Impact Evaluation Study of The Faithful House Programme on Violence Reduction in Families

A mixed-method evaluation of The Faithful House is proposed. In the main, we will conduct a randomized control trial of the education program to determine treatment effect. We will also conduct semi-structured interviews and focus groups with stakeholders. This more qualitative work will be done in order to refine our program and evaluation and to assess the fidelity of the intervention and assess and unintended consequences. Informed consent will be sought from all prospective participants.

The source population is families that receive HIV care from Arusha Lutheran Medical Centre and Selian Hospital. Participation and related compliance is enhanced because subjects come to the hospital for HIV treatment. Couples with children ages 10 to 17 years will be recruited to participate in a Faithful House workshop and study by hospital volunteers. All patients attending the clinic during the recruiting period will be screened based on eligibility criteria. The list of patients screened will be checked against the clinic register to verify that all potential participants were screened and all eligible respondents were invited to participate in the workshop and study.

The list of recruited couples will be randomized to receive the intervention or to serve as the control arm. The control arm will be offered the intervention at the conclusion of the study. The goal is to have N=150 couples in the treatment arm and N=150 couples in the control arm. The rationale for the sample size is twofold. First, given the estimated variances from surveillance studies the sample size permits the estimates of small to moderate effect sizes at 80% power and 5% Type I error. In simple terms, the proposed evaluation is strongly powered to detect politically and clinically meaningful effects. Second, given time, budget, and patient flow characteristics, 300 couples is near the maximum of our capacity. Baseline measures will be taken via survey on all participants in a clinical setting about one month prior to beginning the education program. Post program measures will be taken in a clinical setting three months after the intervention is complete. A context-based pretest and post-test will be administered to only those in the treatment group on the first and last day of the intervention.

Gender and age specific surveys will be used to interview men, women and children and questions on self-reported violence will be included in women and children's questionnaires. Answers to questions on violence will be collected using non-verbal response cards to protect confidentiality and to assure respondents that enumerators will not know their responses.

Analyses will focus mainly on difference in estimates for the primary outcome measures of the self reported incidence by mothers and children (10-17) of acts of violence against women and children including child-witnessed violence in the past three months before the intervention and three months post intervention. Sub-group analysis by gender will be conducted. We will look at within group difference and between group differences. Retention will be assessed by comparing results from the pretest and post-test given on the first and last day of the intervention.

Secondary analyses on secondary outcomes such as parental attitudes toward violence and abuse and changes in intimate partner violence. We will examine the hypothesized theoretical links (e.g., mediation) between observed outcomes and program characteristics. A two-sample t-test will be used to look for differences in incidence of violence between the intervention and control groups. Multiple regression will be used to identify and examine predictor variables.

All data will be collected using electronic pads that contain self-checking logic within the questionnaire. Should missing data exceed 10%; multiple imputation procedures will be employed. However, our experience in similar studies is that imputation was not required because our fielding processes led to very high completions in the field.

Study instruments will include before focus group discussions (FGD) prior to the intervention, baseline, before and after workshop tests, three-month follow-up interviews, and post analysis FGD. Where possible, questions from other family and child violence studies will be utilized to allow for inter-study comparisons, especially focusing on evaluation studies of other The Faith House education programs and on the Violence Against Children in Tanzania National Survey conducted in 2009. All differences will be reported at a p <0.05. Data analysis will be completed using Stata v12. Importantly, we will also conduct a qualitative comparison of our outcome measures with published surveillance estimates in order to learn more about our source population and the potential impact of our program on the area if scaled up. As mentioned above, we will conduct interview and focus groups to improve program development, assess the fidelity of the implementation, and potential unintended consequences. Indeed, on-going study results will be monitored and a priori stopping rules assessed.