Error Assessment of Cytopathologic Diagnosis of EUS-FNA of Pancreatic Ductal Carcinoma
Shantel Hebert-Magee, Amanda Treece, Mohamad Eloubeidi, Isam Eltoum. University of Alabama at Birmingham, Birmingham, AL
Background: Endoscopic ultrasound(EUS)-guided fine-needle aspiration(FNA) of the pancreas is an effective modality in confirming pancreatic adenocarcinoma(PA). However, the extent of error occurring in cytopathologic diagnosis of EUS-FNA of PA is unknown. Moreover, the impact(harm) of these errors on clinical outcomes has not been evaluated.
Design: A database was constructed utilizing the clinical notes, endoscopic, pathologic, and radiographic reports, and tumor bank registry to correlate clinical management, progression, and outcome. Utilizing a modified standardized error classification proposed by Raab et al, we evaluated for type and cause of error, recurrent diagnoses, interobserver agreement of discrepancies, and the impact of diagnostic inaccuracy, particularly harm resulting in delayed diagnosis, management or inappropriate treatment.
Results: 2399 cases from 2000-2010 with corroborating histology existed for 506 cases allowing for evaluation of cytologic error. The database confirmed 274 PA and uncovered incongruity between 40(7.9%) of the cytopathologic and surgical diagnoses. 32 cases(80%) were available for cytohistologic disconcordance review. 81.3%(26/32) were due to cytology specimen sampling error(CSE), of which 96.2%(25/26) were of no/minimal harm, the remaining case had minor clinical significance(no morbidity). Cytology interpretation error(CIE) accounted for 12.5%(4/32), 3 cases were determined to have error which could impact patient outcome (minor/moderate morbidity), and 1 resulted in inappropriate management. CSE/CIE delayed diagnosis on average 26 days(4-186days) in FN cases. CIE of the one FP case resulted in major harm(inappropriate treatment). Specimens which were more prone to cytologic diagnostic error had more passes(avg.5.25) and were either associated with chronic pancreatitis or a mucinous cystic lesion. Individual CIE was not assessed in this study. Interobserver assessment of the discrepant cases showed 27/32 and 5/32 agreement and disagreement with the cytohistologic original interpretation. The kappa scores ranged from.87-.93 for pairwise agreement of discrepant cases.
Conclusions: We found that the clinical impact of CSE/CIE typically results in no/minimal error. However in patients with unusual presentation or suspicion for diagnostic/sampling error was low, the clinical impact was more often consequential(delayed diagnosis/management or inappropriate treatment). Recognizing the impact of error in cytopathologic diagnosis of EUS-FNA PA may create quality assurance practices to reduce error, minimize healthcare cost, and decrease morbidity.
Tuesday, March 20, 2012 1:00 PM
Poster Session IV # 287, Tuesday Afternoon