Review of the Bethesda System Atlas does not improve reproducibility or accuracy in the classification of atypical squamous cells of undetermined significance smears

BACKGROUND. The Bethesda System (TBS) and its accompanying atlas were developed to promote uniform diagnosis and reporting of cervical and vaginal cytology, especially with respect to borderline abnormal smears. The authors assessed whether group study of TBS atlas improves the reproducibility and accuracy of the cytopathologic diagnosis of equivocal Papanicolaou smears. METHODS. One hundred 'atypical' smears were divided into pretest and posttest sets containing equal numbers of negative, atypical squamous cells of undetermined significance (ASCUS), and squamous intraepithelial lesion (SIL) diagnoses based on a five-member panel review. Two comparable teams of four pathologists from George Washington University Medical Center (Washington, DC) and Kaiser Permanente (Portland, OR), each comprised of two more experienced cytopathologists and two less experienced pathologists, independently reviewed the 50 pretest slides and classified the slides according to TBS as negative, ASCUS, or SIL. The teams then conducted group study sessions using TBS atlas. After the review, the pathologists independently classified the 50 posttest slides in a similar manner. RESULTS. Pretest, pair-wise interobserver agreement ranged from 30% to 66% compared with 34-62% for posttest agreement. Absolute percent agreement of reviewers' diagnoses with a previously developed consensus diagnosis based on opinions of a five-expert panel (cytopathologic certainty scale) ranged from 44% to 62% for the pretest set and from 40% to 60% for the posttest set. Comparison of the detection of oncogenic human papilloma virus (HPV) DNA by hybrid capture tube test with smears classified as negative, ASCUS, or SIL revealed that seven of eight reviewers did not demonstrate a stronger association between HPV detection and cytologic diagnosis in the posttest set. CONCLUSIONS. Review of TBS atlas by itself does not appear to improve the reproducibility or accuracy of cytologic diagnoses. The lack of improvement was similar among the pathologists involved regardless of experience level or whether they had a close working relation. (C) 2000 American Cancer Society.