Bayesian analysis of esophageal cancer mortality in the presence of misclassification

Abstract

Background: Esophageal cancer (EC) is one of the most common cancers worldwide. Mortality is a familiar projection that addresses the burden of cancers. With regards to cancer mortality, data are important and used to monitor the effects of screening programs, earlier diagnosis and other prognostic factors. But according to the Iranian death registry, about 20% of death statistics are still recorded in misclassified categories. The aim of this study is to estimate EC mortality in the Iranian population, using a Bayesian approach in order to revise this misclassification.

Methods: We analyzed National death Statistics reported by the Iranian Ministry of Health and Medical Education from 1995 to 2004. ECs [ICD-9; C15] were expressed as annual mortality rates/100,000, overall, by sex, by age group and age standardized rate (ASR). The Bayesian approach was used to correct and account for misclassification effects in Poisson count regression, with a beta prior employed to estimate the mortality rate of EC in age and sex groups.

Results: According to the Bayesian analysis, there were between 20 to 30 percent underreported deaths in mortality records related to EC, and the rate of mortality from EC has increased through recent years.

Conclusions: Our findings suggested a substantial undercount of EC mortality in the Iranian population. So
policy makers who determine research and treatment priorities based on reported death rates should notice of this underreported data.