Evaluation of the National Health Services (NHS) Direct Pilot Telehealth Programme

Evaluation of the National Health Services (NHS) Direct Pilot Telehealth programme: cost-effectiveness analysis

Abstract

Objective to evaluate the cost-effectiveness of a pilot telehealth programme applied to a wide population of patients with chronic obstructive pulmonary disease (COPD).

Design: Vital signs data was transmitted from the home of the patient on a daily basis using a patient monitoring system for review by community nurse to assist decisions on management.

Setting: Community services for patients diagnosed with COPD

Participants: Two Primary Care Trusts (PCTs) enrolled 321 patients diagnosed with COPD into the telehealth programme. 227 patients having a complete baseline record of at least 88 days of continuous remote monitoring and meeting all inclusion criteriawere included in the statistical analysis.

Intervention: Remote monitoring

Methods: Resource and cost data associated with patient events (in-patient hospitalisation, accidentand emergency (A&E) and home visits) 12 months before, immediately before and during monitoring, equipment, start-up and administration were collected and compared to determine cost-effectiveness of the programme.

Main outcome measures: Cost-effectiveness of programme, impact on resource usage, and patterns of change in resource usage.

Results: Cost-effectiveness was determined for the two PCTs and the two periods before monitoring to provide four separate estimates. Cost-effectiveness had high variance both between the PCTs and between the comparison periods ranging from a saving of £140800 ($176,000) to an increase of £9600 ($12,000). The average saving was £1023 ($1280) per patient per year. The largest impact was on length of stay with a fall in the average length of in-patient care in PCT1 from 11.5 days in the period 12 months before monitoring to 6.5 days during monitoring, and similarly in PCT2 from 7.5 days to 5.2 days.

Conclusion: There was a wide discrepancy in the results from the two PCTs. This places doubt on outcomes and may indicate also why the literature on cost-effectiveness remains inconclusive. The wide variance on savings and the uncertainty of monitoring cost does not allow a definitive conclusion on the cost effectiveness as an outcome of this study. It might well be that the average saving was £1023 ($1280) per patient per year but the variance is too great to allow this to be statistically significant. Each locality based clinical service provides a service to achieve the same clinical goal, but they do so in significantly different ways. The introduction of remote monitoring has a profound effect on team learning and clinical practice and thus distorts the cost effectiveness evaluation of the use of the technology. Cost-effectiveness studies will continue to struggle to provide a definitive answer because outcome measurements are too dependent on factors other than the technology.

Keywords: telehealth, telemedicine, cost-effectiveness, remote patient monitoring

1Introduction

Seventy per cent of the total health and social care budget is spent on treatment and care of patients with long term conditions (LTC)[1]. Telehealth services are suggested as a solution to control the spiralling cost and improve clinical outcomes. This involves installing equipment in patients’ homes to measure daily physiological parameters. These are assessed by clinicians (nurse, doctor or a medically trained technician) todetermine the potential for deterioration in health status and assess their well-being.

Measurements for health monitoring can include blood glucose, blood pressure, weight, saturated blood oxygen (SpO2) and electrocardiogram (ECG). Social care monitoring can include activity monitoring, bed and chair usage, falls alarms, and use of utilities. Telehealth equipment can be configured in many ways, and the level and complexity of monitoring technology can vary from a simple phone call to report datathrough to the use of videoconferencing [2].

The term telehealth is used interchangeablyin this document with telemonitoring and remote monitoring. We will refer exclusively to telehealth. Many advantages of telehealth are reported in the literature: improved clinical outcomes, reduced number of unplanned hospital and A&E admissions [3], efficient handling of caseloads, and facilitation of self-management of condition by patients.A meta-analysis confirms these findings [2, 3].

Exploiting technology to good effect proves complex because service re-design and clinical engagement is found to be both expensive and challenging. High quality studies on cost are few and inconclusive. In her review of studies on cost-effectiveness of telehealth, Mistry (2012) reports that many studies lacked adequate methodological rigor and details of design [4]. Mistry argues that this lowered the overall qualityand furtherthat most studies were pilot in nature and presented findings on small sample size (100 patients or less) [4]. Polisena et al. (2009) also conclude that although telehealth might hold the potential to reduce cost, the quality of studies assessing and reporting telehealth cost were poor [5]. Whitten et al. (2002) in a review of cost-effectiveness of telemedicine intervention similarly conclude that there was a lack of evidence demonstrating cost-effectiveness [6]. A recent review on telehealth continues to report that the evidence for cost effectiveness of telehealth is “weak and contradictory” [7].

