Effectiveness of post-partum family planning interventions on contraceptive use and method mix at 1 year after childbirth in Kinshasa, DR Congo (Yam Daabo): a single-blind, cluster-randomised controlled trial

Background: In rural Burkina Faso, a package of six low-technology, post-partum contraceptive interventions (ie, refresher training for providers, a counselling tool, supportive supervision, daily availability of contraceptive services, client appointment cards, and invitation letters to attend appointments for partners), aimed at strengthening existing primary health-care services and enhancing demand for them, doubled the use of modern contraceptives at 12 months post partum (ie, 55% uptake in intervention recipients vs 29% in routine-care users). This study assessed the effect of a similar package but in urban settings of Kinshasa province, Democratic Republic of the Congo, in an effort to reduce the unmet need for post-partum family planning. Methods: Yam Daabo was a multi-intervention, single-blinded, cluster-randomised controlled trial done in six primary health-care centres (clusters) in Kinshasa. Centres were randomly allocated to receive the six-component intervention or standard antenatal and postnatal care in matched pairs (1:1) on the basis of number of monthly births, the ratio of health workers per population in the health zone, and the urban and suburban settings. Only data analysts could be masked to cluster allocation. Health-care facilities were eligible if they provided a continuum of antenatal, delivery, and postnatal care, were well stocked with contraceptives, and were situated close to the main study centre. All pregnant women presenting to the six centres were eligible if they were in their third pregnancy trimester and had no counterindications to deliver in the facility. The main outcome was prevalence of use of modern contraceptives at 12 months after delivery. Analysis was by modified intention-to-treat using generalised linear mixed models or Fisher’s exact test for small groups. Prevalence ratios were adjusted for cluster effects and baseline characteristics. This study was registered with the Pan-African Clinical Trials Registry (PACTR201609001784334). Findings: From July 1, 2016, to Feb 2, 2017, eight of 52 clinics assessed for eligibility met the criteria and were randomised. Of 690 women approached, 576 (83%) women were enrolled: 286 in the four intervention clusters and 290 in the four control clusters. Of them, 519 (90%) completed the 12-month study exit interview (252 in the intervention group and 267 in the control group) and were included in the intention-to-treat analysis. At 12 months, 115 (46%) of 252 women in the intervention group and 94 (35%) of 267 in the control group were using modern contraceptives (adjusted prevalence ratio [PR] 1·58, 95% CI 0·74–3·38), with significant differences in the use of contraceptive implants (22% vs 6%; adjusted PR 4·36, 95% CI 1·96–9·70), but without difference in the use of short-acting contraceptives (23% vs 28%; 0·92, 0·29–2·98) and non-modern or inappropriate methods (7% vs 18%; 0·45, 0·13–1·54). There were no serious adverse events or maternal deaths related to the study. Interpretation: The Yam Daabo intervention package did not have a significant effect on the overall use of effective modern contraceptives but significantly increased implant use in women post partum who live in urban settings in Kinshasa up to a year after childbirth. However, interferences from external family planning initiatives in the control group might have diminished differences between the services received. Such an intervention could be potentially relevant in similar contexts in DR Congo and other countries. Funding: Government of France; UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction. The Yam Daabo study was a pragmatic, cluster-randomised, multi-interventional trial done in eight primary health centres offering reproductive, maternal, neonatal, and child health services in Kinshasa province. The centres were Esengo, Esperodi, Kindele, Libondi, Mama Mosalisi, Ntombwa Ya Maria, Saint Paul, and Tshimungu (figure 1, appendix p 2).14 The central research centre in DR Congo was based at the School of Public Health of the University of Kinshasa, and the overall research coordinating centre was at WHO headquarters (Geneva, Switzerland). Our post-partum family planning intervention package, identified during the formative study phase, comprised three facility-based interventions (ie, refresher training of service providers, regularly scheduled and strengthened supportive supervision of providers, and enhanced availability of contraceptive services 7 days a week), and three individual-based interventions (ie, a post-partum family planning counselling tool, appointment cards for women, and invitation letters for partners to attend appointments). The implementation of three of the six components favoured a cluster design for the trial, with randomisation at the level of the study centre. Furthermore, potential contamination between intervention and control groups for the other three individual-based interventions could be minimised by the cluster design. The intervention package was offered to women allocated to the intervention group and routine care (information about the benefits for mothers and children of spacing births and provision of basic contraceptives) to those allocated to the control group (appendix pp 4–6). Kinshasa province and study area with intervention clusters (red) and control clusters (green) Health facilities were eligible if they provided a continuum of antenatal, delivery, and postnatal care; had