Istituto di Fisiologia Clinica     
Colasante E., Molinaro S., Mariani F. Spatial analysis of drug-related hospital admission: an auto Gaussian model to estimate the hospitalization rates in Italy. In: The Italian Journal of Public Health, vol. 5 (4) pp. 253 - 260. Prex S.p.A, 2008.
Introduction: The aim of this study is to evaluate, even if partially, how much the drug use phenomenon impacts on the Italian National Heatlh System throughout the estimation at local level (Local Health Unit) of the hospitalization rate caused by substance use and abuse such as opiates, barbiturates-sedativeshypnotics, cocaine and cannabis, and keeping in mind the phenomenon distribution in the space and so the fact that what happens in a specific area depends on what is happening in the neighbourhoods close to it (spatial autocorrelation). Methods: Data from hospital discharge database were provided by the Ministry of Health and an auto-Gaussian model was fitted. The spatial trend can be a function of other explanatory variables or can simply be modeled as a function of spatial location. Both models were fitted and compared using the number of subjects kept in charge by Drug Addiction Services and the number of beds held by hospitals as covariates. Results: Concerning opiates use related hospitalizations, results show areas where the phenomenon was less prominent in 2001 (Lombardy, part of Liguria, Umbria, part of Latium, Campania, Apulia and Sicily). In the following years, the hospitalization rates increased in some areas, such as the north of Apulia, part of Campania and Latium. A dependence of the opiates related hospitalization rates on the rate of subjects kept in charge by the Drug Addiction Services is highlighted. Concerning barbiturates-sedatives-hypnotics consumption, the best model is the one without covariates and estimated hospitalization rates are lower then 3 per thousand. The model with only the covariate "rate of subjects kept in charge by Drug Addiction Services" has been used both for cocaine and cannabis. In these two cases, more than a half of the Local Health Units report hospitalization rates lower than 0.5 per thousand. Conclusions: This study has allowed for the development of an indirect indicator for the phenomenon of drug use and it constitutes an effort in answering specific needs for the planning of health policies related to a field that, given the specificity of the phenomenon, is often difficult to detect and quantify by means of more common data analysis techniques. Moreover, it is important to highlight that, being this study a first attempt in applying this statistical methodology to data regarding drug addiction, it needs to be further improved.
Subject Hspitalization rate
Dug addiction
Markov random field
Gaussian automodel
Satial dependence

Icona documento 1) Download Document PDF

Icona documento Open access Icona documento Restricted Icona documento Private


Per ulteriori informazioni, contattare: Librarian http://puma.isti.cnr.it

Valid HTML 4.0 Transitional