Moger Tron Anders, Haugen Marion, Yip Benjamin H K, Gjessing Håkon K, Borgan Ornulf
Department of Biostatistics, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway.
Lifetime Data Anal. 2011 Jul;17(3):445-60. doi: 10.1007/s10985-010-9188-3. Epub 2010 Nov 4.
We present a hierarchical frailty model based on distributions derived from non-negative Lévy processes. The model may be applied to data with several levels of dependence, such as family data or other general clusters, and is an alternative to additive frailty models. We present several parametric examples of the model, and properties such as expected values, variance and covariance. The model is applied to a case-cohort sample of age at onset for melanoma from the Swedish Multi-Generation Register, organized in nuclear families of parents and one or two children. We compare the genetic component of the total frailty variance to the common environmental term, and estimate the effect of birth cohort and gender.
