Behavioural interventions to promote physical activity in a multiethnic population at high risk of diabetes: PROPELS three-arm RCT

Objectively assessed time spent sedentary and in light- and moderate- to vigorous-intensity physical activity

The accelerometer that was used to measure the primary outcome detailed above also provided secondary outcome data on the number of censored steps taken per day, defined as steps taken above an intensity (500 counts per minute) used to distinguish between purposeful and incidental ambulation. 50 Commonly used Freedson cut-off points were used to distinguish between time spent sedentary and time spent in light, moderate and vigorous physical activity. 51 Compliance with the physical activity recommendations of undertaking at least 150 minutes of moderate- to vigorous-intensity physical activity per week was also assessed;52 this was calculated as (1) the accumulation of 150 minutes overall of at least moderate-intensity physical activity to meet the Chief Medical Officer’s updated physical activity guidelines52 and (2) the accumulation of 150 minutes per week of at least moderate-intensity physical activity in bouts of at least 10 minutes in duration.

Objectively assessed time spent in the postures of sitting/lying, standing and walking

Concurrently with, and in addition to, the ActiGraph accelerometer, participants were asked to wear an activPAL3™ device. This is a small, slim, monitor worn on the thigh that uses accelerometer-derived information about thigh position to determine body posture (i.e. sitting/lying, standing or stepping). The activPAL3™ was initialised using the manufacturer’s software with the default settings (i.e. 20 Hz, 10 seconds’ minimum sitting-upright period) and participants were asked to wear the device continuously (24 hours per day). The activPAL3™ was covered with a nitrile sleeve and fully wrapped in one piece of waterproof dressing [Hypafix Transparent (BSN Medical, Hamburg, Germany)] to allow participants to wear the device during bathing activities and was secured to the midline anterior aspect of the upper thigh using hypoallergenic waterproof dressing (Hypafix Transparent). Data were analysed using a bespoke open-source processing package [ProcessingPAL; URL: https://github.com/UOL-COLS/ProcessingPAL/releases/tag/V1.2 (accessed 16 November 2021)]. Data on waking wear time, time spent sedentary, time spent standing and time spent walking were utilised for this study.

Recent Physical Activity Questionnaire

Self-reported physical activity was measured using the Recent Physical Activity Questionnaire (RPAQ), which assesses physical activity across four domains (domestic, recreational, work and commuting) over the previous month. RPAQ has shown moderate to high reliability for assessing physical activity energy expenditure, and good validity in ranking individuals according to their time spent in vigorous intensity physical activity and overall physical activity energy expenditure. 53

Biochemical variables and diabetes diagnosis

Venous sampling was used to assess standard biomedical outcomes, comprising HbA_1c, lipid profile (triglycerides, HDL, LDL and total cholesterol), urea and electrolytes (sodium, potassium, urea and creatinine), and liver function tests (albumin, total bilirubin, alkaline phosphatase and alanine transaminase). Assays were completed in quality-controlled clinical laboratories at University Hospitals of Leicester NHS Trust and Cambridge University Hospitals NHS Foundation Trust. At the Leicester site only, participants were assessed for fasting and 2-hour post-challenge glucose and insulin levels following a 75-g OGTT; the OGTT results will be analysed at a later date and used to provide greater clinical insight into how physical activity affects metabolic health.

For the purposes of the main trials, the classification of glycaemic status was based on HbA_1c values using World Health Organization and NICE guidelines47,48 (see Table 1). During the trial, participants at the Leicester site who were found to have a HbA_1c value in the diabetes range (≥ 6.5% or 48 mmol/mol) were recalled for a second, confirmatory, test, and if diabetes was confirmed they were referred to their clinician. At Cambridge, the response to a HbA_1c value in the diabetes range was to send a letter to the individual’s primary care clinician informing them of the need to confirm a diagnosis.

Protocol for participants found to have type 2 diabetes during the trial

Participants diagnosed with T2D during the trial were retained and continued to be offered all study and interventional procedures, as the primary outcome measure was change in physical activity.

Standard anthropometric and demographic measurements

Height, body weight, body fat percentage and waist circumference were measured to the nearest 0.5 cm, 0.1 kg, 0.5% and 0.1 cm, respectively. Waist circumference was measured using a soft tape measure mid-way between the lowest rib and the iliac crest. Arterial blood pressure was obtained from the right arm of the participant when seated; three measurements were taken and the average of the last two measurements was used. Information on ethnicity, medication history, current smoking status, family history of diabetes in first- and second-degree relatives, and muscular/skeletal injury that prevents physical activity were obtained by self-report. Social deprivation was determined by applying the English Index of Multiple Deprivation (IMD) to participants’ postcodes. For descriptive purposes, data are categorised at the national quintile values to show that the distribution within the recruited population is generalisable to the national average.

Genetics

A blood sample for future genetic analysis was also collected from those who consented.

Cardiovascular risk

Secondary outcomes were used to estimate cardiovascular risk, calculated through the Framingham Risk Score. 54 The Framingham Risk Score has been shown to perform reasonably well in multiethnic UK populations, although it may underestimate risk in South Asian ethnic minorities. 55 However, the secondary outcomes in the PROPELS trial did not allow a more comprehensive risk score to be employed.

Sleep

Participants self-reported on two single-item questions asking about sleep duration in the last night and average sleep duration during a usual week.

Self-reported dietary behaviour

Dietary behaviour was captured in two short questionnaires used in previous research studies by our group, which were administered to participants for self-completion. The questions were based on an abbreviated dietary food frequency questionnaire developed for the European Prospective Investigation of Cancer and Nutrition (EPIC) study and a questionnaire of dietary intentions developed for the international NAVIGATOR (Nateglinide And Valsartan in Impaired Glucose Tolerance Outcomes Research) study. 56,57 The food frequency questionnaire captured portions per week of fresh fruit, green leafy vegetables, other vegetables, oily fish, other fish, chicken, meat, eggs, cheese and wholemeal/brown bread. In addition, alcohol intake (drinks per day) was captured. Dietary intentions captured the degree (on a 5-point Likert scale) to which each participant was actively trying to limit the amount of total fat, saturated fat, sugar or salt in their diet.

Health-related quality of life: EuroQol-5 Dimensions, SF-8 and Hospital Anxiety and Depression Scale

Health-related quality of life was measured using the EuroQol-5 Dimensions, three-level version (EQ-5D-3L),58 and the Short Form (SF-8) Health Survey. 59 The EuroQol-5 Dimensions (EQ-5D) is a standardised questionnaire developed for use as a measure of health outcomes. It defines health in terms of five dimensions: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. It is widely used to calculate QALYs, which are essential for cost-effectiveness analysis. The SF-8 is a self-administered questionnaire that measures eight health domains (general health, physical functioning, role limitations due to physical health problems, bodily pain, vitality, social functioning, mental health and emotional roles) using eight questions. The standard (4-week) recall format was used. Data from SF-8 responses were used to derive a physical component score and a mental component score.

Depression and anxiety symptomology were measured using the Hospital Anxiety and Depression Scale to produce independent subscales for anxiety and depression. 60

Health, medication and smoking status

Medical history and medication status were measured using an interview-administered protocol. Data on family history of diabetes and cardiovascular disease, smoking status and muscular/skeletal injury were obtained by self-report. All adverse events that were reported to the study sponsor (University of Leicester) were also recorded.

Health resources

A health resources questionnaire was used to record the number of times that the participant had seen a health-care practitioner (e.g. a GP, nurse or other health worker) over the previous 12 months and the number of times that they had been to hospital. In addition, the number of contacts and costs associated with the intervention were captured by the research team and were included in the cost-effectiveness analysis of the intervention.

General practice data

The research team attempted to collect relevant biochemical data, diabetes diagnosis data and other medical event data from during the trial directly from participants’ GP practices for those lost to follow-up. Collected data were matched to the nearest follow-up time point.