at least three modern contraceptive methods available, including a barrier method such as condoms, a short-term method such as pills, a long-term method such as an intrauterine device, and referrals for permanent methods; experienced no depletion of stock of contraceptives during the 6 months preceding the study; cumulated an average of at least 30 deliveries per month; were located within a 4 h drive from the research centre; and were willing to participate. All pregnant women attending the clinic for antenatal care were invited to participate in the study. They were eligible if they were in their third pregnancy trimester; their health and pregnancy situation allowed for a birth at the health centre; they had the intention to attend antenatal, delivery, and postnatal care at the health centre; they were not participating in another study; and they provided written informed consent (verbal informed assent for minors aged <18 years and written assent from their parents or guardians). Study approval was obtained by the School of Public Health (University of Kinshasa) ethics committee in DR Congo (reference number ESP/CE/039b/2016) and WHO Research Ethics Review Committee in Geneva, Switzerland (protocol ID RPC757). This study is registered with the Pan African Clinical Trials Registry (PACTR201609001784334). The eight sites were matched in pairs according to the number of monthly births, the ratio of health workers per population in the health zone, and the urban and suburban settings, without specific thresholds for these criteria (appendix p 3). Within each of the four study centre pairs, a site was randomly assigned to the intervention group or to the control group (1:1 ratio). This computer-based randomisation was done four times, once for each pair. No restriction in the randomisation process was required. All eligible participants consecutively presenting at the health centres were included in the clusters. The nature of the interventions did not allow masking to cluster assignment of participants, health staff, research assistants assigned to each centre, and the rest of the research team members. Only data analysts were masked to cluster assignment (ie, they received no information about the cluster allocation and did not interact with the field team). Masking of data was achieved by replacing study site names with numbers and group names with letters. The rationale for our trial approach is based on WHO's seminal 2013 publication of programmatic strategies for post-partum family planning2 and three other systematic reviews (without meta-analysis) on post-partum family planning interventions in low-income and middle-income countries published between 2014 and 2016.6, 7, 8 These publications suggest that the following interventions could have a positive effect on post-partum contraceptive uptake: counselling activities during antenatal care; provision of post-partum family planning information, education, and counselling materials before women are discharged from health facilities after birth, including provision of emergency contraception for women using the lactational amenorrhoea method; promotion by community-based counsellors of exclusive breastfeeding practices before 5−6 months post partum; access to contraceptive methods immediately after birth, including intrauterine devices; provider competencies in quality counselling and the provision of quality services with several readily available products; and long programmes with several contact points between providers and clients across the continuum of care versus short antenatal interventions. The intervention package was designed through participatory action research and the process and contents were detailed elsewhere (appendix pp 4–6).15, 16 The working hypotheses for the selection of the post-partum family planning interventions to be included in the package were as follows. Interventions should strengthen existing antenatal and postnatal care services by means of low-dose, high-frequency interventions. We assumed that they would be more effective than high-dose, low-frequency strategies that promote a specific method over another and restrict services to a narrow timeframe, such as before home discharge after delivery, or during the 6-week postnatal care visits dedicated to post-partum family planning. Drawing from participatory action research principles,17 key actors, including clients and providers, should be meaningfully engaged in the package design, implementation, and research. Such participation would ensure that the package reflects field reality, including restricted clinical capacity and human and financial resources, and is feasible, sustainable, and scalable while also aligned with national health policies. We followed WHO recommendations for medical eligibility of contraceptive use to define modern contraceptives appropriate up to 12 months post partum.18 We categorised modern contraceptives as: long-acting and reversible, including implants and intrauterine devices; short-acting, including injectables, pills, emergency contraception, male and female condoms, and other less commonly used methods; permanent methods (male and female sterilisation); and lactational amenorrhoea. Contraceptives were further defined as modern and appropriate and non-modern or inappropriate. Non-modern contraceptives used traditional methods such as withdrawal and abstinence. Inappropriate contraceptives were lactational amenorrhoea if used after 6 months and calendar-based methods if used during the first 12 months post partum. We assumed that as most women breastfeed up to 2 years, they would not fulfil the initiation requirement