The mixed-methods study will use a stepped wedge cluster randomized controlled trial with randomization of the order of the start of the intervention done at the health center level (clusters) in the Amhara region. The study will begin with a pilot to test the intervention conducted by HEWs at the health post level. If feasible, randomized controlled trial will be conducted with three time periods: baseline period (T1); time period 2 (T2); and time period 3 (T3), each about 6 months in duration. Control period for the first half of clusters will occur at T1. Half of the clusters will start G-ANC first (at T2), and the control period for the second half of the clusters will also occur at T2. The second half of the clusters will start the intervention in the third time period (T3). The first half of clusters will have a choice to continue G-ANC with a second cohort of women during T3 ( Figure 1). We plan to use a stepped wedge study design, because it allows for iterative learning (i.e., adaptive management) during the study to modify the intervention, which was novel compared to traditional cluster RCT. The study was registered on September 23, 2021 on clinicaltrials.gov with registration ID: NCT05054491.

Prior to the intervention, all women in the study HPs will receive the current standard of individual ANC. At the time of the intervention, the health posts in the selected clusters will provide G-ANC to one cohort of 8–10 pregnant women. G-ANC participants will be followed prospectively from the time of study enrollment. All non-enrolled pregnant women in the intervention health posts will continue to receive individual ANC.

The G-ANC study will be conducted in two zones of the Amhara regional state: West Gojjam Zone (South and North Achefer districts) and the South Gondar Zone (Farata, Dera, and Libokemem districts). Twelve health centers in these districts will be selected as G-ANC study centers. G-ANC meetings will take place at the health post level only. Amhara region has an estimated population of 22.5 million ( CSA/Ethiopia & ICF, 2017). Amhara region has 12 administrative zones, and the economy is primarily subsistence farming.

Twelve HCs will be selected in the Amhara region after meeting the eligibility criteria ( Table 1) and categorized as follows: 1) type of location (peri-urban or rural); and 2) client volume for the last one-year (low or high) volume. Three of the five health posts per HC (total 36 health posts) will be selected to participate based on inclusion criteria. Study participants will be: pregnant women attending ANC; HEWs; and facility and regional managers working in study sites in Amhara region.

During the intervention time period, pregnant women who come for their first ANC visit at the health post will receive standard individual care. At the end of this first ANC visit, eligible women who agree to participate in G-ANC will be added to the G-ANC cohort. Approximately 8-10 women will be enrolled in a cohort. Up to two additional pregnant women who meet study criteria can be added to a G-ANC cohort if added to the cohort by the second G-ANC meeting. Six monthly G-ANC meeting will be conducted for each cohort on a fixed day/time at each Health Post(HP). Cohort membership will be fixed (i.e., the same women and facilitators at each meeting). Each cohort will meet for G-ANC meetings in a designated meeting space, set up to ensure privacy during clinical consultations. G-ANC meeting will be conducted by at least one HEW, who is trained in the G-ANC service delivery model.

Each meeting will have the same general structure: self-assessment by the clients themselves and initially taught and supervised by the HEW (such as measurement of weight and blood pressure using an automated cuff); discussion around pre-determined themes using materials designed for G-ANC; individual assessments by the HEW in a private space; and peer support activities that create group cohesion. Table 2 presents the G-ANC meeting schedule and content per meeting. Each meeting is expected to last 90–120 minutes and will be documented as an ANC contact. Social distancing in response to COVID-19 as per national guidelines will be enforced during meetings. Women will have the opportunity for additional private time with the HEW after the G-ANC meeting if desired or needed for health and/or safety. Women will be told that they can and should return at any time if they have questions or concerns. Phone/text reminders will be used to remind participants of G-ANC meetings as a strategy to improve ANC attendance.

The G-ANC intervention will be implemented at the health posts that offer integrated point-of-care testing (POCT) to screen for anemia, syphilis, and asymptomatic bacteriuria (ASB). Availability of these rapid test kits improve the quality of ANC and reduce the need for women to travel to the health center for routine screening. Point-of-care testing was introduced and piloted in these facilities, but it is not yet the national standard of care for ANC. The test kits were available in all HPs at the start of the study and will be provided by the study until completion.