Potential mediators of behaviour change

Diabetes progression

We assumed a conversion rate to T2D in the control study arm of at least 24% over the entire 4 years of the study and would, therefore, have 80% power to detect a 40% reduction in the relative risk of T2D in both intervention study arms compared with the control study arm (α = 0.025). We used an estimated conversion rate at the lower level reported for traditionally defined prediabetes. 5,68

Analysis of the primary outcome: change in ambulatory activity (steps per day) at 48 months

The analysis involved two a priori comparisons in which each intervention arm was compared with the control study arm. Should either of these comparisons reveal a significant difference, a third a priori comparison would have been undertaken to compare the difference between intervention arms; this will be included as a secondary analysis. Data were analysed at the patient level using a modified intention-to-treat protocol; participants with complete data were analysed in the study arm to which they had been randomised, regardless of the dose of intervention they actually received. Analyses of covariance models were used, adjusting for wear time at baseline, wear time at 48 months, number of valid days at baseline, number of valid days at 48 months, the three randomisation stratification variables (centre, ethnicity and sex) and ambulatory activity at baseline as covariates. Missing data at baseline were replaced using the indicator method. 69

Ethnicity and other subgroup analyses for primary outcome

For the primary outcome only, interactions between randomised study arm and (1) sex (men/women), (2) age (< 60 years/≥ 60 years), (3) ethnicity (white European/South Asian/other), (4) family history of T2D (yes/no), (5) prediabetes status at baseline (yes/no), (6) baseline obesity status [< 30 kg/m² (27.5 kg/m² for South Asians), ≥ 30 kg/m² (27.5 kg/m² for South Asians)] and (7) baseline deprivation (split at median IMD score into high vs. low) were tested by including the relevant interaction parameters in the analysis model.

If the p-value of any of the interactions tested above was < 0.05, then the estimates and 97.5% confidence intervals (CIs) of the two intervention effects (Walking Away vs. control and Walking Away Plus vs. control) on the primary outcome were reported in the relevant subgroups.

Sensitivity and per-protocol analysis for the primary outcome

To access the possible impact of excluding those with data lost to follow-up from the analysis, the primary analysis was repeated replacing missing data using multiple imputation by chained equations, across 10 imputed data sets. In addition, to access whether or not adherence to the intervention affected the results, the primary analysis was repeated when restricted to a per-protocol data set in accordance with the following criteria:

• control – all individuals

• Walking Away – attended initial education and at least one follow-up annual refresher session

• Walking Away Plus – attended initial education and at least one follow-up annual refresher session and registered with the text message service and received the initial telephone call and received at least one further telephone call during the trial.

Analysis of continuous secondary outcomes

Ambulatory activity at 12 months and all other continuous secondary outcomes at 12 and 48 months were analysed using the same strategy and at the same time points as described for the primary outcome, with the exception that accelerometer wear values were removed from non-accelerometer-derived outcomes.

Analysis of binary secondary outcomes

The odds of compliance with moderate to vigorous physical activity recommendations at 12 and 48 months were analysed using logistic regression, adjusted for the randomisation stratification variables (centre, ethnicity and sex) and compliance with moderate to vigorous physical activity recommendations at baseline as covariates.

The odds of diabetes at 12 and 48 months were analysed using logistic regression, adjusted for randomisation stratification variables (centre, ethnicity and sex). Those who were diagnosed with diabetes but had a HbA_1c value subsequently recorded in the non-diabetes range were still classified as having diabetes.

Potential mediators of behaviour change and self-reported use of behaviour change strategies

To avoid multiple testing with measures that were not study outcomes, data on illness perceptions, self-efficacy and self-reported use of behaviour change strategies are reported descriptively, but were not subject to statistical testing.

Statistical significance and reporting of data

As the primary outcome involved two comparisons with the control (Walking Away and Walking Away Plus), the significance level was adjusted accordingly to account for this multiple testing. Therefore, statistical significance was considered at a p-value of < 0.025, rather than the tradition p-value of < 0.05. The results of the statistical analyses are, therefore, reported as means (97.5% CIs) to be consistent with the lower significance level.

The cost of the interventions

Two different sets of intervention costs were calculated. One was the expected costs of the interventions as they would be implemented in the real world. The intention in using the costs of interventions as we expect them to be implemented is to correct for budgetary inefficiencies that are an artefact of the trial process and would not be expected to occur in a real-world setting. To estimate these, we combined data collected in the study, data from other similar educational interventions and expert opinion. These costs were used as our base-case costs in both analyses.

The other type of intervention was the cost of interventions in the PROPELS trial, regardless of whether or not these costs reflect realistic costs, if Walking Away or Walking Away Plus were to be implemented. These costs will be presented as a sensitivity analysis.

Model-based analysis

The School for Public Health Research model was used to simulate the long-term incidence of T2D, complications of T2D and other related conditions, such as cardiovascular disease, cancer and depression. The model is well validated for use in diabetes prevention interventions73 and was adapted to the specific requirements of the PROPELS trial in three ways:

1. The rates of progression to T2D in the first 4 years of the model were based on a statistical model derived from data collected in the trial.

2. Daily step count was added as a new population characteristic to incorporate the primary outcome variable of the trial.

3. The existing cardiovascular risk function was adapted to incorporate the independent effect of ambulatory activity on cardiovascular risk, as reported in the NAVIGATOR trial. 20

The model population was simulated based on an analysis of the baseline characteristics of the trial participants. Additional characteristics required for the model that were not collected in the trial were imputed using Health Survey for England data. 74 Owing to over recruitment of South Asian participants in the trial compared with the general population of the UK, analyses were performed separately for South Asian and non-South Asian populations. These results were then aggregated proportionally to the ethnic structure of the UK to provide an estimate of cost-effectiveness that was representative of the UK population as a whole.

The primary model outcomes were aggregated lifetime costs and lifetime QALYs, both of which were discounted at a rate of 3.5% per annum in line with NICE recommendations. 75 Relevant clinical outcomes, including the number of diabetes diagnoses, incidence of diabetes complications and incidence of related conditions (including cardiovascular events), are also reported.

Cost-effectiveness was reported as an ICER and net monetary benefit assuming a willingness-to-pay threshold of £20,000 per QALY, as recommended by NICE. 75 Uncertainty around these results was explored through probabilistic sensitivity analysis (PSA) and value-of-information techniques. We also conducted scenario analyses, as outlined in Chapter 6.

Within-trial analysis

The within-trial analysis was conducted in line with Ramsey et al. ’s76 2015 recommendations for cost-effectiveness analysis alongside clinical trials. Resource use was calculated using data collected as part of the trial, and the cost of this was valued using standard unit sources, including, but not limited to, the British National Formulary77 and NHS Reference Costs. 78 The outcome measure was QALYs. Utility values were calculated by mapping the EQ-5D-5L data collected in the trial to the EQ-5D-3L scores, as recommended by NICE. 75 The resulting scores were valued using the standard EQ-5D-3L tariff for the UK population. 79 The total QALYs were estimated using an area-under-the-curve method. The costs and QALYs accrued after the first year were discounted at 3.5% in line with NICE guidance. 75 As for the model-based evaluation, the results are presented as an ICER and net monetary benefit, assuming a willingness-to-pay threshold of £20,000 per QALY. 75

Trial Steering Committee

The trial was overseen by an independent Trial Steering Committee (TSC), which was responsible for the overall management and oversight of the trial. The TSC was made up of:

• Simon Heller, Professor of Clinical Diabetes, University of Sheffield (chairperson)

• Richard Morris, Professor in Medical Statistics, University of Bristol

• Des Johnston, Professor of Clinical Endocrinology, Imperial College London.

The TSC met approximately every 6 months, normally within 2 months of a meeting of the Data Monitoring and Ethics Committee (DMEC).

