Neoplasms are the second most common mortality cause in Brazil, together with injuries and accidents. Amongst the neoplasms, lung cancer is one of most frequent in men and in women, and it is also one of the most lethal. Moreover, the attributed risk of tobacco as an etiologic agent of this type of cancer is considerably high, making it potentially susceptible to preventive public health measures. The objective of this work was to analyze spatial and temporal patterns of lung cancer in four Brazilian States (Rio Grande do Sul, Santa Catarina, Paraná and São Paulo), from 1996 to 2000. The observed values were obtained from the Brazilian Ministry of Health Mortality Information System (SIM-MS). The expected values were calculated using indirect standardization according to sex and age groups (SMR). The geographic units analyzed were micro-regions defined by the Brazilian Census Bureau (IBGE). A Bayesian model that allowed for space-time interaction was used, using WinBUGS software. Adjusted SMR showed that in Southern Brazil there was a "U" pattern in lung cancer mortality for men, and indicated specific areas that had higher SMR and/or greater SMR growth. The main hypothesis for this result would be different tobacco use incidences, but the absence of this information at a regional level in Brazil did not permit including this variable in our analysis. The results of this paper can be used to guide public policies directed towards reducing the use of tobacco and of mortality from lung cancer.
Lung cancer; Mortality; Epidemiology; Spatial analysis; Temporal analysis; Bayesian methods
