Statistical Issues in Cervical Screening Cytology Device Evaluation

• Activity Number: 611
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Medical Devices and Diagnostics
• Author(s): Kyungsook Kim*+ and Meijuan Li
• Companies: FDA and FDA/CDRH
• Keywords: Cervical Cytology screening ; verification bias

Abstract

Cervical cytology by Papanicolaou (Pap) smears is an effective means of screening for cervical premalignant and malignant conditions. Cervical intra-epithelial neoplasia (CIN) and cervical cancer remain important health problems for women worldwide. The Cervical cytology device outcomes are ordinal multinomial data, and how to appropriately evaluate the device analytical and clinical performance pose many layers of statistical challenges. In this talk, we will discuss some statistical challenges and issues in study design, data analysis, and sample size calculation. For the current practice, the reference method collects the specimens first. Since the collection order is not a random, the systematic bias could happen due to the percentage of abnormal cells may not evenly distributed between the reference method and new device. In addition, over 95% patients in screening population are negative in cervical malignant conditions. In order to have sufficient malignant cases, very large number of patients is needed for the study, yet is may not be feasible or ethical to have all negatives being verified by a clinical reference standard. Also, the outcome of the screening test has se