Data Monitoring and Ethics Committee

A fully independent DMEC reported to the TSC. Although it was highly unlikely that the non-pharmaceutical, non-invasive interventions proposed would result in any substantial negative effects on trial participants, the DMEC held responsibility for the interests of participant safety and data integrity and reviewed all reported adverse events. It also assessed data at 12 months, with predefined rules for stopping the trial for futility if specific criteria were met. These criteria were based on whether or not there was evidence that the intervention was causing harm, defined as a decrease in the primary outcome (ambulatory activity) in the intervention arms compared with the control arm, based on the 99% CI and 95% CI. If a decrease was seen based on the 99% CI in both arms, then the trial would be terminated. If a decrease was seen based on the 95% CI, then secondary outcomes would be considered in making a recommendation for termination. These criteria were not met.

In addition, the DMEC reviewed and signed off the statistical analysis plan. The DMEC comprised Graham Hitman, Professor of Molecular Medicine and Diabetes, Queen Mary, University of London (chairperson); Naveed Sattar, Professor of Cardiovascular and Medical Sciences, University of Glasgow; and Michael Campbell, Emeritus Professor of Medical Statistics, University of Sheffield.

Data integrity

The study was reviewed by the NHS National Research Ethics Service Committee East Midlands – Leicester and the Comprehensive Local Research Network (ethics number 12/EM/0151). All data were entered (through secure web-based access) and held in a specifically designed database in the Leicester Clinical Trials Unit. The database was designed with internal validity and quality control checks; potential errors were highlighted to, and corrected by, the study team. Data were released to the study statistician at predefined time points.

Ethics approval

Ethics approval was granted by the NHS National Research Ethics Service, East Midlands – Leicester Committee, which co-ordinates ethics permissions across the following study and recruitment sites:

• University Hospitals of Leicester NHS Trust

• Leicester City Clinical Commissioning Group

• West Leicestershire Clinical Commissioning Group

• East Leicestershire and Rutland Clinical Commissioning Group

• University of Cambridge

• MRC Epidemiology Unit

• Cambridge and Peterborough Clinical Commissioning Group

• Cambridge University Hospitals NHS Foundation Trust.

Amendments

The PROPELS protocol was subject to 14 amendments, none of which changed the main aims or objectives of the trial; for a list, see Appendix 3.

Trial registration

The trial was registered with ISRCTN as ISRCTN83465245: The PRomotion Of Physical activity through structured Education with differing Levels of ongoing Support for those with pre-diabetes (PROPELS) (https://doi.org/10.1186/ISRCTN83465245).

Walking Away: the initial education session

Walking Away was delivered by two trained educators to groups of up 10 participants, with participants invited to bring a family member or a guest if they wished. Sessions were delivered in a variety of settings chosen for proximity to the recruiting GP surgeries, including at the surgeries themselves, in nearby community centres and at hospital sites.

The curriculum for the initial education session, examples of activities and the underlying theories and behaviour change techniques are presented in Table 3. Walking Away was aimed at increasing participants’ knowledge of diabetes risk, changing their outcome expectations about how physical activity can lead to decreased diabetes risk, and strengthening their self-efficacy for engaging in increased physical activity. The programme was designed to harness key behaviour change techniques related to self-regulation, including goal-setting, action-planning, self-monitoring and barrier identification/problem-solving to ensure that motivation for behaviour change translated into actual behaviour change. Self-monitoring was supported through the provision of a pedometer [Yamax SW200 (Yamax, Shropshire, UK)]. Participants used their daily habitual step count (measured at baseline prior to the education programme) to set personalised steps-per-day activity goals.

Participants were encouraged to increase their activity levels by up to 3000 steps per day, equivalent to around 30 minutes of walking. Goal attainment was encouraged through the behaviour change technique of setting graded tasks, whereby the use of proximal objectives, such as increasing ambulatory activity by 500 steps per day every 2 weeks, is used to work up to overall goals. Participants were encouraged to make an action plan detailing where, when and how they would reach their first proximal goal, to repeat action-planning for each new proximal goal, to wear their pedometer on a daily basis and to self-monitor their ambulatory activity using a specifically designed steps-per-day diary. Although this was primarily a physical activity intervention, a short time was allocated to covering the key dietary messages because participant groups had requested this during the intervention development process for Walking Away.

Walking Away: the annual refresher sessions

As for the original Walking Away RCT, after the initial education session, participants were offered annual group-based maintenance sessions at 12, 24 and 36 months (‘refresher sessions’). Refresher sessions each lasted 2.5 hours and were designed to revisit the key messages of the initial session, strengthen self-efficacy through sharing successes and prompt problem-solving in relation to barriers, goal-setting and self-monitoring using pedometers. The annual nature of the group sessions was designed to fit with primary care pathways, in which annual clinical follow-up is recommended for those with a high risk of chronic disease, such as prediabetes. 43,86

Text messaging system

Text messages were used to prompt pedometer use and provide tailored feedback on meeting steps-per-day goals. The frequency of texts received varied over time to coincide with the initial and annual refresher face-to-face group sessions (Table 4).

The text messaging system was hosted on a secure University of Cambridge server. The program consists of two parts:

1. a set of tables in a MySQL database

2. a set of PHP (hypertext preprocessor) scripts.

The MySQL tables contain data from participants, which were obtained in two ways:

1. data elicited by educators during telephone calls (see Telephone support)

2. data extracted from text messages participants sent to the system, such as the number of steps participants reported having taken and a STOP message sent by a participant indicating that they did not wish to receive any further text messages.

A messages table contained the bank of messages that were developed for the PROPELS trial. Schedule tables specified the days and time slots when text messages should be sent. Matrix tables specified which variables should be used to individually tailor each message.

Messages were tailored in two ways:

1. Different participants received different messages depending on the values of the relevant tailoring variables.

2. Tags were embedded in some messages enabling variable values to be inserted dynamically (e.g. #nname# to insert the participant’s nickname).

Other tables stored the educators’ identification numbers, messages sent by participants to the system and messages sent by the system to participants. Messages sent to participants included query messages, for example asking them to text in their weekly step count.

Participants registered with the system by sending a text that included the keyword ‘PROPELS’ and their nickname to a specified number. The system was fully automated except for the educator interface, which allowed manual data entry. The main PHP (hypertext preprocessor) script was set to run every 15 minutes from 06.05. The first time that it ran each day it set up the text messages to be sent that day. Then, every 15 minutes it checked whether or not the time for sending the messages had been exceeded. If it had, the messages were sent.

A message was ‘sent’ by sending a HTTP (HyperText Transfer Protocol) request to a company called FastSMS, which relayed the message over the mobile telephone network. Other scripts sent birthday and New Year messages, processed incoming texts (relayed from FastSMS) and queried the database to produce reports, for example of messages sent that day.

Telephone support

Telephone support from trained educators was used to support and tailor the text messaging system. An educator telephoned the participant approximately 1 week after the Walking Away session to confirm short- and long-term step goals and an action plan for the next 6 months, and to elicit information needed to enable the text messaging system to be tailored to variables such as confidence in increasing physical activity, previous experience of physical activity and potential mobility issues that prevented walking from being the primary activity. The collected information was captured in an online form and saved to a database for use in the text messaging programme. Participants also continued to receive two telephone calls annually to review their progress.

Structure and intensity of mHealth follow-on support

The structure and intensity of the mHealth intervention used in the PROPELS trial is detailed in Table 4.

This annual structure was repeated each year following each group education refresher session for the 4 years that constituted the intervention period.

The mHealth intervention content and the structure used were developed and piloted within the PROPELS programme of work. This development work, along with examples of the content used, is detailed below.

Development of the mHealth follow-on support intervention: a pragmatic framework for developing and piloting a text messaging intervention

The robust development of mHealth behaviour change interventions can be time-consuming, and yet in RCT protocols87,88 this is often allocated limited time. The time constraints of the PROPELS RCT protocol provided for 12 months to conceptualise, develop and test the follow-on support programme prior to the commencement of the RCT. Our methods offer a pragmatic framework for developing and piloting a text messaging intervention – drawing on relevant behaviour change theory and using rigorous qualitative methods incorporating user engagement – that lends itself well to replication and application to the development of similar interventions.

