Introduction

Healthcare associated infections (HAIs) caused by multidrug resistant organisms (MDROs) are a major concern in hospitals globally [1]. These organisms include methicillin resistant Staphylococcus aureus (MRSA), multi-drug resistant non-fermenters, carbapenem-resistant Enterobacterales (CRE), vancomycin resistant Enterococci and Candida auris. MRSA is a major healthcare-associated pathogen that is endemic in many healthcare settings and is associated with worse health outcomes and economic costs [2, 3]. Emerging threats such as carbapenemase-producing carbapenem-resistant Enterobacterales (CP-CRE) and MCR-1-producing colistin-resistant Enterobacterales have the potential for rapid spread making it critically important for aggressive infection prevention and control measures [3, 4].

Well planned infection prevention and control strategies are critical in preventing MDRO associated HAIs. Singapore is a major travel hub with many people arriving each year to access health services. Decision makers have responded to the threat from MDRO transmission in its healthcare institutions using an extensive range of infection prevention strategies [5]. Universal active surveillance for MRSA and targeted screening for other pathogens play an important role, while detailed manual and non-touch environmental cleaning strategies minimise the risk of spread from surfaces and equipment. Limiting infection transmission through the isolation of patients is another important strategy. There are five specific transmission pathways that could be interrupted by the effective isolation of individuals: patient to healthcare worker (HCW); patient to environment; HCW to patient; environment to patient; and, environment to HCW [6].

Even though the rationale for isolation of patients colonised and or infected with MDROs is strong it appears impractical to provide permanent single-room isolation facilities for all MDRO colonised and/or infected patients in Singapore acute care hospitals. The majority of acute beds in Singapore public hospitals are in ‘Type B’ or ‘Type C’ wards that comprise 4, 6 or 8 beds. Only ICUs, ‘Type A’ wards and specialised isolation rooms are single room with adjoining bathroom and toilet. Temporary isolation spaces were however established for treating COVID-19 patients when the pandemic unfolded in Singapore [7].

We investigate a possible role for temporary ‘pop-up’ isolation rooms that are effective at isolating the patient within a general ward environment. For this paper we consider ‘Rediroom’ a mobile cart that unpacks into an air-filtered isolation room that offers the users hands-free entry [8]. Given that resources for infection control are finite [9] there is a need to identify whether a decision to add this intervention would be cost-effective [10] when compared to current infection prevention efforts.

The study question is by how much are ‘health services costs’ and ‘patient health benefits’ are expected to change from a decision to implement a temporary isolation room into acute care hospitals in Singapore. This will be cosidered in a framework for cost-effectivceness analysis [11]. The finding will be useful for those managing hospitals in Singapore with endemic MDROs, many with increasing incidence, and inadequate isolation capacity [12].

Methods

Health outcomes and costs

Changes to health outcomes are characterised by the reduction in risk of mortality from avoiding a case of HAI. The data for attributable mortality for a case of HAI are from the first national point prevalence survey [14], Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Hospital mortality outcomes for those with and without HAI.

https://doi.org/10.1371/journal.pone.0271739.t001

A one-proportion z-test enabled an estimate and 95% confidence interval of the probability of death for each type of HAI, see S1 Appendix. These are not adjusted for other factors that might affect risk of death. The costs of a bed day in the public system were taken from Singapore Ministry of Health [18], Table 2. The costs of adoption comprise a monthly capital cost plus a single-use canopy cost per patient, with the estimates used shown in Table 2. All costs are relevant for 2018.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Input parameters for the cost-effectiveness model.

https://doi.org/10.1371/journal.pone.0271739.t002

Results

The expected changes to the outcomes from a decision to adopt a temporary isolation room at an assumed effectiveness of 30% reduction in cases are shown in Table 3. On average there will be 478 fewer cases of HAI per 100,000 occupied bed days from a decision to adopt a temporary isolation room. This will release 1,627 bed-days for other uses, and these are valued at SGD$1.33M in savings. One hundred and thirty-six lives will be saved and 1,754 life years gained.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Changes to outcomes from a decision to adopt a temporary isolation rooms, per 100,000 occupied bed days.

