Operational definitions improve reliability of the age-related white matter changes scale

Abstract

Although the age-related white matter changes (ARWMC) scale has been advocated to be applicable to both MRI and CT for assessing the severity of WMC, its inter-rater reliability on CT is only fair. We aimed to operationalize the ARWMC scale and investigate the effect of this operationalization on the reliability and validity on MRI and CT. Operational definitions of the ARWMC scale were derived from Erkinjuntti research criteria for subcortical vascular dementia and Scheltens scale. Using original and operationalized ARWMC scale, eight observers recorded the time for rating per MRI and per CT. We investigated the inter-rater and intrarater reliability as well as validity against volume using data from 97 stroke patients. Inter-rater reliability of the operationalized scale on CT (0.874, 95% confidence interval [0.780-0.934]) was better than the original scale (0.569, 95% confidence interval [0.247-0.775]). Its intrarater reliability on CT (0.869) and reliability on MRI (inter-rater: 0.860; intrarater: 0.838) was comparable with the original scale (CT intrarater: 0.750 and on MRI inter-rater: 0.845; intrarater: 0.853). The time required to administer the operationalized scale (4'2″ for MRI and 1'18″ for CT) was similar to that of the original scale (3'56″ for MRI and 1'16″ for CT). The original scale and operationalized scale also significantly correlated with WMC volume (operationalized scale ρ = 0.613, P < 0.001, original scale ρ = 0.638, P < 0.001). Operational definitions improve the inter-rater reliability of ARWMC scale on CT, and it correlates with volumetric measurement.