Phase II: feasibility study

Objectives

The objectives of the feasibility phase, as stated in Chapter 2, were to:

• assess the proportion of Pakistani and Bangladeshi families and the proportion of all families completing the adapted programme
• assess the acceptability of the programme to Pakistani and Bangladeshi families and families from other ethnic groups
• assess the feasibility of delivery of the adapted programme
• assess the feasibility of participant recruitment, randomisation and follow-up
• assess the feasibility of the collection of cost data from both a health and societal perspective to inform a future trial evaluating intervention clinical effectiveness and cost-effectiveness
• collect data on recruitment, attrition and other relevant measures to inform the parameters of any future trial.

Design

The feasibility study was designed as a small-scale, two-arm cluster RCT, with the weight management programmes as the clusters. This enabled an estimation of the proportion of families completing the adapted intervention programme and the standard programme, an evaluation of the feasibility of intervention programme delivery and processes to be used in a future clinical effectiveness and cost-effectiveness trial and the collection of key data to inform a future trial. All programmes that were planned to be delivered across Birmingham within the study period (September 2015 to April 2016, n = 24) were randomised to be delivered as either the adapted First Steps programme (intervention) or the standard First Steps programme (comparator). Families who were referred to the weight management service in this time period were invited to attend the programme that was most convenient for them, regardless of whether or not they consented to participate in the CHANGE study. Therefore, both intervention and comparator programmes contained a mix of CHANGE study participants and non-CHANGE study participants. The families participating in the children’s weight management programmes within the study period were not aware of whether the programme was the standard or the adapted intervention.

Community researchers

Community researchers (individuals from local Pakistani and Bangladeshi communities with relevant language skills) were again recruited to assist with the feasibility study. They worked alongside the core research team and undertook informed consent procedures, collection of outcome data and evaluation interviews and also provided language support when required.

Researcher training

All community researchers and the CHANGE study research team completed the NIHR e-learning courses on good clinical practice and informed consent in paediatric research. In addition, they completed a 1-day training course on child protection (‘Child Protection Level 2 Recognition & Referral’ run by the BCHCT). Community researchers received two bespoke training sessions on data collection procedures, run by the study research team. The community researcher undertaking interviews (Meanaz Akhtar, of Pakistani heritage) had prior qualitative research experience and received additional training from the research team.

Methods

Study arm allocation

Randomisation of the weight management programmes was conducted by a member of Birmingham Primary Care Clinical Research & Trials Unit (Andrea Roalfe) before the start of the feasibility study to enable service providers to plan delivery of the programmes within the study period. The 24 programmes planned in the study period were randomised to intervention and comparator arms using a 2 : 1 ratio, so that 16 programmes were delivered as the adapted programme and eight were delivered as the standard programme. This randomisation ratio ensured that there was a sufficient number of Pakistani and Bangladeshi families in the intervention arm to enable calculation of the primary outcome of completion (see Sample size). The randomisation was conducted in Stata^® version 13 (StataCorp LP, College Station, TX, USA). The 24 programmes were planned to be delivered over four school half-terms (six programmes per school half-term), and so randomisation was stratified by school half-term (i.e. for each half-term the six planned programmes were randomised with a 2 : 1 ratio). The name ‘First Steps’ was retained for both the standard and the adapted programmes, so that participants remained unaware of which study arm they were in. Programme allocation was communicated to the BCHCT service providers but concealed from the research team who were, therefore, unaware of the allocations at the time of baseline data collection.

Participant recruitment

Families in Birmingham with a child aged 4–11 years who had excess weight (defined as > 91st centile for BMI on the 1990 UK growth reference charts44) and who were referred to the existing children’s weight management service within the study recruitment period were eligible to participate. An additional eligibility criterion was the ability of the child to attend and participate in a group setting. We aimed to recruit 80 children and their families to participate in data collection within the study period (approximately 20% of all families referred to the children’s weight management service). As the intervention was adapted primarily to be more suited to Pakistani and Bangladeshi families, we prioritised recruitment of these families. We aimed for Pakistani and Bangladeshi families to comprise 60% of the study sample so that the success of the programme adaptation could be explored from the perspective of these families.

For families referred to the weight management service, the BCHCT service providers identified the First Steps programme that was closest to their home postcode and wrote to the families to confirm that they had been booked onto a programme. A CHANGE study invitation letter and parent and child information sheets were included with this booking confirmation. A cover letter, translated into Urdu and Bengali, was also included with the study information, which explained that families were being invited to take part in a research study relating to children’s weight management and that they would shortly receive a telephone call in which the study would be explained to them in their preferred language.

Within 3 days of receipt of the booking letter, BCHCT staff conducted follow-up telephone calls to check each family’s intention of attending a programme and to ask if they would be happy for their contact details to be forwarded to the CHANGE study research team. Parents who did not speak English were contacted by telephone in their preferred language. The contact details of those who verbally agreed were then passed to the CHANGE researchers.

The CHANGE research team contacted the families by telephone to ask if they had any questions and if they were happy to take part in the study. If they agreed to participate, an appointment at their home was arranged to obtain informed consent to participate and to undertake baseline assessments. For parents who did not speak English, the telephone calls and measurement appointments were conducted by a community researcher with the appropriate language skills. If required, participant information sheets were verbally translated by the community researchers before consent was obtained.

Consent

At the initial home visit, parental consent was obtained for all children taking part in the CHANGE study. Additional consent and assent were obtained for family members who were present at the home visit. Consent to conduct height, weight and body fat percentage measures was requested from any family members aged ≥ 16 years. Parental consent was requested for family members aged < 16 years. Written assent was obtained from children aged ≥ 8 years and verbal assent was obtained from younger children.

Follow-up

After the initial visit to obtain consent and baseline data (time point 0; T0), study participants received two further home visits to collect outcome data: within 1 month of the end of the programme (time point 1; T1) and 6 months after the end of the programme (time point 2; T2). Participants were telephoned to arrange the follow-up appointments and were then sent a letter of confirmation. They also received a reminder telephone call or text on the day of the appointment. All participants who provided data at T1 were given a shopping voucher worth £10. A flow diagram of the process of participant recruitment and assessment during the study is shown in Figure 6.

FIGURE 6 Flow diagram of feasibility study participant recruitment and assessment. a, Language support available. (Figure not reproduced here; see the original article.)

Safeguarding

If, during the study, the researchers had any concerns regarding the safety or well-being of a child or family member, they were required to document these on a safeguarding incident report form and submit these to the BCHCT. The BCHCT staff then followed their standard safeguarding procedures.

Intervention

The adapted intervention programme was delivered over 6 weeks and is described in detail in Chapter 1 (see Table 3 and Report Supplementary Material 1). Parents and children attended all sessions. Owing to school term time constraints, there were only 5 weeks available for the delivery of some of the programmes. In these instances, session plans for weeks 4 and 5 of the programme were combined to be delivered in one session. Details of how they were combined are in the facilitator’s manual in Report Supplementary Material 1. The study intervention period ran over four school half-terms. Four intervention programmes were run during each school half-term (16 programmes delivered in the intervention arm in total).

Facilitator training

Two intervention facilitators and a third assistant facilitator from the BCHCT children’s weight management service were trained to deliver the adapted First Steps programme. Two training sessions (of 2.5 hours and 1.5 hours) were provided by the CHANGE study co-ordinator and principal investigator, which were completed 1 week prior to the start of the courses (September 2015). The children’s weight management service operational manager also attended the training. The first session introduced the new materials and discussed the plan for each week in detail. The second session was delivered 3 days later to allow the facilitators to review the material. At this session, there was further discussion of the programme and the opportunity for the facilitators to raise any queries or concerns, which were then discussed and resolved. Facilitators were asked to adhere to the session plans and not to make any changes without first discussing them with the CHANGE research team. During the programme delivery period, facilitators were asked to feedback their experiences to the study co-ordinator, which enabled ongoing refinement of the programme sessions and the provision of further support to the facilitators.

Comparator

The standard First Steps programme was delivered for all programmes that were randomised to the comparator arm within the study period. Another facilitator from the BCHCT children’s weight management service delivered this programme over 5–7 weeks (depending on school term constraints), with parents attending all sessions and children attending the first and last sessions. More detail on the comparator programme is given in a TIDieR checklist in Appendix 4. Two comparator programmes were run during each of the four school half-terms within the study period (eight programmes in total).

To avoid contamination between the two arms, the facilitators delivering the adapted programme did not deliver the standard programme during the study period and were asked to refrain from discussing programme content with the facilitator who was delivering the standard programme during this time.

Evaluation of the adapted programme’s acceptability and the feasibility of its implementation

Programme completion

The primary outcome, to evaluate success of the programme adaptation to increase its suitability for Pakistani and Bangladeshi families, was calculated as an estimation of the proportion of these families completing the adapted programme and 95% confidence intervals (CIs). We also aimed to estimate the proportion of families of all ethnicities that completed the adapted programme and the proportion of families that completed the comparator programme. Completion was defined as attending at least 60% of the programme. To enable more precise estimates of these proportions to be made, we used routinely collected, anonymised attendance data from all programmes within the study period. This enabled us to use data from all programme participants, regardless of whether or not they were CHANGE study participants.

Family attendance at the programmes was recorded via parental signature at each session. Attendance information was collated and anonymised by the BCHCT weight management team and provided to the CHANGE study team once all programmes within the study period had been completed. Families were categorised into three groups: (1) completers (attended ≥ 60% of sessions), (2) non-completers (attended < 60% of sessions) and (3) non-attenders (did not attend any sessions).

Observation of programme delivery and facilitator feedback

Direct observations of programme sessions were undertaken by the research team to evaluate implementation and participant engagement and response. Specifically, observers took note of whether or not the programme was delivered as planned, which aspects were more challenging to deliver and how well the various parts of the programme engaged the families attending. We aimed to observe each facilitator delivering each programme session at least once across the study intervention period. Observers also made note of how the context influenced delivery and response (e.g. size of group, the facilitator, the venue, etc.). Facilitators were also asked to feed back their experiences to the study co-ordinator after each week of delivery of the programme, which provided valuable information on the feasibility of delivery of the sessions and the engagement of participants. The information gained from observations and feedback was used to refine further the adapted programme in the first 8 weeks of the study intervention period.

Interviews with programme facilitators, parents and children

At the end of the study intervention period, interviews were conducted with the two facilitators who delivered the adapted programme, with the aim of exploring their experiences of programme delivery, their perceptions of how participants responded to the programme and how this compared with their previous experience of delivering the standard programme.

Interviews were also undertaken with parents/carers who attended the adapted programme to explore their experiences, which parts of the programme they valued most and least and whether or not they were able to make positive behaviour changes as a result of attending the programme. We aimed to recruit 15 parents to participate in interviews, purposively sampling to ensure that approximately half were Pakistani or Bangladeshi and that we included a mix of programme completers and non-completers. We developed the interview schedules to include specific prompts to explore participants’ views of the adaptations made to the programme. Interviews were undertaken by the CHANGE research team (Tania Griffin and Kiya L Hurley, both of white British heritage), or by a community researcher (Meanaz Akhtar, of Pakistani heritage) if a participant required the interview to be in another language. In addition, nine children aged ≥ 8 years who had completed the programme were recruited to participate in interviews in which their experiences of the programme were explored. Again, we purposively sampled to include Pakistani and Bangladeshi children. Participating parents and children each received a shopping voucher worth £10. The interview schedule included participatory techniques that are designed to assist with interaction and recall.60–62 The interview schedules for facilitators, parents and children are shown in Appendix 5. Written consent/assent was obtained from all interview participants.

The interviews were audio-recorded, translated into English (when required) and transcribed verbatim. The interviews that were recorded in English were transcribed by an external transcription company (Clayton Research Support), whereas those conducted in a different language were translated and transcribed by the community researchers and back translated by an independent researcher to check for accuracy.

Data analysis for all participants was guided by the framework approach.63 This more deductive analysis approach was selected, as we were seeking to evaluate the adapted programme and to explore specific aspects of the adaptations that were made. Two researchers (Tania Griffin and Kiya L Hurley) reviewed the transcripts independently and identified codes to apply to the data. The researchers discussed their allocated codes and agreed on a final coding framework that was then applied to all transcripts using NVivo 10. The data were charted and mapped and overarching themes were identified. For the interviews with parents, differences were explored between those who completed the programme and those who did not.

Measurement of outcomes to be assessed in a future trial to evaluate intervention clinical effectiveness and cost-effectiveness

Data on a range of outcome measures were collected from study participants at three time points (T0, T1, T2; see Figure 6). The main purposes of this data collection were to (1) assess the feasibility of collecting these data from participants, (2) assess the completeness of the data collected and (3) gain estimates of parameters to be used in a sample size calculation of a future trial. In addition, we aimed to use the data to explore the direction of effect of the intervention and comparator programmes on these outcomes.

