The study is a multi-country, two-phase, cRCT implemented in Machakos and Kisumu Counties, Kenya and in Nasarawa State, Nigeria. It was designed in collaboration with the Kenyan Ministry of Health (MOH) and the Nasarawa State MOH and Primary Health Care Development Agency in Nigeria.

In each country, 20 health facilities were selected and matched in pairs by: type of health facility (e.g., health center, hospital); client caseload for first ANC visit (i.e., ANC1); location (urban, peri-urban, rural); culturally similar catchment populations; and availability of FP services. The pairs had to be either within the same state (Nigeria) or county (Kenya), but did not need to be geographically adjacent. In Kenya, 10 clusters were assigned to each of the two counties where the study took place.

The unit of randomization was the health facility, which formed the clusters. For each pair of matched health facilities, randomization (1:1) was done to the intervention (group care) or control arm (the current standard of individual care according to country guidelines). Assignment to study arms was done through a lottery using pieces of paper in a basket bearing the words “intervention” or “control”.

In both study arms, all women provided written consent to participate in the study. In each health facility, 50–54 participants were consented and enrolled, along with three health care providers. For the intervention arm, four to five cohorts (groups of women) were formed per facility. Figure 1 summarizes the study design.

The study is being implemented in Machakos and Kisumu Counties in Kenya and in Nasarawa State, Nigeria. Countries were chosen based on interest to the funder. In-country locations were chosen in collaboration with local colleagues, based on adequate security for women to travel to ANC; functional local health system governance and structure; Jhpiego’s ability to leverage existing maternal, newborn, and child health/FP (MNCH/FP) projects to reduce implementation costs; and where possible, avoidance of other ongoing projects that could introduce bias. Table 1 presents key indicators for Kenya and Nigeria, comparing the national and state/county-level indicators in each country as appropriate ^{35, 36} .

In 2016, Kenya had an estimated population of 45 million. Machakos and Kisumu County populations were 1.1 million and 0.96 million, respectively. While ANC attendance is nearly universal, fewer than six in ten (57.6%) pregnant women attend four or more visits during pregnancy ³⁵ . The facility-based delivery is 61.2% nationally, 62.9% in Machakos County, and 69.5% in Kisumu County ³⁵ . Nationally, 52.9% of women receive PNC within two days of giving birth ³⁵ . FP use among postpartum women was estimated at 36% based on analysis of 2008–2009 Kenya Demographic and Health Survey (DHS) data ³⁷ .

Nigeria is the most populous nation in Africa, with an estimated 198 million inhabitants. Nasarawa State has an estimated population of 2.5 million. One-third of pregnant women (33.9%) in Nigeria do not receive any ANC ³⁶ . Half (51.1%) receive four or more ANC visits ³⁶ . The facility-based delivery rate is 35.8% nationally and 40.1% in Nasarawa State ³⁶ . Both ANC and delivery with a skilled birth attendant coverage have stagnated ³⁸ . Quality of ANC is poor ³⁹ . PNC within two days is 39.6% for the mother and 14.0% for the newborn ³⁷ . FP use among postpartum women was estimated at 15% based on analysis of 2008 Nigeria DHS data ³⁷ .

Both countries have national policy documents that promote key evidence-based MNCH/FP interventions—at least four ANC visits and a facility-based delivery with a skilled provider ^{40– 44} .

Eligibility for clusters. The intervention is multilevel with activities targeted to different participants in the selected health facilities.

Health facilities (clusters): A list of health facilities offering ANC and postnatal care (PNC) services in each study location was prepared by the study team, based on data from the national health information systems (HIS), which use the open source software platform District Health Information System 2 (DHIS2). Within the study areas, at the time of site selection, 551 health facilities in Nigeria and 410 in Kenya were providing ANC services. Study criteria dictate that women in group care need to be 20–24 weeks GA at the time of the first group meeting, thus a minimum ANC1 census was deemed the most important criteria to determine the ability to form cohorts. HIS data for the 12 months preceding the study was used to identify potential sites. In Nigeria, all facilities with more than 20 new ANC clients/month who initiated care at or before 20 weeks GA were considered for inclusion. GA at ANC1 was not available in the Kenya HIS. Therefore, assuming approximately 50% of women enter care < 24 weeks (participant criteria), all facilities with a minimum of 30 new ANC clients/month were considered. Members of the study team then visited all facilities to verify HIS ANC census numbers, by examining the ANC registers for the previous three months, and to verify additional selection and matching criteria. Facilities were selected in conjunction with county/state health officials and included consideration of “best match” ability as well as inclusion of urban/peri-urban/rural sites and varied levels of the health system ( Table 2).

