1\. Introduction and research questions
Although systematic reviews suggest the potential effectiveness of in-service learning (i.e. volunteering) and youth development as approaches to preventing teenage pregnancy, no intervention involving toddler-mentoring has been previously evaluated. Teens and Toddlers (T\&T) is an intervention that enables young people at risk of teenage pregnancy to spend time mentoring a toddler as well as participating in counselling and classroom work focused on youth development. This trial aims to answer the following research questions:
1. What is the effect of the T\&T intervention on participants' use of contraception at last sex, sex without contraception over the previous 3 months, expectation of teenage parenthood and youth development (primary outcomes) one year after intervention delivery?
2. What is the effect of the T\&T intervention on participants' experience of teenage pregnancy, use of condoms at last sex, sex without condoms over the previous 3 months, emotional self-reflection and emotional vocabulary , dislike of school and knowledge of contraception (secondary outcomes) one year after intervention delivery?
2\. Intervention, setting and participants
The T\&T intervention is to be delivered in 22 sites in the south-east, north-east and north-west of England. Each site comprised a nursery and secondary school. Participants will be recruited in two cohorts (September 2009 and February 2010). Participants and parents will be asked to give informed consent and participating young women will be surveyed at baseline, and then randomised to intervention or control group (see below).
3\. Sample size
Our power calculation focused on our primary outcome, no contraception use at last sex in the previous 6 months, which we anticipate being of 25% prevalence among our comparison group at follow-up 2, informed by the previous Ripple trial. Assuming this prevalence, in order to detect a halving of risk (in the range of effects reported by for example the Safer Choices and CAS-Carrera studies) with 80% power and 95% precision would require a sample of 180 per arm.
In addition, a further 120 girls are to be recruited at baseline to act as potential reserves should girls allocated to receive the T\&T intervention drop out of the intervention in its first 8 weeks. Where such drop-out did occur, the girl dropping out of the intervention and her matched pair in the comparison group would nonetheless remain in the trial analytic sample in order to minimize selection bias. A reserve pair from the same school, one randomized to replace the drop-out on the intervention and one allocated to comparison group, would augment the drop-outs in the analytic sample. Any girls who drop out of T\&T after 8 weeks will not be replaced on the intervention and will also remain in the trial analytic sample.
The sample will be recruited through the participating schools. Each school will recruit young women in groups (one or two groups per year-cohort depending on the size of the school and the level of involvement). The average size of each group will be 12 girls of whom 6 will be randomised to T\&T and 6 to the control group (with an additional 4 girls recruited as reserves for every group, half of whom would if necessary be randomised to T\&T and half randomised to the control group).
4\. Randomisation
Within each school, girls will complete baseline assessment and be put into matched pairs based on age and sexual experience. Then they will be given a random number using a computer-based random number generator. Within each pair, the girl with the lowest number would be selected for the Teens and Toddlers programme and the girl with the highest number would be in the comparison group. The reserve pairs would be selected in a similar way, each pair would be given a random number. The two pairs with the lowest random numbers would become the reserve pairs, and the six pairs with the highest random numbers would form the main pairs for the evaluation. This randomisation process will be done using anonymised data remotely at the Offices of the National Centre for Social Research. Notification of girls' allocation will then be sent to the schools so that teachers can inform girls. Neither girls, teachers nor intervention providers can be blinded to allocation. Research data collection and analysis staff will be blinded to allocation.
5\. Outcome measures
We have pre-hypothesised primary and secondary outcomes in order to avoid bias through 'data-dredging' for significant results. We have tried to balance having sufficient breadth to reflect the potentially broad benefits of participation in the T\&T programme, with the need to reduce the possibility of false-positive results arising from multiple tests of statistical significance. Full questions are available on request. All measures will be piloted within an independent sample of young people who will be then interviewed regarding their understanding of the meaning of the questions and the meaning of their responses. These outcomes will be measured at follow-up 2 (72 weeks) via pen and paper self-completion questionnaire surveys done under exam conditions in a classroom supervised by two researchers with no teachers present.
6\. Data collection methods
Participants will be surveyed at baseline, at the end of the programme (20 weeks) and one year after the programme is completed (72 weeks). Girls will be offered £10 vouchers to girls in intervention and comparison group for completion of questionnaires. A repeat visit to schools will be made to survey those students missing the previous visit. Multiple forms of contact information will also be taken so that students repeatedly missing from, or moving, school can be tracked down.
7\. Analysis
There is no plan for an interim analysis done to inform whether stopping rules should be implemented because by the time an analysis of follow-up 2 data for cohort 1 is available then cohort 2 will already have experienced the intervention. Analysis of effects estimates will be done on an 'intention to treat' basis with all participants and controls originally recruited to the study plus any reserves brought into the programme to augment drop-outs being included in the analysis regardless of attendance in the programme. Logistic regression will be used to estimate effect sizes for binary outcomes and ANOVA for continuous measures. Each will generate crude and adjusted estimates, the latter adjusting for any significant (P\<0.1) baseline differences in socio-demographic measures or baseline values of outcome measures.
8\. Governance
The trial will be managed by an executive group consisting of researchers from the National Centre for Social Research and the London School of Hygiene and Tropical Medicine which will meet monthly. The Trial Steering Group function will be fulfilled by a committee comprising representatives from each research team, the funders (Department for Education), the providers (Teens and Toddlers) as well as two independent members (Meg Wiggins, Senior Research Fellow, Social Science Research unit, Institute of Education and Sally Hopewell, Senior Research Fellow, Centre for Statistics in Medicine, University of Oxford). This group will meet quarterly to scrutinise trial design and methods and assess data quality. This group will thus including within its function that of a data monitoring committee.
