Digital interventions for hypertension and asthma to support patient self-management in primary care: the DIPSS research programme including two RCTs

Article history

The research reported in this issue of the journal was funded by PGfAR as project number RP-PG-1211-20001. The contractual start date was in January 2014. The final report began editorial review in August 2019 and was accepted for publication in March 2022. As the funder, the PGfAR programme agreed the research questions and study designs in advance with the investigators. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PGfAR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2022 Yardley et al. This work was produced by Yardley et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Yardley et al.

Summary of aims and rationale

The overall purpose of the DIPSS (Integrating Digital Interventions into Patient Self-Management Support) programme was to address the question of how digital interventions (DIs) can be used to provide cost-effective support for patient self-management of long-term conditions in primary care. To address this question, we chose to focus specifically on improving management and, consequently, outcomes for two common, contrasting long-term conditions (i.e. hypertension and asthma). We chose contrasting clinical conditions to allow comparison of patients from different age groups, with very different patterns of symptoms and different self-management regimes, as this would enable us to consider which findings were specific to one condition and which might be more common across different conditions or management regimes. The proposed project team brought together (1) researchers with leading international expertise in e-health, hypertension, asthma, behaviour change and health economics, and in developing, trialling and implementing complex health-care interventions; and (2) patient and public involvement (PPI) representatives, including people with experience of hypertension and asthma, and representatives of two relevant patient organisations [Blood Pressure UK (London, UK) and Asthma UK (London, UK)].

Our programme of research was intended to undertake the rigorous development and evaluation necessary to maximise the likelihood of effective integration of DIs within NHS primary care, while identifying and using best practice methods of designing and delivering DIs to ensure that they were considered accessible and useful by patients and clinicians. Our specific objectives were as follows:

• To identify key features associated with maximising feasibility, acceptability (to patients and health professionals), clinical effectiveness and cost-effectiveness of DIs.

• To examine the range of delivery and support modes that can be used for DIs and assess their relative feasibility, acceptability [to health-care professionals (HCPs) and patients], clinical effectiveness and cost-effectiveness.

• To optimise interventions for hypertension and asthma and to carry out feasibility studies in preparation for full randomised controlled trials (RCTs).

• To undertake a RCT of a DI for hypertension to determine the clinical effectiveness and cost-effectiveness of integrating it into routine care.

Summary of research

We proposed three closely linked parallel workstreams (Figure 1). A behavioural and economic workstream [workstream 1 (WS1)] focused on identifying condition-specific and common factors influencing cost-effective integration of DIs into primary care. This research was embedded in two clinical workstreams that developed and trialled DIs for self-management of hypertension [workstream 2 (WS2)] and asthma [workstream 3 (WS3)].

FIGURE 1.

Research pathway diagram for DIPSS.

Workstream 1 undertook detailed intervention planning to identify factors influencing acceptable and cost-effective integration of DIs into primary care and, hence, the required elements and characteristics of the interventions and support to be offered for hypertension and asthma self-management in WS2 and WS3. To inform our planning, we completed systematic reviews of the relevant quantitative and qualitative literature and also drew on our primary qualitative studies of patient and HCP views and experiences.

Workstream 2 and WS3 developed DIs for self-management of hypertension and asthma (respectively), using iterative qualitative research to ensure that the DIs were viewed as acceptable and useful by patients and primary care staff. We proposed that both WS2 and WS3 would complete feasibility trials of the DIs, using quantitative and qualitative methods to evaluate patient and primary care experiences of different delivery formats. WS2 also carried out a full RCT of the cost-effectiveness for reducing blood pressure (BP) over 1 year of the optimum method(s) of delivering the DI compared with usual care, with an embedded qualitative process analysis.

There were two changes from the original proposal that affected the direction of the research programme. First, the feasibility trial for the hypertension intervention in WS2 was integrated into the main trial as an internal pilot study, as no changes to the intervention or trial procedures were required. Second, we originally proposed to test the following two intervention arms alongside a usual care arm: (1) the most intensive support (i.e. at least three face-to-face consultations and telephone and/or e-mail support, with the option of further support if required) for the DI considered likely to be feasible and cost-effective in routine care; and (2) the least intensive support (i.e. one face-to-face consultation at baseline, with further support up to the level of condition, provided only as required) for the DI considered likely to be effective in routine care. Instead, we trialled only one intervention arm (vs. a usual-care arm), which included a minimal level of nurse support (i.e. a compulsory face-to-face or telephone conversation), with the option of more face-to-face, telephone and e-mail nurse support if required. This was deemed the acceptable and efficient compromise between the two originally proposed.

Changes to the digital, clinical and research context since the research programme commenced

Introduction

Some key elements of the HOME BP intervention were informed by a previous programme of work that developed and trialled non-DIs for managing high BP. 15,33 These elements included the frequency of self-monitoring, the algorithm for interpreting BP readings and recommending appropriate action, and the procedure for the HCP to plan three potential medication changes in advance for each patient. This workstream (i.e. WS2) sought to explore how to adapt these procedures to be implemented successfully via an online intervention. This adaptation process is not simply a matter of transferring written materials into a digital delivery format. It is well established that it is vital to ensure that patients and clinicians find the intervention easy to use and are motivated and confident to implement the procedures correctly with only digital support. 34–36 In addition, a secondary aim of WS2 was to examine whether or not digital support for healthy behaviour change could contribute to better self-management of hypertension. Therefore, WS1 also involved developing and adapting our existing digital healthy behaviour change resources for use by people with hypertension in WS2.

Intervention Development Team and patient and public involvement

The Intervention Development Team included clinicians with expertise in hypertension management, e-health and lifestyle change, health psychologists, web developers, representatives of the charity Blood Pressure UK and PPI contributors. Our PPI contributors, including three patients (Shelley Mason, Keith Manship, Cathy Rice) with hypertension and/or stroke and one public contributor (Samantha Richards Hall) with a general interest in DIs, joined the project team and provided essential advice on the grant proposal. All PPI contributors were subsequently invited to each Management Committee meeting to discuss important issues arising in the planning and development of the HOME BP intervention, including decisions about support for healthy behaviour change, insights into patient burden of self-monitoring, and discussions around approaches to participant recruitment and how to promote accessibility of the patient materials.

Early prototypes of the intervention were shared with the PPI contributors for feedback from a patient perspective and this led to important changes to optimise the intervention, such as the introduction of additional optional information for patients who might like to know more about clinical risks of hypertension. PPI contributors also provided an important patient perspective during debates among the research team, for example some researchers were concerned that the motivational quiz might be irritating or hard to relate to, but PPI contributors felt that the quiz was useful and engaging. PPI contributors promoted a focus on patient priorities throughout this phase of intervention planning and development, and provided the opportunity for rapid feedback on the early development of the intervention to maximise potential to meet patients’ needs.

Objectives

This section will describe the substudies used to inform the planning and development of the HOME BP DI, based on evidence, theory and qualitative research. For each substudy or discrete research activity, we report aims, methods, results and practical implications for how it informed the intervention development. The substudies are described in chronological order as follows:

• Phase 1: collate and synthesise evidence, including primary mixed-methods research, evidence from quantitative review of the literature and evidence from qualitative review of the literature.

• Phase 2: behavioural analysis, identifying facilitators and barriers, and how to address them.

• Phase 3: intervention development and optimisation, alongside developing guiding principles.

• Phase 4: mapping facilitators and barriers on to theory.

The section ends by considering how the INDEX (IdentifyiNg and assessing different approaches to DEveloping compleX interventions) actions for developing complex interventions were met in this planning and development process. 25 The INDEX actions were published in 2019 and comprised 18 recommended actions for intervention developers to consider, collated through a systematic synthesis of intervention development approaches.

Phase 1: collating evidence from primary mixed-methods research, and evidence from quantitative and qualitative reviews of the literature

Phase 2: behavioural analysis – identifying facilitators and barriers, and how to address them (described in Band et al.30)

Phase 3: intervention development and optimisation alongside developing guiding principles

The HOME BP DI included both patient and HCP components. The patients completed two online training sessions, which were designed to raise motivation and teach patients how to self-monitor their BP. After the first online session, patients attended a baseline medication review with their prescriber, in which a three-step medication plan was created. The patient then completed 1 week of practise BP readings to increase confidence, after which they were reminded by e-mail to self-monitor their BP for 7 days every month. Patients entered their readings online, and if the average reading was above target for 2 consecutive months then the prescriber received an e-mail alert recommending that the next medication change in the plan was made. A third optional online session became available after 9 weeks to increase motivation and self-efficacy to engage in healthy lifestyle changes for managing high BP. From this optional session, patients could choose to complete one-off online educational modules on reducing salt, eating a healthy diet or reducing alcohol, or could sign-up to a multisession DI to support physical activity or weight loss40 [with the latter only available to those with a body mass index (BMI) over 25 kg/m²]. Supporters (i.e. nurses or health-care assistants who had completed online training for this role) were asked to send monthly support e-mails to patients throughout the trial, and to provide optional face-to-face support as needed. See Appendix 1 for full details of the intervention.

Phase 4: mapping facilitators and barriers on to theory (described in Band et al.30)

Mapping the HOME BP planning and development process to the INDEX actions

New guidance for complex intervention development has recently emerged,25 based on a taxonomy of approaches to intervention development, interviews, Delphi consultation and workshops with developers and stakeholders. O’Cathain et al. 25 completed a comprehensive review of approaches and produced 18 actions that are recommended for consideration during intervention planning and development. For completeness, Table 2 shows a retrospective mapping of the HOME BP intervention planning and development process to the 18 actions from this guidance25 [see Appendix 1 for a full description of the HOME BP intervention using the Template for Intervention Description and Replication (TIDieR) checklist46].

Completing Table 2 provided a useful prompt and a template for describing aspects of the intervention development process that are important, but are seldom currently described, such as details of the decision-making process and planning for efficient future implementation.

