Household Bargaining and Female Condom Use

Our study took place in Matola, which is the capital of Maputo Province and lies approximately 10km west of Maputo City. HIV prevalence in Maputo Province is high and disproportionately affects women, at an estimated 29.6% for women and 15.8% for men (Ministerio da Saude, 2015). The intervention is run by Pathfinder International, and is aimed at women in populations with high HIV transmission risk.

The programme consists of six interactive group sessions lasting ninety minutes each, held fortnightly over a three month period. Pathfinder trains female health workers from the local area to facilitate the programme, and thus facilitators are socially proximal to the participants. Group sizes range from a minimum of five to a maximum of twelve women aged 18-49 years, with on average 10 women per group, creating an environment conducive to discussion.

The sessions are guided by a curriculum covering: information on female condoms and demonstration of their use on pelvic models; information about other contraceptive methods; information on HIV/AIDS and other STIs; and discussions around consent, negotiation of contraceptive use, intimate partner violence, and women's rights. Female condoms are also added to the set of products carried by local health workers, which already include male condoms, that participants can access freely and discreetly at the end of each session.

2014-09-01

2014-12-31

Key outcome variables are 1) female condom use (measured as: Ever use of female condoms, Use of female condoms in the last 30 days, Current use of female condoms -- with "Current use" being the preferred outcome variable), and 2) frequency of sex acts (measured using the diaries as: Likelihood of a sex act per individual per week)

Secondary outcome variable: male condom use (measured as: Ever use of male condoms, Use of male condoms in the last 30 days, Current use of male condoms)

We conducted a phased-in randomised control trial in Maputo Province, Mozambique, to study the adoption of condoms, and to examine how intra-household bargaining affects the adoption of female condoms when they are introduced in a setting where only male condoms are widely available. Specifically, we assess the short-run impacts of the intervention on women who were assigned to participate in the programme in the second half of 2014, compared to women who were assigned to participate six months later. In addition to baseline and
endline data, we collected weekly sexual diary data for a subsample of the women. This allows us to investigate impacts at the sex-act level, including effects on the frequency of sex acts. To measure bargaining power, we collect information about assets brought by the woman to the relationship, and also use two different survey modules covering decision-making and power dynamics in the relationship.

Pathfinder International began its female condom programme in Matola in 2011. We expanded the programme to four additional neighbourhoods in 2014, using a phased-in experimental design. The healthcare workers who facilitated the programme first conducted
door-to-door recruitment to identify women willing to participate. The eligibility criteria were threefold: women needed to be between 18 and 49 years of age, sexually active, and not pregnant. All recruited women participated in a baseline survey.

Once the entire sample had responded to the baseline survey, participants were assigned to either the first or second wave of group sessions. Participants in the first wave constitute our treatment group. Participants in the second wave constitute our control group. Randomization was done at the level of the facilitator: half of the respondents recruited by each facilitator were allocated to the first round of sessions and half were allocated to the second round of sessions. The reason for stratifying on facilitator was to improve power, and to ensure that there would be enough space for treatment and control participants to attend sessions close to their home.

The treatment group then received the intervention from September-December 2014. The endline survey was conducted in February-March 2015, two to three months after treated individuals had received their last group session. Baseline participants (both treatment and control) were traced and administered the endline survey. Following the endline survey, the control group then received the intervention from
March-May 2015.

During the period between baseline and end of the intervention, a subsample of participants from both the treatment group and the control group participated in weekly coital/sexual diaries.

Randomization was done in office using a computer program: A member of the research team took the list of respondents for each facilitator,
sorted them by a randomly-generated number, and assigned the first half to treatment and the second half to control.

The unit of randomization was the individual, stratified by facilitator. That is, the women recruited per facilitator were randomly assigned to the treatment or the control group -- within facilitators.

No

Planned number of units: 340 women (20 women per facilitator)

Planned number of observations: 340 women (20 women per facilitator)

Planned sample size: 170 women in treatment group, 170 women in control group.

Based on the final realized sample, we have 80% power to detect the following minimum detectable effect sizes at the 5% level in a two-tailed test: ever use -- female condoms 7.6 pp; use last 30 days -- female condoms 3.5 pp; current use -- female condoms 4.9 pp. pp, male condoms 13.9 pp.