This chapter describes the methods used for the feasibility study of the intervention described in Chapter 7. The primary aim of the study was to determine the feasibility and acceptability of progressing to a full-scale comparative evaluation of the Cubs intervention. The objectives of the feasibility study were:
to appraise the willingness of parents and staff to implement and sustain the intervention
to assess the associated costs of implementing cue-based feeding in NNUs
to determine the feasibility and acceptability of conducting a future RCT, including views on important outcomes
to scope existing data-recording systems and potential short- and long-term outcome measures.
Study design and theoretical framework
The mixed-methods feasibility study comprised quantitative and qualitative approaches to meet the above objectives. The study involved implementation of the intervention in three NNUs. The study was designed as a single-arm feasibility study; no comparison group was included. Feasibility outcomes of acceptability, adoption, appropriateness and fidelity were based on the model by Proctor et al.83 We judged feasibility as whether the intervention was practicable to implement based on the views of the participants, as well as the expert opinions of the research team and Stakeholder Advisory Group. We used the Proctor et al.83 definition of acceptability (i.e. that the intervention was ‘agreeable, palatable or satisfactory’).
Setting
The intervention was implemented in three NNUs, two in England and one in Scotland. The sites were selected to provide different sizes and levels of unit, and diversity of populations. For pragmatic reasons, units where members of the research team had established links were selected. Although all three units have full accreditation for the UNICEF UK BFI neonatal standards,68 at the time of selecting the units only site 2 had achieved this status. shows details of the three units.
Characteristics of intervention settings
Implementation
Implementation of the intervention commenced with a site visit by the research team members, who provided an initial 2-hour training session for the research nurse and the key cascaders on the intervention and the study procedures. Each unit was provided with paper and electronic copies of the intervention manual and resources, a pack for the parents of each infant containing the intervention information and resources, and a set of posters to be displayed around the unit. Links to the Cubs film and narrated PowerPoint were also included. It was planned that the intervention would commence once training had been cascaded to most staff. Commencement of the study was delayed for several reasons, including gaining approval from the study sponsor (University of Dundee) taking longer than anticipated and a delay in being able to circulate the film (one of the infants featured in the film died and it was felt that permission should be sought from the parents to retain that section of the film). In one unit, a planned move to temporary accommodation was delayed meaning that instead of taking place before the study started, it took place just as the study was starting. The principal investigator (PI) asked to delay the start of the study until the move was completed and the staff and parents were accustomed to their new environment. Recruitment took place over 7 months (August 2019 to February 2020). All follow-up was stopped in early March 2020, as NHS research was stopped because of the COVID-19 pandemic.
Quantitative methods
Quantitative methods were used to assess recruitment and retention, and clinical outcome data on infants included in the study.
Sample and eligibility criteria
The target population was preterm infants (< 37 weeks’ gestation), including infants of multiple births, who were at least partially enterally fed, in NNUs or transitional care settings. The sample size for the feasibility study of 20 infants from each unit was based on the expert opinion of the study team and discussion with the PIs at each site. This required the recruitment of three or four infants each month, at each site, over the 6-month planned recruitment period, which the PIs assessed as achievable. This would provide enough data on the outcomes, including estimation of recruitment rate, rates of completion of the intervention and follow-up, and clinical outcomes for the infants.
The eligibility criteria were developmentally normal preterm infants born before 37 weeks’ gestation, who were clinically stable and at least partially enterally fed, had an intragastric tube in place at the start of the study, and whose parent(s) consented to their inclusion in the study. Singleton and multiple births were eligible for inclusion and any planned feeding type or method [i.e. at the breast, expressed breast milk (EBM), formula or a combination]. The exclusion criteria were infants born after 37 completed weeks of gestation, infants who were not at least partially enterally fed, preterm infants who had transitioned to full oral feeding, and infants with major congenital anomalies, gastrointestinal disorders (e.g. necrotising enterocolitis), congenital infections and major neurological conditions (e.g. cerebral palsy, seizures, grade III–IV intracranial haemorrhage, periventricular leukomalacia). Infants on high-flow oxygen were also excluded from the study.
