This manuscript describes protocol version 3.0, dated July 29, 2026. Important protocol modifications are documented through version-controlled records and communicated. The completed SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) checklist is provided as Checklist 1.

The study will be conducted in Jimma Zone, which is located in the Oromia Regional State in southwestern Ethiopia, approximately 350 km from the capital city, Addis Ababa. Administratively, the Jimma Zone comprises 21 woredas (districts) and 2 town administrations, with an estimated population of 3.5 million. Most of the population resides in rural agrarian communities.

In the study area, health extension workers (HEWs) paid by the government serve as the frontline workforce at the community level, providing maternal and newborn health education, conducting routine home visits, mobilizing mothers through the Women’s Development Army, and facilitating referrals between households, health posts, and health centers. The HEWs also run “Pregnant Mothers’ Forums,” a community-based platform in which pregnant women gather at the health post for monthly group sessions led by HEWs and midwives. These sessions provide structured health education on a range of maternal and child health topics and serve as an important channel for counseling, peer support, and early engagement with essential maternal health services. These existing Pregnant Mothers’ Forums will serve as the primary platform for delivering the study intervention.

The trial was coordinated by the principal investigator in collaboration with the research team at Jimma University, Ethiopia, and the Swiss Tropical and Public Health Institute (Swiss TPH), Switzerland, as well as the Jimma Zone Health Office and the participating woreda health offices in Jimma Zone, Ethiopia. The principal investigator was responsible for overall trial coordination, communication with participating sites, intervention implementation, regulatory compliance, and reporting. The supervisory team provided scientific and methodological oversight and reviewed trial progress. A local field-coordination team supervised participant recruitment, intervention delivery, follow-up, and adherence to the study protocol. The data-management team was responsible for database development, data-quality checks, secure data storage, and preparation of the analysis data set.

No end point adjudication committee was established because the primary and secondary outcomes were assessed using prespecified operational definitions and standardized data-collection procedures. No formal data monitoring committee was established because the intervention was educational and considered to present minimal risk. Trial conduct, participant safety, data quality, and intervention fidelity were instead monitored by the principal investigator and the supervisory team.

This study is designed as a 2-arm, assessor-blinded, open-label, parallel-group, cRCT. A total of 28 health facilities will serve as the units of randomization and will be allocated in a 1:1 ratio to the intervention and control groups (14 clusters per group).

The study was designed to detect a 20% increase in PNC use, from an expected 28% in the control arm, based on the 2019 Ethiopia Mini Demographic and Health Survey [32]. The sample size calculation assumed an intracluster correlation coefficient of 0.10 and an average of 20 mothers per selected kebele. At randomization, 28 health-facility clusters were allocated equally to the intervention and control arms (14 clusters per arm). Within these clusters, kebeles served as the sampling units. One kebele was selected from each intervention cluster and 2 kebeles from each control cluster, resulting in 14 intervention kebeles and 28 control kebeles—a 1:2 intervention-to-control sampling ratio. With 20 women recruited from each of the 42 selected kebeles, the initial required sample size was 840 women. To account for an anticipated 15% (126/840) loss to follow-up, 3 additional women were added per kebele, resulting in a final target sample size of approximately 966 women. The unequal sampling ratio was adopted to improve statistical efficiency within the available budget, given the substantially higher cost of delivering the intervention than providing usual care.

The study population consists of pregnant women residing in the selected kebeles (the lowest administrative boundaries) of Jimma Zone, Oromia Regional State, Ethiopia, and meet the study’s eligibility criteria. The eligibility criteria guiding the selection of mothers were structured at 2 levels: cluster-level and participant-level eligibility. At the cluster level, health facilities were eligible if they served catchment populations of at least 5000 people, provided more than 50 ANC visits per month, and were located in periurban areas to ensure operational feasibility and reliable follow-up. At the participant level, eligibility was restricted to pregnant women between 20 and 24 weeks of gestation who resided within 1 hour of the selected health facility, thereby enabling timely and optimum exposure to the intervention and adequate follow-up until the postnatal period. Women were excluded if they planned to relocate outside the study area before completion of follow-up.

Health facilities (health centers) will constitute the unit of randomization. Facilities will be randomized to the intervention or control arm using computer-generated random numbers, stratified by woreda, to ensure balance across administrative areas. The randomization sequence will be generated by the first author, who will not be involved in participant recruitment or outcome assessment. Allocation will be concealed until kebele selection and participant enrollment are completed. Outcomes will be measured at the individual level among mothers residing within the selected kebele catchment areas of the randomized facilities.