Assessments were undertaken with children, parents and other family members who were available and gave consent at the data collection visits to their homes. Assessments with children included measures of adiposity, an assessment of pubertal status and objectively measured physical activity and psychosocial measures, which included a quality-of-life utility measure (a key outcome measure in cost–utility analysis). Usual eating patterns of the children were assessed through parental report. Assessments with parents and other family members included anthropometric measures, an assessment of family eating and physical activity behaviours, parenting styles, parental feeding practices and parental self-efficacy. Anthropometric measures on parents and other family members and the measure of family behaviours were included because, in a future economic evaluation of the adapted children’s weight management programme, it would be important to try and measure any benefits of the programme to the wider family.

The complete list of outcome measures collected is shown in Table 4. The assessments were undertaken by trained researchers in the participants’ homes.

Collection of child outcome data

Anthropometric assessments

Height, weight and body fat percentage measurements were collected following standard operating procedures (see Appendix 6). Participants were measured in light clothing without socks and shoes and any headwear was removed. If the participant was wearing religious headwear that they did not wish to remove, it was measured and then deducted from the height measure. Child height measurements were taken on inspiration, and in duplicate, using a Marsden Weighing Group© Leicester Height Measure (HM-250P). To allow for measurement error, a third measurement was taken when the difference between the two measurements was > 0.4 cm. The two closest of the readings were averaged to give the definitive measurement.

Weight and body fat percentage measurements were taken using Tanita^® body composition scales (Tanita^® BC-420MA body composition scales). Participants were asked to pass urine before stepping on to the scales to minimise the effect that a full bladder could have on the fat percentage measurement. BMI was calculated by dividing weight (kg) by height squared (m). BMI z-scores were calculated for BMI and fat percentage using the lambda–mu–sigma (LMS) method and UK reference data.44,76

Waist circumference was measured 4 cm above the umbilicus (a measurement point identified to be the most valid and pragmatic in overweight children77) using a flexible steel tape measure. Children were asked to raise their clothing to access the waist; however, if they refused, an option was given to take the waist measure over thin clothing. Waist measurements were taken at the end of a normal expiration and were taken in duplicate. When the difference between measurements was > 0.4 cm, a third measurement was taken. The two closest of the readings were averaged to give the definitive measurement. The LMS method and UK reference data78 were used to produce an age- and sex-specific waist z-score from the definitive waist circumference measurement value.

Questionnaire assessments

The researchers verbally administered a number of questionnaires to children and recorded the responses (see Appendix 7). Quality of life was assessed using two instruments: the Pediatric Quality of Life Inventory (PedsQL) Generic Core Scales version 4.0 and the Child Health Utility 9D (CHU-9D) questionnaire. PedsQL is a validated questionnaire consisting of subscales measuring physical, emotional, social and school functioning.64,65 These aspects of quality of life are considered separately and all four are combined to assess overall quality of life. Age-specific versions of the PedsQL were used according to the child’s current age. Items were read to the children and responses were obtained either verbally or by using pictorial scales with younger children. PedsQL Likert scales were converted to 0- to 100-point scales, in which higher scores indicate a better quality of life.

The CHU-9D is a validated quality-of-life measure,66–68 which is administered for use in health economic analyses. It is a preference-based measure of health-related quality of life for use in children. It comprises a descriptive system and a set of preference weights, allowing the calculation of quality-adjusted life-years for use in a cost–utility analysis.

A figure rating scale, developed by Collins,69 and adapted by Rand and Resnick,70 was used to assess body image. This scale consists of sex-specific drawings of nine prepubescent figures ranging from very underweight (value = 1) to very overweight (value = 9). Two versions (light skinned and dark skinned) were developed for use in the study. Researchers used the version that was most appropriate for the child’s skin colour. Children were asked ‘Which picture looks the most like you look?’ and ‘Which picture shows the way you want to look?’ to assess ‘self’ and ‘ideal self’, respectively. To calculate children’s body dissatisfaction, the ‘ideal self’ score was subtracted from the ‘self’ score, giving a value of between –8 and +8. A score of 0 indicated body satisfaction, negative values indicated that the child would like to be larger and positive values indicated that the child would like to be thinner.

To assess children’s patterns of food intake, the Children’s Dietary Questionnaire was used (see Appendix 8). This parent/carer-completed 28-item questionnaire is a measure of intake patterns of a variety of healthy and unhealthy foods and is validated for assessment of intake patterns in children aged 4–16 years.71 Four food group scores are calculated from the items: fruit and vegetables, fat from dairy, non-core foods and sugar-sweetened beverages. As the questionnaire was originally developed in Australia, minor adaptations were made, with input from the Parent Advisory Panel and the co-investigator team, to ensure applicability to the diets of the UK-based ethnically diverse study population. Parents were asked to complete the questions independently unless language or literacy support was required. In these cases, the researchers completed the questions with the parents.

Physical activity assessment

Objectively measured free-living individual physical activity was assessed using triaxial accelerometers. At each assessment period, during the home data collection visit, children were asked to wear an accelerometer continuously for 7 days on their non-dominant side, except when sleeping. Two types of accelerometers were used in the study: the wrist-worn GENEActiv© (Activinsights Limited, Kimbolton, UK) and the waist-worn ActiGraph© GT3X+ (ActiGraph, Pensacola, FL, USA). The purpose of including two different accelerometer types was to assess which gave the greatest level of participant compliance and data completeness. GENEActiv monitors are typically worn on the wrist and provide raw acceleration data. ActiGraphs are typically worn on the waist, have been commonly used in research with children and have been validated for use in this age group.79–83 They can also provide raw acceleration data. At the baseline assessments, researchers assigned children either a GENEactiv or ActiGraph monitor at random. The children were then fitted with the same monitor type at the follow-up time points.

Both devices were set to record at 100 Hz in 60-second epochs. Data were analysed using the GGIR version 1.4 package developed in R version 3.3.1 (The R Foundation for Statistical Computing, Vienna, Austria; www.r-project.org; accessed 9 May 2016) through the University of Birmingham BlueBEAR High Performance Computing Service (www.birmingham.ac.uk/bear; accessed 9 May 2016). This package autocalibrates the raw triaxial accelerometer signals (x, y and z) to produce one omnidirectional value, the gravity-subtracted signal vector magnitude (SVMg; mg), using the Euclidean norm minus one (ENMO) formula [i.e. SVMg = √(x² + y² + z²) – 1]. Further information on the GGIR package is detailed at the following web page: (https://cran.r-project.org/web/packages/GGIR/GGIR.pdf; accessed 9 May 2016). A valid day’s wear was defined as > 10 hours of wear time in a 24-hour period. Device- and location-specific ENMO cut-off points were employed to estimate minutes spent in moderate activity (equivalent of four metabolic equivalents), vigorous physical activity (equivalent of six metabolic equivalents) and moderate to vigorous activity (activity at intensity levels over four metabolic equivalents). These were 359.7 mg (moderate activity) and 695.8 mg (vigorous physical activity) for GENEActiv monitors and 249.9 mg (moderate activity) and 464.6 mg (vigorous physical activity) for ActiGraph monitors.84–86

Other data collection

In children aged ≥ 8 years, pubertal status was assessed using a simplified version of the Tanner scale.87 Researchers were asked to indicate if there was any sign of breast development (females) or facial hair (males). In addition, parents were discreetly asked if menstruation had begun for female children. The Parent Advisory Panel was consulted on this approach to ensure its acceptability to participating children and parents.

The BCHCT children’s weight management team provided information on date of birth, sex and postcode for the participating children. This information was verified at the initial home visit. Parent/carers reported their child’s ethnicity, religion and language spoken. The postcodes were converted to the Index of Multiple Deprivation (IMD)88 2015 scores, a measure of small-area deprivation, which was used as a proxy for socioeconomic status. Participants were categorised into five groups using quintile cut-off points for England, so that group 1 included all individuals living in an area that is classified as being in the 20% most deprived areas in England, and group 5 included all those living in areas that are classified as being in the 20% least deprived areas in England.

Collection of parent outcome data

Anthropometric assessments

Height, weight and body fat percentage measurements were conducted with parents and carers who gave consent using the previously described measurement protocols (see Appendix 6). The BMI score was calculated by dividing weight (kg) by height squared (m).

Questionnaire assessments

Parents were given a questionnaire booklet and asked to complete it (see Appendix 8). Researchers provided assistance to parents when necessary. If a parent did not speak English, it was ensured that a community researcher with relevant language skills undertook the data collection visit so that they could verbally translate and complete the questionnaire with the parent. If participants spoke a language that was not spoken by the community researchers, an interpreter accompanied the researcher and verbally translated questions and participant responses, which were then recorded by the researcher.

At baseline, questionnaires included a range of questions to collect sociodemographic information, including parental ethnicity, religion, first language, place of birth (and when they moved to the UK if applicable), age when leaving full-time education, highest educational qualification and employment status.

At all time points, a number of questionnaire subscales relating to aspects of parenting style, self-efficacy and feeding practices were included in the parent questionnaire booklet. Parenting style was assessed using the authoritative parenting dimension of the Parenting Styles and Dimensions Questionnaire.73 This dimension consists of three subscales: connection (five items), regulation (five items) and autonomy (five items). Subscale scores were calculated individually and summed to give a total authoritative parenting score. Higher scores represent a more authoritative parenting style.

Parental self-efficacy was measured using the Parental Locus of Control scale.74 This scale consists of 10 items, which were summed to create a total self-efficacy score. Higher scores represent lower parental self-efficacy.

Nine subscales of the Comprehensive Feeding Practices Questionnaire (CFPQ) were used to measure a selection of parental feeding practices. The CFPQ has been shown to be valid in children up to 12 years of age and in varied cultural contexts.75,89 However, to keep respondent burden to a minimum, only selected subscales of the CFPQ were included in the parent questionnaire. These were: (1) child control, (2) encourage balance and variety, (3) environment, (4) modelling, (5) monitoring, (6) restriction for health, (7) restriction for weight control, (8) teaching about nutrition and (9) involvement. Each subscale contains between three and eight items, totalling 39 items. Higher scores represent greater use of each practice.

All items from the three questionnaire instruments described above were answered on Likert scales from 1 to 5 points. Items indicated as needing to be reverse-scored by the relevant instrument instructions were reversed accordingly. For ease of interpretation, scores for all items in each subscale were summed and divided by the number of items, creating a score range of 1–5 points for each subscale. If one (3- to 5-item subscales) or two items (6- to 15-item subscales) were missing for any responder, the subscale was standardised using the number of items answered as the denominator. When more than 1 or 2 items were missing, a score was not calculated.

Collection of family outcome data

Anthropometric assessments

Height, weight and body fat percentage measurements were taken for any other consenting/assenting family members aged > 4 years who were present at the data collection visits. The previously described measurement protocols were used to conduct these measurements (see Appendix 6).

Family behavioural assessment

The Family Nutrition and Physical Activity (FNPA) questionnaire is designed to assess family behaviours and environments related to children’s nutrition and physical activity.72 The questionnaire has 20 items overall and 10 subscales (two items each). The subscales are (1) family meals, (2) family eating practices, (3) family food choices, (4) family beverage choices, (5) restriction and reward, (6) screen time, (7) creating a healthy environment, (8) family activity, (9) child activity and (10) family schedule/sleep routine. All items of the FNPA were included in the questionnaire booklet for completion by parents/carers.

Collection of data on costs associated with the intervention

Within the feasibility study, we explored methods to measure costs from a societal perspective, including intervention-specific costs, parent productivity costs (e.g. time off work to attend the intervention sessions), associated child-care costs and changes to the family’s weekly food bill.

Intervention costs

For both the intervention and the comparator programmes, resource use data were collected prospectively by the BCHCT staff delivering the programmes and by the study research staff. Methods were developed to capture the resource use and costs associated with the development and production of programme materials for both intervention and comparator programmes. This included costs relating to the development and production of manuals, additional information provided and any other materials used within the programmes. In addition, the costs associated with training the facilitators in the intervention arm were also captured, as were the staffing costs associated with delivering the programmes. Data were also collected on the costs associated with the hiring of premises, with the resources used to promote the programmes and project manage the programmes and with the time spent assembling the materials used within the programmes. Costs associated with travel for staff to and from the venues were also captured.

Costs to families

Spending time attending a weight management programme and spending further time putting the learning into practice leads to a cost being incurred by families. This includes a cost linked to the time spent attending the programme, any associated child-care costs and changes to the family’s weekly food bill. A questionnaire survey was designed to capture this information (see Appendix 9). Specifically, the questionnaire asked parents what they would have been doing if not attending the programme, what arrangements they had made if they had needed time off work, whether or not they had paid for child care for other children when attending the programme, how long their journey to the venue was and how they had travelled, and whether or not their weekly food bill had changed since attending the programme. The questionnaire was administered by the facilitators to all programme participants (regardless of whether or not they were CHANGE study participants) at the final session.

Collection of anonymised outcome data from all children attending the intervention or comparator weight management programmes in the study period

To assist in exploring the direction of effect of the adapted children’s weight management programme on BMI z-score, the BCHCT team provided the research team with routinely collected, anonymised data on sex, ethnicity and pre- and post-programme height and weight measurements from all children who attended the weight management programmes within the study period. This provided a larger sample size for the exploratory analysis of the outcome of BMI z-score in the intervention and comparator groups. The measurement protocols used for the routine programme data collection were different from the study protocols, so the analyses of the study data and the routinely collected data were undertaken separately.