The G-ANC intervention and materials were adapted from Jhpiego’s G-ANC intervention model, which consisted of five monthly two-hour meetings with clinical assessments alongside structured group discussions and activities and facilitated by nurses, midwives, or community health extension workers ( Grenier et al., 2019). It was modified in collaboration with the MOH for the Ethiopian context as a six-meeting model. The model is designed to increase early entry into ANC and increase retention in ANC, in support of changes to the Ethiopian national ANC guidelines that increased from 4 to 8 ANC contacts ( Ministry of Health Ethiopia, 2022). Intervention materials in Amharic language include: a pictorial self-assessment card women used at the beginning of each meeting to record the weight, blood pressure and indicate any danger signs; meeting guide used by the facilitators to guide each meeting; 5-6 A4-size illustrated picture cards per meeting that are designed for nonliterate audiences (total 36 cards), with questions on the back of each card to foster discussion; and a A4 booklet with small versions of all the picture cards that women can take home to share what they’ve learned with their spouses and families; an implementation guide for HEWs to introduce G-ANC in their health post; a simple G-ANC register is part of the intervention package to aid the facilitator to track attendance and record other data; and three monitoring tools, including a meeting observation checklist to provide feedback and assess fidelity. The picture cards were pre-tested with HEWs before the pilot phase, and all the intervention materials will be revised as needed based on the pilot feedback. Each intervention HP will be provided these materials, 12 chairs, a basic set of supplies to run activities, two weigh scales, three user-friendly Microlife PSA blood pressure devices for the self-assessments, and one privacy screen to divide the private clinical consultation area from the larger G-ANC meeting space.

All clients will continue to receive the components of standard care, and no standard therapies will be denied to the intervention group. Likewise, standard, individual ANC care will not be denied if requested at any time.

Training: At the start of each time period, two HEWs in each intervention HP will be trained on the G-ANC intervention (five days) in a highly participatory, facilitated learning environment by study staff. The HEWs will also be trained (1.5 days) on estimating gestational age including hands-on practice to improve identification of eligible women. Midwives from HCs who will provide onsite mentorship to HEWs during G-ANC implementation will be trained for four days on G-ANC focused mentorship by the study team. Key stakeholders, including participating health facility managers and regional managers, will be oriented to G-ANC to build ownership and support.

Primary outcome 1: Proportion of women with at least four ANC visits.

Primary outcome 2: Proportion of women who delivered at health facility.

After exploring the balance by time period/intervention status and correlations, the effect of G-ANC will be estimated using generalized linear mixed effects models with binomial distribution and logit-link.

We will adjust for facility- and individual-level characteristics that show imbalance between the intervention vs. control time periods and can be strongly correlated with the outcome scores. The primary analysis will not adjust for multiplicity and will be performed at 0.05 level of statistical significance.

Secondary outcomes: Satisfaction level of women with ANC service, proportion of women who reported self-efficacy or empowerment.

Analysis of secondary outcome measures will follow a similar plan as the analysis of the primary outcome. Subgroup analysis will be conducted for specific subgroups to determine if the interventions were different. The key subgroups of interest are marital status, education, and economic status.

ANC clients will be recruited via four strategies: 1) pregnant women will be identified as potential study participants through existing local community structures (such as the women’s development army leaders) as a means to explain the study to potential participants and recruit them into the study; 2) All new ANC attendees in the intervention HPs will be screened for eligibility during the first ANC contact with a ANC service provider; 3) All new ANC attendees in the study HC will be screened for eligibility during the first ANC contact with a ANC service provider; 4) eligible women will be identified by G-ANC trained HEWs who make door-to-door visits in the community.

A research assistant will rotate among each of the intervention HPs during the recruitment period to explain the study to women who meet the inclusion criteria and agree to participate in the study.

The research assistants will obtain verbal consent from women in a private space in the clinic using IRB-approved recruitment and consent documents in the local language. Verbal consent will be sought instead of written consent as written consent is seen as a personal identified in the community and verbal consent is preferred. This consenting method was approved by the ethics committee (see section Ethics). The potential G-ANC participants will be informed during the consenting process of the expected number of meetings (i.e., ANC visits), their rights as participants, risks, and benefits. The research assistant will explain the need to collect their names, phone numbers, and addresses, which will be used for study-related follow-up purposes only. Anyone declining participation in the study will be referred to individual ANC care during the intervention period. A baseline questionnaire will be administered to consenting study participants at the beginning of each step (time period).

For HEWs and health facility managers, study staff will explain the purpose based on the inclusion criteria. The HEWs will be oriented to the study including the randomization (they will not be blinded to the order of randomization of their health center and affiliated health posts) and their role in the study including screening ANC first time clients for eligibility to participate. HEWs who will be facilitating G-ANC meetings will be informed that they may be asked to participate in an in-depth interview (IDI) at the end of the study related to implementation of group-based care and its effect on their job and satisfaction levels. Oral consent will be obtained in a private location at the health facility by research assistants.

For the regional health managers, study staff will contact the Regional Health Bureau responsible for overseeing reproductive health programs in Amhara region via phone or in-person and invite the regional health manager to participate in an IDI as per the inclusion criteria. An appointment will be scheduled to obtain informed consent and conduct the IDI, preferably at the Regional Health Bureau’s office. Written consent will be obtained by the research assistant, and the interview conducted in a private location.

