2006 Scientific Session Abstracts
Scientific Session Presentations & Schedule
E-Poster Presentations
Submission Compliance and Guidelines
Copyright Transfer Form (PDF)
Abstract Archive
Structured Data Entry And Reporting Using Misys Healthcare Copath: A Two Year Experience
J. Mark Tuthill (mtuthil1@hfhs.org), Paul Tranchida, Richard Zarbo, Department of Pathology, Henry Ford Medical Center, Detroit, MI
Content: Structuring the data entered on anatomic pathology reports has long been a goal for clear reporting of pathology diagnoses as well as cancer staging information, gross description, and microscopic case features. This has typically been accomplished by use of standardized templates that are appropriately inserted in the various sections of a diagnostic pathology report and populated with case specific information. This results in communicating clear and consistent text for similar diagnostic entities. However, such templates are still effectively "free text" and do not result in database specific field population for the elements in question or advantages of database technology for queries. Working with Misys Healthcare we helped develop and implement structured data entry and structured data reporting modules that lead to database driven structuring of diagnostic pathology information.
Technology: CoPath 2.3b from Misys Healthcare with structured data reporting module; Sybase database back end; Unix database server; Windows 2000 application servers, CoPath 2.3b client workstations running on windows 2000. CoPath management reports were generated by Power Builder 6.0.
Design: Tables were added to the CoPath database allowing the population of a structured data checklist consisting of structured categories each of which is comprised of specified data elements. The default structured data dictionary elements provided by Misys were replaced with Henry Ford Hospital synoptic reports that have been in use for over a decade. These structured elements are based on a combination of: information from the medical literature, AJCC checklists, CAP checklist, and ADASP reporting standards. This results in a structured data checklist with a series of categories each of which has a series of elements. For example, in the breast cancer staging checklist, the first category is "histopathologic type". The available choices include all accepted histopathologies for breast cancer. A particular checklist (or more than one) is added to a CoPath case through a pick list and associated with the particular "part types" present in that case. As designed, a case can have multiple checklists that relate to case multiple parts, or multiple checklists that are associated with one or another of the parts accessioned. Once added to a case, the associated checklist categories are populated with specific elements through pick lists allowing for structuring of the case data. A comment field for limited variable information such as numbers or dimensions is also provided.
Results:Since implementation of the system in October, 2004, we have structured over 3000 cases. Once the required information is available (usually at case finalization), adding the checklist and populating the required data can be accomplished in less than one minute. This has resulted in pathologists populating this data personally at sign out, as opposed to dictation and transcription as typically done with templates. The checklist must be fully populated for each category prior to sign out. Once entered, structured information can be mined using reports built against the CoPath database and can be tailored to parts and categories of interest. Reports run rapidly and are highly specific as apposed to natural language queries run against free text or SNOMED searches.
Conclusion: Structuring pathology data and creating a simple pick list driven interface has resulted in an easy to use, highly efficient way to enter diagnostic pathology information. Data retrieval from the system is highly specified and brisk with searches for the occurrence of a given element taking three to fifteen seconds across a database of 100, 000 cases versus minutes to hours for natural language queries. As pathologists are entering this information directly, transcription errors have been reduced with significant time savings and process improvement. Further, as each category must be populated with an element, we have nearly eliminated errors of omission in cases where the checklists are used.