Stage 2 methods

An adapted version of the following methods was published open access in Matthews et al.73 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: http://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.

The development of the HelpMeDoIt! website and app (stage 1) has been described in Chapter 2, as have the methods and findings. This chapter describes the methods for the feasibility RCT (stage 2). The study protocol can be found online (see Report Supplementary Material 1). Amendments to the original protocol are presented in Appendix 7. The following description of the methods adheres to the TIDieR (Template for Intervention Description and Replication) guidelines.89

Study design and setting

HelpMeDoIt! was a feasibility RCT conducted with adults with obesity living in Greater Glasgow and Clyde Health Board area of Scotland (April 2016–February 2018).

Participants

Participants were eligible for the trial if they met all of the following inclusion criteria and none of the following exclusion criteria. All queries about eligibility were directed to the HelpMeDoIt! trial manager, and discussed among the study team as needed, before randomisation took place.

The inclusion criteria were:

• being an adult aged 18–70 years
• having a BMI of ≥ 30 kg/m²
• trying to lose weight
• having access to a smartphone and the internet.

The exclusion criteria were:

• having a terminal illness
• having had previous bariatric surgery
• living with dementia
• being pregnant
• having poor competence in English (resulting in inability to complete study materials)
• having contraindications to physical activity
• having previously been a participant in stage 1 intervention development
• already being a nominated helper in the trial.

Potential participants were briefly screened by the trial manager over the telephone or by e-mail. Any individual who did not meet the initial criteria for age and BMI (e.g. they had a BMI of 27 kg/m²) was thanked for their interest and directed to other publicly available, evidence-based resources for weight loss. The contact details of individuals who met these initial criteria were passed to a field worker, who then liaised with the participant for the remainder of the eligibility checking process and follow-on steps.

Contraindications to physical activity were assessed using an adapted Physical Activity Readiness Questionnaire.90 Anyone with a medical condition or taking medication or who thought they may have a contraindication to physical activity was advised to check with their GP before commencing any physical activity. We asked women of childbearing age to let the study team know if they became pregnant at any point during the trial. Women already recruited were not excluded from the study if they became pregnant, as the intervention could still help them make healthy lifestyle choices. They were given a leaflet on diet and safe physical activity during pregnancy. In a future trial, the analysis would be intention to treat and we would not exclude women who become pregnant.

Sample size

This was a feasibility study, and the main focus was to assess the acceptability of the intervention and the feasibility of the evaluation methods and to estimate parameters for a larger study. We intended to recruit 120 participants. As we were most interested in the acceptability of the intervention, we recruited using a 2 : 1 ratio, with 80 participants in the intervention group and 40 participants in the control group. We expected a dropout rate of 30%. This final sample size of 84 for analysis was not powered to detect differences between groups for the proposed effectiveness outcomes (BMI, physical activity and diet) but it allowed an estimation of any feasibility proportion (e.g. proportions retained/found the study acceptable/provided outcome data) across the whole sample with a 95% confidence interval (CI) of plus or minus 11 percentage points. This would also allow for the estimation of the mean of a continuous outcome (such as BMI) in the intervention arm with a 95% CI of 0.262 of a standard deviation.

Recruitment and retention

A multipoint recruitment strategy was employed to target a broad range of participants (e.g. age, gender, socioeconomic status). This primarily involved recruiting via (1) primary care, (2) online sources and (3) community sources.

• Primary care recruitment: we collaborated with the Scottish Primary Care Research Network (SPCRN) to recruit participants. The SPCRN staff liaised with GP practices and searched patient databases for potentially eligible participants based on the inclusion/exclusion criteria. Practice staff excluded vulnerable patients based on other known information, for example having dementia or receiving social work input for drug dependency. The SPCRN posted the recruitment pack to the agreed list of patients. The recruitment pack included (1) a GP cover letter informing patients that they had been identified as eligible for this study, (2) a study information sheet, (3) a contact details form and (4) a prepaid envelope. After reading the information, individuals who were interested in taking part were asked to complete the contact details form and return it to the study team using the prepaid envelope.
• Online recruitment: regular adverts were placed on the Glasgow hub of Gumtree (Gumtree.com Limited, London, UK; www.gumtree.com) (a free online community advertising website). The advert was brief, highlighting the key points of the study and the eligibility criteria, and directing interested individuals to contact the trial manager for full details. We also established a HelpMeDoIt! Facebook page and Twitter account for posting up-to-date information and we advertised on online forums, for example new-mother groups and local weight loss groups. People were encouraged to express their interest to the study team.
• Community recruitment: we advertised in the local press and through slimming clubs, weight management clinics and exercise-on-referral services, and by placing study posters in multiple community locations. In addition to the strategies already mentioned, we had planned to target men through barbershops and football grounds, as recruiting men into weight loss trials is known to be challenging, but owing to resource restrictions we were unable do this.91