Sample size

We used routinely collected, anonymised data on programme completion to calculate the primary outcome (the proportion of Pakistani and Bangladeshi families completing the adapted programme and 95% CIs). An analysis of existing service data showed that the mean group size when a programme commenced was 11; therefore, it was expected that approximately 176 families would attend the 16 adapted programmes delivered in the intervention period, 40% of which would be Pakistani or Bangladeshi (approximately 70 families). We calculated that this sample size would allow an estimation of the proportion of Pakistani and Bangladeshi families completing the programme to within 26% precision. For example, if 65% of Pakistani and Bangladeshi families completed the adapted programme, the 95% CI of the estimate of completion would be 52% to 78%. For this calculation, we inflated the variance that we would expect under individual randomisation by the variance inflation factor for cluster trials (with an additional inflation of 11% to account for the varying cluster size90) and assumed a mean cluster size of five. As there were no specific data to inform the estimate of the intracluster correlation coefficient (ICC), we used patterns of ICCs from other sources to inform our calculation.91 We expected a moderate to high prevalence of our outcome (programme completion), which is associated with lower ICCs, but our outcome was a process measure and we had small cluster sizes, both of which factors are associated with higher ICCs.92 Therefore, we assumed a moderate ICC of 0.05.

We planned to recruit at least 80 participants into the study to undergo data collection at the three time points. This number of participants was deemed to be sufficient to test recruitment and data collection methods, and to enable estimation of the key parameters required for a sample size calculation of a future trial.93

Statistical analysis

As the feasibility study was a cluster randomised design, methods were employed throughout the analysis to account for the clustered nature of the data. In accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines,94 an intention-to-treat approach was used for all outcome analyses. Outcomes were considered significant at the 5% level. Statistical analysis was conducted in Stata version 13.

Estimation of programme completion

The primary outcome was the estimate of the proportion of Pakistani and Bangladeshi families that would complete the adapted programme and the 95% CIs (using routinely collected, anonymised programme data). Completion was defined as attending at least 60% of the programme. We adjusted the CIs using robust standard errors to account for the effect of clustering. To implement this calculation, we used the proportion function, with adjustment for cluster, in Stata version 13.

The same methods of analysis were performed to estimate the secondary outcomes to assess programme acceptability: (1) the proportion of all families completing the adapted programme, (2) the proportion of Pakistani and Bangladeshi families completing the comparator programme and (3) the proportion of all families completing the comparator programme.

Participant characteristics

Routinely collected individual child data (age, sex and ethnicity) were summarised at baseline for the CHANGE study participants and non-CHANGE study participants (i.e. children attending a weight management programme but not participating in the CHANGE study). The baseline characteristics of the CHANGE study participants were summarised by intervention and comparator arm. Categorical data are presented as numbers and percentages. Normality of continuous variables was assessed by comparison of the mean and median values and the distribution of values in a histogram plot. When normally distributed, continuous data are presented by their mean and standard deviation. Medians and interquartile ranges (IQRs) are presented for variables that exhibited non-normality.

Completeness of outcome data

To assess completeness of the outcome data collected from participants, the number and proportion of participants completing each measure are presented for each data collection time point.

Intervention and comparator costs

An exploratory analysis was undertaken to refine the focus for the methods for an economic evaluation within a future trial. In line with good practice for economic evaluation, we assessed the incremental costs. Therefore, costs that were common to both arms (i.e. the comparator and the intervention), such as costs associated with staff travelling to and from venues, programme promotion costs, project management and costs associated with assembling materials, were excluded from this analysis. Only the costs that were different between the two arms (i.e. the costs linked to the adaptation) were included in the analysis.

Estimation of parameters to inform a sample size calculation of a future trial

In addition to study attrition, other parameters required for the sample size calculation of a future cluster RCT are estimations of the ICC for the primary outcome and the correlation between baseline and follow-up measures of the primary outcome. Therefore, we used anonymised BMI z-score data routinely collected at the end of the programme to estimate the ICC, and we estimated correlation between baseline and post-intervention BMI z-score using the routinely collected programme data and outcome data from the CHANGE study participants. We also estimated the correlation between baseline and 6-month post-intervention BMI z-scores in the CHANGE study participants.

Exploratory analysis of participant outcome data by intervention and comparator arms

For each study arm, outcome data from children and parents at T1 and T2 were summarised using their means and standard deviations, when normally distributed, and medians and IQRs otherwise.

To explore the direction of effect of the adapted and standard programmes on the outcomes studied, mean differences between T0 and T1 (end-of-programme follow-up), and T0 and T2 (6-month follow-up) were calculated for both study arms, taking into account clustering.

Although an evaluation of intervention effectiveness was not an aim of the feasibility study and it was not sufficiently powered to detect minimum clinical differences in outcomes between intervention and comparator arms, we performed exploratory analyses to estimate differences in outcomes between the study arms. The purpose of undertaking these analyses was twofold: first, to develop methods of analysis for use in a future trial and, second, to explore the direction of effect of the adapted programme compared with the standard programme. Mixed-effects linear regression models were developed, adjusting for clustering and baseline measures. Ninety-five per cent CIs were reported and differences deemed significant at the 5% level. For each outcome, two models are presented. First, a model was developed with the T1 or T2 outcome as the dependent variable, adjusted for the baseline value of the outcome and clustering. Then, a second model was developed, further adjusting for child age, sex, ethnicity and IMD score. In addition, models were developed with the anonymised BMI z-score data routinely collected from all programme participants as the outcome, first, adjusting for clustering and baseline BMI z-score and then, second, further adjusting for child age, sex and ethnicity.

When necessary, data were transformed for use within the models. The following method was used to identify outcome variables that required transformation. First, the mean and median values of each outcome variable were compared to identify data skew. For those outcomes with an obvious skew, a Tukey’s Ladder of Powers plot was generated using the gladder command in Stata version 13.95 When a particular transformation was undertaken, two models were tested: one with the data in their raw form and another with the transformed data. The residuals from each model were calculated, plotted and compared for normality. When transformation of the outcome variable did not improve the normality of the model residuals, the untransformed data were used in preference to aid interpretation of the results. Transformations for a particular outcome were used for all models with that outcome to improve comparability. Outputs of all transformed models were back-transformed.

For the main analyses, only participants with complete data were included in the models. The number of participants included in each model is presented by study arm to allow full consideration of missing data.

Sensitivity analyses

For a variety of reasons (incomplete data collection, attrition, etc.) the numbers of participants included in the models to explore differences in outcomes between intervention and comparator arms were lower than the total number of study participants. For some outcomes, the number of missing data was substantial. Therefore, to explore further the differences between study arms, two imputation techniques were employed. First, the baseline observation carried forward technique was utilised, in which the missing follow-up value is replaced with the observed baseline value.96 This technique allows a conservative effect estimate to be generated. However, concern has been raised over the ability of single-imputation methods, such as baseline observation carried forward, to produce unbiased estimates of variance96 and there is an overconservative assumption that study non-completion is synonymous with no effect.97 Therefore, we used multiple imputation techniques in a second sensitivity analysis. Imputation was conducted for each outcome separately. For the anthropometric and physical activity outcomes at T1 and T2, the following items were included in the imputation process: the outcome of interest, the baseline (T0) value of the outcome (when available), other anthropometric and physical activity measures taken at the relevant time point (i.e. T1 or T2), completion status, pubertal status, age, sex, ethnicity and IMD score. For the questionnaire-based outcomes at T1 and T2, the following items were included in the imputation process: the outcome of interest, the baseline (T0) value of the outcome (when available), completion status, pubertal status, age, sex, ethnicity and IMD score.

We allowed for clustering in the multiple imputation procedure. The procedure involves fitting a conditional model for each variable with missing data in which the variable is linearly regressed on all other variables using mixed-effects linear regression (those with and without missing data). It then uses a Gibbs sampling approach (100 iterations), updating each set of parameters in turn, conditional on the others. A mixed-effect linear regression model was used to analyse each continuous outcome variable conditional on the imputed data. Rubin’s rule is used to pool the effect estimates of 10 sets of imputed data.98 We transformed any variables that showed marked non-normality. Transformations used in the imputation process were consistent with the main analysis. These multiple imputation models were implemented in Stata and REALCOM-Impute (Centre for Multilevel Modelling, Bristol, UK).

Results

Families participating in the First Steps weight management programmes within the study period

In the study intervention period (September 2015 to April 2016), a total of 536 families were referred to the children’s weight management service and invited to attend a programme. In total, 49% of referrals were made through the NCMP, 27% were self-referrals, 13% of families were referred by their general practitioner and 11% were referred by their school nurse. Of the 536 families invited, 243 families (45.3%) attended at least one programme session. This was consistent across the intervention and comparator arms. Forty per cent of all families who were invited to attend were Pakistani or Bangladeshi, with a slightly higher proportion in the intervention arm than in the comparator arm (42% vs. 35%). Of all Pakistani and Bangladeshi families invited, 48.1% (103/214) attended at least one programme session, with a slightly higher proportion attending in the intervention arm than in the comparator arm (51% vs. 41%). The flow of programme participants in the intervention and comparator programmes is shown in Figure 7.

FIGURE 7 Flow diagram of First Steps programme participants. BP, Bangladeshi or Pakistani. (Figure not reproduced here; see the original article.)

Anonymised attendance data from all families who attended at least one session of a programme during this period were used to calculate the proportion of families completing the programmes.

The CHANGE study participant recruitment and follow-up

A total of 92 children and their families consented to participate in the CHANGE study between September 2015 and April 2016. During this period, BCHCT provided contact details for 157 families. Details for 14 families were received once the recruitment target of 80 had been surpassed; therefore, these families were sent a letter explaining that recruitment was complete but that if they were interested to hear more about the research they were welcome to contact the study team. Therefore, 62.9% (92/143) of participants approached by the research team consented to participate in the study.

Of the 92 participating families, 75 (81.5%) received a home data collection visit at T0 (baseline). We were unable to arrange a T0 data collection visit with the remaining 17 families (see Recruitment difficulties). We attempted to contact all participants at T1 and T2. If we were unable to make contact at T1, we still attempted to contact families at T2. Of the 92 families, 82 had outcome assessments at T1 (end of programme, 89.1%) and 60 had outcome assessments at T2 (6 months post intervention, 65.2%). Participant recruitment and follow-up is shown in Figure 8. A total of 19 families participating in the CHANGE study (20.7%) did not attend any programme sessions (14 families, 20.3%, in the intervention arm; five families, 21.7%, in the comparator arm).

FIGURE 8 Flow diagram of the CHANGE study participants. a, Attempts were still made to contact families at T2. (Figure not reproduced here; see the original article.)

Recruitment difficulties

A multistage recruitment process was developed (see Figure 6), which was designed to be implemented by the BCHCT children’s weight management service team and the CHANGE study research team on a continuous basis throughout the study recruitment period, so that all families invited to attend a programme were also invited to participate in the study. However, in reality, not all families were invited to take part in the CHANGE study. This was a result of several factors. During the study period, the BCHCT team had significant administrative staffing issues, which resulted in delays in calling families and incomplete recording of whether or not participation in the CHANGE study had been discussed with the families. In addition, not all families received a telephone call as a follow-up to their programme invitation letter.

The second stage of recruitment also proved problematic. The research team experienced some difficulties in contacting people, even though the BCHCT team had successfully been able to contact them by telephone. Telephone calls were made at different times of the day to optimise the chances of making contact but for some families contact was still not achieved. In some cases, the telephone numbers were no longer active. If contact was not achieved after multiple attempts, a letter was sent to the family asking them to contact us should they be interested in taking part in the study.

Contact details were often received by the study research team either just before participants started a programme or after the programme had started, which posed logistical difficulties in arranging the collection of baseline data. In addition, the study researchers and community researchers encountered multiple delays when attempting to gain the necessary permissions to take consent and collect data from participants. This meant that, particularly at the beginning of the study, the number of researchers able to undertake the initial visits to take consent and collect baseline data (T0) was less than expected. The impact of both of these factors was that it became difficult or impossible for the research team to arrange the baseline (T0) appointment prior to the families starting their programme. A pragmatic decision was made to allow T0 appointments to be undertaken up until the families had attended their second session so that baseline data collection feasibility and acceptability could be tested. However, we were unable to arrange T0 visits to collect baseline data for 17 out of the 92 families who consented to participate, as they had already had the opportunity to attend at least two programme sessions. These families received a first data collection visit at T1 (i.e. after the end of the programme), and their baseline data were recorded as missing. Recruitment to a future trial would need careful planning with all partners involved to ensure that baseline data are collected from participants before they receive the intervention.

