Feasibility cluster randomised controlled trial phase 2: methods

Parts of this chapter have been reproduced from Kipping et al.1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

This chapter describes the methods used in the feasibility cluster RCT.

Sample size, eligibility, recruitment and randomisation

The sample size for this feasibility study was not informed by a power calculation. The sample size of 12 nurseries was chosen to provide information on variability within and between nurseries at baseline and follow-up, response rates and intracluster correlation coefficients (ICCs) in anticipation of a larger trial. A larger sample would not have been appropriate for a feasibility study and 12 participants was deemed a sufficient sample for the process evaluation, which was the most important part of this feasibility study.

The study took place in nurseries in two areas of England (North Somerset and Gloucestershire) and in the homes of children recruited to the study. North Somerset is a rural area adjacent to Bristol, with 14.1% of children living in poverty (per cent of children aged < 16 years in families receiving means-tested benefits and low income in 2012).71 Gloucestershire is a large rural county to the north of Bristol. The health of people in Gloucestershire is generally better than the England average; however, 13.8% of children live in poverty.72

The inclusion criteria for child-care providers were that they must be a day nursery, private nursery school, maintained nursery school, children’s centre with nursery, or preschool, in North Somerset or Gloucestershire. Excluded child-care settings were child minders, crèches, playgroups, primary school reception classes (where schools operate an early-admission policy to admit children aged 4 years) and au pairs. Settings were eligible if they had a minimum of 20 children aged 2–4 years who attend the child-care providers for at least 12 hours per week over 50 weeks of the year, or 15 hours per week in term time.

The NAP SACC UK study was discussed with North Somerset child-care provider managers at meetings convened by North Somerset Council and advertised in the Council’s early years newsletter. These opportunities were not available in Gloucestershire because of timing of recruitment. Child-care providers were sent a letter from the Council, project information sheet, reply envelope and form indicating if they wish to participate and reason for their response. Non-responders were followed up with a reminder and then a telephone call. All interested child-care providers were contacted by telephone to discuss the study following which, if the provider was still interested, they were offered a visit to discuss the intervention and study in more detail. A £200 incentive was provided to all participating nurseries at the end of the study.

Randomisation involved allocation to two arms: NAP SACC UK or usual practice. Twelve child-care providers (six in each area) were recruited to the trial, of which six were randomised to the control group and six to the intervention (three in each area). Allocation was conducted by an independent statistician at the Bristol Randomised Trials Collaboration, blind to the identity of the child-care providers. Stratified randomisation was used to ensure balance for (1) deprivation (using IMDs for the local super output area for 2010, where the provider is located), (2) size of child-care provider [small (≤ 48 children) or large (> 48 children)] depending on number of children attending, details obtained from the Council, Ofsted website or the nurseries own website] and (3) location (North Somerset and Gloucestershire). To achieve reasonably equally sized groups for randomisation, 12 groups were created across each of the two areas, with six lists created for each of low, medium and high IMD score. Each of the six lists by IMD category combinations was divided into two equal groups according to size.

Parents/carers were given a letter, project information sheet, reply envelope and form indicating if they wish to give consent for their child to take part in the study (sent at the end of August 2015). Research staff attended all nurseries late afternoon on between 1 and 3 days (depending on staff availability) to talk to parents about recruitment and answer any questions. All children received a small thank you gift [packet of stickers or a Mr Men™ (Mister Men Limited, London, UK) book] after each of the two data collections to encourage the prompt return of accelerometers.

The NAP SACC UK intervention

The NAP SACC UK intervention is based on the online Go NAP SACC intervention73 (but delivered in person) and was adapted for use in the UK during phase 1 of the feasibility study. It is based on social cognitive theory and the socioeconomic framework. In the feasibility study, the intervention was delivered by health visitors (NAP SACC UK partners) and local experts delivered the workshops for nursery staff. The intervention used in the NAP SACC UK feasibility trial is summarised using the Template for Intervention Description and Replication (TIDieR)74 12-item checklist in Table 5. The intervention also contained a home-based component, informed by other studies of behaviour change with parents/carers of young children use of digital media, and interviews and focus groups with parents/carers, nursery managers and health visitors. The intended behaviour change techniques used in the different elements of the NAP SACC UK intervention are outlined in Appendix 5, Table 32.

