Statistische Demografie

Auf einen Blick Projekte Publikationen Team


Smooth Age-By-Covariate Interaction in Hazard Models

Angela Carollo, Jutta Gampe, Roland Rau, Marcus Ebeling; in Zusammenarbeit mit Paul HC Eilers (Erasmus University Medical Center, Rotterdam, Niederlande), Hein Putter (Leiden University, Medical Center, Niederlande)

Ausführliche Beschreibung

When death rates (or other hazards) are studied, the age-trajectory of the force of mortality is commonly modified by the values of the covariate in a particular (simple) way. The most prominent model is based on the assumption of proportional hazards, and additive hazards models are also considered.

For a continuous covariate, for example Body Mass Index (BMI), the impact of different values of the risk factor can be non-monotone, with very low and very high values being linked to higher mortality, and hence this nonlinear influence should be captured. Furthermore, this effect can vary with age. Analyzing these two aspects jointly requires a modeling approach that can capture the full interaction between the age-trajectory of mortality and the nonlinear covariate influence.

This project employs P-splines to estimate a smooth hazard-by-covariate surface. The method is applied to study the impact of BMI on age-specific mortality. In a further extension, we will study the evolution of the surface over calendar time.


Statistik und Mathematik

Das Max-Planck-Institut für demografische Forschung (MPIDR) in Rostock ist eines der international führenden Zentren für Bevölkerungswissenschaft. Es gehört zur Max-Planck-Gesellschaft, einer der weltweit renommiertesten Forschungsgemeinschaften.