Attrition

There was a high level of attrition in this study, with only 65% of participants followed up at 6 months (T2). Of the 32 participants who were not followed up at T2, one-third actively withdrew from the study and the rest were lost to follow-up, despite multiple attempts to contact them by telephone and letter. There was greater attrition in the comparator arm than in the intervention arm (52.2% vs. 29.0%; see Figure 8). There was also a difference in the proportion of participants actively withdrawing between the two study arms, with a greater proportion of participants withdrawing in the comparator arm than in the intervention arm (26.1% vs. 7.2%). Participant characteristics of those who were followed up compared with those who were not followed up were broadly similar (Table 5). A higher proportion of Pakistani and Bangladeshi participants were followed up than not followed up. We also explored programme attendance and completion by study arm in those who could not be followed up. In the intervention arm, 7 out of the 20 participants (35%) who were not followed up at T2 did not attend any weight management programme sessions and three (15%) attended but did not complete the programme. In the comparator arm, 3 out of the 12 participants (25%) who were not followed up at T2 did not attend any programme sessions and four (33.3%) attended but did not complete the programme.

Participant characteristics

Characteristics of the CHANGE study and non-CHANGE study participants

The characteristics of children invited to attend a First Steps programme in the study period by the CHANGE study consent status are shown in Table 6. Overall, child characteristics of the families consenting to participate in the CHANGE study were broadly similar to those who did not consent to participate. When comparing the characteristics of the CHANGE study participants with those of the non-CHANGE study participants by programme type, for the group allocated to receive the adapted programme (the intervention arm), there was a higher proportion of Pakistani and Bangladeshi families who consented to be in the study than did not consent. However, in the group allocated to receive the standard programme (i.e. the comparator arm), the proportion of these families who consented was lower than those who did not consent. Mean pre-programme BMI z-scores were similar across the CHANGE study participants and non-CHANGE study participants in the two study arms, with a slightly lower mean in the CHANGE study participants than in the non-CHANGE study participants in the intervention arm (2.76 vs. 2.93). There are a large number of missing BMI data for the non-CHANGE study participants, as more than half of those invited to attend a programme did not attend and, therefore, their height and weight measurements were not taken.

CHANGE study participant characteristics

The baseline characteristics of children in the intervention and comparator arms are shown in Table 7. In total, almost half of the study participants were of Pakistani or Bangladeshi ethnicity, with a higher proportion of Pakistani and Bangladeshi children in the intervention arm. Nearly 80% were in the most deprived quintile and this was consistent across the study arms. Mean anthropometric measures were marginally higher in the comparator arm.

Programme completion

The proportions of families that completed the programme in the intervention and comparator arms, estimated from anonymised attendance data from all families attending at least one session of a programme during the study period, are shown in Table 8. The primary outcome, the proportion of Pakistani and Bangladeshi families completing the adapted programme, was estimated to be 78.8% (95% CI 64.8% to 88.2%). The proportion of all families who completed the adapted programme was 76.3% (95% CI 67.0% to 83.6%), which was higher than the estimated proportion of all families completing the standard programme (58.1%; 95% CI 46.5% to 68.8%).

The analysis of programme completion was repeated for only the CHANGE study participants (see Appendix 10), and a broadly similar pattern of completion in the intervention and comparator arms was seen, although the 95% CIs of the estimates are very wide because of the small numbers of participants included in the analysis.

Observation of programme delivery and facilitator feedback

The CHANGE research team directly observed delivery of 12 of the adapted programme sessions. One facilitator was observed delivering all six sessions. The other facilitator was observed delivering all but the first session and was observed twice when delivering session 2 (healthy eating). The facilitators also communicated with the study co-ordinator by e-mail or telephone to feed back on their experiences of programme delivery each week. Overall, the facilitator feedback of session delivery was positive and the facilitators reported a high level of enjoyment when running the sessions.

In the first school half-term of programme delivery, the research team used the observation and facilitator feedback to refine programme materials further. Refinements to the programme and other actions undertaken in response to the observations and feedback are shown in Table 9. After the first school half-term, the intervention materials and the facilitator’s manual were finalised.

Further observations and feedback continued throughout the study intervention period to determine whether or not the programme was being delivered as intended and to identify ongoing difficulties with programme delivery. No major concerns were identified and, in general, the facilitators delivered the programme materials in the way that they were directed to in the manual and training sessions. However, contextual challenges in delivering the programme remained throughout the intervention period. Particular issues were the disruptions caused by families arriving late, which happened at almost all sessions, and the logistical challenges of large group sizes (e.g. when there were > 12 families or when there were many siblings accompanying the children). The presence of interpreters for non-English-speaking participants worked well and the participants who required interpreters were observed to be included and engaged in the sessions.

Interviews with programme facilitators

Both intervention facilitators were interviewed (F01 was female, of Pakistani ethnicity and delivered 12 adapted programmes, and F02 was female, of white British ethnicity and delivered four adapted programmes within the study). One interview took place face to face in the facilitator’s workplace and the other took place via the telephone because of the facilitator’s availability. The interviews were 46 minutes and 41 minutes in length.

Facilitator views of the adapted intervention

Overall experiences

Both facilitators reported that they enjoyed delivering the adapted programme, which was consistent with the feedback they gave during the running of the programmes. They both acknowledged that the programme was very different from the standard programme that they had previously delivered. Much of the facilitators’ praise for the new programme focused on the more interactive and flexible delivery, which was a significant change from the standard programme (although there were interactive elements in the standard programme; see Appendix 4). The facilitators also felt that the families attending responded well to the interactive nature of the programme:

I think a lot of the aspects of the course worked brilliantly and I really enjoyed the interaction of it so overall I enjoyed it and I much prefer it as a course.

It was a nice refreshing change to see that we could facilitate rather than actually talk, erm, some of the families to death at times so it was, it was I would say nice. Erm, and as I, like I said it was a refreshing change to sort of try and deliver something new. Erm, and to be honest I, I would say that it was, erm, a pleasant I would say response from a lot of the families as well.

With the new programme I would say that there was a lot more opportunity for the families to get involved so a lot more sort of, erm, I would say practical activities that we could get them involved into as well, so that was nice. Although we did have some practical things in the old programme, erm, I would say that maybe some of the things we didn’t sort of hit the nail on the head with certain things which you guys [researchers] did.

Both facilitators highlighted that the programme sessions provided a good environment for social interaction between participants and that parents would share experiences and support one another in sessions. This meant that their role was much more facilitatory and less directive than previously:

There, erm, a better I would say, erm, social environment between I would say the participants because we gave them I think more opportunities as well though in this course. Although they still had opportunities in the older course, I would say they did have a few more extra ones here where we’d bring in the games and things as well. So I think it was a, it was a nice chance for them to, I would say more chances for them to actually gel together and, and, and speak to one another.

What tended to happen was one parent would say ‘from last week do you remember talking about this? And blah, blah, can you tell me a bit more about it?’ And then another parent would join in and go ’you know my daughter does that and then for me when . . .’ it worked brilliantly. I could just sit back and let them do it and for the most part a lot of good discussion was had. Certainly in that when the children were removed and they went off to play in the physical activity session when it’s just me and the parents and that’s when it works beautifully. If one parent does pipe up with something and then they all start talking and I liked that because it took it away from me going ‘right, do you do this or do you do that?’.

The provision of new, high-quality printed materials (display boards, participant folders, etc.) was well received by the facilitators. They felt that it gave the programme a more professional edge and helped the families feel that they were important, thus encouraging their engagement with the programme. The facilitator delivery guide was also highly valued:

I think because the [display] boards themselves which I love, the materials and everything fantastic, the boards are brilliant . . . and just the whole fact that they’re getting materials you know the ones that we were giving [in the standard programme] were very good but these were of a quality that’s kind of far in excess of the ones that we were giving previously and I think that kind of gives them a sense of ownership and importance and they’re coming to something and a little bit of excitement and involvement for the children.

We’ve loved the resources, colour copies and the folders. We’ve been enjoying all of that part.

And for me to have my guide, my facilitator guide so I’m sitting there with that one guide constantly and making notes and thinking about it, um, I like that as well, that’s a big help.

Challenges of delivering the new course were also discussed. Difference in delivery style (i.e. from ‘classroom’ to a more interactive style) and running all sessions with both parents and children present were raised by one facilitator as big challenges:

Wow, massively challenging because it was very different, the course previously was only involving the parents and it was a very much classroom-based type thing, ‘you listen to me, this is what I would like you to do’ kind of thing. We met the children only at the beginning and at the end, so we didn’t really have to interact with the children as much, um, we only took their weight and height, you know their weights and measures at the beginning and the end, we were trying basically to persuade the parents that this is how they should be adapting their lifestyle to suit their children and to improve the health of their children, but it was very much more weight focused as well, so we had to think of a way to approach the subject because they were there because their children were overweight and that was always there. So the focus I think, and the delivery was very different, so when this course came along it was a lot more involved on our part and because the children were going to be there . . . the big difference is that the method of delivering it was going to have to be very different, so that I liked, um, and then delivering it I enjoyed.

Saturday programmes

One of the adaptations of the new programme was more availability of programmes on a Saturday. With the standard programme, 25% of sessions were delivered on a Saturday, but with the adapted programme this proportion was increased to 50%. Both facilitators recognised that this had a positive impact on the success of the adapted programmes, as they believed that it enabled the parents to be more relaxed, outside the pressures of the working week, and allowed them to bring other family members along and, therefore, to embed the behaviour change messages more firmly into the family:

Families are, are very relaxed about Saturdays and, and I would say that they have been very successful . . . I think it’s the same as any of us though ‘cause on a weekday I just, on my working days oh God, I can’t take any extra on, you know. So I can see that for, for some families although they’re still doing it they might be quite stretched and then quite stressed out by the time they’re there, they’re not really concentrating. So I don’t know . . . they’re just, you know, a bit more relaxed and they’re probably just in a better mind frame, erm, to concentrate on something, something that’s gonna benefit their family.

It does make a massive difference because the family can come, the whole family can come, whereas during the week you know even some adults find it difficult to take that hour and a half you know they’re working or you know they’re at home but they’ve got other you know commitments . . . but an hour or so on a Saturday morning before the weekend starts if they’re motivated to come to it, which most of them are it has to be said, it seems like the obvious time to do it and you do get, you get siblings along and the messages apply to them just as much as it applies to the person who’s been nominated to come.

Children in attendance

A key adaptation was the presence of parents and children at all programme sessions (children attended only the first and last sessions of the standard programme). The facilitators were clear that this had a positive impact on the programme for a number of reasons. From a delivery perspective, they felt that the delivery needed to focus on engaging the children to keep them interested, and this would have a positive impact by removing the spotlight from parents and lessening the perception that their parenting was being criticised. They also felt that the children motivated the parents to keep attending the programme sessions, and they valued the opportunity to witness interactions between parents and children, as this helped to shape the support they gave the families:

. . . the children were going to be there, and if you can engage the children then you stand a lot more chance of engaging the parents because sometimes it’s very difficult to persuade an adult not that they’re doing something wrong but you’re always treading that line between you’re kind of not being a very good parent are you?. . . but when the children are there you can approach it in a very different way, so when we were given this course it was obvious that that was going to be, the big difference is that the method of delivering it was going to have to be very different, so that I liked.

The children drive it. If the children can drive it, ‘I want to go, I want to go, I want to go’ then the parents will bring them, they just will whereas if the parent is going ‘do you know, I can’t be bothered to go’ they just won’t but if the child is there going ‘you, I want to go’ so having the children there I think it has made the whole course has made a difference to attendance I think, definitely.

Erm, and also as a, get, having the kids involved was such a big plus for us because like I said, there were lots of reasons why we’re, it was, you know, it was good to see them with their parent and what their relationship was like and act with them. We’re there, you’d act, you’d know whether they’re doing the, the targets or not which was hilarious because you’d just see in their faces they weren’t.

However, the facilitators identified that having younger children in the group was a big challenge. They felt that the content of the programme was beyond the comprehension of 4- and 5-year-old children and this sometimes resulted in them losing interest and disrupting the session:

We had very young children there so I think we need to think about it in terms of the age of the children and age appropriateness, which I don’t think we thought about at all, any of us, when you’ve got 4- and 5-year-olds there they’re not interested because they’re 4 and 5, whereas when they’re kind of 6 and 7 they can sit and talk to you and listen.

In actual fact, talking age of children, some of the age groups we did realise . . . that we found that the 5-year-olds were losing interest very quickly . . . It was that the very young ones were so in, in reflection to what I’ve just said, because some of the content was still quite, I would say, I would say quite hard for them to understand when we’re talking about things like carbohydrates and things for a 5-year-old that’s just gonna go straight past them. They would lose interest but it became sort of a bit of challenging.

Programme content

Both facilitators expressed concerns about the ‘healthy eating’ session (week 2). There was a feeling that all the nutrition messages were packed into this session, whereas with the standard programme this content was spread throughout all sessions. This session had been revised so that the content was reduced but the facilitators remained concerned about its delivery. One facilitator highlighted that trying to give the planned amount of information to the participants in this session created a conflict with the programme’s ethos of listening to, and being responsive to, participants:

. . . but we only had about an hour to deliver what we were previously delivering in 6 weeks, and it’s a lot of information for anyone to take in. Even as a nutrition professional there was a lot of information to kind of get in for an hour so to deliver it to parents, and for them to go away. It was kind of they were reeling from all the information that we were trying to give them.

