New application of demographic and health survey in nutrition modeling


This post is the first in a series around “Evaluation in the 21st Century,” this year’s theme for the annual Evaluation Conference, hosted by the American Evaluation Association.

With the technological leaps we’ve seen in the early years of this century—including the rise of computing power and the Internet—as well as more sophisticated survey methods, public health practitioners and evaluators have access to incredible amounts of data. A key task for evaluators today is to think of new ways to use existing data for purposes beyond their original intent. In the spirit of 21st century evaluation, analysts at the SPRING project decided to use a novel approach to answer a complex question by using source data from the public health golden standard dataset—the Demographic and Health Survey (DHS)—to develop their model.

dagostino blogNutrition, non-communicable disease, and the links between the two are garnering increased attention in public health today. Birth outcomes (for example, having a low birth weight) and nutrition early in life have marked effects on health later, and the USAID-funded SPRING project was interested in investigating that link. In 2012, SPRING conducted a large modeling exercise to answer some compelling questions about the effect of infant and young child nutrition programs on later-life non-communicable disease rates. As part of the modeling exercise, analysts needed to be able to link family planning use with birth outcomes in specific countries, but struggled to find the right approach. While public health researchers routinely study family planning and birth outcomes independently, it was difficult to get access to data which accurately measured both family planning use and birth outcomes together.

To address this issue, SPRING decided to split the question into two pieces, using the length of interpregnancy intervals (IPIs) as the key link between the two pieces of information. An IPI is the length of time between the birth of the first child in a timeline and conception of the second child, and is routinely collected as part of the DHS. Though researchers aren’t completely clear on how IPI affects birth outcomes like birth weight, they hypothesize that it is because a longer IPI allows women’s bodies to replenish micro- and macronutrients between pregnancies (Wendt et al., 2012)

SPRING conducted the analysis with the DHS data by linking family planning use and IPI (both collected in the reproductive calendar), then linking IPI with reported birth weight. Though we would have wanted to make the direct link between FP use and birth weight, even the DHS has trouble collecting complete data around birth weights. Only one-third of the women with two pregnancies in the reproductive calendar who provided complete information about FP use and IPI were able to report the birth weight of their children. Splitting this into two steps allowed as much use as possible of the available data.

By sharing this work more broadly, we can showcase a new way of using existing data. Our adapted methods and our outcomes also serve as a call to action to public health researchers to think about new ways of measuring indicators, such as birth weight, which have a huge effect on health throughout the lifecycle, but because of recall difficulty, can be very complex to measure on a regular basis.

Learn more on the SPRING website, or visit Alexis at the #eval13 Poster Session from 6-8 pm on Wednesday, October 16.



Data for the modeling exercise was taken from the International Institute for Population Sciences (IIPS) and Macro International. 2007. National Family Health Survey (NFHS-3), 2005-06, India: Key Findings. Mumbai: IIPS. Available at:

Wendt, Amanda, Cassandra M. Gibbs, Stacey Peters, and Carol J. Hogue. 2012. “Impact of Increasing Inter-pregnancy Interval on Maternal and Infant Health.” Paediatric and Perinatal Epidemiology 26: 239–258.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.