http://lattes.cnpq.br/9978011179725606; AZEVEDO, Jullianna Vitório Vieira de.
Resumo:
This work aimed to evaluate the effects of seasonal climatic variations in the incidence of respiratory diseases by influenza (PI) in the elderly population in the cities of Campina Grande, metropolitan region of João Pessoa (RMJP), Monteiro and Patos. Generalized linear models from the linear Poisson regression to relate the dependent variable set to the records of hospitalizations for causes associated with influenza and the independent variables (rainfall, average air temperature and relative humidity) to analyze the relations established by modeling has been used. Aditionally, was applied ANOVA variance test with a significance level of 0.05 of probability to determine which independent variables is more significant. Also the residual generated by adjusting the models in order to identify the distribution that best fitted the data were analyzed. The nonparametric Mann-Kendall trend analysis for the time series of hospitalizations for causes associated with influenza as well as the unit root test Dick-Fuller (DF) for stationary analysis was applied. Once determined the time series characteristics was applied to the methodology Box and Jenkins (1976), was used in this study ARIMA model and evaluation of the autoregressive models generated applied to the punitive indices AIC (Akaike's Information Criterion) and BIC (Bayesian Information Criterion). All static analysis was performed using R software. In general is possible to identify that the highest peaks of hospitalizations for PI occur in autumn and winter. Therefore, these results suggest an association between the cold and hospitalizations for IP. In most municipalities studied, the increase in morbidity rates for influenza and associated causes aged 65 and over show a possible lack of effect of vaccination campaigns. The statistical model is presented as an alternative in the analysis and prediction of cases of hospitalization due to IP, contributing to public policy, helping in decision-making avoiding economic and human losses.