The present study used an early economic model to compare the cost impact of using Xpert Carba-R compared with chromogenic media culture for screening patients at high risk of CPE infection or colonisation. The results show that, in the scenarios explored, Xpert Carba-R is cost saving compared with media culture.

While Xpert Carba-R has higher technology costs, there are cost savings associated with preventing unnecessary isolation and freeing up space. Current guidance recommends that individuals are preemptively isolated while awaiting test results, where possible [19]. Culture testing results can take up to 3 days, resulting in a long preemptive isolation time. Xpert Carba-R results take 50 min so reduces preemptive isolation time and thus the time to accommodate negative patients on the ward. As a result, the general ward stay costs are higher with Xpert Carba-R compared with chromogenic media culture. However, our analysis still indicates that £818 can be saved for each person screened with Xpert Carba-R, due to the reduction in preemptive isolation time. It is also likely that the cost of Xpert Carba-R will be discounted in practice, meaning the unit costs of Xpert Carba-R will be lower than those used in the model and cost savings could be even greater.

Moloney et al. [10] also used an early economic model to estimate the cost savings of using a PCR-based diagnostic testing strategy compared with chromogenic media culture for CPE identification in the UK NHS [10]. The study found that a PCR testing strategy saved £462.13 per person for patients spending 5 days in hospital and £847.37 per person for patients spending 15 days in hospital. The similar cost savings reported support the findings of the present study, which assumed an 8.30-day and 14.10-day average hospital stay for uncololonised and colonised patients, respectively. However, Moloney et al. [10] assumed that only one negative result was required before patients could be accommodated on the main ward, reducing the need for preemptive isolation while awaiting further screening results. Also, the PCR assay used in the model was not specifically Xpert Carba-R, but it used performance estimates based on a range of PCR tests. This could explain why the cost savings estimated for patients spending 5 days in hospital were much lower than the present study.

Unnecessary isolation is associated with bed closures, so the reduction in unnecessary isolation could free up resources and staff time, as well as improving patient satisfaction and quality of life [10]. It is often not feasible to preemptively isolate patients, due to limited availability of facilities, so they may be placed in multi-occupancy bays using standard infection control precautions and contact precautions until screening results are available. If there are a large number of patients with CPE and it is not possible to isolate each patient, cohorting can be used whereby patients with the same acquired CPE organism can be cohorted together in one area. However, this can become problematic if patients with different resistant genes are isolated together[19].

Another way in which Xpert Carba-R could be cost saving compared with culture testing is by reducing onward transmission. Performing active screening as per the guidance can prevent onward transmission, resulting in resource-use savings. However, the faster speed at which Xpert Carba-R can produce reliable results may be advantageous by allowing quicker implementation of infection control measures and further reducing onward transmission. In the model, it was assumed that the time to result with Xpert Carba-R was half a day. However, Xpert Carba-R can provide results in 50 min [3], which means that the model likely underestimates the true capacity and patient flow advantages of using Xpert Carba-R. The opportunity cost of releasing required isolation spaces was not included in this model, though is anticipated to be of considerable benefit to strained services.

The model did not explicitly explore contacts and onwards transmission, as data to inform this were challenging in a heterogenous group. Those who were truly colonised, and either misdiagnosed or not isolated, were assumed to be able to transmit. Transmission was assumed to be limited to the reproduction number during the hospitalised period, which is likely to be extremely conservative and does not consider the likely large number of contacts that will be made in the first 24 h after admittance. It is, therefore, expected that the cost savings associated with preemptive isolation and quick, accurate identification with Xpert Carba-R are likely to be far greater than those demonstrated in this study.

In England, there is variation between trusts in the CPE carriage rate. Previous studies in London have identified a carriage rate of 0.50% compared with a carriage rate of 11.00% in studies in Manchester [11]. Our study assumes that all of those who are high risk will be identified as such, but the proportion who are considered high risk is not well understood. Thus, scenarios were run to examined to understand whether the conclusions would be affected if this were to change and they were not. Where prevalence is high, it is likely that the increased speed to obtain results may have additional benefits that were not modelled, such as efficient use of limited isolation areas. This may also impact onwards transmission in high prevalence areas.

An earlier, interim version of this model was described in abstract form where underlying assumptions were different from those described in this analysis. Clinical feedback on the earlier version increased our understanding of evidence found in the literature, and the model was refined to be as described in this analysis. The total time horizon was reduced, the number of days to result with media culture was increased, no patients were assumed to move to an intensive care setting and people who were not screened were not isolated, as these were not felt to reflect clinically likely scenarios. Regardless, where these scenarios were modelled, Xpert Carba-R was estimated to be cost-saving.

There are several limitations with the current model. The cost of onward transmission associated with CPE has not been explicitly modelled. Instead, it was assumed that transmission was limited to the reproduction number during the hospitalised period. This is likely to be a very conservative estimate as it does not consider the potentially large number of contacts and the cost, clinical or operational consequence of onward transmission or outbreaks.

Another important factor is that different hospitals may have used different types of chromogenic media. As a result, the sensitivity and specificity of chromogenic media may vary between hospitals and impact the associated costs. As the population considered in this analysis were very heterogenous, a high-level view was taken and the authors are aware that different hospitals use different technologies.

In addition, it is important to note that the literature available to support economic modelling in this indication are sparse. As such, simplifying assumptions were required, though the authors took care to test the sensitivity of results to the assumptions as much as possible. As the true value of many parameters could potentially take different values in different scenarios and settings, a probabilistic approach was taken to reflect this uncertainty and scenarios were examined to understand as much as possible about the deterministic results.

This model did not explicitly examine hospital capacity or the number of contacts that people make during their stay, for simplicity. We believe that our approach is conservative and likely to underestimate the true benefit of using Xpert Carba-R, due to the speed of results and resources released. Where hospitals are operating with limited capacity to preemptively isolate people or where prevalence is high, it is likely that a very fast and accurate result that is available on the ward will have a positive effect on capacity and reduce onward transmission.