Overview
What was already known
- Previous economic analyses of remote home monitoring for patients with COVID-19 have reported the resources used and the amount spent per patient monitored.
What this chapter adds
- The mean running cost per patient monitored under CO@h services was slightly lower compared with CVW services (£527.5 vs. £599.1).
- For both CO@h and CVW services the mean cost per patient monitored at home was lower in sites using both tech-enabled and analogue modes of data submission compared with the sites using analogue-only modes.
- Mortality rates were slightly higher for the patients under CO@h services in comparison with patients under CVW services (0.9% vs. 0.7% respectively).
- The majority of the staff involved in running CO@h and CVW services were clinical staff. Over 50% of staff (clinical and non-clinical staff combined) were employed at band 5 or below in the CO@h service, whereas in CVW services there were slightly more staff at band 6 or above.
Background
While cost analyses have been carried out for remote home monitoring of some conditions,140 there is a gap in the literature regarding effectiveness, costs, resources and workforce associated with remote monitoring for COVID-19.36 The aim of this study was to evaluate the costs of implementing remote home monitoring for patients with COVID-19 during wave 2 of the pandemic (October 2020–April 2021). Costs related to setting up and running the CO@h and CVW services were identified and measured and stratified by the mode of technology used for patients’ data submission, whether tech-enabled and analogue modes were used, or analogue-only.
The research questions that this analysis addressed were:
- What were the numbers of patients involved (triaged, monitored, deteriorated, escalated, and number of deaths) in the CO@h and CVW services for the period October 2020 to April 2021?
- What was the composition of the workforce involved in providing these services, and did this vary between tech-enabled and analogue and analogue-only data submission modes?
- What were the costs of setting up the CO@h and CVW services and how do these costs vary between tech-enabled and analogue and analogue-only data submission modes?
- What were the mean costs per patient monitored when running the CO@h and CVW services, and how do these costs vary between tech-enabled and analogue and analogue-only data submission modes?
- How much time did sites spend on activities required to run the CO@h and CVW services and how did these times vary between tech-enabled and analogue and analogue-only data submission modes?
Methods
Data collection
We gathered retrospective information using a data collection form provided to the 28 sites that was also used for the analyses in Chapters 4–6. Designated site leads and data leads at each research site were contacted and asked to complete the form in Microsoft Word, and return it electronically to a member of the research team. During May to June 2021, we collected data for the period 1 October 2020 to 30 April 2021 (see Chapter 2 for further information). We collected information regarding number of patients involved (triaged, monitored, deteriorated, escalated, and number of deaths), and the number of staff and resources used for setting up and running the sites (see Project website documentation – Study instruments).
To explore the differences in staff, resources, patients involved and patient outcomes further between tech-enabled and analogue versus analogue-only data submission modes, additional data were collected from four of the 28 sites. All of these four sites were using both tech-enabled and analogue data submission modes. We conducted an in-depth analysis of the time spent per patient for specific activities of the programme and calculated the cost per patient for all the activities. The data collection form for these four sites included additional questions regarding the total number of patients monitored using each data submission mode, staffing (role/band, time spent per patient on every activity) and the frequency of contacts made with patients (see Project website documentation – Study instruments).
Data analysis
Patient and staff numbers
Sites were asked to report the number of patients onboarded on to both services (CO@h and CVW), the number of patients whose health deteriorated and were escalated to different pathways, as well as the number of deaths reported among monitored patients (see Chapter 2 for more detailed information). For sites running both CO@h and CVW services we have summarised the number of patients in each of these services.
We examined the numbers of staff by grade (converted into FTE numbers of staff) during the running of the programme. Staff were grouped into clinical and non-clinical groups. Clinical staff were from a range of backgrounds including medicine and nursing, and non-clinical staff had either an administrative role [e.g. any managerial role, information technology (IT) assistant, etc.] or other support (e.g. drivers, technical instructors). Clinical staff were divided into three subgroups (i.e. bands 5 and below, bands 6–7 and bands 8 and above) and non-clinical staff into two subgroups (bands 5 and below, bands 6 and above). In the workforce analysis, sites were stratified into four groups based on the numbers of patients onboarded: very small (200 patients or fewer), small (201–600 patients), medium (601–1000 patients) and large (1001+ patients). Owing to data limitations, we assumed that the services were running for the same time period in all sites.
