Experimental Design
Research questions
- How does business-oriented compare to pro-social messages in motivating THPs to join a (formal) community health program?
- What is the impact of the community health program delivered by THPs?
- How do THPs compare to CHWs in delivering the community health program?
Sampling
The study builds on a census of all THPs in the Bafata region, selected by the Ministry of Public Health and UNDP Guinea-Bissau due to its high malaria prevalence. The study builds on a census of all THPs in the Bafata region, selected by the Ministry of Public Health and UNDP Guinea-Bissau due to its high malaria prevalence. In the census, we assess THPs' interest in joining the program, experimentally varying how the program is advertised. We reach a total of 1,437 THPs. A reinforcement phone contact follows the initial contact. After that, we implement a behavioral measure of interest in participating in the program, as we ask THPs to call the program to express their interest. We also conduct a face-to-face survey at the end of the project, in which we gather a measure of willingness to meet with representatives of the formal health system and traditional medicine. From the sample of villages where we interviewed at least once from May 2023 to May 2024, we sample all the villages where at least one THP expressed interest in participating, and we know of at least one active CHW. For the baseline, we survey 324 villages with a viable THP and CHW. At the endline, we plan to sample six representative households from each village, defined as the catchment areas of the corresponding THP/CHW. Only women aged 18 to 49 will be eligible for interviews, as they are the primary target population of the intervention and the CHWs’ strategy in the region.
Assignment to treatment
This project includes two randomized interventions. In the first set, we randomize messages to the THP population in Bafata, emphasizing (i) the program’s business-improvement benefits or (ii) its community health benefits. The randomized messages on the potential benefits of the THP integration program are submitted at the end of the census.
In the main set of randomized interventions, we assign selected villages in equal proportions to Treatment 1 (T1), Treatment 2 (T2), and the Control group (C). Randomization is conducted at the village level, using three stratification variables: health area, recruitment message type, and village size. Health areas are determined geographically, with some smaller areas aggregated to form larger strata. Within each stratum, we group villages into blocks of six, allowing random assignment to treatment and control groups in a balanced manner. For the remaining villages, those outside initial blocks are sorted by geographic proximity and further grouped into blocks of three based on recruitment message type and village size within each stratified variable construct. We therefore propose an experimental design that enables comparability between villages with an integrated THP, villages with an activated CHW, and villages with none, as follows:
- T1: THPs promote the adoption of recommended behaviors, such as using long-lasting, impregnated mosquito nets, and refer to health centers pregnant women to receive intermittent preventive therapy and any patient presenting malaria symptoms.
- T2: In this group, CHWs will receive the same training and will be tasked with the same activities as the THPs of T1.
- C: The control group will serve as a comparison group to both T1 and T2.
We will test the following hypotheses:
- H1: T1 improves health-related outcomes over C;
- H2: T2 improves health-related outcomes over C;
- H3: T1 improves health-related outcomes over T2.
Data
Regarding the first part of the project, testing the impact of pro-business and pro-social messages on the adoption of the program by the THPs, we implement a set of phone surveys and behavioral measures of willingness to integrate the program. We will continue assessing the messages’ impacts for the remainder of the project, i.e., the evaluation of the main intervention. Our measurement will include a broad range of data sources, as reported below.
- THPs and CHWs baseline, midline, and endline surveys
The surveys follow a consistent structure across baseline, midline, and endline. At baseline, we collect socio-demographic and professional data, assess service range and pricing for THPs, and document CHWs’ program participation and reports. We administer a knowledge test on malaria and antenatal care, measure diagnosis practices, beliefs about modern and traditional medicine, cooperation with health centers, trust in the health system, and time use, and assess willingness to pay for participating in an event with the MoH and the representatives of THPs.
The midline surveys focus on post-training outcomes, evaluating THPs and CHWs’ understanding of the program’s goals and activities, health knowledge, beliefs about modern and traditional medicine, and intentions to exert effort in the program and collaborate with health centers. We also document implementation costs for a cost-effectiveness analysis. The endline surveys assess THPs’ and CHWs’ awareness of the program’s goals and activities. We repeat the knowledge test on malaria and antenatal care and measure knowledge of health centers’ services. We assess time spent on the program, patients visited, referrals made, adherence to best practices, and cooperation with health centers. We also elicit beliefs about modern and traditional medicine, trust, and motivation. We assess service range, pricing, and revenues for THPs and document CHWs’ program participation and reports.
- Household endline survey
The household endline survey targets women aged 18-49, collecting socio-demographic data and interactions with THPs and CHWs. It includes a knowledge test on malaria and antenatal care, evaluates the adoption of preventive behaviors, antenatal care, health center visits, and follow-up on referrals. The survey also measures individual and household beliefs about modern and traditional medicine, trust in traditional versus formal health providers, and health outcomes.
- Referral vouchers
We will distribute two-way identified patient referral vouchers to THPs and CHWs as part of treatments (T1 and T2) to track referrals to health centers. The vouchers are distributed as part of the intervention and collected at the endline at the health centers to compare the effectiveness of both healthcare providers in making referrals at the village level.
- Health centers' baseline, midline, and endline surveys
The health center survey includes modules about working hours, services offered and their
pricing, equipment available, and personnel employed at all the health centers in the area of our study. We also inquire about the center’s relationship with the THPs and CHWs in the area.
- Health centers' logbooks
We will collect data in all the health centers’ logbooks in our study area at midline and endline. Logbooks with registries on antenatal care, birth-giving, postnatal care, family planning, and vaccination are organized by patient-date entries and homogeneous across facilities.
Outcome variables
We assess treatment effects across multiple outcomes at the THP, CHW, household level, and health center levels. The first intervention focuses on THPs’ willingness to participate in the program and final participation. The main intervention comprises multiple outcomes of interest at the THP, CHW, and household levels. The primary outcomes focus on health knowledge—particularly related to malaria and antenatal care—trust in modern and traditional health practitioners and beliefs about health centers, modern medicine, and traditional medicine. Additionally, we track adherence to best practices, cooperation with health center staff, and overall engagement with the program, assessed through time dedicated to the program, number of patients visited, and referrals to health centers. Secondary outcomes at the THP level include service range, pricing, and revenues in their activity, while CHWs include participation in the other activities of the CHW program and reports delivered.
At the household level, primary outcomes cover health knowledge, trust in modern and traditional health practitioners, beliefs about health centers, modern medicine, and traditional medicine. Moreover, we will assess preventive health behaviors, antenatal care, and visits to health centers. We also monitor follow-up on health center referrals and overall health outcomes. Secondary outcomes include individual and household-level health outcomes.