Terms of Reference for Baseline Data Collection on Cluster Randomized Trial of the impact of Birth Preparedness Savings (BPS) Program on Health Facility Delivery in Gombe Northeast Nigeria at PACT

Background

Specifically, in 2016, SAQIP established 120 mothers’ groups across 23 of 114 political wards in four Local Government Areas (LGAs) in Gombe. Each group has approximately 25 women of reproductive age. These women are offered Pact’s unique WORTH intervention - an award-winning micro-banking program that empowers women to lift themselves and each other out of poverty17 – for 18 months. Every week, members of the groups receive obstetric and newborn health education, and lessons on how to read and write.  Every two weeks, the groups conduct community banking sessions where all members are required to make small savings, and when group funds grow large enough, members may begin taking loans to start small businesses or expand an existing one17. The members are also encouraged to save voluntarily into the group’s purse. This voluntary savings can be withdrawn at any time to meet life’s exigencies. SAQIP provides no capital or seed money to these women. The groups only receive notebooks for recordkeeping and lock-boxes for storing their savings and ledgers17. The groups are facilitated by two types of volunteers. The maternal newborn and child health (MNCH)/ village health worker (VHW) volunteer facilitates the groups’ health education sessions. The empowerment worker (EW) volunteer facilitates the community banking sessions. These volunteers receive a monthly stipend for transportation from SAQIP.

To date, SAQIP has formed 660 mothers’ groups across 57 of the 114 wards in Gombe. The project has also now decided to offer some of the groups a birth preparedness savings (BPS) program to further motivate and increase members’ ability to pay for and deliver their babies in health facilities. If the BPS works to improve health facility delivery amongst members of the mothers’ groups, SAQIP hopes to scale the program to all groups. Coupling birth preparedness and complication readiness strategies with savings-led initiatives may improve skilled birth attendant utilization18. And while some evidence suggests women who prepare for child birth have a greater likelihood for institutional delivery19–22, the impact of birth preparedness and complication readiness amongst pregnant women remains ambiguous22–24; and requires more studies to understand21,25. This protocol describes the approach to assess the impact of BPS on health facility delivery amongst SAQIP’s mothers’ groups’ members in Gombe state.           

Objectives

The primary objective of the study is to determine whether BPS increases health facility delivery amongst SAQIP’s mothers’ groups’ members who had a live birth in the previous 12 months compared with similar members in mothers’ groups not offered BPS.

The secondary objective will answer the following questions:

• What is the effect of SAQIP’s mothers’ groups’ intervention on health facility delivery irrespective of BPS?
• What is the difference in health facility delivery between mothers’ groups established at different time periods?
• How much do pregnant women save in preparation for childbirth and how is this distributed considering their poverty status?

Methodology

Study design and setting

A cluster randomized controlled trial will be conducted. The trial will have two arms and the ward will be the unit of randomization. The ward is chosen as a unit of randomization due to budget reasons. Also, it is easier to control for contamination that can be introduced by VHWs by making the ward the unit of randomization. The hypothesis that SAQIP’s Birth Preparedness Savings (BPS) program will increase health facility delivery amongst pregnant members of the mothers’ groups will be tested. The study is set in Gombe state, northeast Nigeria, and will be implemented over 12 months. Gombe State is one of the 36 states of the Federal Republic of Nigeria, located in the center of the north east of the country on latitude 9”30’ and 12”30’N, Longitude 8”5’and 11”45’E. It is bordering Borno, Yobe, Adamawa, Taraba, and Bauchi states. Gombe has 11 Local Government Areas and 114 political wards26.

Study population

SAQIP has established 660 mothers’ groups’ in 57 of 114 political wards in Gombe state. Each ward has approximately 11 mothers’ groups. Of the 660 mother’s groups (with 16,500 members), 220 were established between 2016 and 2017(old mothers’ groups) and the remainder in 2018 (new-mothers’ groups). This study will recruit wards from the 57 SAQIP has mothers’ groups in. The study populations are all members of the mothers’ groups with a live birth in the last 12 months prior to the survey. Across the old and new mothers’ groups, of the 16,500 members, approximately 2805(17%) are expected to have had a live birth in the last 12 months15.

Inclusion Criteria

All 57 wards and 2,805 members with a live birth in the last 12 months will be eligible to participate in the study.