Objectives

This section will describe the evaluation of the HOME BP intervention during a 12-month RCT. Aims, methods, results and implications are described for each discrete piece of research as follows:

• A RCT to assess the clinical effectiveness and cost-effectiveness.

• A process evaluation exploring how patients and HCPs experienced and implemented the intervention in practice, including:

  • a patient qualitative process study, examining the perceived benefits and burdens of using the intervention for patients

  • a patient quantitative process study, examining engagement and usage of the HOME BP intervention by patients

  • a HCP mixed-methods process study, exploring HCPs’ experiences of and adherence to using the intervention.

The section finishes with a conclusions section, which draws the findings together.

Randomised controlled trial to assess clinical effectiveness and cost-effectiveness

Results

Figure 2 shows the flow of participants through the trial.

FIGURE 2.

Flow of participants through HOME BP trial. a, Partial withdrawals withdrew from the intervention but consented to be followed up; and b, partial withdrawals in usual care consented to follow-up. Reproduced with permission from McManus et al. 47 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from McManus et al. 47 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Table 3 provides baseline characteristics of the sample.

TABLE 3 - Baseline characteristics of HOME BP participants

SD, standard deviation.

Baseline characteristic	Randomised group
	Intervention (N = 305)	Usual care (N = 317)
Age (years), mean (SD)	65.2 (10.3)	66.7 (10.2)
Female, n/N (%)	145/305 (47.5)	143/317 (45.0%)
Ethnicity, n/N (%)
White	285/304 (93.8)	299/317 (94.3)
Black African	5/304 (1.6)	3/317 (1.0)
Black Caribbean	0/304 (0.0)	1/317 (0.3)
Indian	3/304 (1.0)	0/317 (0.0)
Pakistani	1/304 (0.3)	3/317 (1.0)
Other	10/304 (3.3)	11/317 (3.5)
Index of Multiple Deprivation quintile, n/N (%)
1–3 (most deprived)	36/304 (11.8)	27/317 (8.5)
4–7	108/304 (35.5)	125/317 (39.3)
8–10 (least deprived)	160/304 (52.6)	166/317 (52.2)
Diabetes, n/N (%)	24/278 (8.6)	32/291 (11.0)
Of which: type 1	1/278 (0.4)	1/291 (0.3)
Systolic BP (mmHg), mean (SD)	151.7 (11.8)	151.6 (11.1)
Diastolic BP (mmHg), mean (SD)	86.4 (9.6)	85.3 (9.9)

The 12-month follow-up rate was 89% in both groups. Systolic BP at 12 months was significantly lower in the intervention group (138.4/80.2 mmHg) than in the control group (141.8/79.8 mmHg), with a difference between groups of –3.4 mmHg [95% confidence interval (CI) –6.1 to –0.8 mmHg]. For diastolic BP, the between-group difference at 12 months was –0.5 mmHg (95% CI –1.9 to 0.9 mmHg) (Table 4). Exploratory subgroup analyses suggested that the intervention had a larger effect in younger participants. Self-reported adverse effects showed no differences between the two groups. According to a self-reported symptoms scale, which was used as an indication of side effects, a significantly higher proportion of the intervention group reported weight loss at 12 months, but this was not born out on objective measurement of weight. Although engagement with self-monitoring was relatively high across the sample (with 80% of the sample completing both training sessions and at least three complete sets of BP entries), less than one-third of the sample chose to register on an optional programme for healthy behaviour change.

TABLE 4 - Mean BP at baseline, 6 months and 12 months

SD, standard deviation.

BP measurement	n	Mean (SD) BP (mmHg) at time point
		Baseline	6 months	12 months
Systolic BP
Usual care	282	151.65 (11.10)	140.87 (15.98)	141.83 (16.76)
Intervention	271	151.74 (11.82)	138.69 (17.04)	138.43 (15.99)
Diastolic BP
Usual care	282	85.27 (9.88)	80.18 (10.32)	79.77 (10.10)
Intervention	271	86.44 (9.65)	79.88 (9.68)	80.22 (10.07)

A within-trial cost-effectiveness analysis was conducted from an NHS perspective using data collected on use of services and on the intervention. The reduction in BP of 3.45 mmHg combined with an increased cost in the intervention arm of £38 led to incremental cost per unit of BP reduction in the base case of £11 (95% CI £5 to £29). The increased cost per patient of £38 in the intervention arm was, almost entirely, due to the cost of the intervention (£39.73) (Table 5).

TABLE 5 - NHS cost, primary outcome, QALY and incremental cost-effectiveness: mean value per patient based on differences observed between the usual-care and the intervention arms: imputed and complete case results

Results	Base case (imputed)	Alternative (complete cases)
NHS cost (£)
Usual care	100	100
Intervention	138	138
Difference	38	38
Difference in primary outcome at 12 months (mmHg)	3.45	3.54
Cost/BP (£)	11	11
QALY difference	–0.01	–0.01
Cost/QALY (£)	–3800	–3800

The base case relies on the imputed values; however, the complete-case results were the same for cost and only slightly different for the clinical outcome (see Table 5).

Table 5 also shows a small QALY loss of 0.01 in the intervention arm. Given the combination of higher cost and very slightly reduced QALYs, this means that the intervention arm was dominated by the usual-care arm. However, as QALY differences with regard to improved BP control at 12 months are of less interest than QALY differences with regard to BP control in the longer term, the results of the life-long modelling reported below are of more interest.

The base case included use of NHS services relating to BP, and this comprised the full range of NHS services, including hospital admissions. Although few such admissions were recorded, some were elective procedures that had to have been planned before entry to the trial. Consequently, only hospital admissions that occurred after a change of medication were included in the base-case costing. To test the sensitivity of results to this assumption, a scenario was costed that included all hospital BP-related service use, regardless of timing. Although this scenario made little difference overall, the scenario reduced both the cost difference and the incremental cost-effectiveness.

The mean cost per patient in primary care was similar to the base-case costs, indicating that primary care accounted for almost all costs (Table 6). Within primary care, costs were split roughly 60 : 40 between costs attributable to consultations and costs for prescriptions. Patients in the intervention arm had slightly higher prescription costs, associated with changes in medication and/or dose. However, these costs did not increase the cost of primary care consultations because of the role of the DI. These trends were as might be expected. Further analysis of these changes is planned for a separate publication, which will include changes in the time spent by patients in managing their hypertension.

TABLE 6 - Mean cost per patient in primary care by arm in primary care, disaggregated by consultations and prescription costs

Randomised group	Mean cost (£) per patient
	Consultations	Prescriptions	Total primary care
Usual care	62.5	34.8	97.3
Intervention	55.8	40.7	96.5

As the benefits of reduced BP take the form of lowered risk of cardiovascular disease, long-term modelling was required to capture these effects. The most comprehensive approach involves estimation of lifetime benefits measured in terms of QALYs. Life-years reflect reduced mortality and quality adjustment allows for the effects of non-fatal cardiovascular events.

Rather than develop a new long-term model, we fed the results of the randomised trial into a pre-existing model, which was developed by one of the lead clinicians in the present study for previous trials of BP interventions. 50 The model TASMINH450 (Telemonitoring and Self-Monitoring of Blood Pressure for Antihypertensive Titration in Primary Care) is a Markov patient-level simulation undertaken in TreeAge 2018 (TreeAge Software, Inc., Williamstown, MA, USA). The simulation tracks the costs and consequences of individual patients passing through the model, with characteristics (taken from the trial) free to vary between patients. The model was run over the maximum lifetime of the patients (maximum of 65 years; minimum trial inclusion criteria was age 35 years), a time horizon sufficient to capture all relevant long-term costs and consequences.

All patients started in the well/no event health state. Within a 6-month time cycle, a patient had a risk of suffering a fatal or non-fatal cardiovascular event or of dying from other causes. The possible cardiovascular events in the model were stable angina, unstable angina, stroke, myocardial infarction and transient ischaemic attack. A 10-year cardiovascular risk was calculated for each individual patient, with the distribution of coronary heart disease and stroke events dependent on age and sex. Patients who suffered a non-fatal cardiovascular event transitioned to a post-event cardiovascular health state and additional clinical events were not modelled. Once a cardiovascular event had occurred, mortality risk was adjusted accordingly. The impact of each intervention in terms of event reduction was applied as a relative risk, taking into account the mean differences in systolic BP observed in the HOME BP trial.

Owing to a cost difference of £402 and a QALY difference of 0.044, the results from inputting the HOME BP trial results into the long-term TASMIN4 cost-effectiveness model put the incremental cost-effectiveness ratio (ICER) at just over £9000 (Table 7). The probability of being cost-effective at different levels of willingness to pay was explored using a cost effectivness acceptability curve. The probability of being cost-effective was 66% at willingness to pay £20,000 per QALY, rising to 80% as willingness to pay increased to £50,000. Such results compare well with those assessed for use in the NHS by NICE, albeit with a sizeable degree of uncertainty.

TABLE 7 - Base-case results for the HOME BP intervention vs. usual care, base case, over patients’ lifetime

Strategy	Total cost (£)	Incremental cost (£)	QALY	Incremental QALYs	Incremental cost/QALY
Usual care	2685		11.562
HOME BP intervention	3087	402	11.606	0.044	9107

See Appendix 2 for the full cost-effectiveness analyses and see Report Supplementary Material 1 for a Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist.

The intervention group had significantly more changes to their antihypertensive medication than the usual-care group during the trial, both in terms of dose increases (relative risk 2.03, 95% CI 1.54 to 2.69) and new drugs added (relative risk 1.46, 95% CI 1.12 to 1.91); however, this had minimal impact on costs.

Further details of the secondary outcomes are reported in the main trial publication. 47

Process evaluation exploring how patients and health-care professionals experienced and implemented the intervention in practice

Health-care professionals mixed-methods process study: exploration of health-care professionals’ experiences of, and adherence to, using the intervention

Aim

The aim was to develop a detailed understanding of adherence levels, factors influencing adherence and barriers to and facilitators of implementing the intervention in primary care.

