Effect of circulation and discussion of cervical smears on agreement between laboratories

The study objective was assessing whether circulation of smears and discussion of those with differing interpretation can increase reproducibility between laboratories. The study included: the blind interpretation of a first set of 194 smears among seven laboratories, the discussion of smears with discrepant diagnoses during the previous phase and the blind interpretation of a second set of smears of same size and characteristics. After discussions, the overall weighted kappa increased in five laboratories (substantially in three : +50%, +27% and +20%). However, no change was observed in one laboratory and a slight decrease (-4%) in another. The latter interpreted the second set of smears at a longer time interval from discussions. Agreement improved for all diagnostic classes except low grade intraepithelial neoplasia (LSIL). Overall, the intervention increased diagnostic agreement, but its effect varied with laboratory and by diagnostic class and could be transient. Continued programmes of smear exchange and discussion appear to be advisable.