Aims/hypothesis The aim of the study was to investigate the relationships between childhood type 1 diabetes and socioeconomic conditions, which might provide clues to the aetiology of the disease. Materials and methodsIn an ecological study, we investigated the relationships between socioeconomic conditions and the incidence of type 1 diabetes incidence among children aged 0-14 years in North Rhine-Westphalia (NRW), Germany, between 1996 and 2000 at the level of the 33 districts. Incidence data were obtained from the population-based NRW diabetes register and regional socioeconomic data from official statistics. Associations were assessed by Poisson regression models and Bayesian conditionally autoregressive regression models (CAR). Results In simple Poisson regression, population density, proportion of non-German nationals in the population, measures of income, education and professional training, and deprivation scores were significantly associated with diabetes risk (p<0.01). An increase of about one interquartile range (IQR) in population density, proportion of non-German nationals or household income was associated with a 9-12% decrease in diabetes incidence. A rise of about one IQR in income ratio, measures of education and professional training, or in deprivation score (high values correspond to high deprivation) was associated with an 8-12% incidence increase. There was a significantly 'linear' increasing incidence trend across five deprivation classes (relative risk: 1.06; 95% CI: 1.03-1.09). All associations were confirmed when overdispersion and spatial autocorrelation were accounted for in Poisson and CAR models. Conclusions/interpretationsThe results raise the possibility that the risk for type 1 diabetes is higher for children living in socially deprived and less densely populated areas. Subsequent investigations are necessary to verify the observed ecological relations at the individual level and to identify the causal factors behind the socioeconomic indicators. Diabetologia (2007) 50:720-728
In many practical situations, simple regression models suffer from the fact, that the dependence of responses on covariates can not be sufficiently described by a purely parametric predictor. Moreover observations may be spatially or temporarily correlated and unobserved heterogeneity between individuals or units may be present. We propose a general class of structured additive regression models (STAR) allowing for a flexible semiparametric predictor. STAR models cover a number of well known model classes as special cases, including generalized additive mixed models, geoadditive models, varying coefficient models and geographically weighted regression. Non-linear effects of continuous covariates and time trends are modelled through Bayesian versions of penalized splines, random walks or flexible seasonal components. Correlated spatial effects can be estimated based on Markov random fields or two dimensional penalized splines. I.i.d. Gaussian effects are used to deal with cluster specific unobserved heterogeneity. Our Bayesian approach allows to treat all functions and effects within a unified general framework by assigning appropriate priors with different forms and degrees of smoothness. Inference is performed on the basis of a generalized linear mixed model representation. This can be viewed as posterior mode estimation and is closely related to penalized likelihood estimation in a frequentist setting. Variance components, corresponding to inverse smoothing parameters, are then estimated by using marginal likelihood. Numerically efficient algorithms allow the computation of these estimates even for fairly large data sets. As a typical example we present results on an analysis of data from a forest health survey: Each year the damage state of a population of trees is measured as a binary response, and the site of each tree is available on a lattice map. Continuous covariates are the age of the tree, canopy density at the stand and calendar time.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.