Methods

A mixed-methods approach was adopted for the HCP process evaluation. The HOME BP intervention was used by 125 HCPs across 70 general practices, and adherence data were collected either automatically by the online program or from the patients’ medical notes. The following measures of adherence were used:

• Percentage of automated recommendations to increase patients’ medication in response to home readings that were actioned.

• Percentage of patients for whom a three-step medication plan was created.

• Percentage of medication changes that were issued remotely (by letter or e-mail).

Health-care professionals also completed self-report questionnaires before and after completing compulsory online training at the start of the trial, capturing perceived self-efficacy, outcome expectancies and perceived intervention acceptability for patients. The data were analysed using a combination of correlations, Mann–Whitney U-tests and chi-squared tests.

Qualitative semistructured process interviews were conducted with 27 HCPs during the trial, including GPs, nurse prescribers, nurses and health-care assistants. To begin with, all prescribers and supporters were invited to an interview (17/25 accepted our invitation) and, subsequently, purposive sampling was used to target HCPs with more patients in the study and HCPs who were acting as both a prescriber and a supporter. The interviews were transcribed and analysed using thematic analysis. 41

The quantitative and qualitative data were analysed separately to maximise the strengths of each research method and were, subsequently, integrated using triangulation in which key findings from each analysis were compared to establish whether they were in agreement, partial agreement (i.e. complemented one another), disagreement or silence. 58

Findings

The sample for the quantitative analyses included 62 prescribers (i.e. GPs or nurse prescribers; 35% female), 58 supporters (i.e. nurses or health-care assistants; 95% female) and 5 prescriber–supporters who performed both roles (60% female). The subsample of HCPs who took part in qualitative interviews included 13 prescribers (38% female), 11 supporters (91% female) and 3 prescriber–supporters (100% female).

In terms of adherence to the target behaviour of escalating patients’ antihypertensive medication in response to average home readings being above target for 2 consecutive months, 405 e-mail recommendations were sent to HCPs, of which 215 (53%) were actioned. Comparisons of recommendations actioned against recommendations that were not showed that cases where systolic BP was closer to the threshold of 135/85 mmHg, and cases in which the patient had already been recommended a medication change previously, were less likely to be actioned. Meanwhile, patient age did not appear to make a difference.

Figure 3 shows the distribution of average systolic BP readings in cases where the recommendation for a medication change was not adhered to. Figure 3 shows that in 181 of 190 cases not adhered to, the patient had a mean BP reading below 150 mmHg (note that 150 mmHg is the target for the national Quality and Outcomes Framework in UK general practice59), although there were a few higher means that did not result in a change.

FIGURE 3.

Distribution of average systolic BP readings in cases where the recommendation for a medication change was not adhered to. Note that the cases in which systolic BP was below 135 mmHg triggered a medication change recommendation due to the diastolic mean exceeding the target, rather than the systolic. SD, standard deviation.

The qualitative data also indicated that borderline readings were a reason for not changing patients’ medication when recommended, and other reasons included concerns about the lack of context in which the readings had occurred (e.g. whether or not the patient had recently experienced a stressful event or illness) and preferring to recommend healthy behaviour change instead.

Adherence to planning three medication changes in advance for each patient was 82%, but the qualitative interviews highlighted several issues with implementing this procedure in practice. Some HCPs found it challenging to plan three medication changes for more complex patients, and there were also concerns about planning in advance when side effects or changes in health might mean that the medication change is no longer appropriate. A few HCPs described negative experiences of having to update the three-step plan, which could create additional work for them and cause anxiety for the patient.

Notifying patients of medication change remotely (i.e. by e-mail or letter) occurred in 38% of cases, whereas in all other cases the HCP spoke to the patient by telephone or face to face. Adherence to sending a monthly support e-mail to patients in the trial was 56%. Qualitative data indicated that HCPs had some concerns about contacting patients remotely, for example patients might not receive or value the information.

Practice implications

This mixed-methods evaluation suggested that there were practical issues with creating a three-step medication plan for some hypertensive patients and that this process might need more flexibility to improve implementation in practice. The processes of changing patients’ medication when their BP was above-target and supporting patients’ BP management via e-mail appeared straightforward to enact, but some HCPs were doubtful about the benefit of changing their working processes in this way. It may be that changes at the organisational level to BP targets and normalising e-mail support for patients might facilitate implementation of these processes.

Conclusions

Overall, the HOME BP intervention appeared to be both clinically effective and cost-effective, with significant reductions in BP compared with usual care, for a low cost per unit of BP. The reduction in BP found in this trial is important in terms of long-term health outcomes, with an anticipated reduction of 10–15% of patients suffering a stroke and of 5–10% of patients experiencing coronary-related events.

The detailed process evaluation of patients’ and HCPs’ experiences of implementing the intervention suggested some ideas for optimising the intervention, including:

• a guided discussion at baseline to increase patients’ and HCPs’ motivation to change medication when average BP exceeds the threshold, and to address some of the common concerns for patients about taking more medication

• additional support for patients with continuously raised BP readings to encourage patients to maintain engagement with self-monitoring

• acknowledgements for HCPs when patients have received information sent remotely to reassure HCPs that the information has been received and to increase the feasibility of remote support.

Objectives

In this section, we describe the intervention planning (WS1), development and testing (WS3) of the My Breathing Matters intervention, a digital self-management intervention for adults with asthma. The intervention objective was to improve functional quality of life of primary care patients with asthma, by supporting illness self-management by pharmacological means (e.g., medication adherence, appropriate health-care service use) and non-pharmacological means (e.g. breathing retraining, cognitive–behavioural therapy/stress reduction, healthy behaviour change). Development was carried out in four phases. The phases have not been reported elsewhere (apart from the development of the ‘guiding principles’61) and are, therefore, fully described in this section. The phases are described in chronological order and for each phase we report on the aims, methods, results and practical implications for how the findings informed intervention development. The four phases were as follows:

1. Collate and synthesise evidence to inform intervention planning, including a systematic review of quantitative evidence, a qualitative meta-ethnography and primary mixed-methods research.

2. Create an intervention plan, which involved developing guiding principles and carrying out behavioural analysis, to identify barriers to key behaviours and specify how these will be addressed.

3. Create an intervention prototype and use iterative qualitative interviews (i.e. think-aloud and retrospective interviews) to optimise the intervention.

4. Map the evidence onto behavioural barriers and intervention components onto theory.

At the end of this section, we demonstrate how the INDEX actions for developing complex interventions were met in this development process. 25

Intervention Development Team and patient and public involvement

The intervention development process was guided throughout by an Intervention Development Team that involved asthma-focused clinicians, psychologists, web developers, PPI contributors and representatives of Asthma UK, a key stakeholder organisation. Members of Asthma UK were invited on our Steering Committee and Intervention Development Team to provide their extensive knowledge of asthma and available self-management resources, and to help recruit PPI contributors. Asthma UK would also provide a suitable channel for national intervention dissemination once the acceptability and effectiveness of the intervention is established.

Our PPI contributors included two people with asthma (David Russell and Mark Stafford-Watson) and one public contributor with a general interest in DIs (Samantha Richards Hall). The PPI contributors provided feedback on study materials (e.g. interview topic guides, participant information sheets) and detailed feedback on the intervention prototype, and changes were made following their feedback. Two PPI contributors (David Russell and Samantha Richards Hall) provided their feedback on the key findings and final interpretations of the mixed-methods process evaluation and one PPI contributor (David Russell) was a co-author on the associated manuscript for this work, published in npj Primary Care Respiratory Medicine. 62

Phase 1: collate and synthesise evidence

Phase 2: creation of an intervention plan

Phase 3: creating and optimising the intervention

Phase 4: mapping the evidence onto behavioural barriers and the intervention components onto theory

Results

Appendix 4 shows the mapping in a behavioural analysis table (see Table 31, columns 3–6). When mapped onto the BCW, the My Breathing Matters intervention components were shown to target all six target constructs: (1) psychological capability, (2) physical capability, (3) reflective motivation, (4) automatic motivation, (5) physical opportunity and (6) social opportunity. The intervention components mapped onto six intervention functions (i.e. education, persuasion, training, modelling, enablement, and environmental restructuring) and 22 different BCTs. The intervention mapped onto all four core constructs of NPT (i.e. coherence, cognitive participation, collective action and reflexive monitoring).

For the logic model (Figure 4), three types of variables proposed to mediate the impact of the My Breathing Matters intervention on asthma-related quality of life were identified: (1) behavioural adherence, including effective engagement with DIs and improved pharmacological and non-pharmacological management (e.g. preventer medication adherence, engagement with a personal asthma action plan, attendance at annual asthma reviews, engagement with breathing retraining, engagement with cognitive–behavioural stress management practice and engagement with healthy behaviour change); (2) physiological mediators (e.g. improved asthma control, improved lung function and fewer exacerbations); and (3) psychological mediators (e.g. reduced stress, improved mood and improved enablement).

FIGURE 4.

Logic model of the My Breathing Matters intervention to improve quality of life in patients with asthma. A, Uptake and engagement facilitation; b, pharmacological support; and c, non-pharmacological support. BR, breathing retraining; PAAP, personal asthma action plan.

Mapping the My Breathing Matters intervention development process to the INDEX actions

As in the HOME BP intervention development (see Chapter 2), we retrospectively mapped the My Breathing Matters intervention development process to the 18 recommended actions for consideration during intervention development provided by O’Cathain et al. 25 (Table 10).

Conclusion

Similar to the HOME BP intervention, the combination of these approaches to intervention development helped ensure that the intervention was optimally persuasive, motivating and feasible to implement in practice for adults with asthma. A full description of the My Breathing Matters programme using the TIDieR checklist is provided in Appendix 5. 46 A demonstration of the My Breathing Matters intervention can be accessed via URL: www.mybreathingmatters.co.uk (accessed 29 July 2022).

