NOURISH was an Australian RCT administered by Queensland University of Technology (QUT) in Brisbane, Queensland, and Flinders University in Adelaide, South Australia. The methods used have been detailed in the NOURISH protocol paper
[12]. A brief overview is given here, with a more detailed description of methods directly relevant to this paper provided below. First-time mothers were first approached after delivery in the major maternity hospitals in each city. Mothers were later recontacted and consenting mothers and their infants were randomly allocated to either the intervention or control group following baseline measurements (Time 1) when the infants were 2–7 months of age. Follow-up measurements were conducted at two points: Time 2 (infants aged ~14 months) and Time 3 (infants aged ~24 months). Follow-up was six months after completion of each of the two intervention modules. At all three time points data collected included maternal and child weight and height/length measured by study staff at study-specific measurement clinics or home visits, self-reported maternal and infant behaviours assessed by questionnaire and infant food intake assessed by a telephone 24-h recall and two 24-h food diaries.
The NOURISH intervention comprised a comprehensive skills-based program that used a cognitive behavioural approach and focused on the feeding and parenting practices that mediate children’s early feeding experiences. The intervention was delivered via two modules: modules 1 and 2 commenced when the children were approximately 4–6 and 13–16 months of age, respectively and were delivered over 12 weeks. Each module comprised six interactive group sessions (40 groups across both modules and sites) of 1–1.5 h duration. These sessions were co-led by a dietitian (n=13) and psychologist (n=13) at community child health clinics to reduce implementation costs, provide participants with convenient local access and engage the child health nurses. Facilitators received standardised training, used a comprehensive facilitator manual and standard presentation materials, and participated in fortnightly supervision teleconferences to promote intervention quality and integrity. The format was consistent with child health service delivery models at the time in Queensland and South Australia. The control group received self-directed access to usual community child health services that at mothers’ initiative potentially included child weighing, individual appointments with a child health nurse or access to information via a web site or a telephone help line.
Approval for NOURISH was obtained from 11 Human Research Ethics Committees (HRECs) that covered both universities and all the recruitment hospitals (QUT HREC 00171 Protocol 0700000752) and the trial was registered with Australian New Zealand Clinical Trials Registry (ACTRN12608000056392).
Recruitment
A two-stage recruitment strategy (referred to as Stage 1 and Stage 2) was used to access a consecutive sample of first-time mothers. Eligible mothers who had delivered a healthy term infant (>35 weeks, >2500g) were approached whilst still in hospital (Stage 1) and consent sought for later contact. In Australia 99% of births occur in a hospital or associated birthing centres
[13] and a comprehensive or universal home visit program was not available in either city. Mothers were primiparous, aged at least 18 years, with no documented history of domestic violence or intravenous drug use, or self reported eating or psychiatric disorders. Competent written and spoken English, and the ability and willingness to attend sessions at designated venues (in the event of being allocated to the intervention) were also eligibility requirements. The NOURISH intervention was conceptualised as a universal rather than targeted nutrition intervention and hence the selection criteria were designed to identify healthy mothers and babies for whom a community-based group intervention would be suitable.
Due to state-specific ethical and hospital requirements, mothers were approached by hospital midwives in Brisbane and research staff and students in Adelaide. The intention was to screen all women who delivered, including on weekends, and approach all mothers who met the selection criteria. The study was described to the mothers verbally and via a pamphlet. Mothers were invited to provide details for later contact regarding enrolment in the study. Those who agreed to later contact completed a consent form and completed a three page questionnaire; providing demographic data as well as basic information on lifestyle, health and tobacco and alcohol use during pregnancy, self reported pre-pregnancy weight status, intended feeding method and details of two alternative contacts. These mothers were provided with a NOURISH branded folder that contained the information pamphlet and a change of address card. Those who did not agree to be recontacted were asked to provide a sub-set of the same demographic data, as well as their self-reported pre-pregnancy weight status and intended feeding method.
