Key points are not available for this paper at this time.
OBJECTIVE: The main objective of this study is to investigate how the direction and strength of the association between infant mortality and its predictors are changing over time in Bangladesh using a nationally representative sample for the period 2011-2014. DESIGN, SETTING AND PARTICIPANTS: Data from two repeatedly cross-sectional Bangladesh Demographic and Health Surveys (BDHSs) for the years 2011 and 2014 were used. A total of 7664 (with 312 infant death) and 7048 (with 264 infant death) complete cases, respectively, from BDHS 2011 and 2014 datasets were included in the study. METHODS: Cox's proportional hazard model with robust standard error (SE) that adjusts for the complex survey design characteristics was implemented to assess how the risk factors associated with infant mortality change their paths. RESULTS: Results reflected that administrative division remained as a potential risk factor of infant death for both periods. Household's socioeconomic status, father's employment status, age difference between parents turned out to be potential risk factors in 2014, though they did not show any significant association with infant death in the year 2011. In contrast to 2011, mothers' individual-level characteristics such as age at childbirth, education, media exposure, employment status did not remain as significant risk factors for infant death in 2014. Younger fathers increased the burden of death among infants of adolescent mothers. At higher order births, the burden of infant death significantly shifted from rural to urban areas. From the year 2011 to 2014, urban areas achieved socioeconomic equity in infant survival, while the extent of inequity was increased in rural areas. CONCLUSION: Community-based programmes should be designed for urban mothers who are expecting higher order births. To eradicate the socioeconomic inequity in infant survival, the government should design strong and sustainable maternal and child healthcare facilities, especially for rural areas.
Rahman et al. (Sun,) studied this question.