Erm, but that was the only thing that I really struggled with was week 2 to actually get that, all that information across and it didn’t really, then it kind of went against what we were trying to do, which was giving them a chance to sort of interact with us, because there was so much.

One facilitator felt that the programme was imbalanced in terms of having too great a focus on physical activity and not enough of a focus on nutrition. However, this was in contrast to the views expressed by the parents and children:

I think that’s my biggest frustration about the course itself is that we’ve moved too far in down the road to saying physical activity is key, and it is key but they need to go hand in hand and we are losing the practical food tips, so that’s my frustration I think, um, with the course because overall I love it and I much prefer delivering this one than I did the previous one which was a lot more stilted and a lot more formal but the balance has shifted . . . we should talk about the food again and the physical activity and try and marry them both together.

The same facilitator also expressed her frustration at the shift in focus of the adapted programme away from weight management and towards healthy behaviours. In contrast to the findings from the qualitative study in phase I, she felt that a focus on weight was necessary and led to better outcomes in terms of weight loss:

So [on the standard programme] we were giving them practical [advice] – look at this, did you realise, no I didn’t, I’m going to change it for that, and it was like a peg, they could grab onto that one and do something about it whereas [on the adapted programme] we always veered away from the words weight, overweight, and it was healthy lifestyle, which is fine but the results at the end anecdotally I don’t think were as good in that regard.

Although they preferred the adapted programme overall, both facilitators clearly identified aspects of the standard programme that they valued highly but that had been omitted in the nutrition content in week 2 of the adapted programme. In particular, the use of visual props, such as showing the actual amount of sugar in a fizzy drink, was felt to be of great value in delivering key messages. In addition, one facilitator felt that the messages on how to implement changes were less clear to participants than in the standard programme, and some of the more practical tips had been lost in the adapted programme:

Everybody is astounded when they see those kinds of sugar, it just, something just drops, the penny drops even though they think they know until they actually see it physically and I think we missed that massively and the plate, the portion plate, not the Eatwell plate but there’s a portion plate, because when you look at the size of the plate as well and I think visually, props-wise, we needed to incorporate more of that stuff . . . The parents all know cans of coke have got sugar in them, but until you pick up the sugar bag and put it alongside it, that’s when the shock registers as to exactly how much there is.

The only thing that I would say is that when it came to sort of the nutrition side of things, I still found that things were a little bit vague . . . when we talk about carbohydrates and things, I still think they needed a little bit more direction as to how, you know, like how many portions they should be having or just that staying a little bit clinical, erm, with what we’re expecting them to do, which we didn’t really, in the new programme we didn’t really go into, which is the only thing that I was a bit, er, sceptical about.

Both facilitators reported that many parents returned to discussing nutrition and healthy eating in the third session when their children were not present (‘making changes’, week 3). One facilitator found this very helpful as the parents themselves were bringing up points from the previous session and it gave her an opportunity to tie in healthy eating messages with changing behaviours. In contrast, the other facilitator felt that this was a sign that the messages were not covered in enough detail and that parents were frustrated:

. . . and they would say [in week 3], you know, ‘Oh’, you know, ‘This is the issue that we’re having with food again’, so they’d bring that back up and it was kind of nice because although we were, we were discussing and delivering behaviour change, it kind of brought in stuff from the previous session and an opportunity for them to ask questions without their children being there . . . it was nice for them because they sort of felt like there was a bit of a break, you know, a breakout for them where they’d actually have a chance to, to ask us, you know, ‘This is what we’re struggling with’, or, you know, not necessarily just behaviour, just in general all their habits and how to break them.

I have found in my behaviour, the behaviour session [week 3] when the children go away and I’ve got the parents there we tend to spend the first half of that session talking about food, portion size, tell me about how many meals they should be having a day and it’s kind of well we did cover that last week but they’re frustrated that we didn’t cover it in enough detail or give them something like, I’m going to show you my hand, this is the size of the plate, you know the fist, any kind of practical tips as to how much they should eat, when they should be serving their children.

Programme website

The facilitators had differing opinions on the value of the website. One was very positive about the website as a resource for families, whereas the other felt that there were barriers to its use among the programme participants, although she acknowledged that it was a good resource for older children who may have more information technology skills than their parents:

Apart from the fact it had my face in I loved it, no I think it’s great because we’d always been asking for a resource that we could give them um anyway, so I think it’s a great idea and I do like it, and I think it’s easy to get around, um, because the courses are so short I think sometimes they’re left hanging and go well, what do we do without seeing you every week now? And to be able to say to them you’ve got your website, you’ve got my number, both those things, everything is there and if you can’t find the answer there then you can phone us and we can support you because that’s you know that’s what they want, they want some support, and having that is brilliant.

Erm, I would say that some of them were a little bit sort of not, erm, what’s the word, sort of computer whizz, so some of the parents like, well who had language barriers and things wouldn’t really, I would say, go onto a website . . . But some of the kids that were, I would say, at an age where they understood ‘cause kids, you know, kids are computer whizzes so they knew what they were doing, they, they would say they’ve been on, erm, some of them had looked at recipes.

Interviews with parents and children attending the adapted programme

Participant characteristics

Fifteen interviews with sixteen parents who attended the intervention were conducted. One mother and father, from the same family, were interviewed together. There was a mix of mothers and fathers, ethnicities and levels of programme attendance, with 11 participants classed as programme completers and five as non-completers. Unfortunately, we were unable to recruit any Bangladeshi parents to participate in an interview. Three participants were interviewed in another language (two in Urdu, one in Mirpuri). Nine children were interviewed, seven of whom were children of parents who were also interviewed. All children had completed the programme and were between the ages of 10 and 12 years at the time of interview. Participant characteristics are summarised in Table 10. The average length of the interviews with parents was 20 minutes, ranging from 14 to 29 minutes. The average length of the interviews with children was 19 minutes, ranging from 12 to 28 minutes.

Parent and child views of the adapted programme

Overall experiences

In general, parents and children reported attending the programme as a positive experience and felt that it had encouraged them to make behavioural changes. Several participants discussed the wider impact that the programme had within the family, with many reporting behavioural changes that had occurred in other family members. These wider impacts were not confined to families who completed the programme, with some participants who had only partially attended reporting that there were behavioural changes within the family as a result of attending sessions:

. . . the whole thing it was delivered so softly it was just about making sensible choices, informed choices, you know, and I think that we got all the tools that we needed to do that, you know, we were told everything and the way it was delivered was superb, I can’t say that enough it was just from start to finish it was a really good course. Er, and yeah we’ve continued doing it and we’ve made changes and we’re continuing to make changes.

All I can say is it was a brilliant experience for my daughter because she’s made a major change in her life and I think as she grows older she will realise, you know, how to keep on going that was instead of going the other way she was doing, that’s the only other thing I can really say.

. . . it was really good, it helped a lot, erm . . . I enjoyed going to it because it was fun, yeah.

. . . but I think whatever we learnt, we will actually stick to it erm, in our family, her dad is one that he loves his food and drink, so changing him was quite hard, but even like, if he has a choice with a meal he might actually chose a water instead of the fizzy drink now, whereas before that wouldn’t have even been an option . . . I think you know, even whilst I was doing the course I was mostly focusing on my daughter, but after the course I thought I don’t actually have any physical activity implemented on myself so I’ve started doing my walks, going to the gym, doing a bit of aerobics.

I think all of it was because I could see her weight and I could see what I could do as a parent to help her and all my family. And my husband made a few changes in his diet as well, like he picked up that, because he used to eat unhealthy and now he’s eating healthily and he’s going to the gym as well, so I think so it has worked for us.

Er, I’ve sort of, er, changed portion size as well for the 2 year old and I haven’t changed the size of the plates yet, but portion size I’ve changed, er, fruit and vegetables, they eat a lot more fruit, they’ve always even before we started the programme they really love their fruit but the vegetables with the two younger ones was a problem, but with [child name] he eats almost all vegetables.

Some parents felt that they already knew some of the information presented within the programme, but they acknowledged that they also acquired new knowledge:

They explained quite well, I mean what to eat and what should be avoided. Although I already knew about this but it is common when we attend such events we always learn some new information, so this helped me a lot.

But the stuff was pretty much a lot of stuff we’d already knew, you know, and we picked up and stuff already, so – but there was a – quite a few beneficial things, like stuff like, erm, for instance, erm, like the [brand name] jars, like we didn’t – we knew they had sugar but we didn’t know how much and stuff like that, and then the sugar stuff smart [application] and stuff like that.

The group-based aspect of the course was generally well received by both parents and children. Parents valued the support they gained through sharing experiences with and learning from other parents, and children enjoyed the opportunity to make new friends:

It was nice to be around other people that had the same like issues with weight management stuff with their children, because the support was good, as in, you know, to be around other parents and, you know, just generally when you talk to the parents and stuff, you kind of like – it makes you realise that you’re not the only family with this issue and stuff, so the support helped and stuff.

It was brilliant because I’ve, good idea, because for me, it’s just I’m learning things and I learned through other parents as well, if I listen to them I get some ideas as well and they learned something from me as well.

In line with the facilitators, parents found the session in which the children undertook activities separately (week 3) to be very beneficial, as it allowed more opportunity to develop relationships with other parents:

The week that [name of child] went off to do her stuff, I think you [talking to child] really enjoyed it. And it was quite a – it was quite useful as well – it was quite useful as well to have an opportunity to talk about things while [name of child] was busy doing something else.

Programme timing

Programmes delivered on Saturdays were generally well received. One parent expressed concern at taking her child out of school on a weekday during an important school year and so chose a programme on a Saturday that was much further away. Some parents and children reported that a 9.00 a.m. start on a Saturday morning was challenging for them:

Programme structure

Parents were mostly happy with the six 90-minute programme sessions delivered over 6 weeks. A minority of parents commented that they would prefer a different structure, either a more intensive programme with sessions close together (e.g. over the summer holidays) or a programme delivered over a longer time period:

Maybe that it was just once a week basically. Maybe, yeah. ‘Cause now that we’ve got a 6 week [school holiday] that could take place now and the children off – off school.

I think personally that there should be a programme for overweight children that they should be in touch with them regularly, rather than, you know, just a session – like a few sessions clustered together, or you – do you know what I mean? Like a longer time period to kind of lengthen that.

Barriers to programme attendance

Parents who did not complete the programme generally recalled logistical reasons for not attending. This was consistent with the findings in the phase I qualitative study. However, one parent felt that his child was uncomfortable with being identified as overweight and that he did not identify himself as being the same as the other children who were attending the group:

I wanted to join again but my car broke down. However, they said if you want to come please let us know, though I wanted to go but I had this issue so I said I will try to come . . . In the beginning I attended two sessions and then one more, but then I could not go. I wanted to go with my friend but I could not go in the last three sessions.

Because of getting from school to go it was – and because going to college and stuff, it’s kind of – was kind of impossible to attend every session.

. . . from first impression I find all of them they’re fat kids and this kind of – this, erm, first impression, it can affect him. He thought, ‘Am I one of them? I’m not fat. I’m a sportsman’. So from that side I think that can affect him saying, ‘No, I’m not the one to attend this kind of course.’

Children in attendance

The inclusion of children at all programme sessions was generally very well received. Parents valued the fact that children were directly hearing messages relating to healthy eating and physical activity from a professional, and some also felt that it provided an opportunity for them to start conversations around weight management with their child:

It was actually quite useful because most of the time children won’t listen to parents, but when they see a professional explaining they take on board. And I think it was actually tool for me to reach him, so it’s literally opened our communications between us so it was quite useful really.

. . . I thought you know, we might as well go for it and see what happens and when we did go, finally when they finished, I thought it really, really helped, I think that bit of encouragement from somebody else rather than the mum saying to the child, I think it really helps.

Parents also identified some difficulties with having children present at the sessions. In line with the facilitators’ views, some parents felt that the wide age range of children presented a difficulty, with younger children becoming disengaged and older children becoming bored:

I definitely think, it probably didn’t have the same, er, the same accessibility or the same outcome with the younger children, I felt that they were you know they weren’t engaged in it, they were more playing with toys and all of that, so it didn’t really benefit them too much. And as a result because they were getting into mischief and the parents were being drawn away from it and possibly missing key points, er, but yeah for the older children, I’d probably say 9 upwards they were fully engaged and they were listening to what was going on, you know, and they were taking it in. I don’t think it was much use for the younger, if anything it was more of a hindrance.

Erm, I think a little bit, because [child’s name] was one of the, I’m not really sure if she was one of the eldest but she was a Year 6 child and she’s quite bright anyway so a little bit of it was too basic for her, so she found that, like occasionally you know once the main part of it, she found it a little bit boring so maybe if you were to do it again, split it up into the age groups.

They just got a bit bored, didn’t they, after it? It’s quite – it was about an hour and a half, and I think it was just too long for some of the little ones to sit and listen.

