Methods to improve the reliability of histopathological diagnoses in the placenta

The interobserver reliability of histopathological features in the placenta was examined. Two pathologists independently reviewed slides from 250 placentas. The pathologists were given a morphological description of the placenta, but were blinded to clinical status, gestational age and original diagnoses. A protocol for diagnosis and grading of features was first developed and pilot-tested. Definitions and criteria were refined and elaborated. A range of features was examined including inflammatory lesions, features indicative of reduced uterine blood flow and other miscellaneous histopathological changes. Weighted kappa coefficients were calculated. The effect of multiple features on reliability was examined by stratifying on the presence of a second feature and calculating stratum-specific kappa coefficients. Results indicated good to excellent agreement for diagnoses of chorioamnionitis, cord vasculitis, funisitis and villitis (kappa(w) range 0.70-0.83). Agreement between observers was more variable for the diagnosis of reduced uterine blood flow states. Excellent agreement was observed for the diagnosis of meconium staining of the placenta (kappa(w) = 0.79). In general, lower levels of agreement were observed for features in the presence of a second feature. Reproducible measures are a prerequisite to using placental histopathology for diagnostic and prognostic information. This study demonstrated reliable placental diagnoses can be achieved through a standardised protocol.