2006 Scientific Session Abstracts
Scientific Session Presentations & Schedule
E-Poster Presentations
Submission Compliance and Guidelines
Copyright Transfer Form (PDF)
Abstract Archive
Exploring Alternative Database Sources for Cancer Disease Identification
Michael Graiser, PhD1, (michael.graiser@emoryhealthcare.org); Vasili Egnatashvili, MD2, Rochelle Victor1, Christopher Flowers, MD1, David Kooby, MD2 1 Winship Cancer Institute, Department of Hematology and Oncology, Emory University School of Medicine, Atlanta, GA; 2 Department of Surgery, Emory University School of Medicine, Atlanta, GA
Content: Data from retrospective analysis is often necessary for planning prospective studies. Capturing appropriate patient populations to perform retrospective analysis can be difficult. Since 2002 the Winship Cancer Institute utilized a linked database application to perform cancer-related searches. Identifiers including ICD-9 diagnosis codes, ICD-O site and histology codes and text searches from electronic medical records were studied. Queries were analyzed to examine the data sources that would aid in identifying all Emory patients with hepatocellular carcinoma (HCC).
Technology: GeneSys SI (GSI) is a Java and web-based oncology database application, which receives daily updates with patient diagnosis and treatment data from the Emory Hospital and Clinic patient administrative systems, lab, pharmacy, Clinical Trials, Cancer Registry and electronic medical record reports. The MRS Cancer Registry system (IMPAC Medical Systems, Inc., Cambridge, Massachusetts) is Emory’s Cancer Registry application. EeMR is Emory Healthcare’s electronic medical records application (Cerner Corporation, Kansas City, Missouri). TAPIOCA is an Emory-developed web-based system storing pathology reports which are searchable using the ht://Dig WWW search engine software (Free Software Foundation, Boston, Massachusetts).
Design: GSI was used to identify HCC patients utilizing Cancer Registry ICD-O site and histology codes, ICD9 diagnosis codes from The Emory Clinic and Emory Hospital and free-text searches of pathology and other medical record reports. The MRS application was queried to find HCC patients based on ICD-O histology codes. TAPIOCA was used to find HCC patients using the text string “hepatocellular carcinoma”. Disease verification was conducted by examining electronic medical records in GSI, EeMR or TAPIOCA.
Results: HCC studies with 17 GSI queries found 1731 patients of whom 479 were verified as HCC+. The query run from the MRS Cancer Registry database returned 740 patients of whom 572 were HCC-verified from EeMR and another 75 verified from TAPIOCA (total 647 HCC+). The TAPIOCA text search found 1011 pathology reports as follows: 707 with a single diagnosis of HCC, 71 with a differential diagnosis including HCC, 68 with a clinic history of HCC and 165 negative for HCC. The Cancer Registry query identified an additional 341 HCC+ patients after which an additional 246 HCC+ patients were identified from TAPIOCA.
Conclusion: This study confirms our previous finding that ICD-O histology codes and free-text searches of pathology reports are highly effective sources for identifying cancer patients. Future endeavors at improving diagnosis identification strategies are necessary to identify pure populations of patients for retrospective evaluation. These may include enhancing search capabilities of electronic pathology reports.