For the in-depth interviews of women and HEWs involved in G-ANC, qualitative data will be collected at the end of the study after all G-ANC meetings have been conducted. Oral consent will be obtained from the HEWs and eligible women 3–6 months postpartum to be enrolled for the qualitative portion of the study.

The intervention in this study is for women to take part in group ANC. Some risks, discomforts, and inconveniences may apply and will be minimized as outlined in Table 3.

We used the existing data from Amhara region for two primary study outcomes (i.e., four or more ANC contacts (31.5%) and facility-based delivery (27.1%) to calculate the number of clusters (i.e., HCs) to be randomized to be able to detect an odds ratio of 2 comparing post-intervention to pre-intervention proportion with 80% statistical power using stepped wedge design with three time periods and two steps for switching from standard care to intervention ( Figure 1). For the complete stepped-wedge design, the calculations indicate that a sample of 12 HCs with three time periods/two steps, six HCs switching from control to treatment at each step, and with an average of 30 women per cluster (i.e., three HPs with 10 women in each) per time period achieves 84% power to detect an odds ratio of 2 at 0.05 level of statistical significance. The underlying pre-intervention proportions and the post-intervention proportions associated with odds ratio of 2 are shown in Table 4.

The within-HC correlation (ICC) is assumed to be 0.100. This level of correlation is consistent with the one reported by Ousman et al., where clustering was done at enumeration area level with an ICC of 0.11 for ANC visits ( Ousman et al., 2019). The total sample size for ANC clients is then 12 HCs x 3 HPs x 10 women/HP x two time periods=720 women. An odds ratio of 2.0 translates to 50–60% improvement in completion of four or more ANC contacts and 60–70% improvement in facility-based delivery in the G-ANC arm compared to pre-intervention.

The study will be conducted in 12 health centers. The overall sample size is a total of 770 women. The pilot will enroll 50 women (i.e., 10 women per HP, 5 HPs), and the RCT will include a maximum of 720 women ( Table 5). Sample size for the qualitative data collection is indicated in Table 6.

Health centers with catchment health posts that provide POCT for ANC services will be listed and randomized to begin the intervention at T2 (i.e., the first randomized set of facilities) or T3 (i.e., the second randomized set of facilities). These facilities will be randomized 1:1 before T1 to two different intervention start times (i.e. two sequences), stratified by location and client volume, using randomly permuted blocks of size 2 and 4. We will use centralized randomization as the allocation concealment mechanism: A remote and independent statistician in the Johns Hopkins Bloomberg School of Public Health (JHSPH) will generate and implement the allocation sequence, and communicate it to the study sites after confirming the eligibility of the clusters. HEWs will identify pregnant women and provide the list of women to ACIPH. ACIPH field workers will get consent and enroll eligible women.

Blinding: Study staff and those assessing outcomes will not be blinded to knowing the study time period to which facility is randomized. The participants will also not be blinded about the type of care they will be receiving (group ANC or routine/conventional care).

G-ANC at the health post level is a new intervention, therefore the study will be rolled out using a phased approach.

Pilot phase: Prior to study activation, we will pilot G-ANC in up to five purposively selected health posts in Amhara region that will not be included in the study. This pilot phase will allow us to adapt G-ANC materials to adjust for the literacy levels of the HEW and conditions at the health post. After the materials have been adapted, up to two HEWs per health post (total 10 HEWs) will be trained to facilitate G-ANC. In the pilot phase, we will enroll a maximum of 50 women (up to 10 per HP x up to 5 HPs). The pilot phase will also allow us to determine whether: HEWs are able to facilitate the G-ANC meetings; there are enough eligible ANC clients at the health post level; we are able to enroll enough women at the HP who are of a similar gestational age; and women come back to the HP for the G-ANC meetings. After completing at least two G-ANC meetings in the pilot, we will determine the feasibility of moving to Phase 2 of the study.

Trial phase: Phase 2 includes T1, T2, and T3. During T1, pre-intervention data will be collected from the first set of 18 HPs by interviewing pregnant women during their pregnancy and again postpartum. The women in T1 will not receive the intervention. During T2, the G-ANC intervention will be implemented in the first set of 18 HPs (first 6 clusters), and pre-intervention data collection will be conducted in the second set of 18 HPs (i.e., second 6 clusters). During T3, the G-ANC intervention will be implemented in the second set of 18 HPs. If the first set of 18 HPs chooses to continue the G-ANC intervention with a second cohort, an additional six months of data collection will occur during T3. Participant timeline is depicted on Figure 2.

Once study participants are enrolled, the participant timeline is about 8.5 months: enrollment with a baseline questionnaire; follow-up for about six months during ANC; and end-line survey up to 6 weeks post-delivery. Participants in the qualitative research will have a single contact for the IDIs.

The overall duration of the entire study is 18 months. The implementation process will be documented in these sites, and a transition plan will also be developed for Jhpiego to handover implementation to the government.

