The importance of cytologic intrarater and interrater reproducibility: The case of ductal lavage

Kala Visvanathan, Deborah Santor, Syed Z. Ali, In Soon Hong, Nancy E. Davidson, Kathy J. Helzlsouer

Research output: Contribution to journalArticlepeer-review

Abstract

The reproducibility of a test result is a critical component of its clinical utility. Little information is available concerning the intrarater reproducibility of cytologic assessments. This study evaluated the reproducibility of cytologic interpretation of epithelial cells obtained from ductal lavage (DL), a minimally invasive method used to obtain sample cells from breast tissue. Two cytospin slides were made for each duct sampled. Slides with <10 cells were considered inadequate to make a diagnosis; the remaining slides were classified into mildly atypical, markedly atypical, and malignant cells. Each pair of slides were classified by the more serious diagnosis. DL samples from 100 ducts were independently blind-reviewed by two experienced cytopathologists. All abnormal slides and a random sample of normal slides and slides identified as inadequate for diagnosis (n = 43) were rereviewed. The κ for intrarater agreement was 0.59 ± 0.10 for cytopathologist 1 and 0.33 ± 0.08 for cytopathologist 2. The κ for interrater agreement of slides from 100 ducts was 0.46 ± 0.07. The interrater agreement of the slides that were rereviewed was κ = 0.27 ± 0.09. Fair to moderate intrarater and interrater agreement of DL cytology was observed. Low intrarater and interrater cytologic consistency may compromise the interpretation of clinical studies of DL.

Original languageEnglish (US)
Pages (from-to)2553-2556
Number of pages4
JournalCancer Epidemiology Biomarkers and Prevention
Volume15
Issue number12
DOIs
StatePublished - Dec 2006

ASJC Scopus subject areas

  • Epidemiology
  • Oncology

Fingerprint Dive into the research topics of 'The importance of cytologic intrarater and interrater reproducibility: The case of ductal lavage'. Together they form a unique fingerprint.

Cite this