All individuals who expressed an interest in the study received a participant information sheet (see Report Supplementary Material 2) and had at least 1 week to consider taking part. After individuals were screened for initial eligibility by the trial manager, a trained field worker contacted them by telephone to check their full eligibility and to arrange a face-to-face appointment for baseline data collection. Prior to any data being collected, participants had the opportunity to ask questions before signing an informed consent form (see Report Supplementary Material 3). Field workers met with participants at a place of their choice, which included their home (in which case our lone-working policy was followed) or a room at our research unit. All field workers were trained in study procedures, had undertaken recent training in good clinical practice,92 and held NHS Research Passports allowing them to liaise with participants who were recruited via primary care.

A key aspect of the HelpMeDoIt! intervention was the recruitment of ‘helpers’. A helper was an individual whom participants nominated as someone they thought could support them with their weight loss goals. If an individual agreed to be a helper, they were directed to a web link via which they were able to access an information sheet about the study (see Report Supplementary Material 4) and provide informed consent using an online form (see Report Supplementary Material 5). This indicated their consent to be a helper and for the study team to keep their contact details (for the purposes described below), and to signify whether they were willing to be contacted with regard to completing an interview at a later date (for which they would be given separate information and give separate consent). Contact details for the study team were available via the web link and information sheet so that helpers could call or e-mail if they had any queries or needed further information. After providing consent to be a helper, they were asked to enter brief demographic and contact details. We asked for their postal address, mobile phone number and e-mail address (where available); this information was required so that we could give them updates on participants’ progress and send them reminders as well as voucher payment for interviews. The demographic data were used to describe the characteristics of the helpers and to sample them for the qualitative interviews according to gender, age and relationship with participant.

We developed a retention strategy to optimise the number of participants with data at follow-up. This included offering flexibility regarding where data collection took place; sending newsletters and birthday cards; obtaining mobile numbers and alternative contact details; offering a reduced data collection (‘minimum data set’) to participants who were reluctant to complete the full follow-up; and giving £20 voucher payments as a thank you for each point of data collection.

Randomisation

We were most interested in exploring the feasibility of the intervention and so randomised participants in a 2 : 1 ratio into intervention and control groups. Participants were allocated using a mixed randomisation/minimisation algorithm to ensure balance with respect to gender and BMI (< 40 and ≥ 40 kg/m²). In blocks of 15 participants, 12 were assigned according to the minimisation algorithm (designed to maintain as close to a 2 : 1 allocation ratio within strata defined by each minimisation factor) and three were allocated (in a 2 : 1 ratio) at random. The minimisation/randomisation schedule (the order in which participants were allocated by minimisation or randomisation) was prepared by a statistician in the Robertson Centre for Biostatistics (University of Glasgow) using the method of randomised permuted blocks. This statistician did not carry out the final analysis for the study. Participants were remotely allocated by field workers using an automated telephone service operational 24 hours per day. Allocation was undertaken in the presence of the participant after they had completed the relevant baseline data collection procedures. On group allocation, participants were allocated a unique randomisation number.

The intervention group

Participants in the intervention group received access to the HelpMeDoIt! smartphone app and website for 12 months. The app and website content has been described in detail in Chapter 2. Participants were advised that they could continue to access other sources of lifestyle change/support external to the app, such as attending weight loss groups and fitness classes.

Exploring the feasibility of participants also acting as helpers

The HelpMeDoIt! study aimed to explore how the intervention might work in a real-world setting. Because the intervention had a social support focus, it was possible that two or more friends/relatives would wish to lose weight together and support each other. It was unknown if participants would choose to act as helpers, but it was important to allow for and explore this for several reasons, including the following: (1) this approach may have potential benefits for participants as a result of increased support and motivation; (2) participants who also act as helpers might have more beneficial outcomes than participants who do not act as helpers; and (3) it would allow the identification of a spillover effect in line with the diffusion of innovation theory.93 Our study therefore allowed participants in the intervention group to act as helpers for a friend or relative. Their helper would then also have access to the participant aspect of the intervention (i.e. so that they could be both participants and helpers to each other). Participants in the control group could not act as helpers as this would enable them to view the intervention content. In addition, individuals who were already helpers for someone else could not sign up as a participant as they might have been randomised to the control group and therefore would have already seen the intervention content.