Eligibility for participants. Study participants include pregnant women, health care providers, and sub-national health managers. The inclusion criteria are detailed in Table 2. There were no additional exclusion criteria.

Pregnant women: Pregnant women 15 years and older attending ANC1 in any of the study health facilities. Additional criteria on GA and availability were applied depending on the study arm.

Health care providers: Providers working in the participating health facilities primarily providing ANC and PNC services.

Sub-national health managers: Included either government officials who work in sub-national level health management positions or health facility in-charge in study settings.

The intervention is provision of an alternative service delivery model that delivers care in a group, compared to the current standard of care provided individually (i.e., one-on-one between a provider and an ANC client). Women in the intervention arm were invited to attend five G-ANC meetings (Phase 1) and four Healthy Mother, Healthy Baby (HMHB) meetings (Phase 2) in the year following birth.

Phase 1: G-ANC. At the first ANC visit (ANC1), pregnant women are assessed for GA. If eligible, women in the intervention group are offered the option of G-ANC for their subsequent ANC. Interested women are placed in a cohort of 8–15 other women of similar GA, all of whom will be 20–24 weeks pregnant at the time of a pre-scheduled cohort start date (Meeting 1). Thereafter, they attend subsequent routine visits together every month as G-ANC meetings. Five total scheduled meetings are held and recorded as ANC contacts, each lasting approximately two hours.

In each intervention facility, two ANC providers (nurses or midwives) and one CHEW were trained as G-ANC facilitators. In both countries, CHEWs are facility-based and provide some level of clinical care. Two facilitators were assigned to each cohort and were advised to co-facilitate meetings for that group throughout the intervention.

Prior to initiating the study, inputs were made to all facilities, in both arms, in the same quantities, to ensure comparable capacity for service delivery ( Table 3). The inputs included a select set of ANC commodities that would ordinarily be provided through the public sector health system. In addition, a one- to two-day refresher training on ANC and PPFP counseling was provided for all ANC providers at control sites and for G-ANC facilitators at intervention sites. Intervention facilities received additional G-ANC specific inputs as outlined in Table 3.

Clinical care in both study arms was provided according to national clinical standards and guidelines ^{47– 50} . No changes were made to clinical care protocols or content in any health facility. No additional activities were undertaken in either arm to improve the quality of clinical care (e.g., support for management of complications; integration of PPFP in postnatal wards).

At the first meeting, women agreed on what day and time they would like subsequent meetings to be and whether husbands should be invited to Meeting 3 ( Table 4). Women who have or develop complications requiring additional individual care are encouraged to continue attending group meetings.

Longitudinal group care registers were introduced in the intervention facilities and used for each cohort. They serve a dual purpose of data collection and reinforcement of key actions by providers and women, such as taking IFA supplements, sleeping under a LLIN, birth planning, and making plans regarding healthy timing and spacing of pregnancy.

Phase 2: HMHB. Healthy Mother, Healthy Baby meetings extend group care into the first year after birth ( Table 4). The same cohorts continue to meet, with their babies. Meetings are scheduled every three months (total of four HMHB meetings). Because of the staggered timing of births for women in a cohort, HMHB meetings are not designed to replace standard PNC visits (i.e., four visits in the first six weeks) as recommended by WHO and included in national guidelines of both countries ⁵¹ . Women in both arms are encouraged to follow country guidelines regarding immediate PNC, as well as national immunization schedules (when they do not correspond to scheduled group meetings).

In 2017, the G-ANC facilitators were trained to facilitate HMHB meetings and remained attached to the same cohorts in the intervention arm. Prior to the first cohort delivering, two to three providers from each facility in both arms will receive clinical training in PPFP methods.

