The Analysis of Rates Using Poisson Regression Models

Abstract

Models are considered in which the underlying rate at which events occur can be represented by a regression function that describes the relation between the predictor variables and the unknown parameters. Estimates of the parameters can be obtained by means of iteratively reweighted least squares (IRLS). When the events of interest follow the Poisson distribution, the IRLS algorithm is equivalent to using the method of scoring to obtain maximum likelihood (ML) estimates. The general Poisson regression models include log-linear, quasilinear and intrinsically nonlinear models. The approach considered enables one to concentrate on describing the relation between the dependent variable and the predictor variables through the regression model. Standard statistical packages that support IRLS can then be used to obtain ML estimates, their asymptotic covariance matrix, and diagnostic measures that can be used to aid the analyst in detecting outlying responses and extreme points in the model space. Applications of these methods to epidemiologic follow-up studies with the data organized into a life-table type of format are discussed. The method is illustrated by using a nonlinear model, derived from the multistage theory of carcinogenesis, to analyze lung cancer death rates among British physicians who were regular cigarette smokers.

Understanding the Effects of Age, Period, and Cohort on Incidence and Mortality Rates

“…The emphasis of this discussion is the interpretation of results from fitting regression models to data that include age, birth cohort, and period or year of diagnosis. The details of model fitting are described elsewhere (1,15,16,19,27), and no specialized programs are necessary. Unfortunately, the results of these analyses can be confusing, and indeed contradictory.…”

Methods of Mapping and Identifying Small Clusters of Rare Diseases with Applications to Geographical Epidemiology

“…These techniques are familiar in many disease atlases, for example, Kemp et al 1985; Gardner et al 1983; Smans, Boyle, and Muir 1990. For rare diseases there is the problem that fluctuations marked on a map may represent merely random variation and appropriate methodology has been developed: empirical Bayes estimates of the standard ratios (Clayton and Kaldor 1987) and Poisson regression modeling to test for heterogeneity (Frome 1983). In addition, adjustment for spatial autocorrelation can be incorporated using modeling (Clayton and Kaldor 1987;Grifith 1988;Pocock et al 1980) and cluster indices computed (Kemp et al 1985; Abel and Becker 1987).…”

A specialist leukaemia/lymphoma registry in the UK. Part 1: incidence and geographical distribution of Hodgkin's disease

“…This process involves a large number of statistical tests; the P values should therefore be interpreted with caution and are referred to as 'nominal P values'. Global testing of differences of O/E ratios by area unit are also based on the Poisson distribution; the method is that of Poisson regression (Frome, 1983), using GENSTAT. It should be noted that an explicit comparison was made of OIE ratios with E calculated as above (i.e.…”