The control group

The control group received leaflets about the health benefits associated with healthy eating and physical activity behaviour change (see Report Supplementary Material 6 and 7). They did not receive any prompted advice on social support or personalised content. They were not restricted in any way regarding their involvement in other weight loss activities, for example joining a slimming club. We collected data on this as part of the process evaluation. Control participants were offered access to the website and app after follow-up was complete at 12 months.

Progression criteria from feasibility to full trial

The feasibility of the evaluation methods, the feasibility and acceptability of the intervention, and the intervention’s potential to be further developed and delivered in a fully powered RCT were the key questions addressed by this study. Feasibility was assessed using the progression criteria outlined in Table 8. These criteria were finalised within the Trial Management Group and approved by the Trial Steering Committee at the end of stage 1, prior to beginning stage 2. There was substantial debate around criterion 5. On the one hand, current research on app use indicates that around 21% of users engage with an app only once.94 However, if only a minority of participants engage with the app and it is effective for them, then it may have a cost-effective and worthwhile impact on public health when rolled out across the population. On the other hand, we wanted to see a reasonable proportion of participants engaging with the app sufficiently to set goals and identify helpers, even if the subsequent interactions with their helpers were not made via the app. The cut-off point of visiting the app at least twice was chosen as we felt that, with this level of engagement, participants could have exposure to the key intervention components, in particular the idea of goal-setting, monitoring and recruiting helpers to aid in weight loss goals, and that this level of exposure may be enough to kick start the process of behaviour change.95

The results of the study are reported in Chapters 4–6 and were used to assess whether or not the progression criteria had been achieved. Final assessment of the study results against the progression criteria was completed by the Trial Steering Committee.

Feasibility measures and exploratory outcomes

All feasibility measures and exploratory outcomes are outlined in Table 9. These were collected using a case report form (CRF) and questionnaire completed face to face with a study researcher in the participant’s home or an interview room in the university (with the exception of one measure collected over the telephone; see Table 9). Data were collected using hard-copy documentation (see Report Supplementary Material 8 and 9).

Feasibility measures

Feasibility and acceptability of the HelpMeDoIt! intervention and study methods was the key focus of this trial. These relate to the overall progression criteria (see Table 8) and are summarised in Table 9. The majority of these measures will be discussed further in the qualitative and process evaluation methods sections below.

Exploratory primary outcomes

We explored the feasibility of assessing multiple outcome measures, including potential primary outcomes for a future trial. We collected data in relation to three outcomes that could be used as the primary outcome: BMI, physical activity and diet (see Table 9). We explored which of these was the most feasible by assessing acceptability and data completeness. BMI was expected to be the most appropriate measure because it is objective, cheap to measure and more accurately measured than diet and physical activity, and has higher response rates. As measuring diet110 and physical activity111 in community-based trials is challenging, we assessed two ways of measuring these outcomes. Each outcome was measured at baseline and 12 months. Twelve months was chosen as the primary outcome point for the feasibility trial, which would be the same for any future effectiveness trial. This longer-term follow-up is important as, although there may be some benefits of short-term weight loss, weight loss maintenance is key to longer-term health benefits and cost savings to the NHS and society.112

Body mass index (kg/m²) was calculated from measures of height and weight. Height was measured using a Seca Leicester Height Measuring Stadiometer (Seca GmbH & Co., Hamburg, Germany), with the participant facing forward, wearing no shoes and with their head in the Frankfort plane (parallel to the floor). Height was recorded once, in centimetres, to one decimal point. Weight was measured, without shoes on, using Tanita HD 352 High Capacity Low Profile Electronic Weighing Scales (Tanita Europe BV, Amsterdam, the Netherlands). Scales were calibrated before first use. Weight was recorded once, in kilograms, to one decimal point.

