Using Community Health Nurses and Workers for Home Visits to Pregnant Women to Improve Maternal and New Born Health in Ghana

Methods: Assignment of interventions (for controlled trials) Implementation of intervention, Randomization process Randomization of the 20 electoral areas will be done by a statistician. There will be block randomization of clusters depending on the availability of either a Community Health Nurse midwife or a staff midwife within the Electoral area. The block randomization will ensure adequate distribution of intervention and control arms across all the categories of healthcare services available.

Participants will be assigned to intervention or control once informed consent is given, based on their electoral area. All participants in a cluster will be recruited at a health facility located within the electoral area. Participants will not know which group they belong at the time of consent to reduce selection bias. Following random allocation and assignment to either intervention or the control arm, the groups will continue their usual antenatal care of at least four ANC visits as part of the components of Focused Antenatal Care (FANC) services at facility level. Participants will also have access to all on-going health interventions available for all pregnant women. The interventional arm will in addition to their usual care, receive up to three home visits by the CHW/CHN where specific health information will be provided to the woman.

The demographics of women in the two arms will be compared to determine if any differential take-up resulted in unbalanced groups. There may be other consequences of differential take-up that are not evidenced by a comparison of participant demographics; this will be a limitation of the research but cannot be further mitigated.

Sample size In calculating the sample size, the weighted average of the baseline SAB rate of 37% from 2017 actual data for four health sub-districts and the intra-cluster correlation of 0.033, reported from three studies by Pagel et al in 2011 were used.

Hence for each of the two arms, there will be 10 electoral areas (clusters) in which 46-50 women will be recruited for a total sample size of 456 per arm. The total sample size of 912 mothers means the trial would have 80% power to detect a relative increase of 35% in the rate of SAB with an alpha of 0.05.

The study aims to recruit 46-50 women per cluster. Should no women drop-out of the study, the sample size required to detect the size of effect sought with enough statistical power would be 912 (46-50 per cluster).

Blinding The data collectors will be blinded; they would not know which women were in the intervention or control arms. All data on the primary outcome will be verified by the researcher, who cannot be blinded. Also, it is impossible to blind the pregnant women and the CHW/CHNs.

Involvement of District staff The CHWs and CHNs will conduct the visits; one visit in the second trimester and two in the third trimester- a total of three home visits per pregnant woman. The district officers will assist in supervising the CHW and CHN visits. Other staff who will be blinded, will be trained to collect outcomes data.

Methods: Data collection, management, and analysis Data collection methods Initial data on participants' background characteristics will be documented at the health facility where recruitment occurs. This will be done by the researcher using a pre-tested data collection questionnaire (see Appendix 2) and assisted by midwives. Outcome data will be collected from the pregnant women at the end of the trial by blinded district officers for reliability and validity. This will be done using a questionnaire administered in the woman's home. The blinded data collection officers will collect outcomes data only. They will not know which group the women belong to but the researcher (who will only collect data to verify the primary outcome and visit acceptability) will.

The data collectors will be trained to respond appropriately in circumstances where mother experienced a distressful event during pregnancy and /or delivery (offering empathy and adapting the data collection process intended to avoid any further distress) and contact the researcher immediately. The researcher will then ensure the social welfare officer is linked with the affected woman and her family and the needed support provided.

Data Quality All deliveries will be verified by the researcher at the facility or place where the birth occurred. Also, antenatal visits will be verified from the Maternal health record book.

Data entries, coding, storage Data will be collected on paper by blinded data collection officers. The hard copies will be submitted to the researcher at the end of each day's field work and stored in locked file cabinets at the research unit of the district health directorate. Entries will be made in MS Excel by the researcher. The Excel database (which will be anonymous) will be updated to the Warwick email regularly and Supervisors copied. The hard copies or anonymous soft copies will be made available to the DMEC for data verification. Check codes using Excel formulae will be built into the database to help prevent data entry errors and data loss. The entered datasets will be validated by an independent statistician.

Statistical analyses Statistical tables will be generated, and appropriate statistical tests used to analyze the variables of interest. Data analyses will be done using STATA version 14.0. Dr Nicholas Parsons and Dr Peter Kimani have provided professional statistics support and, have agreed to provide further support with the Randomization process through the school's Statistics Clinic.

The main measure of association to be determined is the Odds Ratios of SAB (OR) in intervention arm compared to control. The estimated effect of the intervention will be presented as OR together with its 95% confidence interval. Same will be computed for the controls. The ORs will be adjusted for clustering . The analysis for each outcome will be by intention-to-treat and will account for clustering by Electoral area.

Description of subject involved The participants involved are pregnant women. Only women who reside in Ellembelle will be enrolled in the study. As such women should have lived continuously in the Electoral area for at least 3 months prior to enrolment into this study. Enrolment will proceed for eligible participants only after informed consent has been given.

Trial monitoring The Trial Steering Committee (TSC) The TSC has seven members comprising, the investigator, the research supervisors who are experienced researchers at the University of Warwick, and four independent policy makers and research experts from the Ghana Health Service and Ghana's Ministry of Health. They will provide technical guidance on the conduct of the trial and support timely and efficient dissemination of findings.

The members of the TSC are

1. Dr Marion Okoh-Owusu (Researcher) Contact details: +233206301994; Email: m.okoh-owusu@warwick.ac.uk
2. Dr Celia Brown (Research Supervisor- Warwick) Contact: celiataylor@warwick.ac.uk
3. Prof Paramjit Gill (Research Supervisor- Warwick)
4. Dr Abraham Hogson M.D, PHD (Research Division- Ghana)
5. Dr Alberta Biritwum-Nyarko MD, MPH, MGCP (Ghana Health Service)
6. Dr Emmanuel Odame MD, MPH, MGCP, (Ministry of Health, Ghana)
7. The Senior Biostatistician, Ghana Health Service

Data Monitoring and Ethics Committee (DMEC) The trial data will be monitored by the DMEC, which comprises three independent members from Ghana and supported by project supervisors.

DMEC members are:

1. Mr Isaac Korankye- Snr. Biostatistician (Ghana Health Service)
2. Mr Francis Nuoh- Snr. Biostatistician, Ghana Health Service
3. Mr Prince Oppong-Darko - (Health Information Officer)
4. Dr Celia Brown (Research Supervisor- Warwick)
5. Prof Paramjit Gill (Research Supervisor- Warwick)

The DMEC will receive confidential data on the trial, monitor progress and give the relevant advice where needed. The CHW/CHN trial is a health education intervention delivered in the pregnant woman's home at the community level, and its evaluation depend on skilled births occurring within a year, hence the DMEC will not conduct any interim analyses.