Lessons from Statistics without Borders Maternal and Infant Care Survey
Monica Dashen is a stay-at-home mother of three children who worked at the Office of Survey Methods and Research in the Bureau of Labor Statistics for years. She now enjoys applying her knowledge and skills to humanitarian projects.
In 2011, the military junta that had ruled Myanmar (and changed the name from Burma) for more than two decades, undertook vast reforms. The 2008 military-drafted constitution remained, and the new president—Thein Sein—was a high-ranking general. But the new government formally became civilian and the opposition leader, Aung San Suu Kyi, now holds a seat in parliament. In the process of democratization, Thein Sein, signed peace treaties with various ethnic minority rebels. Recently, Myanmar held a national election.
Meanwhile, Global Community Service Foundation (GCSF), located in Virginia, has been running a small charity in the Inle Lake area of Myanmar. In 1995, GCSF started building homes in the area and, in 2010, launched a maternal and infant care program. Retired government midwives (dubbed GCSF midwives) regularly travel to a prescribed set of villages and distribute prenatal vitamins, as well as advise the women about getting tetanus shots and worm pills from their local government midwife.
Now that the government is in the process of democratization, GCSF would like to expand its maternal and infant care program to more Inle Lake villages. To do so, GCSF needed to show the effectiveness of their program and find ways to scale up. GCSF partnered with Statistics without Borders (Maria Suchowski and the author) to formulate a program evaluation and find ways of expansion.
This SwB-GCSF collaboration sought to find out whether the GCSF program matters and identify critical knowledge gaps of maternal and infant care among mothers and health care workers. The plan was to compare the main part of Inle Lake, serviced by GCSF, to the lower part, not serviced by GCSF. Using the two-stage cluster sampling method, SwB interviewed 322 females capable of having children about their fertility history, medical care, knowledge, and beliefs.
The survey was fairly straightforward, as it covered the well-known maternal and infant care turf. The two most time-consuming components were questionnaire design and survey implementation.
No “off-the-shelf” questionnaire existed. The two major maternal and infant care surveys—Demographic Health Survey (DHS), sponsored by USAID, and Multi Indicator Cluster survey (MICS), sponsored by UNICEF—did not meet GCSF’s needs, so SwB designed a questionnaire from scratch, which was difficult because GCSF did not keep detailed records and it was not clear what the midwives did and did not do. To circumvent this problem, SwB conducted an initial study in which GCSF providers told us about their services, training, schedules, supplies, and so forth. Unfortunately, the midwives did not provide a detailed account of their services. To fill in the gaps, SwB relied on the DHS to identify possible services such as checking for worms and high blood pressure. SwB also conducted a focus group involving mothers from a GCSF-sponsored village and conversed with a USAID Burmese doctor about maternal and infant care.
The questionnaire was not built in isolation in the United States and implemented in Myanmar. Rather, it was piloted three times and underwent wording changes after each pilot. Knowing the survey would be the first survey the mothers ever took, SwB sought to make it as Burmese as possible. For example, an interviewer pointed out that asking whether families eat together was silly, as that is the custom. The intent was to find out whether women ate table scraps, so the question was deleted.
SwB’s efforts paid off. The Burmese mothers wanted to be involved in the full survey and were willing to spend the time. There were 53 recorded refusals/vacancies. Of the 53, 39 were reported as vacancies. A closer look at the data indicated the mothers were either at work or at a festival. During the interview, the mothers chatted about their birth history and dietary habits and beliefs. To prevent the fetus from having diarrhea, for instance, many pregnant mothers listed at least 10 types of beans they avoided, along with an extensive list of other healthy foods. (The women reported eating rice with a smattering of meats and vegetables.) Later, the translators surrendered their red pens and simply wrote “beans,” instead of translating each bean type.
Two lessons were learned in the implementation of the survey. The first lesson is that it took more time than expected due to the following three obstacles:
- GCSF staff and the interview team had to be brought up to speed on survey practices. For example, SwB members devised manuals based on the MICS and DHS manuals that describe implementation of standard practices in a developing country. In another instance, the interview team underwent a two-day training session that involved familiarization with the questionnaire and interview protocol.
- Access to all 20 villages (10 GCSF and 10 non-GCSF-sponsored villages) had to be gained. The GSCF staff—the original interview team—could only go to the villages they serviced. After many attempts to find interviewers who could visit all the villages, SwB finally garnered interviewers who were nurse aides at a local clinic.