Physical activity was measured using ActiGraph GT3X (ActiGraph, LLC, Pensacola, FL, USA) accelerometers, objective activity monitors that measure the duration, intensity and frequency of physical activity. Participants were asked to wear the accelerometer on their right hip for 7 days during waking hours (except when swimming or bathing). Participants received their accelerometer during a face-to-face visit; this enabled the researcher to demonstrate the correct placement of the device. Data were collected in 1-second epochs at a sample rate of 100 Hz, and converted to 15-second epochs for analysis using ActiLife 6 software (ActiGraph, LLC, Pensacola, FL, USA).113 Non-wear time was identified by > 60 minutes of continuous 0 counts and removed before analysis. Data were included for analysis when the accelerometer had been worn for a minimum of 4 days, and with a minimum wear time of 10 hours per day. Freedson cut-off points98 were used to determine the amount of time spent sedentary and in moderate to vigorous physical activity. The 7-day Physical Activity Recall questionnaire114 was used to measure physical activity by self-report. Participants, guided by the researcher, reported their activity over the previous 7 days in relation to moderate, hard and very hard exercise. This measure has been validated for use in adult populations,99 and researchers adhered to the protocol published by Sallis et al.114

Diet was measured with the Dietary Instrument for Nutrition Education (DINE) questionnaire,96 a validated seven-item questionnaire that explores the frequency of consumption of different food types, for example bread and rolls, cereals and meats. Field workers asked participants to report the frequency with which they ate specific foods. The frequencies were scored using DINE guidelines96 to produce an overall score for fat and fibre. Diet was also measured using repeat 24-hour dietary recall97 collected by a researcher over the telephone on 4 separate days within a 10-day period (including one weekend day). Participants self-reported their food intake, prompted by the researcher, for the previous 24 hours. Researchers were guided by photographic and textual examples of portion sizes, which they could use as prompts over the telephone. Participants’ food intake was entered into weighed intake software programme (WISP) dietary analysis software (version 4.0; Tinuviel Software, www.tinuvielsoftware.co.uk/wisp4.htm; accessed 1 May 2018)115 and analysed for energy intake, macronutrients and fibre. The 24-hour dietary recall measure was found to be a poor method of data collection at baseline and, therefore, it was not used at follow-up. Full details of this decision are provided in Chapter 6.

Secondary outcomes

Waist circumference was measured using a 2-metre flexible tape measure with a buckle around the mid-point between the iliac crest and the inferior margin of the lower rib. Hip circumference was measured around the widest point of the buttocks. Measurements were recorded twice in centimetres to one decimal point (e.g. 95.2 cm). A third measure was taken if the difference was > 0.5 cm.

Health-related quality of life was measured using the five-item EuroQol-5 Dimensions, three-level version (EQ-5D-3L), questionnaire and quality-of-life thermometer.100 This measure is used frequently in health-related research to explore five dimensions: mobility, self-care, usual activities, pain and anxiety/depression. An additional measure of capability well-being was carried out using the ICEpop CAPability measure for Adults (ICECAP-A)101 scale. This is a new scale which, compared with the EQ-5D-3L, explores less clinically related changes in quality of life over four dimensions: feeling settled and secure, being independent, achievement and progress, and enjoyment and pleasure. This was considered an appropriate measure in our population owing to the potentially large range of participant characteristics. Mental health was measured using the General Health Questionnaire – 12 items (GHQ-12),102 a validated and frequently used 12-item self-report questionnaire. We gathered data on NHS resource use and participant-borne costs using a specially designed resource questionnaire. These data helped us establish key cost drivers of the intervention.

Questionnaires assessing potential mediators of change were used to explore their feasibility in a future trial. These included the Exercise and Eating Habits Social Support Scales (social support),103 the Weight104 and Exercise Efficacy Lifestyle Scales (self-efficacy)105,106 and the Treatment Self-Regulation Questionnaire (motivation).107

At 12 months we administered the USE questionnaire86 to assess the usefulness, ease of use, ease of learning and satisfaction with the app and website. We also asked participants to complete the Heaviness of Smoking Index (HSI)108 and Alcohol Use Disorders Identification Test (AUDIT-C)109 questionnaires at 12 months, using this opportunity to assess the feasibility of additional questionnaires for data collection. These may be helpful in identifying other potential lifestyle changes made by participants in a future trial related to ‘spillover’ effects of the intervention.

