Validity and Reliability of Retrospective NIH Stroke Scale Assessment for Initial Stroke Severity

Yoon, Sung Sang; Bu, Seon Hee; Park, Key Chung; Chang, Hye Jung; Kwon, Young Dae

Abstract

BACKGROUND: The retrospective severity scoring system of NIHSS for acute stroke patients was found to be valid and reliable by using patients' medical records in studies abroad. However, in Korea, medical records are often summarized and contain missing information which makes it difficult to conduct a restropective outcome study. The purpose of this research was to evaluate the validity and reliability of the retrospective NIHSS scoring system according to patients' medical records with written clinical histories and physical admission notes.
METHODS
An algorithm for the retrospective NIHSS scoring system was developed and applied to 75 patients with acute ischemic stroke. Missing data on physical examination results were scored as normal. One neurologist who was blinded to this study measured the NIHSS score prospectively on the initial patient examination. After the patient's discharge, two other neurologists, blinded to the patient's clinical condition, evaluated the NIHSS score independently based on the information collected from the patient's physical admission notes. The criterion-related validity was evaluated by the Pearson Correlation Coefficient, and the measure of agreement between two raters was evaluated by the Kappa Statistic.
RESULTS
The criterion-related validities of the retrospective NIHSS scoring system were high in the total and each itemized scores, except for the items of LOC command, limb ataxia, dysarthria, and neglect. The interrater reliabilities were also high except for the items of LOC command, limb ataxia, and dysarthria.
CONCLUSIONS
The retrospective NIHSS scoring algorithm was found to be a reliable and unbiased tool even when some physical examination elements are missing from the written medical records.