Figure 1 presents the participant flow through the trial. There were no eligibility criteria for communities and all community leaders agreed to support participation in the study with none excluded. Using previously described methods these communities were matched on expected births, by nursing teams, into eight clusters that were then randomized in order that each nurse team had at least one intervention and control group. Four clusters were randomized to intervention, four to control. All intervention clusters received the intervention with no control clusters receiving the intervention, as planned. Enrollment began October 10, 2018 and ceased September 24, 2019. The mean size of intervention clusters was 25 women with a 4.7 standard deviation. The mean cluster size of control clusters was 27 women with a 15.4 standard deviation. Of the 141 women approached to participate in the study in the intervention clusters, 26 were not eligible and 7 declined to participate. Of 108 women who received the intervention, 3-month outcome data was available for 101 of them. In the control clusters, of 143 women approached for study participation 24 were not eligible and 19 declined to consent. Of the 100 women who were enrolled, none were lost to follow-up by 3 months.
Table 1 describes the overall study population and the subpopulations of women by intervention arm. Women were young, with a median age around 22 years old, the majority had had some education (around 90%), and most were married (around 88%). The largest subpopulation of the cohort was primiparous (around 39%) and the majority had received greater than four prenatal visits (about 81%) through the Madres Sanas program. About two-thirds of women experienced vaginal birth and were delivered by a nurse or physician, about half delivered in a facility, and around three-fourths gave birth to infant weighing at least 2500 grams. The majority of babies were born alive and remained alive by the 72-hour postpartum visit (around 93%), most were alive at the 40-day enrollment visit (about 95%), around 7% of women were already sexually active by the time of enrollment, and over one-third of women did not desire future fertility (about 37%). Randomization was effective and study arms were not statistically different by any of these characteristics when checked (p-values not shown).
Figure 2 visually illustrates the contraceptive choices of women in the intervention arm of the trial who were offered free contraceptives in their homes at the time of their 40-day postpartum visit as well as the method they were using by 3 months post-enrollment. Seventy-four of the 108 women (68.5%) in the intervention clusters initiated a contraceptive method at enrollment; 2 opted for condoms (1.9%), 5 for pills (4.6%), 37 for the injection (34.3%), and 30 for the implant (27.8%). The remaining 34 women (31.5%) declined to initiate a method with no missing data at that timepoint. By the 3-month post-enrollment timepoint, the continuation rates of these methods were 50%, 80%, 76%, and 90%, respectively. In the intervention population, by 3 months, 18 women were still not using a method (16.7%) and 7 (6.5%) had been lost to follow-up. Of those that had initiated, continued, or started a second method, 3 women (2.8%) were using natural family planning, 1 (0.9%) condoms, 4 pills (3.7%), 46 (42.6%) the injection, 28 (25.9%) the implant, and 1 (0.9%) sought female sterilization.
Table 2 presents secondary outcomes collected at 3 months in the intervention arm. Of women still on their initial method of choice, 90% were very satisfied. Of those who discontinued (n = 11), 3 discontinued due to side effects, 2 due to partner preference, 1 did not know why she discontinued, 2 did not want to contracept anymore, 1 sought sterilization, and 1 forgot to take her pills. Of the women who discontinued and did not initiate another method (n = 4), each one had a different reason, including: partner preference, could not continue due to resource constraints, did not know, and did not want to contracept anymore. In the population of women who originally declined a method at the 40 day visit (n = 34), 19 subsequently initiated a method by seeking it in the community, and 100% of them reported they were very satisfied on that method. Of the remaining 15 women still using no method by 3 months (44% of those who originally declined), 13 reported a reason for not initiating a method, which included: not having a partner or due to partner preference, not knowing, unable to obtain a method because the health post was closed, declining to start because menses had not resumed, not wanting to use a method, and not wanting to “ruin” her uterus.
Figure 3 and Table 3 illustrate the experience of women in the control clusters. Of the 100 women followed through 3 months, with no lost to follow-up, 44 (44%) had not initiated a method. Reasons these women cited for not choosing a method were: not having a partner or due to partner preference, could not continue due to resource constraints, did not know, did not want to, wanted to menstruate first, wanted to wait until the baby was older, was already using or wanted another method (natural family planning, intrauterine device, sterilization), and forgot or was afraid to start a method. In the remaining 56 women who chose to start a contraceptive by 3 months, 53 sought the injection, 1 used natural family planning, and 1 woman obtained the implant. These 56 women were largely very satisfied or a little satisfied with the method they chose (n = 53), with 2 women reporting a little dissatisfaction and 1 woman reporting she was very dissatisfied.
Table 4 presents the primary outcome of the paper, which was comparing implant uptake at 3 months by study arm. First, however, it illustrates overall contraceptive use by 3 months and how that varies by study arm. In control clusters 56.0% of women were using a method compared to 76.8% of women in intervention clusters, p < 0.001, OR 3.6 [1.9,6.9]. It then shows how that usage breaks down by method of contraception including modern and traditional methods. Before this current study regarding implant usage, we observed high usage of short-acting methods, which included condoms, pills, and the injection. The use of these methods varied by study arm with 94.6% of women in control clusters opting for these methods compared to 61.5% of women in intervention clusters, p <0.001, OR 0.09 [0.02,0.3]. Finally, for our primary outcome, implant usage by 3 months was statistically significantly different between study arms with 2 women from control clusters using the method compared to 28 women in the intervention arm, p < 0.001, OR 18.8 [4.3,81.4]. Of note, no adverse outcomes or pregnancies have been reported to date (data not shown).