Implementation of a paper-based participant tracking system will be used to accurately assessing participant retention in the study. The G-ANC registration and participating registration book will be used to track ANC attendance. This will provide the HEWs with a listing of all participants who have missed scheduled G-ANC meetings.

Both quantitative and qualitative data collection approaches will be used in this study. The following data will be collected in the pilot phase and then during study implementation (Phase 2).

Once enrollment begins, all new ANC attendees in the selected health posts will be screened for eligibility during the first ANC contact with a health care provider. Screening and Enrollment Questionnaire will be completed. Basic demographic data will be collected from all eligible consented women (all who meet the inclusion criteria). All data will be de-identified for those who decline study participation. Baseline and endline data will be collected from all eligible and voluntary women participating in all study time periods. During the first period, data will be collected from 18 control health posts. In the second period (T2), data will be collected from 18 intervention and 18 control health posts. In the third period, (T3) data will be collected from 18 intervention health posts.

Data quality will be monitored through the use of trained research assistants and supervisors, data quality supervision from, and use of an electronic data collection program to minimize data entry errors. The study team will select respondents for the IDIs at the end of the study based on preliminary findings from the quantitative survey. IDIs with women will be conducted at the end of T2 and T3, after completion of all of the G-ANC meetings ( Table 7). Specific topics/domains to explore with G-ANC participants include: satisfaction with care; any perceived changes in self-efficacy or empowerment; health literacy; improved communication and/or trust with providers and the health system as a whole; factors in uptake/non-uptake of postpartum family planning (PPFP); and the overall impression of the G-ANC model.

IDIs will be conducted with HEWs who were trained in G-ANC and facilitated G-ANC meetings in the intervention sites and health managers. Specific topics/domains to explore with providers will include: attitudes towards providing group care; perceived changes in their communication and relationship with patients (if any); changes in empowerment, self-efficacy, and satisfaction related to their perceived ability to do their job well; changes in workloads; sustainability of group care; suggested changes to the model /logistics required to offer group care; perceived effects of group care on colleagues and patients. IDIs with health facility managers and regional health managers will be conducted at the end of T2 and T3 in the intervention areas.

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 with considerations for the step-wedge extension ( Hemming et al., 2018; Hemming et al., 2019).

Quantitative data analysis. The primary analysis will be conducted using intention-to-treat (ITT) analysis. After exploring balance by time period/intervention status and correlations, the effect of G-ANC will be estimated using generalized linear mixed effects models with binomial distribution and logit-link. The model will include G-ANC as the intervention indicator, time period and their interaction as the primary predictors, and HC as the random intercept. We will adjust for facility- and individual-level characteristics that show imbalance between the intervention vs. control time periods and can be strongly correlated with the outcome scores. The amount of missing data will be assessed for each variable and overall, for the sample. If more than 5% of data on covariates are missing, multiple imputation procedures will be used assuming data are missing at random ( Sterne et al., 2009).

Qualitative data analysis. Following each audio recording, either the moderator or an official transcriber familiar with local language and English will transcribe from the audio recording. Analysis will be ongoing. Coding of textual passages will be done using open code 4.03 Umeå: Umeå University , an open-source tool known for qualitative analysis.

Following the computer system validation, the data manager will proceed to build, test, validate the master plan, and document testing of the study database using the SurveyCTO platform. SurveyCTO will automatically capture quantitative data and synchronize with the server. SurveyCTO will be programmed so that automated quality checks and data classification systems will be run. Data will be maintained in a way that preserves their accuracy, integrity, and legibility with user access restrictions and data retrievable by designated personnel only. A security system that prevents unauthorized access to the data will be maintained. The database management system will be password protected, with each member of the research team responsible for data management having their own password. After data collection, Addis Continental and Johns Hopkins Bloomberg School of Public Health (JHSPH) will de-identify the data. Quality control reports will be sent to the study team bi-weekly indicating missing, overdue data and outstanding queries during the baseline and end-line survey time periods. Throughout the fieldwork, field supervisors and the research team will observe the research assistants conducting interviews and carry out field editing. By checking the research assistants’ work regularly, the field supervisor can ensure that the quality of the data collection remains high throughout the data collection period. The de-identified dataset will be made available on ClinEpiDB or another open access site.

Research assistants will be trained in the study in a multi-day training workshop, covering the important points of the approved study protocol, including privacy and confidentiality. We will review and discuss all relevant content in the JHSPH Ethics Field Training Guide. We will review data collection procedures, which will ensure ethical conduct of research and data integrity. Ethical approval was obtained from the Addis Continental Institute of Public Health institutional ethical review board, Addis Ababa, Ethiopia (ACIPH/IRB/006/2021); and the JHSPH Institutional Review Board, Baltimore, Maryland, United States of America (IRB14448). ACIPH and JHSPH IRB amendments will be obtained for important protocol modifications, if any.