The primaryconcern is thatoutcomes are based on small pilot studies that run over short periods of time (typically 3-6 months) [7],and with little consideration of the impact of services introduced by the pilot. Outcomesare expected to be different when at scale over a prolonged periodand integrated into normal healthcare services. To establish whether this is true, the Whole systemDemonstrator project in the UKundertook a randomized trial with 6000 patients. The conclusion is thata telehealth service in combination with standard care is not cost effective when compared with standard usual care. However, only savings are published withoutany comparison with the costs of the telehealth services. Theauthors recommend that future research on cost-effectiveness of telehealth should be “targeted towards specific patient population and sub population” and should explore the “association between area level factors, patient characteristics (demographics, needs levels for each index condition), and the variation in their service use and cost.” [8].

Further studies [9] also conclude that additionalhome telemonitoring care is not cost effective compared to usual care. This study [9] also reportsvariation in cost for the groups; there was significant difference in costspre and during the project; and many patients had no admission to hospital during the period of the study.

In contrast, this paper reports on the cost-effectiveness of a telehealth programme when applied to a wide population of patients with chronic obstructive pulmonary disease (COPD).It takes into account the characteristics and structure of service delivery.

2Methods

2.1Programme Description

National Health Service Direct (NHSD)launched a pilot telehealth programme for patients with COPD in two Primary Care Trusts (PCTs) (November 2010 in PCT1 and March 2011 in PCT2) in order to evaluate cost-effectiveness of a service. The remote monitoring was managed centrally through NHSD.

(PCTs were the administrative bodies in England responsible for commissioning primary, community and secondary health services from providers. Until 31 May 2011 they also provided community health services directly. Collectively PCTs were responsible for spending around 80 per cent of the total NHS budget. Primary care trusts were abolished on 31 March 2013 as part of the Health and Social Care Act 2012, with their work taken over by Clinical Commissioning Groups.)

The NHSD telehealth programme included: planning and administration of the programme; developing operating policy and procedures and technical requirements; device commissioning and management of the installation process; developing clinical process workflow for the call centre; and reporting and management of data elements for evaluation. The PCTs were responsible for: the operational and service elements of the programme at their respective sites, and participation in planning and reporting of results for programme evaluation.

Evaluation for the NHSD telehealth programme was designed by the joint research team of Brunel University and Chorleywood Health Centre, with input from NHSD and the PCTs. The evaluation was designed so that the majority of data from the patient was gathered through the Health Buddy in order to: simplify implementation (no paper), reduce impact on PCT staff workload, and make the project as close to normal service as possible. The PCTs were selected on the basis of having well-established community teams to manage patients with COPD. In this way we would be able to determine the impact of introducing the technology and not the introduction of service. Ethical approval was gained from NHS Research Ethics Committee.

The BOSCH® Health Buddy remote monitoring system was used for the programme. Peripheral devices included temperature, weight, oxygen level, and blood pressure. It also provided structured questionnaires on symptoms for respiratory distress, cough and sputum, their severity and user acceptability for both patient and nurse professionals. Health education messages were transmitted to the patient.

2.2Population Identification

The evaluation was commissioned by NHSD to determine cost-effectiveness and other outcomes of a telehealth programme. Two PCTs participated and a total of 321 patients were enrolled into the monitoring programme. Of these, 227 were included for statistical analysis using the criteria of having a complete baseline record of at least 88 days of continuous remote monitoring.

The decision to refer a patient to the telemonitoring service was made by the PCT, and was based on the inclusion criteria as in section 2.2.1.

2.2.1Inclusion criteria for receiving the monitoring

Patients having a primary diagnosis of COPD and at least one of the following:

•Two or more admissions or emergency attendances in the 12 months

•Difficulty following a medication regime

•Difficulty following a treatment regime

Patients had to satisfy the following criteria:

•Have a telephone

•Reasonable cognitive ability that enables them to report observations

•Be able to understand English (to use the Health Buddy)

•Reasonable dexterity (to use the Health Buddy and peripherals)

•Not have a severe visual impairment

2.2.2Inclusion in the statistical data set

•Completion of all baseline measures

•Three months (88 days) of continuous remote monitoring

2.3Patient Management

Each PCT was responsible for referring a patient to the telemonitoring service and how the patient was managed within the service. Each patient was provided with the Health Buddy unit and a finger pulse oximeter and instructed to take one SpO2 measurement each day. In the same measurement session, the Health Buddy would ask questions relating to the symptoms of COPD, such as the type of cough and sputum production. Devices to monitor other vital signs were provided according to the needs of the patient. Data was reviewed daily by the nurse and was used to assist decisions on the management of the patient.

2.4Resource and Cost Utilisation

Resource utilisation data was obtained from multiple sourcesand include A&E visits, ambulance usage, and hospitalisation. Cost reports were submitted by NHSD, PCT1 and PCT2.

For comparative purposes, each patient acted as their own control. To avoid the effects of seasonal variation and expected deterioration in health, two control periodswere compared. For each patient we selected:an equal period of monitoring taken to start 12 months prior tomonitoring and so cover the same time of the year(seasonal);and an equal period taken to end immediately before the start of the remote monitoring (deterioration). The levels of service usage for each patient were found for both these periods and the monitoring period.

