Please select Print from the file menu or right mouse click and choose print from the menu to print this page.

[1561] Variation in Diagnostic Immunohistochemistry (DIHC) Testing within a Multi-Hospital System

D Kavalieratos, DM Grzybicki, J Ho, SS Raab. University of Pittsburgh, Pittsburgh, PA

Background: Wennberg and others have reported small geographic area variation in a number of clinical practices, such as operations performed and tests ordered. This variation exists even when patient populations are similar, indicating suboptimal utilization of clinical services. Our goal was to determine the frequency of DIHC test ordering in a multi-hospital system composed of different practice types.
Design: We used a retrospective review design to examine DIHC ordering practices in a hospital system composed of 12 unique pathology groups (4 academic and 8 community-based) for the 2004 calendar year. Overall, 196,777 cases were accessioned. Proportions of total cases examined with DIHC performed and average numbers of stains per case were calculated. The Chi-square test was used to examine differences in DIHC ordering practices by group setting, individual institution, pathologist, specimen type, anatomic site, and test panel. The correlation coefficient was used to examine relationships between inter- and intra- institutional diagnostic malignancy rates and case mix (based on current procedural terminology [CPT] codes) and number of DIHC stains.
Results: The frequency of DIHC use was hospital and practice type dependent (P < 0.001) with academic and community hospitals ordering DIHC in 10.4% and 2.0% of cases, respectively. DIHC ordering within these groups was highly variable with one academic hospital ordering DIHC in 18.5% of cases and several community hospitals ordering DIHC similarly to academic hospitals. The mean number of stains per case was variable across academic hospitals (range: 2.7 to 5.8 stains) and community hospitals (range: 0 to 11.9 stains) (P < 0.001). The hospital frequency of DIHC use correlated with malignancy frequency and case mix; however case mix and/or malignancy frequency could not fully account for variable interinstitutional ordering patterns. For example, academic and community hospitals ordered DIHC in 29% and 19% of prostate needle biopsy cases (P < 0.001) respectively, even though the proportion of malignant cases was similar. Pathology groups ordered different DIHC panels for the same scenario; for example, to diagnose metastatic melanoma, one hospital ordered AE1/3, S100 and HMB45 and a second hospital ordered MelanA and S100.
Conclusions: We found high variation in DIHC ordering practices across a diverse group of hospitals in the same medical system. This variability indicates a lack of standardization and suggests the existence of suboptimal use in a yet undetermined portion of cases.
Category: Quality Assurance

Tuesday, March 27, 2007

Poster # 185, Tuesday Afternoon