Aims and objectives

In this section, we describe the evaluation of the My Breathing Matters intervention during a 12-month feasibility RCT. The aims, methods, results and implications are described for each discrete piece of research as follows:

• A feasibility RCT to assess feasibility of trial procedures and data analysis.

• A mixed-methods process evaluation to explore the acceptability of the My Breathing Matters intervention, including how patients experienced and used the intervention.

The section finishes with a conclusions section, which draws together all the findings.

Feasibility randomised controlled trial to assess feasibility of trial procedures and data analysis

Results

Table 11 provides the baseline demographic characteristics of the study population by group and Figure 5 shows participants’ flow through the trial. Follow-up data at 3 months was gathered from 91% of patients (intervention group, 36/44; control group, 44/44; total, 80/88) and at 12 months was gathered from 90% of participants (intervention group, 36/44; control group, 43/44; total, 79/88). At 12 months, four patients formally withdrew from the study, one patient was withdrawn as they were no longer eligible (i.e. they were referred to secondary care) and four patients did not complete their 12-month follow-up questionnaires. Nine adverse events and three serious adverse events were reported, which were unrelated to the study.

TABLE 11 - Baseline demographic characteristics of the My Breathing Matters study population per group

Percentages are reported from 42 participants, as two intervention participants did not provide these data.

Characteristic	Overall sample (N = 88)	Randomised group
		Intervention group (N = 44)	Control group (N = 44)
Age (years), mean (SD)	56.6 (15.2)	57.0 (14.2)	56.3 (16.2)
Female, n (%)	53.0 (60.2)	27.0 (61.4)	26.0 (59.1)
BMI (kg/m2), mean (SD)	29.5 (6.1)	28.9 (5.9)	30.1 (6.3)
Length of diagnosis (years), mean (SD)	24.0 (17.5)	25.2 (17.2)	22.8 (17.8)
Ethnicity, n (%)
White	84 (95.5)	42 (95.5)	42 (95.5)
Other	4 (4.5)	2 (4.5)	2 (4.5)
Smoking status, n (%)
Current	9 (10.2)	7 (15.9)	2 (4.5)
Former	29 (33.0)	13 (29.5)	16 (36.3)
Never	50 (56.8)	24 (54.5)	26 (59.1)
Age left education (years), n (%)	18.5 (5.3)	19.4 (7.0)a	17.7 (2.7)
≤ 16	40 (46.5)	18 (42.9)	22 (50.0)
17–18	22 (25.6)	9 (21.4)	13 (29.5)
> 18	24 (27.9)	15 (35.7)	9 (20.5)
Index of Multiple Deprivation, mean rank (median decile)	17,192 (5.5)	17,231 (6.5)	17,212 (5)

FIGURE 5.

A CONSORT (Consolidated Standards of Reporting Trials) flow diagram for the My Breathing Matters feasibility trial. This figure has been reproduced with permission from Ainsworth et al. 70 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/.

This figure has been reproduced with permission from Ainsworth et al. 70 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/.

The mean Mini AQLQ and ACQ scores (and SDs) and the percentage of patients achieving a minimal clinically important difference improvement in Mini AQLQ scores of ≥ 0.5 at each time point are presented in Table 12. Patients in the intervention group who completed the 12-month follow-up measures (n = 36) had mean improvement in asthma-related quality of life (i.e. Mini AQLQ score) of 0.35 (95% CI 0.10 to 0.60), compared with an improvement of 0.21 (95% CI –0.09 to 0.51) in the control group. The between-group difference (controlling for baseline differences) was 0.18 higher (95% CI –0.21 to 0.56) in the intervention group, indicating better quality of life. In the ACQ 12-month analysis, the between-group ACQ score was 0.14 lower (95% CI –0.40 to 0.11) in the intervention group, indicating better control. These findings are not significant, but indicate consistent trends to improvement in both asthma quality of life and asthma control in the intervention group when compared with the control group. There was no statistically significant difference in the number of patients who showed minimal clinically important difference improvement at 3 or 12 months across groups. There was no suggestion of an effect on physiological measures of lung function. The data quality check, and subsequent examination by research team clinicians, found that reviews completed by the practice nurses lacked the detail to quantify the amount of medication prescribed. Specifically, the reviews did not provide information on the number of inhalers issued on each prescription or specify the device (e.g. metered dose inhaler or dry powder inhaler) prescribed in enough detail. In any future trial, data collection plans would need to ensure that these data were collected.

TABLE 12 - Mean Mini AQLQ and ACQ scores, and percentage of patients achieving a minimal clinically important difference at baseline and at 3 and 12 months

Measure	Intervention group			Control group
	Baseline (n = 44)	3 months (n = 36)	12 months (n = 37)	Baseline (n = 44)	3 months (n = 44)	12 months (n = 43)
Mini AQLQ, mean (SD)	4.85 (0.94)	5.51 (0.85)	5.29 (0.98)	4.78 (1.09)	5.30 (1.07)	5.00 (1.25)
Mini AQLQ minimal clinically important difference improvement (%)		47.2	38.9		47.7	39.5
ACQ, mean (SD)	1.35 (0.66)	0.98 (0.65)	1.00 (0.59)	1.56 (0.91)	1.28 (0.87)	1.26 (0.69)

Mixed-methods process evaluation to explore the acceptability of the My Breathing Matters intervention

Conclusion

Overall, the findings demonstrated that a fully powered confirmatory RCT to assess the effectiveness of the My Breathing Matters intervention is feasible. The outcomes evaluation supported the need for a confirmatory RCT, with some optimisation of specific trial procedures for recording health utilisation data to improve the quality of the data collected for a health economics analysis. The My Breathing Matters intervention appeared to be acceptable to adults with asthma and there was good user engagement with the intervention. Future iterations of the intervention should include modifications to its information structure to facilitate easy, but structured, access to content important to users.

Objectives

Our objectives (see Chapter 1) were to develop and trial DIs to support patient self-management of two common contrasting health conditions, that is, hypertension and asthma (objectives 3 and 4, WS2 and WS3). Through the process of planning, developing and evaluating these interventions, we also aimed to generate a better understanding of what features and methods for implementing DIs could make them acceptable, feasible, effective and cost-effective to integrate into primary care (objectives 1 and 2, WS1).

Workstream 2 and WS3 fully achieved the aims of developing and trialling interventions that proved acceptable, feasible and (in the fully trialled intervention) also cost-effective. In-depth theory-, evidence- and person-based planning and development processes were undertaken for both interventions, drawing on extensive qualitative research, PPI input, review of the evidence and behavioural analysis to ensure that key behavioural barriers to engaging with the target behaviours were addressed. Feasibility studies were successfully carried out for both DI, as an internal pilot trial for the hypertension DI and as a standalone feasibility study for the asthma DI. Both studies suggested that fully powered RCTs were feasible, with only minor modifications to the intervention and the asthma health utilisation data collection methods.

A RCT recruited 622 patients in primary care to explore the clinical effectiveness and cost-effectiveness of the DI for hypertension. Eighty-nine per cent of patients completed the 12-month follow-up and the DI was found to be effective, with the intervention group having significantly lower BP at 12 months than the usual-care group. The intervention cost was £11 per unit reduction in systolic BP, which is less than what is reported previous similar studies. 50,51

Chapter structure

Implications of our findings for future digital health intervention research reflects on the generalisable insights that we obtained through this programme of research into how to develop engaging and useful DIs using the PBA, theory and evidence. Implications of our findings for integrating digital interventions for hypertension and asthma into primary care then considers the clinical implications of our research in terms of hypertension and asthma, and The contribution of patient and public involvement provides a reflection on the important contributions PPI made to this programme. Finally, Strengths and limitations describes some strengths and limitations of the research programme, and Summary and recommendations for future research summarises the recommendations for future research.

Implications of our findings for future digital health intervention research

Implications of our findings for integrating digital interventions for hypertension and asthma into primary care

The contribution of patient and public involvement

The aims of PPI involvement in the programme were to ensure that (1) patients’ needs were taken into account throughout the research, (2) the interventions were reassuring, motivating and enjoyable to use, (3) the research studies were accessible and feasible for participants to take part in and (4) the research was more likely to lead to sustainable change. Our PPI contributors were an integral part of all stages of the research cycle, including designing and undertaking the research, and interpreting and disseminating the findings. We worked closely with PPI contributors one to one and during group meetings throughout the programme, as we sought written input on research documents and interventions.

Patient and public involvement contributors provided significant and valuable input throughout the DIPSS research programme and we gained important generalisable learning about how to combine PPI with our theory-, evidence- and person-based approaches. Boxes 1 and 2 share insights into our PPI process. Box 1 gives specific examples of how our HOME BP PPI contributor Cathy Rice improved our research and Box 2 presents Cathy Rice’s own reflections on the process of contributing to the HOME BP intervention.

Strengths and limitations

Summary and recommendations for future research

In summary, the DIPSS research programme achieved the objectives of developing highly acceptable and feasible DIs for the contrasting conditions of asthma and hypertension. As intended, we showed that the intervention for hypertension was both clinically effective and cost-effective, and the intervention for asthma had good potential for a full effectiveness trial. Our research is already beginning to influence future clinical research and practice through further implementation. In addition, the theory-, evidence- and person-based approaches to intervention development that we refined through this research programme were shown to be successful in enabling us to identify and address important contextual barriers to and facilitators of engagement with the intervention by HCPs and patients. Therefore, we have documented, reported and are very actively disseminating these methods to the wider community of intervention developers in the public and private sector.

Contributions of authors

Lucy Yardley (https://orcid.org/0000-0002-3853-883X) (Professor, Health Psychology) was the chief investigator, and initiated, led and had overall responsibility for the programme.

Kate Morton (https://orcid.org/0000-0002-6674-0314) (Senior Research Assistant, Health Psychology) was lead researcher on the qualitative meta-ethnography, the qualitative and mixed-methods hypertension process analyses and on PPI engagement for the HOME BP intervention.