The number of FTE staff involved per site could be affected by other factors such as the type of service, data submission mode and the size of the site. We checked for all these factors and reported on the results of the multivariate regression analysis (we regressed the total number of FTE staff, and FTE staff in different groups, against all these factors jointly in an ordinary least squares regression analysis).
Costs
We calculated the costs incurred for setting up and running the CO@h and CVW services, also examining the variation by data submission mode (tech-enabled and analogue vs. analogue-only). The costs of setting up the sites were reported as mean costs per site. Running costs were calculated as mean costs per patient for the period (October 2020 to April 2021) and included: (1) staff that were involved in running the service, (2) pulse oximeters, and (3) other monthly expenditures related to the maintaining of the tech-enabled platforms. Staff running costs were based on the information about staff hours and band.141 We assumed that the unit cost for staff employed at Band 3 and below was equal to that of band 4 staff.141 The number of pulse oximeters was based on the number of patients monitored. It was assumed that 70% of pulse oximeters would be returned and reused. This decision was based on the results from the patient and carer survey (n = 1069) that showed that 69% of patients or carers had been asked to return the pulse oximeter (range: 21.5–100% by site). The oximeters were provided by NHSEI, with the unit cost per pulse oximeter varying between £20 and £25. However, 15 of the 26 sites purchased all pulse oximeters from alternative sources while 2 of the 26 sites purchased part of them.
The mean running cost per patient could be affected by factors like the type of service, data submission mode, seniority of FTE staff and total number of patients monitored. As above, to investigate the effect of all these factors on the mean running cost, we used an ordinary least square regression model.
Additional detailed information provided by the four selected sites was used to investigate the time spent on different activities per patient associated with each patient data submission mode (see Project website documentation – Study instruments).
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.
Measuring cost-effectiveness
In the study protocol we originally planned to undertake a cost–utility analysis of remote home monitoring for patients with COVID-19.55 In the absence of any impact on mortality or hospitalisation being demonstrated by the primary quantitative analyses (see Chapters 5 and 6), a cost–utility analysis was considered uninformative. We therefore focused on the programme costs alone.
Results
A total of 26 of the 28 sites returned the data collection form, of which 13 provided CO@h services, 4 provided CVW services and nine both CO@h and CVW (integrated) services.
Patients onboarded and the impact on outcomes
The number of patients triaged for both CO@h and CVW services (26,126 and 1,761 respectively) was higher compared with the number of patients monitored (17,424 and 1,711). The percentage of the monitored patients over total triaged was much higher for CO@h (149.9%) if compared with CVW services (103.3%). This may be because hospitals had a less heterogenous pool of people qualifying for services and had much more prior information about them. The proportion of the patients escalated and deaths over those monitored was higher for the CO@h services compared to the CVW services (16.6% and 10.8% respectively for escalated patients and 0.9% and 0.7%, respectively, for the number of deaths; Table 10).