- SwB had to gain cooperation from the respondents. Permission from local village heads was given and blessings from a top monk were granted after we gave gifts of rose-scented soaps. Also, goodwill was generated for the survey after GCSF made arrangements for and covered the costs of a mother delivering a healthy baby at a local hospital. The mother, who was in the final pilot study, reported a series of miscarriages and stillbirths.
The second lesson learned is that implementing a full-scale survey is local; it cannot be done remotely. On-the-fly decisions were made based on field conditions. Someone familiar with the survey process should be present at the start. During training, the importance of interviewing all women of child-bearing age, regardless of marital status, was pointed out. Questions about single women arose. After some back and forth, it was clear that women living alone were a rarity. Single women usually reside with their parents or in-laws. (A widowed woman typically remains with her in-laws after her husband’s death.) As it turned out, there were some women living alone who were interviewed.
The team interviewed 322 women residing in 20 villages. Homes were selected using the random walk method. To prevent response fabrication, the supervisor confirmed all interviewer routes and conducted re-interviews. The supervisor usually interviewed health care workers as the team travelled down the lake.
Implementation is a collaborative process. No one person can do it alone, and the investment in training time pays off. During training, the team was instructed not to make village substitutions; however, it arrived to empty villages early on. The mothers were at a nearby village for an all-day festival. But by calling the village head beforehand to ask the moms to stay and arriving earlier, the team was able to meet with the mothers another day.
The lack of records proved to be an obstacle in the initial data analyses, and no pre-natal care differences were found between the GCSF and non-GCSF-sponsored villages. Turning back to the records, it became apparent that the exact number of times midwives visited each village to distribute vitamins or perform checks was unknown. Eventually, SwB identified those villages served the longest—five years. The three villages served for less than two years were dropped from the analyses. Looked at this way, the data showed the GCSF program does matter. Mothers from GCSF-sponsored villages reported receiving more pre-natal care services than those from non-GCSF-sponsored villages.
The post-natal and neo-natal care questions derived from a conversation with a USAID Burmese doctor provided a direction for program expansion. Analyses indicated many of the mothers were unaware of post-natal danger signs (or maternal problems after birth). However, the doctors and nurse aides felt obliged to clarify misconceptions after each interview.
Finally, the GCSF-SwB survey showed the need for additional medical training for auxiliary midwives, community health care workers, and village women.
For more information about or to volunteer, visit the Statistics without Borders’ website.
SwB Helps NGOs Focusing on Vulnerability of Women and Children
Statistics without Borders (SwB) researchers along with SciMetrika—a nongovernmental organization (NGO) with technical capacity but few resources—focused on the social and economic consequences of the Haitian earthquake. In a 2014 Statistical Journal of the IAOS article, “A Nationally Representative Economic Survey Five Months after the Haitian Earthquake,” Ryung Kim, James Ashley, and Mary Corcoran found that more women were unemployed than men and there was an increase in the number of orphans. Such findings provide for more targeted interventions.
SwB members also have focused on innovative methods in field research. When a list of addresses was incomplete in war-torn Sierra Leone, SwB researchers Sowmya Rao and Gary Shapiro, together with Theresa Diaz, the principal UNICEF investigator, devised a way to draw a representative sample for a child mortality survey. Global positioning systems (GPS) mobile devices allowed the interviewers to map the houses in various neighborhoods. Later, when a random sample was selected, the interviewer returned to the household with the device to collect data, as the survey was already programmed in. Such innovations cut costs and increase efficiency. Their results were published in the 2012 CHANCE article “Use of GPS-Enabled Mobile Devices to Conduct Health Surveys: Child Mortality in Sierra Leone.”
In another instance, SwB researchers Michiko Wolcott, Joseph Pollack, and Minh Tran, together with Cat Graham and Chris Thompson of Humanity Road, collaborated to harness the power of social media in emergency situations. Using typhoon Haiyan as a case study, the group sought to provide a quick method for locating victims and defining priority damage areas by tweet counts. Such methods replace the more time-consuming search and rescue methods. Their results were published in a 2015 Statistical Journal of the IAOS article, “A Guide to Social Media Emergency Management Analytics: Understanding Its Place Through Typhoon Haiyan Tweets.”
Together, these examples highlight the general vulnerability of women and children in emergencies and the need for more targeted intervention of these populations in any emergency recovery process.