Merging multiple hospital databases into a single shared database

Abstract

This work concerns the merging of eight hospital databases in the Finnish Ostrobothnia region. Each of them used their own database solution, making it hard for them to analyze data between hospitals. The purpose of this project is to make it possible for them to compare and analyze the data of all participating hospitals, making it possible to find new useful datapoints, and making it easier to find anomalies that previously would have been almost impossible to find when constrained to a single hospitals data. The work is needed because before this, no cross hospital analyzes were possible on a larger scale, and patient trends in the region were constrained to each hospital, making some of them more obfuscated than necessary, and the participating hospitals determined that this was something they all wanted to improve. Due to GDPR and patient rights, no patient identifying information was included in the scope of data, and restrictions had to be made on which people had right to see which data. The project was completed by first determining the scope of what database data that should be included on the table and column level for each hospital. Once that has been determined, the chosen data is gathered using SQL fetches, modified to a homogenous format using Python scripts and extracted to csv files. The resulting files are then sent to the shared server over SFTP and saved to the shared database using Python. After the data has been transferred to the database, then the Neotides reporting program is integrated to work with this merged data. Once the work was done, analysis of the participating hospitals patient data as a whole became possible, and patient flows between hospitals became possible to follow. This made it possible to analyze long term effects of procedures with greater accuracy, since the data was no longer constrained to patients revisiting the same hospital for follow up procedures and complications. The hospitals could also analyze their strengths and weaknesses compared to one another much easier and more accurately than before. These results among other things makes it possible for hospital management to improve their hospitals performance both for quality of care as well as long term cost per patient. To make even more precise analysis possible, the hospitals would have to work together to standardize their data collection practices in some way, making the starting point more homogenous. This would also make it easier for the hospitals to sometime in the future share visit data and other data with each other more easily for more than just management level reports and analyzes. Further work could also include increasing the scope of data to collect, and inviting more hospitals to be a part of it.