The structured, iterative process that we used to develop the PROPELS follow-on support programme involved undertaking concurrent and sequential research with the target population while maintaining a strong focus on the integration of theory and evidence. The methods that we employed combined features of published mHealth development frameworks89,90 and multiple iterative phases of qualitative research similar to a user-centered design process. 88 Our framework for intervention development and piloting was informed by the model by Dijkstra and De Vries89 for developing computer-generated tailored interventions (to conceptualise the programme) and the mHealth development and evaluation framework by Whittaker et al. 90 and Fjeldsoe et al. 91 The high level of engagement with our target population enabled us to refine the design to optimise its acceptability to users.

The four phases of the development process are described below.

Phase 3: pretesting

Based on the findings of phases 1 and 2, we created model text messages and conducted four further focus groups with participants (n = 20; age range 52–77 years) to test these. Eligibility criteria for these were the same as in phase 2, but we also invited participants from the control arm of the Walking Away study who had not previously attended the programme;80 the recruitment and consent procedures were identical to those described for phase 2. Before attending a pretesting focus group, participants were sent a pedometer and activity diary by post and were encouraged to record their number of steps per day for 1 week. Participants were asked to bring along a mobile telephone when attending the focus group.

A topic guide covered participants’ experiences of Walking Away, as in phase 2, and also explored experiences of wearing the pedometer and recording steps. During the focus group, participants were sent example text messages (Figure 1) to provoke reactions in situ and generate think-aloud87 reactions and discussions about different message types. Data were analysed using the approach described in phase 2, with the phase 2 coding framework developed further to reflect the phase of development.

FIGURE 1.

Example text messages. Reproduced with permission from Morton et al. 100 This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mhealth and uhealth, is properly cited. The complete bibliographic information, a link to the original publication on http://mhealth.jmir.org/, as well as this copyright and license information must be included. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Morton et al. 100 This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mhealth and uhealth, is properly cited. The complete bibliographic information, a link to the original publication on http://mhealth.jmir.org/, as well as this copyright and license information must be included. The figure includes minor additions and formatting changes to the original figure.

During the pretesting focus groups, participants were sent a range of text messages that were developed as a result of phases 1 and 2, which included:

1. reminder texts – reminding participants to wear the pedometer and log daily steps and instructions for texting in step counts

2. feedback texts – giving feedback on behaviour, including social reward and positive reinforcement

3. motivational texts – messages using behaviour change techniques to strengthen motivation for physical activity (e.g. habit formation, commitment, reframing physical activity beliefs)

4. information texts – information about health consequences

5. problem-solving texts – contained response options to a list of predefined barriers when a goal was not met (see Figure 1).

Depending on the response option that they chose, participants were sent a tailored motivational or information text.

The key themes emerging from the phase 3 pretesting focus groups that informed the final content of the PROPELS follow-on support programme are described in the following sections.

Self-monitoring of physical activity

Most participants reported that self-monitoring their daily steps with the pedometer had increased their awareness of their own activity and, therefore, had increased their motivation to be more active:

I found the pedometer really, really useful. I didn’t wear it all the time, but once I wear it I make sure I do 10,000 steps. If I looked at it half-way through the day and think I’ve only done 5000 then I went out for a walk purposely just to get the figures up.

FG4

Some participants reported finding the pedometer demotivating or disheartening, especially those with mobility problems, who felt that their step count was always low because they could not engage in walking as their primary activity:

I wish that it was not just dependent on the steps. Because we do all sorts of other things rather than just steps.

FG7

For such individuals, the self-monitoring process should allow for other activities to be counted (e.g. swimming and gardening).

Text message type, language and frequency

Reminder texts

It was clear from participants’ feedback that establishing the appropriate frequency of reminder texts was critical to avoiding the intervention becoming off-putting and perceptions that it was ‘Big Brother’-like or ‘checking up on you’, especially once wearing the pedometer had become a habit. This suggested that reminders should become less frequent as the intervention progresses:

If you’ve got something constantly . . . well, not constantly, but weekly reminding you to do something then you’re still there doing it. And possibly if you’re doing it for several weeks then you’ll get actually used to wearing it and putting it on. It’s like putting your clothes on. You put your socks on, put your pants on, ‘oh I’ll put my thing [pedometer] on’. It’s all getting used to what you’re doing, like with your lifestyle.

FG5

Prompting texts

A text message to prompt people to input their weekly step count was generally accepted to be a useful motivational tool:

I suppose the very fact that we would be doing it [texting in step counts] we are creating a certain level of discipline which we didn’t have before.

FG4

Feedback texts

The model feedback messages sent when a step goal was achieved (i.e. positive reinforcement) were well received and helped to foster a sense of accountability:

We are all school kids in a sense, in our heads, so if someone says you did well it’s really encouraging.

FG5

We tested a variety of feedback messages to be used in the event that participants did not achieve their step goal. The consensus was that these should be fairly light-hearted, positive and encouraging. Messages emphasising discrepancy between the person’s current behaviour and their goal were well received, provided that the texts also offered encouragement and support by, for example, including positive elements along with more negative feedback:

. . . you’ve got to put in, you know, the positive that eliminates some of the negativity out of the messages. So this one was ‘thanks for the text, keep wearing your monitor and logging your steps, try to increase your activity to ensure’ . . . it’s not quite positive enough.

FG4

Several participants remarked that humour could be useful in providing feedback when a step goal was not achieved:

You can’t castigate somebody but you can try and get some laugh out of it from some point of view, saying ‘get off your bottom and go for a walk!’.

FG6

However, they also recognised that messages could be received differently and that the use of humour was risky, especially when participants’ confidence was low:

. . . if I read that and I was in the wrong mood I’d take that as you’re telling me what to do, and I’d say ‘b*****r off’.

FG4

Motivational texts

The feedback on motivational messages was far from uniform. Overall, participants indicated that the gentle language and suggested content of the motivational messages rendered them acceptable, avoiding ‘being told you’ve got to do it’. Some of the model messages were perceived as dated (e.g. recommendations to avoid using a remote control to change the TV channel) or irrelevant (e.g. tips about using stairs at home: ‘. . . but I live in a bungalow!’). Participants tended to prefer practical tips and suggestions for increasing activity over more motivational suggestions (e.g. ‘try writing down your barriers to activity this week’). In one focus group, participants suggested using generally supportive messages not necessarily linked to physical activity or health:

I know why I’m doing it [to reduce the chances of T2D] so we don’t need reminding of it all the time.

FG7

Information texts

Participants were in agreement that messages were too focused on the health consequences of inactivity and that a focus on benefits other than weight and reduced risk of T2D would be preferable:

You could just say ‘good morning, this is PROPELS, hope you have a nice day’ or whatever . . . just simple – it doesn’t need to really say anything.

FG6

. . . when you’ve got a weight problem like I’ve got, I don’t need to be reminded – I’m doing my best!

FG4

Problem-solving texts

The predefined response format was viewed by some as unsuitable for problem-solving:

It’s like one of those PPI messages [spam text messages about reclaiming mis-sold insurance] — I hate those!

FG4

However, others valued the idea of being able to easily text in the reason why they had not achieved their goal. Participants generally liked the tailored and personalised texts that were triggered by their responses to the problem-solving texts (e.g. in response to selecting the illness or injury response option, participants would receive the message ‘Take it easy this week. We hope that you feel better soon’).

Tailoring

The concept of receiving individually tailored text messages was very popular, in particular in relation to goal progress and/or achievement:

You should get the one [text message] that’s relevant to you. If you’re doing more [steps], if you’re achieving your target or doing more, you still get one, but it should be different.

FG4

Participants observed that different people need different support, especially in terms of confidence and self-discipline in adhering to an activity plan. They suggested that people who are struggling to meet their goal and/or have mobility problems limiting the amount of walking that they could achieve should receive messages that are less direct or less pushy.

Language and frequency

The general feedback was that the language used in texts needed to be formal, friendly and polite, with use of the participant’s name but limited use of emoticons:

I’m just warning you that it might be interpreted that you are shouting at us because in text language, capitals [letters] is shouting.

