First we describe the Families for Health trial then focus on the methods we used to recruit participants and evaluate the different recruitment methods.
The families for health study design
Families for Health is a family-based group intervention for the treatment of children aged 6–11 who are overweight or obese. The intervention is 10 weeks long, 2½ hours per week, run in a community venue with parents/carers and children attending parallel groups. Four trained facilitators, two for the children’s group and two for the parents/carers group, run each programme. In contrast to other similar interventions, the Families for Health programme emphasises parenting skills, relationship skills and emotional and social development, combined with information about lifestyle. The Families for Health trial evaluated the effectiveness and cost-effectiveness at 12 months of the Families for Health programme delivered in the NHS using a randomised controlled trial. Participants across three former West Midlands NHS Primary Care Trusts (Sites A, B and C) were randomly allocated to receive the Families for Health programme or usual care. The aim was to run six Families for Health programmes (two in each site). Usual care could vary across the three sites, but consisted of either a community-based group intervention or a one-to-one intervention. Ethical approval for the study was obtained from the National Research Ethics Services (NRES) Committee West Midlands – Coventry & Warwickshire Research Ethics Committee (REC), (reference 11/WM/0290). Further study and intervention details are published elsewhere [12].
Inclusion criteria
The trial aimed to recruit 120 families (40 from each site) over 12 months (between March 2012 to February 2013) [12]. The initial inclusion criteria stipulated that families must have at least one overweight (≥91st centile for BMI) or obese (≥98th centile for BMI) child aged 7–11 years, based on the UK 1990 BMI charts [13]. Families were excluded where the parent/carer or child had insufficient command of English to be able to participate in the group, if the child had a metabolic or other recognised medical cause of obesity, or if the child had severe learning difficulties and/or behavioural problems that would make participation difficult.
Researchers used a three-step procedure to obtain informed consent, giving parents/carers and children time to consider whether they wished to participate [12]. First, after each enquiry, potential participants were given or sent by post information sheets about the trial (child and parent versions). Second, after a minimum of 3 days, parents/carers were contacted by telephone to ask whether they were interested in taking part in the trial and to answer any questions. Third, a researcher visited the parent/carer(s) and child(ren) at their home and obtained the parents/carers’ written consent and the child’s written assent. Researchers were trained in informed consent. Informed consent was obtained from all randomised participants.
Planned recruitment strategies
Recruitment of families started in mid-March 2012 across Sites A and B, while recruitment in Site C was delayed until mid-April 2012 due to delays in setting up the intervention. We used passive recruitment using the media (advertisements in local newspapers and on local radio) and active targeted recruitment both by sending letters to families with an overweight or very overweight child recently measured in the National Child Measurement Programme (NCMP) and by referrals of relevant families from health-care professionals.
Changes to eligibility criteria
In the first month of recruitment, some parents of 6-year-old children were keen to take part, and were disappointed when told they were not eligible. The programme suits younger children, with its focus on active games, the activities around healthy eating and circle time. We decided to change the inclusion criteria to children aged 6–11 years, to allow 6-year-olds to take part. This change was approved by the funders, National Institute for Health Research (NIHR) Health Technology Assessment (HTA), and received ethical approval.
Changes to projected recruitment rate
Recruitment was intended to be completed in 12 months, but instead took 24 months. In April 2013, because of slow recruitment, the target recruitment rate was changed from ten to six families per month. Recruitment in Site A was extended to February 2014, running an additional seventh Families for Health programme to increase the number of study participants. We finally recruited 115 families, five short of the original target. Figure 1 shows the projected graph of planned recruitment at the start of the study with a target of recruiting ten families per month, the revised plan for recruitment of six families per month, and the actual cumulative total.
Changes to recruitment methods
Because of the slow recruitment rate, we adopted additional methods of recruitment. These included, as active methods, targeted letters from GPs to families identified in their records as having an overweight or obese child in Site C and similar letters from local Change4Life advisors in Site B. (Change4Life is a campaign in England and Wales with information aimed at families to encourage healthy eating and physical activity: www.nhs.uk/Change4Life/). Additional passive methods included placing flyers and posters at schools and in school newsletters, in the community and at GP surgeries and the research team attending public events to raise awareness of the trial. As recruitment continued, we found that word of mouth became a useful recruitment method. Additional file 1 gives examples of the recruitment methods.
Design of recruitment study
Data collection
At the first enquiry from a potential participating family, the parent/carer was asked where they heard about the study. For health professional referrals, the source of the referral was noted. Once a family was recruited, we collected quantitative and qualitative data at baseline, and the 3-month and 12-month follow-ups. At the baseline visits, participants completed a recruitment questionnaire, which included the question, ‘Where did you hear about the research study?’ (GP, school nurse, health visitor, paediatrician, school, media, flyer/leaflet, friends/family, letter from NCMP, other). During analysis, in cases where there was a discrepancy between the questionnaire response and the response given when they first enquired about the study, the response at their first enquiry was used. Where parents/carers cited more than one source of information about the study, we recorded the method that was the final prompt for them to enquire about the study with the research team.
One-to-one semi-structured interviews were undertaken with parents/carers during a control family’s 3-month follow-up visit or an intervention family’s interview after accessing a Families for Health programme (at 3 months from baseline, or later if there was a delay in accessing the intervention). Consecutive sampling included representation of all Families for Health groups aiming for diversity of age, ethnicity and gender of the children, family size and whether they completed the intervention or not. The mean duration of parent/carer interviews was 20 minutes (range 4 to 62 minutes). Interviews were digitally recorded, subject to the permission of each participant, and transcribed verbatim.
Data analysis
Transcripts were checked for accuracy against the initial recording. All transcripts were anonymised, and data were analysed using NVivo 10. Coding of data was thematic, based on the interview schedules with the addition of emergent themes [14]. We developed a coding framework during meetings of the research team, ensuring that important topics in the data were captured. The initial codes were then developed into themes and reviewed again in comparison with the coded extracts [15]. At least 10 % of the coded data was cross-checked by another member of the research team, to reduce researcher bias. For this paper, we report those data coded under ‘Why joined the study/intervention’, which included anything to do with how the family had heard about the study and what motivated them to take part.
Family attendance for the Families for Health intervention was recorded each week by the facilitators. At follow-up visits, participants were asked what intervention they attended and how many sessions. For the purposes of this paper, the attendance records completed by facilitators are used for Families for Health participants, and self-reported attendance for families in the usual care arm.
Statistical analysis involved comparing the baseline characteristics of those recruited by active and passive methods. A chi-squared test was used to compare categorical variable baseline characteristics, unless 20 % of the expected frequencies of categories was less than 6, in which case Fisher’s exact test was used. Differences in continuous variable baseline characteristics between actively and passively recruited participants were compared by means of independent sample T-tests.