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

[1555] Improving Interobserver Reproducibility in Pap Test and Cervical Biopsy Interpretations

DM Grzybicki, C Jensen, KR Geisinger, JE Janosky, EM Wojcik, CH Stone, CN Booth, G Carter, FA Meier, RJ Zarbo, SS Raab. University of Pittsburgh, Pittsburgh, PA; University of Iowa, Iowa City, IA; Wake Forest University, Winston-Salem, NC; Loyola University, Maywood, IL; Henry Ford Health System, Detroit, MI

Background: Interpretive variability is substantial for Pap test and histopathologic cervical specimens, and lack of agreement and standardization results in diagnostic errors and over and under treatment. Our goal was to determine the efficacy of specific interventions in improving interpretive reproducibility in cytotechnologists and pathologists.
Design: A total of 38 cytotechnologists and 52 pathologists from 5 labs established baseline reproducibility in interpreting Pap test and cervical biopsy specimens, performed interventions to improve, and then re-measured reproducibility using standardized slide study sets. Interventions were lab-specific, interactive and non-interactive, varying from one-on-one discussions and multi-head microscope conferences to didactic lectures. Interactiveness of individual interventions was graded on a 1 to 8 scale. We measured the differences between pre and post-intervention for pairwise agreement of lab cytotechnologists, cytologists, and surgical pathologists.
Results: Individual lab cytotechnologist, cytopathologist, and surgical pathologist pre and post-intervention site specific median values were 0.52 and 0.58, 0.52 and 0.46, and 0.53 and 0.48 respectively. The number of labs that showed post-intervention improvement in cytotechnologist, cytopathologist, and surgical pathologist agreement was 4, 1, and 1, respectively. Lab cytotechnologists who improved their interpretive reproducibility used interactive interventions (> score 3), and non-interactive interventions either resulted in lower or no change in interpretive reproducibility. Many surgical pathologists strongly resisted reproducibility evaluation.
Conclusions: Interpretive reproducibility of cervical specimen diagnoses may be improved, and improvement necessitates interactive interventions that appear to be more acceptable to cytotechnologists than pathologists. Current pathologist culture is a major barrier to designing and implementing effective interventions to improve interpretive reproducibility.
Category: Quality Assurance

Tuesday, March 27, 2007

Poster # 197, Tuesday Afternoon