The aim of this application is to develop software for estimation in generalized linear mixed models (GLMMs). This general class of models have important uses in macro-epidemiology, including (a) the estimation of disease maps, (b) the study of """"""""ecological"""""""" relationships between environment and disease using data points derived from regions, towns or small areas, (c) the study of temporal trends in incidence and mortality rates using age-period-cohort models and more general interaction models, (d) the forecasting of future rates, and (e) the joint dependence of disease rates on spacial location and on time. The models also have important uses in analytical studies of the complex multi-level explanatory variables commonly encountered in occupational studies and genetic epidemiology. Application of these methods has been hindered by their computational intractability. Recent developments of Monte Carlo techniques for inference have provided invaluable yardsticks for assessing less computationally intensive approximation methods.
We aim to develop, and to make freely available, software for both of these approaches. We also aim to collect a series of illustrative examples of their use in cancer epidemiology and to prepare and deliver a short course on statistical methods in macro-epidemiology whose aim would be to train epidemiologists in the uses of GLMMs in the areas of study outlined above.
| Ecochard, R; Clayton, D G (1998) Multi-level modelling of conception in artificial insemination by donor. Stat Med 17:1137-56 |
| Songini, M; Bernardinelli, L; Clayton, D et al. (1998) The Sardinian IDDM study: 1. Epidemiology and geographical distribution of IDDM in Sardinia during 1989 to 1994. Diabetologia 41:221-7 |