Kate Greenwell (https://orcid.org/0000-0002-3662-1488) (Senior Research Fellow, Health Psychology) was lead researcher on the asthma mixed-methods process evaluation.

Beth Stuart (https://orcid.org/0000-0001-5432-7437) (Associate Professor, Medical Statistics) led on the quantitative statistical analysis for both trials and was a member of the Programme Management Group.

Cathy Rice (https://orcid.org/0000-0001-5961-2413) (PPI contributor) provided PPI input throughout the research programme, and drafted and revised the relevant PPI sections of this report (see Plain English summary and Chapter 6).

Katherine Bradbury (https://orcid.org/0000-0001-5513-7571) (Senior Research Fellow, Health Psychology) co-led the hypertension intervention development and supervised the qualitative HOME BP process analyses.

Ben Ainsworth (https://orcid.org/0000-0002-5098-1092) (Lecturer, Health Psychology) was lead researcher on the development of the My Breathing Matters intervention and the asthma feasibility trial.

Rebecca Band (https://orcid.org/0000-0001-5403-1708) (Senior Research Fellow, Health Psychology) co-led the development of the HOME BP intervention and the HOME BP quantitative process evaluation.

Elizabeth Murray (https://orcid.org/0000-0002-8932-3695) (Professor, e-Health and Primary Care) led the hypertension systematic review and was a member of overall the Programme Management Group, contributing expertise in implementing DIs in primary care.

Frances Mair (https://orcid.org/0000-0001-9780-1135) (Professor, General Practice and Primary Care) led the asthma systematic review and was a member of overall the Programme Management Group, contributing expertise in implementing DIs in primary care.

Carl May (https://orcid.org/0000-0002-0451-2690) (Professor, Medical Sociology) was a member of the Programme Management Group, contributing expertise in NPT and implementation of DIs.

Susan Michie (https://orcid.org/0000-0003-0063-6378) (Professor, Health Psychology) was a member of the Programme Management Group, contributing expertise in BCTs.

Samantha Richards-Hall (https://orcid.org/0000-0002-7665-4268) (PPI contributor) was a member of the Programme Management Group, providing public contributor perspectives throughout the research programme.

Peter Smith (https://orcid.org/0000-0003-4423-5410) (Professor, Social Statistics) was a member of the Programme Management Group, providing senior expertise in statistical methods.

Anne Bruton (https://orcid.org/0000-0002-4550-2536) (Professor, Respiratory Rehabilitation) was a member of the Programme Management Group and co-led the asthma workstream, including the intervention development.

James Raftery (https://orcid.org/0000-0003-1094-8578) (Professor, HTA) was a member of the Programme Management Group and led the health economics analysis.

Shihua Zhu (https://orcid.org/0000-0002-1430-713X) (Senior Research Fellow, Health Economics) assisted with the health economics analysis.

Mike Thomas (https://orcid.org/0000-0001-5939-1155) (Professor, Primary Care Research) was a member of the Programme Management Group and led the asthma workstream.

Richard J McManus (https://orcid.org/0000-0003-3638-028X) (Professor, Primary Care) was a member of the Programme Management Group and led the hypertension workstream.

Paul Little (https://orcid.org/0000-0003-3664-1873) (Professor, Primary Care Research) was a member of the Programme Management Group, helped initiate and oversee the programme, and co-led the hypertension workstream.

All authors made substantial contributions to design, or acquisition of data, analysis and interpretation of data, all authors were involved in the drafting of the manuscript or revising it critically for important intellectual content, and all authors approved the final version of the manuscript to be published.

Funding

The HOME BP and My Breathing Matters interventions were developed using LifeGuide software, which was partly funded by the NIHR Southampton Biomedical Research Centre (BRC).

Publications

McLean G, Murray E, Band R, Saunderson K, Hanlon P, Little P, et al. Digital interventions to promote self-management in adults with hypertension: protocol for systematic review and meta-analysis. JMIR Res Protoc 2015;4:e133. https://doi.org/10.2196/resprot.4648

Band R, Morton K, Stuart B, Raftery J, Bradbury K, Yao GL, et al. Home and Online Management and Evaluation of Blood Pressure (HOME BP) digital intervention for self-management of uncontrolled, essential hypertension: a protocol for the randomised controlled HOME BP trial. BMJ Open 2016;6:e012684. https://doi.org/10.1136/bmjopen-2016-012684

McLean G, Band R, Saunderson K, Hanlon P, Murray E, Little P, et al. Digital interventions to promote self-management in adults with hypertension systematic review and meta-analysis. J Hypertens 2016;34:600–12. https://doi.org/10.1097/HJH.0000000000000859

McLean G, Murray E, Band R, Moffat KR, Hanlon P, Bruton A, et al. Interactive digital interventions to promote self-management in adults with asthma: systematic review and meta-analysis. BMC Pulm Med 2016;16:83. https://doi.org/10.1186/s12890-016-0248-7

Band R, Bradbury K, Morton K, May C, Michie S, Mair FS, et al. Intervention planning for a digital intervention for self-management of hypertension: a theory-, evidence- and person-based approach. Implement Sci 2017;12:25. https://doi.org/10.1186/s13012-017-0553-4

Bradbury K, Morton K, Band R, May C, McManus R, Little P, Yardley L. Understanding how primary care practitioners perceive an online intervention for the management of hypertension. BMC Med Inform Decis Mak 2017;17:5. https://doi.org/10.1186/s12911-016-0397-x

Morton K, Dennison L, May C, Murray E, Little P, McManus RJ, Yardley L. Using digital interventions for self-management of chronic physical health conditions: A meta-ethnography review of published studies. Patient Educ Couns 2017;100:616–35. https://doi.org/10.1016/j.pec.2016.10.019

Bradbury K, Morton K, Band R, van Woezik A, Grist R, McManus RJ, et al. Using the Person-Based Approach to optimise a digital intervention for the management of hypertension. PLOS ONE 2018;13:e0196868. https://doi.org/10.1371/journal.pone.0196868

Morton K, Dennison L, Bradbury K, Band RJ, May C, Raftery J, et al. Qualitative process study to explore the perceived burdens and benefits of a digital intervention for self-managing high blood pressure in Primary Care in the UK. BMJ Open 2018;8:e020843. https://doi.org/10.1136/bmjopen-2017-020843

Ainsworth B, Greenwell K, Stuart B, Raftery J, Mair F, Bruton A, et al. Feasibility trial of a digital self-management intervention ‘My Breathing Matters’ to improve asthma-related quality of life for UK primary care patients with asthma. BMJ Open 2019;9:e032465. https://doi.org/10.1136/bmjopen-2019-032465

Greenwell K, Ainsworth B, Bruton A, Murray E, Russell D, Thomas M, Yardley L. Mixed methods process evaluation of my breathing matters, a digital intervention to support self-management of asthma. NPJ Prim Care Respir Med 2021;31:35. https://doi.org/10.1038/s41533-021-00248-6

McManus RJ, Little P, Stuart B, Morton K, Raftery J, Kelly J, et al. Home and Online Management and Evaluation of Blood Pressure (HOME BP) using a digital intervention in poorly controlled hypertension: randomised controlled trial. BMJ 2021;372:m4858. https://doi.org/10.1136/bmj.m4858

Morton K, Dennison L, Band R, Stuart B, Wilde L, Cheetham-Blake T, et al. Implementing a digital intervention for managing uncontrolled hypertension in Primary Care: a mixed methods process evaluation. Implement Sci 2021;16:57. https://doi.org/10.1186/s13012-021-01123-1

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, CCF, PGfAR or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the PGfAR programme or the Department of Health and Social Care.

Brief name

The HOME BP intervention.

Why?

High BP or hypertension is a common condition in the UK, affecting approximately one in four adults, with many patients taking medication for their high BP remaining above national targets. 91,92 Raised BP is a key risk factor for cardiovascular events, including heart attack and stroke. 93 Clinical inertia is a recognised contributor to uncontrolled BP,8 and this occurs when a HCP chooses not to increase antihypertensive medication during an appointment, despite the patient’s BP reading being above target. Reasons for clinical inertia can include uncertainty about the accuracy of readings taken in clinic, concerns about side effects and a patient’s reluctance to take more medication. Previous evidence15,33 has shown that non-DIs can reduce BP by encouraging patients and GPs to make medication changes in accordance with a pre-agreed plan, based on an algorithm for home BP readings. The HOME BP intervention aimed to provide a cost-effective feasible digital solution for managing uncontrolled BP by translating these effective procedures into an online intervention.

The target behaviours to reduce BP included self-monitoring BP at home, changing medication when BP was above-target and optional healthy behaviour changes. Social cognitive theory39 and NPT35 were used to explain how the intervention would change the target behaviours.

What materials?

The HOME BP intervention comprised three online training sessions for patients, monthly online BP entry and feedback pages, two online training sessions for HCPs and automated e-mails that acted as reminders and prompts for action for patients and HCPs.

Each participant received an OMRON M3 BP monitor (Milton Keynes, UK) to use for self-monitoring.

The online materials were as follows.

What: procedures

The intervention procedures are shown in Figure 6.

FIGURE 6.

HOME BP intervention procedures.

Figure 6 shows that patients completed the first training session online and then had an appointment with their prescriber to agree three potential medication changes. The prescriber recorded the three planned medication changes in a template, which was saved to the patient’s notes. At this time, the patient also collected their BP monitor at their general practice. At home, the patient could then login to the HOME BP intervention and complete session 2, which trained patients to take two morning readings for 7 days, record these on paper and then enter the second reading from each day on to the HOME BP intervention.

Following completion of session 2, patients were prompted to take 1 week of practise home readings and enter these on the HOME BP intervention. The intervention offered the opportunity for patients to send their nurse a message via the HOME BP intervention about their practise readings if they wanted to. Patients also received an e-mail that reminded them that they could make an appointment to see their nurse if they wanted to talk about how to use the BP monitor. These steps were taken to help promote patients’ self-efficacy in home readings and to ensure that patients felt confident to take their own BP going forwards.