Process evaluation

The process evaluation explored, in detail, the feasibility and acceptability of both the intervention and the study design. The process evaluation was conducted following the MRC guidelines for process evaluation of complex interventions116 and examined the following elements: (1) intervention-related findings on context, fidelity, exposure, reach, programme theory and logic model; and (2) study-related findings on recruitment, retention and contamination. Report Supplementary Material 10 illustrates the different domains of the process evaluation, as well as examples of key questions and the methods for addressing those questions. In brief, some quantitative data informed the process evaluation (e.g. intervention use statistics); the remaining process data were gathered from qualitative interviews with participants and helpers.

Quantitative process data

Web/app analytics

App and website use data were collected for both helpers and participants to assess engagement with the intervention. Key use data included the number of logins to the website and app by the helper and participant; the duration of logins; the average sessions per user; which web pages were viewed and how often; the number of goals set using the app; the frequency of weight updates; the number of helpers nominated; contacts between helpers and participants via the app; the number of views of ‘progress charts’ by participant and helper; and patterns of use over time. Demographic data on the helpers, including their relationship to the participant, were also included in the process evaluation.

Qualitative process data

Qualitative interviews with participants

We planned to interview up to 30 participants at 6 months and up to 20 participants at 12 months (dependent on data saturation). Participants were purposively sampled for a range of characteristics (e.g. level of app/website use, age, gender). We also specifically interviewed participants who did not take up the intervention to explore their reasons for this. Semistructured interview guides were used at both time points to explore participants’ insights into the acceptability of the outcome measures, the acceptability and usability of the app and website, patterns of use, the impact of the intervention on behaviour, the support received from helpers, and the barriers to use (see Report Supplementary Material 11). Interview schedules were guided by the HelpMeDoIt! programme theory and helped us explore potential mediators of change. All participants at baseline were guided to draw a sociogram (a diagrammatic representation of their social network). This was used as a prompt with interview participants to explore the potential sources of social support they had in their social network. This also informed a social network analysis, which was not part of the initial funding bid or timeline, and will be published at a later date.

Qualitative interviews with helpers

At 6 months we planned to interview up to 20 helpers, purposively sampled for a range of characteristics (e.g. level of app/website use, age, gender). A semistructured interview guide was used to explore helper insights related to acceptability; guidance provided for being a helper; types of support they provided to their friend; challenges of supporting their friend and/or using the app and website; and changes in their own health behaviour as a result of being a helper (see Report Supplementary Material 12).

All interviews were completed by trained researchers over the telephone or face to face in a meeting room in the university. Interviews were audio-recorded and transcribed verbatim. A separate informed consent process was undertaken for the qualitative interviews, whereby participants and helpers were sent a separate information sheet detailing what was expected of them during the interview. They were also sent a copy of the consent form, which they were asked to read. Prior to beginning the interview, the researcher checked if the participant had any queries and gained (1) verbal consent if over the telephone (audio-recorded) or (2) written consent if face to face.

Economic evaluation

The economic evaluation aimed to (1) identify and measure the key cost drivers of the intervention and control arms, including health, social care and societal resource use; and (2) identify suitable outcome measures and summarise mean costs to inform the design of an economic evaluation in a definitive trial. The following costs were collected at baseline and 12-month follow-up.

Intervention costs

Intervention costs were identified and measured for the development of the intervention. The cost of the design of the mobile application and maintenance for the purposes of the feasibility study are reported.

Health, social care and personal resource use

All resource use data were collected from participants with a specially designed resource use questionnaire (see Report Supplementary Material 8 and 9). This collected information on (1) primary care services provided in the NHS (e.g. GP visits, practical nurse, dietitian); (2) secondary care services provided in the NHS [e.g. accident and emergency (A&E) attendances, hospital stay, visit by consultant]; (3) and personal costs (e.g. household income spent on different food and drinks items, including groceries, alcohol, takeaway food, meals out and lifestyle activities). Medication use at baseline and follow-up was also recorded. Mean group costs were calculated by attaching the unit costs to frequency of resource use per group participant. Medians and ranges for resource use quantities and costs are reported.