Mothers who gave consent at Stage 1 were recontacted via mail when their infant was aged approximately 4 months (Stage 2). The mail-out was posted in a NOURISH-badged envelope and comprised a (i) personalised cover letter, (ii) four page participant information statement (extended version of the pamphlet provided in hospital), (iii) two page consent form, (iv) green coloured data collection form for those wishing to consent including a checklist of infant health problems and a mental health screener (the Kessler 10
[14]) to assess eligibility, (v) peach coloured data collection form for completion by non-consenters, asking their reasons for non-consent and five questions regarding current feeding, (vi) form for consenters to indicate preferences for the venue, day and time for baseline measurement, (vii) change of address card, and (viii) reply-paid envelope. The cover letter explained the colour coding of the forms, and advised mothers that they had equal chance of winning a $AU50 baby product voucher irrespective of whether they returned a completed green (consent) or peach (non-consent) form. At least three attempts were made by telephone, email, and text messaging to contact mothers who did not respond to the mail-out. Forms were re-sent to mothers who reported they had mislaid or not received the Stage 2 mail. Potential participant dyads were required to be still living locally (i.e. within travelling distance of group sessions), with no serious infant health problems, and a score on the Kessler 10 Psychological Distress Scale (K10;
[14]) under 30 (not indicative of high maternal psychological distress).
Rationale for recruiting two cohorts
Two waves of recruitment occurred, resulting in two participant cohorts. Recruitment of Cohort 1 took place at four Adelaide hospitals and three Brisbane hospitals between February and June 2008. The second wave of recruitment (Cohort 2) took place at two of the same hospitals in Adelaide and one additional hospital (September 2008 – March 2009) and at one of the same hospitals in Brisbane (January – April 2009). From a small pilot (N=105), we anticipated a 60% Stage 1 consent rate, with 70% of these converting at Stage 2 to full enrolment in the trial. For Cohort 1, we exceeded this at Stage 1 with 74% consenting for re-contact (Brisbane: 74%, Adelaide: 72%), but the Stage 2 conversion rate to enrolment (including non-consent as well as those who became ineligible and those unable to contact) was only 31% (Brisbane: 35%, Adelaide: 25%). As a result, Cohort 1 delivered 53% of our recruitment target (N=830), raising concerns about statistical power. Ethical approval was gained for a second round of recruitment (referred to as Cohort 2) at three hospitals in Adelaide (two of the same and one additional) and at one Brisbane hospital. The decision to use only a subset of original hospitals for Cohort 2 recruitment was based on response rates from Cohort 1 which were, at least in part, related to the size, level of engagement and research culture and infrastructure of the hospitals. The Stage 1 consent rate for Cohort 2 was 79% (Brisbane: 88%, Adelaide: 71%).
Other than use of a subset of the original hospitals as outlined above, recruitment procedures for Cohort 2 did not differ from Cohort 1. The exception was an additional mail-out to Cohort 2 mothers when infants were approximately six weeks old (compared with first re-contact at approximately 4 months for Cohort 1). It was hoped that this intermediate contact would improve Stage 2 conversion rates, however the Stage 2 conversion rate (including non-consent, ineligible and unable to contact) of 35% for Cohort 2 (Brisbane: 38%, Adelaide: 31%) was still well below expectations. The mail-out reminded mothers of the study aims and that they would be re-contacted in the next 4–6 weeks. Material mailed included the first NOURISH newsletter, a magnet with the study logo and contact details, and a Stage 2 non-consent form with reply paid envelope. A page of testimonials (de-identified but with permission) from mothers already in the study was included. The idea to include these testimonials came from feedback from Cohort 1 mothers who were participating in Module 1 in Adelaide. Cohort 2 received the same Stage 2 documents as Cohort 1. Again, at least three attempts were made to contact mothers who did not respond to the mail-out.