Meeting frameworks and key content. The meeting frameworks for G-ANC and HMHB were designed to incorporate three key components of group care: 1) individual clinical assessment; 2) participatory, facilitated learning, and; 3) peer support. The frameworks were informed by evidence-based learning principles, social behavior change and communication principles, and social support theories. They were designed to include highly facilitative, nonhierarchical, patient-centered participatory approaches with attention to education and literacy levels and community cultural norms.

Clinical assessments include self-assessments (e.g., blood pressure, weight, and danger signs), and brief, individual, private assessments with the provider, behind screens within the group meeting space. To address the main topic of each meeting, large illustrations cards are used to facilitate discussion around key actions or danger signs. Questions are posed to the group while viewing each illustration to: promote sharing of knowledge, examine cause and effect, agree on actions to take, discuss barriers and solutions, and agree to share information with others. The women receive take-home booklets with small versions of the same illustrations to use as memory aids and resource guides for sharing information with family and friends. Other meeting components are designed to specifically support group identity and cohesion; promote empowerment, self-efficacy, and agency; and promote self-reflection and planned action.

Prior to introducing G-ANC in the intervention sites, 30 facilitators in each country were trained together by study staff in a five-day workshop offsite using the G-ANC framework and materials developed for this study. They were also trained to use self-reflective quality assurance and improvement tools, which included a scorecard for facilitation skills, a fidelity checklist, and prompts for identifying strengths and weaknesses and things to change or try at the next meeting.

Program officers (POs) were employed by the study to mentor facilitators, oversee research assistants, and support daily implementation of the study at all study facilities in their defined geographic area. POs are nurse-midwives who speak the local languages. POs aimed to attend 80% of the first three cohorts at each facility to provide mentoring on the G-ANC model and feedback on facilitation skills. Debriefs immediately after group meetings followed a structured coaching outline, incorporating reflections from the G-ANC facilitators.

Different outcomes are targeted for the different phases of the trial. The primary outcome for Phase 1 is the proportion of facility-based delivery among women enrolled in intervention (G-ANC) arm versus control (standard individual care) arm. For Phase 2, the main outcome is the proportion of women using a modern FP method (defined for study purposes as condoms, oral contraceptives, injectables, implants, IUDs, and sterilization) at 12 months postpartum, comparing the two study arms.

Secondary outcomes include additional measures of: service utilization; provision of quality ANC; experience of care by women and providers (measures of satisfaction and positive experience); MNCH/FP-related health literacy and uptake of healthy behaviors (e.g., sleeping under a LLINs, taking IFA supplements, early initiation of breastfeeding, knowledge of danger signs, knowledge of lactational amenorrhea method criteria); and improvement in agency, self-efficacy, and empowerment, specifically related to birth planning and PPFP ( Table 5).

All ANC service providers at all study sites were oriented to the study prior to initiation of enrollment. Enrollment ended January 2017 in Nigeria and June 2017 in Kenya. One research assistant (RA) was hired and assigned to each health facility in both study arms. Standardization of study procedures was facilitated by a study manual ( Supplementary File 1), which was provided to all study staff.

Health care providers. Study staff approached eligible ANC and PNC health care providers in participating health facilities, explained the purpose of the study, and obtained written consent prior to starting study enrollment. Health care providers were not blinded to group assignment. In intervention sites, two health care providers and one CHEW were recruited. In control sites, all health care providers who provide ANC services were included.

Pregnant women. Research assistants were stationed at their assigned site during ANC service provision. ANC providers documented GA, methods used to calculate GA, and estimated date of delivery for every new ANC client. Those < 24 weeks GA were referred by the ANC provider to speak to the RA for further eligibility screening. No personal identifiers were retained for clients > 24 weeks GA, or those who did not ultimately consent.

Each RA conducted screening and consent in a location in the facility that allowed for audio and visual privacy. Women were accorded an opportunity to hear more about the study through use of a client study information sheet and screening was done in the most appropriate of five languages (English; Kiswahili, Kamba, and Luo in Kenya; English and Hausa in Nigeria). After confirming eligibility, the RA then collected basic demographic data. No personal identifier information was collected prior to the consent process.