Recruitment procedures
Infants eligible to be included in the study were identified by the care team, usually a neonatal nurse, who introduced the parents to the study and provided them with the participant information sheet. Parents interested in participating in the study completed a reply slip, which was passed to the site research nurse. The research nurse discussed with the parent the information in the participant information sheet, responded to any questions and gained written consent from parents for their infant to be included in the study. Consent for the qualitative aspects of the study was taken at the same time.
Data collection and outcome measures
For recruitment and retention assessment, the site research nurses kept a log of the total number of infants screened, number eligible, number recruited and reasons for non-recruitment. Baseline and follow-up data were collected by research nurses from the infant’s care records, as shown in .
Data on feeding outcomes were taken from the ‘Our Feeding Journey’ record designed specifically for the study. Outcomes included cues observed, interval between feeds, type of feed (breast milk, formula, combination) and method of feeding (at breast, bottle, tube, other). The research nurses scanned the documents for transfer to the research team and parents could keep the document if they wished. The 2-week follow-up data were collected via a telephone call or e-mail by the site research nurse.
Data analysis
Analysis took place after all data were entered in the database and the database had been locked. The number of missing data were examined to inform decisions about outcome measure selection for a definitive trial. Outcomes were summarised as means, medians and standard deviations for quantitative variables, and percentages and denominator for categorical variables. The distributions of outcomes were explored and transformations used where appropriate. Baseline characteristics for infants were tabulated in total and by site. The primary analysis consisted of descriptive measures at baseline, during intervention, discharge and at the 2-week follow-up post discharge. The changes from the start of the intervention to discharge included age at discharge, days from intervention to full oral, days from intervention to discharge, weight at transition to full oral feeding (g), weight at discharge (g) and weight at the 2-week follow-up (g). The mean weight per day of the intervention was also calculated. All analyses were performed using IBM SPSS Statistics version 26 (IBM Corporation, Armonk, NY, USA). The results of the quantitative data are presented in Chapter 9.
Feeding data from the ‘Our Feeding Journey’ documents were entered into a database. Owing to the large number of data collected, we entered data from a 24-hour period each week, starting on day 2 of the intervention (to account for infants starting the intervention at different times on the first day) and then each following seventh day until the infant was discharged. In addition, to capture the last full 24 hours on the intervention, data from the day before discharge were also included in the analysis. Missing data (i.e. where not all feeds were documented every day) made it difficult to achieve an accurate analysis of cue-based compared with scheduled feeding based on times of feeds. Therefore, on the advice of the Stakeholder Advisory Group, we selected three infants (one from each site) with the most comprehensive data to provide more detailed case studies. Each infant’s transition to oral feeding journey is unique; therefore, the case study approach was used to provide insight to the experience of transitioning to oral feeding. The case studies were not intended to be representative of the full range of participant experience. For each individual case study, we summarised data collected at baseline, during the intervention and at follow-up. Start and stop cues were analysed using percentages to identify the most common cues recognised by staff and parents.
Qualitative methods
The quantitative data were supplemented with qualitative data as part of a process evaluation to provide a more in-depth understanding of implementation outcomes of acceptability, feasibility, adoption and fidelity.83 The study also included qualitative data from interviews with parents of infants receiving the intervention and staff involved in its implementation. Further data on fidelity were collected through observations.
Sample and recruitment
We aimed to conduct semistructured interviews with a subset of a total of 30 parents from the 60 infants included in the study, comprising 10 from each site. For the observation we did not define a sample size but aimed to conduct 18 hours of observation (six 3-hour observations) in each site. The consent for the semistructured interviews and observations was taken at the same time as the consent for the infants to be included in the study.