| Throughput and outcome | CO@h service | CVW service | ||
|---|---|---|---|---|
| Patients (n) | Monitored patients (%) | Patients (n) | Monitored patients (%) | |
| Patients triaged | 26,126 | 149.9 | 1761 | 103.0 |
| Average number of patients triaged per site | 1188 | 135 | ||
| Patients monitored | 17,424 | 100.0 | 1711 | 100.0 |
| Average number of patients monitored per site | 792 | 132 | ||
| Patients deteriorated and escalated | 2898 | 16.6 | 185 | 10.8 |
| Deaths | 160 | 0.9 | 12 | 0.7 |
| Patients deteriorated and escalated | Patients (n) | Deteriorated and escalated patients (%) | Patients (n) | Deteriorated and escalated patients (%) |
| Emergency department | 2240 | 77.3 | 135 | 73.0 |
| Admitted to the hospital | 1257 | 43.4 | 161 | 87.2 |
| ICU | 403 | 13.9 | 7 | 3.5 |
| Primary care | 1806 | 62.3 | 73 | 39.4 |
Of the 22 sites running the CO@h service, 16 used tech-enabled and analogue modes, and 6 used the analogue-only mode (see Report Supplementary Material 2). Of the 13 sites running a CVW service, 6 used the tech-enabled and analogue data submission modes and 7 the analogue-only mode. Our results show that from the total number of patients onboarded in the CO@h services 76.8% were followed up using the tech-enabled and analogue data submission mode while in the CVW services 60.8% of patients onboarded were followed up using analogue-only data submission mode.
The structure of the workforce involved in providing CO@h and CVW services
The total number of FTE staff involved in running the CO@h service was on average 7.4 per site [range: 1.0–39.0] and for the CVW service this was 2.1 per site [range: 0.1–4.8]. Staff involved were a mix of medical consultants, emergency department staff, general practitioners, nurses, advanced nurse practitioners (ANPS) and medical students. Our analysis of the staff employed in running the services showed that there were no clear patterns in the differences in terms of clinical/non-clinical staff or seniority of the staff involved between CO@h and CVW services.
Table 11 provides the mean number of FTE staff by role and band, stratified by type of service and size of site (see Report Supplementary Material 2 for the total number of FTE staff involved). For both CO@h and CVW services, most of the staff involved in monitoring the patients were clinical (68.4% for CO@h service and 86.7% for CVW service).
| Type of site (by no. patients)a | Sitesa | Staff FTE (mean n) | Non-clinical staff (FTE equivalent) | Clinical staff (FTE equivalent) | |||||
|---|---|---|---|---|---|---|---|---|---|
| Non-clinical staff n (%) | Band 6 + n (as % over total) | Band ≤5 n (as % over total) | Clinical staff mean n (%) | Band 8 + n (as % over total) | Band 6–7 n (as % over total) | Band ≤5 n (as % over total) | |||
| CO@h service | |||||||||
| The total sample | |||||||||
| Very small sites (0–200 patients) | 3 | 3.7 | 1.1 (28.7) | 0.6 (16.6) | 0.4 (12.0) | 2.6 (71.3) | 0.1 (2.5) | 0.9 (24.9) | 1.6 (43.9) |
| Small sites (201–600 patients) | 6 | 4.2 | 1.3 (31.6) | 0.2 (4.3) | 1.1 (27.3) | 2.8 (68.4) | 1.3 (6.4) | 1.1 (26.3) | 0.4 (10.2) |