In the intervention arm, assignment to a cohort was done based on GA at the time of the pre-determined cohort start date (i.e., Meeting 1). Eligible women unable to attend the first meeting were not consented or assigned to a group; they were advised to continue with standard individual ANC. If a woman was eligible for the study, but there were no open cohorts (e.g., the appropriate cohort was full), she was advised to continue with standard individual ANC. In the control arm, all eligible women who were willing to participate were consented. Written consents were obtained for all study participants in their language of choice (from the five options). A baseline survey was administered immediately after the consent process.

If it was not possible to recruit a minimum of eight women at least three days before the scheduled cohort start date, the cohort was cancelled, and the assigned women re-evaluated for eligibility into another cohort. Women who become ineligible to participate due to cohort cancellation were advised to continue with standard individual ANC. Their data were then coded to reflect cohort cancellation and retained for reporting purposes. However, they were removed from the study sample, and no additional follow-up was done for these women.

Sub-national health managers and health facility in-charge. Sub-national health managers (i.e., county-level in Kenya; state-level in Nigeria) responsible for overseeing reproductive health programs were recruited by study staff via phone at the end of Phase 1 to participate in the in-depth interviews (IDIs). Priority was given to managers known to have had exposure to G-ANC at the end of Phase 1. An appointment for a face-to-face interview was scheduled with each respondent.

Sample size calculations were done separately for the two countries based on the assumptions that were relevant to the local situation. In each country, the goal was to ensure that the study was adequately powered and each cluster could recruit at least four cohorts.

Quantitative: The study is powered to detect a 15 percentage point difference in the primary outcome of interest—proportion of deliveries in a health facility—between the intervention and control arms at the end of Phase 1. For the purpose of sample size calculations, DHS data were used to estimate the average proportions of facility-based delivery in the study regions: 61% for Kenya and 40% for Nasarawa State in Nigeria respectively ^{35, 36} .

Intra-class correlation was estimated at 0.03 for both countries, which will lead to a design effect of 2.45, assuming an average of 30 new ANC clients per month in Kenya and 20 new ANC clients per month in Nigeria for each of the 20 health facilities in each country (i.e., 10 facilities per study arm). After factoring an attrition rate of 20%, sample sizes of 1,026 and 1,076 women were determined for Kenya and Nigeria, respectively. The study is powered at 85% for Kenya and 80% for Nigeria to maintain enrollment sufficient for four cohorts per health facility. Three health care providers were recruited per facility (120 total), including only those who routinely provide ANC services.

For Phase 2, we calculated the change in proportion of women who are using a modern FP method at 12 months after birth that would be detectable given the sample size enrolled for Phase 1. Current estimates of PPFP use from the most recent DHS were not available for Kenya or Nigeria. For Kenya, the rate therefore was estimated at 65%, based on rates of PPFP one year after birth in the 2014 endline results from the Kenya Urban Reproductive Health Initiative ⁵² . For Nigeria, sample size estimation used the contraceptive prevalence rate (modern methods) of 16% for Nasarawa State from the 2013 Nigeria DHS. Based on assumed 20% attrition in Phase 1 and an additional 20% attrition by the end of Phase 2, the study will be able to detect a 10-percentage point difference in PPFP use at 12 months in Kenya and nine percentage point difference in Nigeria between study arms.

Qualitative: Approximately 15 IDIs and six focus group discussions (FGDs) will be conducted per phase in each country. In sampling for both IDIs and FGDs, the aim is to draw a sample reflective of the full range of responses to the G-ANC intervention model. For Phase 1, the IDIs and FGDs were conducted with women and health care providers both near and far from district referral hospitals to capture variation in challenges related to training, logistics, supervision, transport, and care seeking based on location.

