Validity of a computerised population registry of dementia based on clinical databases

The handling of information through digital media allows innovative approaches for identifying cases of dementia through computerised searches within the clinical databases that include systems for coding diagnoses. The aim of this study was to analyse the validity of a dementia registry in Gipuzkoa based on the administrative and clinical databases existing in the Basque Health Service. This is a descriptive study based on the evaluation of available data sources. First, through review of medical records, the diagnostic validity was evaluated in two samples of cases identified and not identified as dementia. The sensitivity, specificity and positive and negative predictive value of the diagnosis of dementia were measured. Subsequently, the cases of living dementia in December 31, 2016 were searched in the entire Gipuzkoa population to collect sociodemographic and clinical variables. The validation samples included 986 cases and 327 no cases. The calculated sensitivity was 80.2% and the specificity was 99.9%. The negative predictive value was 99.4% and positive value was 95.1%. The cases in Gipuzkoa were 10 551, representing 65% of the cases predicted according to the literature. Antipsychotic medication were taken by a 40% and a 25% of the cases were institutionalised. A registry of dementias based on clinical and administrative databases is valid and feasible. Its main contribution is to show the dimension of dementia in the health system. El manejo de la información mediante soportes digitales permite abordajes innovadores de la identificación de los casos de demencia mediante búsquedas automatizadas en las bases de datos clínicas con sistemas de codificación de los diagnósticos. El objetivo de este trabajo fue analizar la validez de un registro de demencia en Gipuzkoa basado en los sistemas de registro administrativos y clínicos existentes en el Servicio Vasco de Salud. Es un estudio descriptivo basado en la evaluación de las fuentes de datos disponibles. Primero, mediante revisión de historias clínicas se evaluó la validez diagnóstica en dos muestras de casos identificados y no identificados como demencia. Se midió la sensibilidad, especificidad y valor predictivo positivo y negativo del diagnóstico de demencia. Posteriormente se buscaron los casos de demencia vivos a fecha 31 de diciembre de 2016 en toda la población guipuzcoana y se recogieron variables sociodemográficas y clínicas. Las dos muestras de validación incluyeron 986 casos y 327 no casos.