Pitching the sessions at the right level to engage both the children and the parents was also identified as a challenge, particularly around more complex concepts such as food labelling. This activity in particular evoked differing responses from participants, ranging from those who found it challenging but learned from it to those who felt that it was too complex for the children to grasp:

To be honest with you we both learned, it’s whatever applicable to the children was also applicable to the parent as well. And I think one of the useful things which I never really pay attention initially was the food labelling, was that fat and how many grams salt, it’s not something I normally pay attention, I just grab them.

I think she didn’t really pick on it that much, for her if you put a package in front of her it’s still going to be a package. She did sort of understand that if she had too much it’s bad for her, that’s it, but I think it was more for adults.

Erm, it was quite fun at first. Err, it was quite fun altogether because we had to go round the room and we had to find the different foods. Then you had to find out how much they were labelled in, err how much sugar and fats are there in there.

Um, I didn’t like the food labelling activity because you didn’t get to write the um, you had to quickly write the words down and um, you had to like, you would get muddled up and things. And uh, you would get confused. But with the sheets I’ve got in my folder, I know which is a healthy snack and which is not.

Physical and other activities

In contrast to the facilitators’ views, parents felt that there was still not enough focus on physical activity, despite the incorporation of physical activity into each session and a dedicated physical activity session for children. Children really enjoyed the physical activities, along with the food-making activity. Some parents and children reported that there was too much sitting and listening for children, despite the increased interactive content:

Yeah, but I thought there’d be more like activities for the kids and stuff, I think they only had activities on 1 day and the other time they had, that’s when they had to make snacks, so the rest of the time they were sitting there and [child’s name] would say, do we have to go because it’s so boring just sitting there.

Like, erm, when you got to do some exercise for only 1 week, and sitting down, that’s not much fun, it was supposed to be fun but it wasn’t that fun . . .

Some parents suggested that local community facilities could be used, such as the park or local leisure centre:

I think it was good, erm, I think at the time the weather didn’t really support us so we couldn’t really go out but, erm, if you were to do it again, maybe one of the sessions could have been at the park or at a leisure centre or something to actually, erm; they did go into the playground for a while, but if you were to do like a full session outside with the parents, maybe that would have helped, but we learnt that the children need, the amount of exercise they need each day which is quite a lot if you think about it.

You know, it was in the morning, weekend morning, was I think my son also plays with his friends, er, and I think if he can – you know, I’m thinking if they can – if they can do that the same activities in the park, the student with the kids always playing there, some kind of – some activities, they always, erm, start that kind of, erm, erm, er, activities in the park. So if they can move the place to the park maybe.

Programme website

All participants were asked about their use of the website. Only one parent and some of the children reported having used the website; however, it is possible that there was some confusion between the programme website, developed as part of the adapted intervention, and the nationally available Change4Life website (https://registration.change4life.co.uk; accessed 20 April 2017). This suggests that this may have been a less useful aspect of the programme and possibly signposting to existing online resources, such as Change4Life, may be a more cost-effective option:

Yes, but not as much as I should have if I’m honest, I used it three or four times and we did use it, er, we did use it every now and again just to look at food, the sugar content and the fat content because of the items on the site because some of the – some of the cards and things like that.

Summary of findings from interviews with facilitators, parents and children

The experiences of all those who were interviewed were generally very positive. Participants reported behaviour change within their families, even if they did not attend all programme sessions, and facilitators valued the flexibility of programme delivery. In keeping with the findings of the qualitative study in phase I, the weekend delivery of the programmes generally worked well for families, and the adaptation of children attending all programme sessions was felt to be of value by the facilitators and the participating parents. As in phase I, parents acknowledged the value of children hearing messages directly from an authoritative figure. However, both parents and facilitators drew attention to the fact that having such a wide age range of children attending presented difficulties. In particular, it was felt that the programme content did not keep the younger children engaged. Facilitators, parents and children all welcomed the interactive elements and valued the social and peer support aspects of the programme, both of which had been highlighted as important elements in the phase I qualitative study. Parents and children especially liked the physical activity and food-making activities but wanted more of these. Again, this is consistent with the phase I findings, in which parents identified that they wanted physical activities within the programme. However, these findings in phase II imply that the adaptation of introducing physical activity content into the programme did not go far enough and that families felt that there should be even more opportunity to undertake physical activity within the sessions. In contrast, the facilitators felt that there needed to be more content in relation to nutrition, especially practical advice when preparing food for the family. Contrary to the findings in phase I, one of the facilitators felt that the shift in the focus of the programme away from weight management was unhelpful. However, this viewpoint was not articulated by any parents or children. The programme website was not well used. Overall, the programme was valued by those delivering and participating.

Feasibility and completeness of outcome measurements

Feasibility of data collection

Data collection in the participant’s home

A number of difficulties were experienced with the data collection visits to the participant’s home. The home visit appointments were made by telephone a few days in advance, with a follow-up confirmation letter and a reminder text or telephone call on the day of the appointment. Despite this, on several occasions the families were not at home when the researcher arrived. Some participants gave reasons (e.g. unplanned hospital trips, unexpected school commitments); others did not give an explanation. It was possible to rearrange some of these data collection appointments, but some families were lost to follow-up as repeated attempts to contact the family and rearrange the visit failed.

Initially, data collection home visits were planned to be undertaken by one researcher and a lone-working policy was implemented. Following the first few visits, concerns were raised by researchers, as many appointments were during hours of darkness and often they had to park their car some distance from the home and walk with heavy equipment in unfamiliar areas. Therefore, it was agreed that first visits to a home should be attended by two researchers and that a risk assessment should be undertaken for lone researchers to make the subsequent visits.

Participant burden

At the start of data collection, the baseline home visits lasted a median length of 60 minutes, ranging from 40 to 90 minutes. In some cases this was too long, with participants becoming fatigued before data collection was completed. Researchers noted early on that the number of questionnaire items was too high for the adult participants. In response to this, the data collection process was modified in January 2016. The questionnaire booklet was divided into two. One contained core data questions (sociodemographic information, Children’s Dietary Questionnaire, FNPA questionnaire) and was completed at the appointment and the other (containing questions on parenting style, parental self-efficacy and parental feeding practices) could be completed by parents at the measurement appointment, or at a later date, and returned to the research team in a Freepost envelope.

For non-English-speaking participants, community researchers (or, when necessary, accompanying interpreters) translated all the questionnaires. Therefore, these participants did not have the option of completing the second questionnaire booklet at a later time. The data collection appointments took even longer for these families to allow for translation time and for all questionnaires to be completed. This remained an unresolved issue in the study.

Concealment of study arm allocation from researchers undertaking data collection

Programme allocation was unknown to the study team at the time of baseline measurements. However, once participants had started a programme, it became apparent to the study team which arm they were in through conversation with the participants and the presence of the programme materials within the home. This was less of a problem with the community researchers, as they were unaware of the differences between the intervention and comparator programmes. In a future trial, study arm could be concealed from the researchers undertaking the measurements if they are separate from the researchers who are co-ordinating the study and undertaking process evaluation of the intervention.

Other issues

Ramadan is a period of fasting during daylight hours that is observed by many Muslim families, and in 2016 it was within the planned T1 measurement time period. It became clear during the study that the participating Muslim families did not want a data collection appointment during Ramadan. Therefore, we delayed the appointments with these families until after the end of this period.

Data completeness: sociodemographic data

The completeness of sociodemographic data collection from participating children and parents/carers is shown in Table 11. Most sociodemographic data that were requested were provided. Data on parental education level and employment status were less complete, with approximately 85% of participants providing these data. Data completeness for birth place was high, but data completeness for the age at which parents moved to the UK was lower (74.5% of those born outside the UK provided this information).

Data completeness: outcome measurements

Table 12 shows the number and proportion of families who provided valid data and the average score/value for each outcome.

Anthropometric measurements: children

All children consented to height and weight measurements being taken and, therefore, a BMI z-score was calculated for all children. At each time point, two children had invalid body fat percentage measurements (e.g. the child did not want to remove their socks); therefore, the number of children with body fat percentage z-scores is marginally lower. Of the anthropometric measurements, waist circumference had the lowest compliance at each time point, with 15–28% of children declining this measurement. Children were asked to raise their clothing to access the waist; however, if they refused, an option was given to take the waist measure over thin clothing. When utilised, this was noted on the case report form (T0, n = 4; T1, n = 2; T2, n = 1).

Questionnaires

All child-completed questionnaires (the PedsQL, the Figure Rating Scale and CHU-9D) were completed at the home visit appointment and had good completion rates [ranging from 87% (T2, PedsQL score) to 100% (T0, CHU-9D score)].

Parent-completed questionnaires were split between those that were completed at the home visit appointment (the Children’s Dietary Questionnaire and the FNPA) and those for which parents had the option to complete after the home visit and return by post (from January 2016; the authoritative dimension of the Parenting Styles and Dimensions Questionnaire, Parental Self-efficacy scale and various subscales of the CFPQ). Completion rates for those questionnaires completed at a home visit ranged from moderate (62%, T1, FNPA score) to good (97%, T2, Children’s Dietary Questionnaire: fruit and vegetables score). However, those questionnaires that could be completed later and returned by post had substantially lower completion rates, ranging from poor (43%, T2, the authoritative parenting score) to moderate (67%, T0, parental self-efficacy score).

Physical activity measurements

Two types of accelerometer were tested: the wrist-worn GENEActiv and the hip-worn ActiGraph. The aim of using both accelerometer types in this feasibility study was to determine whether or not there was a difference in compliance between the two monitor types. GENEActiv monitors are less established in the literature than the hip-worn ActiGraphs; therefore, it was planned to use the GENEActiv monitors on approximately 75% of participants. Table 13 highlights the difference between the monitor types at each time point. At T0, the return rate and proportion of children who wore a monitor for at least 1 valid day (> 10 hours) were similar between monitors. However, at T1 and T2, both were lower for the hip-worn ActiGraph. For those who wore a monitor, the number of valid days of wear time recorded was similar between monitor types (Figure 9).

FIGURE 9 Physical activity monitor: median valid days of wear at each time point. (Figure not reproduced here; see the original article.)

Anthropometric measurements: parents and other family members

The purpose of undertaking anthropometric measurements (height, weight and percentage of body fat) with parents and other family members was to assess the feasibility of collecting data that would enable evaluation of the impact that the intervention had on the wider family. The collection of these data proved problematic. At baseline (T0), of the 75 families who had a data collection home visit, 64 (85%) had at least one other family member and 37 (49%) had at least two family members who consented to measurements. These proportions reduced to 78% and 34% for T1 and 60% and 28% for T2, respectively. A major issue was that the family members who consented to be measured at each time point were often not the same, that is, one sibling was measured at T0 and another sibling was measured at T1, etc. The implication of this is that it would not be possible to use these data to evaluate the impact that the intervention had on the BMI or percentage of body fat of other family members, as the data are so incomplete.

Costs associated with the intervention programme

Development of the facilitator manual

A manual for facilitators delivering the adapted programme was designed in phase I of the study. Small refinements to the manual will be made in response to the findings of the feasibility study, but the cost of doing so will be minimal and will be classed as a programme development cost, so will not contribute any further cost to the intervention if ‘scaled up’. Therefore, these costs were not included in the overall costs of the intervention.

Programme set-up costs

To deliver the intervention programme, two training sessions were held with the facilitators. This involved introducing the new materials and discussing the plan for each week. The costs associated with the training of facilitators were classed as set-up costs, as staff received training at the start of the programme but this was not ongoing. Hence, if the programme was rolled out more widely, some initial staff training would be needed but this would not be required on a regular basis.

For both the intervention and comparator programmes, items of equipment were purchased for use during the sessions. These equipment costs were also classed as set-up costs, as the equipment was then used at each subsequent session. Table 14 presents the set-up costs for both the intervention and comparator programmes. The total set-up costs were £178 for the comparator arm and £940 for the intervention arm. The additional costs for the intervention arm are mainly related to staff training and the production of display boards.

Delivery costs

There were several elements that were common to both the intervention and comparator arms, relating to the promotion of the programmes, project management and assembling materials. However, our focus was on collecting costs that were different between the intervention and comparator arms. These included the provision of adapted materials, the hire of premises and staffing costs.

Both the intervention and the comparator participants received printed materials. With the exception of the BMI charts, these were different for each arm. For the intervention arm, resource use linked to the provision of the materials was successfully collected. For the comparator arm, however, the resource use linked to the participant materials was challenging to collect, as this represented ‘routine practice’ for the BCHCT children’s weight management team. For example, it was challenging to separate out resource use that was linked to providing photocopied information for the comparator sessions with other routine photocopying undertaken by the team. As a consequence, the material costs for the comparator arm were estimated based on feedback from the BCHCT staff alongside analysis of the comparator session plans.

For both the intervention and the comparator arms, average resource use was estimated based on an assumed full attendance of families at each session. In other words, once a family had signed up to at least one session, it was assumed that materials would be prepared for them for each subsequent session. For the intervention arm, 169 families attended at least one session; this number was 74 families for the comparator arm. The session materials provided to the families in each arm, and the incremental cost associated with the provision of these materials for the intervention arm, are detailed in Table 15.

