Surgeon interrater reliability in the endoscopic assessment of cistern scarring and aqueduct patency

Abstract

OBJECTIVE The success of endoscopic third ventriculostomy with choroid plexus cauterization may have associations with age, etiology of hydrocephalus, previous shunting, cisternal scarring, and possibly aqueduct patency. This study aimed to measure interrater reliability among surgeons in identifying cisternal scarring and aqueduct patency. METHODS Using published definitions of cistern scarring and aqueduct patency, 7 neuroendoscopists with training from Dr. Warf in Uganda and 7 neuroendoscopists who were not trained by Dr. Warf rated cistern status from 30 operative videos and aqueduct patency from 26 operative videos. Interrater agreement was calculated using Fleiss' kappa coefficient (κ). Fisher's 2-tailed exact test was used to identify differences in the rates of agreement between the Warf-trained and nontrained groups compared with Dr. Warf's reference answer. RESULTS Aqueduct status, among all raters, showed substantial agreement with κ = 0.663 (confidence interval [CI] 0.626-0.701); within the trained group and nontrained groups, there was substantial agreement with κ = 0.677 (CI 0.593-0.761) and κ = 0.631 (CI 0.547-0.715), respectively. The identification of cistern scarring was less reliable, with moderate agreement among all raters with κ = 0.536 (CI 0.501-0.571); within the trained group and nontrained groups, there was moderate agreement with κ = 0.555 (CI 0.477-0.633) and κ = 0.542 (CI 0.464-0.620), respectively. There was no statistically significant difference in the amount of agreement between groups compared with Dr. Warf's reference. CONCLUSIONS Regardless of training with Dr. Warf, all neuroendoscopists could identify scarred cisterns and aqueduct patency with similar reliability, emphasizing the strength of the published definitions. This makes the identification of this risk factor for failure generalizable for surgical decision making and research studies.