FG5

It makes it sound as though you’re talking at us, rather than a computer.

FG6

Participants felt that less is more when it comes to the frequency of messages. Overall, they perceived the approach of sending daily messages to be too heavy-handed and potentially demotivating:

. . . otherwise if you are going to get this [text message] daily you’re going ‘oh another one’ and you get fed up with it.

FG6

Phase 3 findings

• Further emphasised the importance of personalising and tailoring messages according to key variables (e.g. previous levels of physical activity, mobility issues that limit physical activity, individuals’ confidence in increasing physical activity, goal achievement/progress).

• Shaped the content of the messages (i.e. what type of benefits to focus on in the motivational messages).

• Informed the frequency of messaging and sequencing of the follow-on support programme.

• Highlighted the importance of including other activities (e.g. cycling and swimming) to maintain engagement of participants who did activities other than walking.

Sampling and recruitment

We purposively sampled five education sessions at the 12-month time point so that we could achieve a range in terms of educator, location and participant demographics. When booking participants into an education session that was scheduled to be observed, the booking administrator informed the participant about the observation; if a participant wanted to attend a session that was not being observed, they were offered a different session. On the day of the sessions, the researcher introduced themselves to the participants and explained the purpose of the observations; during the session, they observed from the back of the room to minimise distraction and interference.

Data capture

The researcher took anonymised handwritten field notes while not interfering with the education session. 111,112 The focus of the observation was engagement with the intervention components, including with the education session itself, and how the participants talked about their experience and levels of physical activity (including pedometer use, diary and/or text messaging, where appropriate) during the first 12 months. Key modules in the PROPELS curriculum were ‘the participants’ story’ and ‘physical activity’, which facilitated participants to talk about the goals that they had set, the challenges that they faced, the strategies that they used to overcome these and the challenges that they had not overcome. The field notes informed the topics to be explored in the qualitative interviews later.

Sampling and recruitment

Final sample

Intervention costs

Two sets of intervention costs were calculated: one relating to the estimated costs of delivering Walking Away and Walking Away Plus, given that it is likely to be delivered in real-world practice, and the second relating to the cost of delivering Walking Away and Walking Away Plus, as per the PROPELS trial protocol. The former set was developed in consultation with personnel involved in delivering Walking Away and related interventions, and was intended to represent the most likely costs faced by the NHS were Walking Away or Walking Away Plus to be implemented. The intention is to correct for any artificially high or low implementation costs that may be an artefact of the trial process. The real-world costs were used in the base-case analyses.

The Walking Away intervention costs included educator and administrative staff time, physical resources (e.g. pedometers and teaching materials), venue hire and refreshments, staff training and participant expenses. In addition, there were further costs that applied only to the Walking Away Plus arm, namely the cost of follow-up telephone calls and text messages.

The trial management group anticipated that, in a real-world setting, staff would need to be retrained less frequently than the 4 years of the trial would suggest. Therefore, the cost of training staff was annuitised over 10 years (in line with assumptions made for similar structured education interventions114) using the following annuity calculation:

where r = discount rate (3.5%) and n = expected lifetime of the training (10 years).

In the trial, participants from both intervention arms attended the same education sessions; therefore, the common costs were totalled and divided proportionally to the number of participants randomised to each arm. The Walking Away Plus additional costs were added to this to yield a total cost per arm that was divided by the number randomised to produce a per-participant cost for each arm.

Perspective, discounting and time horizon

In line with the NICE methods guide,75 the analyses took an NHS and a Personal Social Services perspective; future costs and QALYs were discounted at a rate of 3.5% per annum; and the time horizon was a lifetime horizon in the primary analysis and a 4-year time horizon in the secondary analysis.

Outcome measures

Costs and QALYs were estimated for each strategy.

The primary outcome measure was the ICER, which is change in cost/change in QALYs. As there are three strategies in this decision problem, the strategies were ordered by the total number of QALYs produced and ICERs were calculated comparing each strategy to the next least effective strategy. Any intervention that was dominated (i.e. produced fewer QALYs at a higher cost than another intervention) or was extendedly dominated (i.e. if a strategy’s ICER is higher than the ICER of the next most effective intervention) then these strategies were removed when calculating ICERs.

The value of information was calculated for this decision problem. We calculated the value of expected value of perfect information (EVPI). This statistic is the value to a decision-maker (e.g. NICE) of eliminating the uncertainty in all parameters in the economic analysis. Therefore, it provides an upper limit on any future research to reduce uncertainty on these sets of strategies. We also calculated the parameter EVPI, which is the value to the decision-maker of eliminating all uncertainty in a subset of parameters in the model, for step count, diabetes diagnoses and HbA_1c.

The model population

To model the cost-effectiveness of Walking Away and Walking Away Plus in the general population, we ran the model twice for each analysis: once with a population representing the South Asian trial subpopulation, and once with a population representing the non-South Asian trial subpopulation. The results from the two subgroups were weighted using data from the UK government117 on the relative size of these two populations in the UK. For example, if we found that a strategy resulted in 10 discounted QALYs being accrued in the non-south Asian population and 11 discounted QALYs being accrued in the South Asian population, and if 10% of the population was South Asian, then the total number of discounted QALYs accrued for that strategy would be 10.1, because (11 × 0.1) + (10 × 0.9) = 10.1.

Each of these subpopulations was synthesised based on an analysis of the trial participants at baseline, using the following steps:

1. assessment of the distribution of each continuous variable required by the model (e.g. height, weight and systolic blood pressure)

2. transformation of any non-normally distributed variables to generate approximately normal distributions for each variable

3. construction of a multivariate normal distribution using all variables (both continuous and categorical) simultaneously to preserve correlations between variables

4. sampling of each variable for all simulated individuals

5. for continuous variables:

   1. – generation of cut-off points for each categorical variable using the proportional frequency of each category in the PROPELS data

   2. – the sampled variable was converted back into the appropriate category using these cut-off points.

6. back-transformation of any transformed variables to the natural scale.

Step count modelling

At baseline, step count was sampled from a multivariate normal distribution as described previously. At 12 and 48 months, step count was simulated using a beta-regression. Beta-regressions predict two outcomes: a mean effect and dispersion in the mean effect. Variance can be predicted for any mean effect and dispersion using the following formula:

Furthermore, beta-regressions require that the data that are being predicted are on a scale between 0 and 1. We transformed step count so that 0 was equivalent to 1 step lower than the lowest value in the PROPELS trial (464 steps per day) and 1 was equivalent to 1 step per day more than the highest value in the PROPELS trial (25,341 steps per day).

The key advantage of this approach is that it allows heterogeneity to be included in the simulation because we know the expected number of steps per day and the expected variance in the number of steps per day for every individual. This means that we do not have to assume that an average treatment effect is appropriate, with the level of heterogeneity in identical individual outcomes being driven by the data.

In line with the statistical analysis, to estimate step count at 4 years we controlled for two indicator variables for randomised study arm (Walking Away vs. control, Walking Away Plus vs. control), the three randomisation stratification variables (centre, ethnicity and sex), ambulatory activity at baseline and ambulatory activity at 1 year as covariates. Unlike in the statistical analysis, we did not include wear time at baseline, wear time at 48 months, number of valid days at baseline or number of valid days at 48 months as covariates in the regression model, as predicting these covariates in our simulation model would have increased the computational complexity of the simulation and the causal linkages between these parameters and measured step count were unclear. We applied these control variables to both the mean effect and the dispersion parameters of the beta regression, with the exception of ethnicity for the dispersion parameter, as the models failed to converge when this covariate was included.

To estimate step count at 1 year, we controlled for two indicator variables for randomised study arm (Walking Away vs. control, and Walking Away Plus vs. control), the three randomisation stratification variables (centre, ethnicity and sex) and ambulatory activity at baseline as covariates. We applied these control variables to both the mean effect and the dispersion parameters of the beta-regression, again with the exception of ethnicity for the dispersion parameter.