Retention strategies
A range of retention strategies were used, that can be categorised into three main themes: (i) participant convenience, (ii) keeping in contact and (iii) tangible incentives. The specific strategies used are detailed below:
iConvenience strategies were those intended to make participation as easy as possible. For instance, all mothers were able to select the health care clinic location, day and time most convenient for them to attend. Appointment confirmation, maps and information regarding parking, the questionnaire and, where necessary, a consent form were provided by mail. Mothers were sent text message reminders (via MessageNet;
http://www.messagenet.com.au) regarding measurement appointments. In addition, home visits to conduct anthropometric measurements were offered to participants with transport or time constraints, both after hours and on weekends. Home visits conducted were n=18 (Time 1), n=102 (Time 2) and n=147 (Time 3). Convenience strategies specific to mothers in the intervention group were also used. For instance, mothers in the intervention group were given a card containing details of all six information session dates and times for easy display on a fridge or pin board, and were sent a reminder text message before each session. For most intervention group mothers, the measurement and intervention sessions were at the same venue, but if not maps and parking information were again provided. Free child care was provided at some Module 1 sessions in Adelaide only (as it was freely available at the venue) and at all Module 2 sessions in both cities. Providing this service for Module 2 sessions was anticipated to improve attendance and reduce the distractions that would arise from having young children present.
iiKeeping in contact strategies were those intended to facilitate contact. A study-specific mobile telephone and email address were used and monitored several times each day to ensure timely responses to participants. Participants were given postcards and reply paid envelopes to encourage them to notify NOURISH of change of contact details. Mothers who moved away from either city were encouraged to remain in the study, by being sent instructions for measuring their child at home and reply paid envelopes. Alternate contacts were followed up if participant contact was lost, and letters were sent by Registered Mail as a final attempt to reach participants who were difficult to contact. Newsletters were sent quarterly to participants in both the intervention and control group, with content that was deliberately unrelated to nutrition, growth or weight status. Topics included returning to work after having children, introducing a second language, travelling with children, children’s book ideas, and tips for choosing a car seat.
iiiTangible incentives included: (i) bibs and magnets with the NOURISH logo, email contact address and telephone number; (ii) donated moisturiser samples for mother and baby, and (iii) plastic measuring spoons that were also intended to be used for completing the food diaries. Teabags were attached with the second questionnaire. Appreciation certificates with space for a child’s hand-print and magnetic photo frames bearing the NOURISH logo were sent to participants still active at Time 3.
Baseline assessment and allocation
Consenting mother-infant dyads had baseline (Time 1) weight and height/length measurements taken by trained research staff at government child health clinics, when the child was approximately four months old. As discussed above we provided as wide a range of clinic locations, days and times as was feasible within the constraints of staff availability and costs and access to clinic space. From a specified list, mothers chose the clinic, time and date that they wished to attend. A statistician external to the study was provided with a spreadsheet containing identification numbers of mothers who had been measured and the de-identified clinic at which they had been measured. The statistician stratified by the clinic attended, the rationale being that mothers would be most likely to attend the clinic closest to where they lived, and as such the clinic attended would serve as a proxy for socio-economic status. Thus, dyads were randomised in blocks of four within each stratum. Mothers were informed by mail to which group they had been allocated and those in the intervention group were provided with venue information and a schedule of the group sessions. In all communication with mothers the terms ‘information’ and ‘monitoring’ were used to describe the intervention and control groups respectively.
Data treatment and statistical analyses
Participants listed in Figure
1 as discontinued at Time 2 either actively withdrew prior to Time 2 or could not be contacted (or did not provide any data) at Time 2 or 3. Those listed as discontinued at Time 3 provided some data at Time 2 but either actively withdrew between Time 2 and 3 or could not be contacted and/or did not provide any data at Time 3.
To assess differences between (i) mothers who participated and mothers who did not; (ii) mothers allocated to control and mothers allocated to intervention group, and (iii) mothers who completed follow-up (Time 3) and mothers who did not, independent samples t statistics and likelihood ratio chi-square statistics were calculated for continuous and dichotomous data collected at Stage 1, respectively. All analyses were conducted using PASW/SPSS Version 18. Level of significance was specified as α=.05 (two-tailed).