Uncovering trajectories of childhood adversity with a latent Markov model approach, and their assioation with premature mortality

Specialeforsvar ved Clara Clipet-Jensen

Titel: Uncovering trajectories of childhood adversity with a latent Markov model approach, and their association with premature mortality
An application on nation-wide Danish register data from 1980-2015

Abstract:  Growing evidence in many versatile _elds of research suggests an association between adverse childhood experiences and somatic diseases related to mortality. This thesis sets out to implement a statistical method for lifecourse epidemiology, that enables the study of adverse childhoods and their potential association to premature death. With a Latent Markov (LM) model approach, children are categorised into latent classes of child-hood adversity that exhibit similar traits, and transition between them over the course of childhood. An adaptation of the LM model is conceived for time-to-event data, accom-modating the presence of an outcome of interest that decreases the sample size, such as premature death. The model is applied on nation-wide Danish register-based data from 1980-2015 of about 2.3 million children observed longitudinally from birth to 15 years of age. Discrete-time survival analysis shows that a trajectory reecting very adverse childhood results in higher hazards of dying prematurely than a trajectory of an outright non-adverse childhood.

Vejledere:   Susanne Ditlevsen, Naja Hulvej Rod, SUND
Censor:        Søren Andersen, Novo