Maternal Death Analysis: Trend Analysis with Rare Events

Undergraduate: Hannah Inman

Faculty Advisor: Jane Monaco
Department: Biostatistics

---

Every year, the North Carolina State Center for Health Statistics (SCHS) collects data on women between the ages of 10 and 50 who died within a year of giving birth. These deaths are collectively referred to as "maternal deaths" and have a variety of causes, including drugs, automobile accidents, and the pregnancy or birth itself. This project uses SCHS data to analyze maternal deaths by cause from the years 1999 to 2013. According to these data, the number of deaths from year to year was between 36 and 75. Of those deaths, an average of 22 women died of pregnancy or birth-related causes each year. The question is: Has there been any consistent trend over time due to any particular cause of death? Analysis of these data is affected by several factors: small sample sizes, changes in regulatory requirements for reporting, and the need to maintain the privacy of the women and families involved. Descriptive statistics and linear regression models were calculated using statistical software (SAS 9.4). Initial findings indicate that there has been no linear increase or decrease in maternal deaths due to some causes, such as automobile accidents and homicide. However, there has been a statistically significant linear increase in deaths due to drugs. An objective statistical analysis of maternal death data is critical to developing public policy measures to prevent such deaths.