After entering their week of practise readings, patients received reminders to monitor their BP every 4 weeks during the 12 months, except after 3 consecutive months of well-controlled readings at which point the reminders dropped to once every 8 weeks. Patients received automated feedback on their readings as soon as the readings were entered, and the readings were also shared with the prescriber and supporter. The algorithm calculated an average, and a medication change was recommended if the average was above-target for 2 consecutive months. Prescribers were trained in the procedure for changing medication during their online training, and the steps were reinforced in the e-mail they received at the time a change was needed. The steps involved checking that the planned change was still appropriate for the patient and issuing the new prescription along with a template letter that included instructions for the patient on how to make the change and any blood tests that might be needed.

Throughout the 12-month intervention, supporters were asked to send monthly e-mails to patients using predesigned templates, which could be edited to personalise the e-mail. This monthly remote support was designed to help patients feel supported when self-monitoring their BP at home and to reinforce the benefits of adhering to self-monitoring and medication change.

Nine weeks from when the patient was randomised, an automated e-mail was sent to alert patients that the optional healthy behaviour change session was now available, and there was also an option to view session 3 from their homepage when they logged into the HOME BP intervention. If patients wanted, they could book an appointment with the supporter at this point to discuss their choice of healthy behaviour change.

Who provided?

The intervention was delivered by a prescriber (i.e. a GP or nurse prescriber) and a supporter (i.e. a practice nurse or a health-care assistant) at each general practice. Each prescriber and supporter was required to complete an online training session prior to recruiting any patients to the intervention, the content of which is described in Chapter 3. General practices were reimbursed for taking part in the research study.

How?

The majority of the intervention was delivered online individually. Some components of the intervention involved face-to-face or telephone appointments with a HCP, which could be initiated by the patient or the HCP.

Where?

The intervention was implemented in a primary care context in the south of England from 2015 to 2018. Online components were completed by patients at home or by HCPs at their general practice. Patients needed to have internet access to be able to take part. Any general practice was eligible to sign up to the study.

When and how much?

The intervention was designed to last 12 months. Patients were asked to take their BP readings for 7 days, twice every morning. Patients would then have 3 weeks off before starting again. Patients record their 7 days of readings online at the end of the week. If their BP average was well controlled for 3consecutive months, then their self-monitoring frequency dropped to once every 8 weeks.

It was anticipated that each online training session for patients would take approximately 30 minutes to complete.

Tailoring

Blood pressure targets were tailored to take account of age and diabetes as follows:

1. Patients aged < 80 years without diabetes: 135/85 mmHg.

2. Patients with diabetes: 135/75 mmHg.

3. Patients aged ≥ 80 years without diabetes: 145/85 mmHg.

The optional healthy behaviour session (i.e. session 3) was tailored to only offer the fifth option of weight loss if the patient’s BMI was > 25 kg/m².

Modifications

The intervention procedures and materials were only minimally modified during the course of the study. Small changes were made in line with feedback from process interviews during the internal pilot trial. The changes included the addition of a stress-related quiz question in session 1 because of the prevalent perception among participants that stress was the main cause of their high BP, additional e-mails about the benefits of healthy behaviour changes to increase uptake, and the option to launch a healthy behaviour change module from the intervention homepage, rather than needing to complete the optional healthy behaviour change session.

How well: planned?

Online usage data were captured automatically by the intervention software LifeGuide:

1. The number of patients completing the core training (i.e. sessions 1 and 2).

2. The number of patients completing session 3 (optional).

3. The number of patients completing 7 days of practise readings.

4. The number of BP entries made by patients.

5. The number of prescribers completing the core training.

6. The number of supporters completing the core training.

7. The number of recommendations made to change patients’ medication.

The software enabled patient and HCP engagement to be assessed. In addition, reviews were conducted of the patients’ medical notes at the end of the study to explore prescribers’ adherence to planning three medication changes in advance and implementing recommended medication changes.

Qualitative process evaluations were undertaken to explore patients’ and HCPs’ experiences of implementing the intervention.

How well: actual?

Two detailed mixed-methods process evaluation studies were undertaken to explore patients’ and HCPs’ adherence and experiences of implementing the HOME BP intervention. The findings are described in detail in Chapter 3.

Introduction

The HOME BP trial aimed to assess the cost-effectiveness of the HOME BP DI relative to usual care. Within-trial results are reported in a cost-effectiveness analysis in terms of cost per unit of BP reduction. As the intended effects are reduced risk of stroke, heart failure and unstable angina, these longer-term effects were estimated in a model that was previously developed for that purpose. 1 The analysis is expressed in terms of incremental cost per QALY gained. The perspective in both short- and long-term analyses is that of the NHS, but the role of non-NHS costs on patients was explored in the within-trial analysis. Costing was based on unit costs in 2018/19. Bootstrapping was used to estimate mean values for costs and QALYs.

The main results of the trial are reported fully elsewhere,47 as are details of the longer-term model. 50 Although key elements are summarised here, please see these publications47,50 for fuller details.

The next section reports on the within-trial analysis of the HOME BP trial. A later section reports on the long-term economic modelling, along with exploration of various scenarios, before making comparisons with other similar analyses and drawing some general conclusions.

Costing

Costing was based on various resource use headings (Table 13). Data were collected from a review of case notes at the end of the trial. The cost estimates derived were for each patient and were included as a cost variable in the statistical analysis. Only service use data relating to BP were included. These data covered cardiovascular disease and possible side effects of antihypertensive medications, including dizziness and falls. Clinical advice was used to adjudicate on the inclusion/exclusion decisions. The number of patients using each service is shown along with costs of BP-related services (see Table 13).

As the trial ran between 2015 and 2018, the year for costing was taken as 2018. Discounting of costs was not considered necessary as the follow-up period was limited to 12 months.

Few patients recorded use of A&E, outpatient or inpatient departments in relation to BP. Inpatient admissions received particular attention because of their relatively high cost. Five inpatient admissions that were potentially related to BP were recorded. A review, carried out blind to which arm patients were in, showed that the dates on which all these admissions occurred were all before any changes in medication by patients in the trial. These admissions were, therefore, taken to reflect decisions made, or conditions in place, before the start of the trial. For that reason, these admissions were excluded from the base case, but were included in a sensitivity analysis.

The same logic of excluding service use that occurred before any medication changes in the trial was applied to outpatient and A&E visits. Therefore, the base or preferred case costing included only inpatient, outpatient and A&E use that occurred after any medication changes in the trial. Again, this decision was made and applied before the data were unblinded. The sensitivity analysis included all BP-related use of these services, regardless of their timing in relation to medication changes.

Costing drugs

Data were collected on all BP-related drugs that patients were taking at baseline, including name, dose and duration, along with subsequent consultations and changes in medication. BP-related drugs were defined as antihypertensives, broadly defined (see list available on request). Clinical advice was sought in unclear instances. We assumed that patients who did not have a change of medication during the trial continued their baseline medication unchanged for the duration of the trial. When data were missing with regard to duration of medications for patients who had changes in their medications, the durations were assumed to continue unchanged to the next change of medication or to the end of the trial, whichever was reached first. These assumptions involved a minority of patients (around 100–150 patients), most of whom had no change in their drugs.

The NHS Drugs Tariff June 201894 was copied for each BP drug mentioned in the trial. Each patient’s use of drugs was costed, from baseline and through changes during the trial, at the dose-specific prices listed in the NHS Drugs Tariff June 2018. 94 When branded rather than generic drugs were recorded, the generic rather than the brand price was used. This reflected common NHS prescribing practice, and also avoided bias due to the occasional prescription of brand rather than generic drugs. As very few brand names were listed, this assumption made little difference. The few drugs that did not have a generic version were costed using the stated price for the brand by dose.

If a drug was not included in the NHS Drugs Tariff June 2018,94 then the British National Formulary (BNF)95 was used; however, this applied in only a small number of cases.

When the dose recorded in the trial was not listed in the NHS Drugs Tariff June 2018or the BNF,94,95 then it was assumed that the dose could be made up by several dosages. This meant patients might have to use two pills rather than one, but in a few cases it involved up to four drugs (e.g. 4 × 25 mg of atenolol for 100 mg of atenolol).

Costing primary care contacts

The data recorded were in response to the following two questions. First, ‘Who did the patient speak to’, which was followed by three options: (1) GP, (2) practice nurse or (3) health-care assistant. Second, ‘How’, which was followed by four options: (1) face to face, (2) telephone, (3) letter or (4) e-mail. This led to 12 options (GP-face to face, GP-telephone, practice nurse-face to face, practice nurse-telephone, health-care assistant-face to face, health-care assistant-telephone, etc.). A unit cost (Table 14) was estimated for each option and was applied to all BP-related contacts.

Personal Social Services Research Unit (PSSRU) unit costs for 2017/18 (p. 12796) put the cost per hour of GP time from £181 to £243 (depending on in/exclusion of direct care costs and with/out qualification costs). A cost per consultation is provided for only GPs and not for other staff and this varied from £28.30 to £37.40 for an average 9.22-minute consultation.

The costing in Table 14 used the cost per GP consultation of £34.30, that is, including qualification but not direct costs. The higher figure that included direct costs was not considered appropriate as it appeared to include the overhead costs, which are also included in the estimate of the cost of the practice nurse.

The cost per hour of a practice nurse in the PSSRU data96 ranged from £36 to £42 per hour, depending on qualifications. We used the higher nurse cost per hour, including qualification (as with the GP estimate). Assuming the same duration of consultation as for a GP, this put the cost per face-to-face nurse consultation at £6.50.

As the term ‘health-care assistant’ is not listed by the PSSRU,96 this was costed as a band 4 nurse (i.e. the lowest grade for which the PSSRU provided a unit cost).

