Currently, Indonesia relies primarily on a national MMR estimate generated from the IDHS in 2008. The design and size of the IDHS sample meant only 62 PRDs, over a 5 year 'window', were found, and hence a very imprecise estimates even at national level. Given the diversity of Indonesia in terms of health services and health profile, this national average is likely to conceal more than it reveals, and because it is averaged over the last five years it will not capture recent changes. The lack of precise, local, and current estimates of maternal mortality in Indonesia is a major problem when trying to assess local needs, or the impact of programmes to reduce maternal risks - which was one motivation for developing MADE-IN/MADE-FOR. But it is also a major problem when it comes to assessing the validity of any measurement method, including MADE-IN/MADE-FOR. We have assessed the reliability, feasibility and efficiency of MADE-IN/MADE-FOR using internal consistency checks, and assurance of high quality data capture processes.
The key characteristic of MADE-IN/MADE-FOR is the use of existing networks of village informants to report vital events retrospectively. In combination with existing information on population and fertility, this allows a precise, local estimate of MMR to be obtained quickly from a 'one-off' survey. A companion 'population survey' allows investigation of risk factors, but increases costs. There is clearly the potential to develop the process into a sustainable ongoing system, which might result in higher quality data since events would be more recent - for instance verbal autopsy interviews would always be conducted within a reasonably short period of the deaths. Many countries in the region have also used villagers in health programs including in recording and reporting systems. A study in Cambodia, for example, involved Village Health Volunteers in a community-based surveillance system , and found that it could successfully fill the gaps of the current health facility-based disease surveillance system. A similar method has also been used in India .
The use of two informant networks in the MADE-IN/MADE-FOR is an important innovation, as it allows capture re-capture estimation. This makes the survey 'self-calibrating' in that it can estimate its own coverage, in a situation where it is clear that no single method captures all deaths. A study in Cambodia which tried to compare two surveys methods (a community based survey and a household survey) for estimating maternal and perinatal mortality found detection failures in both surveys, as high as 30-40% . A study in the USA found only 62% of maternal deaths were identified through death records .
The huge advantage of MADE-IN/MADE-FOR over household surveys like DHS, is that costs are reduced by a factor of 10-100. For the DHS, the level of investment required to ensure a high-quality survey, such as long periods of training, extensive pilot testing, separate household listing teams prior to the survey, and maintaining data quality tables during fieldwork, along with considerable technical assistance needed, can result in costs easily exceeding US$150 per household interview , implying $12 or more per woman-year of exposure (calculated from mean number of WRA in a household from the Indonesia DHS which is 2.4 and a 5 year 'window'), while MADE-IN/MADE-FOR only cost $0.1 per woman-year of exposure. It should, however, be noted that the DHS collects much more information than MADE-IN/MADE-FOR.
This two step method is very similar to the rather loosely defined RAMOS (reproductive age mortality study) method which is often regarded as the gold standard for estimating maternal mortality in developing countries, if conducted properly [17–19]. Both methods start by identifying deaths to WRA together with information on time of death related to pregnancy status. The difference is that the first step of RAMOS uses all available methods - existing records (e.g. vital registration, health facilities reports, burial records), and any appropriate types of informants, to gather data on deaths to WRA. MADE-IN/MADE-FOR in contrast, restricted itself to using Kader and RT informants only - to keep costs down. The second step of RAMOS, investigating the cause of death to WRA, is very similar to MADE-FOR, except the latter only involves visits to probable PRDs. The MADE-FOR step is crucial in verifying PRDs, as otherwise false-positives and undetected matching cases severely inflate the capture-recapture MMR estimate.