We aimed also to include a range of health-care practitioners involved in caring for infants as they transitioned from tube to oral feeding and supporting parents. Our target sample was 10–15 staff members from each unit, including neonatal nurses, nursery nurses, infant feeding co-ordinators, doctors, and speech and language therapists. Members of staff were recruited through an e-mail invitation to all staff members, which included the participant information sheet as an attachment. Staff members who were willing to be interviewed were requested to contact the site research nurse to ask any questions and arrange a convenient time and venue for the interview. Written consent was taken at the outset of the interviews. For the non-participant observation of practice, the site research nurse identified a day and time when there was at least one infant enrolled in the study on the unit. At least 24 hours prior to a planned, non-participant observation session, the research nurse contacted (either face to face or by e-mail) all staff who were rostered to be on duty during the observation to inform them that it was planned and to provide the participant information sheet. Written informed consent was taken immediately prior to the planned observation.
Data collection
The purpose of the interviews with parents was to explore in depth their experiences and views of cue-based feeding, the acceptability of the intervention, whether or not it was implemented as intended, parental satisfaction with care and support for infant feeding, how parents would feel about a future randomised trial and their views on important outcomes for such a trial. A semistructured topic guide was developed based on the above aims, the objectives of the study and a framework of implementation outcomes.83 Interviews were conducted face to face by the research nurses on the NNU. As the study progressed and it became clear that arranging the interviews with parents while they were on the NNU was challenging, an amendment to the protocol was made to offer a telephone interview following discharge. However, no parents took up this option. The interviews lasted 30–60 minutes and were audio-recorded with the consent of the participants.
The purpose of the interviews with staff was to explore their views and experiences of the training provided, of implementing cue-based feeding in the context of their unit (including tailoring or modification of the intervention), and to assess acceptability, adoption, appropriateness and fidelity. A further purpose of the interviews was to assess willingness to support a future randomised trial and to identify important outcomes for such a trial. An interview topic guide was developed based on the above aims, the study objectives and implementation outcomes.83 Interviews were conducted at a time and place that was convenient for the interviewee. Interviews lasted 40–60 minutes and were audio-recorded with the consent of the participants.
The purpose of the observations was to assess fidelity to the intervention. The observations were guided by a checklist and field notes with a focus on fidelity, and interactions between staff and parents concerning feeding. Observations were conducted by the site research nurses.
In terms of gathering data on costs, on the advice of the health economist member of the Stakeholder Advisory Group, we included questions in the interview topic guides to assess any additional time or resource needed by staff or parents, and any out-of-pocket expenses incurred by parents and families.
Data analysis
The audio-recordings of interviews were transferred securely from the sites to the University of Dundee and were transcribed verbatim. The transcripts were checked and anonymised by a member of the research team. We used a framework approach to analyse the interview and observational data.84 This comprised a seven-stage approach: transcription, familiarisation, coding, developing a working analytical framework, applying the analytical framework, charting data into the framework matrix, and interpreting the data. The analysis was supported by qualitative data analysis software (NVivo; QSR International, Warrington, UK). The analytical framework was derived deductively using the implementation outcomes described by Proctor et al.83 and inductively (i.e. incorporating new themes in the data through open coding). This approach to analysis enabled comparison by themes across multiple accounts as well as retaining the context of individual experience. To enhance reliability of the coding, two researchers independently coded the first few transcripts before agreeing a set of codes to apply to all transcripts. The research team met regularly during the analysis to discuss interpretation of research material. As well as using the Proctor et al.83 implementation outcomes across staff and parent interviews, the staff transcripts were analysed using a normalisation process theory85 framework to assess context, attitudes, organisational support and barriers to the intervention.
Ethics approvals
Approval for the feasibility study was given by the North of Scotland NHS Ethics Committee (19/NS/0055), the Health Research Authority and NHS Research Scotland. Research and development approval was given by the relevant NHS trusts in England and health board in Scotland.
Summary
In this chapter we have described the methods used for the feasibility study. This includes the study objectives, design and theoretical framework, the implementation process, the research settings and the methods for the quantitative and qualitative components of the study. The following two chapters present the findings of the quantitative analysis (Chapter 9) and the qualitative analysis (Chapter 10). These findings were then taken forward to the final stage of the research (WP 4), as described in Chapter 11.
