Colombani F, Pereira E, Bettaieb J, Gobin L, Cowppli-Bony A, Hoppe S, Coureau G, Picat M Q, Salamon R, Monnereau A, Savès M
Registre général des cancers de la Gironde, université Bordeaux-Segalen, institut de santé publique, d'épidémiologie et de développement, 146, rue Léo-Saignat, 33076 Bordeaux cedex, France.
Rev Epidemiol Sante Publique. 2013 Feb;61(1):1-9. doi: 10.1016/j.respe.2012.04.006. Epub 2013 Jan 30.
Hospital-based cancer registries (HBR) of Comprehensive cancer centers (CCC) named "Enquête permanente cancer (EPC)" could represent an important information source of data for a population-based cancer registry (PBR). The main purpose of this study was to evaluate the concordance between data from HBR of the CCC for regions of Bordeaux and South-Western France and data from PBR of the General cancer registry of Gironde and the completeness for the year 2005, in order to explore the possibility for immediate integration of data collected by the HBR in the database of the PBR and to reduce the systematic consultation of medical records for cancer cases notified by the HBR.
We estimated the proportion of concordant cases between the HBR and the PBR and the unweighted and weighted Cohen's Kappa coefficient for certain variables.
Among 1151 cases notified by the HBR, 1095 (positive predictive value of HBR: 95%) were included in the PBR database. For these admissible cases, sex, birth date and place for residence postal code at diagnosis were concordant in more than 94%. Concordance of cancer incidence date was 50% but difference between two dates was lower than 1 month in 91% of cases. Concordance of cancer topography and morphology was respectively 72% and 78%. Cancer progression, estimated by TNM classification was concordant in 96% of cases. The Kappa coefficient was 0.95 for pT, and respectively 0.92 for pN and M.
Automatic data integration from the HBR database to the PBR database without consulting medical record will be possible in Gironde. But, topography and morphology will be systematically recoded using pathologist records. And, if cancer progression (coded by TNM classification) is missing in the HBR database, medical records will be consulted in order to complete this data in the PBR database.
名为“癌症永久调查(EPC)”的综合癌症中心(CCC)的医院癌症登记处(HBR)可能是基于人群的癌症登记处(PBR)的重要数据信息来源。本研究的主要目的是评估波尔多和法国西南部地区CCC的HBR数据与吉伦特省综合癌症登记处的PBR数据之间的一致性以及2005年数据的完整性,以探讨将HBR收集的数据立即整合到PBR数据库中的可能性,并减少对HBR通报的癌症病例进行病历系统查询的情况。
我们估计了HBR和PBR之间一致病例的比例以及某些变量的未加权和加权科恩卡帕系数。
在HBR通报的1151例病例中,1095例(HBR的阳性预测值:95%)被纳入PBR数据库。对于这些可接受的病例,性别、出生日期和诊断时居住邮政编码所在地的一致性超过94%。癌症发病日期的一致性为50%,但在91%的病例中两个日期之间的差异小于1个月。癌症部位和形态的一致性分别为72%和78%。通过TNM分类估计的癌症进展在96%的病例中是一致的。pT的卡帕系数为0.95,pN和M的卡帕系数分别为0.92。
在吉伦特省,无需查阅病历即可将HBR数据库中的数据自动整合到PBR数据库中。但是,部位和形态将使用病理学家的记录进行系统重新编码。并且,如果HBR数据库中缺少癌症进展(按TNM分类编码),将查阅病历以在PBR数据库中完善此数据。