Statistical Analysis Identifies Influential Factors for Neonatal Mortality in Ghana’s Ashanti Region

By Lina Sorg

During the neonatal period—defined as the first 28 days of a newborn’s life—infants are particularly susceptible to infection, illness, and other post-birth risk factors. Neonatal death rates (death of a live-born child within this month) are low in developed countries but concerningly high in developing countries. In fact, the “Good Health and Wellbeing” indicator of the United Nations’ Sustainable Development Goals specifically aims to reduce neonatal mortality worldwide to at least as low as 12 deaths per 1,000 live births by 2030. This is an ambitious goal, given that the neonatal death rate in low and middle-income countries is roughly 40 to 50 deaths for every 1,000 live births. In contrast, the rate in high-income countries is between two and four deaths for every 1,000 live births. Infant deaths stem from a variety of factors, including infections, preterm birth, low birth weight, intrapartum-related complications like asphyxia, and birth defects.

In a minisymposium presentation at the 2023 SIAM Conference on Computational Science and Engineering, which is currently taking place in Amsterdam, the Netherlands, Elizabeth Amankwah of Kwame Nkrumah University of Science and Technology conducted a statistical analysis of neonatal mortality in the Ashanti Region of Ghana. The number of children who die in Ghana within 28 days of birth declined from 43 deaths per 1,000 live births in 2014 to 27 deaths per 1,000 live births in 2017. However, the Ashanti Region was the only region in Ghana that experienced a growth in mortality during this time period; neonatal mortality increased from 42 deaths per 1,000 live births in 2014 to 52 deaths per 1,000 live births in 2017. “Many Ghanaians don’t see their babies grow up,” Amankwah said. “We wanted to see what was happening so we could add something to society and help the Ashanti Region.”

To assess the maternal, neonatal, and health care system variables that impact newborn fatalities in Ghana’s Ashanti Region, Amankwah developed a model—with neonatal mortality as the dependent variable—to predict neonatal mortality and identify the prevalent contributing factors. She used 2017 data from the Ghana Demographic Health Survey, which recorded 7,408 live births from a population of 25,062 women aged 15 to 49. Amankwah examined a variety of independent variables, including maternal age, infant gender, educational level, antenatal facility, antenatal care provider, religion, and weight. According to the data, most infants die within the first six days of life. Mothers who are over 30 years old have higher rates of neonatal mortality than their younger counterparts. Infant mortality rates were also higher for mothers with little (primary/middle) to no education, and more male babies die in the Ashanti Region than females.

Amankwah utilized a chi-square test and ran each independent variable against neonatal mortality (the dependent variable). She determined whether the independent variables were important based on their p-values, which she tested at a significance level of five percent based on the following hypotheses:

• \(H_0=\) There is no association between <independent variable> and neonatal mortality.
• \(H_1=\) There is an association between <independent variable> and neonatal mortality.

"There is no link between the independent factors and the dependent variable for the independent variables whose p-values are greater than the specified 0.05,” Amankwah said. 

Next, Amankwah employed a stepwise logistic regression analysis in R to predict the survival of a baby in the first 28 days of life. “The full model was reduced to a restricted model that only includes variables that—in the presence of others—are significant to the dependent variable,” she said. The regression hence excluded all independent variables except the sex and birth weight of the child, both of which were deemed significant. Amankwah then conducted an outlier test, multicollinearity test, and linearity test to verify the model’s accuracy.

In the future, Amankwah hopes to continue her study by comparing the 2017 findings to data from other years. “Proximate factors are highly significant in determining whether a child is alive or dead,” she said. “Socioeconomic factors were found to be less important predictors of a child’s death when compared to the other variables. The weight and sex of the child are the factors causing an increase in neonatal mortality in the Ashanti Region.” 

Lina Sorg is the managing editor of SIAM News.