The cost of materials for the intervention arm was higher than that for the comparator arm. This was because the materials for the intervention arm were printed in colour and were bespoke products. Families in the comparator arm were given folders that had no cost for BCHCT, and all handouts were printed in black and white. Additional incremental costs for the intervention arm were attributable to the additional venue hire and staffing costs for the intervention compared with the comparator. The intervention involved additional support from an assistant and increased provision at the weekends. Details of these incremental costs are shown in Table 16.

Table 17 summarises the additional costs associated with the intervention, compared with the standard weight management programme. Costs are presented in terms of average additional (incremental) costs per session and per attender (family).

Costs borne by families

We attempted to capture the costs incurred by those attending the intervention and comparator programmes via a questionnaire survey delivered at the end of the programme. However, the questionnaire was completed by all those who attended the final session, including by parents who were not participating in the CHANGE study. The questionnaire was completed by 96 parents, 13 of whom had attended a comparator programme and 83 of whom had attended an intervention programme. Respondents were asked about the main activity they would be engaged in, if they had not attended the First Steps programme. This question was completed by 66% of respondents. Overall, the highest number stated that they would be looking after the home or family (n = 37 respondents) and only four people stated that they would otherwise have been in paid employment (all from the intervention group). Just one of the respondents who had taken time out of paid employment stated that alternative arrangements had been made to cover their work. None of the respondents stated that they had paid someone to look after dependants while they attended the programme.

As part of the questionnaire survey, data were collected on whether or not participants had noted a change in their weekly food bill since they started the programme. This question was completed by 53% of respondents. Of those who completed the question, a slightly higher number stated that they had noticed a change in their food bill (n = 28 respondents) than stated that there had not been a change (n = 23 respondents). Of those who noticed a change in the amount of their food bill, the numbers saying that the bill had increased (n = 8 respondents) and decreased (n = 10 respondents) were similar. The amount of change to families’ weekly food bills was cited as ranging from an increase in the weekly shopping bills of £25 to a decrease of £5.

Parameters to inform sample size estimation for a future trial to evaluate intervention effectiveness

Using the anonymised data on BMI z-score from the children’s weight management programme providers (n = 154), we estimated the ICC to be 0.03 (95% CI 0.002 to 0.40) at the end of the programme, and the correlation between start-of-programme and end-of-programme BMI z-score to be 0.98. Using data from only the CHANGE study participants, the correlation between baseline (T0) and end-of-programme (T1) BMI z-scores was 0.97 and between baseline and 6 months post intervention (T2) was 0.91.

Exploratory analysis of the effect of the adapted programme on outcome measures

Mean changes (and standard deviation) in outcome scores by study arm allocation are presented in Table 18. In both arms, the mean changes of all anthropometric outcomes were in a favourable direction at the end of the programme (T1) and at 6 months after the end of the programme (T2). Only very marginal changes were seen in physical activity indicators, with the average acceleration decreasing in both study arms and moderate to vigorous physical activity decreasing by 1 minute per 24 hours at 6 months in the intervention arm and increasing by just over 2 minutes per 24 hours in the comparator arm. The PedsQL scores increased in both study arms, as did the CHU-9D utility scores. Body dissatisfaction scores decreased in both arms (indicating a favourable direction of effect). The direction of change of dietary indicators was mixed in both study arms, but changes were marginal. The FNPA scores increased in both arms and there was a mixed picture of changes of the authoritative parenting style, parental self-efficacy and parental feeding practices scores (low numbers of participants completed these questionnaire items). The mean changes seen in the comparator arm were generally slightly larger than in the intervention arm. However, there are a number of issues that need to be taken into account when considering these findings. First, the feasibility study was not designed to evaluate programme clinical effectiveness. The intervention was refined during delivery and, because of the logistical issues we encountered at T0, some participants had their baseline measurements taken when they had already commenced the programme. In addition, there were very small numbers in the comparator arm and, of particular note, there was differential attrition across the two study arms (29% and 52% at 6 months in the intervention and comparator arms, respectively). The combination of these factors makes it impossible to draw any meaningful conclusions from the differences in mean changes in outcomes seen across the two study arms. In addition, regression to the mean may, in part, explain the changes seen.

We also undertook exploratory analyses to compare outcomes in intervention and comparator arms at T1 and T2 by developing mixed-effects linear regression models (using intention-to-treat analysis). The models are presented in Tables 19 and 20. At both time points, the point estimates of the mean differences were of small magnitude. In the adjusted models, all CIs of the estimates included a value of zero at T1. This was also the case at T2, with the exception of the quality-of-life physical functioning score and the CHU-9D score. Again, it is not possible to draw meaningful conclusions from these models because of the feasibility study design, issues with timing of the baseline data collection, small sample size and differential level of attrition in the two study arms.

Sensitivity analyses

Two imputation methods were undertaken to estimate missing values at T1 and T2: (1) baseline observation carried forward and (2) multiple imputation techniques. The sensitivity analyses for T1 and T2 are presented in Appendix 11 but, in brief, the mean differences in outcomes remained very small.

Analysis of anonymised body mass index z-score data from programme participants

Anonymised BMI z-scores for all programme participants within the study period were provided by the BCHCT weight management service team. Data were available only for children who attended the first and last sessions of the programmes, as this was when the measurements were undertaken (n = 154; n = 115 in the intervention group and n = 39 in the comparator group). Using these data, the mean changes in BMI z-score at programme end (T1) were –0.09 and –0.13 for the intervention and comparator arms, respectively. The mixed-effects linear regression analysis showed a mean difference of 0.05 (95% CI –0.01 to 0.10) in the model adjusted for baseline BMI z-score and clustering, and 0.04 (95% CI –0.02 to 0.10) in the model that was further adjusted for age, sex and ethnicity. This is similar to the analysis with the CHANGE study participants only (see Table 19). Again, there were differential levels of programme completion, which meant that there were fewer data available at the end of the programme in the comparator group than in the intervention group, which precludes any meaningful conclusions being drawn from this analysis.

Discussion

Summary of the feasibility study findings

Feasibility and acceptability of the adapted programme

The NHS children’s weight management provider service was successfully able to deliver the adapted intervention programme. The delivery structure of six 90-minute sessions over 6 weeks was generally found to be acceptable to participants and feasible in terms of cost and delivery for the service provider. The adaptation of more weekend provision was accommodated by the service providers, although there were higher costs attached to staffing and venue hire at weekends. Weekend sessions seemed to suit the participants who attended them, although some families still identified challenges to attending at weekends.

The facilitators showed a high degree of enthusiasm for the programme and valued the high-quality materials, the interactive and flexible nature of delivery and the inclusion of children in the programme. The participating families also positively viewed these aspects and, in particular, drew attention to the value of becoming connected with other families, sharing experiences and supporting each other in their endeavours to change behaviours within the family. These aspects of the programme were adaptations that were explicitly identified in the theoretical intervention development process in phase I of the study. In contrast to the phase I findings, one facilitator felt that the de-emphasis on weight management in the adapted programme was unhelpful and possibly lessened the impact that the programme had. However, the stigma surrounding overweight has been found to be a barrier to attendance of children’s weight management programmes100 and the need to reframe messages in children’s weight management programmes has been highlighted.101 This, together with our findings in phase I of the need for an emphasis on healthy lifestyles, rather than weight, supports our approach in the adapted programme.

The views of the programme facilitators and participants diverged when it came to the balance of nutrition and physical activity content. Despite adapting the programme to include physical activity sessions, participating parents and children wanted even more physical activities within the sessions and felt that these and other ‘doing’ activities, such as making healthy snacks, were the ‘fun’ elements of the programme. However, the facilitators felt that the nutritional content and messages had been somewhat lost from the standard programme and that there was too much emphasis on physical activity.

The session that was observed to be most problematic to deliver was the healthy eating session, and this was confirmed by the facilitator feedback. The session was refined in response to this and delivery was improved, although the facilitators still reported difficulties with its delivery at the end of the study period. Thus, further minor refinements to streamline the session may be required in the future.

The interviews with parents and children did not reveal any differences in programme acceptability across different ethnicities and, similarly, direct observation and the views of facilitators were consistent with the universal acceptability of the programme. This evidence supports the underlying principle of having flexibility and responsiveness in the programme, which was identified as important in the adaptation process.

A key indicator of the acceptability of the programme to participants was the estimation of the proportion of families who completed the programme. The definition of 60% attendance was adopted for this study in recognition of the fact that families with children may have other commitments or unexpected events that may prevent them from attending all sessions, even if they intend to. The primary outcome of interest was completion in Pakistani and Bangladeshi families, as the intervention adaptation process focused primarily on increasing programme acceptability for these families. However, a key secondary outcome was to estimate completion rates in families of other ethnicities and, overall, to ensure that in focusing the adaptation process on Pakistani and Bangladeshi communities, the programme had not become less acceptable to other families. The proportion of families who completed the adapted programme was, overall, 76% (79% of Pakistani and Bangladeshi families and 75% of other families). These point estimates of completion were much higher than for the comparator programme (in which 58% of all families completed the comparator programme) and also compare favourably to historical, routinely collected data on attendance (49% of families who attended at least one session completed the programme in the year preceding the study). These findings support the high acceptability of the adapted programme and are consistent with the qualitative data from interviews with programme participants. These levels of programme completion also compare favourably with what is reported in the current children’s weight management literature. A review of 23 studies of child weight management interventions reported a median drop-out rate of 37% (range 4–83%);31 however, these were all clinic-based interventions. In terms of community-based group child weight management programmes, reported programme drop-out rates have frequently been high (> 40%).102,103 In the UK, the MEND programme has been rolled out on a large scale in the last decade. This is a community-based group programme for 7- to 13-year-old children and their families, comprising 20 sessions. A study of participants, undertaken over a 4-year period, reported that 59% of participants completed the programme (defined as > 75% attendance).32

Feasibility of trial methods

Study design

The feasibility study was designed as a cluster RCT to enable assessment of the proportion of families that completed the adapted and the standard programmes in the two study arms. We were able to execute this study design within the existing structure of the children’s weight management service. The cluster design enabled participants to attend the most convenient programme for them and was appropriate for a trial in which both the intervention and the comparator are group-based programmes. However, in a future trial to evaluate clinical effectiveness, consideration needs to be given to the comparator arm. The difference in effect size between two programmes that one may expect to see may preclude undertaking such a trial because the sample size would need to be very large. A more feasible approach to a future clinical effectiveness evaluation would involve having a comparator arm that has no active intervention programme. In this scenario, the required study design would be a trial randomised at the individual level. Therefore, using the data generated by this feasibility study, we have undertaken a sample size calculation for a future individually randomised controlled trial that compares the adapted group-based programme with no intervention. One parameter required for this calculation, which we have not been able to estimate in the feasibility study, is the likely difference in effect between a group receiving the intervention and a group with no intervention. We have, therefore, drawn on available literature to determine this parameter. We selected a mean difference in BMI z-score between study arms of 0.15 because, although there are currently no data to support a clinically meaningful threshold for change in BMI z-score,104 significant improvements in a range of cardiovascular disease markers have been reported with this level of BMI z-score reduction.105

To test a mean difference in BMI z-score of 0.15 in a study population with a BMI z-score standard deviation of 0.6 (as reported in this study) at 90% power and 5% significance, allowing for a correlation between pre- and post-intervention measurements of 0.7 (which is a conservative estimate), 169 participants per study arm would be required. However, the proposed trial, although individually randomised in design, would have clustering in the intervention arm because of the group-based nature of the intervention programme and, therefore, given the findings of this study, attrition also needs to be taken into account in the sample size calculation. Assuming that the programme is delivered to groups of 10 and allowing for 30% attrition, the expected cluster size after attrition would be 7. Therefore, assuming an ICC of 0.03 (as estimated in this study), the sample size in the intervention arm would be 199 [= 169 × 1.18 (= 1 + 6 × 0.03)]. This equates to approximately 199 participants being randomly allocated to the intervention arm across 28 (= 199/7) programme groups and 199 participants being randomly allocated to a control arm.

Participant recruitment and follow-up

We achieved our participant recruitment target of 80 in this study. We did not expect difficulties in recruitment as we were recruiting families that had already agreed to attend a children’s weight management programme. However, the process of recruitment proved problematic for a number of reasons and resulted in difficulties in undertaking baseline assessments with families before they started a programme. In a future trial, it would be necessary to plan carefully with all partners involved in participant recruitment to ensure that there is enough time to gain consent and undertake baseline assessments before the start of the intervention programme. If an individually randomised design was adopted, then randomisation of participants would need to be undertaken after baseline assessments and before the participants start a programme.