Outcome measures

Primary outcomes

The primary outcome of this feasibility RCT was the acceptability of the intervention and the trial methods, assessed against pre-set progression criteria (see Progression criteria). Progression criteria were agreed with the NIHR Public Health Research funding board and an independent TSC and were assessed using a range of methods.

Secondary outcomes

The secondary outcomes were measured at baseline (T0) prior to the intervention and 8–10 months after the baseline measurements (T1) (follow-up was deliberately staggered to assess the feasibility of following up children aged 4 years as they moved to school). These assessments were to inform the choice of primary outcomes for a full-scale trial and particularly whether or not the outcomes require data collection from parents/carers and children, or if the outcomes could be the environmental audit and zBMI using data linkage from the National Child Measurement Programme (NCMP). The outcomes included:

• child accelerometer-measured activity
• children’s height and weight
• child food and drink intake, specifically of fruit and vegetables, snacks and sugar-sweetened drinks
• child screen time
• EPAO instrument score
• parental and nursery staff mediators
• family costs
• health-related quality of life (HRQoL).

Data collection

As this feasibility study was not designed to assess efficacy, effectiveness or cost-effectiveness, data collection was designed primarily to assess feasibility and acceptability. All measurements were completed at baseline and again at follow-up, except for the baseline nursery audits and demographic information provided on consent forms. Data collection methods comprised the following.

Accelerometry measured activity

We used ActiGraph GT1M accelerometers (Actigraph, Pensacola, FL, USA), which have been extensively validated for assessment of PA among children. Accelerometers were fitted by NAP SACC UK researchers on the day of data collection, and instructions for their use were given to parents/carers on the same day.

Anthropometric measures of children

All anthropometric measurements were completed by DBS-checked trained fieldworkers with a member of nursery staff present. Weight was measured without shoes in light clothing to the nearest 0.1 kg using a Seca digital scale (Seca, Birmingham, UK). Height was measured to the nearest 0.1 cm without shoes using a portable Harpenden stadiometer (Holtain Ltd, Pembrokeshire, UK). Fieldworkers were trained to ensure correct position for height assessment. The ethics committee requested that the research team wrote to nursery managers alerting them to any children who were on or above the 99th zBMI centile for their age and sex, with context regarding the concerns about obesity in children, and advising them to follow their usual child protection procedures.

Children’s food and drink intake specifically fruit and vegetables, snacks and sugar-sweetened drinks

Dietary assessment was performed using the Child and Diet Evaluation Tool (CADET) prospective ticklist record for all foods consumed in a 24-hour period, an instrument validated for use in intervention studies with young children.75,76 The CADET assesses the intake of 15 food groups, and CADET data were recorded by researchers undertaking observation of the children at all snacktimes and mealtimes at the nurseries and by parents/carers completing the CADET form at home for the child’s food and drink consumption before and after nursery. At baseline, the data were collected on paper forms with four children observed by one researcher. At follow-up, the data were recorded on paper and transferred to Google Nexus (Google Inc., Mountain View, CA, USA) tablets in the nursery. In a sample of four nurseries, CADET data collection at home was piloted over a weekend in addition to the weekday of nursery data collection; for two of these nurseries, the data were returned by parents on paper and in two nurseries they were collected from the parents by a researcher over the telephone. For the two nurseries where CADET was provided over the telephone, a NAP SACC UK researcher made up to four attempts to contact the parents/carers by telephone.

Sedentary time

Sedentary time was assessed by gathering details of screen time and quiet play time on a weekend day; the questions were based on those used in the Toybox study.77 Sedentary time was assessed via questions at the end of the CADET tool completed by parents/carers.

