Determinants of Child Mortality in Kenya: Stratified Cox Proportion Hazard Model

This study explores determinants of child mortality in Kenya using the 2008/09 Kenya demographic and health survey (2008/09 KDHS) data. Cox proportional hazard model is used and stratification method was used to handle the non-proportionality of the hazards. Kaplan Meir log rank test is used to explore the data and determine the significant variables that have association with child mortality to be used in the model.

The study establishes that there is improvements in child survival over the years with variables such as mother and father’s education, place of delivery, wealth index , type of residence among other traditionally associated with child mortality being insignificant. Nyanza province shown a lot of improvement as most of earlier studies had put it as the risk province compared to central province. The results for region of birth shows that children born in north eastern province are 1.256 times to dies before their fifth birthday compared to central province. Central province is the second followed by coast. Eastern province had the least child mortality at 0.365. Male children were found to be 1.336 times more likely to die before their fifth birthday compared to females. Children born to families that share toilet were 1.245 times at risk of death before their fifth birthday compared to those who don’t share their toilets.

For marital status, compared to the married, children born to single parents were 1.124 times at risk while those born to separate parents were 2.063 times at risk of death before their fifth birthday.  Size of the child at birth was also significant with small children being 1.931 times at risk compared to the average. Every additional birth of a child to a mother in five years’ time increases the risk of death of that child by 1.66 times. Ethnicity, number of children born, age of the mother and literacy were also found to be the major determinants of child mortality.