Full text
| |
| Author(s): |
Ana Carolina Cintra Nunes Mafra
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Faculdade de Ciências Médicas |
| Defense date: | 2011-06-21 |
| Examining board members: |
Ricardo Carlos Cordeiro;
Claudia Torres Codeço;
Júlio de Motta Singer;
Djalma de Carvalho Moreira Filho;
Ronaldo Dias
|
| Advisor: | Ricardo Carlos Cordeiro |
| Abstract | |
The search for understanding some epidemiological phenomena often involves an tool called spatial analysis of risk. The study of space in which certain outcomes occur allows the researcher to consider information that can not be collected through questionnaires or medical records. It also puts questions about what makes a certain area within the study region was associated with greater risk or protection for the outcome studied. Many techniques are used for this kind of study as the generalized additive models that fit the spatial analysis of the risk with others informations of interest. But now, epidemiological studies that consider the spatial distribution of risk are analyzed only with dichotomous outcomes, such as when it classifies the individual in case or control. This is a limitation that this study aims to overcome when presenting an analytical process of the spatial distribution of risk when you have a multinomial response variable. In addition to presenting this new tool, this study analyzed two outcomes: first, from a case-control study of precarious workers in the city of Piracicaba in which the response was: severe cases, mild cases or controls. Another illustration comes from a cross-sectional study on mosquito breeding sites in the Southern District of Campinas, where we met many breeding sites, few or no breeding sites. First, it is necessary a discussion on the appropriateness of each multinomial model to some epidemiological studies. It also discusses the choice of one among several multinomial models and shows the way to interpret the results of the analysis. We present the computational functions for the analytical process to make this method accessible to other researchers (AU) | |