Environment and Policy Assessment and Observation instrument score

The EPAO instrument was developed for the NAP SACC programme and assesses child-care nutrition and PA environments, policies, and practices. It was developed using the standards, recommendations and research literature on which the NAP SACC intervention itself was based. It has been tested for validity and reliability in nursery settings in the USA.78 The EPAO, modified to accommodate the expanded best practices in 2014, consists of a 1-day observation and review of pertinent centre documents using 189-item questions and 16 free-text sections, with the average of all subscale scores representing total nutrition, indoor play, outdoor play and sedentary time scores. The EPAO from the USA was adapted to ensure consistency with the NAP SACC UK intervention and consistency with English descriptions of food. The EPAO instrument was administered at each of the nurseries by researchers who received specific training. This task was usually shared between two fieldworkers (morning and afternoon) as the form was long and required constant attention.

Parental and nursery staff mediators

Parental and nursery staff knowledge (e.g. nutrition, oral health, PA and sedentary behaviours), self-efficacy and motivation were assessed using tools created for this study. The reliability and validity of the tools are being explored in a separate study to inform whether or not they need further refinement for use in a full-scale trial. A paper questionnaire covering parent/carer mediators was completed by parents/carers.

Data linkage

The feasibility of zBMI data linkage with the NCMP data held by Local Authorities in England66 in the first year of primary school at age 4–5 years was examined in principle. All parents/carers of consented children were sent a further consent form in the post at the end of the study to request consent to link their child’s height and weight data measured through the NCMP (the data linkage did not take place because the study finished before the data were collected in schools).

Intervention data collection procedures comprised the following.

The home component

This was evaluated with respect to data from the record of parental use of the website, goal-setting, sent text messages, sent e-mails and Facebook group activity.

The Review and Reflect tool

Nursery staff completed a Review and Reflect tool at the beginning and end of the intervention. This provided an indication of the staff’s assessment of any changes in the nursery environment, policy and practice relating to nutrition, PA, sedentary behaviours and oral health. This tool is based on the original and revised NAP SACC self-assessment tool.

Goal-setting forms

Nursery staff completed a form to outline their eight intended goals at the beginning of the intervention and this was sent to the research team.

Data were collected by the research team and casual fieldworkers. Training for all staff took place prior to data collection, which covered background of the trial, the structure of nursery visits and role of the fieldworkers, practice measurements in pairs, an explanation of the study protocols and standard operating procedures including the call-in protocol, the policy on working hours, the lone working policy, and the protocol for concern about a child observed during fieldwork. Fieldworkers for follow-up data collection were trained prior to data collection. Google Nexus tablets were trialled for collecting nursery CADET data at all nurseries during follow-up data collection. The tablets were encrypted and made secure with a passcode. Data were entered offline and uploaded later when the tablet was connected to the internet.

Data storage and management

The secure, web-based electronic data capture system REDCap was used to store study data. This tool was created specifically for clinical research. The study team built and modified electronic forms in the database with support from a university clinical data systems manager.

Data from questionnaires were stored in anonymised form using participant identification (ID) numbers. REDCap contains functionality that allowed the team to e-mail survey links directly to participants. It was used in this way to collect follow-up parent/carer questionnaire data, for those parents/carers who provided an e-mail address on their child’s consent form (the majority). NVivo data analysis software was used to store, code and undertake thematic analysis of anonymised transcripts of the process evaluation data. All data collected will be made freely and openly available for other researchers to use via the data.bris Research Data Repository (where consent was given for data to be shared); this will take place at an appropriate time with respect to a possible full trial.

Data analysis

The statistical analyses were primarily descriptive, providing estimates of eligibility, recruitment, intervention delivery and retention rates among this study population, with 95% confidence intervals (CIs) calculated to incorporate between-provider variation where appropriate. Summary statistics were calculated for the outcome measures using means and standard deviations (SDs) by allocation arm and key demographic variables to inform the sample size and recruitment plan for the main trial, if warranted. Differences were explored for what was identified to be the primary outcome for a future trial by study location (North Somerset/Gloucestershire) and deprivation (high vs. low). Comparisons were made between those who completed the study and those who dropped out to investigate if this is a potential source of bias. Missing data were not imputed for the purposes of the feasibility study. However, the extent of missing data was examined and described to inform the full trial. Stata^® v14 (StataCorp LP, College Station, TX, USA) statistical software was used for all analyses.