A multistage sampling approach will be used to select woredas, health facilities, kebeles, and study participants (Figure 2). First, in collaboration with the Zonal Health Department, 14 woredas were purposively selected based on safety, security, and logistical feasibility. All primary health care facilities within the selected woredas were subsequently screened using predefined criteria (catchment population >5000 and reliable accessibility). From the eligible facilities, 28 health facilities (2 per woreda) were selected using probability proportional to size and randomized, with the number of pregnant women in each facility’s catchment serving as the size measure.

Subsequently, the 28 health facilities were grouped into 14 matched pairs according to woreda, with each pair comprising the 2 selected health facilities within the same woreda. Within each matched pair, 1 health facility was randomly allocated to the intervention arm and the other to the control arm. This matched-pair cluster randomization resulted in 14 health-facility clusters in each study arm. For each selected health facility, all kebeles with at least 23 eligible pregnant women were identified. One kebele was randomly selected from each intervention health-facility cluster, whereas 2 kebeles were randomly selected from each control health-facility cluster. Thus, 42 kebeles (14 in the intervention arm and 28 in the control arm) served as sampling units nested within the 28 randomized health-facility clusters. Each selected kebele inherited the treatment allocation of its corresponding health facility and was not randomized independently.

Finally, within each selected kebele, a sampling list of eligible pregnant women was compiled with support from HEWs and verified by a health extension supervisor to minimize coverage bias. From this list, a fixed number of 23 women per kebele were randomly selected using simple random sampling, producing a final target sample of 966 women across the 42 kebeles.

Recruitment and enrollment will be carried out by trained data collectors who will verify eligibility, provide study information, and obtain informed consent. From the list of eligible women, 23 will be selected using simple random sampling, after receiving a detailed explanation of the study purpose, procedures, potential risks and benefits, voluntary participation, and confidentiality safeguards in their preferred local language. If a selected woman declines participation, a replacement will be randomly drawn from the remaining eligible list. Women who agree to participate will provide written or thumbprint-informed consent and will be enrolled immediately.

Several strategies will be implemented to minimize potential sources of bias. Randomization at the health-facility level will help balance both measured and unmeasured characteristics across groups, thereby reducing selection bias. Eligibility screening will be conducted using a standardized tool to ensure consistent application of inclusion and exclusion criteria. Data collectors will receive comprehensive training on interviewing techniques, nonleading questioning, and the use of standardized data collection instruments to minimize interviewer bias. To reduce information and recall bias, data will be collected prospectively at predetermined time points and linked with facility records whenever feasible. Data collectors will remain blinded to intervention allocation to the extent possible, particularly during outcome assessment. Additionally, recruitment and follow-up procedures will be closely supervised to ensure adherence to study protocols, enhance fidelity, and minimize variability across sites.

Multiple approaches will be used to minimize loss to follow-up throughout the study period. Eligible women will be enrolled only if they reside within 1 hour of the study facility and do not anticipate moving outside the study area before completing the follow-up period. At enrollment, detailed locator information, including phone numbers, alternative contacts, and physical directions, will be recorded to facilitate timely tracking. Data collectors and HEWs will maintain scheduled home visits and reminder calls to ensure continuity of follow-up through pregnancy and the postnatal period. Engagement with community structures, including mothers’ forums and the Women’s Development Army, will further support retention by reinforcing communication and community-level follow-up. The study team will monitor follow-up completion weekly, and any missed visits will trigger immediate tracing efforts.

The conceptual framework presents the hypothesized pathways through which the intervention is expected to influence PNC use. Specifically, the intervention is expected to strengthen attitudes toward PNC, subjective norms, perceived behavioral control, self-efficacy, outcome expectations, goal setting, social support, and behavioral intentions, ultimately promoting PNC use (Figure 1).

The total planned duration of the study is 9 months, with recruitment begun in November 2025 and will continue for approximately 3 months. Data will be collected at 3 time points: baseline (20‐24 weeks of gestation), immediately after the intervention, and at 6 weeks postpartum, and the intervention will be implemented over a 4-month period. Each participant will be followed longitudinally from recruitment at 20 to 24 weeks of gestation until 6 weeks postpartum. Participants in the intervention arm will be invited to attend all intervention sessions.

Participants in the control clusters will continue to receive routine antenatal and maternal health education as part of the existing standard of care, delivered by HEWs and midwives during routine ANC and Pregnant Mothers’ Forums sessions. Routine education consists of unstructured counseling based on the national maternal and newborn health guidelines, covering general pregnancy, childbirth, and PNC messages, and is delivered according to usual practice without the structured flipbook, behavioral strategies, or theory-informed components of the intervention. To minimize contamination, intervention and control clusters are geographically separated, providers are trained only in the procedures assigned to their respective clusters, and intervention materials are not distributed outside intervention sites.

