Our analysis focused on evaluating the costs of remote home monitoring services for COVID-19 patients in England from October 2020 to April 2021. We assessed the costs of these services from the NHS perspective, considering the above-mentioned period. Follow-up events, such as hospital or intensive care admissions, were not included due to data limitations.

When evaluating new interventions, the National Institute for Health and Care Excellence (NICE) recommends taking into account all relevant NHS costs that could be affected as a result [16]. These include changes in staff numbers and time spent on different activities, training and education requirements, support services (like laboratory tests) and overall service capacity or facilities (including hospital beds, diagnostic services, etc) [17]. Careful consideration of these costs and identification of any constraints on the resources needed to support implementation may significantly impact decision making during prospective evaluations and service planning [18, 19].

We took an NHS perspective to measure costs. All costs were calculated in 2021 UK£ (November 2021: 1GBP = $1.368USD; 1GBP = 1.184EUR). Therefore, we first identified the following items of resource use to include in our analysis: staff; utilisation and maintenance of remote monitoring methods (including digital remote monitoring systems); oximeters; other medical equipment (e.g., thermometers, defibrillators); and other non-medical equipment (e.g., office and IT equipment, stationery). This aligns with NICE methodological guidelines, which recommend breaking down costs into appropriate generic organisational categories and budgetary categories [17] (see also Fig. 1).

The description and categorisation of costs required in setting up and running the CO@h and CVW services

The costs were separated into (i) those related to setting up CO@h and CVW services and (ii) those related to running the service delivery. This distinction was made in line with previous methods used to analyse the costs of telemedicine interventions [20]. The setting-up costs encompassed investments in the remote technology-enabled monitoring platform, procurement of other medical and non-medical equipment required for setting up, as well as staff-related costs (including hours worked and staff involved in setting up the services). The running costs refer to regular and ongoing expenses incurred during the operation of services, including costs related to medical equipment provided to patients, staff involved in service delivery and the maintenance of the technology-enabled platforms. All costs and resource use items were additionally categorised by the mode used to monitor patients (technology-enabled with analogue mode or analogue-only mode).

We quantified costs using a detailed bottom-up approach [21] (see also Fig. 1). This entailed gathering data on resource utilisation and costs for each. Resources used were collected in the appropriate physical units (e.g., hours spent per nurse, physician or other staff by employment band or seniority, etc.) in line with current guidelines and best practices for health economic evaluations [17].

In this section, we explore how the costs of the intervention were determined, using unit costs from reliable sources and data collected from questionnaires. Unit costs extracted from routinely published sources or collected questionnaires were multiplied by the volume of resource use for each item [22]. They were then summed across all items to calculate costs. Due to short-term follow-up (< 1 year), costs were not discounted.

Staff-related costs were calculated using published unit costs per hour and the information collected from study sites about staff hours and salary bands [23]. Clinical staff were divided into three sub-groups based on their salary band (Band 5 or below, Bands 6–7, or Band 8 or above), while non-clinical staff were divided into two sub-groups (Band 5 or below, Band 6 and above). The unit cost per pulse oximeter was reported as varying between £20 and £25 (US$27.36–S$34.2) and we used the mid-point of this range (£22.5; US$30.78) unless a specific unit cost was reported by the site. Unit costs for other non-staff-related resources used (e.g., unit cost of thermometers that might have been given to enrolled patients) were reported through the collected questionnaires.

The costs for similar implementation sites were aggregated (we took the weighted mean value, weighted by the number of patients), enabling us to evaluate the overall financial impact of the intervention and understand how site-specific complexities may influence these costs.

We gathered information retrospectively using an electronic data collection form. The form was sent to 28 purposively selected implementation sites (COVID‐19 remote home monitoring services delivered in NHS trusts or primary care providers) that were representative of a range of regions, in terms of levels of deprivation, population size and urban/rural mix and ethnic group make-up [6]. Data were collected for the period from 1 October 2020 to 30 April 2021. The form distinguished between the setting-up and running stage of the COVID-19 remote home monitoring services, considering the information on the staff and resources used (Fig. 1).

The information on the number of patients triaged and monitored, those whose health deteriorated and were escalated for treatment (following health deterioration) and the number who died was also collected and used to calculate the mean costs per patient.

We calculated the costs incurred for setting up and running the CO@h and CVW services, also examining the variation by data submission mode. We adjusted the resources used and costs incurred per patient by applying weights based on the number of patients monitored by each service (CO@h and CVW) and the total number of patients at each site.

The setting-up costs were only reported as a mean cost per site (as the duration of the running period and the lifespan of digital platforms may exceed our study period). The running costs were calculated as mean costs per patient for the period October 2020 to April 2021. The mean costs per patient were calculated by dividing the total costs of each service by the number of patients for each site. We adjusted the resources used and costs incurred per patient by applying weights based on the number of patients monitored by each service (CO@h and CVW) and the total number at each site. For implementation sites running both CO@h and CVW services, we calculated the mean setting-up and mean per-patient running costs of these services separately.

The number of pulse oximeters used during the study was based on the number of patients monitored. An assumption was made that 70% of these pulse oximeters would be returned and reused. This was based on results from a patient and carer survey (n = 1069), which showed that 69% of patients or carers had returned the pulse oximeter (range 21.5–100% by site) [6]. Therefore, only 30% of the costs of these pulse oximeters were accounted for during the study period.

The mean running cost per patient could be affected by factors such as the type of service, data submission mode, seniority of staff and the total number of patients monitored. We investigated whether the mean running cost per patient was associated with these factors by using a multivariable regression analysis. This method allows us to quantify the relationship between the number of users and costs, adjusting for other service characteristics. The regression analysis controls for the type of service, data submission mode, seniority of full-time equivalent (FTE) staff and the total number of patients monitored to explore the main determinants of the costs. Bootstrapping with 10,000 iterations was used to estimate the standard errors of the regression model coefficients. We investigated the sensitivity of our findings to alternative specifications of staff seniority. Statistical tests and regression analyses were conducted using Stata v.18.

To extend the analysis to a national level, we estimated the total national costs of the CO@h and CVW services by combining our findings with data from other parts of the study [24]. For example, we utilised data from the number of people who fulfilled the inclusion criteria for CO@h and CVW services and multiplied these by the estimated proportion of patients using the CO@h and CVW services and the mean running cost per patient for each service. Finally, the findings of this study were triangulated with other quantitative and qualitative evaluation studies of the same COVID-19 remote home monitoring services [6].