The study uses a mixed methods data collection approach (quantitative and qualitative). The data sources are client and service provider surveys; data extraction from facility records, G-ANC registers, patient held records; and FGDs and IDIs with providers, women, and sub-national health managers. Phase 1 data collection ended January 2018 in both countries. Phase 2 data collection concludes in October 2018. The two phases of the study use similar data collection approaches. Table 6 summarizes data collection by phase. Each participant was assigned a unique study ID that is used to link all the data from the same individual in the database. Qualitative data will be used to complement the quantitative data through more in-depth exploration of selected thematic areas.

Study data were collected and managed using electronic data capture tools hosted at Jhpiego through the REDCap ^TM (Research Electronic Data Capture) platform. REDCap is a secure, web-based application designed to support data capture for research studies.

Quantitative data

Surveys ( Supplementary File 1): Baseline surveys were completed at the time of enrollment, during the woman’s first ANC visit. Face-to-face interviews with recently delivered women (RDW) were conducted at the end of Phase 1 (3–6 weeks post-delivery) to gather information regarding the women’s ANC and delivery experience. For the RDW survey, RAs contacted women by mobile phone to schedule a time to visit them at home to conduct the survey, based on the expected day of delivery, from REDCap ^TM. RAs made up to five attempts to contact the women on different days of the week at different times of the day over a two-week period to schedule an interview. If phone numbers were not functional or women were not successfully contacted after 5 attempts, physical tracing was done using the information provided at enrollment.

Women are also scheduled to be surveyed after Phase 2 to collect data on MNCH/FP care received in the 12 months after birth. Data are captured electronically using tablets loaded with REDCap ^TM software. During both phases, a participant is determined to be lost to follow-up if both phone and physical tracing efforts are unsuccessful.

Data extraction: Per the study protocol RAs extracted data from different sources including: longitudinal group care registers; women-held records (Kenya only); and other relevant facility-based health records. The data were captured electronically through REDCap.

Qualitative data. Focus group discussions and IDIs were scheduled and conducted in 2017 toward the end of Phase 1 and will be conducted toward the end of Phase 2 in 2018. An international qualitative research team lead, trained, and oversaw local qualitative researchers familiar with MNCH/FP-related issues. The FGDs and IDIs were audio-recorded and then transcribed after obtaining consent from the study participants.

Focus group discussions: FGDs for health care providers explored attitudes toward providing group care; perceived changes in their communication and relationship with patients; changes in empowerment, self-efficacy, and satisfaction related to their perceived ability to do their job well; sustainability of group care; suggested changes to the model and logistics required to offer group care; and perceived effects of group care on colleagues and clients. For the women attending group care, FGDs explored: satisfaction with care; perceived changes in self-efficacy or empowerment; health literacy; improved communication and trust with providers and health system as a whole; factors in uptake and continuation of PPFP; and overall impression of the group care model. Phase 2 FGDs will focus on similar topics; with a further focus on the ways HMHB influenced beliefs and actions around PPFP, nutrition and early childhood development practices, couple communication, and development of social capital.

In-depth interviews: IDIs were conducted with a select group of women on pre-determined topics that might have been difficult to explore through FGDs. To explore how problem recognition and care seeking was influenced by G-ANC, criteria for selection included: non-responders, (i.e., women who only came to one or two meetings [those whose barriers we don’t understand]); women who attended G-ANC meetings but didn’t deliver in a facility; and women who experienced complications during pregnancy. The IDIs are conducted until saturation of new information is reached. In addition, sub-national health managers were also selected for IDIs. Phase 2 IDIs will follow a similar pattern, including inclusion of those who came to meetings but did not accept PPFP at one year postpartum and exploration of care seeking for sick infants.

Data collection tools were pretested and revised as necessary before use. A data management plan for the study is in place to assure data quality at all stages. The key components of the plan include the use of logic checks and skip patterns; verification of completeness of each record by POs prior to server upload; verification by POs of a random 20% of study subjects with submitted survey data that surveys were in fact administered to them; and real-time data review and monitoring by study investigators based in Kenya, Nigeria, and USA on the REDCap ^TM cloud platform. In Nigeria, where women tend to deliver at the same facility where they receive ANC, POs also cross-checked 20% of self-reported facility-based deliveries with facility records. This verification was not possible in Kenya where women commonly seek ANC and delivery care at different facilities, including the private sector.