Fidelity monitoring will be implemented to ensure that the intervention is delivered as intended and to minimize variations in implementation across clusters. In the intervention kebeles, fidelity will be assessed using standardized checklists completed by supervisors, periodic observations of educational sessions, and brief participant feedback forms. These tools will document adherence to session content, delivery quality, participant engagement, and use of culturally adapted materials.

In the control kebeles, fidelity monitoring will focus on ensuring that routine care continues without exposure to intervention components. Supervisors will conduct brief periodic checks to verify that no intervention materials or messages are being delivered and that standard care practices remain unchanged. A simple checklist will be used to document compliance with routine care protocols.

The study team will implement procedures to identify and appropriately manage any maternal or newborn complications, including stillbirth, occurring during the study period. Although the intervention is educational and is not expected to increase clinical risks, participants may experience pregnancy-related complications as part of routine care. Any participant reporting or identified with complications will be referred promptly to the appropriate health facility for clinical assessment and management according to standard care protocols. The study team will document such events and report serious adverse events in accordance with ethical and regulatory requirements.

In the event of a stillbirth, the mother will receive appropriate support and counseling through the existing health system. Health workers involved in the study will provide compassionate communication, psychological first aid, and emotional support, and mothers requiring additional psychosocial support will be referred to available counseling services according to local practice and facility capacity. Participation in the study will not interfere with access to routine maternal and psychosocial care.

Electronic data capture and computerized processing were used to improve data quality, reduce entry errors, and facilitate real-time monitoring of study indicators [35]. Trained data collectors, independent of intervention delivery, will administer all questionnaires using ODK Central. Baseline interviews will capture sociodemographic characteristics, obstetric history, newborn care knowledge, intention to seek PNC, behavioral determinants derived from TPB and SCT, and mobile phone access or prior exposure to PNC messages. During the intervention period, HEW and midwives will document session attendance, fidelity to session content, and participant engagement, with supervisors conducting periodic observations to ensure adherence to the protocol. Follow-up assessments at 6 weeks postpartum will measure PNC use, newborn care practices, newborn care knowledge, behavioral determinants, knowledge of maternal danger signs, and participant satisfaction.

The primary outcome of PNC use, attendance at three or more PNC visits, will be measured using a structured maternal interview administered at the mother’s home at 6 weeks postpartum. Mothers will be asked whether they received a PNC check within 48 hours of birth and the total number of PNC contacts within 6 weeks. To minimize reporting bias, these responses will be verified against facility registers and maternal health cards issued during PNC visits.

The secondary outcome of newborn care knowledge will be assessed using a 22-item structured questionnaire (Multimedia Appendix 1) derived from the WHO PNC guidelines and previously published literature using a modified questionnaire developed by McConnell et al [36]. Items cover thermal care, breastfeeding, cord care, hygiene, and recognition of maternal and newborn danger signs. Responses are scored as correct/incorrect (Annex I); total scores will be summed and categorized into “good” vs “poor” knowledge based on predefined cutoffs.

This research project will implement a comprehensive data management plan. We will use digital data-capture and processing workflows to enable secure, automated handling and analysis of complex biophysical and survey data (eg, ODK with cloud backups and audit trails) [35]. All raw and processed data will be systematically collected, documented, and stored in secure, access-controlled repositories, in compliance with Swiss TPH guidelines. Metadata will be standardized following the FAIR (Findable, Accessible, Interoperable, and Reusable) data principles to facilitate future reuse and reproducibility. To prevent data loss due to hardware failure or human error, regular backups will be performed using both local and cloud-based storage solutions.

An audit trail will be maintained throughout the project to document all modifications made to the data, clearly recording when, how, and by whom changes were made, thereby enhancing transparency and reproducibility. A comprehensive data dictionary will accompany all data sets, providing clear definitions of variables, coding schemes, and units of measurement to ensure consistency across collaborators and future users. Data cleaning procedures will be conducted in Stata (version 18; StataCorp LLC) and fully documented using well-annotated do-files. Additionally, a logbook will be maintained in both digital and physical formats, as appropriate, to capture day-to-day research activities, data issues encountered, decisions made, and updates to protocols, further supporting the audit trail and ensuring full traceability of the research process.

Missing data will first be examined to determine their extent and patterns. If the proportion of missing data is minimal, complete case analyses will be performed. If missingness is substantial and considered compatible with a missing-at-random mechanism, multiple imputation will be considered as a sensitivity analysis.