| Medium sites (601–1000 patients) | 5 | 9.7 | 3.8 (39.1) | 0.6 (5.7) | 3.2 (33.4) | 5.9 (60.9) | 0.6 (5.9) | 2.9 (29.3) | 2.5 (25.7) |
| Large sites (1001 + patients) | 8 | 9.8 | 2.7 (27.4) | 0.2 (2.2) | 2.5 (25.2) | 7.1 (72.6) | 1.1 (22.7) | 3.3 (33.3) | 2.8 (28.6) |
| Total sample | 22 | 7.4 | 2.3 (31.6) | 0.3 (4.5) | 2.0 (27.1) | 5.1 (68.4) | 0.9 (11.3) | 2.3 (30.5) | 1.9 (25.9) |
| Tech-enabled and analogue | |||||||||
| Very small sites (0–200 patients) | 3 | 3.7 | 1.1 (28.7) | 0.6 (16.6) | 0.4 (12.0) | 2.6 (71.3) | 0.1 (2.5) | 0.9 (24.9) | 1.6 (43.9) |
| Small sites (201–600 patients) | 2 | 4.0 | 2.8 (69.3) | 0.1 (2.2) | 2.7 (67.1) | 1.2 (30.7) | 0.3 (6.4) | 1.0 (24.3) | 0.0 (0.0) |
| Medium sites (601–1000 patients) | 5 | 9.7 | 3.8 (39.1) | 0.6 (5.7) | 3.2 (33.4) | 5.9 (60.9) | 0.6 (5.9) | 2.9 (29.3) | 2.5 (25.7) |
| Large sites (1001 + patients) | 6 | 5.9 | 2.0 (34.7) | 0.2 (3.2) | 1.9 (31.5) | 3.9 (65.3) | 1.3 (22.7) | 2.2 (37.4) | 0.3 (5.3) |
| Total tech-enabled and analogue | 16 | 6.4 | 2.5 (38.8) | 0.4 (5.7) | 2.1 (33.1) | 3.9 (61.2) | 0.7 (11.3) | 2.0 (31.2) | 1.2 (18.6) |
| Analogue-only | |||||||||
| Small sites (201–600 patients) | 4 | 4.2 | 0.6 (13.8) | 0.2 (5.3) | 0.4 (8.5) | 3.7 (86.2) | 1.9 (43.9) | 1.2 (27.2) | 0.6 (15.1) |
| Large sites (1001 + patients) | 2 | 21.6 | 4.6 (21.4) | 0.3 (1.4) | 4.3 (20.1) | 17.0 (78.6) | 0.2 (1.0) | 6.5 (30.0) | 10.3 (47.7) |
| Total analogue-only | 6 | 10.0 | 1.9 (19.3) | 0.2 (2.5) | 1.7 (16.8) | 8.1 (80.7) | 1.3 (13.1) | 2.9 (29.2) | 3.9 (38.5) |
| CVW service | |||||||||
| The total sample | |||||||||
| Very small sites (0–200 patients) | 11 | 1.8 | 0.2 (11.5) | 0.02 (1.3) | 0.2 (10.2) | 1.6 (88.5) | 0.2 (10.4) | 0.5 (28.4) | 0.9 (49.7) |
| Small sites (201–600 patients) | 2 | 3.7 | 0.7 (18.1) | 0.0 (0.0) | 0.7 (18.1) | 3.0 (81.9) | 0.1 (2.9) | 2.9 (79.0) | 0.0 (0.0) |
| Total sample | 13 | 2.1 | 0.3 (13.3) | 0.02 (0.9) | 0.3 (12.3) | 1.8 (86.7) | 0.2 (8.4) | 0.9 (42.0) | 0.8 (36.3) |
| Tech-enabled and analogue | |||||||||
| Very small sites (0–200 patients) | 5 | 1.9 | 0.2 (12.7) | 0.04 (2.0) | 0.2 (10.7) | 1.6 (87.3) | 0.1 (4.5) | 0.6 (34.0) | 0.9 (48.8) |
| Small sites (201–600 patients) | 1 | 4.0 | 1.3 (33.2) | 0.0 (0.0) | 1.3 (33.2) | 2.7 (66.8) | 0.2 (5.3) | 2.5 (61.5) | 0.0 (0.0) |
| Total tech-enabled and analogue | 6 | 2.2 | 0.4 (18.9) | 0.03 (1.4) | 0.4 (17.5) | 1.8 (81.1) | 0.1 (4.7) | 0.9 (42.3) | 0.8 (34.1) |
| Analogue-only | |||||||||
| Very small sites (0–200 patients) | 6 | 1.8 | 0.2 (10.4) | 0.01 (0.6) | 0.2 (9.8) | 1.6 (89.6) | 0.3 (15.5) | 0.4 (23.5) | 0.9 (50.6) |
| Small sites (201–600 patients) | 1 | 3.4 | 0.0 (0.0) | 0.0 (0.0) | 0.0 (0.0) | 3.4 (100.0) | 0.0 (0.0) | 3.4 (100.0) | 0.0 (0.0) |
| Total analogue-only | 7 | 2.0 | 0.2 (7.9) | 0.01 (0.5) | 0.1 (7.4) | 1.9 (92.1) | 0.2 (11.8) | 0.8 (41.8) | 0.8x (38.5) |
Table 11 also shows that for all sites running CO@h services, 53.0% of staff were employed at band 5 or below (27.1% non-clinical and 25.9% clinical staff). Sites using tech-enabled and analogue modes had a higher proportion of non-clinical staff at band 5 or below (33.1% non-clinical and 18.6% clinical staff) compared to analogue-only mode (16.8% non-clinical and 38.5% clinical staff).
