The JT was implemented in the Ouémé department in Benin and was evaluated using a stratified two-stage sampling design at baseline and endline. The Ouémé department was selected for the intervention because it offered a new area to test the approach and was geographically distinct from the TJ pilot site (Couffo department). Additionally, there have been no other social and behavior change efforts in the study area beyond national efforts, such as the government’s annual FP campaign. The Atlantic department served as a control or comparison site. It was matched in the same way with Ouémé on socio-demographic characteristics. Although contraceptive use rates vary considerably across sub-regions in Benin , FP indicators were relatively similar between the two sites, with rates of contraceptive use at 10.4% in Atlantique and 9.0% in Ouémé; unmet need rates were 28.2% in Atlantique and 32.4% in Ouémé .
During the first stage, 32 villages (16 intervention villages out of 44 in the Ouémé department, 16 comparison villages in the Atlantique department) were drawn with a probability proportional to the sampling size based on on estimates of the adult population (15-59 years) according to national statistics of 2015. census data. The researchers stratified the sample by region and village size for the second stage with support from TJ staff (Table 1). In each sampled village, all households were enumerated to form the sampling frame for the random selection of individual households to be included in the study. All occupants of the household were listed, and one eligible woman and one eligible man were selected for interview. Six hundred and fifty (650) households were sampled in the intervention area, Ouémé, and 627 in the comparison sites in the Atlantic. Survey participants included women aged 18-44 who were married or in union and men who were married or in union with women aged 18-44 living in the communities studied. Men and women were not necessarily couples.
The cross-sectional household survey data was collected in May 2015 before the start of the intervention activities and again fourteen months later in September 2016 after the end of the intervention activities. Baseline and end-of-study tools for men and women can be found in supplementary files. Data collectors used portable tablets to collect data with secure submission to a secure cloud server to store collected data. In cases where eligible people were not home, interviewers returned home up to two times to conduct the survey. If no qualified person was found, interviewers replaced the household. Initially, 519 women and 505 men were interviewed in Ouémé, and 524 women and 525 men were interviewed in Atlantique. At the end of the study, 523 women and 522 men were interviewed in Ouémé, and 523 women and 523 men were interviewed in Atlantique.
The study protocol was reviewed and approved by the Health Research Ethics Board of the Institute of Applied Biomedical Sciences in Benin and by the Institutional Review Board of Georgetown University in the United States in 2012.
The JT assessment included four primary outcomes, which were assessed using binary measures of (1) current use of modern contraception; (2) intention to use a modern FP method; (3) real need met in FP; and (4) perceived met FP needs.
Current use of modern contraception was based on methods available in Benin, including female and male sterilization, oral contraceptives, intrauterine devices, injectables, implants, female and male condoms, diaphragms, spermicidal foams/jelly, method standard days and lactational amenorrhea method (LAM). Intention to use contraception was assessed by asking respondents, “Do you think you will use any method to delay or avoid getting pregnant at any time in the future?” Real and perceived met need outcomes were defined following a decision framework on the calculation of unmet need . Real need met included women who were using a modern method at the time of the interview and who did not wish to become pregnant within a year. Perceived met need included women who were using a traditional FP method or a non-LAM form of breastfeeding (and believed they were protected from pregnancy) and did not wish to become pregnant within a year. Benin’s traditional methods were included in the survey: periodic abstinence, weaning, herbal teas, traditional rings and traditional belts.
To determine whether needs were being met for modern FP, the survey included a series of questions about current pregnancy status, pregnancy intention, current FP use, and, if applicable, the method. The survey also included questions for pregnant women regarding desire for the current pregnancy, and for all respondents not using a modern method, the reasons for not using a modern method.
Intermediate outcome measures
Intermediate outcomes included: individuals’ self-efficacy and their perceived ability to access contraception; couple communication on fertility and FP; social media dissemination behaviors, including public communication about FP and fertility; social media outreach metrics such as seeking advice and sharing advice with others; and normative attitudes and beliefs about fertility, FP, and gender. Table 3 presents a selection of indicators used to measure the various intermediate results.
Sociodemographic characteristics of participants included: age in years (18-24, 25-34, 35 and older); education (none, primary, secondary or higher); religion (Christian, traditional, Muslim, none); ethnicity (linked to the Fon/Fon, Adja, Yoruba); number of living children (none, one, two, three, four, five or more); and have co-wives (no or yes).
The analysis proceeded in four stages, using models stratified by sex. We first assess the balance between the intervention and comparison groups on baseline characteristics using Pearson’s chi-square tests of independence. We then perform bivariate tests of independence to assess differences between baseline and endline on intermediate and major FP outcomes: current use, intention to use, actual met need, and perceived met need. Next, we use logistic regression to examine the effect of TJ’s primary activities on the program’s intermediate and primary FP outcomes. These activities included: 1. Interpersonal Communication Activities (IPC, which relate to Components 2 and 5 of the TJ ToC); 2. Listen to advocacy from influential people (Leaders, which refers to Component 3); and 3. Hear the radio broadcasts of the TJ group discussions (Radio, which relates to component 4). All models controlled for age, education, religion, number of living children, and co-wives. Ethnicity was excluded from the analyzes due to collinearity.
Finally, we also performed difference-in-differences (DID) analyzes to compare the change in outcomes between baseline and follow-up in the intervention site versus the change over time in the control group. . The DID approach is a powerful statistical method that controls for both observed and unobserved time-invariant factors falsely correlated with treatment (intervention) [42, 43]. DID analyzes allow researchers to assume that, subject to the covariates in the model, the observed change in the comparison sites represents what would have happened in the treatment sites if the intervention had never taken place.