Academic journal article Central European Journal of Public Health

Comparing Models of the Effect of Air Pollutants on Hospital Admissions and Symptoms for Chronic Obstructive Pulmonary Disease

Academic journal article Central European Journal of Public Health

Comparing Models of the Effect of Air Pollutants on Hospital Admissions and Symptoms for Chronic Obstructive Pulmonary Disease

Article excerpt

SUMMARY

There is an increasing interest in the use of hospital admission for Chronic obstructive pulmonary disease (COPD) in studies of short-term exposure effects attributed to air pollutants. However, little is known about the effect of air pollutants on COPD symptoms. This study was undertaken to determine whether there was an association between air pollutant levels and both hospital admissions and symptoms for COPD. For model comparison, we present Generalized Linear Model, Generalized Additive Model and a general approach for Bayesian inference via Markov chain Monte Carlo in generalized additive model. Furthermore, for comparing the predictive accuracy, Artificial Neural Networks (ANN) approach is given.

Key words: chronic obstructive pulmonary disease, Generalized Additive Model, Bayesian, WinBUGS, hospital admission, air pollution

(ProQuest: ... denotes formulae omitted.)

INTRODUCTION

Chronic obstructive pulmonary disease (COPD) is a group of diseases characterized by airflow obstruction that can be associated with breathing-related symptoms (e.g., cough, exertional dyspnea, expectoration, and wheeze). There is an increasing interest in the use of hospital admission data in studies of shortterm exposure effects attributed to air pollutants. Considerable attention has been paid to vulnerable individuals such as subjects suffering from a chronic obstructive pulmonary disease (COPD). Numerous studies have investigated the relationship between air pollution and hospital admissions for COPD (1-18).

Generalised Additive Model (GAM) (19) has become the most widely used method for assessing the short-term health effects of air pollution. GAM models provide a flexible alternative to parametric regression models. GAM provides a powerful class of models for modelling nonlinear effects of continuous covariates in regression models with non-Gaussian responses. A huge variety of competing approaches are now available for modelling and estimating nonlinear functions of continuous covariates. Prominent examples are smoothing splines (20), local polynomials (21), regression splines with adaptive knot selection (22-24) and P-splines (25, 26). Currently, smoothing based on mixed model representations of GAMs and extensions is extremely popular (27-30). Schwartz and Marcus focuse on GAM for model selection in multiple Poisson regression for modelling associations between air pollution and increases in hospital admissions for respiratory disease (7).

The Bayesian inference for generalized additive model enjoys the flexibility of nonparametric models and the exact inference provided by the Bayesian inferential machinery. It is this combination that makes Bayesian nonparametric modelling so attractive (30, 31). Bayesian approaches are currently either based on regression splines with adaptive knot selection (32-37) or on smoothness priors (20, 38, 39). Crainiceanu et al. provide a simple set of programmes for the implementation of nonparametric Bayesian analysis in WinBUGS using Penalized Spline Regression (40). Brezger and Lang provide Bayesian semiparametric regression based on smoothness priors and Markov chain Monte Carlo (MCMC) simulation techniques (41).

While adverse effects of exposure to air pollutants and hospital admissions for COPD are well studied, little is known about the effect of air pollutants on COPD symptoms. This study focuses on modelling air pollution and both symptoms and hospital admissions (assuming that consecutive outcomes are independent) for COPD to the Afyon Respiratory Disease Hospital between 1 October 2007-30 September 2009.

The goal of this paper is not to discuss generalized additive model, Bayesian methodology, or provide novel modelling techniques. Firstly, we compare GLM (Multiple Poisson Regression), GAM and GAM with a Bayesian approach in WinBUGS (42), which has become the standard software for Bayesian analysis for modelling the association between air pollution, and both symptoms and hospital admissions for COPD using Akaiki Information Criteria (AIC), Bayesian Information Criteria (BIC), and Deviance Information Criteria (DIC). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.