Clinical Question: Are PAPNET readings of conventional Papanicolaou (Pap) tests better than manual readings by cytologists?

Setting: Other

Study Design: Diagnostic test evaluation

Synopsis: Investigators compared PAPNET readings with manual readings on 21,747 conventional Pap tests. All samples were read using both PAPNET (a computerized system for reading conventional Pap tests but not ThinPrep slides) and a manual reading by one of 20 cytologists, with randomization of the order of reading. Blinding of the readings was not described, but a similar number of slides were reclassified from both groups. Concordant results always were considered accurate. There were 372 discordant readings. The gold standard was additional manual readings by two expert cervical cytologists and histologic results of biopsies.

The gold standard evaluation determined that PAPNET reading diagnosed four additional true-positive results and two additional false-positive results than manual reading (not statistically significant). With use of PAPNET, there was a small decrease in the proportion of Pap tests considered unsatisfactory for reading (1.86 percent versus 2.13 percent for manual reading; P = .002; number needed to screen = 400).

Bottom Line: PAPNET is as accurate as manual readings by cytologists for abnormal diagnoses. The choice of using PAPNET or a manual reading by a cytologist should be based on cost. (Level of Evidence: 1b)