2005 Scientific Session Abstracts
Error Reduction, Efficiency Improvement, and Cost Savings Through Automated Data Entry
J. Mark Tuthill (mtuthil1@hfhs.org), Paul Tranchida, Richard Zarbo, Ruan Varney, David Chen, Department of Pathology, Henry Ford Medical Center, Detroit, MI
Content: Data collection is a part of the workflow in all pathology departments. Three major problems in the gathering of data are the introduction of error during collection, the inefficiency of visual recognition and manual documentation of the data, and the cost. In this project we sought to address these issues by implementing digital scanning and optical recognition technology to extract data from paper. There were three scenarios where we applied this solution: (1) User surveys involving employees of the pathology department; (2) Clinical information data gathering for research studies; and (3) Capturing patient identification from surgical pathology requisition forms.
Technology: Form scanning workstation consisting of a personal computer running Windows 2000 connected to a high-speed scanner. The software application Cardiff TELEform Enterprise with EForms (Provided by Scantron Corporation, Irvine, CA) was installed on a database server running Windows 2003 Server, and Microsoft SQL. Web pages for user access were hosted on a similar server.
Design:
Surveys : Using TELEform Designer we produced a scannable version of our customer satisfaction and employee computer proficiency surveys.
Pathology Error Grant:Two scannable forms were created to document the clinical information for each patient studied. Each form had approximately eighty fields distributed over three pages, with each page containing a "Page Link" field which ensured correct association of each page's data to the appropriate patient regardless of scanning order.
Clinical Requisitions: The scanning program was configured to recognize the existing paper requisition forms. The application was programmed to locate and interpret a bar code label placed by pathology staff containing the patient identification and accession number as well as an area where clinical staff entered demographic information. The scanning software crosschecked the pathology and clinical information, and on validation exports case information plus linked to the document images.
Results: For each form, the information was read using a combination of optical character recognition of user handwriting as well as interpretation of markings indicating user choice in multiple-choice fields. Preliminary impressions of using automated data entry with the hardware/software setup described above confirm its utility saving time and improving staff efficiency, the accuracy of data collected and decreased cost of collection due to time savings.
Conclusion: Automated data collection using optical character recognition can help to reduce the error rate and inefficiency of collecting and validating information from users in a variety of settings that are frequently encountered in a pathology department.