USE OF AN AUTOMATED PROCESS TO SCAN AND INTERPRET MANUALLY-CODED DATA FORMS IN PHARMACOECONOMICS RESEARCH
Author(s)
Reardon G1, Mozaffari E2, 1RxI Associates, Worthington, OH, USA; 2Pharmacia Corporation, Kalamazoo, MI, USA
OBJECTIVE: The reliability of a computer program to automatically scan and interpret data from manually-coded, data-collection forms was evaluated in a resource-utilization study of patients undergoing total knee-replacement surgery and receiving low molecular-weight heparin therapy to prevent deep vein thrombosis. METHODS: Detailed in-patient data was extracted by technicians via manual chart review and recorded on 13-page forms. To expedite the analysis, Teleform 6.2 was used to design the data-collection form and later, to automatically scan and interpret each form returned. RESULTS: The reliability of this technology was evaluated for the first 100 eligible-patient forms received. Of all field entries, Teleform automatically processed 89.9% without an operator verification prompt. The remaining ambiguous fields were viewed and edited on screen by an operator. Data for each completed form was then forwarded to an output data set. The output data-set records were compared against screen images of the original data forms to determine the reliability of this process. An average of 183 field entries was recorded for each patient and scanned by Teleform. 99.0% of the fields in the output data set matched completely with the handwritten field entries on the original forms. This reliability varied slightly with the type of field analyzed. Check box, date, numeric, and alphanumeric fields had reliability scores of 100.0%, 99.8%, 99.6% and 96.5%, respectively. CONCLUSION: This program shows promise for organizations that routinely collect information on hard-copy forms, though the required learning curve and cost hampers its value for the occasional user or for small projects.
Conference/Value in Health Info
2001-11, ISPOR Europe 2001, Cannes, France
Value in Health, Vol. 4, No. 6 (November/December 2001)
Code
PMI9
Topic
Real World Data & Information Systems
Topic Subcategory
Health & Insurance Records Systems
Disease
Multiple Diseases