This is a multi-centre, open, parallel group, randomised (1:1) controlled superiority trial of full milk feeds versus gradual milk feeds. Mothers and infants will be recruited from around 40 neonatal units (predominantly local neonatal units (LNUs) and Neonatal Intensive Care Units (NICUs) in the UK. A list of participating sites can be found at www.feed1.ac.uk. The full protocol and other trial documentation, including participant information sheets and consent forms, can also be found on the trial website. An embedded Study Within a Trial (SWAT) is also included which is investigating methods of training sites. Details of the SWAT can be found on the SWAT repository (17) (18) and will be reported separately.
Eligibility
Inclusion criteria include i) infants born at 30+0 to 32+6 weeks gestation (inclusive) and ii) infants <3 hours (180 minutes) old (since recorded time of birth). Infants requiring respiratory support (such as via continuous positive airway pressure) or other supportive treatments will be included in the study if the clinician is in equipoise about the infant being randomised to either the “full milk” or the “gradual milk” arm. Similarly, well infants should only be included if the attending clinician is in equipoise about the best feeding regime and the infant being randomised to either “full milk” or “gradual milk” groups. Exclusion criteria include i) infants with known congenital abnormalities of the gastrointestinal tract or other congenital conditions that make enteral feeding unsafe, ii) infants who are small for gestational age (SGA) (birthweight <10th centile) and have evidence of reversed end-diastolic flow on antenatal umbilical artery doppler ultrasound, iii) mothers who have participated in the trial during a previous pregnancy.
Interventions
For infants in the full milk group (intervention), fluids will be started as milk at 60 ml/kg/day, increased as per their individual requirements and in line with standard neonatal practice. The choice of feeding intervals will be determined by local policy and clinician’s preference. Wherever possible, mother’s expressed breast milk will always be the first preference for infant milk feeds. It is likely that mother’s breast milk will need to be supplemented with additional milk, i.e. either infant formula milk or donor breast milk in the first few days. The decision as to the type of milk used will be made by the mother and the site, in line with the site’s local policy. For infants in the gradual milk group (control), fluids will be given in accordance with standard practice at the site. This may include milk feeds, starting at a maximum of 30 ml/kg/day on day 1 with a minimum of 30 ml/kg/day of supplementary IV fluids or parenteral nutrition. Adherence to the randomised allocation will be monitored on a regular basis by the Trial Management Group.
Outcomes
The primary outcome is length of infant hospital stay. Since hospitals could apply different discharge criteria, a secondary outcome of time until objective discharge criteria are met has also been included. The statistician, blinded to treatment allocation, will use the date at which each infant first met all three of the following criteria: i) current weight >1700g, ii) infant is able to take at least one full suck feed assessed as adequate, iii) infant has been off additional temperature support for >24 hours. Daily data will be collected to determine the day on which the infant achieved all the three features to determine the time until objective discharge criteria are met. Secondary outcomes are listed in Table 1.
Table 1 – secondary outcome measures
Outcomes assessed whilst in hospital
|
• Time to regain birth weight
• Time taken to maintain full milk feeding (>140 ml/kg/day for 3 consecutive days)
• Number of days of peripheral cannula and IV cannulae inserted until full milk feeding
|
Outcomes assessed at hospital discharge
|
• Survival
• Incidence of microbiologically confirmed late-onset sepsis
• Necrotising enterocolitis (Bell’s stage 2 or 3) (19)
• Growth (change in Z scores from birth for length, weight and head circumference)
• Breast feeding
• Breast milk feeds
• Number of days of parenteral nutrition
• Number of central venous lines inserted and central line days
|
Outcomes assessed at 6 weeks corrected gestational age
|
• Survival
• Breast feeding
• Breast milk feeds
• Parental satisfaction and wellbeing, assessed using Preterm Birth Experience and Satisfaction Scale (p-BESS) questionnaire (4)
|
Outcome measures included have been guided by the Core Outcomes in Neonatology (COIN) core outcome set (20). Discussions are ongoing to obtain separate funding in order to collect longer-term follow-up data at two years corrected gestational age. In addition, the research team are in discussions with researchers based in other countries, such as Canada and Australia, to set-up parallel studies. This will be to provide sufficient power to assess outcomes of NEC and sepsis; these outcomes occurred in 1% and 12% of infants retrospectively in the SIFT Trial among those who would have been eligible for FEED1.
Sample size and recruitment
Our sample size calculation has been based upon detecting a between group difference in means of length of stay of two days. Input from our parent representatives suggests that from a family perspective even a short reduction in length of hospital stay could make a huge difference to parents. In addition, from a cost-saving perspective, a reduction in length of stay of 2 days for this population of infants could result in £5.6m annual savings for the NHS in England and Wales and >12,000 days of increased neonatal cot capacity.
