner serum separator (item 454214), K₃EDTA (item 454217) and fluoride-EDTA (item 454085) tubes are used in the authors’ institution. Ethylenediaminetetraacetic acid was measured using a validated assay on an Architect c16000 (Abbott Laboratories, USA) ⁶ based on a previously described method. ¹⁴ The inter-assay coefficient of variation (CV) for EDTA was 6.8% at 0.16 mmol/L and 3.8% at 0.26 mmol/L for in-house fresh frozen plasma-based controls during the last month of the study. The inter-assay CV for potassium was 0.8% at 2.7 mmol/L, 0.9% at 3.7 mmol/L and 0.6% at 6.5 mmol/L for Technopath Multichem S Plus during the last month of the study. The monthly inter-assay CV for EDTA and potassium for the above-specified concentrations during the study period was ≤8% and ≤1.5%, respectively.

Data were tabulated in Excel 2019 (Microsoft corp.) and statistical analysis was performed using SPSS Statistics for Windows version 28 (IBM Corp.). Since data were non-parametric (Shapiro–Wilk test p < 0.001), the Mann–Whitney U test was used to measure the change in frequency of EDTA contamination before and after the introduction of the new comments. The threshold for statistical significance was 5%. Data are expressed as medians with inter-quartile ranges (IQR).

Results

U&Es were analysed on 420,745 and 380,945 samples during the 52 weeks before and 43 weeks after the introduction of the new comment. Of these, 279 and 116 samples had EDTA ≥0.15 mmol/L, respectively.

U&Es per week were similar (P = 0.066) before [9069 (8248–9400)] and after [8484 (6876–9730)] the intervention. The frequency of samples with EDTA ≥0.15 mmol/L per week decreased [5 (3–7) versus 2 (1–4); P < 0.001] after the introduction of the new comment. The frequency of samples with EDTA ≥0.15 mmol/L per week per 10,000 U&Es decreased by 58% [5.6 (3.1–9.2) versus 2.3 (1.1–4.4); P < 0.001] after the introduction of the new comment.

Discussion

A UK national survey of specimen contamination reported a median of 76 EDTA-contaminated samples per annum (median 19 per quarter, SD 57.6) per laboratory. Apart from one laboratory that measured serum EDTA, participating laboratories identified EDTA contamination using surrogate markers (hyperkalaemia, hypocalacaemia, hypomagnesaemia and hypophosphatasia). ⁴ In the present study, we report a higher frequency of EDTA-contaminated samples than the survey median probably because EDTA measurement leads to greater and more certain identification of EDTA-contaminated samples.

The laboratory’s previous comment only informed the requestor of EDTA contamination. As part of this study, the comment was amended to inform on clinical practice by highlighting that EDTA contamination occurs almost exclusively with open phlebotomy and a recommendation was made to use closed phlebotomy (vacutainer) or with open phlebotomy to observe the correct order of tube fill.^7,8,10 This educational comment was associated with a 58% reduction in the frequency of EDTA-contaminated samples adjusted for laboratory workload. Although the study period was during the COVID-19 pandemic and national blood tube shortage, weekly U&E requests were similar before and after the introduction of the educational comment. It is, however, possible but unlikely that other changes related to COVID-19 and national blood tube shortage such as different patient population, change in requesting practices and different staff performing phlebotomy could have influenced the reduction in EDTA-contaminated samples. However, the annual frequency of EDTA contamination during the study period prior to the introduction of the new comment was comparable to the annual frequency of EDTA contamination from 2015 to 2019.

Interpretative comments are widely believed to add value to laboratory practice by improving patient safety, user satisfaction and outcomes.^15–20 They are appreciated by clinical colleagues and may be of educational value.^21,22 The provision of interpretative comments is considered an important element of the post-analytical phase and the International Standards Organization’s guidance for medical laboratories (ISO 15189:2012) advises the provision of interpretative comments on results where applicable.^15,23 However, evidence for the efficacy of interpretative comments in achieving outcomes is limited and generally focuses on aiding result interpretation, facilitating clinical management decisions and modifying test-requesting behaviour.^{16,18–20,24,25} This study is the first, to our knowledge, reporting an association between implementation of educational laboratory comments and reduction in a pre-analytical error. No other factors that may have affected EDTA contamination, such as changes in phlebotomy devices, phlebotomy training, and sample handling and rejection practices, occurred during the study period. It is, therefore, most likely that the reduction in the frequency of EDTA-contaminated samples was due to the educational comments. The study design, however, does not prove causality and the lack of a control group may be considered a limitation. Identification of a control group, however, would be difficult in a real-world single centre study particularly given the absence of other laboratories offering definitive EDTA measurement to detect EDTA contamination. ⁴ The reduction in EDTA contamination in this real-world scenario may, however, also be considered a strength since findings of tightly controlled studies may not necessarily reproduce or be applicable in routine practice. ²⁶

A limitation of this approach is that the person reviewing the results (generally doctors and nurses) may be different from the person who collected the sample (generally phlebotomists but also junior doctors and nurses) and therefore the feedback may not reach the individual collecting the blood sample. In this study, however, the feedback comments were associated with a reduction in the frequency of EDTA-contaminated samples probably because in our centre most EDTA-contaminated samples are collected by doctors as was identified in a previous study. ¹⁰ Other potential ways to decrease EDTA sample contamination include reviewing local phlebotomy practices and devices, assessing accessibility of closed phlebotomy devices and optimizing phlebotomy practice including awareness of ‘order of tube-fill’.^5,10 None of these, to our knowledge, took place during the study period.

Another study limitation is underestimation of EDTA contamination, since our laboratory uses a single criterion of serum potassium ≥6.0 mmol/L to reflex EDTA measurement, which was unchanged during the study period. This approach may identify all clinically significant spurious hyperkalaemia attributable to K-EDTA contamination but may not identify all samples with clinically significant K-EDTA contamination, for example K-EDTA contamination causing pseudonormokalaemia masking hypokalaemia.^2,6 The laboratory’s current approach may also not identify spurious hypocalcaemia, hypophosphatasia or hypomagnesaemia in samples in which U&E have not been requested. It is, therefore, likely that the frequency of samples with EDTA contamination in the laboratory is higher than reported in this study. Modification of the laboratory’s algorithm to additionally reflex EDTA based on delta change in potassium and clinically significant hypocalcaemia and hypomagnesaemia may improve the detection of samples with EDTA contamination. This, however, will increase the EDTA tests performed and decrease the pre-test probability of a ‘positive’ EDTA test.

In summary, evidence-based, explicit, simple and standardized educational automated laboratory feedback comments were associated with a 58% reduction in EDTA-contaminated samples by improving phlebotomy practice. We suggest that this approach may be usefully employed to reduce other pre-analytical errors.