Multivariate disease mapping of seven prevalent cancers in Iran using a shared component model.

Background: The aim of this study was to model the geographical variation in incidence and risk factors of seven prevalent cancers in Iran.
Methods: The data for cancers of esophagus, stomach, bladder, colorectal, lung, prostate, and female breast along with their risk factors in all 30 provinces of Iran for the year 2007 were included into study. Smoking, overweight, inadequate consumption of fruits and vegetables, socioeconomic status and low physical activity were studied as risk factors. Standardized incidence ratios were estimated using full bayesian model. In addition, the shared component model was used to explore the spatial pattern of the cancers and to estimate the relative importance of their corresponding risk factors.
Results: Fars and the Northwestern provinces were observed as high risk areas and Hormozgan (a Persian Gulf coastal province), Sistan and Baluchestan, South Khorasan, and Kerman provinces located in Southeast were areas of low risk for most cancers. For all five risk factors, larger effects on incidence of the relevant cancers were found in the Northern provinces compared to other areas. Smoking, overweight, inadequate consumption of fruit and vegetable, socioeconomic status, and low physical activity were found to have more effects on incidence of stomach, breast, esophagus, and breast cancers, respectively.
Conclusions: Most of the high risk areas for seven cancers were in accordance with the results for spatial patterns of related risk factors and their relative weights on relevant cancers. The multivariate shared component model of the seven cancers achieves a considerable improvement in terms of Deviance Information Criterion over the individual modeling of diseases.