As the beta-regression produces data on the mean and variance for each individual in the model, we simulated each individual’s step count by randomly sampling this from their individualised beta-distribution (based on their expected mean step count and expected dispersion). We did this separately for each individual’s step count at 1 and 4 years. After we had sampled each individual’s step counts (on the 0–1 scale), we transformed these simulated values back into steps per day. Between 0–12 and 12–48 months, step count was assumed to be linear. Beyond 48 months, step count was extrapolated based on four different possible scenarios. First, an underlying trajectory for step count was estimated by constructing an ordinary least squares linear regression model of step count and age at baseline, adjusted for sex, site, intervention arm and ethnicity from the PROPELS data. The coefficient for age from this model was used as the underlying annual change in step count in a no-intervention scenario for each individual. The observed step count was then compared with this underlying trajectory in the following four scenarios (Figure 6):

• A. Step count increased between 0 and 12 months and then declined to 48 months, but was above the underlying trajectory at 48 months. In this case, step count would continue to decline at the same rate as between 12 and 48 months until it converged with the underlying trajectory, at which point the individual switched to the underlying trajectory.

• B. The inverse case of (A) – step count decreased between 0 and 12 months and then increased to 48 months but remained below the underlying trajectory. In this case, the step count would continue to increase at the same rate as between 12 and 48 months until it converged with the underlying trajectory.

• C. An individual experienced improvement during the trial such that their 48-month step count was above the underlying trajectory and step count had increased between 12 and 48 months. In such a case, the individual would not converge with the underlying trajectory were their rate of change to continue. In such a case, the annual rate of change for the underlying trajectory would be applied after 48 months such that the individual step count trajectory was parallel to the underlying trajectory.

• D. The inverse case of (C) – the individual experienced a decline in step count from 12 to 48 months and their 48-month step count was below the underlying trajectory. In this case, the annual rate of change for the underlying trajectory was applied after 48 months such that the individual’s trajectory was parallel to the underlying trajectory.

FIGURE 6.

Example step count trajectories. In each scenario, line A represents the individual’s trajectory between years 1 and 4, line B represents the underlying trajectory, and line C represents the modelled step count.

Step count and QRISK

It was assumed that the unadjusted QRISK2 algorithms currently in the model adequately described the baseline risk of cardiovascular disease (CVD) for an individual with the average baseline step count in the PROPELS trial.

A QRISK2 modifier was calculated for each individual in the model, based on model 3 (the fully adjusted model) from the NAVIGATOR analysis. 20 This analysis was a Cox’s proportional hazards model, which concluded that the hazard of cardiovascular events was reduced by 10% for every 2000-step increment at baseline and that a change from baseline of 2000 steps in 12 months was independently associated with an 8% reduction in cardiovascular event rate.

From this, per-step hazard changes were calculated by finding the 2000th root of 0.9 and 0.92, respectively. Individual hazard ratios for each patient were calculated, which adjusted for their baseline ambulatory activity and the change in step count that they experienced in the RCT. These hazard ratios were recalculated every year. The formulae for calculating each individual’s hazard ratio is as follows:

Therefore, each individual had a hazard ratio at baseline depending on their baseline ambulatory activity, which was updated every 12 months depending on their change in ambulatory activity. The individual-estimated rate of having a CVD event within the next year (as currently calculated in the existing model structure using the QRISK2 algorithm) was multiplied by the annual hazard ratio to adjust the risk of CVD events according to individual step count.

HbA_1c trajectories

As for step count, we used beta-regressions to estimate HbA_1c levels at 1 and 4 years due to their advantages in being able to incorporate heterogeneity. HbA_1c was transformed so that 4.5% (0.1% lower than the smallest value in the PROPELS trial) was equal to 0%, and 9.2% (0.1% higher than the largest value in the PROPELS trial) was equal to 1.

To estimate HbA_1c levels at 4 years, we controlled for two indicator variables for randomised study arm (Walking Away vs. control, and Walking Away Plus vs. control), HbA_1c at 1 year, the three randomisation stratification variables (centre, ethnicity and sex) and HbA_1c at baseline as covariates. We applied these control variables to both the mean effect and the dispersion parameters of the beta regression.

To estimate HbA_1c at 1 year we controlled for two indicator variables for randomised study arm (Walking Away vs. control, and Walking Away Plus vs. control), the three randomisation stratification variables (centre, ethnicity and sex) and HbA_1c at baseline as covariates. We applied these control variables to both the mean effect and the dispersion parameters of the beta regression.

Between 1 and 4 years we assumed that HbA_1c levels changed linearly between the two time points. After 4 years, we assumed that HbA_1c in the treatment was maintained or declined in the same way that we have assumed that step count was maintained or declined (see Step count modelling). From the fifth year onwards, the HbA_1c levels of individuals are estimated using the Breeze et al. 115 trajectory models.

Modelling uncertainty

To assess uncertainty in the model, we ran both deterministic analyses and PSAs. In the deterministic case, we set all model parameters to their mean values. The model ran one fixed simulated population. In the PSA, we varied all model parameters simultaneously within their distributions and performed many different model runs. We used this set of PSA results to estimate mean costs and QALYs, and also calculated 95% CIs around these point estimates. Initially, we ran 1000 different sets of parameters. We assessed the results for stability using Hatswell et al. ’s118 criteria, and performed more PSA runs when needed.

We further explored the sources of model uncertainty using value-of-information analysis, which estimated the potential cost to the decision-maker of making the wrong decision owing to uncertainty around the results. We used Sheffield Accelerated Value of Information71 to do this. We assumed that the maximum number of people affected by the decision was equivalent to the number of people expected to be offered the National Diabetes Prevention Programme (a similar intervention that is currently being rolled out nationally), namely 100,000 per year,119 and that the decision would be relevant for 10 years.

Threshold analysis

We conducted a threshold analysis to determine the justifiable cost of the interventions for them to be cost-effective relative to usual care at a maximum acceptable ICER of £20,000 per QALY. To do this, we used a rearranged incremental net monetary benefit equation:

Costs

The costs that each participant incurred included the cost of the intervention (methods described previously), the cost of health-care resource use and the cost of medications taken during the trial period. Data on resource use were collected in the trial and data on unit costs were obtained from nationally representative sources. Within the trial, participants were asked about resource usage within the past 12 months at 0, 12 and 48 months. Resource use in years 2 and 3 was assumed to be the same as at 48 months. Costs incurred in years 2, 3 and 4 were discounted at 3.5% per year. 20

Resource use

Self-reported data on health-care resource use at baseline and at 12 and 48 months were taken from the PROPELS data, namely the number of times in the last 12 months that they had used each of the following health-care resources: GP, practice nurse, other health-care professional, attendance at accident and emergency, outpatient clinics, day hospital and inpatient stays. Costs of these resources were obtained from the PSSRU121 and NHS reference costs122 and are summarised in Table 12.

Medications

Self-reported data on medication use at baseline and at 12 and 48 months were taken from the PROPELS data. It was assumed that a participant who was taking one or more of these medications was doing so long term and was expected to be taking the medication for the duration of the trial. If a participant was found to have been taking a medication at 48 months, but not at 12 months, the date of their commencing medication was assumed to be midway between these time points (i.e. 30 months). The cost of 1 year of treatment on each medication type was taken from the Regional Drug and Therapeutics Centre (Newcastle) Cost Comparison Charts. 79 It was assumed that participants had been prescribed the most commonly prescribed preparation of each medication class (as reported in the Prescription Costs Analysis120). The annual cost of treatment with the drugs included is shown in Table 12.

Utilities

The benefits of the intervention arms were estimated QALYs, calculated from EuroQol-5 Dimensions, five-level version (EQ-5D-5L), scores. 123 In line with current NICE recommendations, utility values were generated by mapping the EQ-5D-5L domain scores to EQ-5D-3L value set. Individual 5L dimension scores were mapped to 3L scores using the van Hout algorithm. 122 The mapped 3L scores were aggregated into health state indices, which were valued using the Dolan et al. 79 valuation model.

