Conexión de registros sanitarios: base poblacional de salud de Andalucía / Health record linkage: Andalusian health population database

OBJETIVO: Describir el desarrollo de un sistema de información que conecta datos procedentes de múltiples registros, sanitarios y otros, para su uso con fines asistenciales, de administración, gestión, evaluación, inspección, investigación y salud pública. MÉTODO: Conexión determinística de datos pseudonimizados de una población de 8,5 millones de habitantes, procedentes de Base de datos de usuarios, Historia clínica electrónica DIRAYA, Conjunto mínimo básico de datos (hospitalización, cirugía mayor ambulatoria, urgencias hospitalarias y hospital de día médico) y sistemas de información de salud mental, pruebas de imagen, pruebas analíticas, vacunas, pacientes renales y farmacia. Se utilizó un codificador automático para los diagnósticos clínicos y se definieron 80 enfermedades crónicas para su seguimiento. La arquitectura del sistema de información constó de tres capas: datos (base de datos Oracle 11g), aplicaciones (MicroStrategy BI) y presentación (MicroStrategy Web, librerías JavaScript, HTML 5 y hojas de estilo CSS). Se implantaron medidas para la gobernanza del sistema. RESULTADOS: Se incluyeron datos de 12,5 millones de personas que fueron usuarias entre los años 2001 y 2017, con 435,5 millones de diagnósticos. El 88,7% de estos diagnósticos fueron generados por el codificador automático. Los datos se presentan mediante informes predefinidos o consultas dinámicas, ambos exportables a ficheros CSV para su tratamiento fuera del sistema. Analistas expertos pueden acceder directamente a las bases de datos y realizar extracciones mediante SQL o tratar directamente los datos con herramientas externas. CONCLUSIÓN: El trabajo ha mostrado cómo la conexión de registros sanitarios abre nuevas posibilidades en el análisis de datos

OBJECTIVE: To describe the development of an information system that connects data from multiple health records to improve assistance to patients, health services administration, management, evaluation, and inspection, as well as public health and research. METHOD: Deterministic connection of pseudonymized data from a population of 8.5 million inhabitants provided by: a users database, DIRAYA electronic medical records, minimum basic data sets (inpatients, outpatient mayor surgery, hospital emergencies and medical day hospital), mental health information systems, analytical and image tests, vaccines, renal patients, and pharmacy. An automatic coder was used to code clinical diagnoses and 80 chronic pathologies were identified to follow-up. The architecture of the information system consisted of three layers: data (Oracle Database 11g), applications (MicroStrategy BI) and presentation (MicroStrategy Web, JavaScript libraries, HTML 5 and CSS style sheets). Measures for the governance of the system were implemented. RESULTS: Data from 12.5 million health system users between 2001 and 2017 were gathered, including 435.5 million diagnoses, 88.7% of which were generated by the automatic coder. Data can be accessed through predefined reports or dynamic queries, both exportable to CSV files for processing outside the system. Expert analysts can directly access the databases and perform queries using SQL or directly treat the data with external tools. CONCLUSION: The work has shown that the connection of health records opens new possibilities for data analysis

[Health record linkage: Andalusian health population database]. / Conexión de registros sanitarios: base poblacional de salud de Andalucía.

OBJECTIVE: To describe the development of an information system that connects data from multiple health records to improve assistance to patients, health services administration, management, evaluation, and inspection, as well as public health and research. METHOD: Deterministic connection of pseudonymized data from a population of 8.5 million inhabitants provided by: a users database, DIRAYA electronic medical records, minimum basic data sets (inpatients, outpatient mayor surgery, hospital emergencies and medical day hospital), mental health information systems, analytical and image tests, vaccines, renal patients, and pharmacy. An automatic coder was used to code clinical diagnoses and 80 chronic pathologies were identified to follow-up. The architecture of the information system consisted of three layers: data (Oracle Database 11g), applications (MicroStrategy BI) and presentation (MicroStrategy Web, JavaScript libraries, HTML 5 and CSS style sheets). Measures for the governance of the system were implemented. RESULTS: Data from 12.5 million health system users between 2001 and 2017 were gathered, including 435.5 million diagnoses, 88.7% of which were generated by the automatic coder. Data can be accessed through predefined reports or dynamic queries, both exportable to CSV files for processing outside the system. Expert analysts can directly access the databases and perform queries using SQL or directly treat the data with external tools. CONCLUSION: The work has shown that the connection of health records opens new possibilities for data analysis.