Observer variation, dysplasia grading, and HPV typing: a review

Area of Science:

• Gynecologic Pathology
• Cytopathology
• Biostatistics

Background:

• Cervical squamous dysplasia assessment involves subjective interpretation, leading to observer variation.
• Quantifying this variation is crucial for reliable diagnosis and patient management.

Purpose of the Study:

• To statistically analyze observer variation in cervical squamous dysplasia grading.
• To explore the impact of classification systems on diagnostic agreement.
• To review the validation of the binary Bethesda system.

Main Methods:

• Statistical analysis using the kappa statistic to quantify inter-observer agreement.
• Information theory applied to understand sources of variation.
• Comparison of consensus diagnoses with human papillomavirus (HPV) typing for validation.

Main Results:

• The number of diagnostic categories significantly influences the kappa statistic.
• Information theory provides insights into the reasons for observer variation.
• The binary Bethesda system demonstrates utility in classifying cervical squamous dysplasia.

Conclusions:

• Standardized classification systems like the Bethesda system are essential for reducing observer variation in cervical dysplasia diagnosis.
• Statistical methods are vital for assessing and improving diagnostic reliability in pathology.