Modelling of Factors Associated With Perinatal Mortality Rates in Tanzania-analysis of Tanzania Demographic Health Survey 2015-16 Using Bayesian Approach

Abstract

Background: Perinatal mortality is a big problem facing child survival in Tanzania. The problem has been empirically associated with numerous factors. The challenge which exists is the statistical association of these factors with perinatal mortality. This study aimed to investigate the determinants of perinatal mortality in Tanzania based on the TDHS 2015-16. Methodology: The study used data from the Tanzania Demographic Health Survey whereby an assessment of the determinants of perinatal mortality was carried out using bivariate chi-square test, normal Bayesian logistic regression and multilevel Bayesian analysis in R-INLA. The multilevel relative risk was computed using log-binomial models in R-INLA. The chi-square test and regression models were undertaken in R version 4.0.3. Results: A total of 13,266 women aged between 15-49 years and 10,233 children born in the previous 5 years period were included in the study. Overall, 187 stillbirths and 215 neonatal deaths were recorded in the study period, giving a total of 402 perinatal deaths. Perinatal mortality was significantly higher in children of higher birth orders (2+) and maternal body mass index 25+, but lower in children with previous birth intervals of 15+ months. In hierarchical Bayesian analysis, body mass index, previous birth interval and birth order were independently associated with perinatal mortality. Birth order 2-3 (adjusted Odds Ratio (AOR) 5.19 95% CI (5.17,5.21), birth order 4-5 AOR 3.39 95% CI (3.37,3.40), birth order 6+ AOR 5.43 95% CI (5.40,5.46) and body mass index 18.5-24.9 (adjusted Odds Ratio (AOR) 2.22 95% CI (2.20,2.24), body mass index 25-29.9 AOR 2.86 95% CI (2.81,2.91) and body mass index >30 (AOR 2.86 95% CI (2.83,2.89) were associated with increased risk of perinatal mortality. On the other hand, pregnancy intervals 15-26 months, 26-38 months and 39+ months were associated with lower risk of perinatal mortality with AORs

0.336 95% CI (0.335,0.338), 0.340 95% CI (0.338,0.341) and 0.099 95% CI (0.096,0.102) respectively. Conclusion: Perinatal mortality remains a significant problem in Tanzania. We have identified some of the risk factors, indicating the need to invest in educating mothers about previous childbirth intervals, birth order and body mass index. Reduction of perinatal mortality should also be attributed to interventions that focus on the mother’s occupation, age at first birth and education level. Health planners and managers could consider putting a strategic plan in action to assist in improving maternal health in Tanzania, thereby reducing the risk of perinatal deaths.