Participant attrition was a major issue, with one-third of participants being unable to be followed up at the 6-month time point. The majority of these were lost to follow-up rather than having actively withdrawn from the study, despite multiple attempts to contact them by telephone and letter. The level of attrition seen in this study is within the range of attrition seen in other children’s weight management intervention studies. The Cochrane review12 of 64 studies reported that participant drop out at 6 months ranged from 1% to 42%, with only 31 studies reporting a follow-up of > 80%. The high attrition seen in this study may, in part, be explained by the characteristics of the study population: Birmingham has high levels of population turnover.106

Attrition poses a major challenge for a future trial to evaluate intervention clinical effectiveness. In this feasibility study, as well as there being a high level of attrition overall, there were differential levels of attrition in the two study arms (29% in the intervention arm vs. 52% in the comparator arm). This makes the interpretation of outcome comparisons between the two arms more complex and necessitates data imputation, as demonstrated in our exploratory outcome analyses. Differential attrition across study arms has been less commonly reported in other studies.12 There are no obvious solutions to the problem of attrition. We designed our data collection processes to minimise loss to follow-up by offering appointments in the participant’s home, which has been shown to be an effective approach to reduce attrition;107 thus, it was disappointing to retain only two-thirds of the participants until the end of the study. In future research, we would recommend requesting as many contact details as possible, including all telephone numbers and postal and e-mail addresses. Rewards for participants at the final follow-up point should also be considered. In addition, the sample size calculation of a future trial should take into account a significant level of attrition.

Outcome data collection

Overall, the collection of outcome data worked well. The home visits appeared to be acceptable to participants, and they were feasible to undertake. However, the visits were resource intensive, with a great deal of staff time spent travelling to participants’ homes and there being a necessity for two researchers on the first visit. This is a consideration in the design of a future trial.

One issue that arose with the collection of outcome data was the difficulty in concealing the study arm that the participants were allocated to from the researchers undertaking the assessment visits. At baseline, this was in part because some participants had already commenced a weight management programme before the data collection visit took place. In a future trial of similar design to this feasibility study, separation of the research staff undertaking the outcome measures and the core research team responsible for the delivery of the study overall would help in blinding those undertaking the measures. However, in a trial comparing a programme with no intervention, blinding of staff undertaking measurements at follow-up could be challenging.

The feasibility study was planned to enable us to explore the feasibility of collection of a wide range of outcome data. It became clear during the study that the number of questionnaire instruments that we were requiring parents to complete was too much of a burden. In general, home visits lasting for more than 1 hour became wearisome for the families and the researchers. In response to this, we prioritised certain questionnaire instruments to be completed at the visit and asked parents to complete the others at a later time and post them back to the research team. This significantly affected data completeness, with only 43–58% of participants completing the second group of questionnaires at the follow-up time points. The other issue with splitting the collection of questionnaire data in this way was that it did not work for non-English-speaking participants. In these cases, the researchers either stayed to complete all the questionnaires or, if the participant became too fatigued, they left the second group of questionnaires blank. There is also the possibility that the heavy burden on participants at the data collection visits contributed to the high attrition seen in the study. Thus, in a future trial it would be important to consider carefully which outcome data are essential and to resist the temptation to measure too many outcomes.

Overall, sociodemographic and anthropometric data were relatively complete, as were the questionnaire measures that were directly administered to the child participants. Apart from the FNPA questionnaire, the parent questionnaires that were administered during the home visit were relatively well completed. In a future full-scale trial, we would recommend the use of BMI z-score as the primary outcome, given the completeness of data collection; however, it would be important to include other anthropometric measures as secondary outcomes, given the limitations of BMI as a measure of adiposity.

The anthropometric measurements with parents and other family members proved particularly problematic. The number of family members agreeing to be measured reduced over the three time points, and often those agreeing at each time point were different family members. Collection of family data in a future trial would need careful planning with prior agreement at the outset over which family members will be present to be measured at each data collection time point. This would be extremely logistically challenging.

Process evaluation methods

The implementation of the adapted programme was directly observed by researchers, who attended all programme sessions and also observed delivery by both facilitators. In addition, there was contemporaneous feedback from the facilitators on their reflection on delivering the session and how it was received by the participants. We also undertook a series of interviews with facilitators and participants. In this study, these methods were used primarily to assess the feasibility of delivering the adapted programme and how acceptable it was to the participants, as well as enabling refinement of aspects of the intervention during the study intervention period. However, these methods also enabled the collection of key process evaluation data that would be required to assist in the interpretation of the evaluation of outcomes in a future trial. In line with the UK’s Medical Research Council guidance on the process evaluation of complex interventions,108 we clearly articulated the causal assumptions underlying the intervention programme in phase I of this study and have collected data that enable evaluation of intervention fidelity, the degree of tailoring, participant responsiveness to the various elements of the intervention programme and the influence of context on both delivery and participant response. The experience of collecting these process data, together with the previous process evaluation experience of the research team,109 will inform the development of detailed tools to assist in process evaluation in a future trial, such as observation checklists and standardised facilitator feedback forms.

Economic evaluation methods

The economic component of the study aimed to design processes for, and capture data on, the costs and outcomes associated with the intervention and comparator programmes to inform the methods for a future economic evaluation alongside a definitive trial. On the whole, processes were successfully put in place to measure the costs associated with the adapted programme. Data were successfully collected on the costs associated with producing the adapted programme materials. However, some challenges were also encountered. In particular, difficulties were encountered in terms of capturing data on resource use for the comparator programmes. As this represented ‘routine practice’ for staff delivering the programmes, it was difficult to separate out the costs associated with conducting the standard programme from other day-to-day activities. In a future trial, if the same design were adopted, we would recommend having separate processes for capturing the costs associated with producing materials for comparator programmes.

As part of the study, we aimed to collect data on the costs borne by families in attending the weight management programme. These data were collected via a questionnaire distributed at the end of the programmes. Problems were incurred by collecting data in this way. Very few questionnaires were returned by those attending the comparator programmes and the questionnaire was completed by all programme participants at the final session, rather than by only those who had participated in the CHANGE study. We would recommend that, in a future trial, these data are collected alongside the outcome data rather than separately at the end of the programme.

The core outcome data that could be used in a future economic evaluation were collected successfully (see Outcome data collection). This would permit different types of economic evaluation to be conducted in a future definitive trial. First, a cost-effectiveness analysis could be conducted, using the primary clinical outcomes included in the trial, for example estimation of the cost per mean change in BMI z-score. In addition, collection of CHU-9D data would allow a cost–utility analysis to be undertaken as recommended by NICE. However, as discussed (see Outcome data collection), it was not feasible to collect the data to capture the impact that the intervention had on the wider family (anthropometric measurements on parents and other family members). The design of the economic evaluation alongside a future trial would adopt two approaches: a within-trial analysis that would estimate the cost-effectiveness using data generated by only the trial on the basis of an intention-to treat principle and, if appropriate, a longer-term model-based economic evaluation extrapolating both the costs and outcomes over a lifetime. The model-based economic evaluation would need to follow standard guidelines for decision-analytic modelling by adequately capturing the uncertainty with long-term intervention effects, future costs and outcomes linked to disease associated with being overweight and structural uncertainty within the model.

Exploratory analyses of intervention effect on participant outcomes

This feasibility study was not designed to assess intervention clinical effectiveness; however, we undertook exploratory analyses of our outcome data. Although we did show a favourable direction of intervention clinical effect at the end of the programme and at 6 months, the effect size was small. Given the continuing refinement of the intervention programme during its delivery, the high and differential study attrition and the lack of ‘true’ baseline data for some participants (as some T0 measurements were taken after the start of the programme), we cannot draw any conclusions from these analyses, other than that there is a signal that the adapted programme results in some reduction in BMI z-scores and other anthropometric measures.

Strengths and limitations

The cluster randomised design of this feasibility study enabled a detailed evaluation of both the feasibility of delivery and acceptability of the adapted intervention programme and the feasibility of the processes and methods required for a future clinical effectiveness and cost-effectiveness evaluation. Estimation of participant retention gave an indication of overall programme acceptability. The inclusion of a comparator arm gave us the opportunity to directly compare participant retention in the adapted programme with that in the standard programme. In their evidence synthesis on the adaptation of health promotion programmes for minority ethnic groups, Liu et al.35 highlighted the need for this type of direct comparison. The qualitative evaluation methods that we employed enabled us to explore, in detail, the engagement of programme deliverers and participants and resultant change in behaviours. We were also able to use this information to refine the programme within the feasibility study.

We were able to test recruitment and follow-up over 6 months, as well as the processes involved in the collection of a wide range of outcome measures and cost data that would need to be collected in a future economic evaluation of the programme. We gained valuable insights into the areas that would require particular attention when designing a future trial and what modifications would be required to the methods used in this study.

Although our adaptation was primarily focused on increasing the relevance and acceptability of the children’s weight management programme to families from Pakistani and Bangladeshi communities, we tested its acceptability in an ethnically diverse population, predominantly from socioeconomically disadvantaged areas. This was important, as the adapted programme was designed to be delivered flexibly and to be responsive to families from all ethnic and cultural backgrounds. In addition, we designed the study to be inclusive of participants who did not speak English. The use of community researchers from Pakistani and Bangladeshi communities at the data collection visits was a particular strength, with these community researchers providing language skills to, and having a shared cultural understanding with, families from these communities.

There are limitations to this feasibility study. The programme adaptation process focused specifically on families from Pakistani and Bangladeshi communities. Therefore, we cannot assess whether or not a programme that was adapted with a broader focus would be even more acceptable and relevant to other families. However, the study was designed to focus on Pakistani and Bangladeshi families, as data from the existing weight management service in Birmingham highlighted that families from these communities were more likely to drop out of the programmes than families from other communities. In addition, our findings indicate that all families were more likely to complete the adapted programme than the standard programme, indicating high acceptability across all ethnicities, and this is supported by the qualitative data obtained from the feasibility study participants.

We did not focus on initial attendance at the programme, once a family had been booked to attend. Within this study, some adaptations were made to the programme recruitment materials and processes but these were applied to both intervention and comparator programmes. Despite this, initial programme attendance remained low, with < 50% of the families that were booked to attend actually attending any sessions in both arms of the study. Even of the families who agreed to participate in the CHANGE study, 20% did not attend any First Steps programme sessions. This is clearly another concern for the future provision of children’s weight management services, but it was beyond the scope of this feasibility study and, thus, needs to be a focus of future research.

The cluster randomised design of the study enabled us to assess programme completion in both the adapted and the standard programmes. However, it did not enable us to evaluate the acceptability to study participants of individual randomisation or allocation to a control arm in which no intervention is received, neither did it provide us with an estimate of difference in effect between intervention and control (no intervention) arms to inform the sample size calculation of a future trial. The study design incorporated home visits to collect data from participants in order to minimise attrition. This is a resource-intensive approach and would increase the cost of a future trial. In addition, despite this approach, there were high levels of attrition. During the study, we did consider an alternative and potentially lower cost method: attending the first and last programme sessions to undertake data collection with study participants. However, after discussion with the Study Steering Committee, it was agreed that this approach would present problems in blinding researchers to the study arms that the participants were in and participants who did not attend or complete the programmes would still require home appointments for data collection.

Consideration of progression to a trial to assess clinical effectiveness and cost-effectiveness

Within the CHANGE study protocol, we specified that we would consider the following factors when making the decision to proceed to a full clinical effectiveness and cost-effectiveness trial of the adapted children’s weight management programme:

• the estimate of the proportion of families completing the modified programme (Pakistani and Bangladeshi families and all families)
• the feasibility and acceptability of the culturally adapted components of the programme to the ethnically diverse study sample
• the feasibility and acceptability of trial processes
• the feasibility and acceptability of outcome measures and other evaluation methods
• the identification of appropriate methods for an economic evaluation in a full-scale trial.

We have demonstrated increased participant completion of the adapted programme compared with the standard programme, both in the feasibility study and also compared with historical service data. In addition, our qualitative data have shown that the programme is feasible to deliver and highly acceptable to service providers and participants. There was an incremental cost associated with the delivery of the adapted programme, but this was relatively modest (£33 per family attending).

We were able to successfully recruit participants from a pool of families referred to the Birmingham children’s weight management service. However, the number of families that withdrew or were lost to follow-up was high, which has implications for the feasibility of a full-scale trial. We were able to successfully undertake a cluster randomised study design with interventions being delivered in both study arms. However, as previously discussed (see Study design), the likely design of a future trial would involve individual randomisation with no or minimal intervention in the control arm, and this study does not provide data on the acceptability of these trial processes.

Data collection methods overall were feasible; however, in a future trial, fewer outcome data would need to be collected to minimise participant burden. The process evaluation methods that we undertook were both feasible and acceptable and yielded valuable data that would assist in the interpretation of outcomes in a future trial. We also successfully collected cost and outcome data, which would be required for a future economic evaluation of the adapted programme.

We have, therefore, broadly fulfilled our predefined progression criteria. However, we also feel that it is important to consider the changing commissioning landscape of children’s weight management programmes in the UK. In the last 15 years, the commissioning of community-based children’s weight management programmes that address weight-related behaviours has been widespread and the NICE guidelines,17 published in 2013, which were based on extensive and systematic evidence reviews, explicitly recommended the provision of family-based multicomponent lifestyle weight management services for children and young people. However, in the last 2 years, with reductions in public health budgets, a lack of long-term evidence for the effectiveness of programmes and difficulties with uptake,36 anecdotal evidence suggests that the provision of such services has become less widespread. In order to undertake a future trial of our adapted intervention programme, we would require collaboration from commissioners, which may prove challenging in the current climate.