Women who were beginning a new method of contraception at one of six urban clinics in Accra and Kumasi, Ghana during the enrollment period (from September 15 to November 2, 2017) were invited to participate in our study. Eligible participants were identified to study team members by clinic staff, and were approached by trained research assistants. Those who were interested in participating were taken to a small room near the waiting room to complete a pre-visit survey. After their counseling session, they were interviewed again. Participants were called at 3, 6, 9, and 12 months post-enrollment (December 2017-November 2018) to determine continuation. Records were matched by participant phone number.
In their pre-visit survey, participants—all women between ages 18 and 49—were asked about their demographics (age, gravidity, parity, marital status, educational level) and questions to assess their previous use of contraception, their knowledge about contraceptive methods, and their method preferences. Participants also completed the Desire to Avoid Pregnancy (DAP) scale. This newly developed scale is designed to capture the sometimes contradictory feelings a woman has about a potential pregnancy. Participants are asked how much they agree with a range of statements such as “If I had a baby in the next year, it would be bad for my life,” “Thinking about having a baby within the next year makes me smile,” and “If I had a baby in the next year, it would be hard for me to manage raising the child.” We included this scale in both the initial and follow-up contacts, as we hypothesized that side effects would be differentially tolerated by women with different levels of desire to avoid pregnancy. For example, women who are highly motivated to avoid pregnancy might be more tolerant of menstrual disruption, while those who hold more positive feelings toward a potential pregnancy might discontinue more readily.
In the post-visit survey, participants were asked if they were leaving with a method, and if so, which method. They were also asked if this was the method they wanted to be using. During follow-up, participants were asked if they were still using the same method they had adopted at the beginning of the study. Those who indicated they were not using the same method were asked why, with responses recorded in an open-ended format. During analysis, we coded these responses into method-related factors and pregnancy desire. Participants were also asked if they were using any method of contraception, and if so, which method, with responses recorded in an open-ended format. During analysis, these were coded into other modern methods, emergency contraception (EC), condoms, and no method. Participants who were using another modern method were coded as having switched, while those who reported no method or switching to condoms only or EC were coded as having stopped.
Due to the impact of side effects on contraceptive discontinuation rates, all participants were asked at follow-up if they had experienced any side or undesirable effects with their method. Those who answered yes were asked to describe the side effects in an open-ended format. These were coded into menstrual side effects (decreased bleeding, spotting, prolonged bleeding), weight changes, and dizziness/headaches. Participants were further asked if they had been expecting these side effects.
To assess level of satisfaction, at each follow-up encounter, all participants were asked if they received the method they wanted, if they would choose this method again if they had the opportunity, if they would recommend their method to a friend, and their level of decision-making at the initial visit (using the question “During this visit, who made the decisions about what birth control method you would use?” with “The provider,” “Mostly the provider,” “The provider and me together,” “Mostly me,” and “Me” as choices). The response to “would you choose this method again?” was summed across the number of times the participant answered the question and the mean was assessed. Finally, all participants answered the DAP scale at each follow-up, and their mean DAP score across the number of times they answered the DAP was assessed.
Analysis
Bivariate logistic regression analysis was performed to identify factors associated with the outcome variable (discontinued for reason other than pregnancy or desired pregnancy). Factors that were found to be significantly associated at the .1 level, as well as those deemed important to control for, were entered into a multivariate logistic regression analysis. We used marginal effects via the margins command in Stata, as well as odds ratios, to interpret the results of the regression.