The PSSRU provide an estimate of £8.10 for a GP telephone call, based on a small study to do with triage in general practice, and this seems a reasonable proportion, as £8.10 of £34.3 is approximately 25%. The same proportion was applied to the cost per face-to-face consultation for practice nurses and HCAs.

A small number of letter and e-mail contacts were recorded. These were costed at £1.67 and £1.00 each, respectively, as a fixed cost, regardless of who they were from. A small number of non-BP-related primary care contacts were recorded and these were reviewed and included/excluded prior to unblinding by two clinicians (RMcM and PL).

All patients in the trial had an initial GP consultation to do with entering the trial. However, this consultation was not included in the costing, as it was considered a research cost and applied equally to both arms of the trial. If the intervention were to become routine practice, then the assumption implied here is that discussion of patients’ use of the intervention would occur in a routine consultation.

Costing inpatient episodes

The data collected showed all BP-related inpatient admissions by specialty and reason (Table 15). Checking the dates of these admissions showed that all inpatient admissions occurred before any changes of medication occurred in the trial. A decision was made pre-unblinding that these admissions should be omitted from the base-case cost, but are included as part of a sensitivity analysis (see Sensitivity analysis).

The two angioplasty admissions were allocated to an angioplasty Healthcare Resource Group (HRG) using the National Tariff, which puts the cost in 2017/18 at £2404 for the most common HRG (EY41D Standard Percutaneous Transluminal Coronary Angioplasty with CC score 0–3).

Similarly with pacemaker insertion, the cost of the most common HRG was put at £2814 (HRG EY06E Dual Chamber Pacemaker Insertion with CC score 0–2). 97

The two admissions linked to falls were classified under HRG WHO9G (Tendency to Fall, Senility or Other Conditions Affecting Cognitive Functions, without Interventions with CC score 0–1). One admission was classed as routine with a cost of £1844 and the other was classed as a short stay with a cost of £533.

Costing accident and emergency attendances

Data were collected on all attendances at A&E by reason. Attendances deemed most likely to be BP related are listed in Table 16. The most common reason was chest pain. As discussed above, A&E attendances occurring before any change in medication were excluded from the base-case costing following the logic outlined above for inpatients. The cost per A&E attendance was put at £160 in 2017/18 by NHS Improvement. 97

Costing outpatient attendances

Data were collected on all outpatient attendances. Outpatient attendances most likely relating to BP are shown in Table 17.

Outpatient visits occurring before any change in medication for trial patients were excluded, following the logic outlined in Costing inpatient episodes. Excluding outpatient visits in this way led to only a small number of outpatient visits being included. The cost per outpatient attendance was put at £125 in 2017/18 by NHS Improvement. 97 More detailed unit costs are available for A&E, outpatients and inpatients. The higher-level averages have been used as a first cut, but see more detailed unit costs below for inpatients. Data are not available to disaggregate A&E attendances, but may be applicable to outpatient attendances depending on inclusion/exclusion criteria.

Sensitivity analysis

As discussed above, the base-case scenario was costed on the basis of BP-related service use in primary care, including outpatients and A&E attendances that occurred after a medication change in the trial. This latter criterion excluded the five inpatient admissions that might have been BP related.

An alternative scenario was costed, based on including those inpatient, outpatient and A&E visits plausibly related to BP.

A third scenario explored costing only primary care costs, that is, excluding the relatively small numbers of outpatient and A&E attendances that occurred after a medication change in the trial, and this gave results almost identical to the base-case scenario, reflecting the paucity of use of these services.

Non-NHS costs

Although data were collected on the amount of time patients spent using the website, we decided not to cost it because of the finding from the process evaluation that showed that patients valued the DI in terms of perceived benefits (e.g. reassurance and improved health), but also experienced burdens (e.g. worry about health). Time did not appear to be an important feature. Furthermore, patients’ perceptions of illness and treatment perceptions about hypertension appeared to influence perception of benefit or burden. Valuation of such issues would require going well beyond estimates of time spent online.

We offer suggestions under Recommendations for research on how these matters might be developed.

On diet and lifestyle, very few patients recorded any changes, and this finding was consistent with the overall finding that the key changes arising from the intervention were to do with changes in prescribed drugs. Therefore, no further costing was deemed necessary.

The result was that we are able to present only cost-effectiveness estimates from an NHS perspective. We note that this is the perspective required by NICE.

Costing the HOME BP intervention

Blood pressure monitors were provided by OMRON at a lower cost than those sold commercially. The cost to the trial was £23, whereas the commercially available units cost around £65 (Boots UK Limited, Nottingham, UK, 2018 price). As these monitors last for several years, usually taken as 4 or 5 years, an estimate of the cost for 1 year, that is, the same as for other cost headings, is required. This costing can be calculated in two ways: straight-line depreciation (offset the same amount each year, e.g. 25% for each year if over 4 years) and the annuity method. The £23 cost incurred in the study is shown in annual terms using both methods and in time frames of 4 and 5 years in Table 18. The methods make fairly little difference, ranging from £4.60 to £6.26.

The other element of the intervention had to do with the website. Following advice from the principal investigator, the cost of the website was based on the cost of the employee who programed and maintained LifeGuide. This cost was estimated based on the programer’s salary and the proportion of time spent on maintaining server, spread over then current 10 projects:

• 5% of £48,677 (level 5/spine point 43) for 3 years (HOME BP was live from 2015 to 2018) = £7300.

• Divided by the number of participants in the intervention arm (n = 305) = £23.90.

This gives the figure of £23.90 (see Table 18). As this figure does not appear to include overhead costs (rent, utilities, etc.), a 40% overhead has been added (see Table 18).

Depending on the overhead issue, as well as writing off the monitor over 4 years, puts the cost of the intervention at £30.16 or £39.72 (see Table 18). Slightly different results due to different assumptions are also shown in Table 18.

The higher cost of £39.72 was chosen for the base case on the grounds that using the higher cost meant that estimates of cost-effectiveness would err on the high rather than the low side.

Quality-adjusted life-years

The data from EuroQol-5 Dimensions (EQ-5D), which were collected at baseline and at 6 and 12 months, were used to estimate QALYs. Although the health gain from reduced BP has to do with long-term effects of reduced risk of cardiovascular events and death, these data help indicate whether or not any short-term changes might result from the intervention. The more relevant long-term cost-effectiveness of the intervention was estimated in long-term modelling, which is in Long-term cost-effectiveness. For completeness, the within-trial results for incremental cost per QALY are reported here.

Data collected using EQ-5D were used to estimate QALYs via preferences based on a survey of the public. The results indicated a very small QALY, statistically non-significant, loss in the intervention group relative to the control group (Table 19).

Some EQ-5D scores were missing. Full values were available for 89% of patients, with no difference by arm. 47 The principal analysis of the primary outcome used raw and adjusted data, and was agreed in a statistical analysis plan before final data lock. 47 The primary analysis used general linear modelling to compare systolic BP in the intervention and usual-care groups at follow-up, adjusting for baseline BP, practice (as a random effect to take into account clustering), BP target levels and sex. Analyses were on an intention-to-treat basis and used 100 multiple imputations by chained equations for missing data. The imputation model included all outcome and stratification variables.

The QALY values were then imputed based on BP and these values were used in both the within-trial analysis and in the longer-term modelling.

Results

This within-trial cost-effectiveness analysis was conducted from an NHS perspective using data collected on use of services and on the intervention. The reduction in BP of 3.45 mmHg, combined with increased cost in the intervention arm of £38, led to an incremental cost per unit of BP reduction in the base case of £11 (95% CI £5 to £29) (see Table 5). The increased cost per patient of £38 in the intervention arm was due almost entirely to that of the intervention (£39.73).

Complete-case and imputed analyses (see Table 5) gave almost identical results, with only very small differences. The uncertainty around this estimate was simulated probabilistically using 10,000 runs (Figure 7).

FIGURE 7.

Scatterplot of joint distribution of incremental mean cost from NHS perspective and mean BP reduction from baseline (mmHg) over 12 months.

The base case included use of NHS services relating to BP. This included the full range of NHS services, including hospital admissions. Although few such admissions were recorded, some were elective procedures that had to have been planned before entry to the trial. Consequently, only those hospital admissions that occurred after a change of medication were included in the base-case costing. To test the sensitivity of results to this assumption, a scenario was costed that included all hospital BP-related service use, regardless of timing. This scenario made little difference overall, but reduced both the cost difference and the incremental cost-effectiveness.

A small, non statistically significant decrement in QALYs was observed in the intervention arm (see Table 5), which, when combined with its higher cost, meant that the intervention arm was dominated by the usual-care arm. However, as QALY differences to do with improved BP control at 12 months are of little interest compared with those in the longer term, the results of the lifelong modelling provide a more robust and plausible estimate of cost effectiveness. The mean cost per patient in primary care was similar to the base-case cost, indicating that primary care accounted for almost all the costs (see Table 6). Within primary care, costs were split roughly 60 : 40 between costs attributable to consultations and costs for prescriptions. Patients in the intervention arm had slightly higher prescription costs, associated with changes in medication and/or dose. However, these costs did not increase the cost of primary care consultations because of the role of the DI. These trends were as might be expected. Further analysis of these changes is planned for a separate publication, which will include changes in the time spent by patients in managing their hypertension.

The resulting cost effectivness acceptability curve (Figure 8) indicatated that the probabilities of being cost-effective for the intervention against the usual care are 51%, 90% and 98% at thresholds of £10, £20 and £50 per unit of BP, respectively.

FIGURE 8.

Cost-effectiveness acceptability curve of the intervention and usual-care groups based on BP from baseline over 12 months.

Long-term cost-effectiveness

A published long-term cost-effectiveness study1 of self-management of BP with and without telephone support, broadly similar to the HOME BP intervention, published in 2019, provided a relevant long-term model. Use of this model1 was facilitated by the overlap of investigators (RMcM) in these two trials.

