Quality assurance using patient split samples: recommendations for primary healthcare laboratories

Introduction

The use of point-of-care testing (POCT) is rapidly growing in laboratory medicine, offering advantages like accessibility, faster results, immediate decision-making, and ease of use. However, as for all medical laboratory testing, accurate POCT results are critical, since inaccurate results can lead to misdiagnosis, inappropriate treatment decisions, and adverse patient outcomes [1], 2]. In primary healthcare settings, where POCT is conducted by staff with varying laboratory expertise, robust quality assurance (QA) protocols are essential to ensure reliable results [3].

Key factors of QA protocols for POCT include internal quality control (IQC) and participation in external quality assessment (EQA) programs, along with education and guidance on test operation and result interpretation [4]. Additionally, some primary healthcare laboratories perform quality control by using patient split sample comparisons, also known as “parallel testing”, with a central laboratory. This process involves comparing results from the same patient using two different measuring systems, often using two different measurement procedures, typically analyzing a capillary or venous whole blood sample on a POCT instrument and a venous sample (serum, plasma or whole blood) on a central laboratory instrument. However, there are few published recommendations on when and how these split sample comparisons should be conducted, and to our knowledge, there are no publications about what are considered acceptable deviations of the compared results.

A recent publication from International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) committee for POCT [5] recommends routine split sample comparisons where appropriate, but without specifying detailed procedures. Australia and the United Kingdom recommend split sample testing as an alternative to EQA in cases where EQA programs are unavailable, though specific guidelines on implementation are lacking [6], 7]. However, there are some specific recommendations such as in Denmark, where systematic monthly comparisons with defined performance specifications are recommended for selected analytes as a replacement for EQA [8]. However, these recommendations are currently under revision. Further, the Canadian society of clinical chemists point of care testing interest group have published guidelines on how often split sample testing should be performed and the number of patient samples required, depending on the complexity of the POCT instrument and the testing setting [9]. However, these guidelines recommend quite comprehensive split sample routines making it challenging for the primary healthcare laboratories to follow. In addition, no recommendations are given on performance specifications or what to do if results are outside the limits.

The Norwegian organization for quality improvement of laboratory examinations (Noklus) with about 3,500 participants has more than 30 years of experience in ensuring POCT quality in primary healthcare [10]. Noklus provides EQA schemes, along with education and guidance for primary healthcare laboratories. It also offers online quality assurance (QA) protocols to support comprehensive quality management of POCT, including IQC [4]. However, formal recommendations for conducting patient split sample comparisons with central laboratories are lacking.

The aim of this study was to develop practical and easy-to-follow recommendations for primary healthcare laboratories on when, and how, to compare POCT results with central laboratory results using patient split samples as part of the QA protocol, and to establish acceptability limits for the compared results.

Materials and methods

Results

Discussion

This study presents recommendations for primary healthcare laboratories on when, and how, to compare POCT results with central laboratory results as part of the QA protocol using patient split samples. To our knowledge, this is the first study presenting pragmatic and user-friendly guidance for non-laboratory personnel and includes recommendation of which acceptability limits to use. The patient split sample approach serves as a valuable quality assurance tool in primary healthcare when IQC or EQA is not suitable or possible, ensuring the accuracy and reliability of POCT by comparing results with those obtained from a central laboratory.

While split sample testing is designed to verify POCT accuracy, it is important to acknowledge that central laboratory measuring systems may also produce inaccurate results compared to a reference measurement procedure. Therefore, before implementing split sample testing, it is advisable to review the laboratory’s performance in EQA programs to better interpret potential discrepancies. It is also important to assess the stability of the measurand in question to ensure that it remains stable during transport.

One primary application of split sample testing is in situations where no EQA programs are available for the POCT device (scenario A, Table 1). In this scenario, we recommend using at least two patient samples in clinically relevant concentration levels (e.g., one normal and one high/low level) approximately 2–4 times per year. The results should be recorded and assessed against predefined acceptability limits. If one or more results are outside the limits, root cause analysis should be conducted using a dedicated troubleshooting checklist (Table 3).

Some POCT instruments lack dedicated IQC control materials, while others provide controls that behave significantly differently from patient samples, limiting their effectiveness. In such cases, split sample testing provides a practical alternative for monitoring the POCT instrument (scenario B, Table 1). For practical reasons, we recommend using one patient sample in clinically relevant concentration level on a weekly or monthly basis. The frequency can be determined by using the scoring system recommended for IQC routines in primary healthcare [21], however, daily split sample testing is not recommended, as it would result in an unnecessary burden on workload and offers minimal additional benefit [22].