Across all sites running the CVW services, 51.3% of all staff (50.4% clinical and 0.9% non-clinical) involved were employed at band 6 and above. A similar trend was observed for the sites using analogue-only mode where 54.1% of all staff involved (53.6% clinical and 0.5% non-clinical) were employed at band 6 and above whereas for the tech-enabled and analogue mode most staff involved (51.6%; 34.1% clinical and 17.5% non-clinical) were at band 5 and below.
Our analysis of the effect of other factors (i.e. type of services, data submission mode and size of the site) on the number of FTE staff involved revealed that, as expected, a key determinant was the size of sites. Note that these findings should be considered with caution due to the small sample size (N = 35 sites) (see Report Supplementary Material 6).
The costs of setting-up and running the sites
Table 12 presents detailed information on the mean running cost per patient triaged and monitored over the study period for both CO@h and CVW services (see Report Supplementary Material 6). These data show that most funds during the setting-up phase were spent on non-staff items (e.g. the technology platform, medical and other equipment). Our analysis of running costs showed that the mean cost per patient monitored using the CO@h service was lower compared with the CVW service (£527.5 vs. £599.1). The mean cost per patient monitored was lower for sites using tech-enabled and analogue compared with the sites using analogue-only data submission mode for both CO@h (£515 vs. £561) and CVW (£584 vs. £612) services.
| Cost per patient | Total Covid-19 Care at Home | |||||
|---|---|---|---|---|---|---|
| CO@h service | CVW service | |||||
| Mean cost per patient | 10th percentile | 90th percentile | Mean cost per patient | 10th percentile | 90th percentile | |
| Triaged | £351.80(£73.70) | £67.60 | £512.20 | £581.80(£139.90) | £74.40 | £1366.80 |
| Monitored | £527.50(£110.50) | £101.40 | £768.00 | £599.10(£144.00) | £76.60 | £1407.40 |
| Tech-enabled and analogue mode | ||||||
| Triaged | £343.40(£83.80) | £78.80 | £464.10 | £567.00(£208.10) | £129.00 | £1152.50 |
| Monitored | £514.90(£125.60) | £118.20 | £695.80 | £583.90(£214.30) | £132.80 | £1186.80 |
| Analogue-only mode | ||||||
| Triaged | £374.10(£165.20) | £68.90 | £821.10 | £594.50(£204.10) | £130.10 | £1220.90 |
| Monitored | £560.90(£247.80) | £103.20 | £1231.10 | £612.10(£210.20) | £134.00 | £1257.10 |
The regression analysis of the effect of factors (i.e. type of services, data submission mode, seniority of FTE staff and the number of the patients monitored) on the mean running cost per patient revealed that, as expected, a key determinant was the number of patients monitored. Another significant factor was the type of service (i.e. CO@h or CVW). Even though the type of the data submission mode and the FTE difference between the senior and junior staff involved were not statistically significant predictors of mean running costs per patient, the direction of their effect was consistent with the descriptive findings. Note that these findings should be considered with caution because of the small sample size (N = 35 sites) (see Report Supplementary Material 6).