Quality-adjusted life-years

Per-participant QALYs were calculated from the utility values using the area-under-the-curve method. In line with the NICE methods guide,20 QALYs for years 2, 3 and 4 were discounted at 3.5% per year.

Cost–utility analysis

The incremental costs and benefits of the interventions were calculated using seemingly unrelated regression equations, implemented using the systemfit package in R. 119 The equation system comprised two equations:

The fitted covariates included site, age, sex, ethnicity, and baseline costs and QALYs.

Results are presented as incremental costs, incremental QALYs, ICERs and incremental net monetary benefit, assuming a willingness-to-pay threshold of £20,000 per QALY in line with NICE guidelines. 75

Scenario analyses and multiple imputation

In addition to the two sets of intervention costs described above, the analyses were performed using both complete cases and imputed data. For the latter, data were imputed using the mice (multiple imputation by chained equations) function in R. 124 Five imputed data sets were generated using the predictive mean matching option. The imputation models included all of the covariates used in the systemfit models. As it was not possible to run systemfit models directly using imputation models, the systemfit models were run separately for each imputed data set and the results were pooled manually using Rubin’s rules. 124

Trial intervention costs

For reasons of practicality, participants in both the Walking Away and the Walking Away Plus arms attended the same structured education sessions. Therefore, in costing the interventions, it was assumed that the costs incurred in providing each arm were the same, except for the telephone and text message support costs, which were assigned only to the Walking Away Plus arm. The full details and unit costs used in the microcosting of the interventions are provided in Appendix 11, Tables 49 and 50.

Educators

The cost of providing an educator for 1 hour was £20.37 (including on-costs). The sessions lasted 4 hours, giving a per-session cost of £81.48. A total of 259 sessions were delivered by two educators and 123 sessions were delivered by one educator.

Delivery of the interventions was supported by a training and mentoring package, which comprised:

• Training to deliver –

  • the initial Walking Away structured education sessions (both intervention arms)

  • the Walking Away update sessions (both intervention arms)

  • the follow-up telephone calls (Walking Away Plus only).

• Mentoring to promote the fidelity of –

  • the Walking Away sessions (both intervention arms)

  • the telephone calls (Walking Away Plus only).

Training was delivered by two members of NHS staff: one at band 7 and one at band 8a. Fidelity promotion was carried out by the band 8a staff member only. For clarity, ‘trainers’ refers to the staff members who delivered the training sessions and ‘educators’ refers to the individuals who attended the training sessions and delivered the structured education sessions in the interventions.

Walking Away delivery training

All educators (eight in Leicester and four in Cambridge) attended the Walking Away training. One session was run in each site. Nine educators (five in Leicester and four in Cambridge) attended the update training sessions. One update training session was run in each site. Details of the total costs of running these training sessions are provided in Table 13.

Telephone call delivery training

Initially, all educators were trained to deliver the telephone calls. Throughout the study, there was a high staff turnover for this part of the intervention, as educators were unwilling to make the telephone calls, and replacement staff were trained to take over. The total telephone call training costs were £4071, details of which are given in Table 13.

Walking Away fidelity training

There were two types of sessions to promote fidelity of the Walking Away education sessions: in year 1, educators were observed delivering the intervention; in years 2–4, they met with trainers for mentorship and support. The total Walking Away fidelity training cost was £3393, details of which are given in Table 14. Fidelity assessment of the telephone calls was conducted separately, and this cost £1803, details of which are provided in Table 13.

Support staff costs

Two types of support staff, administrators and project managers were employed to assist organisation of the interventions. The trial management team estimated that it took each administrator 15 minutes to book an appointment for a participant, and 30 minutes per education session to book the venue, prepare resources and process invoices. Administrators at Cambridge were paid university grade 2 salaries and at Leicester the staff were at NHS band 3; project managers were at university grade 6 at Cambridge and NHS band 6 at Leicester. For Cambridge staff, the mean salary (including on-costs) for each grade was used to estimate annual salary. Hourly rates for grade 2 staff were calculated based on a 35-hour working week. In the case of Leicester staff, the mean salary for a band 6 hospital-based professional staff member including overheads was taken from the Personal Social Services Research Unit (PSSRU). Equivalent figures for band 3 were not available, so the hourly rate for a band 3 staff member was estimated using the ratio of hourly pay to hourly total cost for a band 4 staff member. The costs of employing these staff are shown in Table 15.

Text message system

The total number of text messages sent in the trial was 96,252. Each text cost 3.5 pence to send. In addition, there was an annual licence fee of £118 for the system, giving a total cost of text messages of £3841.

Participant expenses

Participant travel expenses were reimbursed on request at both sites. Leicester did not reimburse mileage but did pay for participants’ taxis, whereas Cambridge did reimburse mileage but did not pay for taxis. The total cost of travel expenses across the two sites was £6454.19.

Administration consumables and teaching resources

Participants in both Cambridge and Leicester were sent appointment letters by post. The Leicester site also sent participants a map showing where to go for the appointment. Costs were associated with teaching resources (paper, pens and booklets) and sets of demonstration materials. In total, the cost of administration and consumables came to £11,163.

Venue costs

In addition to the cost of staff to deliver the sessions, costs were incurred for venue hire and providing refreshments. Some venues provided refreshments in house, whereas at other sessions the refreshments were provided by the PROPELS team. The total cost of booking venues came to £9924.

Additional costs

In addition, each participant received a pedometer that cost £10.50.

Real-world cost assumptions

Following communication with personnel involved in delivering Walking Away and related interventions in a real-world setting (specifically the DESMOND National Director, Bernie Stribling, who is familiar with the intervention landscape for all Walking Away sites in the UK and Ireland), a second set of costs was estimated to reflect the expected cost of delivering the interventions outside the trial. The differences between the two sets of costs are summarised in Table 16. It is assumed that, in the real-world setting, either Walking Away or Walking Away Plus would be commissioned.

Per-participant costs

The per-participant costs are summarised in Table 17. The total within-trial cost of delivering Walking Away was £286 per participant and the cost of delivering Walking Away Plus was £338 per participant. The real-world costs of delivering the interventions are estimated to be slightly lower than the within-trial costs (Walking Away £257, Walking Away Plus £322). There are two primary reasons for the real-world costs being so close to the within-trial costs of delivering Walking Away. First, the staff members who would be expected to deliver the interventions in the real world are more expensive than the staff used in the PROPELS RCT. Second, we expect that fewer sessions per year will be delivered in real-world settings, increasing the effective per-participant cost of training the educators.

Within-trial costs

Cost-effectiveness

The ICERs from the within-trial analysis for the real-world costs and complete-case data are given in Table 19. When one QALY is valued at £20,000, in line with NICE guidelines, the within-trial analysis would indicate that Walking Away is the most cost-effective option. These results are highly volatile and are sensitive to the assumptions made about missing data and the calculation of the costs. For example, when trial costs are used and the missing data are imputed, usual care would be the most cost-effective option when one QALY is valued at £20,000. It should be noted that all within-trial results are highly uncertain.

This uncertainty is illustrated on the cost-effectiveness plane in Figure 7. Interventions on the cost-effectiveness plane to the right of the dotted line are cost-effective at a willingness-to-pay threshold of £20,000, and those to the left are not. As the 95% CIs for cost and QALYs for both interventions cross the dotted line, it is not possible to be certain if either intervention is cost-effective at this threshold.

FIGURE 7.

Cost-effectiveness plane showing within-trial adjusted incremental costs and QALYs.

Similarly, the cost-effectiveness acceptability curve in Figure 8 demonstrates this uncertainty. The graph shows the probability that each of the three trial arms is cost-effective at varying values of a QALY. At a low willingness-to-pay threshold, usual care has the highest chance of being the most cost-effective intervention because it is the cheapest option. As the value of a QALY increases, the QALY gains in Walking Away and Walking Away Plus become more valuable and their relative chance of cost-effectiveness increases. Between the upper and lower limits of the NICE threshold, namely £20,000 and £30,000 per QALY, the three arms have a roughly even chance of being the most cost-effective option.