Exclusion criteria

New mothers’ groups formed during and/or after the baseline study; and new mother’s groups members that join the groups after the baseline study, will be excluded from the baseline survey.

Randomization 

This study uses randomization to assign wards to intervention or control arms, using computer-generated random numbers. This allows for causal attribution of the change observed in health facility delivery amongst women with a live birth in the last 12 months due to the intervention. Using the ward as the unit of randomization prevents spillover effect from the intervention to control groups.

Random allocation of wards will take place before baseline measures and will be conducted by a member of the study team blinded to any identifying information about the wards and mothers’ groups. To control for possible geographic confounders, the wards will be stratified by new and old mothers’ groups, and rural/urban characteristics. Additional confounders like the characteristics of VHWs facilitating the groups will be controlled for during analysis using regression techniques.

The intervention cannot be offered to all groups due to administrative, logistics, and budget constraints. Furthermore, considering spillover, although it would be possible for members of the groups to visit other wards for personal reasons, this is not considered a significant threat to the study. We will also reduce potential contamination by carefully allocating VHWs to mothers’ groups. A VHW may only be assigned several Mothers’ groups all within the same ward to further control for contamination i.e. the VHWs assigned mothers’ groups offered BPS will not be assigned to mothers’ groups not offered BPS.

Sample strategy and power calculations

All 57 wards will be considered eligible to participate in the study. The eligible wards will be stratified by urban-rural classification. Twenty-four wards will be randomly selected from the 57 wards to participate in the study using proportionate sampling technique. The 24 participating wards will be randomized into 12 treatment and 12 control wards. All mothers’ groups within each treatment ward will receive the BPS intervention and business will continue as usual in the control wards (Figure 1). The sample size of 24 wards has a power of 80% (α=0.05) to detect a difference of ≥19% for the change in the proportion of MG members with a live birth in the last 12 months who delivered in a health facility when baseline and end-line estimates are compared for intervention and control wards. The sample size calculated was performed using the following equation for the comparison of two proportions:

The equation was calculated according the following parameters:

Parameter

Value

Definition

α

0.05

Significance level

β

0.8

Desired power of the test

Tail

1

One-tailed or two-tailed test

 

1.64

Z-value corresponding to the desired significance level of the test

0.84

Z-value corresponding to the desired power of the design

m

40

Expected # women with a live birth in the last 12 months per ward

ρ

0.2

The coefficient of variation of true proportions between clusters within each group

 π0

0.45

True (population) proportion in the absence of the intervention

π1

0.59

True (population) proportion in the presence of the intervention

c 

12

Number of clusters/wards in each group

The ρ was estimated from the November 2016 end-line data of the firth 120 mothers’ groups established by SAQIP across 23 LGAs in Gombe state. Using the end-line dataset, an ANOVA was performed on the proportion of women with a live birth in the last 12 months that delivered in a health facility, treating the binary outcome data of ‘did not deliver in a health facility /delivered in a health facility’ as 0/1 dummy variables. The π0 was derived from IDEAS’ August 2017 household survey results table. The household survey was conducted across the same 57 wards SAQIP is implementing its MG intervention.

Outcome measure for the study

• Health facility delivery amongst members of mothers' groups who had live births in the last 12 months.
• Post-natal care visit within 48hrs amongst pregnant members who had a live birth in the last 12 months.
• Average number of ANC visits amongst members with a live birth in the last 12 months.
• Willingness to deliver babies in a health facility amongst currently pregnant members of the mothers' groups.

Data collection strategy

Overall, your firm is expected to propose a data collection and quality strategy for this study. Though not cast in stone, we however expect the minimum criteria below:

Data collectors

Data collectors must be trained and familiar with the purpose of the study. Questions on the data tools must also be interviewer administered and captured using electronic data capturing devices. The data collectors must:

• Have experience collecting data for similar or larger studies;
• Have at least, an HND or equivalent academic qualification;
• Be females;
• Reside within or are very familiar with the target LGAs; and
• Speak Hausa fluently

The electronic data capturing tool must allow the data collectors send data from the field using the internet, to a secured database - in real time. SAQIP must have access to the database through an online portal at all times during the study.

Data Tools

The same questionnaire used first and second wave studies will be utilized for this study. The questionnaire has been adapted from IDEAS’ Gombe Household Survey tool used in 2012, Nigerian Literacy Survey tool, Nigeria Demographic Health Survey tool of 2013, and Grameen’s Foundation Progress out of Poverty Index tool.