We have no reason to believe that the intervention will cause harm. However, there might be some risk of embarrassment to women who share their personal information with group members; such information might leak outside the group. Additionally, the study team is tracking occurrence of anticipated and unanticipated events as outlined in the research plan. These include miscarriages, stillbirths, neonatal deaths, and maternal deaths.

Data analysis and the reporting of results for this study will be conducted in accordance with norms for analyzing cluster randomized trials as described in the Consolidated Standards of Reporting Trials guidelines ⁵³ .

Quantitative data analysis. Descriptive analysis of quantitative variables will be done using measures of central tendency (mean, median), measures of dispersion (range, standard deviation), and proportions (frequencies, percentages) as appropriate. Next, differences between the two study arms will be assessed using chi square test. Potential confounding factors, such as the women’s age will be assessed.

The primary outcome analysis will be done based on the intention-to-treat analysis. Participant’s data will be analyzed based on the groups to which they were randomized to compare the proportions of facility-based deliveries between the two study arms. The endpoints will be analyzed using a generalized estimating equations multivariable logistic regression model. The model will adjust the estimates for imbalances in the baseline characteristics.

Secondary analysis will be conducted to explore the uptake of services offered at ANC and during the 12-month post-delivery period, such as FP and child immunization. All statistical analysis will be performed using the R statistical software ⁵⁴ . Subgroup analysis will be conducted for specific subgroups to determine if there was a differential effect of the interventions. The amount of missing data will be assessed for each variable and overall for the sample. If more than 5% of data are missing, multiple imputation methods will be used, assuming data are missing at random ⁵⁵ .

Qualitative data analysis. Following each audio recording, either the moderator or an official transcriber familiar with local language and English will transcribe from listening to the audio recording. The transcription will be done, and personal identifiers redacted. The transcribed data will be checked for quality by study staff conversant in both English and local languages used. Transcriptions will be in English. Analysis will be ongoing to inform decisions on saturation and thus there is no need to wait until all qualitative data are collected. Coding of textual passages will be done in Atlas-ti or other similar qualitative data analysis software. Two to three qualitative data analysts will code the textual information on health topics. The emerging themes will be summarized in tables and other appropriate formats. Audio recordings will be retained until when transcriptions analysis are complete, then destroyed. The electronic versions will be stored by Jhpiego in secure servers along with the electronic quantitative data files.

The study team plans to disseminate findings among national and sub-national stakeholders through in-country dissemination events and globally through peer-reviewed journal manuscripts and international conference presentations. About 10 publications, including conference abstracts and journal manuscripts, are planned. Study findings will be presented to national and sub-national health officials to determine if and how group care will be integrated into policy and included as a strategy for service delivery. The study will contribute to the body of knowledge that will inform decision makers locally and globally on whether G-ANC is a feasible service delivery model that is more acceptable and effective than individual ANC.

Ethical approval was obtained from the Kenya Medical and Research Institute, Nairobi, Kenya; Nasarawa State MOH Research Ethics Committee, in Nigeria; and the Johns Hopkins Bloomberg School of Public Health Institutional Review Board, Baltimore, Maryland, United States of America (IRB000007078). The study is registered with the Pan African Clinical Trials Registry, PACTR201706002254227 ( www.pactr.org).

Enrollment of study participants and collection of baseline data began in October 2016, with the first group meetings occurring at the end of the month. Enrollment was completed in January 2017 in Nigeria and June 2017 in Kenya. Overall 1,075 and 1,013 pregnant women were enrolled in Nigeria and Kenya, respectively. Figure 2 and Figure 3 show the enrollment cascade for the two countries.

In Nigeria, G-ANC meetings for Phase 1 were completed in May 2017 with data collection for Phase 1 completed in September 2017. Phase 2 HMHB meetings were completed in May 2018, and data collection in July 2018.

In Kenya, Phase 1 G-ANC meetings were completed in October 2017 and data collection in January 2018. Phase 2 HMHB meetings and data collection will be completed in November 2018. Final study results will be available in February 2019.

The study was registered retrospectively on May 2, 2017, by the Pan African Clinical Trials Registry PACTR201706002254227