Limitations of current health information systems

Objectives: To explore the potential for using hospital patient data collected from a number of operational information systems and aggregated into central databases for knowledge discovery purposes. Of particular interest was its potential to provide information that would enable the hospital to explore the relationships between nursing skill mix, processes of care and patient outcomes relative to AR-DRG or ICD-10-AM patient categories. Methods: Relevant information systems in use such as the HBCIS, Transition II and routinely collected patient outcomes data, available data definitions, database structures, information systems design and data retrieval methods were evaluated for the purpose of their ability to extract the desired data in the right format to answer the research questions posed. The system of interest (Transition II) is a large complex oracle seven relational database containing 564 Tables and 6450 data columns. We concentrated only on the variables describing nursing process factors, patient demographics and characteristics and outcomes. These data were downloaded with the view to explore the use of data mining techniques specifically to support this research. This paper has focused on the information systems in use and problems associated with their data retrieval. Results: The downloaded data could not be reliably identified as some of these data were not extractable from their primary source and had previously been processed. Nor was it possible to relate data from the main system with data from other systems as there was no mechanism, such as standard patient identifiers, to reliably relate such data on a per patient basis. Transition II does not meet basic data warehouse requirements and thus it was not possible to explore the use of a number of data mining techniques for knowledge discovery purposes. Crystal, the report writer in use with this system provides routine reports and can also be used to extract data and answer any number of research questions, however the user needs to have a solid understanding of the data elements available and the data base/system structures to do this in an informed and reliable manner. Conclusions: Many of our legacy systems only serve the purpose for which they were designed and limit additional usage. They collect and store an enormous amount of potentially useful data but are inflexible. Attempts at reliable data extraction for a variety of other purposes such as decision support, practice evaluation or knowledge discovery revealed many limitations and shortcomings. In addition users need extensive training to make good use of the available data from such systems. This research has highlighted the need for optimum system design to suit multiple uses from operational data collected throughout an organisation. Database structures and data dictionaries need to be well documented to enable users to realise many potential benefits of such large and costly information systems.