Reference costs

Reference costs for items of resource use were identified from readily available UK reference sources including the Personal Social Services Research Unit reference cost book.117

Health economics outcomes

The economic analysis assessed the feasibility of using the EQ-5D-3L instrument100 and the ICECAP-A101 instrument as a means of capturing any short-term effects on health-related quality of life or capability well-being. This involved exploring data completeness and response rates. Participant responses were converted to a utility score using the value set elicited from UK general population. The EQ-5D-3L is the preferred measure of health-related quality of life in adults by NICE technology appraisal.118 The ICECAP-A is a complementary measure of capability well-being.

The economic evaluation adhered to guidelines for good economic evaluation practice as outlined by Gold et al.119 Specific guidance was also sought from the NICE economic evaluation public health reference case120 as it was anticipated that this intervention was likely to have an impact on costs and outcomes beyond the NHS and Personal Social Services and thus require a broader public sector evaluative perspective.

Quantitative analysis

A statistical analysis plan (see Report Supplementary Material 13) was drafted by the Robertson Centre for Biostatistics, University of Glasgow, and reviewed by the Trial Management Group and Trial Steering Committee during the trial. It was finalised and approved by the principal investigator and trial statistician before the analyses commenced. All quantitative analyses were carried out using R1 for Microsoft Windows v3.2.2 or higher (Microsoft Corporation, Redmond, WA, USA).121 The mice122 and interplot123 packages were also used. A summary of the analyses is as follows.

Baseline characteristics

These were summarised overall and by randomised group. Participant characteristics were summarised in relation to sociodemographic, lifestyle, occupational, health status and quality-of-life variables.

Feasibility measures

These were the primary focus of the analysis. Study-related data, including recruitment, randomisation and retention at 12-month follow-up, were reported overall and by randomised group, with 95% CIs. The association between baseline factors and follow-up was assessed using logistic regression, with follow-up (yes/no) as the response variable. Intervention-related data, including use of the intervention, was summarised for the intervention group, overall and in relation to selected baseline characteristics. The availability and utility of data relating to use of the app and website was explored, and a range of summary measures was presented in the final statistical outputs.

Exploratory outcomes

These were summarised overall and by randomised group, and compared using linear regression models, with randomised group, the baseline measurement of the outcome, age and gender as predictor variables. Regression models also adjusted for the minimisation factors. The residuals from each regression model were assessed for normality. When necessary, the outcome measure (at follow-up and at baseline) was transformed to improve model fit. All analyses were conducted under intention-to-treat principles and complete-case analysis used, unless > 20% of cases were lost as a result of missing data, in which case multiple imputation was also performed. These analyses were exploratory and underpowered, so no formal hypothesis testing was performed. P-values are presented for descriptive purposes as a guide to the interpretation of the magnitude of reported associations. Effect sizes were reported in line with CONSORT guidelines for reporting feasibility and pilot studies.124

Process evaluation measures

These were descriptively analysed to summarise use of the app and website. A per-protocol analysis was conducted using simple proxies for adherence (e.g. website login) to explore the treatment effect associated with adherence.

Health economics cost data

These were analysed as follows. Resource use data were summarised and described using mean values and variation around these estimates. Key fixed and variable costs of developing the intervention were described and summarised. EQ-5D-3L and ICECAP-A outcome data were reported by within-attribute response rates, mean values and associated variance. Within-trial economic analyses were performed using Stata^® 12.0 (StataCorp, College Station, TX, USA).

Qualitative analysis

Qualitative data analysis explored the feasibility and acceptability of the intervention, the extent to which participants and helpers engaged with it, the perceptions of how the intervention influenced behaviour, the value of helpers’ support, and contextual factors. Qualitative data were analysed by two researchers who independently coded using an adapted version of Braun and Clark’s125 approach to thematic analysis. Twenty per cent of the interviews were double-coded by two researchers, who resolved disagreements by discussion. The resultant coding framework was discussed between the researchers and also within the larger study team to finalise the themes and subthemes. The results of the qualitative analyses were combined with intervention use data to explore and refine the HelpMeDoIt! programme theory. The aim of analysing the programme theory in this way was to better understand the mechanisms and key contextual factors to consider when refining the intervention and evaluation design. A refined version of the programme theory and updated logic model can be found in Chapter 6. Qualitative analyses were supported by NVivo 10 (QSR International, Warrington, UK).

Withdrawal of participants

Participants had the right to withdraw consent to participate in any aspect of the HelpMeDoIt! study at any time. Participants had three options for withdrawing from the study: (1) withdrawal from the study intervention, (2) withdrawal from study follow-up or (3) withdrawal from the entire study and removal of their data from analysis. A participant withdrawal form was completed by the participant, field worker or trial manager, noting the withdrawal option and also the reason for withdrawal (if given).

