Hajizadeh Nastaran, Baghestani Ahmad Reza, Pourhoseingholi Mohamad Amin, Ashtari Sara, Najafimehr Hadis, Busani Luca, Zali Mohammad Reza
Department of Biostatistics, Faculty of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran.
Physiotherapy Research Centre, Department of Biostatistics, Faculty of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran.
Galen Med J. 2019 Jul 9;8:e1223. doi: 10.31661/gmj.v0i0.1223. eCollection 2019.
Some errors may occur in the disease registry system. One of them is misclassification error in cancer registration. It occurs because some of the patients from deprived provinces travel to their adjacent provinces to receive better healthcare without mentioning their permanent residence. The aim of this study was to re-estimate the incidence of gastric cancer using the Bayesian correction for misclassification across Iranian provinces.
Data of gastric cancer incidence were adapted from the Iranian national cancer registration reports from 2004 to 2008. Bayesian analysis was performed to estimate the misclassification rate with a beta prior distribution for misclassification parameter. Parameters of beta distribution were selected according to the expected coverage of new cancer cases in each medical university of the country.
There was a remarkable misclassification with reference to the registration of cancer cases across the provinces of the country. The average estimated misclassification rate was between 15% and 68%, and higher rates were estimated for more deprived provinces.
Misclassification error reduces the accuracy of the registry data, in turn causing underestimation and overestimation in the assessment of the risk of cancer in different areas. In conclusion, correcting the regional misclassification in cancer registry data is essential for discerning high-risk regions and making plans for cancer control and prevention.
疾病登记系统可能会出现一些错误。其中之一是癌症登记中的错误分类。出现这种情况是因为一些来自贫困省份的患者前往相邻省份接受更好的医疗服务，却未提及他们的常住地。本研究的目的是利用贝叶斯校正法对伊朗各省的错误分类进行重新估计，以得出胃癌的发病率。
胃癌发病率数据取自2004年至2008年伊朗国家癌症登记报告。采用贝叶斯分析方法，以错误分类参数的贝塔先验分布来估计错误分类率。根据该国各医科大学新癌症病例的预期覆盖范围选择贝塔分布的参数。
该国各省癌症病例登记存在明显的错误分类。估计的平均错误分类率在15%至68%之间，贫困程度较高的省份估计的错误分类率更高。
错误分类误差会降低登记数据的准确性，进而导致不同地区癌症风险评估出现低估和高估。总之，校正癌症登记数据中的区域错误分类对于识别高危地区以及制定癌症控制和预防计划至关重要。