As this was a feasibility study, the sample size was not based on a power calculation, so the sample cannot provide a usefully precise estimate of the intervention effect. However, we were able to determine likely recruitment rate of nurseries and children, adherence to the intervention, and intracluster correlations in anticipation of a larger trial.

For qualitative data we aimed for a maximum variation sample of interviewees to achieve a broad perspective on the salient issues. Emergent issues from earlier interviews and focus groups were explored in subsequent interviews and the number of interviews undertaken was determined by data saturation (no new issues or themes emerging from within/across participants). All interviews and focus groups were audio-taped and transcribed verbatim. All the transcripts were read and reread in order to gain an overall understanding of participants’ views and experiences. Data were coded in NVivo software, and analysed thematically, allowing comparisons to be made within and across the interviews. Quotations which best represented the nature of each theme were extracted. In the process evaluation of the feasibility trial a total of 40 interviews were carried out and transcribed. One transcriber working within the University of Bristol was paid to transcribe 11 interviews; an outside transcriber with previous experience of working with the University transcribed 24 interviews, and the remaining five interviews were transcribed by one of the NAP SACC UK team members.

Physical activity

Children wore accelerometers for 5 days including week and weekend days. Periods of 60 minutes with zero values were interpreted as time that the monitor was not worn.79 A day was considered valid if 8 hours of data were recorded.80 Mean minutes of sedentary time (using two thresholds of 0–25 and 0–199 counts per 15 seconds using the criteria proposed by Evenson and Puyau81,82 were used to inform choice in a full trial). Mean minutes of light, moderate to vigorous intensity PA were then processed (thresholds of 200–799, and ≥ 800 counts per 15 seconds). Mean accelerometer counts per minute (which provide an indication of the overall volume of PA in which the children engage) were also calculated as this approach facilitates comparison with studies that may have applied a different cut-point. The accelerometer data were checked for outliers; informed by previous studies with children we excluded implausibly high values, such as might occur when a participant uses a trampoline, using a cap of 11,714 counts per minute.83 Any values above this were capped at this maximum number. Consequently, the activity was still recorded as vigorous activity, but any biologically implausible peaks of activity were removed.

Anthropometric measures of children

Analysis consisted of height, weight, zBMI and proportion of overweight and obese (as determined by the UK1990 age and gender reference charts84 at 85% and 95% centiles, respectively), with further sensitivity analysis conducted using the International Obesity Task Force thresholds.85

Children’s food and drink intake specifically fruit and vegetables, snacks and sugar-sweetened drinks

Colleagues at the University of Leeds, who designed the CADET tool, coded and analysed the CADET tick list food diary in DANTE, a Microsoft Access^® (Microsoft Corporation, Redmond, WA, USA)-based food diary analysis programme designed by the Nutrition Epidemiology Group at the University of Leeds.75 Frequency was assumed to be equal to one portion. Portion sizes were based on average food intakes of children from the National Diet and Nutrition Survey86 and were age and gender specific. Nutrient values were obtained from the food composition tables of McCance and Widdowson and their supplements.87

We assessed children’s diet quality based on their adherence to the nationally recognised Children’s Food Trust voluntary food and drink guidelines for early years settings.28 This is in line with NAP SACC UK aiming to increase nursery staff’s knowledge of nutrition and improve dietary provision consistent with the Children’s Food Trust. Furthermore, examining diet by way of adherence to dietary guidelines instead of total intake means that the results can be interpreted against these guidelines and, therefore, have practical implications for public health policy. The Children’s Food Trust guidelines are underpinned by a nutrient framework88 and encompass four food groups on which to base meals and snacks: starchy foods; fruit and vegetables; non-dairy protein sources (meat, fish, and alternatives); and milk and dairy foods, with additional guidance on desserts, pudding and cakes, drinks, fat, salt and sugar. Appendix 6, Table 33, contains a list of food items included in each food group as defined by the Children’s Food Trust. Consequently, some food items were included in more than one food groups (e.g. baked beans were included in both the fruit and vegetables and the non-dairy protein sources food groups).