For the comprehensive cost analysis, pertinent information was defined as useful by the project partners. Two templates were made for the project; the first recorded data about patient usage of health services other than home monitoring, and the second recorded data about the telemonitoring service. These were completed by the PCTs. Analysis of resource usage was separated into two recognisable groups defined by clinical event outcomes, all-cause and COPD specific clinical events. This was based on coding whenever possible. Identifying trends and changes attributable to remote monitoring was to be aided by comparing the service user data of high cost users (as defined by a-priori admission to hospital) before and during monitoring. Two data collections were made, one midway through the project to verify methods and data quality and at the end for final analysis.Resource utilisation and costs included those to establish the programme (training, planning meetings) and for on-going running (administration).

2.4.1Health resource utilisation

Analysis of health resource utilisation was performed by using each patient as their own control and comparing metrics before and during monitoring. In order to compensate for seasonal effects and because COPD is a degenerative disease, data was taken for an epoch of equal duration 12 months before the monitoring period and immediately before monitoring.

2.4.2Healthcare system cost

The telehealthprogramme targeted a population of patients with known high cost and likelihood of future high cost. The eligibility criteria were designed to select those having high resource utilisation and consequent cost. The purpose of the investigation was to compare the savings in resource with the cost of the programme.

Cost data included the initial cost for start-up (attending planning meetings, staff training), on-going management of the programme, and increased personnel time to examine data and visit patients. We assumed a monthly service charge of £100 ($125) per patient was made by NHS Direct to provide the equipment, manage the data and provide the service. The cost also included installation and decommissioning.

2.4.3Healthcare system cost-effectiveness

For this study we chose to undertake define cost-effectiveness as a comparison between the costs to provide the remote monitoring service and the direct savings that in health cost that could be attributed to to the introduction of the service. Service costs included equipment, remote monitor service costs and additional staff resources; direct savings were attributed to reduction in utilization of health service resource including hospitalization avoided, reduced in-hospital stay, and reduced contacts with health services.

Our data did not include quality of life, and so we were unable to undertake cost-benefit analysis.

3Results

3.1Baseline data

Based on the inclusion criteria, 107 patients from PCT1 and 120 patients from PCT2 were included in the statistical data set.The average length of monitoring in PCT1 was 203 daysand in PCT2 was 215 days, giving a total of 21,738 and 25,764monitoring days for each PCT respectively. The mean age of patients on entry to monitoring was 70.9 (s.d. 8.9)with 112 male, and 112 female participants, and was the same for each PCT.Data on gender was not provided for 3 patients.

Health service utilisation data used in this study was provided by the PCTs. Resource usage for hospitalisation and A&E visits was separated by cause of admission (COPD, all other causes).

The study was designed to evaluate service, rather than a controlled trial, and patients were enrolled and withdrawn based on clinical judgement. This resulted in wide variation length of participation. We chose the minimum period for inclusion in the data set in order to be of sufficient duration that we might capture clinical events and was comparable to the period of monitoring in other home monitoring services, such as used for hospital discharge. This period is typically 90 days, but inspection of the data showed that a choice of 88 days of continuous remote monitoring would include several further patients.

3.2Initial and Start-up Costs

Each PCTprovided detailed costs associated with establishing the programme. This included a breakdown of the impact on all staff involved, together with all aspects such as training, attending meetings, planning meetings, additional administration and travel. These costsare reported in Table 1.

There were differences in approaches and costs between the PCTs. PCT1 allocated all COPD patients to a member of its long term conditions team, whereas PCT 2 allocated the less severe patients to its community matrons and the more severe patients to the respiratory team. PCT 1 also allocated significantly greater time to training than PCT 2. It also appears that they appointed dedicated administration support. As a consequence, PCT 1 reported significantly higher start-up costs.

