Table 4.
Univariable and Multivariable Predictors of Interobserver Reproducibility in the Hematoxylin-Eosin (H&E) Diagnosis of Squamous Versus Nonsquamous Carcinoma
| Variable | Univariable Predictors, OR (95% CI) | Multivariable Predictors, OR (95% CI) | Interpretation |
|---|---|---|---|
| Tumor factors | |||
| Gradea | 0.722 (0.646–0.808) | 0.812 (0.714–0.922) | Decreased agreement in more-poorly differentiated tumors |
| H&E slide factors | |||
| Qualityb | 0.573 (0.494–0.665) | 0.689 (0.576–0.823) | Decreased agreement if one or both pathologists felt the slide was of low quality |
| Confidencec | 2.21 (1.86–2.63) | 2.02 (1.68–2.41) | Increased agreement if both pathologists were highly confident |
| Pathologist factors | |||
| Difference in pathologist experienced | 0.904 (0.864–0.945) | 0.901 (0.860–0.944) | For every 5 y in experience difference, there is a 10% decrease in the odds of agreement |
| Cumulative pathologist experiencee | 1.03 (1.01–1.07) | 1.03 (1.00–1.06) | For every 5 y of combined experience, there is a 3% increase in the odds of agreement |
| Pulmonary pathology expertf | 1.38 (1.21–1.58) | 1.21 (1.04–1.41) | Increased agreement if both are expert |
Abbreviations: CI, confidence interval; H&E, hematoxylin-eosin; OR, odds ratio.
Grade based on the original pathologic diagnosis (well-differentiated, < moderately differentiated, < poorly differentiated); the OR for 1 unit increase in grade.
The OR comparing cases where one or more pathologists felt the quality of slide was insufficient to cases where both felt the quality was sufficient.
The OR comparing cases where both pathologists had high confidence to any other confidence pairing.
The OR for difference in experience between the pathologists per 5-year block.
The OR for combined sum of experience for the pathologist pair per 5-year block.
The OR comparing cases where both pathologists were experts compared with neither being expert.