We operationalised these guidelines (hereon known as NAP SACC UK Nutrition Best Practice Standards) when possible, based on the number of main meals and snacks children consumed on the day of observation separately. Main meals included breakfast, lunch or tea (dinner), and snacks included morning and afternoon snack. We assigned children a score of 1 for each respective standard they met and 0 if they did not meet a standard. The scores were summed to derive an overall NAP SACC UK Nutrition Best Practice Standards score (range 0–9 for two meals, 0–8 for one meal, 0–5 for two snacks and 0–3 for one snack).

Individual dietary intakes of fruit and vegetables, snacks and sugar-sweetened drinks during nursery attendance were assessed, but we were not able to accurately calculate intake of these foods while children were at home. Although parents were asked to report their children’s food intake using a home food diary, the response rate of this assessment was low, and many parents filled in the diary on a different date from the nursery observation. As a result, we were not able to estimate the daily dietary intake of the aforementioned foods in order to compare it by setting (nursery vs. home), in line with the presented results on PA. Therefore, and also taking into account that the intervention involved training nursery staff to follow the CFT’s guidelines, the NAP SACC UK Nutrition Best Practice Standards score was considered as the dietary outcome of the study. We computed descriptive statistics for adherence with the NAP SACC UK Nutrition Best Practice Standards by intervention arm.

Screen time

Screen time and time spent in quiet play were separately categorised into < 1 hour, 1–2 hours and > 2 hours. In the absence of UK guidelines for screen time this categorisation was informed by the American and Australian guidelines that children aged 2–5 years should have screen time or television time limited to 1 hour89 or < 1 hour90 per day.

EPAO

Analysis consisted of total EPAO score, EPAO subscores and scores for the components identified for change by the nursery.

Parent-completed questionnaire about child and family

Details were summarised on parent socioeconomic status, family variables (siblings), hours child spends at nursery per day and per week, age of child, parent ethnicity and child ethnicity.

Health economics

Our aim was to pilot measures of resource use and estimate more precisely the cost of the intervention to inform a full-scale trial. We measured key cost components for the public sector, child-care providers and parents. We delineated resource use (e.g. hours), unit costs (e.g. cost per hour) and calculated mean provider and parental costs in the intervention and control groups. We estimated incremental costs and 95% CIs for descriptive purposes. We anticipated that the NAP SACC UK intervention would be unlikely to affect HRQoL in the short term, and any effect on HRQoL is more likely in the long term through changes in dietary habits, PA and BMI. However, we piloted a measure of HRQoL in order to explore any changes over the course of the trial and examine associations between PA and HRQoL.

The HRQoL was measured using the Paediatric Quality of Life Inventory (PedsQL) 4.0 for children aged 2–4 years91 at baseline and 8–10 months later. The instrument has been tested for reliability and validity in community settings91 including comparing HRQoL between preschool children with or without obesity.92 The inventory has 21 items rating HRQoL in four domains: physical function, emotional function, social function and nursery function. Parents rated their child’s functioning related to each item on a scale of 0 to 4 (0 = never a problem, 1 = almost never a problem, 2 = sometimes a problem, 3 = often a problem, 4 = almost always a problem). The PedsQL items were reverse scored and linearly transformed onto a scale of 0–100, for ease of interpretation, so that higher PedsQL scores indicated better HRQoL. Missing observations were imputed, as per the PedsQL scoring instructions.93

Family expenditure on PA of the 2- to 4-year-old child was collected from the parent questionnaire at baseline and 8–10 months later. Parents were asked to list all the physical activities their 2- to 4-year-old child had participated in over the past week and the associated cost, if any, of each activity (entry fees, membership, etc.). A subsequent question asked how many miles in the car were travelled to get to the activity. Expenditure on transport was calculated using the HM Revenue and Customs tax rates per business mile, costed at 45p per mile.94 The cost of transport to the activities was added to the expenditure on physical activities to generate the total weekly expenditure on PA of the 2- to 4-year-old. Parents were also asked to estimate the time spent on their child being physically active including travel and waiting time. This time was converted into hours per week.