As for all QA results, the split sample results should be recorded, and deviating results should be investigated with appropriate corrective actions taken if clinically necessary. Identified errors should be corrected, and it must be assessed whether previously reported patient results were affected and should be recalled. With appropriate guidance, primary care laboratories should be capable of managing QA processes effectively, even when staff have limited laboratory experience.

Although routine IQC procedures recommend additional quality checks when new reagents are introduced or after instrument maintenance [21], 23], patient split sample testing may not be practical in these situations due to the delayed availability of central laboratory results. Instead, an alternative approach involves analysing a patient sample using both the old and new reagent or pre- and post-maintenance, allowing potential discrepancies to be detected. These results should be documented and reviewed over time, similar to standard analytical quality assurance.

In scenario C (Table 1), the primary goal of split sample testing is to ensure individual patient result accuracy rather than analytical quality assurance. In this situation, it is assumed that the primary care laboratory participates in EQA programs and use IQC materials at a regular basis or uses a split sample approach as recommended in this paper. It is important that user errors and errors related to the operation procedure are eliminated before sending the sample to the laboratory for comparison as outlined in Table 4. If errors are detected, these should be corrected and the sample re-analysed. If no errors are detected and the POCT result appears inconsistent with the clinical presentation, confirmation through laboratory testing (such as a split sample comparison) may be warranted.

If significant discrepancies between POCT and laboratory results are observed, preference should be given to the laboratory result given that the quality control routines in the central laboratory are sound. In cases of recurring deviations which may be caused by e.g. interferences present in the sample from the patient, consideration should be given to permanently analysing that patient’s samples at the central laboratory. Possible causes of persistent discrepancies include patient-related interferences, such as e.g. the presence of antiphospholipid antibodies affecting POCT INR measurements [24].

There are currently no published recommendations for which acceptable limits to use in split sample comparisons. In this study, we have used a pragmatic approach when establishing the acceptability limits to ensure ease of use in primary healthcare settings. The limits were determined taking into account several factors, including biological variation, state-of-the-art, EQA limits, and recommendations from publications, using both the total allowable error (TEa) concept and the maximum allowable uncertainty (MAU) concept [25], 26]. Because these recommendations are intended for non-laboratory personnel, the aim was to provide easy-to-remember numbers and similar limits for as many measurands as possible. Therefore, we did not establish the exact number for each measurand.

The limits must account for various factors influencing differences between POCT and laboratory results and is therefore wide. Consequently, the split sample approach serves as a less sensitive quality assurance tool than traditional EQA and IQC and can thus only detect large errors. This is why we recommend that this approach should never be the preferred choice of performing quality assurance of POCT in primary healthcare [22], it should only be done in special circumstances as described above (scenario A-C). Although sending patient samples to a clinical laboratory may be less expensive than purchasing commercial control materials, we do not recommend replacing traditional IQC and EQA with the split sample approach since it might be less sensitive, and the interpretation of the results can be more challenging. In addition, ethical guidelines generally require that patients provide oral (or written) informed consent before an additional venous sample is collected (for measurement of the same measurand) solely for quality assurance purposes. However, if a venous sample has already been obtained for POCT, usually no such consent is needed.

The strength of this study is that it reflects the experience and opinion of the whole Noklus organisation and not just the opinion of the authors. The recommendations are a product of the organisation’s principles for education and guidance of the primary healthcare for more than 30 years [10]. The limitations are that no experts outside Norway contributed, and that the recommendations are valid only for quantitative tests such as for example glucose meters, CRP and HbA_1c POCT devices, it does not include qualitative rapid tests such as for example SARS-CoV-2 and Streptococcus A, or semiquantitative test such as urine strips.

In conclusion, recommendations for patient split sample procedures have been developed to ensure reliable POCT results in primary healthcare laboratories. These procedures provide a quality assurance framework that can be used in special situations making it possible to detect significant errors and ensure safe patient management. Importantly, they are designed to be simple and user-friendly for non-laboratory personnel, facilitating widespread adoption in primary healthcare settings. Future work will monitor the usability of these recommendations among the Noklus’ participants.