[1966] Inter- and Intra-Observer Agreement in Diagnosing Dysplasia in Barrett's Esophagus: Comparison of Routine Glass Slide vs. Digital Image Examination

Scott H Bradshaw, David Driman, Marc P Dupre, Vincent G Falck, Richard Kirsch, Abdelrazak Meliti, Bich Nguyen, David Owen, Amir Rahemtulla, Robert H Riddell, Heidi Sapp, Genevieve Soucy, Catherine Streutker, E Celia Marginean. TOH, Ottawa, ON, Canada; St Luc, Montreal, QC, Canada; St. Michael's, Toronto, ON, Canada; LHSC, London, ON, Canada; Mount Sinai, Toronto, ON, Canada; CLS, Calgary, AB, Canada; SHR, Saskatoon, SK, Canada; VGH, Vancouver, BC, Canada; BCCA, Vancouver, BC, Canada; QEII HSC, Halifax, NS, Canada

Background: Whole-slide imaging offers promise for Internet-based telepathology consultations between institutions and for use in daily sign out. The aim of this study is to investigate the inter- and intra-observer variability in diagnosis of Barrett's esophagus (BE) dysplasia, using digital images and glass slides.
Design: Sixty cases of BE with diagnoses including negative for dysplasia (NEG), low grade dysplasia (LGD), indefinite for dysplasia (IND), high grade dysplasia (HGD), intramucosal carcinoma (IMC) and invasive carcinoma (ICA) were scanned. The digital and glass slides were reviewed by 12 expert GI academic pathologists, with a 30 day washout period between reviews.
Results: Table 1 shows the inter-observer Kappa statistic using six diagnostic categories, and a simplified three category system: 1 NEG, 2 LGD/IND and 3 HGD/IMC/ICA. Table 2 shows the kappa statistic for the intra-observer variability when comparing glass slides vs. digital images. Overall, the inter-observer agreement is similar for glass and digital images, slightly better for glass slides. Agreement varies from slight to moderate using six categories, and improves to fair to substantial using three categories. The intra-observer agreement between glass and digital slides is similar to inter-observer agreement within the same media.

Inter-observer kappa
DxGlass (6)Digital (6)Glass (3)Digital (3)
NEG0.540.480.540.48
IND0.150.160.37*0.28*
LGD0.210.15  
HGD0.470.400.74**0.68**
IMC0.450.36  
ICA0.390.43  
Overall0.380.330.550.48
*Combined IND/LGD; **Combined HGD/IMC/ICA



Intra-oberver kappa
Observer6 Dx3 Dx
10.370.58
20.280.47
30.430.52
40.310.39
50.270.37
60.300.45
70.500.6
80.340.5
90.390.57
100.390.6
110.320.53
120.280.41
Overall0.350.50



Conclusions: There is less inter-observer variability with three categories than with six. It is unclear whether the intra-observer variability between glass and scanned images is due to a change in media or simply the result of baseline intra-observer variability within the same media. Future work will evaluate the intra-observer variability with no change in media. The potential for digital slide diagnosis in clinical practice warrants further validation of this technology.
Category: Quality Assurance

Monday, March 4, 2013 9:30 AM

Poster Session I Stowell-Orbison/Surgical Pathology/Autopsy Awards Poster Session # 298, Monday Morning

 

Close Window