The model, hereafter referred to as the TASMINH4 model, was a Markov patient-level simulation undertaken in TreeAge to model the different strategies. This type of Markov model tracks the costs and consequences of individual patients passing through the model, with characteristics (taken from the trial) free to vary between patients. The model was run over the maximum lifetime of the patients (maximum of 65 years; minimum trial inclusion criteria was age 35 years), a time horizon sufficient to capture all relevant long-term costs and consequences.

Each patient had characteristics created by randomly sampling the trial patient-level data by means of a uniform distribution. These characteristics affected their probability of subsequent model events. For instance, males had a higher risk of cardiovascular disease than females. The model was run with a large number of simulated patients (n = 50,000) to account for interpatient variability and to adequately model a representative clinical population.

Model structure

All patients started in the well/no event health state. Within a 6-month time cycle, a patient had a risk of suffering a fatal or non-fatal cardiovascular event or dying from other causes. The possible cardiovascular events in the model were stable angina, unstable angina, stroke, myocardial infarction and transient ischaemic attack. Ten-year cardiovascular risk was calculated for each individual patient, with the distribution of coronary heart disease and stroke events dependent on age and sex. Patients who suffered a non-fatal cardiovascular event transitioned to a post-event cardiovascular health state and additional clinical events were not modelled. Once a cardiovascular event had occurred, mortality risk was adjusted accordingly. The impact of each intervention in terms of event reduction was applied as a relative risk, taking into account the mean differences in systolic BP observed in the HOME BP trial.

Results (long term)

Owing to a cost difference of £402 and a QALY difference of 0.044, the results from inputting the HOME BP trial results into the long-term TASMIN4 cost-effectiveness model put the ICER at just over £9000 (Table 20). The key inputs from the HOME BP trial were a 3.45 mmHg difference in BP and a cost difference of £38. The small QALY decrement in the intervention arm in the trial was offset in the longer-term modelling by reductions in cardiovascular events and deaths (see Table 20).

The range of the increments in the different runs of the model are shown in a scattergram form in Figure 9. The cost-effectiveness acceptability curves (Figure 10) show a 66% probability of the intervention being cost-effective at £20,000 per QALY, rising to 72% at £30,000 (note that these relatively low probabilities imply considerable noisiness or wide CIs).

FIGURE 9.

Scattergram of repeated runs of long-term model.

FIGURE 10.

Cost-effectiveness acceptability curve of long-term cost per QALY.

Scenario analyses

Different scenarios explored the implications of varying the inputs to the model, including the cost difference and the timescale (Table 21), as well as the number of events averted (Table 22).

As expected, when the cost difference in the base case (£38/year) was increased, so too did the QALY ICER. When the difference was increased to £100, the ICER rose to £26,432. A cost difference of £77 led to an ICER of just under £20,000, taken by some as a willingness-to-pay threshold.

In addition, when the time frame was reduced from lifetime in the base case, the ICER increased (see Table 23) from just over £9000 to just under £67,000. This implies that the health gains occur mainly after 5 years.

The long-term cost-effectiveness of the intervention depends on the number of cardiovascular events averted. Table 22 shows the number of events predicted in the model for each arm along with the proportions.

A total of 7.4% of individuals in the usual-care strategy had a non-fatal stroke, compared with 6.9% of individuals in the HOME BP arm. The majority of people (approximately 71–72%) did not have an event and died from other causes. Differences in outcomes were from the 28% of individuals who had an event.

The cost per major event avoided was about £36,000 (£402/0.0112).

Discounting made a difference, as might be expected. When costs were discounted 3.5%, then the base-case QALY ICER was £4508 per QALY and the difference in QALYs is 0.0891. When neither costs nor QALYs were discounted, then the ICER was £6573 per QALY.

Comparisons with other similar studies

Two studies were considered to be relevant: (1) TASMINH450 and (2) HITS a Scottish randomised trial of telemonitoring to control BP. 98,99 The TASMINH450 and HITS98,99 studies are briefly summarised in Table 23.

The reported comparisons from TASMINH450 were self-management compared with usual care and telemonitoring compared with self-management, and these are summarised along with HOME BP and HITS in Table 23. Several points are noted.

First, the reductions in BP are broadly similar across the three trials, ranging from 3.5 to 4.7 mmHg. The differences are more apparent in terms of their costs, which varied by a factor of £3 between TASMINH450 and HOME BP, with the HITS trial98,99 having a much higher cost.

Second, the relevant comparison between TASMINH450 and HOME BP would be that between telephone and usual care; however, this was not presented in the published analysis, which followed the standard practice of reporting interventions in terms of next best. However, the relevant ICER for that comparison was just over £3000, which is well below that for HOME BP, and this prompts the question of how the relative cost compared.

The costs in Table 23 refer to the differences between TASMINH450 and HOME BP in total cost after 12 months, including both the cost of the intervention and of services used. Neither TASMINH450 nor HOME BP provide precise estimates of the intervention for several reasons. Costs of interventions in trials reflect those incurred during the trial, which might well be different if those interventions were used in routine practice. The emphasis in trials is on delivering a new untried intervention, which may incur costs that would not otherwise be incurred. Furthermore, if used in routine practice, most interventions, and particularly interventions that are digital, would benefit from economies of scale. With the interventions under discussion, separation of the intervention and service use cost was difficult, with some elements of the intervention, such as follow-up reminders or telephone calls, being recorded as service use, rather than part of the intervention. For these reasons, we do not consider that the cost differences between the various interventions shown in Table 23 for TASMINH450 and HOME BP should be treated as precise. Rather, the cost differences provide an indication of the likely low cost of the interventions.

Turning to the HITS trial,98,99 two points are worth making. The HITS trial98,99 showed a similar reduction in BP as TASMINH450 and HOME BP, but a higher per patient cost (£109 vs. £38 for HOME BP) (see Table 23). This higher cost may reflect the HITS trial’s98,99 reliance on telemonitoring, which may have been more expensive at that time. Although we have not analysed the cost difference in detail, the long-term scenario analyses reported above included one scenario with a cost difference of £100, close to that in the HITS trial. 98,99 The scenario resulted in an incremental cost per QALY of £26,432. Although well above the incremental cost per QALY in the two other trials (i.e. TASMINH450 and HOME BP), this might still be considered as worthwhile value for money, particularly if the cost of providing the intervention might have declined since then.

Limitations

The usual limitations to do with randomised trials apply in that every trial is a specific one-off experiment. Against that, two trials50,98,99 have come to similar conclusions. All three trials (i.e. TASMINH4,50 the HITS trial98,99 and HOME BP) were fairly big and recruited from general practice in different locations. The overall conclusion must be that self-monitoring, supported to some extent by telemonitoring or a website, can lead to improvements in BP. Although the improvements are modest, as might be expected given the populations (i.e. most patients were already on medication for BP), they seem to be clinically worthwhile and reasonably cost-effective.

All three trials (i.e. TASMINH4,50 the HITS trial98,99 and HOME BP) shared a limitation to do with the costing of the intervention and its knock-on cost impact in the short run. Although this limitation is, to some extent, inevitable, the results indicated that such interventions can be provided at a modest cost per patient, which would be very likely to show economies of scale and reduced cost per patient if made widely available.

Recommendations for research

First, more comprehensive modelling of the long-term effects of BP reduction would appear to be useful, perhaps supported by an individual-level meta-analysis of the relevant trials. The informal review of the most relevant trials50,98,99 showed that the interventions trialled resulted in QALY gains, which seemed to be mainly in the longer term, a finding consistent with cost-effectiveness modelling carried out for NICE. Modelling might also be improved by more detailed comparisons between the relevant models and by monitoring the extent to which the projected reductions in cardiovascular events are confirmed in practice.

Second, more attention to the costing of the range of relevant interventions would be helpful. As discussed above, randomised trials providing a novel intervention designed only for that trial are not an ideal way to study what such interventions might cost in routine care, particularly if provided to large numbers of people.

Third, as noted above, although we aimed to include costs incurred beyond the NHS, we were unable to do so. Very few patients reported cost effects due to changes in lifestyle, which is consistent with the overall finding that the main impact of the intervention was on use of the appropriate medications. Although we collected data on the amount of time patients ‘spent’ using the web support, the process evaluation found that patients using he website experienced benefits, as well as costs. Research might usefully explore improved ways of measuring these contrasting elements.

Finally, research is needed on how digital aids for treating hypertension can best be located within the range of other self-management aids in other diseases and conditions. Following the COVID-19 pandemic, a shift to remote consultation seems likely to continue, as does increased provision of digital supports.

Conclusions

The reduction in BP in the HOME BP intervention was similar to that in other comparable trials. 50,98,99 The cost of the intervention was modest at just under £40 per patient. Although this is probably an overestimate, given that it was based on providing a novel service for relatively few people, it, nonetheless, delivered benefits that would be considered cost-effective in terms of NICE and the NHS. Long-term modelling puts the incremental cost per QALY at just over £9000. If included in a suite of DIs, which seems increasingly possible, the cost per patient would probably reduce.

More generally, post-COVID-19 and in line with demographic trends, self-management seems likely to become more widely used in modernised health services. The work reported here provides evidence of both its clinical effectiveness and cost-effectiveness.

Contributions

James Raftery designed, costed and wrote this appendix, with input from Richard J McManus. Shihua Zhu provided input on the within-trial analysis and Sue Jowett and Richard J McManus provided input on the long-term modelling.

Predictors of systolic blood pressure at 12 months

The number of BP entries a patient made, the number of medication changes recommended and medication necessity beliefs to predict systolic BP at 12 months are provided in Tables 24–26.

Predictors of number of blood pressure entries

Patients’ self-reported medication adherence (MARS), patients’ perceived concerns and patients’ perceived necessity of BP medication (BMQ) predicted the number of BP entries they made during the study, controlling for age and baseline BP (Tables 27–29).