We investigated whether the mean running costs per patient at each site were associated with the total number of patients monitored. Figures 23 and 24 show a positive relationship (slope coefficient for the CO@h service 0.54 (95% CI 0.19 to 0.89), and for CVW service 2.10 (95% CI 0.16 to 4.03)) between the number of patients monitored and the mean running cost per patient for the CO@h and CVW services, respectively. This suggests that running costs increase as the number of patients monitored increases. However, this finding should be treated with caution for several reasons. First, the number of sites involved in this analysis is small. Second, as evidenced by the two figures, there are outlying sites in terms of mean costs and number of patients monitored. Third, the figures are based on running costs only; if set-up costs were included, these costs would fall as the number of patients monitored increased. Fourth, we do not have data on the maximum patient capacity of sites on which total running costs would be based.
FIGURE 23 The mean cost per patient for CO@h services. Source: Based on data provided by each site; 22 sites used CO@h services. (Figure not reproduced here; see the original article.)
FIGURE 24 The mean cost per patient for CVW services. Source: Based on data provided by each site; 13 sites used CVW services. (Figure not reproduced here; see the original article.)
Time spent on monitoring activities
This part of the analysis used data that were available only from the four sites where we undertook a more detailed investigation of the monitoring activities. All four sites used tech-enabled and analogue and analogue-only data submission modes. The total number of patients monitored at these four sites was 3366, of whom 2824 were monitored using tech-enabled and analogue mode and 542 were monitored using analogue-only mode (see Report Supplementary Material 2).
The most time-consuming activity was patient’s monitoring, and on average less staff time was spent on this under tech-enabled and analogue mode compared with analogue-only (68.0 minutes vs. 72.5 minutes respectively; note that this activity by staff might be removed in the case of self-monitoring by patients). Based on time spent and staff working on each activity (see Report Supplementary Material 6), we calculated the average cost of each activity per patient, for the tech-enabled and analogue and analogue-only modes (£115.1 vs. £98.3, respectively; if the patient monitoring costs were removed, as might be the case in a self-monitoring service with no monitoring by staff, and assuming all other staff inputs/costs remained constant, these costs would fall to £80.0 and £60.8, respectively; see Report Supplementary Material 6). Please note that because of the amount of missing data, these findings should be viewed with caution.
These four sites reported on average 8.7 contacts per patient using tech-enabled and analogue mode throughout the time they were in the CO@h service, and 15.5 contacts per patient using the analogue-only option.
Discussion
Key findings
A total of 26 of the 28 sites contacted provided information for this analysis, of which 13 provided the CO@h service, 4 provided the CVW service and 9 provided both services. In total, 17 sites used tech-enabled and analogue and 9 analogue-only data submission modes. The mean running cost per patient monitored under the CO@h service was lower compared with CVW service. The main driver of the cost per patient was the number of FTE staff and the grade of staff involved, and this was regardless of the type of service adopted. A comparison with findings from wave 1 (March to August 2020)36 showed that the mean cost per patient monitored remained fairly constant from wave 1 to wave 2 for the CO@h service (£553 vs. £527) and increased from wave 1 to wave 2 for the CVW service (£400 vs. £599).