Table 1 Resource Utilisation and Cost

Setup / Running
Salary
(£/hour) / External
Training
(Hours) / Internal
Meetings
(Hours) / Planning
(Hours) / Admin
(Hours) / Total Hours / Total Salary
(£) / Salary Cost+30%
(£) / Meeting Travel
(£) / Patient Travel
(£) / Total Cost
(£)
PCT 1
Nurse staffing
Long term Conditions Team
8A / 29.05 / 1.00 / 31.50 / 6.00 / 0.00 / 38.50 / 1118.43 / 1453.96 / 6.56 / 45.92 / 1506.44
7 / 25.73 / 100.00 / 101.50 / 4.00 / 115.50 / 321.00 / 8259.33 / 10737.13 / 13.76 / 697.16 / 11448.05
6 / 18.78 / 31.00 / 0.00 / 0.00 / 0.00 / 31.00 / 582.18 / 756.83 / 0 / 0 / 756.83
5 / 15.59 / 18.00 / 0.00 / 0.00 / 0.00 / 18.00 / 280.62 / 364.81 / 0 / 0 / 364.81
4 / 12.14 / 1.00 / 0.00 / 0.00 / 2112.50 / 2113.50 / 25657.89 / 33355.26 / 0 / 0 / 33355.26
Total for PCT 1 Respiratory Team / 46667.99 / 47431.39
PCT 2
Nurse staffing
Respiratory Team
8C / 40.66 / 0.00 / 30.50 / 5.00 / 0.00 / 35.50 / 1443.43 / 1876.46 / 0 / 14 / 1890.46
8B / 34.39 / 0.00 / 16.25 / 2.00 / 0.00 / 18.25 / 627.62 / 815.91 / 0 / 0 / 815.91
8A / 28.85 / 2.00 / 73.25 / 44.15 / 4.50 / 123.90 / 3574.52 / 4646.88 / 0 / 81 / 4727.88
7 / 23.39 / 34.75 / 7.00 / 50.00 / 120.30 / 212.05 / 4959.85 / 6447.81 / 0 / 45 / 6492.81
6 / 19.59 / 32.75 / 5.50 / 11.00 / 91.50 / 140.75 / 2757.29 / 3584.48 / 0 / 225 / 3809.48
4 / 13.42 / 0.00 / 0.00 / 0.00 / 1.00 / 1.00 / 13.42 / 17.45 / 0 / 27 / 44.45
3 / 11.54 / 7.00 / 4.00 / 0.00 / 163.50 / 174.50 / 2013.73 / 2617.85 / 0 / 0 / 2617.85
Total for PCT 2 Respiratory Team / 20006.84 / 20398.84
Community Matron
8C / 40.66 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0 / 0 / 0.00
8B / 34.39 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0 / 0 / 0.00
8A / 28.85 / 4.00 / 26.00 / 15.00 / 97.50 / 142.50 / 4111.13 / 5344.47 / 0 / 892 / 6236.47
7 / 23.39 / 1.00 / 0.00 / 0.00 / 0.00 / 1.00 / 23.39 / 30.41 / 0 / 0 / 30.41
6 / 19.59 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0 / 0 / 0.00
4 / 13.42 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0.00 / 0 / 0 / 0.00
3 / 11.54 / 4.00 / 4.00 / 0.00 / 89.00 / 97.00 / 1119.38 / 1455.19 / 0 / 9 / 1464.19
Total for PCT 2 Community Matron / 6830.07 / 7731.07
Total for all PCT 2 staff / 26836.91 / 28129.91
Total for all staff / 73504.90 / 75561.30

3.3Comparison of Outcomes

We determined that inpatient stays were the single dominant cost for health resource utilisation. Changes seen in A&E and outpatient usage were small and thus savings were small in comparison. Cost analysis was only performed for inpatient stays.

3.4Health Resource Utilisation - Inpatient Stays

Table 2 Overall health resource utilisation

12 Months Before / Immediately Before / During / Change 12 Months Before / Change Immediately Before
COPD / Other / COPD / Other / COPD / Other / COPD / % / Other / % / COPD / % / Other / %
A&E / 3 / 80 / 8 / 112 / 7 / 99 / 4 / 133 / 19 / 24 / -1 / -13 / -13 / -12
Outpatients / 144 / 78 / 149 / 86 / 122 / 83 / -22 / -15 / 5 / 6 / -27 / -18 / -3 / -3
Inpatient all
Stays / 64 / 43 / 100 / 52 / 71 / 39 / 7 / 11 / -4 / -9 / -29 / -29 / -13 / -25
Bed days / 603 / 209 / 581 / 329 / 398 / 195 / -205 / -34 / -14 / -7 / -183 / -31 / -134 / -41
Cost / 385920 / 133760 / 371840 / 210560 / 254720 / 124800 / -131200 / -34 / -8960 / -7 / -117120 / -31 / -85760 / -41
Average stay / 9.4 / 4.9 / 5.8 / 6.3 / 5.6 / 5 / -3.8 / -40 / 0.1 / 2 / -0.2 / -3 / -1.3 / -21

Total resource utilisation data (Table 2) indicates a decrease in total inpatient bed days during the monitoring periodwhen compared with both the period 12 months before (34%) and immediately before monitoring (31%).Thisresults in savings of £131,200($164,000) and £117,120($146700) respectively.There is a difference in the cause of the savings made. The savings made during monitoring when compared with data from 12 months before are due largely to reduced average length of stay (-40%) despite a comparative increase in the number of stays (11%). Comparison of hospital usage data during monitoring and immediately before monitoring shows a reduction in the number of admissions (-29%) with a negligible reduction in length of stay (-3%).