https://doi.org/10.1371/journal.pone.0271739.t003

The joint distribution of expected change ‘total costs’ and ‘life years’ gained is shown in Fig 1. The mean change to total costs is -$SGD329,432, indicating overall that the cost savings from fewer cases HAI exceed the implementation costs. For this cost saving there are 1,754 life years gained. There is a 67% probability that adoption will be cost saving and 100% probability it will be cost-effective against the threshold value of $SGD80,000 per life year gained.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Joint distribution of the expected change to total costs and health benefits from a decision to adopt temporary isolation rooms, per 100,000 occupied bed days.

https://doi.org/10.1371/journal.pone.0271739.g001

The lowest possible value for the effectiveness of the intervention is that 1% of cases of HAI are prevented; the impact of this assumption on the results are shown in Table 4. On average there will be 16 fewer cases of HAI per 100,000 occupied bed days, 54 bed-days are released for other uses, 4.57 lives will be saved and 59 life years gained. The mean change to total costs is $968,967 for a return of 59 years of life. The incremental cost per life year saved is $16,519. There is a zero probability that adoption will be cost saving but a 100% probability that adoption will be cost effective. When mortality benefits are additionally halved the ICER increases to $33,190 per life year gained and the probability that adoption is cost effective remains at 100%. For these scenarios the conclusion is that adoption is a cost-effective decision.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Changes to outcomes with 1% effectiveness used.

https://doi.org/10.1371/journal.pone.0271739.t004

Discussion

The findings reveal the adoption of a temporary ‘pop-up’ isolation room only needs to reduce the cases of healthcare acquired infection by 1% to be a cost-effective decision in Singapore public hospitals. It is likely that the real world effectiveness will exceed this, and so the economic benefits will likely be larger. If adoption achieves a 30% reduction in cases, the expectation is that health services costs would reduce by approximately $330,000 per 100,000 bed days, and there would be many lives saved and substantial health benefits. Who actually enjoys the benefit from the cost savings will depend on the funding model of the hospital and the country, but it is likely that hospitals, government funders and patients themselves would benefit.

A strength of this study is that we included a full economic evaluation for a potentially important technology, which considered both the costs to hospitals and health benefits to patients, and quantified and presented the value of an adoption decision with transparency. This contrasts the majority of the infection prevention and control economic evaluations published in literature, which are often partial evaluations of only hospitalisation costs [26]. While we only compared the decision to adopt to ‘existing practices’, our analyses can be expanded to additionally consider other novel infection prevention and measures.

This study is based on assumptions applied to a model which has limitations compared to a prospective, pragmatic randomised trial. Yet this design would be impossible to blind and complex and slow to implement. Furthermore the time taken would possibly realise opportunity costs in lost savings and lost health gains [27]. In this study we did not consider other factors which could affect implementation. It is possible that there could be a net loss of total beds in a shared cubicle where the typical distance between beds is 1.5 metres. User acceptability will also impact the success of a strategy featuring temporary isolation rooms which must be aesthetically appropriate, comfortable, functional and not associated with stigma. The rooms need to work within nursing workflows, allied health, medical and portering requirements. The advantages of temporary ‘pop-up’ isolation room as compared to making permanent building reconfigurations are most likely related to costs and speed of deployment.

The quality of the data used for the model parameters is good, with the data gathered for the first national prevalence survey [14] utilised for this analysis. The excess length of stay parameters were estimated using a state-based model that appropriately includes the timing of key events of HAI, death and discharge from hospital [20]. A recent review found analyses that use time fixed methods for the estimation of these outcomes generate biased, inflated, outcomes [28]. The estimates of excess mortality due to infection are naïve as they were not adjusted for other known factors associated with increased mortality. For instance, patients who die with an infection are likely older, with more severe disease and more comorbidities compared to those without HAIs. To address this, we conducted a scenario analysis that halved the probability of death, which in effect halved the health benefits estimated by the model, and found that our conclusions regarding cost-effectiveness were maintained.

As our analyses were robust to uncertainty arising from model parameters and to plausible scenarios, we conclude that our study provides some evidence that the that the adoption of a temporary ‘pop-up’ isolation room is likely to be cost-effective to Singapore public acute care hospitals, and may potentially result in reduction of healthcare costs. It is plausible that adoption of this technology is a good decision for other countries in the region, where infection prevention infrastructure is less developed and unlikely to advance in the short or medium term.

Supporting information

Acknowledgments

We are Grateful to Dr Kalis Marimuthu and the steering committee of the HAI PPS surveillance. We are grateful to Kyaw Zaw Linn, Surveillance Coordinator at National Center for Infectious Diseases for organising summary data.