Predicting prenatal care rates in rural Ethiopia

June 12, 2023—Through predictive models, it may be possible to identify pregnant women in low-resource settings who are at high risk of failing to attend prenatal care, in order to develop interventions to encourage their attendance, according to a new study led by Harvard T.H. Chan School of Public Health.

The study was published on May 31 in JAMA Open Network. Harvard Chan School co-authors included several members of the Department of Epidemiology, including Grace Chan, associate professor; Bryan Wilder and Frederick Goddard, visiting scientists; Clara Pons-Duran, postdoctoral research fellow; and Delayehu Bekele, department associate. Sebastien Haneuse, professor in the Department of Biostatistics, was also a co-author.

No prior studies have aimed to build models that identify women in poor countries at risk of skipping prenatal care visits during their pregnancies, according to the co-authors. To fill in this gap, they analyzed health data from 2,195 women in Amhara Region, Ethiopia, who were pregnant between December 2018 and March 2020, finding that 582 of them failed to attend at least one prenatal care appointment during their pregnancy. Based on this information—and taking into account socioeconomic, demographic, nutritional, obstetric, and medical-history related factors—the researchers then developed six models to predict prenatal care attendance among women in the rural region.

The models showed modest performance, the study found. They were most successful at identifying women at high risk of failing to attend prenatal care appointments, but not specific enough to identify women at moderate risk. According to the researchers, more data—such as an increased sample size, or information on how much women used health services before getting pregnant or during previous pregnancies—would improve the models.

Nonetheless, the researchers deemed the models a step in the right direction, not just for reproductive and maternal health, but for all areas of health in low-resource settings. “Our study opens the possibility to start exploring the development and validation of easy-to-use tools to predict health-related behaviors in settings with scarcity of resources,” they wrote.

Read the study: Development of Prediction Models for Antenatal Care Attendance in Amhara Region, Ethiopia