Integrating Brazilian health information systems in order to support the building of data warehouses

AbstractIntroductionThis paper's aim is to develop a data warehouse from the integration of the files of three Brazilian health information systems concerned with the production of ambulatory and hospital procedures for cancer care, and cancer mortality. These systems do not have a unique patient identification, which makes their integration difficult even within a single system.MethodsData from the Brazilian Public Hospital Information System (SIH-SUS), the Oncology Module for the Outpatient Information System (APAC-ONCO) and the Mortality Information System (SIM) for the State of Rio de Janeiro, in the period from January 2000 to December 2004 were used. Each of the systems has the monthly data production compiled in dbase files (dbf). All the files pertaining to the same system were then read into a corresponding table in a MySQL Server 5.1. The SIH-SUS and APAC-ONCO tables were linked internally and with one another through record linkage methods. The APAC-ONCO table was linked to the SIM table. Afterwards a data warehouse was built using Pentaho and the MySQL database management system.ResultsThe sensitivities and specificities of the linkage processes were above 95% and close to 100% respectively. The data warehouse provided several analytical views that are accessed through the Pentaho Schema Workbench.ConclusionThis study presented a proposal for the integration of Brazilian Health Systems to support the building of data warehouses and provide information beyond those currently available with the individual systems.

摘要
本研究旨在构建数据仓库，整合巴西三套医疗信息系统的相关数据，这些系统分别覆盖癌症诊疗的门诊与住院服务产出以及癌症死亡数据。由于上述三套系统均未配置统一的患者标识，即便在单个系统内部开展数据整合也面临较大阻碍。

研究方法：本研究使用了2000年1月至2004年12月期间，巴西里约热内卢州的公立住院医疗信息系统（SIH-SUS）、门诊信息系统肿瘤学模块（APAC-ONCO）以及死亡信息系统（SIM）的数据。各系统的月度数据均以dbase文件（dbf）格式汇总存储。随后，将隶属于同一系统的所有文件导入MySQL Server 5.1数据库的对应数据表中。通过记录链接（record linkage）技术，实现了SIH-SUS与APAC-ONCO数据表的内部关联以及二者间的跨表关联，并将APAC-ONCO数据表与SIM数据表完成关联。最终，借助Pentaho工具与MySQL数据库管理系统构建了目标数据仓库。

研究结果：本次记录链接流程的敏感度与特异度均超过95%，分别接近100%。该数据仓库提供了多维度分析视图，可通过Pentaho Schema Workbench进行访问。

研究结论：本研究提出了一套巴西医疗系统数据整合方案，可为数据仓库的构建提供技术支撑，并能获取单系统无法独立提供的整合式信息。