Celine Guimas

  1. The project

Maternal and Neonatal Health (MNH) is a major area for health development and despite expansions in health services across developing countries, improvements in MNH in these countries have been slow. The Transparency for Development (T4D) project utilizes local knowledge to design a transparency, accountability and participation intervention that is motivated by the desire to improve MNH and empower citizens in rural communities. Furthermore, the project aims to understand whether these types of interventions work and why they work (or not).

Launched in 2013, this project stems from the collaboration between the Harvard Kennedy School’s Ash Center, the Results for Development Institute (R4D) and the University of Washington. The T4D project focuses on maternal and neonatal health in Indonesia and Tanzania.

The T4D intervention uses a modified version of a “community scorecard” which highlights local key indicators related to health facility providers and uptake of key MNH services. The intervention is comprised of different activities (cf. Figure 1) which aimed to increase community’s awareness and health providers’ accountability. Social actions included examples like organizing a community education campaign on the importance of antenatal care or confronting a healthcare provider who had been stealing medicine.

These actions may impact and improve health outcomes by:

1) Increasing utilisation of health services;

2) Improving content of clinical care;

3) Motivating people who receive lower quality care at one outlet to choose to seek care at a higher quality outlet.


Source: Arkedis et al. (2016, p. 11)

EDI collected the data for the baseline survey of T4D evaluation in Tanzania. In Indonesia, the baseline data were collected by SurveyMeter using comparable tools to those used in Tanzania. This evaluation seeks to determine the effects of the intervention: (1) on the utilisation of health care services related to maternal and child health, (2) on the content of these health care services, (3) on the health outcomes and (4) on the perceived and actual citizen empowerment and efficacy. If there are significant effects, the evaluation also aims to highlight: (5) the main mechanisms through which these effects happen and (6) the role of context in shaping or determining these mechanisms.

  1. Summary of fieldwork in Tanzania

The impact evaluation design consists of a Randomized Controlled Trial (RCT) in both Indonesia and Tanzania, with 100 treatment and 100 control villages in each country. The baseline data collection  involved three separate surveys: a health facility survey, a household survey, and a community survey.

In Tanzania, EDI conducted the baseline survey across Tanga and Dodoma regions. 77 villages were in Dodoma and 123 in Tanga. Between March and July 2015, EDI interviewed 153 health facilities, 200 communities as well as a total of 3000 households.

The health facility survey was administered in health dispensaries, which are the lowest level in the Tanzania’s pyramidal official health system. The household survey was conducted with women who had given birth in the 12 months prior to the survey and the community survey was administered to a small group of approximately 5 participants (typically the village chairman, village executive officer, village councillor(s), member(s) of the village health committee, and other local leaders).

In order to randomly select the respondent of the household survey, EDI first conducted a door-to-door listing exercise. In each sample village, 3 “kitongoji” (i.e. 3 sub-villages) were randomly selected. The listing used a zigzag type procedure such as the village guide selected a corner of the kitongoji and the enumerator then moved across the entire kitongoji. Once eligible respondents were listed, sample respondents were selected using random number tables. The sample respondent list was then entered into surveybe CAPI software and pulled through as reference table used to pre-fill the household questionnaire.

Surveybe software has been used for the entire baseline survey and has been a flexible and convenient tool. During the household survey, GPS coordinates were taken at each household in order to ease future tracking and data collection. Images of the antenatal care (ANC) cards owned by the mothers were also captured in order to provide visual evidence for data users. Surveybe’s multimedia functionality was also particularly useful during the facility survey to enable data users to visualise the level of cleanliness, equipment, and organisation of delivery rooms.

  1. Key findings of the baseline survey in terms of MNH outcomes:
  • ANC access in Tanzania was relatively low with 43% of recent mothers who completed the recommended 4 ANC visits during pregnancy, compared with 90% in Indonesia. However, the vast majority of Tanzanian women (98.4%) had received some form of ANC care during pregnancy.
  • The percentage of pregnant women who delivered at a health facility – one of the key indicators for the health of babies – was low for both Indonesia and Tanzania (55% and 56% respectively).
  • Post-natal care access was higher in Indonesia than Tanzania, but quality of care was low in both countries. In Indonesia, only 59.7% of babies had received comprehensive post-natal care with a skilled provider. In Tanzania, this falls to 36.1% of babies.
  • In terms of general citizen empowerment, 3 out of 4 households in both countries reported feeling they could make their lives better. Health providers were perceived as having high responsiveness to complaints, but local government was perceived as having low responsiveness.

The endline survey in Tanzania and Indonesia is planned to be conducted in mid-2018.

To read the Full Baseline report (Arkedis et al., 2016, BASELINE REPORT, Transparency for Development, September 2016), click here.