A prototype for the detection of inconsistencies in hospital databases

Health professionals need both data with quality and tools to manage all the information that could help them in decision making. This study aimed to develop a prototype that could contribute to the identification of data quality problems in clinical and administrative health databases. In addition to the definition of requirements and business rules, used methods involved the development and testing of a prototype. The prototype was tested using a national database containing more than 11 million discharges in Portuguese acute care public hospitals. Several problems and inconsistencies were found with the use of the prototype. For instance, 82,422 episodes (15%) had diagnostic codes that do not exist in the lookup ICD-9-CM table. Results show that this prototype contributes to the detection of data quality problems. In addition, this prototype can be an incentive for the improvement of the architecture of nowadays information systems, highlighting the importance of the developing of mechanisms for the detection and validation of healthcare data.