The need for training when using the NIH Stroke Scale

The Need for Training when Using the NIH Stroke Scale: Large numbers of mul ticentre acute stroke trials are under way or planned requiring a valid, ef ficient, and reliable measure of patients' status and outcome. An establish ed measure widely used in neurological deficits of acute stroke patients is the National Institutes of Health Stroke Scale (NIHSS), which can reliably be applied by well-trained raters. The question as to whether without syst ematic training or detailed instructions the NIHSS is still reliable has no t yet been addressed. Therefore we applied the NIHSS in 22 subacute stroke patients. Two of the raters had been trained extensively, two raters had no further information on the scale than given in the original NIHSS. The int erobserver reliability among trained raters was substantial (kappa=0.61) an d comparable to prior studies. Reliability was substantially poorer among t he untrained raters (kappa=0.33) as well as between trained and untrained r aters (K=0.45). Among the untrained raters a maximum difference between tot al scores for the same patient of 4 or more points was found in 4 patients and between trained and untrained raters even in 12 patients, Our results s how, that the established good interrater reliability of the NIHSS applies only to trained raters and not to untrained raters. One major problem for t he untrained raters was the lack of instructions for the assessment of the individual items gaze, visual field and dysarthria in comatose or obtunded, unresponsive and aphasic patients. We conclude that the NIHSS cannot be re liably applied whithout systematic training and knowledge of detailed instr uctions. A standardised use of the NIHSS is mandatory in future trials.