Data on food expenditure were collected through the parent questionnaire, where parents were asked to estimate how much they spent on food in either the last week or the last month. Data were converted to weekly expenditure. Food expenditure was separated into five domains: supermarket shop, convenience shop, takeaway food, café/restaurant food and other (e.g. school dinner money). Parents were also asked if the spending on food in that period had been typical.

Health-care use was collected through the parent questionnaire in which parents were asked to detail visits to health-care providers in the last month. These data were collected at baseline and at the 8- to 10-month follow-up. This included, general practitioner (GP)/family doctor, therapist (e.g. physiotherapist, dietitian) and hospital inpatient and outpatient visits. The costs of health-care use of the 2- to 4-year-old children were calculated by multiplying the mean number of visits by the associated Department for Health and Social Care NHS Reference Costs 2015–201695 and the Personal Social Services Research Unit Health and Social Care Unit Costs 2016.96 At follow-up, a subsample of 35 parents were randomly selected (by the trials unit) to recall health-care use over the past 5 months (rather than the last month) to compare parent recall. These data were divided by five to give a monthly average.

Data were collected on parent time taken off work and the associated costs. Parents were asked how many days they had taken off work in the past month associated with the health of their 2- to 4-year-old child. Time taken off work was separated into three domains: paid time off work, unpaid time off work and annual leave days off work. The cost of paid time off was calculated by taking the average weekly salary from the Average Salary of Hours and Earnings Survey 2016.97 The national median weekly salary of £539.00 was divided by five to provide a day rate of £107.80. This number was then multiplied by the number of paid days’ work parents took off because of their 2- to 4-year-old child’s health.

The costs to nurseries of preparing for and delivering the intervention were estimated using the reported time that nursery staff attended the workshop, recorded time with the health visitors and additional information from the nursery logs. The cost of the intervention to the nurseries was estimated using the average hourly rate for nursery workers and managers.98 However, it was at the discretion of the nursery how they pay staff to attend the workshop. Not all nurseries paid their staff to attend the workshop.

All data relating to cost and HRQoL were combined with the parental mediator questionnaire and given to parents by nursery staff to return by post to the research team. At follow-up, all parents/carers who had provided an e-mail address on their child consent forms were e-mailed a link to complete the parent/carer questionnaire online, using the study’s database. Non-responders were sent a paper copy. Data were analysed using Stata MP 14.2 statistics and data analysis package. Mean differences at follow-up were tested using paired t-tests for food expenditure, PA expenditure, PedsQL score and child health-care costs. Costs and outcomes are presented in a cost–consequence table.

Process evaluation

Qualitative data collection and analysis

The aim of the process evaluation was to examine the feasibility and acceptability of the NAP SACC intervention and study design according to three prespecified progression criteria (see Chapter 2, Progression criteria). The design of the process evaluation used three relevant parts of the RE-AIM framework (Reach, Adoption and Implementation).99 The key process evaluation questions (progression criteria) and the data collection methods used are outlined in Table 6. Data were collected through (1) semistructured interviews with key stakeholders (i.e. NAP SACC partners, nursery managers and staff, and parents), (2) non-participant observations (of nursery workshops and meetings between NAP SACC partners and nursery managers) and (3) document analysis of NAP SACC partner logs, Review and Reflect forms and goal-setting forms.