All analyses will be conducted according to the intention-to-treat principle. Descriptive statistics will first be used to summarize participants’ baseline characteristics by study arm. Continuous variables will be presented as means and SDs or medians and IQR, while categorical variables will be summarized using frequencies and percentages. Baseline comparability between intervention and control groups will be assessed descriptively using standardized differences and appropriate statistical comparisons. Composite scores for newborn care knowledge and theoretical constructs, including attitudes, subjective norms, perceived behavioral control, behavioral intention, and self-efficacy, will be generated from multi-item measures. Internal consistency will be assessed using Cronbach α, and construct scores will be standardized (z scores) to improve comparability across domains.

The primary outcome, PNC use (coded as yes/no), will be analyzed using logistic regression models, while secondary outcomes, including maternal and newborn care knowledge scores, will be analyzed using logistic and linear regression models, respectively, according to the outcome distribution. Both unadjusted and adjusted models will be estimated. Adjusted models will include prespecified baseline covariates considered clinically or epidemiologically relevant or associated with the outcomes. Given the cluster-randomized design, all primary analyses will account for within-cluster correlation by using robust standard errors clustered at the health facility level. Intervention effects will be reported as odds ratios or regression coefficients with corresponding 95% CIs and 2-sided P values.

Mediation analysis will be conducted to assess whether changes in theory-based behavioral constructs mediate the effect of the intervention on PNC use and newborn care practices. Structural equation modeling or generalized structural equation modeling, depending on the distribution of the outcome variables, will be used to estimate the direct effect of the intervention on outcomes and the indirect effects operating through intermediate behavioral determinants. Indirect effects will be estimated using bootstrapped 95% CIs, and total, direct, and mediated effects will be reported.

Sensitivity analyses will be conducted to assess the robustness of intervention effect estimates under alternative modeling assumptions, including mixed-effects models, accounting for facility-level clustering, covariate-adjusted models using prespecified baseline characteristics, and alternative approaches to handling missing data, where appropriate.

Ethical approval was obtained from the Institutional Review Board of Jimma University, Ethiopia (approval number JUIH/IRB/0553/26), prior to study initiation. Additional ethical approval was granted by the Ethics Committee Northwest and Central Switzerland (EKNZ), Switzerland (approval number AO_ 2025‐00079) on September 28, 2026. The study protocol was further reviewed through the Swiss TPH Human Research Protection (HPR) process. The trial was prospectively registered at ClinicalTrials.gov (identifier: NCT07095296) before participant recruitment.

Written informed consent was obtained from all participants before enrollment. For participants younger than 18 years, written assent was obtained in addition to written consent from a parent or legal guardian, in accordance with Ethiopian ethical and legal requirements for research involving minors. Before consent was obtained, trained data collectors explained the study objectives, procedures, duration, potential risks and benefits, and participants’ rights, including the voluntary nature of participation and the right to withdraw at any time without affecting access to routine health services.

All study procedures were conducted in accordance with the ethical principles of the Declaration of Helsinki and applicable national and institutional regulations. Participant privacy and confidentiality were strictly maintained by assigning unique identification codes, removing personal identifiers from study data sets, storing study records securely with restricted access, and reporting findings only in aggregate form. No identifiable personal information or images will be disclosed.

The study was funded by the Eckenstein-Geigy Professorship in June 2025. Participant recruitment was conducted from November 2025 to January 2026 across 28 health facilities in Jimma Zone, Ethiopia. All 28 health-facility clusters were randomized, with 14 allocated to the intervention arm and 14 to the control arm. A total of 945 pregnant women were enrolled.

The intervention was delivered from December 2025 to May 2026, and participant follow-up was conducted from December 2025 to July 2026. Follow-up data collection was completed in July 2026. Data cleaning and statistical analysis began in July 2026. The primary and secondary analyses are scheduled for completion in fall 2026, and the trial findings are expected to be submitted for publication in fall 2026.

The mean age of participants was 26.84 (SD 5.23) years, with the majority residing in rural areas (n=834, 88.3%) and being married (n=937, 99.2%). Baseline characteristics were generally similar between trial arms regarding age, residence, marital status, and obstetric history, although notable differences were observed in maternal occupation, educational attainment, and household income.

The intervention was delivered according to the prespecified protocol. Minor operational adaptations were made to participant scheduling and field implementation; however, these adaptations did not change the intervention content, cluster allocation, outcome definitions, or prespecified analytical procedures. No major protocol deviations affecting the conduct or integrity of the trial were identified.

Baseline sociodemographic and obstetric characteristics were summarized descriptively for the overall study population and by trial arm (Table 2).

The key milestones and implementation timeline of the study, from funding and preparatory activities through recruitment, intervention delivery, follow-up, data analysis, and dissemination, are presented in (Table 3).