Making sense of a haemolysis monitoring and reporting system: a nationwide longitudinal multimethod study of 68 Australian laboratory participant organisations

Introduction

Haemolysis refers to the breakdown of red blood cells (also known as erythrocytes) and the release of haemoglobin into the surrounding fluid [1]. Haemolysis is one of the most common causes of preanalytical errors, which can affect the integrity of a blood specimen and the reliability of laboratory results [2], [3]. The presence of haemolysed specimens has major implications for the quality and safety of patient care [4], [5], constituting an area of major importance for laboratories in Australia and around the world [2]. Benchmark data about the prevalence and variation of haemolysis across laboratories can make a valuable contribution to the development of safe practises to reduce haemolysis and potential errors in laboratory results. Reduction in haemolysis rates can enhance the effectiveness of laboratory services and their contribution to safe and quality patient care.

Benchmark data at local, regional and national level have not been widely available. In 2009, 156 European laboratories participated in a survey about practises for identifying and rejecting haemolysed specimens [6]. The survey was launched under the auspices of the European Scientific Advisory Board (now Global Preanalytical Scientific Committee) [7]. In the United States, 772 laboratories reported their haemolysis identification and rejection practises to a similar survey conducted by Howanitz et al. [8]. Both surveys found that haemolysis practises varied widely, and about 70% of laboratories reported a haemolysis rate below 2.9%. However, these surveys provided information only at a single point in time and did not track changes over time.

The key incident monitoring and management systems (KIMMS) project in Australia has established an ongoing patient safety project to monitor laboratory incidents, including haemolysis. The project was originally developed by the Royal College of Pathologists of Australasia and was made available through the Royal College of Pathologists of Australasia Quality Assurance Programs (RCPAQAP) and funded by the Australian Department of Health via a Quality Use of Pathology Program (QUPP) grant. Laboratories from across all Australian States and Territories submit data on a quarterly basis to a centralised repository to establish a benchmark of incidents and monitor data for a variety of different types of preanalytical errors (e.g. patient misidentification, incorrect specimen labelling, haemolysed specimens and other specimen problems) and postanalytical errors (e.g. results going to an incorrect destination or recipient and amendment or retraction of laboratory results already issued). Although the programme is open to laboratories overseas, at this stage there are only Australian-based participants.

The KIMMS database provides an opportunity to investigate existing laboratory practises for identifying and reporting haemolysis and to establish benchmark data over time. One of the barriers preventing the aggregation of benchmark data is the wide variation of criteria for the identification and rejection of haemolysed specimens currently used by laboratories in KIMMS. The aims of this study were thus (1) to assess suitability of KIMMS for quality monitoring of haemolysis across laboratories and (2) to devise a standardised reporting and analysis method to examine the prevalence and variations in practise of haemolysis rejection.

Materials and methods

Results

Prevalence and variation of specimens rejected due to haemolysis

Given different methods were used for accessioning and for counting haemolysis rejections, the haemolysis rejection rates were reported separately (see Table 5) for the five groups using different combinations of accessioning practises and methods for counting haemolysis. The largest group of participant organisations (n=38) with the most accessions reported (~40 million) who defined accessions assigned by episode and haemolysis rejections counted by specimen reported a mean haemolysis rejection rate of 177 rejections per 100,000 accessions. The second largest group of participant organisations (17 organisations with ~27 million accessions, accessions assigned by episode, haemolysis rejections counted by episode) reported a mean haemolysis rejection rate of 270 rejections per 100,000 accessions.

To examine the haemolysis rejection rates over time, we focused the subsequent analysis on the 38 participant organisations who used the most common method of accessioning and counting haemolysis rejections, i.e. accessions assigned by episode, haemolysis rejections counted by specimen. Haemolysis specimen rejections per 100,000 episodes for each quarter and year of the study period are presented in Figure 1. The quarterly haemolysis rejection rate between 2011 and 2014 fluctuated between 100 and 233 rejections per 100,000 accessions. The variations of haemolysis rejection rates between 38 participant organisations were very similar over time as shown in the 95% CIs in Figure 1.

Figure 1:

Numbers of haemolysis specimen rejections per 100,000 episode accessions from the first quarter (Q1) of 2011 to the fourth quarter (Q4) of 2014 for the 38 KIMMS participants who assigned accessions per episode and counted haemolysis rejections per specimen (overall mean=177 specimen rejections per 100,000 episode accessions as shown by the horizontal line, i.e. 0.18% of all accessions).

Discussion

To our knowledge, KIMMS maintains the largest existing longitudinal haemolysis data set in the world. The results of this study revealed considerable variation between participants with respect to where the majority of accessions came from, which type of staff performed the actual specimen collections, what method was used for haemolysis detection, where it occurred in the laboratory process and how rejections for haemolysis were counted. Broadly speaking, the prototypical KIMMS participant organisations received most of the specimens from inpatients in public hospitals, received a majority of specimens that had been collected by phlebotomists, identified haemolysed specimens within each laboratory department using the HI result generated by a chemistry analyser, assigned accessions according to the episode, counted haemolysis rejections per specimen and recorded this information manually in LIS.

This study also identified areas that could improve the quality of future reporting and monitoring of haemolysis rejections. First, the most important conclusion is that there is a need to standardise reporting methods among the KIMMS participants. Of 68 participating organisations, a variety of methods were used for accessioning and counting haemolysis rejections, this systematically influenced the haemolysis rejection rates and made it inappropriate to compare haemolysis rejection rates between laboratory participants that had used different methods. Without standardisation of these recording metrics, deriving useful comparisons, benchmarks and targets will not be possible. Participants varied both in how they assigned accessions (whether it was according to the specimen or to the episode, definitions see Table 1) and how they counted haemolysis rejections for submission to the KIMMS database (by episode, by specimen or by test). Episodes can contain multiple specimens upon each of which multiple tests might be requested. Because the haemolysis rejection rate was calculated by dividing the number of haemolysis rejections by the number of accessions processed by the participant for a given period, differences in how these were defined systematically altered the resultant rates. Participants who counted haemolysis rejections per test recorded a lower rate of haemolysis rejections (16 rejections per 100,000 accessions for two participants) compared with other participants, but this low rate is probably an artefact of how they capture and report their data.

Second, KIMMS participant laboratories did not report haemolysis detection methods and their operational haemolysis cutoff parameters, including the criteria used for haemolysis rejections. Nor did they report the assay methodology and/or the instrumentation that was used, which may determine the point at which haemolysis interferes with the accuracy of a result. These factors would have influenced the rate of haemolysis rejection reported.

Third, we identified a need for the KIMMS participants to report blood specimen accessions separately. Haemolysis mainly affects blood specimens, while participants contributing data to the KIMMS database reported their activity by the number of accessions for all types of specimens (including tissue specimens, and faeces, etc., for which haemolysis was not necessarily relevant). Although the majority of the specimens collected in the laboratories were blood specimens and haemolysis is one of the most common causes of preanalytical errors [2], [3], laboratories’ haemolysis rates would be systematically affected by the proportion of their activity that was performed on other than blood specimens. When the haemolysis rate is calculated using the total number of accessions (regardless of specimen type), laboratories whose activity was dominated by tests on blood specimens would report artificially higher rates of haemolysis than laboratories that did a smaller proportion of tests on blood specimens.

Fourth, providing regular feedbacks to laboratory participants will help them to track their performance over time, to enable comparison and benchmarking with their peers. We devised a reporting and analysis method to examine the haemolysis rates for participants. Based on participants’ reporting methods, five haemolysis rejection rates were calculated by dividing the number of haemolysis rejections by the number of accessions.

Lastly, we identified the lack of information about KIMMS participants and their haemolysis detection and reporting practises. We conducted a survey to retrospectively collect the missing information (Table 2). This information should be reported when participants begin to report to KIMMS and then regularly updated (e.g. on a quarterly basis). The findings from this study highlight the need for newer data collection methods for haemolysis identification reporting, which take advantage of the widespread use of electronic LIS in laboratory services. We found that two thirds of KIMMS participants recorded the haemolysis rejections manually in their LIS. A 2014 survey of Australian biochemistry laboratories reported that 97% of respondents were using an electronic LIS [10]. Using routinely and automatically collected and recorded data from the LIS reduces the likelihood of errors in the data submitted [11], [12] and facilitates the process of sharing data with the KIMMS database, potentially allowing for more frequent update intervals and a faster feedback loop. Various electronic laboratory systems, while undoubtedly useful, are yet to embrace all of their potential because of problems with the interoperability between systems and the aggregation of data [13], [14]. Aggregating data across these multiple systems, and from laboratories employing different criteria and definitions, would greatly facilitate the monitoring of laboratory quality performance and provide a valuable means to generate research evidence.

Despite the limitations of KIMMS data, the overall haemolysis rates, ranging from 0.02% to 1.3% depending on the ways of reporting (Table 5), concur with other literature. Based on the data from the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC), the median haemolysis rates varied from 0.270% to 0.850% during the period of 2009–2013 [15]. A literature review located 37 studies published between 2000 and 2014, which reported an overall rate of haemolysis and methods used to detect haemolysis (HI and/or visual inspection) [16]. A considerable variation in haemolysis rates reported in the literature was identified. Seven studies reported a haemolysis rejection below 1% of all accessions [17], [18], [19], [20], [21], [22], [23], six between 1% and 3% [24], [25], [26], [27], [28], [29], 20 between 3% and 20% [9], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48] and another four above 20% [49], [50], [51], [52]. Haemolysis detection methods, and free plasma haemoglobin levels used for this detection, are likely to be partly responsible for this variation, in addition to the differences in study designs and study populations. Although further research is called for to understand these factors related to variation of haemolysis rates, future studies should aim to generate benchmark data of haemolysis rates with consideration of the haemolysis detection methods used.

As noted above, the wide range of practises in use for recording both accessions/specimens and haemolysis will confound the development of useful benchmarks and standards for good laboratory practise. As a consequence of this study, a guideline on managing and reporting of haemolysed specimens [53] has already been developed by the Australian Association of Clinical Biochemists and KIMMS custodian, the Royal College of Pathologists of Australasia, and was made available through the RCPAQAP. It is anticipated that this guideline will find its way into routine pathology practise across Australia, and through this, the ability to set appropriate standards for haemolysis, and for laboratories to be able to measure their performance against these standards, will become possible.

This study provided a foundation for further thorough investigation of all other quality indicators, i.e. pre- and postanalytical errors, recorded in KIMMS. Haemolysis is one of 21 quality indicators recorded in KIMMS. The research methods used in this study can be applied to assess other KIMMS quality indicators and potentially generate other benchmark data. Standardisation of haemolysis detection and reporting criteria would make it possible to compare KIMMS data with the other established international data, such as the data from the IFCC [54], [55]. Similarly, standardised detection and reporting facilitate laboratories by tracking their performance and improving the safety and efficiency of laboratory processes and, consequently, patient care through targeted interventions to reduce errors.

Conclusions

The KIMMS project has established a unique repository to benchmark incidence and monitoring data over time for a variety of different types of pre- and postanalytical errors, including haemolysis. However, there is a need to standardise reporting methods to enable comparisons of haemolysis rejection rates between different laboratories. Since haemolysis can affect the integrity of the specimen and the reliability of laboratory results, standardised reporting of the occurrence of haemolysis may support the improvement of the quality and efficiency of the preanalytical laboratory process.