Kliniken & Institute … Institute Heidelberger Institut… Groups & Projects Working Groups Epidemiology and… Research Biostatistics Spline regression

Using penalized spline to model age- and season-of-birth-dependent effects of childhood mortality risk factors in rural Burkina Faso

Project team: Heiko Becher

External Collaborator: Göran Kauermann, Pavel Khomski, Universität Bielefeld
Bocar Kouyaté, Burkina Faso

Funding: own funds

Funding period:  2007-2009

Status: completed

Several previous studies have identified risk factors for childhood mortality in high risk areas, such as Sub-Saharan Africa. Among these are lifestyle factors related for example to nutrition or sanitation. Others are related to social class, ethnicity and poverty in general. Few studies have investigated a dependence of these factors by age and season of birth which was the focus in this study.

We performed a survival analysis of 9121 children born between 1998 and 2001 in a rural area of western Burkina Faso. The whole population is under demographic surveillance since 1993. All cause mortality was used as the endpoint. Follow-up information data until the age of five was available. Recently developed spline regression methods were adopted for analysis.

Ethnic group, religion, age of mother, twin status, sex, and distance to next health center were used as covariables all of which having a clear effect on survival in standard Cox regression analysis. Ethnicity is more related to death at early age, as well as age of mother. The effect of the risk factors considered also appear to be related with season of birth. With penalized spline regression, a more detailed risk pattern analysis is obtained.


Becher H, Khomski P, Kauermann G, Kouyaté B: Using Penalized Splines to Model Age- and Season-of-Birth-Dependent Effects of Childhood Mortality Risk Factors in Rural Burkina Faso. Biometrical Journal 2009; 51:110-122.