The Modified Fisher Scale Lacks Interrater Reliability

Christopher Melinosky, Department of Neurology, Lehigh Valley Health Network, 1250 S Cedar Crest Blvd, Suite 405, Allentown, PA, USA. christopher.melinosky@lvhn.org.
Hope Kincaid, Network Office of Research and Innovation (NORI), Lehigh Valley Health Network, Allentown, PA, USA.
Jan Claassen, Department of Neurology, Neurological Institute, New York Presbyterian Hospital, Columbia University, New York, NY, USA.
Gunjan Parikh, Department of Neurology, Program in Trauma, University of Maryland School of Medicine, Baltimore, MD, USA.
Neeraj Badjatia, Department of Neurology, Program in Trauma, University of Maryland School of Medicine, Baltimore, MD, USA.
Nicholas A. Morris, Department of Neurology, Program in Trauma, University of Maryland School of Medicine, Baltimore, MD, USA.

Abstract

BACKGROUND: The modified Fisher scale (mFS) is a critical clinical and research tool for risk stratification of cerebral vasospasm. As such, the mFS is included as a common data element by the National Institute of Neurological Disorders and Stroke SAH Working Group. There are few studies assessing the interrater reliability of the mFS. METHODS: We distributed a survey to a convenience sample with snowball sampling of practicing neurointensivists and through the research survey portion of the Neurocritical Care Society Web site. The survey consisted of 15 scrollable CT scans of patients with SAH for mFS grading, two questions regarding the definitions of the scale criteria and demographics of the responding physician. Kendall's coefficient of concordance was used to determine the interrater reliability of mFS grading. RESULTS: Forty-six participants (97.8% neurocritical care fellowship trained, 78% UCNS-certified in neurocritical care, median 5 years (IQR 3-6.3) in practice, treating median of 80 patients (IQR 50-100) with SAH annually from 32 institutions) completed the survey. By mFS criteria, 30% correctly identified that there is no clear measurement of thin versus thick blood, and 42% correctly identified that blood in any ventricle is scored as "intraventricular blood." The overall interrater reliability by Kendall's coefficient of concordance for the mFS was moderate (W = 0.586, p < 0.0005). CONCLUSIONS: Agreement among raters in grading the mFS is only moderate. Online training tools could be developed to improve mFS reliability and standardize research in SAH.