Data from audits and previous studies suggest that the distribution of length of hospital stay in this population is approximately normal. To detect a difference in means of two days between the two groups with 90% power, 1:1 allocation, and 5% two-sided significance requires 1778 infants, assuming a standard deviation of 13. Allowing for 2% non-collection of the primary outcome data due to death, no consent for data collection after oral assent and infants remaining in hospital at the end of data collection and accounting for clustering will require 2088 infants and recruitment of approximately 1770 women. The inflation to account for clustering assumes that 15% and 1.4% of pregnancies will be twin and triplets respectively, and that the intracluster correlation coefficient for length of hospital stay for infants from the same pregnancy is 0.82.
Participant enrolment and consent
The flow of women and infants throughout the trial is shown in Figure 1. As the infant must be randomised within 3 hours of birth, a time during which the woman is recovering from giving birth and that is emotionally fraught and potentially difficult for families, a two-stage consent pathway will be used. Wherever possible, women will be approached antenatally and asked to consider participation in the trial. At around the time of antenatal counselling consultation, women will be given study information, have the opportunity to discuss the study and ask questions and give full written informed consent if they are willing to do so. For these women, once they have given birth, infant eligibility will be checked and the infant(s) will be entered into the trial, unless the mother expresses she has changed her mind. For some women, receiving information and providing consent antenatally may not be possible due to the rapid and unexpected nature of preterm birth. During labour and postnatally, these women can be approached via an oral assent pathway. This involves a member of the neonatal team inviting the woman to participate and giving the minimal information required to make a decision. A shorter, simplified participant information sheet and/or a short animation film may be used to support this discussion. The decision to participate will be documented in the medical notes and full written informed consent will be obtained at a later date (ideally <72 hours). This two-stage consent pathway has been used successfully in a previous neonatal trial (21) and features as part of guidance from the Royal College of Obstetricians and Gynaecologists (RCOG) (22). Sites will be fully trained in the two-stage consent pathway and will be provided with a range of supplementary materials to support oral assent conversations which include a recorded webinar, video showing examples of discussions and additional documentation. Women are asked to give their optional consent to be contacted for longer-term follow-up of their infant(s) and for later educational outcomes.
Randomisation and blinding
The unit of randomisation is the mother to ensure that a mother-infant(s) dyad is treated as a unit. This will also enable antenatal consent and facilitate early commencement of the intervention. In addition it will ensure that siblings from multiple pregnancies are assigned to the same group. Parents have told us this is important to them as they would not like to feed their infants differently unless there was a medical reason to do so. Randomisation will be performed on a 1:1 ratio, using a secure web-based system, developed and maintained by the Nottingham Clinical Trials Unit (NCTU), which will conceal allocation sequence. Randomisation will use a minimisation algorithm, with a random element, to ensure balance on important prognostic factors: neonatal unit, single or multiple birth, gestational age at birth, birthweight centile and whether IV fluids were started prior to randomisation. Randomisation will be undertaken by the Principal Investigator, clinician or other study team member within 3 hours from the recorded time of birth. This is to ensure infants randomised to full milk feeds can receive the intervention with minimal risk of receiving IV fluids and should help prevent contamination between groups.
It is not possible to blind investigators and families due to the nature of the intervention. However, the trial statistician will remain blinded throughout the trial. To objectively assess the primary outcome, a secondary outcome measure to objectively assess discharge criteria is included and a blinded endpoint review committee (BERC) will also be established to examine cases of late-onset sepsis and necrotising enterocolitis where the diagnosis is unclear.
Trial assessments and procedures
All trial assessments and procedures are outlined in Table 2. Most outcome data will be collected during the infant’s hospital admission. A daily feeding log will be completed until the infant receives 140 ml/kg/day of feeds, sustained for 3 consecutive days. For infants who are transferred to another hospital, i.e. a continuing care site, a paper transfer pack will accompany the infant with information on trial participation and data collection will continue. In order to collect data at 6 weeks corrected gestational age, an online questionnaire (or paper if preferred) will be sent to women; reminders will be sent to increase response-rates.
Table 2 – Schedule of assessments
|
TRIAL PERIOD
|
|
Antenatal pathway
|
Postnatal pathway
|
Treatment period
|
Follow-up
|
TIMEPOINT
|
Before birth
|
After birth
|
Day1
|
Day2
|
Day3
|
Etc.
