Kurin Lock for blood culture collection

There are 14 publications, comprising 12 studies, that make up the clinical evidence

3.1 The evidence base consists of 12 studies reported across 14 publications with 4 full-text peer-reviewed publications, 5 abstracts and 5 posters. The EAG critically appraised the 4 full-text publications using the JBI Case Series critical appraisal checklist. The remaining 10 abstracts and posters were not formally critically appraised because of a lack of detail. The EAG considered 3 of the full-text publications to be low quality and the other study to be medium quality. For full details of the clinical evidence, see section 4 of the assessment report in the supporting documentation.

It was unclear in the studies how people were selected for blood culture sampling and how contaminated cultures were identified

3.2 Most of the studies did not specify how people were selected to have blood culture collection. Only 1 of the studies described how laboratory analysis would identify contaminated blood cultures (false positives). Most studies were based in the US, and there was variability in clinical practice for referrals to collect blood culture samples and laboratory analysis to identify contaminated blood cultures. So, the evidence may limit the generalisability of the study results.

Three UK NHS-based studies (not peer reviewed) reported blood culture contamination rates

3.3 There was limited published evidence on blood culture contamination rates in the NHS and the impact Kurin Lock has on this. One of these studies, based in the UK, was unpublished. The UK evidence estimated baseline contaminations of between 5% and 9%. Atta (2022) reported that the contamination fell from 9% to 3.1% with Kurin Lock use, while Hodson (2022) reported a statistically significant change from 6% at baseline to 1.9%.

Other outcomes are estimated based on blood culture contamination rates so the impact of Kurin Lock is uncertain

3.4 Length of stay was not a formal outcome in any of the included studies, but it was briefly discussed in 4 studies. Of these, Atta 2022 and Parsons 2023 were UK NHS based and 2 studies were from the US (Baxter 2020 and Burnie 2021). Both Atta 2022 and Parsons 2023 were posters that based their results on Alahmadi 2010, which investigated length of stay costs associated with false-positive blood cultures in a general hospital in Northern Ireland between July 2007 and July 2008. Kurin Lock was not used in this study, but the bed day findings were used to estimate cost savings in Atta 2022 and Parsons 2023.

3.5 Similarly, the use of antibiotic treatment was not a formal outcome in the published studies. But it was briefly referred to in 3 studies (Baxter 2020, Burnie 2021 and Ostwald 2021a/2021b). The company economic model used vancomycin treatment based on data from studies based in the US. The use of antibiotic treatment in the model was based on contaminated blood culture rates rather than the direct impact of Kurin Lock.

3.6 Staff adherence was discussed briefly in 2 studies reporting the relationship between adherence to using Kurin Lock and the blood culture contamination rate over a 4-week period. Another study reported that staff adherence ranged between 70% to 75% during a trial use of Kurin Lock.

Kurin Lock is cost saving compared with standard blood culture collection in both the EAG and company models

3.7 The company submitted a decision tree comparing Kurin Lock with standard blood culture collection in an emergency department in a mixed population setting. In the model, after blood culture collection, empirical antibiotic treatment was started in a proportion of the population based on clinical suspicion of bacteraemia. A length of hospital stay was assumed for everyone who had a blood culture taken. The time horizon of the model was the length of stay in hospital, which could be up to 9 days. The decision model showed that Kurin Lock reduced contaminated blood cultures and led to a shorter length of stay as well as reduced antibiotic treatment compared with standard blood culture collection. This resulted in a cost saving of £73 per person in the company base case and a cost saving of £8 per person in the EAG base case. The main driver for the model was the difference in length of hospital stay between Kurin Lock and standard blood culture collection, and the associated cost.

Length of stay and unnecessary antibiotic use are not formal outcomes in the evidence on Kurin Lock

3.8 Length of stay and unnecessary antibiotic use are not formal outcomes in the evidence on Kurin Lock. So, data for these parameters were taken from other sources, based on false-positive tests. The length of stay was taken from a US emergency department setting (Skoglund 2019). The length of stay for a person with a true negative blood culture in an emergency department was 5 days, and was 7 days for people with a false-positive blood culture and 9 days for people with true positive blood culture. The probability of starting antibiotics and the choice of antibiotic (vancomycin) were also from this paper. The underlying bacteraemia risk was from US data and was assumed to be 7.4%.

There is some evidence of cost savings in the UK but there are limitations to this data

3.9 Atta (2022) and Parsons (2023) based their projected cost savings on the results from Alahmadi 2010, which investigated the costs associated with false-positive blood cultures in a hospital in Northern Ireland, rather than collecting resource use data during the trials. Alahmadi (2010) found there was a cost saving of about £5,000 per contaminated blood culture. The EAG considered that this result was driven by the high proportion of people in the Alahmadi study who were in an intensive care unit, where bed day costs are usually higher than in other hospital settings. This suggested that the cost savings may be overestimated in the Atta (2022) and Parsons (2023) studies. The baseline contamination rate is from Atta (2022) for the company base case. For full details of the cost evidence, see section 4 of the assessment report in the supporting documents.

The EAG changed the decision model parameters to make it more appropriate for decision making

3.10 The EAG agreed with all the clinical parameters in the company model apart from the choice of antibiotic. Clinical experts noted that in practice a wide range of antibiotics may be given. The EAG selected gentamycin for the economic analysis. The change in antibiotic in the EAG model did not have a significant impact on the cost savings of implementing Kurin Lock compared with the company model.

3.11 The most significant change to the decision model was the change of the hospital stay cost. The daily hospital costs in the company base case weighted for the population in the emergency department was £881. This uses a daily cost of a short stay from patient-level data for 1 NHS trust. The EAG considered the hospital stay costs to be high and used an alternative approach to calculate them. It applied a non-elective short stay cost for the first day of admission. For subsequent days, it calculated excess stay costs in line with approaches used previously in NICE assessment reports. This resulted in £1,044 for the first day of admission and £377 daily for the rest of the stay, weighted for the population.

The EAG's changes to the model make Kurin Lock less cost saving

3.12 The EAG base case resulted in a cost saving of £8 per person when using Kurin Lock, whereas the company model reported a cost saving of £73. A probabilistic sensitivity analysis using a 20% variance on the EAG base case showed a 62% probability of Kurin Lock being cost saving. A one-way sensitivity analysis showed that the length and cost of stay, rate of blood culture contamination at baseline and reduction in rate of blood culture contamination from using Kurin Lock all have the potential to make Kurin Lock cost incurring or cost neutral. The results from the sensitivity analysis indicate that at baseline contamination rates of less than 3%, there is low probability of Kurin Lock being cost saving. Contamination rates of more than 9% have a high probability of Kurin Lock being cost saving. The sensitivity analysis (see section 11 of the assessment report in the supporting documents) demonstrated that many factors can influence the cost saving potential of Kurin Lock, and this reflects the uncertainty in the savings in different scenarios.