of calendar-based methods of having at least three regular menstrual cycles before 12 months. Participants received individual-based interventions during third-trimester antenatal care visits and postnatal care follow-up visits according to national practice (typically on clinic discharge at 24–48 h after delivery, then at 1 week, 6 weeks, then at months 6 and 9 post partum) before trial exit at month 12. At each visit, providers were instructed to use the counseling tool to offer information and services and give women an appointment card for the following visit. Providers discussed the invitation letter for the partner during the first study visit and let the participants choose whether to take it. In all study sites, research assistants held a journal where they recorded events that could have an influence on the implementation or outcomes of the trial for its interpretation, but they were not included in the analysis and did not modify the trial (eg, staff turnover, depletion of stocks of contraceptives, local civil unrest). The primary outcome was prevalence of use of modern contraceptives at month 12 post partum. We amended our original protocol14 to extend the follow-up period from month 9 to month 12 post partum to allow for better comparability with other published research, after the initial funder's research grant deadline was chosen. This extension was approved by the ethics committees at WHO headquarters and country level. We also reported key secondary outcomes after delivery: prevalence of modern contraceptive adoption before discharge from the health facility (assessed within 48 h after delivery); at 1 week (assessed on day 7); at 6 weeks (assessed at 45 days), which coincided with the visit that women are encouraged to attend to specifically discuss post-partum family planning options as per national recommendations; and at 6 months (assessed on the first day of month 6), which corresponded to the latest point when transition from the lactational amenorrhoea method to another modern method should occur. Contraceptive method mix was also reported for each measurement point. Serious adverse events had to be announced to the principal investigator who assessed whether they were related to the study. We used the following assumptions when calculating the target population size. Women in the intervention group wanting to restrict or space their pregnancies would already use a modern contraceptive method at 6 months after delivery when the lactational amenorrhoea method would no longer be suitable. Therefore, although the main study outcome was prevalence of modern contraceptive use at 12 months post partum, we used country data at 6 months to determine our population size. The population size was estimated using the 2013–14 Demographic and Health Survey data:11 we assumed a 5% uptake of modern contraceptives in the control group (on the basis of reported prevalence of modern contraceptive use of 5% in women at 6 months post partum),12 and an increase to 20% in the intervention group (on the basis of reported prevalence of modern contraceptive use of 15% in women in the general population, to which we added 5 percentage points given the high unmet need for post-partum family planning).19 Assuming an intracluster correlation coefficient of 0·02 (no specific evidence existed for its value from the literature),20 the experimental group and the control group each had four study sites with at least 60 participants per site.21 This number allowed for a statistical power of 93% to detect a difference of 15% to a level of significance of 5% at the individual level (and not at the facility level). Assuming a 10% participant loss to follow-up, each facility recruited at least 70 pregnant women for a cohort of at least 280 pregnant women in each study group, and in total at least 560 participants. The research coordination team at WHO in Geneva developed the paper-based study case report forms with inputs from the country researchers who tested advanced drafts with an appropriate sample of mock clients from sites included in the formative study phase. Data for the trial were collected by trained research assistants when women attended clinic visits or by phone or a home visit when they did not. Data entry was done by the research team of the School of Public Health in Kinshasa and checked in Geneva by use of OpenClinica (version 3.11). Data from all eligible participants were analysed with IBM SPSS Statistics (version 21.0), R (version 3.4.3), and WINPEPI (PEPI-for-Windows, version 11.50). Generalised linear mixed models (log binomial and log Poisson) were used to assess the effect of the package on prevalence ratios (PRs) of main outcomes with 95% CIs, comparing intervention and control groups, while accounting for clustering and adjusting for potential confounders (women's baseline characteristics that were imbalanced between groups). Fisher's exact test was used when samples were small and models did not converge. WINPEPI was used to obtain a post-hoc global estimation of the intracluster correlation coefficient. We did an intention-to-treat analysis that included all women irrespective of whether they continued to visit the clinic after enrolment. Because women had to see a provider to receive the intervention package, we planned to do a per-protocol analysis focused on participants who attended all recommended follow-up visits, but this was not done because of low trial completion by participants in both groups. The study funder had no role in study design, data collection, data analysis, data interpretation, or writing the article. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.