|
Discharge
|
6 weeks corrected age
|
ENROLMENT:
|
|
Eligibility screen
|
X
|
X
|
|
|
|
|
|
|
Informed consent
|
X
|
|
|
|
|
|
|
|
Oral assent1
|
X1
|
X1
|
|
|
|
|
|
|
Randomisation
|
|
|
X
|
|
|
|
|
|
Informed consent2
|
X2
|
X2
|
|
|
|
|
|
|
Baseline data
|
|
|
X
|
|
|
|
|
|
INTERVENTIONS:
|
|
Full enteral feeds or Gradual feeds
|
|
|
X
|
X
|
X
|
X
|
|
|
ASSESSMENTS:
|
|
Daily feeding log
|
|
|
X
|
X
|
X
|
X
|
|
|
Number of painful procedures
|
|
|
X
|
X
|
X
|
X
|
|
|
Discharge criteria
|
|
|
X
|
X
|
X
|
X
|
X
|
|
Late-onset Invasive Infection3
|
|
|
X3
|
X3
|
X3
|
X3
|
|
|
Gut Signs4
|
|
|
|
|
|
X4
|
|
|
DISCHARGE DATA:
|
|
Growth (z scores - weight, length, and head circumference)
|
|
|
|
|
|
|
X
|
|
FOLLOW-UP:
|
|
Healthcare visits
|
|
|
|
|
|
|
|
X
|
Types and modes of feeding
|
|
|
|
|
|
|
|
X
|
Parental satisfaction
(p-BESS)
|
|
|
|
|
|
|
|
X
|
1 Women who present in the late stages of labour or who have already given birth should be consented first via oral assent followed by later written informed consent
2 Women who given oral assent trial should provide written consent within 72 hours of providing oral assent, where possible
3 Each episode of microbiologically-confirmed or clinically-suspected late-onset invasive infection should be reported throughout the treatment period until hospital discharge
4 To be reported if this infant has received at least 5 days of treatment for gut signs, if they are transferred with gut signs, or if they have died from gut signs
Adverse events
Serious adverse events, including death, will be reported as such by the Principal Investigator, or delegate, reporting within 24 hours of being made aware of the event. Adverse reactions collected as outcomes will not need to be reported separately. Late-onset sepsis (micobiologically confirmed or clinically suspected), necrotising enterocolitis (Bell’s stage 2 or 3) or other known complications of prematurity will not be reported as serious adverse events.
Data management
All trial data will be collected by site staff delegated to do so and entered onto a trial specific database with infants only identified by their unique trial number and initials. Using MACRO (Elsevier), the database will be developed and maintained by NCTU staff. Access to the database will be restricted and secure. Sites will be provided with paper workbooks in order to assist them with data collection. Missing or spurious data will be queried in a timely manner throughout the trial. In order to facilitate contact with families at 6 weeks corrected gestational age, contact details will be collected and entered into an online secure system, developed and maintained by NCTU staff. This data is held separately to the deidentified trial data collected. Access to contact details is restricted to those involved in the follow-up phase, as authorised by the Chief Investigator.
Statistical analysis
Analysis and reporting of the trial will be in accordance with the Consolidated Standards of Reporting Trials (CONSORT) (23) guidelines. All analyses will be outlined in a detailed Statistical Analysis Plan agreed prior to database lock. The primary comparative analyses will be conducted according to randomised allocation with due emphasis on confidence intervals for between-group comparisons.
The primary outcome will be analysed using linear mixed models to compare the mean length of hospital stay between groups, adjusting for minimisation variables and accounting for the correlation between outcomes for infants born from a multiple pregnancy. The estimated between group effect will be presented using the difference in means, with a 95% confidence interval. Secondary outcomes will be analysed similarly using appropriate multilevel regression models, dependent on the type of outcome variable. The between group effect will be reported using an appropriate effect estimate along with a corresponding 95% confidence interval.
The primary approach to between-group comparative analyses will be by modified intention-to-treat (i.e. including all participants who have been randomised and without imputation of missing outcome data). In particular, the primary analysis will exclude the small number of deaths that might occur before discharge, but sensitivity analysis will be performed by imputing the primary outcome with the worst observed length of stay for infants who died prior to discharge to check that this does not influence the findings. A further sensitivity analysis will assess the effect of compliance with the allocated feeding strategy through complier average causal effect (CACE) analysis.
Appropriate interaction terms will be included in the primary regression analyses in order to conduct subgroup analyses according to gestation at birth and birth weight centile, but this analysis will be regarded as exploratory as the study is not powered to detect interactions. Interpretation of any subgroup effects will be based on the treatment-subgroup interaction and 95% confidence interval.
A within-trial economic analysis will be conducted from an NHS and personal social services perspective. Resource data will be collected prospectively and unit costs will be obtained from routine sources. The main cost-effectiveness analysis will be based on the cost per reduction in days in care. A longer-term projection of costs and benefits will be estimated through decision analytical modelling. Appropriate sensitivity analyses will be undertaken to account for any uncertainty.
Monitoring and governance
An independent data monitoring committee (DMC) will review unblinded trial data, including safety data, on an intermittent basis. The role of the DMC is outlined in a charter which outlines their terms of reference. Overall independent oversight will be provided by a Trial Steering Committee (TSC).
On-site monitoring will not be conducted routinely throughout the trial. Instead, regular central monitoring of trial data will be undertaken and this data will be used to assess if sites have met any triggers to activate a triggered monitoring visit. The Trial Management Group (TMG) are responsible for reviewing central monitoring reports and agreeing if triggered on-site monitoring visits are required. Any trial conduct audits will be carried out by the sponsor as per their local auditing plans.
Patient and public involvement (PPI)
Our research team includes a parent who has experience of having a preterm infant, born at 31 weeks gestation, and the research manager for Bliss, the UK’s largest premature and sick baby charity, who represent parents with experience of preterm birth. In addition to our immediate research team PPI members, we have a group of three parents who provide additional parent input, and a PPI member on the independent trial steering committee. We involved parents throughout the design stage of this trial and continue to do so whilst it is being conducted.