Non-participant observation

All observations were conducted by Jane Collingwood (trial co-ordinator, maternity cover), with the exception of the NAP SACC UK partner training and one nutrition workshop (in nursery 3), both conducted by Rowan Brockman (research associate). The purpose of the nursery workshop observations was to assess the fidelity, feasibility and acceptability of the workshops by documenting its delivery and its receipt by nursery staff. Observations focused on who and how many staff attended, the workshop format, elements that worked well/less well, and the behaviour, interest and engagement demonstrated by the workshop facilitators and participants. Observations were recorded using a semistructured observation schedule in order to document prespecified information while also allowing for unstructured qualitative observations to be included.

The purpose of NAP SACC partner/nursery manager meeting observations was to assess the fidelity, feasibility and acceptability of these meetings and focused on the style of interaction (e.g. mutual decision-making or led by participant, supportive or critical, etc.), aspects that worked well/less well, and the behaviour, interest and enthusiasm displayed by the NAP SACC partner and manager. Observations were recorded using a semistructured observation schedule.

Semistructured interviews

Semistructured interviews were conducted with key stakeholders (NAP SACC UK partners, nursery managers and staff, and parents) to assess the fidelity, feasibility and acceptability of both the intervention and the research design. We aimed to interview all NAP SACC UK partners and all nursery managers, as well as one member of nursery staff from each intervention nursery. We also sought to interview 20 parents from both intervention and control nurseries.

Interviews with NAP SACC UK partners, nursery managers and nursery staff were conducted face to face or over the telephone, according to the participants’ preferences. All parent interviews were conducted over the telephone. Semistructured interview guides specific to each stakeholder group (Table 7) were used for all interviews to ensure that key topics were covered while allowing participants to discuss other issues they felt were important. Interviews were audio-recorded on an encryptable device and transcribed verbatim. All interviews were conducted by Rebecca Langford, with the exception of four parent interviews conducted by Ruth Kipping.

Document analysis

Three sets of documents were collected and analysed as part of the process evaluation: NAP SACC UK partner logs, Review and Reflect forms and goal-setting forms. NAP SACC UK partners were asked to log costs incurred in their participation in the NAP SACC UK study. Partners logged the date, type (e-mail, telephone, face-to-face meeting) and duration of contacts with each nursery. They also recorded travelling distance and time for the economic evaluation. These logs were analysed alongside other process evaluation data in relation to the key progression criteria.

Review and Reflect forms were either completed individually by the nursery manager and then discussed with the NAP SACC UK partner or completed alongside discussion with the partner during one of their meetings. These forms were reviewed and updated at a follow-up meeting between the manager and partner. Data from these forms were used to document changes made within the nurseries and to triangulated with other forms of data (e.g. interviews).

Forms documenting the goals set by the nurseries were also collected and compared against data from the follow-up Review and Reflect forms and interview data to assess progress made.

Data analysis

Interview transcripts were entered into NVivo software to aid data management and analysis. Qualitative interview data were analysed using techniques associated with thematic context analysis and grounded theory. Initially, Rebecca Langford read through all transcripts to familiarise herself with the data. Rebecca Langford then coded four transcripts line by line to identify themes and develop an initial coding framework. Rebecca Langford and Ruth Kipping independently applied this coding framework to two further transcripts. The results from this independent coding were compared, any discrepancies discussed and revisions were made to the framework. Rebecca Langford applied the revised framework to all subsequent transcripts, making revisions where necessary in discussion with Ruth Kipping as appropriate. Observation data were entered into a Microsoft Excel^® (Microsoft Corporation, Redmond, WA, USA) spreadsheet for analysis, with these data being compared and contrasted with interview data, where relevant, to support cross-checking and triangulation. Data from NAP SACC UK partner logs, Review and Reflect forms and goal-setting forms were entered into the REDCap data management system. Data from interviews, observations and document analysis were triangulated to identify confirmatory or contradictory data.

All analysis was conducted by Rebecca Langford, in discussion with other members of the research team. Rebecca Langford is an experienced qualitative researcher who started work on the study to conduct the process evaluation after the intervention had been delivered and the majority of follow-up data had been collected. Rebecca Langford had no other role in the design or management of the trial or in collecting (quantitative) outcome data. This distance from the wider NAP SACC study allowed her to take a critical and objective view of the intervention and research process.