Effect of infant mortality on fertility in Kenya

Abstract

This study examines the effect of infant mortality on fertility. Specifically the study sought to examine whether the effect of infant mortality on fertility depends on the level of fertility and also to examine whether the effect of infant mortality on fertility varies across parity. This study used data from birth histories of women restricted to last five years prior to 2008/9 KDHS. The method of analysis used in this study was Cox proportional hazard regression. The results indicated that the death of a child leads to a higher risk of having another birth in both regions and across parity. However effect is confounded by length of previous birth interval and duration of breastfeeding. Breastfeeding was found to be associated with a higher risk of having another birth in both the fertility regions and also across the parities though with varying magnitudes. Short preceding intervals were also associated with a higher risk of having another birth in both fertility regions and also across the birth orders. The effect of infant mortality may be working through the curtailment of breastfeeding to affect the probability of having another birth. There is also evidence of behavioural effects through non-use of contraception after curtailment of breastfeeding. These results suggest existence of both biological and behavioural effects though more studies are needed before firm conclusions can be made regarding the distinguishing of biological and behavioural effects. The main policy implications that can be drawn from this study are that child survival programmes should be improved as part of the strategy to lower fertility in Kenya since the death of an infant leads to a higher risk of having another birth. More emphasis should also be placed on exclusive breastfeeding since short breastfeeding durations led to higher risks of the woman having another birth. There is need for further research on the effect of breastfeeding and amenorrhea and also contraceptive use as this study didn’t include the two variables because of limitations on breastfeeding data on the specific times of cessation of breastfeeding and the breastfeeding status of all children. The data on contraceptive use was also limiting as there is the need for the specific times in which there was contraceptive use in terms of adoption and discontinuation.