Frequency and variability of haemolysis reporting

Background and aims

At an international level, this project aimed to produce a detailed analysis of haemolysis by reporting the incidence rates found in the literature. At a national level, analysis of haemolysis rejection rates was determined using the Key Incident Monitoring & Management Systems (KIMMS) dataset. Locally, the rate of haemolysis was assessed within five study hospitals in Sydney according to clinical and patient characteristics, with the impact that haemolysis had on patient outcomes evaluated. Structured interviews were conducted with a number of laboratories to understand variation in identifying and measuring haemolysis and policies for dealing with haemolysed specimens.

Internationally, the literature review identified superior practices to prevent haemolysis (site, tourniquet duration, type of collector, etc.). Nationally, there was a variation in how laboratories assign and count haemolysis rejections. At the five local hospitals, the overall rate of haemolysis was 1.7%. Patients who were triaged in the ED with the most urgent category had a higher rate of haemolysis than less urgent cases, with ED length of stay 18 minutes longer for patients who experienced one or more haemolysed specimens. Haemolysis was a significant cause for repeat Potassium and Troponin testing, resulting in a significantly shorter time between tests. Specialised collectors resulted in lower rates of haemolysis, suggesting the use of phlebotomists may result in lower downstream costs for repeat testing and shorter stays in hospital.

Aims

• Compare the reported frequency and prevalence, risk and detection variability for haemolysed specimens using Key Incident Monitoring & Management Systems (KIMMS) data sources from contributing biochemistry laboratory data sources (nationally); and the data from a single pathology provider servicing hospitals in metropolitan Sydney and regional NSW.
• Measure the levels of haemolysed specimens involving Troponin from EDs and the number of tests not reported using linkage of hospital data sources and pathology service data, and to examine the impact on test request repeats, and consequent variables affecting patient care (e.g. test rates per patient ED encounter and ED length of stay).
• Investigate the measures employed by the pathology service laboratories to identify variation and their impact on the quality and effectiveness of laboratory processes.

Design and method, benefits and findings

Stage 1 of the project was undertaken using data extracted from the KIMMS database that described the haemolysis rejection incidence rate at 68 participant groups of laboratories across Australia. Stage 2 of the project was undertaken in five hospitals in metropolitan Sydney. The five hospitals were serviced by a single pathology provider which provides comprehensive biomedical laboratory services including the following laboratory specialties: Anatomical Pathology, Blood Bank, Chemical Pathology, Microbiology, Haematology, Molecular Genetics and Immunology. In addition to the LHD encompassing the five study hospitals, the pathology laboratory service also serviced four other LHDs and, in 2012, employed over 1000 staff. Stage 3 of the project was undertaken at the same five study hospitals as Stage 2, and at a selection of regional NSW laboratories belonging to the same pathology service.

Benefits

This project has the potential to benefit a range of different stakeholders in the healthcare system.

Patients/consumers:

Patients and consumers benefit from having the highest quality pathology services. The occurrence of haemolysed specimens can result in invalid or unavailable test results which necessitate follow-up phlebotomies which can be unpleasant and increase the risk of iatrogenic injuries.82 There can also be delays in diagnosis and treatment caused by repeat phlebotomies and this can lead to longer stays in the ED.

Clinicians:

The evidence scan provided in this report describes the clinical and patient characteristics associated with higher rates of haemolysis, particularly how different phlebotomy methods, draw sites, and equipment can result in a higher probability of a specimen being haemolysed. These findings can inform harmonised phlebotomy practices and the development of an evidence-based best-practices model for blood collection.

Hospital pathology laboratories:

Haemolysis is a disruptive event in pathology laboratories and results in laboratory resources being spent on communicating with clinicians and processing additional repeat tests. Having a better understanding of the phlebotomy and patient characteristics associated with haemolysed specimens can facilitate the adoption of an evidence-based best-practices model for blood collection. Measuring, benchmarking, and comparing the rates of haemolysis in different clinical and patient contexts, in different laboratories and across time, is a critical process in monitoring and quality improvement. Performance benchmarks can enhance harmonisation, with the potential to improve quality of practice across sites.

Hospital management:

The analyses of haemolysis rates across sites and different clinical contexts, in particular differences in haemolysis rates between the ED and inpatient settings and between specimens collected by laboratory phlebotomists and clinical staff, can inform hospital decisions regarding the allocation of resources and responsibilities in patient care. In particular, hospital management can consider the clinical contexts where laboratory staff can be deployed to conduct phlebotomies.40

Government Departments of Health and LHDs:

This report revealed that there is variation in practice in how phlebotomies are conducted and that these methodological differences result in different rates of haemolysed specimens. The results of this report can inform the creation and adoption of evidence-based best-practice training and mentorship programs for clinical staff and for protocols for equipment and procedures used during phlebotomies.

Findings

• The analysis of Australia-wide KIMMS data describing the rate of haemolysis rejections revealed that there was variation in how laboratories assigned accessions and counted haemolysis rejections. The group of laboratories accounting for the majority of participants and accessions (laboratories which assigned accessions by episode, and counted haemolysis rejections by specimen) reported a mean haemolysis rejection rate of 0.18% of accessions. The second largest group of participants (accessions assigned by episode, haemolysis rejections counted by episode) reported a mean haemolysis rejection rate of 0.25% of accessions.
• In the analysis of detailed pathology data from five study hospitals in Sydney, the overall haemolysis rate was 1.70% of accessions when considering all accessions (in the Biochemistry and Haematology laboratories only). The overall rate was 2.47% when considering only biochemistry specimens that had been assessed for haemolysis, and the rate was 6.37% when considering only biochemistry specimens that had been assessed for haemolysis and that had had a Potassium test ordered and that had been received from the ED.
• When using the scope of all biochemistry specimens that had been assessed for haemolysis, the overall rate of haemolysed specimens was approximately three times higher for clinical staff (2.33%) than it was for laboratory phlebotomists (0.79%).
• Patients who were triaged in the most urgent triage category (Triage 1) had the highest rate of haemolysed specimens at a rate of 8.28% across all EDs. There was little difference between the overall rate of haemolysed specimens for the other triage categories which ranged between 5.78% for Triage 5 to 6.27% for Triage 4 presentations.
• In the EDs, 2,962 repeat Potassium tests (39.7%) occurred after the preceding Potassium test was haemolysed, and in these cases there was a median interval of 2.2 hours between the previous test and the repeat Potassium test. This was a significantly shorter time than when the previous Potassium test was not haemolysed (median interval of 6.3 hours).
• In the EDs, 1,296 repeat Troponin tests (10.8%) occurred after the preceding Troponin test was haemolysed, and in these cases there was a median interval of 2.5 hours between the previous test and the repeat Troponin test. This was a significantly shorter time than when the previous Troponin test was not haemolysed (median interval of 5.1 hours).
• After adjusting for all the baseline characteristics, we estimated the ED LOS for patients was, on average, 18 minutes longer for patients who experienced one or more haemolysed specimens, than for those who did not.
• The outcome of this project was to produce a detailed analysis of the prevalence and variation of haemolysis at an international level by performing an evidence scan and reporting the incidence rates found in the existing literature, then conducting analyses of the haemolysis rejection rates at a broad national scale using the KIMMS dataset, and finally, at a more specific level, assessing the rate of haemolysis according to clinical and patient characteristics, within five study hospitals, and the impact that haemolysis had on patient outcomes such as ED LOS.

Resources

Frequency and variability of haemolysis reporting