Bayesian adjustment for over-estimation and under-estimation of gastric cancer incidence across Iranian provinces

AIM To correct the misclassification in registered gastric cancer incidence across Iranian provinces in cancer registry data. METHODS Gastric cancer data is extracted from Iranian annual of national cancer registration report 2008. A Bayesian method with beta prior is implemented to estimate the rate of misclassification in registering patient's permanent residence in neighboring province. Each time two neighboring provinces with lower and higher than 100% expected coverage of cancer cases are selected to be entered in the model. The expected coverage of cancerous patient is reported by medical university of each province. It is assumed that some cancer cases from a province with a lower than 100% expected coverage are registered in their neighboring province with more than 100% expected coverage. RESULTS The condition was true for 21 provinces from a total of 30 provinces of Iran. It was estimated that 43% of gastric cancer cases of North and South Khorasan provinces in north-east of Iran was registered in Razavi Khorasan as the neighboring facilitate province; also 72% misclassification was estimated between Sistan and balochestan province and Razavi Khorasan. The misclassification rate was estimated to be 36% between West Azerbaijan province and East Azerbaijan province, 21% between Ardebil province and East Azerbaijan, 63% between Hormozgan province and Fars province, 8% between Chaharmahal and bakhtyari province and Isfahan province, 8% between Kogiloye and boyerahmad province and Isfahan, 43% Golestan province and Mazandaran province, 54% between Bushehr province and Khozestan province, 26% between Ilam province and Khuzestan province, 32% between Qazvin province and Tehran province (capital of Iran), 43% between Markazi province and Tehran, and 37% between Qom province and Tehran. CONCLUSION Policy makers should consider the regional misclassification in the time of programming for cancer control, prevention and resource allocation.