Survival Models: Application to Predicting the Breast-feeding Duration of Mothers in Urban Addis Ababa
No Thumbnail Available
Date
2001-05
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa, Ethiopia
Abstract
The benefi ts of breast-feeding for the health of the infant, as an inexpensive and appropriate
source of nutrients, for its simulation of a strong relationship between the mother and the
child, and as a major source of protection against pregnancy through suppression of ovulation
have been very well documented. But, only few have attempted to examine the factors that
determine the duration of brca s t-fcc~ i ng in Ethiopia. Thus using the appropriate stati stical
model thi s study analyzes the socio-demographic factors that determine the duration ofbrcaslfeeding
in Urban Addis Ababa.
Data on breast-feeding from the 1995 Fertility Survey of Urban Addis Ababa carried out by
the Central Stati sti cal Authori ty is used for the analysis. The breast-feeding duration of the
31.78% of the mothers is found censored. Therefore, for the overall examination of the factors
the Cox-regression model is used.
The prevalence and the med ian duration ofhreast-feeding using the Kaplan-Meier method are
estimated to be 95% and 14 months, respect ively. The log-rank test is used to examine the
breast-feeding differentials. To know the joint e ffect of the factors and to identify the more
important factors a multivariate analysis is undertaken. An approach based on statistical
modelling without prioritizing the variables, the ' Hierarchic' method is employed.
The final Cox's regression model shows that educational level of the mother is the variable
which signi ficantly affects the duration of breast-feeding. It is this variable with a negative
associat ion that best detennincs the breast-feeding durat ion. The shortest duration is observed
for mothers with higher levels of education. Finally, re levant discussion is made and the focus
group to which effective breast-feeding promotion campaign should be targeted is
recommended.
Description
Keywords
Survival Models: