The Study of Relative Incidence Risk of Esophageal Cancer in Iranian Counties by the Use of the Bayesian Space-Time Multilevel Model

The age standardized incidence rates of esophageal cancer in women and men in Iran have the fifth and eighth ranks, respectively. Due to the high incidence rate of esophageal cancer and the necessity of knowledge of its geographical distribution, the incidence of this cancer was mapped at province and county levels in the present study.
The study population of this empirical and ecological study included all cases of esophageal cancer from 2005 to 2008 in Iran. The Bayesian Multilevel Space-Time Model was used to measure the relative risk of esophageal cancer. The data consisted of three geographically nested levels in which the countries were nested in the provinces and provinces in time (year). Data analysis was conducted using the Integrated Nested Laplace Approximations (INLA) Bayesian approach in R software.
Findings: The total number of registered esophageal cancer cases in Iran was 7160 from 2005 to 2008. Based on the results of the three-level model, with the adjustment of all risk factors, the annual variation trend was significant and cumulative.
Considering the better fit of the three-level model with the adjustment of all risk factors (physical activity, smoking, fruit and vegetable intake, and overweight), the use of multi-level models with the adjustment of more effective risk factors is recommended in future studies. This investigation showed that Northeast, Northwest, and some areas in Central Iran had the highest risk of esophageal cancer incidence.