FIGURE 8.

Cost-effectiveness acceptability curve showing within-trial data.

Model population

The characteristics of the simulated population compared with the real PROPELS data at baseline are shown in Appendix 10, Table 48. Overall, there was good agreement between the simulated and the real data.

Step count trajectories

The figures below show the step count trajectories in the South Asian (Figure 9) and non-South Asian (Figure 10) populations in the deterministic model runs. The average per-day step count at baseline in the non-South Asian population was 7164 (in all three arms). At 12 months, both Walking Away and Walking Away Plus arms were given increases in step count to reflect the treatment effect of the interventions, resulting in average daily step counts of 7342 and 7563, respectively. By 48 months, the average daily step count had declined to 6981 and 7174, respectively.

FIGURE 9.

Step count trajectories in the South Asian population.

FIGURE 10.

Step count trajectories in the non-South Asian population.

In the Asian population, the baseline step count was 7042 steps per day. At 12 months, it rose to 7178 and 7399 steps per day in the Walking Away and Walking Away Plus arms, respectively. Over the next 3 years, to the end of the trial period at 48 months, step count declined in both groups, to 7075 and 7301 steps per day, respectively.

Beyond 4 years in both subpopulations, both Walking Away and Walking Away Plus individuals converged with their individual underlying trajectory, and then continued to decline in daily step count at a rate of 67 steps per day per year, as predicted by the ordinary least squares regression model. At an aggregate level, there is some noise to this effect as individuals die, and those with a low step count (and higher cardiovascular risk) are at higher risk of mortality. Note that average step count beyond 30 years is omitted from these graphs as the small numbers of individuals make the figures unstable.

HbA_1c and diabetes diagnoses

Table 20 shows the rates of diagnosis of T2D in the deterministic model runs in the non-South Asian and South Asian populations. In both subpopulations, the total number of within-trial diagnoses was largest in the Walking Away arm, followed by the usual care arm, followed by the Walking Away Plus arm. This was expected because our logistic regressions (see Appendix 12, Tables 51–68) produced odds ratios for rates of diabetes diagnosis comparing Walking Away with usual care of 1.55 (97.5% CI 0.52 to 4.63) at year 1 and 1.58 (97.5% CI 0.74 to 3.39) at year 4. Comparing Walking Away Plus with usual care, the odds ratios for the rates of diabetes diagnoses were 0.72 (97.5% CI 0.19 to 2.68) at year 1 and 1.25 (97.5% CI 0.57 to 2.74) at year 4. In all three arms, rates of diabetes diagnosis over a lifetime were higher in the South Asian population than the in the non-South Asian population.

As shown in Figures 11 and 12, the average HbA_1c level (%) was very similar across all three trial arms. The trajectory for the first 4 years (i.e. the data generated from analysis of the PROPELS trial) was not notably different from that for the subsequent years of the model (which was extrapolated using equations based on an analysis of the Whitehall data set125).

FIGURE 11.

The HbA_1c trajectory in the South Asian population.

FIGURE 12.

The HbA_1c trajectory in the non-South Asian population.

Deterministic results

In the deterministic analysis, mean parameter values were used and the population was fixed. The costs and QALYs for each subpopulation and the weighted population are given in Table 21. All results are presented per person. In both the South Asian and the non-South Asian subpopulations, Walking Away is both more expensive and less effective than usual care, that is it is dominated. In both subpopulations, Walking Away Plus is more expensive and slightly more effective than usual care, giving ICERs that are well in excess of the £20,000-per-QALY threshold.

Probabilistic sensitivity analysis

The PSA results for each subpopulation are given in Table 22. In both the South Asian and the non-South Asian subpopulations, Walking Away is both more expensive and less effective than usual care, that is it is dominated by usual care. In both subpopulations, Walking Away Plus is more expensive and slightly more effective than usual care, giving ICERs that are well in excess of the £20,000-per-QALY threshold.

TABLE 22 - Probabilistic sensitivity analysis results (South Asian, non-South Asian and weighted populations)

Population	Costs (£)	QALYs	Incremental costs (£)	Incremental QALYs	ICER (£)
South Asian population
Usual care	24,489	10.712	–	–	–
Walking Away	24,994	10.697	–	–	Dominated by usual care
Walking Away Plus	24,907	10.713	418	0.0014	306,971
Non-South Asian population
Usual care	22,494	9.245	–	–	–
Walking Away	22,910	9.235	–	–	Dominated by usual care
Walking Away Plus	22,838	9.252	344	0.0074	46,163
Weighted population
Usual care	22,598	9.323	–	–	–
Walking Away	23,018	9.312	–	–	Dominated by usual care
Walking Away Plus	22,945	9.330	347	0.00705	49,273

The cost-effectiveness planes (Figures 13–15) for both subpopulations and the weighted population show that the PSA mean ICERs for both Walking Away (dark blue crosses) and Walking Away Plus (light-blue crosses) are to the left of the dotted line that indicates a cost-effectiveness threshold of £20,000 per QALY; that is, they are not cost-effective at this threshold. However, the purple and light-blue clouds of points that indicate individual PSA runs show that, for some combinations of parameters, the results are to the right of the line, indicating that both or either of the interventions could be cost-effective at this threshold.

FIGURE 13.

Cost-effectiveness plane for the model results in the South Asian population.

FIGURE 14.

Cost-effectiveness plane for the model results in the non-South Asian population.

FIGURE 15.

Cost-effectiveness plane for the model results in the weighted population.

As shown by the cost-effectiveness acceptability curve in Figure 16, among the non-South Asian population, at a threshold of £20,000 per QALY there is a 52.0% chance that usual care is the most cost-effective option, a 9.8% chance that Walking Away is the most-cost-effective option and a 38.2% chance that Walking Away Plus is the most cost-effective option. At a threshold of £30,000 per QALY, the corresponding probabilities are 45.5%, 11.9% and 42.5%.

FIGURE 16.

Cost-effectiveness acceptability curve (non-South Asian population).

Among the Asian population (Figure 17), at a threshold of £20,000 per QALY there is a 61.8% chance that usual care is the most cost-effective option, an 11.7% chance that Walking Away is the most cost-effective option and a 26.4% chance that Walking Away Plus is the most cost-effective option. At a threshold of £30,000 per QALY, the corresponding probabilities are 54.6%, 14.0% and 31.4%, respectively.

FIGURE 17.

Cost-effectiveness acceptability curve (South Asian population).

Among the weighted population (Figure 18), there is a 55.6% chance that usual care is the most cost-effective option at £20,000 per QALY, a 9.5% chance that Walking Away is the most cost-effective and a 34.9% chance that Walking Away Plus is the most cost-effective. At a threshold of £30,000 per QALY, the corresponding probabilities are 48.6%, 11.4% and 40.0%, respectively.

FIGURE 18.

Cost-effectiveness acceptability curve (weighted population).

Clinical events

Table 23 shows the PSA average lifetime per-person incidence of T2D, CVD and T2D complications in each of the two subpopulations.

Intervention cost threshold analyses

Table 24 shows the results of the intervention cost threshold analysis. Walking Away Plus would need to be delivered at a cost of £116 per person to be cost-effective at a threshold of £20,000 per QALY. Because Walking Away is less effective than usual care, it would have to be less expensive than usual care by £370 to be cost-effective at £20,000 per QALY.

Value of information

The total per-person EVPI was found to be £279.56. Assuming that the maximum number of interventions that would be delivered per year was 100,000, and that the decision relevance horizon was 10 years, the total EVPI for the UK is £279,559,484. This can be interpreted as the total value to the UK of research to eliminate all uncertainty about the cost-effectiveness of Walking Away and Walking Away Plus.

The expected value of perfect parameter information (EVPPI) for the key groups of parameters taken from the PROPELS trial are presented in Table 25 and Figure 19. The sources of uncertainty that have the largest potential impact on the decision are the parameters required to predict HbA_1c and T2D diagnoses.

FIGURE 19.

The EVPPI of parameters gathered from PROPELS data.