Data management

A data management plan was developed to ensure safe and accurate data management. All study data were gathered by trained researchers using hard-copy paper questionnaires and stored in a locked filing cabinet in our secure research unit. Hard-copy data were anonymised using participant study ID. Data were entered by approved study personnel into a secure online database using a unique participant ID, so that study personnel remained blinded to group allocation. Appropriate elements of the database had automated error-checking facilities to ensure that only valid data were entered. We performed single data entry, 10% of which was cross-checked by double entry. Full double data entry would have been undertaken had a > 5% error rate been detected. Our error rate was 2% and so additional data entry verification was not performed. Audio-recordings were stored on a password-protected server. Files were then shared for transcription via a secure password-protected online cloud, where files were available for only 24 hours. Identifiable information was removed from the transcripts of the qualitative data.

All data will be kept for 10 years in line with University of Glasgow Research Governance Framework Regulations for clinical research. These data will be stored confidentially on password-protected servers. The final data set was accessed only by approved members of staff from the research team and Robertson Centre for Biostatistics, University of Glasgow. Approved members of the software company, who had signed a data protection agreement, had access to limited participant contact details to enable them to effectively manage software errors. The low-risk nature of this study meant that a Data Monitoring and Ethics Committee (DMEC) was not required. Our Trial Steering Committee covered the functions of the DMEC, particularly in relation to ethical issues, patient safety and continuation of the trial.

Assessment of harms

The intervention was low risk to participants. Participants were advised to discuss any health concerns with their GP. We encouraged field workers, participants and helpers to report negative outcomes or experiences to the study team. The issue of ‘harm’ was also explored in the interviews. We developed a standardised operating procedure for dealing with adverse events should these be reported.

Ethics approval

Ethics approval for stage 1 (see Chapter 2) was granted by the University of Glasgow Medical, Veterinary and Life Sciences College Ethics Committee (reference 200140108). Ethics approval for stage 2 was granted by the NHS West of Scotland Research Ethics Committee (reference 15/WS/0288) (see Report Supplementary Material 14). Research governance approval was given by NHS Greater Glasgow and Clyde Health Board.

Patient and public involvement

Patient and public involvement (PPI) was central to all aspects of the HelpMeDoIt! study, including the design and management of the research and the development of participant information resources, as well as contributing to the reporting and dissemination of the research findings.

Patient and public involvement informed the planning stage and preparation of the grant application. Patient representatives who had been involved in other behaviour change trials provided insights related to feasibility, outcomes, content and target population. They reported that this type of intervention had the potential to overcome some issues faced in other studies, for example cost and time to attend intervention groups and other family members or friends undermining behaviour change efforts. They felt that some questionnaires used to collect physical activity and diet data were burdensome and that testing out different measures in this study would be useful. They also felt that it would be important to explore engagement and motivation in the longer term.

Once the study commenced, one PPI member, Selin Campbell, was recruited to join our Trial Management Group and attended monthly meetings. Selin was an active contributor and provided valuable input into the ongoing management of the study and the development of participant information resources. She also helped with piloting the interview schedule, and provided ongoing feedback related to the intervention development and delivery, evaluation methods and interpretation of data. Selin also provided feedback on the interim report of the stage 1 findings and helped the Trial Management Group finalise the HelpMeDoIt! dissemination strategy.

We also had one active PPI member, Kenneth Wallace, on our Trial Steering Committee. Kenneth helped finalise and approve the progression criteria and contributed to assessment of both sets of criteria, namely from stage 1 to stage 2, and from stage 2 to a full trial. He was involved in the ongoing governance and oversight of the study, as well as providing feedback on the interim report of stage 1 findings. He also provided helpful recommendations on how to improve the HelpMeDoIt! app and website for a future trial.

Presentation of the findings

Although feasibility and acceptability in relation to the progression criteria was the focus of the feasibility trial, the remainder of the report presents the findings, beginning with the standard quantitative outcomes and economic analysis results (see Chapter 4), followed by the qualitative results (see Chapter 5), and the process evaluation results (see Chapter 6). In the final discussion chapter (see Chapter 7), the findings are integrated and considered in relation to the overall progression criteria and the research questions.