SAQIP will however minimally update the tool to capture data that will throw more light on learnings that have emerged during the course of program implementation. Any major alterations to the tool however, will require that the tool and/or that section of the tool be piloted before field work. All modifications to the tool will be done by SAQIP only. Your firm is however expected to make recommendations when necessary.

Field work

All fieldwork must be implemented by the trained data collectors and conducted in accordance with the sampling plan described in the methodology section above. The fieldwork must also be carried out within the approved period for the assignment. The work plan for the field work will be approved by Pact and the timelines therein must be adhered to. Your firm would provide logistical support to fieldwork teams, including transportation, communications, meals, and other incidental expenses.

Data Quality

Data quality must be ensured throughout all the phases of data collection and transfer. It is important that all questionnaires are completely and appropriately filled. The dataset is expected to be free of errors. All errors must be thoroughly investigated and corrected by your firm before submission of the dataset. SAQIP has the right to refuse the data and request that the vendor repeats the data collection exercise at no cost to SAQIP. Checks should be put in place to ensure that invalid entries or incompletely filled questionnaires cannot be submitted. Data collectors must be accompanied by supervisors during field work.  The generated data will be the sole property of Pact Nigeria and cannot be shared or used for any purpose without the written permission of Pact Nigeria.

Data Files

Before submitting final data files to SAQIP, your firm should conduct consistency checks and data cleaning based on pre-determined rules. Your firm should generate a list of these rules and submit to SAQIP for approval during the fieldwork period. Your firm should implement these cleaning rules before delivering the final files. The final data files should be submitted as dta, .xls, and .dat files. All cleaning logs should accompany the final data files. In addition, your firm should prepare data dictionary/codebook for the data which list the variables and their codes.

Report writing

After data collection is completed, your firm will prepare a report that summarizes the data collection procedures, any challenges and solutions, and other relevant information. The final report should be divided into a section describing the fieldwork and a section with the results of the data analysis.

Ethics

Ethical approval for this study will be obtained by SAQIP from Gombe State Ministry of Health’s Ethical Review Board. However, it is the responsibility of your firm to obtain administrative approval from Gombe State Primary Health Care Development Agency and the respective LGAs PHC Departments before contact is made with the respondents of the study. It is expected that your firm will also ensure proper community entry and/or obtain permission from the heads of the households during field work.

Additionally, informed consent must be obtained from all respondents.  Participation in the study must be strictly voluntary. Your firm will ensure full confidentiality of all the information obtained from the respondents during the survey.  The entire raw and anonymized dataset must be transferred to Pact in password enabled electronic files, after field work is completed.

Deliverables

Specifically, your firm will deliver the following deliverables:

Reporting Requirements

(Activities and/or Outputs)

To be completed by no later than these Due Dates

Delivery Instructions

• Data collectors’ roster and qualifications

June 4

Soft copy

• Data collection tools for the study

June 6

Soft copy

• Training materials for data collectors’ training

June  10th   

Soft copy

• Scripted and desk-checked PDA

June 10th

Physical presentation of PDAs

• Interim data files

June 19th

SPSS

• Final data files structured in .dta, .xls, and .dat (including paradata) as well as structured metadata documentation, cleaning logs, codebook

June 24th

Soft copy

• Report comprising of a comprehensive field report and descriptive statistics

July 2nd

Soft copy

Key tasks for your firm

The tasks to be completed by the firm include, but are not necessarily limited to the following:

• Propose a data collection and quality plan for the study.
• Recruit qualified and experienced data collectors.
• Train the data collectors on the purpose and tools for the study.
• Liaise with stakeholders as needed.
• Implement the SAQIP-Approved Work Plan for the assignment;
• Ensure data quality throughout the process.
• Proactively inform Pact of any significant modifications to the intervention/project that could affect the evaluation and any difficulties that may arise in implementing the approved study design.
• Submit a final study report.

Remuneration

Major cost drivers will cover firm’s fee rates and travel costs. Pact Nigeria will bear the cost of relevant workshops and the printing and dissemination of the study report. Pact will not provide logistical support and expects that your firm will work independently.

Line of reporting

The firm will report to Pact Nigeria’s Director for Monitoring Evaluation Research and Learning, who will be responsible for communicating with the wider Pact team.