The mean costs per patient in the sites using tech-enabled and analogue were lower compared with analogue-only sites, mainly due to shorter time needed to collect patient medical information. However, as other studies have shown, the potential for technology to save costs must be treated with caution unless the digital assessments are shorter in time and they are proven to decrease follow-up and deterioration of symptoms, thus leading to lower commitment of staff time for overall service delivery.142,143
Comparison with other studies
Previous studies have found that remote patient monitoring for conditions other than COVID-19 was associated with improved clinical outcomes and medication compliance and reduced the number of emergency room visits, hospital admissions and patient travel.144–146 A comparison with data from wave 1 (March to August 2020)36 showed that escalated patient rates for the period October 2020 to April 2021 was higher for the CO@h service (16.6% vs. 10.0%). Whereas for the CVW service the escalated patients’ rates were slightly higher for wave 2 in comparison with March to August 2020 period (10.8% vs. 12.2%). The mortality rates for the period October 2020 to April 2021 were 0.9% for the CO@h and 0.7% for the CVW, slightly lower than the rates previously reported for the period March to August 2020 (1.1% and 0.9%, respectively).36
A previous systematic review found that there was no evidence showing that telemedicine and telecare interventions can be cost-effective compared with conventional health care.147 However, it has been also highlighted that cost-effectiveness analysis of remote monitoring of patients for different diseases is often integrated into other home health-care services making it challenging to isolate and allocate its cost appropriately.147 The evidence on resources used and costs of remote home monitoring for patients with COVID-19 specifically is also very limited. Our previous literature review on similar interventions revealed that very few studies have conducted an economic analysis and in most cases this is limited to describing costs incurred and resources used.32 In general, previous studies have stated that similar interventions may be cost-saving by reducing downstream costs (in terms of avoiding hospital admissions and bed-days).31 Our findings are consistent with a recent Australian study where staffing costs were the main contributors to the total costs of monitoring.148 In other health-care contexts, the use of telehealth for the provision of health care has previously shown to be associated with cost savings, especially in reaching remote areas.149 However, any potential cost saving of technology relies on the assumption that digital assessment is shorter in time and decreases follow-up contacts and deterioration of symptoms, which may result in new admissions to hospital. Caution is warranted as such assumptions may not always hold.142
Strengths and weaknesses
The main strength of this analysis is that we achieved good geographical coverage of sites across England (see Chapter 4), obtained primary data on CO@h and CVW services use and distinguished between the tech-enabled and analogue versus analogue-only data submission modes. Additionally, this study fills a literature gap regarding COVID-related remote monitoring services, since it is the first to provide this level of detail about costs, workforce and activities.
There are several weaknesses with this study. First, we were not able to perform a full economic analysis (i.e. cost–utility analysis) as the effectiveness analysis found no evidence of the CO@h service on mortality or downstream hospital utilisation (see Chapters 5 and 6). Second, the study design did not allow the use of patient-level data, which means we were unable to examine patient-level variation in service costs. Third, cost- and resource-use data provided by some sites were incomplete. For this reason, the in-depth analysis using the four sites was very limited and the results should be treated with caution. Fourth, the selection of the four sites used for the in-depth analysis was based on purposive and convenience sampling, and this may insert selection bias. Fifth, there was wide variation in service models across sites and much variation in how services were designed and delivered (see Chapter 4). Sixth, the generalisation of our findings to all patients with COVID-19 is difficult as pre-existing conditions and level of severity were not taken into consideration. Seventh, the limited number of sites which have been used for this analysis does not allow further grouping in terms of their size, considering both the number of patients and the length of time establishing this programme. Finally, any interpretation of results should be taken with caution as we do not have the information on the maximum capacity of the sites and can only observe the current flow of patients. For sites with a higher planned capacity the underutilisation of resources will mean that the cost per patient may go down with the increase in number of patients.
Implications
Our analysis showed a lower cost of remote monitoring per patient with COVID-19 during wave 2 of the pandemic in the UK. Our results indicated that the tech-enabled and analogue monitoring was less costly, in terms of mean cost per patient, than analogue-only monitoring. It offers a model for future research in the way it covers a range in the size of sites and raises questions about differential practices within the overall envelope of remote monitoring – differences that are necessarily present in any geographically diverse and large-scale health system.
Future interventions on remote monitoring for other health conditions need to consider their impact on primary and secondary care. Such services may differ in costs, band of the staff involved and worked hours based on the health sector of primary impact. Consideration of the differences between the CO@h and CVW services should also be considered given the different pool of patients they serve and the level of information the staff have on them. Future studies should also gather information on total capacity and how this relates to costs; given the fixed costs in delivering services, this may provide a better view on actual costs and may also possibly help future planning. Future research, including more sites, may provide additional evidence and shed light on the areas where the availability of this study’s data was poor. Availability of routine data on health outcomes (instead of having to use site questionnaires) would also facilitate future research. Overall, there is pressing scope for seeing remote monitoring within a more holistic view of patients’ and staff experiences.